AD_________

Award Number: DAMYDl7-94-J-4129

TITLE: Identification of BRCAl and 2 Other Tumor Suppressor Geneson Chromosome 17 Through Positional Cloning

PRINCIPAL INVESTIGATOR: Raymond L. White, Ph.D.

CONTRACTING ORGANIZATION: University of UtahSalt Lake City, Utah 84102

REPORT DATE: April 2000

TYPE OF REPORT: Final

PREPARED FOR: U.S. Army Medical Research and Materiel Command

Fort Detrick, Maryland 21702-5012

DISTRIBUTION STATEMENT: Approved for public release;distribution unlimited

The views, opinions and/or findings contained in this report arethose of the author(s) and should not be construed as an officialDepartment of the Army position, policy or decision unless sodesignated by other documentation.

20010322 155

Form ApprovedREPORT DOCUMENTATION PAGE OMB No. 074-0188

Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining

tVe data needpd, and completing and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for

reducing this burden to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of

Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503

1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED

IApril 2000 Final (1 Jul 94 - 17 Mar 00)

4. TITLE AND SUBTITLE 5. FUNDING NUMBERS

Identification of BRCAI and 2 Other Tumor Suppressor Genes DAMDI7-94-J-4129on Chromosome 17 Through Positional Cloning

6. AUTHOR(S)Raymond L. White, Ph.D.

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION

University of Utah REPORT NUMBER

Salt Lake City, Utah 84102

E-MAIL:ray.white@hci.utah.edu9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING / MONITORING

AGENCY REPORT NUMBER

U.S. Army Medical Research and Materiel CommandFort Detri&k, Maryland 21702-5012

11. SUPPLEMENTARY NOTES

12a. DISTRIBUTION / AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE

Approved for public release; distribution unlimited

13. ABSTRACT (Maximum 200 Words)

The overall goal of this project is to identify genes involved with the development and progression of breastcancer. This goal has remained unchanged since the start of the project, however the discovery of BRCA1 in1994 together with technological advances in gene expression profiling has influenced our strategy to achievethis goal. In the early part of the project our search for tumor suppressor genes was directed by genetic or LOHmapping strategies followed by positional cloning of candidate genes. As we proposed in our revised statementof work (SOW), we have now focused our efforts entirely on microarray-based comparisons to identify breastcancer related genes. Our laboratory has gained access to this technology through collaboration with MolecularDynamics and now we have established microarray spotting and scanning systems with over 40,000 minimallyredundant sequence-verified human cDNA clones. Once candidate genes have been identified, they will befurther characterized using a retroviral-based conditional expression system to assess changes in characteristicspertaining to morphology and growth rate by culturing human breast epithelial cells in an extracellular matrix.

14. SUBJECT TERMS 15. NUMBER OF PAGESTumor Suppressor Genes, LOH Mapping, Chromosome 17, PhysicalMapping, Genetic Mapping, CDNA Screening, Humans, Anatomical 81Samples, Mutation Detection, Breast Cancer

16. PRICE CODE

17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACTOF REPORT OF THIS PAGE OF ABSTRACT

Unclassified Unclassified Unclassified Unlimited

NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)Prescribed by ANSI Std. Z39-18298-102

FOREWORD

Opinions, interpretations, conclusions and recommendations arethose of the author and are not necessarily endorsed by the U.S.Army.

__Where copyrighted material is quoted, permission has beenobtained to use such material.

__Where material from documents designated for limiteddistribution is quoted, permission has been obtained to use thematerial.

__Citations of commercial organizations and trade names in thisreport do not constitute an official Department of Armyendorsement or approval of the products or services of theseorganizations.

In conducting research using animals, the investigator(s)adhered to the "Guide for the Care and Use of LaboratoryAnimals," prepared by the Committee on Care and use of LaboratoryAnimals of the Institute of Laboratory Resources, nationalResearch Council (NIH Publication No. 86-23, Revised 1985).

X For the protection of human subjects, the investigator(s)adhered to policies of applicable Federal Law 45 CFR 46.

X In conducting research utilizing recombinant DNA technology,t-he investigator(s) adhered to current guidelines promulgated bythe National Institutes of Health.

X In the conduct of research utilizing recombinant DNA, theinvestigator(s) adhered to the NIH Guidelines for ResearchInvolving Recombinant DNA Molecules.

in the conduct of research involving hazardous organisms, theinvestigator(s) adhered to the CDC-NIH Guide for Biosafety inMicrobiological and Biomedical Laboratories.

PI - 1ignature 'Date

(4) TABLE OF CONTENTS

Front Cover ................................................................................................... 1SF 298, Report Documentation Page ...................................................................... 2Foreword ..................................................................................................... 3Table of Contents............................................................................................. 4Introduction ................................................................................................... 5Body ........................................................................................................... 6Key Research Accomplishments........................................................................... 12Reportable Outcomes ....................................................................................... 12Conclusions.................................................................................................. 12References.................................................................................................... 13Appendices:

Materials and Methods.................................................................................. 15Tables .................................................................................................... 16Figures .................................................................................................... 16Reprints .....................................................................................................Publications.............................................................................................. 18List of Salaried Personnel .............................................................................. 18

4

(5) INTRODUCTIONThe identification and characterization of genes involved in development and progression of

breast cancer is critical to an understanding of the biological mechanisms that regulate growth ofcells in breast epithelium. The emphasis of this project has been originally directed towards theidentification and characterization of several breast cancer tumor suppressor genes located onchromosome 17. Common for each of these genes was that their approximate physical location hadbeen delineated either through genetic linkage analysis of hereditary segregation patterns or bybeing the target of recurring deleterious events observed as loss of heterozygosity (LOHl) in sporadictumors. Since the discovery of the BRCA1 gene by other groups, our experimental focus have beendirected towards the identification of four presumed tumor suppressor genes defined by LOH onchromosome 17, as proposed in our revised statement of work (SOW). Of these four loci, two arelocated at each extremity of the chromosome, while the remaining two areas of LOHl we haveidentified flank either side of the BRCA1 locus at 17q21. However, contrary to hereditary tumorsuppressor genes whose position may be determined by genetic linkage analysis, the localization oftumor suppressor genes by LOH is circumstantial. LOHl does not provide direct evidence as towhether the region of LOH actually contains a gene critical to carcinogenesis or whether thedeletion is coincidental. Statistically significant areas of LOH indicative of a tumor suppressor genecan only be determined by analyzing large numbers of specimens. At the same time it holds true thatcoincidental deletions provide grounds for mis-interpretation of the position of an actual tumorsuppressor gene. Despite those drawbacks, when our proposal was submitted, this paradigmrepresented the most efficient approach to identifying tumor suppressor genes. However, it hasbecome exceedingly clear that the massive resources applied in various genome centers to establishboth physical and expression maps covering the entire human genome far exceeds the amount ofdata that we were able to generate even within the modest areas of chromosome 17 where we haveconcentrated our attention. The past three years have been a period of transition for the project andearly on during this period we evaluated the efficiency of our strategy compared to newer, moreglobal strategies to identify genes critically involved with tumor development and progression. Wehave for a long time realized that LOHl guided search for tumor suppressor genes is vulnerable tocritique on several levels. Most importantly among these the rather poor delineation of the tumorsuppressor loci, which consequently requires a significant expansion of the physical mappingcomponent of the project and thus dramatically augment the number of candidate genes to evaluate.A different and much broader strategy for identifying any kind of gene involved in tumor formationand progression, which is not dependent on prior knowledge of location, employs genetic profilingusing high density microarrays. Not only does this technique permnit detection of the expressionprofiles of each of a large number of genes in parallel, but it potentially also provides a mechanisticview of how regulatory pathways are controlled. As we proposed in our newly revised statement ofwork (SOW), we have now focused our efforts entirely on microarray-based comparisons to identifybreast cancer related genes. This includes development of robust fluorescent labeling andhybridization protocols as well as the preparation and testing of redundant human cDNA targetsamples for deposition on the microarray slides. Our laboratory has gained access to this technologythrough collaboration with Molecular Dynamics who has provided us with an array robot and ascanning device. In addition, during the past year, we have gained over 40,000 minimally redundantsequence-verified human cDNAs for deposition on the microarray slides. As proposed, we havecompared expression profiles from several distinct breast cell lines. By scanning 2400 clones on ourfirst slide array we have discovered 29 genes and ESTs that reveal altered expression patterns as a

5

consequence of conditionally expressed dominant-negative P3-catenin in primary breast cells. Oncethe full-length sequence of each of the differentially regulated genes has been ascertained andprioritized, they will be introduced into appropriate cell lines to study their effect on cellmorphology and gene regulation. Experiments have also been carried out to observe regulatoryaffects of experimentally transformed breast cells grown in matrigel and comparisons of a largercollection of primary cells and established cell-lines is being designed.

(6) BODYThe following progress were made during the funded years

Statement of Work - revised October 23, 1995Task 1, Isolation and characterization of two potential tumor suppressor genes

approximately 1Mb proximal and 1Mb distal of BRCA1.During the first twelve months of this fundind period, several discoveries have been made

with respect to breast cancer. Most significant was the discovery of Mild et al. (1994) and Futreal etal. (1994) of the BRCA1 gene itself; however, the genomic environment of BRCA1 has alsorevealed several interesting features. Dr. Solomon's group showed that the 5' of BRCA1 gene is invery close proximity to the 5' of the 1A1-3B gene (Brown et al., 1995), raising the possibility ofstudying transcriptional control (transcriptional dominance, positive or negative interference)among these two genes. Our group has found that the L21 riboprotein is located within the BRCA1locus, in an intron flanking BRCA1 exon 14. We have made other potentially very interestingobservations by analyzing primary prostate cancers for allelic loss; those results have suggested thepresence of two additional tumor suppressor genes in the immediate vicinity of BRCA1 (Brothmanet al. 1995, see attached reprint). This new information placed us in a position where the resourcesavailable through this grant were adequate to pursue to identify the two potential tumor suppressorgenes immediately flanking BRCA1 and located within our existing physical map of the BRCA1region. We have identified and published possible candidate genes from these regions during thisperiod: two novel members of the DLG family of genes, DLG2 (Mazoyer et al. 1995, see attachedreprint), and DLG3 (Smith et al. 1996, see attached reprint) and an ADP-ribosylation factor (Smithet al. 1995, see attached reprint).

Common to all members of the Drosophila discs-large family of genes are three distinctstructural domains. At the N-terminal are 1-3 somewhat degenerate Drosophila homology regions(DHRs). DHR motifs are approximately 90 amino acids long and have been shown in vitro to bindcytoskeletal proteins of the band 4.1 family. A region with homology to src oncogene motif 3 (SH3)is found in the central part of the discs-large proteins. This motif, approximately 60 amino acids inlength, is known to be a site of protein-protein interactions. Finally, a guanylate kinase domain(GK) is found at the C-termini of the Dlg proteins. The function of this domain is the catalytictransfer of phosphate from ATP to GMP, forming GDP. In Drosophila, Dlg has been shown to be atumor suppressor (Woods et al., 1989). Additional, and quite compelling, evidence for thebiological importance of the DLG family of genes was recently discovered when it was found thatAPC, the tumor suppressor gene responsible for adenomatous polyposis coli (Groden et al., 1991),interacts on the protein level with the human DLG homolog (Matsumine et al., 1996).

ADP-ribosylation factor (ARF4L) is, based on its predicted protein structure, believed to beinvolved in membrane trafficking and protein secretion. Six protein domains have been identified,

6

three of which are involved with phosphate/magnesium binding, while the remaining three areinvolved with guanine nucelotide binding.

In an experiment designed to directly determine if DLG2, DLG3 and ARF4L are the targetsfor the LOH observed in sporadic breast cancers, we obtained paired tumor and normal tissuesamples from 10 individuals who underwent surgical removal of malignant carcinomas. Weextracted DNA and RNA from these samples, and subsequently tested the paired DNA samples forloss of heterozygosity (LOH). As indicated in figure 1, we used five highly polymorphic geneticmarkers located along the long arm of chromosome 17 (Albertsen et al. 1994a, see attached reprint),three of which lie relatively close to the BRCA1 locus 17 (Albertsen et al. 1994b, see attachedreprint). From among the ten tumors we identified 3 which displayed LOH around BRCA1.According to the established model for LOH involving tumor suppressor genes, the allele remainingin the tumor sample would harbor the deleterious mutation. Using RNA extracted from thesetumors we prepared first-strand cDNAs specific to each of the three genes and submitted thesetemplates for automated sequencing on an AB1373A sequencer (Applied Biosystems, Foster City,CA). As none of the samples we have sequenced have revealed any mutations, we have no evidenceso far to indicate that either DLG2, DLG3 or ARF4L is the primary target in the region of LOHflanking BRCA1.

Another part of our project that has yielded good progress during this period is theidentification of two novel genes from the region of LOH encompassing the plakoglobin locus. Wediscovered these genes in collaboration with Dr. Robert Callahan at the National Cancer Instituteusing the P1 clones 50H1 and 122F4 identified in our laboratory (Albertsen et al. 1994b, seeattached reprint). The first of these genes is the presumed human homologue of the mouse FKBP65gene (Coss et al., 1995). Genes of the FKBP family derive their names from the immunosuppressantmacrolide antibiotic FK506, because they mediate its activity (in part) by binding to a ubiquitousfamily of highly conserved intracellular receptors termed immunophilins (Sigal and Dumont, 1992).Although FK506 is known to block various signal transduction pathways in normal T-cells, FKBPgenes (including FKBP65) are expressed in most tissues that have been analyzed. The biologicalrelevance of human FKBP65 to cancer development remains unclear, but once its full-lengthsequence is ascertained we will undertake a detailed analysis of the gene and its functional domains.The second gene we identified in the plakoglobin region was ascertained by coincidence. As part ofthe process of determining the genomic structure of the FKBP65 gene, we sequenced severalgenomic subclones derived from P1 phage clones 50H1 and 122F4. While analyzing the sequenceof one of these subclones, named 1H2M, I identified a small collection of human ESTs that shared asegment of almost 300 nucleotides of perfect homology to 1H2M. Further analysis and databasecomparisons extended the DNA sequence to approximately 1300 nucleotides, and revealed that thenovel gene shared homology with no other currently known vertebrate gene. However, the proteintranslation of the nucleotide sequence showed 40% homology, over a segment of almost 200 aminoacids, to an uncharacterized gene from C. elegans. It is impossible at present to predict thebiological relevance of the novel gene with respect to tumor formation, but the high degree ofprotein homology preserved across such distant species suggests a fundamental and probably criticalrole.

Another gene that occupied a significant amount of our time and effort was DOC-2, whoseidentification and characterization was published (Albertsen et al. 1996, see attached reprint). A788-basepair segment of DOC-2 was originally identified by differential display between ovariancarcinoma and normal ovarian epithelial tissue; its expression was greatly reduced or entirely absent

7

in a panel of 10 ovarian tumors (Mok et al., 1994). Our ascertainment of DOC-2 was based oncDNA screening of a fetal retina library (Stratagene # 937202) with P1 clone 124D3. One of thecDNA clones we identified, 1RA1, harbored a large segment of the genuine DOC-2 gene; however,we did not realize immediately that the 1RA1 cDNA clone was a chimera between DOC-2 andDLG3. Consequently, our original attempt to verify the chromosomal location of the 1RA1 cloneclearly indicated that the clone was located in the BRCA1 region. It was not until we were in theprocess of determining the genomic structure of DOC-2 that we realized our mistake and found thecorrect genomic location of DOC-2 on chromosome 5. Nevertheless, the complete sequence,genomic characterization, and chromosomal location of DOC-2 have been attributed to the presentArmy grant.

Additional expressed sequences from the two regions in question exist, and we intend to dothis by screening for mutations in these genes in sporadic tumors which reveal LOH. Alternatively,the genes will be tested for their biological roles in tissue culture systems, either through DNA oligoantisense-mediated suppression of normal gene expression or by transfecting breast epithelial cellswith conditionally induceible expression vectors carrying the genes.

Statement of Work - revised October 23, 1995Task 2, To identify and characterize the breast cancer tumor suppressor genes on distal

17q and distal 17p using physical reagents identified by the CEPH.We initiated a physical mapping project to refine the rather crude physical map of 17p. We

have chosen to focus solely on this region in the initial stages of the physical mapping project fortwo reasons: a) the 17p region of LOH is better characterized and thereby provides a betterpossibility for identifying the tumor suppressor gene(s) located there; and b) with only limitedmanpower available to satisfy all aspects of our research proposal it would be unwise to furtherdilute our efforts by simultaneously attempting to refine the physical map of the 17q region. Duringthe first year two developments relating to the chromosome 17p region led us to initiate a refinedphysical mapping of l7pter. Of greatest importance was the publication of two novel candidatetumor suppressor genes in a quite narrowly delimited region of l7p13.3 (Schultz et al., 1996).While it is not clear whether either of these genes is the target for LOH near the telomere of theshort arm of chromosome 17, an important conclusion can be drawn: the very distal location of thepresumed tumor suppressor locus at the extremity of the chromosome means that the generalnumeric relation between genetic distance measured in cM to the physical length measured in Mb,which exists in the central parts of chromosomes, can not be applied in this case because thefrequency of genetic recombinations is elevated in telomeric regions. This phenomenondramatically skews the relationship between genetic distance and physical distance, such that a largegenetic distance actually is contained within a relatively short physical segment of DNA. Schultz etal., (1996) showed that the 3.5 cM of genetic distance between D17S5 and D17S28, which normallywould be expected to represent a 3.5-Mb genomic region, in reality is contained within a singlecosmid (30kb). If the genetic compression observed in this short telomeric region of the short armof chromosome 17 extends beyond the confined segment delimited by the markers D17S5 andD17S28, the task of refining the physical map in the approximately 15-cM region which harbors thepresumed tumor suppressor gene(s) should be relatively simple. To this end we are now in theprocess of obtaining yeast artificial chromosomes (YACs) that have been localized to telomeric 17pthrough the genome mapping efforts at the CEPH and the Whitehead Institute at MIT. By crossreferencing these YACs with the genetic markers we have developed and mapped to this region

8

(Gerken et al. 1995, see attached reprint), we can identify the exact extent of the existing physicalcoverage of the region. The information obtained in that way will allow us to localize potential gapsin the physical maps and will provide guidance as to where the genomic coverage must be expandedto complete the physical map of the region.

Our initial strategy for identifying tumor suppressor genes in areas of LOH on chromosome17 was based on a positional cloning strategy. This strategy is divided into three separate stages.During the first stage a physical map of a the area of interest is constructed using genomic cloneslike YACs, BACs or Pls. Following this stage expressed sequences are ascertained, and as the laststage the candidate genes are analyzed for mutations in breast tumors to determine if indeed they arethe sought-for tumor suppressor gene. While we successfully have used this approach in the past(Cawthon et al. 1990; Groden et al. 1991; Joslyn et al. 1991; Viskochil et al. 1990), we recognizethat this strategy is both labor intensive and only moderately efficient. The inherent inadequacies ofthe strategy have been further accentuated by the efforts of various genome centers to physicallymap and sequence the entire genome and to position expressed sequences along each chromosome,which have resulted in a very extensive overlap with our efforts. This situation has been difficult forus to continue and has required us to redefine the means and methods necessary to achieve theoriginal objective of our project; to identify and characterize genes involved in breast cancerdevelopment and progression. Having evaluated the situation, we determined that we wouldhave tochoose one of the following two experimental paths: a) determining the full-length sequence ofpreviously mapped genes and screening them for mutations in breast tumors, or b) implementing anovel cancer gene identification strategy with a wide scope and of high efficiency. We chose tofollow the latter path and in the following sections we describe the two approaches we haveimplemented to identify differentially expressed genes and to determine genetic expression profiles.

Statement of Work - revised July 28, 1997Task 1, To apply the Microarray Scanning technique to identify genes displaying

altered expression levels among breast cell lines.During the past 3 years most of our efforts have been focused on the completion of Task 1 in

our newly revised statement of work (SOW). As part of this effort is has been our primary goal toestablish microarray analysis as a reliable and reproducible technology to compare gene expressionprofiles and identify differentially regulated genes. Work has been carried out in three areas being a)development of fluorescent labeling and hybridization protocols, b) selection and establishment of acollection of target genes, and c) gene expression comparisons between several different breast celllines.

Development of probe labeling and hybridization protocols.One of the most critical factors for a successful microarray experiment is the preparation of

fluorescently labeled first strand cDNA to probe the microarray. The Molecular DynamicsMicroarray System allows for dual color hybridization and the two fluorescent dyes we used areCy3 and Cy5 from AP-Biotech. The labeling procedure itself is a multi step procedure that involvesextraction of total RNA, mRNA purification and first strand cDNA synthesis. The fluorescent dyesconjugated to dCTP are incorporated into the cDNA to generate the fluorescent hybridization probe.We have found that SuperScript II results in better probes than AMV and MMLV reversetranscriptase, probably because it lacks proof reading ability. A variety of hybridization formatshave also been tested. The most critical features of a successful hybridization with a complex probe

9

are to achieve and maintain high probe concentration during hybridization. To increase the probeconcentration we have determined that the smallest practical hybridization volume is 20ýd under a22mmx60mm coverslip. Steps must also be taken to maintain proper salt concentrations in thehybridization buffer during incubation (i.e. eliminate evaporative effects). We have found thatsealing the coverslip to the slide during hybridization leaves a fluorescent signature that affects thehybridization signal. In the procedure we currently employ the coverslip is not sealed, but theincubation takes place in a closed humidified chamber.

Selection and preparation of target cDNA clones.A key reagent in microarray analysis is the collection of genes deposited on the array slide

against which the levels of expression are measured. In our last annual report, we mentioned that wehad obtained 23,000 minimally redundant cDNA clones from Genome Systems (St. Louis) andResearch Genetics (Huntsville, Al). The procedures we have implemented to process this largenumber of clones are divided into multiple steps. In the first, mini cultures of each clone weregrown and each plasmid DNA was purified in the 96-well plate format. Using the plasmid as atemplate in a PCR reaction in conjunction with vector based primers, the insert cDNAs wereamplified. Following gel analysis and colorimetric analysis of the PCR yield, the PCR productswere purified, still using the 96-well format, and deposited to the array slides. Although these23,000 cDNA clones had been successfully used for our microarray experiments (such as Karpf etal. 1999, see attached reprint), we were later informed that about 30% of cDNA clones had beenmislabeled by the company. However, fortunately during the past 6 months we were able to gain anew set of over 40,000 minimally redundant sequence-verified human cDNA clones from ResearchGenetics. These sequence-verified cDNA clones are now successfully processed and ready for thelarge-scale microarray experiments.

Analysis of primary breast cells with conditionally induced dominant fi-catenin.P3-catenin has resently been shown to act as an oncogene in certain biologic models and to

study the gene regulatory effects of the cancer pathway driven by Wnt signaling in breast cells,human primary epithelial mammary cells were transfected with a conditionally suppressibledominant-negative 03-catenin construct as mentioned above. Excessive stimulation of this pathwayand certain types of mutations have been shown to involve the formation of a persistenttranscriptionally active complex of 13-catenin and LEF1. mRNA from several different primarybreast cultures, with and without induced 13-catenin, was prepared and fluorescently labeled for dualcolor hybridization on microarray slides representative of 2400 unique cDNAs. Twenty-ninedifferent genes have been identified in repeated experiments as differentially expressed and arelisted in Table 1. To confirm that the microarray observations are correct Northern analysis of eachdifferentially expressed gene is being undertaken. In Figure 1 we show that Psoriasin, whichdisplayed a 3-4 fold increase in abundance judging from the microarray analysis reveal similardifferences in abundance when tested by Northern analysis.

Statement of Work - revised July 28, 1997Task 2, Characterization of growth and morphologic effects of conditionally expressed

candidate cancer genes in cultured breast cells.Comparison of three conditional expression vector systems.To identify a conditional expression system we compared a dominant-negative P-catenin

construct (N-terminal deletions, dN131 and dN151) in three different conditionallyinducible/suppressible vectors, pRetro ON, pRetro OFF (Clontech) and LINX (Hoshimaru et al.

10

1996). These vector DNAs were transfected into the Phoenix Amphotropic packaging cell line (Dr.Garry Nolan) and the resulting infectious viral particles produced were used to infect primaryhuman mammary epithelial cells. After two weeks of drug selection in either 0.5 [tg/ml Puromycin(for pRetro ON and pRetro OFF) or 100 [tg/ml G418 (for LINX), cells were treated with or withoutlOng/ml doxycycline (DOX) for 48 hours. Total cell extracts were subjected to Western analysisusing anti-C-terminal 13-catenin antibody (Figure 2a). To demonstrate that induced dominant 13-catenin is functional, cells were infected with LINX vector control and LINX with inducibledominant 13-catenin gene (dN131) and subsequently transiently transfected with luciferase reporterconstruct, TOP FLASH, containing LEFM responsive element in its promoter (obtained from Dr.Hans Clevers). As shown in Figure 2b, the relative activity of the LEFl-reporter construct showed2.5 fold increase in activity following induction of 13-catenin. Based on these results we haveselected the LINX for future experimentation.

Characterization of growth and morphologic effects in cultured breast epithelial cells.The consequences of mutations in regulatory genes, such as p53, Rb and BRCA1 genes,

should be associated with observable cellular and molecular changes. Bisssell et al. have shown thatculturing human mammary epithelial cells in an extracellular matrix called Matrigel provides afunctionally relevant microenvironment conductive to form three-dimensional structures collagen(Weaver et al. 1996, Weaver et al. 1997). Normal cells will form differentiated spheroids with acentral lumen. These structures express certain cell lineage markers such as sialomucin at the apicalmembrane and type IV collagen at the basal membrane. Breast cancer cells, however, aredisorganized and lack polarized expression of these markers (Weaver et al. 1996, Weaver et al.1997). In collaboration with the Bissell laboratory, we have established three-dimensional culturesystem in our laboratory. As an initial attempt to test the sensitivity of this culture system onmorphology and growth due to the altered expression of tumor suppressor or growth control genes,we created human mammary epithelial cells (HMEC) deficient for pRB by infecting primaryoutgrowth from breast organoids with the human papillomavirus type 16 (HPV16) E7 gene(Spancake et al. 1999, see attached reprint). HPV16 E7 binds to and inactivates pRB, and alsocauses a significant down-regulation of the protein. Culturing normal HMIEC in a reconstitutedbasement membrane (rBM) provides a correct environment and signaling cues for the formation ofdifferentiated, acini-like structures. When cultured in this rBM, HMEC+E7 were found to respondmorphologically as normal HMEC and form acinar structures. In contrast to normal HMEC, manyof the cells within the HMEC+E7 structures were not growth arrested as determined by a BrdUincorporation assay. pRB deficiency did not affect polarization of these structures as indicated bythe normal localization of the cell-cell adhesion marker E-cadherin and the basal deposition of acollagen IV membrane. However, in HMEC+E7 acini we were unable to detect byimmunofluorescence microscopy the milk protein lactoferrin or cytokeratin 19, both markers ofdifferentiation expressed in the normal HIMIEC structures. These data indicate loss of RB in vivowould compromise differentiation, predisposing these cells to future tumor-promoting actions,suggesting that this culture system would be suited to examine the function of other tumorsuppressor and growth control genes.

11

(7) KEY RESEARCH ACCOMPLISHMENTS"* DLG2, DLG3, and ARF4L genes were mapped to BRCA1 flanking region. However, we

have no evidence so far to indicate that either DLG2, DLG3 or ARF4L is the primary targetin the region of LOH flanking BRCA1.

"* We have established the microarray spotting and scanning system with over 40,000minimally redundant sequence-verified human cDNA clones.

"* We have established a reliable retroviral-based conditional expression system."* We have established a system to examine the growth and morphological properties of

human breast epithelial cells cultured in an extracellular matrix (Matrigel).

(8) REPORTABLE OUTCOMESManuscripts

Mazoyer S, Gayther SA, Nagai MA, Smith SA, Dunning A, van Rensburg EJ, Albertsen H,White R, & Ponder BA. A gene (DLG2) located at 17q12-q21 encodes a new homologue ofthe Drosophila tumor suppressor dIg-A. Genomics. 1995 Jul 1;28(1):25-31.

Smith SA, Holik PR, Stevens J, Melis R, White R, & Albertsen H. Isolation and mapping of agene encoding a novel human ADP-ribosylation factor on chromosome 17q12-q21.Genomics. 1995 Jul 1;28(1):113-5.

Brothman AR, Steele MR, Williams BJ, Jones E, Odelberg S, Albertsen HIM, Jorde LB, RohrLR, & Stephenson RA. Loss of chromosome 17 loci in prostate cancer detected bypolymerase chain reaction quantitation of allelic markers. Genes Chromosomes Cancer.1995 Aug;13(4):278-84.

Smith SA, Holik P, Stevens J, Mazoyer S, Melis R, Williams B, White R, & Albertsen H.Isolation of a gene (DLG3) encoding a second member of the discs-large family onchromosome 17q12-q21. Genomics. 1996 Jan 15;31(2):145-50.

Albertsen HM, Smith SA, Melis R, Williams B, Holik P, Stevens J, & White R. Sequence,genomic structure, and chromosomal assignment of human DOC-2. Genomics. 1996 Apr15;33(2):207-13.

Karpf AR, Peterson PW, Rawlins JT, Dalley BK, Yang Q, Albertsen H, Jones DA. Inhibition ofDNA methyltransferase stimulates the expression of signal transducer and activator oftranscription 1, 2, and 3 genes in colon tumor cells. Proc Natl Acad Sci U S A. 1999 Nov23;96(24): 14007-12.

Spancake KM, Anderson CB, Weaver VM, Matsunami N, Bissell MJ, & White RL. E7-transduced human breast epithelial cells show partial differentiation in three-dimensionalculture. Cancer Res. 1999 Dec 15;59(24):6042-5.

(9) CONCLUSIONSDuring the funded period, the overall goal of our project is to identify genes involved with

the development and progression of breast cancer. This goal has remained unchanged since the startof the project, but the discovery of BRCA1 in 1994 together with technological advances in geneexpression profiling has influenced our strategy to achieve this goal. In the early part of the projectour search for tumor suppressor genes was directed by genetic or LOH mapping strategies followedby positional cloning of candidate genes. As we proposed in our revised statement of work (SOW),we have focused our efforts entirely on the development of microarray-based comparisons toidentify breast cancer related genes and further characterize the change in morphology and growth

12

of human mammary epithelial cells conditionally expressing the breast cancer related genes usingthe Matrigel culture system.

(10) REFERENCESH. Aberle, C. Bierkamp, D. Torchard, 0. Serova, T. Wagner, E. Natt, J. Wirsching, C. Heidkamper,

M. Montagna, H.T. Lynch, G.M. Lenoir, G. Scherer, J. Feunteun and R. Kemler. "The humanplakoglobin gene localizes to chromosome 17q21 and is subject to loss of heterozygosity in breastand ovarian cancers". Proc. Nati. Acad. Sci. USA 92: 6384-6388.

H.M. Albertsen et al. "Genetic mapping of the BRCA1 region on chromosome 17q21." Am. J. Hum.Genet. 54:516-525 (1994a).

H.M. Albertsen, S.A. Smith, S. Mazoyer, E. Fujimoto, J. Stevens, B. Williams, P. Rodriguez, C.S.Cropp, P. Slijepcevic, M. Carlson, M. Robertson, P. Bradley, E. Lawrence, Z.M.Sheng, R.Hoopes, N. Sternberg, A. Brothman, R. Callahan, B.A.J. Ponder and R. White. "A physical mapand candidate genes in the BRCA1 region". Nature Genetics 7: 472-479 (1994b).

H.M Albertsen, S.A. Smith, R. Melis, B. Williams, P. Holik, J. Stevens, and R. White. "Sequence,genomic structure, and chromosomal assignment of human DOC-2." Genomics 33:207-213(1996).

A.R. Brothman, M.R. Steele, B.J. Williams, E. Jones, S. Odelberg, H.M. Albertsen, L.B. Jorde, L.R.Rohr and R. Stephenson. "Loss of chromosome 17 loci in prostate cancer detected by polymerasechain reaction quantitation of allelic markers". Genes Chromosomes and Cancer 13: 278-284(1995).

M.A. Brown, K.A. Jones, H. Nicolai, M. Bonjardim, D. Black, R. McFarlane, P. de Jong, J. P. Quirk,H. Lehrach and E. Solomon. "Physical mapping, cloning, and identification of genes within a 500-kb region containing BRCA1". Proc. Nati. Acad. Sci. USA 92: 4362-4366 (1995).

L.H. Castilla, F.J. Couch, M.R. Erdos, K.F. Hoskins, K. Calzone, J.E. Garber, J. Boyd, M.B. Lubin,M.L. Deshano, L.C. Brody, et al "Mutations in the BRCA1 gene in families with early-onsetbreast and ovarian cancer." Nature Genetics 8: 387-91 (1994).

R. Cawthon R et al. A major segment of the neurofibromatosis type 1 gene: cDNA sequence,genomic structure, and point mutations. Cell 62:193-201 (1990).

I.M. Chumakov and 61 co-authors. "A YAC contig map of the human genome". Nature 377: 175-297(1955) (Supplement)

M. Coss, D. Winterstein, Rn. Sowder, S. Simek. Molecular cloning, DNA sequence analysis, andbiochemical characterization of a novel 65-kDa FK506-binding protein (FKBP65). J Biol Chem270:29336-41 (1995).

C.S. Cropp, Z.-M. Sheng, H.A. Nevanlinna, H. Albertsen, R. White and R. Callahan "Genomiclocation and structure of plakoglobin". -in preparation.

C.S. Cropp, H.A. Nevanlinna, S. Pyrhi5nen, U.-H. Stenman, P. Salmikangas, H. Albertsen, R. Whiteand R. Callahan. "Evidence for involvement of BRCA1 in sporadic breast carcinomas" CancerResearch 54: 2548-2551 (1994).L.S. Friedman, E.A. Ostermeyer, C.I. Szabo, P. Dowd, E.D. Lynch, S.E. Rowell and M.C. King.

"Confirmation of BRCA1 by analysis of germline mutations linked to breast and ovarian cancerin ten families". Nature Genetics 8: 399-404 (1994).

P.A. Futreal, Q. Liu, D. Shattuck-Eidens, C. Cochran, K. Harshman et al. "BRCA1 mutations inprimary breast and ovarian carcinomas". Science 266: 120-122 (1994).

13

S.C. Gerken, H. Albertsen, T. Eisner, L. Ballard, P. Holik, E. Lawrence, M. Moore, X. Zhao and R.White. "A strategy for constructing high-resolution genetic maps of the human genome: A genetic

map of choromosome 17p ordered with meiotic breakpoint mapping panels." American Journal of

Human Genetics 56: 484-499 (1995).J. Groden et al. Identification and characterization of the familial adenomatous polyposis coli gene.

Cell 66:589-600 (1991).M. Hoshimaru, J. Ray, D. Sah, F. Gage. Differentiation of the immortalized adult neuronal

progenitor cell line HC2S2 into neurons by regulatable suppression of the v-myc oncogene.

Proc. Natl. Acad. Sci. USA 93:1518-23 (1996).G. Joslyn et al. Identification of deletion mutations and three new genes at the familial polyposis

locus. Cell 66:601-13 (1991).A.R. Karpf, P.W. Peterson, J.T. Rawlins, B.K. Dalley, Q. Yang, H. Albertsen, and D.A. Jones.

Inhibition of DNA methyltransferase stimulates the expression of signal transducer and activator

of transcription 1, 2, and 3 genes in colon tumor cells. Proc. Natl. Acad. Sci. USA 96:14007-

14012 (1999).C. Kirkpatrick and M. Peifer. "Not just glue: cell-cell junctions as cellular signaling centers".

Current Opinion in Genetics and Development 5: 56-65 (1995).S.M. Mazoyer, S.A. Gayther, M.A. Nagai, S.A. Smith, A. Dunning, E.J. van Rensburg, H. Albertsen,

R. White and B.A.J. Ponder. "A gene located at 17q12-21 encodes a new homologue of theDrosophila tumor-suppressor dlg-A protein" Genomics 28: 25-31 (1995).

Y. Miki, J. Swensen, D. Shattuck-Eidens, P.A. Futreal, K. Harshman, S. Tavtigian, Q. Liu, C.

Cochran, L.M. Bennett, W. Ding, et al. "A strong candidate for the breast and ovarian cancer

susceptibility gene BRCA1". Science 266: 66-71 (1994).J.C. Murray, K.H. Buetow, J.L. Weber, S. Ludwigsen, T. Scherpbier-Heddema, F. Manion, J. Quillen,

V.C. Sheffield, S. Sunden, G.M. Duyk, J. Weissenbach, G Gyapay, C. Dib, J. Morrissette, G.M.Lathrop, A. Vignal, R. White, N. Matsunami, S. Gerken, R. Melis, H. Albertsen, K. Ward, R.Plaetke, S. Odelberg, D. Ward, J. Dausset, D. Cohen, H. Cann. "A comprehensive human linkage

map with centimorgan density". Science 265: 2049-2054 (1994).J. Simard, P. Tonin, F. Durocher, K. Morgan, J. Rommens, S. Gingras, C. Samson, J.F. Leblanc, C.

Belanger, F. Dion et-al "Common origins of BRCA1 mutations in Canadian breast and ovarian

cancer families". Nature Genetics 8: 392-398 (1994).S.A. Smith, P.R. Holik, J. Stevens, R. Melis, R. White and H. Albertsen. "Isolation and mapping of a

gene encoding a novel human ADP-ribosylation factor on chromosome 17q12-q21." Genomics 28:113-115 (1995).

S.A. Smith, P.R. Holik, J. Stevens, S. Mazoyer, R. Melis, B. Williams, R. White and H. Albertsen."Isolation of a gene (DLG3) encoding a second member of the disc-large family on chromosome17q12-21." Genomics 31:145-150 (1996).

C.L. Sommers, E.P. Gelman, R. Kemler, P.Cowin, and S.W. Byers "Aletrations in -CateninPhosphorylation and Plakoglobin Expression in Human Breast Cancer Cells". Cancer Res. 54:3544-3552 (1994).

K.M. Spancake, C.B. Anderson, V.M. Weaver, N. Matsunami, M.J.Bissell, and R.L.White. E7-

transduced human breast epithelial cells show partial differentiation in three-dimensionalculture. Cancer Res. 59:6042-6045 (1999).

L.K. Su, B. Vogelstein, K.W. Kinzler. "The APC tumor suppressor protein associates with catenins".

Science 262: 1734-1737 (1993).

14

B.J. Williams, E. Jones, X.J. Zhu, M.R. Steele, R.A. Stephenson, L.R. Rohr and A.R. Brothman."Evidence for a tumor suppressor gene distal to BRCA1 prostate cancer." Urology (submitted)(1995).

THE UTAH MARKER DEVELOPMENT GRouP: D. Adamson, H. Albertsen, L. Ballard, P. Bradley, M.Carlson, P. Cartwright, T. Elsner, D. Fuhrman, S. Gerken, L. Harris, P. R Holik, A. Kimball, J.Knell, E. Lawrence, J. Lu, A. Marks, N. Matsunami, R. Melis, B. Milner, M. Moore, L. Nelson,S. Odelberg, G. Peters, R. Plaetke, R. Riley, M. Robertson, R. Sargent, G. Staker, A. Tingey, K.Ward, X. Zhao and R. White. "A collection of ordered tetra-nucleotide repeat markers from thehuman genome." American Journal of Human Genetics 57: 619-628 (1995).

D.F. Woods and P.J. Bryant. "Molecular cloning of the lethal (1) discs-large (1) oncogene ofDrosophila". Dev. Biol. 134: 222-235.

V. Weaver, A, Fischer, 0. Peterson, and M. Bissell. The importance of the microenvironment inbreast cancer progression: recapitulation of mammary tumorigenesis using a unique humanmammary epithelial cell model and a three-dimensional culture assay. Biochem Cell Biol74:833-51 (1996).

V, Weaver, 0. Petersen, F, Wang, C. Larabell, P. Briand, C. Damsky, and M. Bissell. Reversion ofthe malignant phenotype of human breast cells in three-dimensional culture and in vivo byintegrin blocking antibodies. J Cell Biol 137:231-45 (1997).

D. Viskochil et al. Deletions and a translocation interrupt a cloned gene at the neurofibromatosistype 1 locus. Cell 62:187-92 (1990).

(11) APPENDICESMaterials and methods

Cell Culture and retroviral infection.Surgical discard material from reduction mammoplasties was minced with opposing

scalpels, placed in digestion buffer, and incubated in spinner flasks at 37'C until stroma dissolved(approximately three to five hours). Digestion buffer contained 1 unit/ml Collagenase D (RocheMolecular Biochemicals, Indianapolis, IN), 2.4 units/ml dispase (Roche Molecular Biochemicals),and 6.25 units/ml DNase (Sigma, St. Louis, MO) in Dulbecco's phosphate buffered saline. Thedigested material plus 10% fetal calf serum (FCS) was centrifuged at 800 rpm for 10 min. Theresulting pellet was resuspended in wash buffer and the organoids were separated from stromal andblood cells by sequential sedimentation at lx g. Organoids were cultured immediately or frozen inDulbecco's Modified Eagle Medium: Nutrient Mixture F-12 (Ham) 1:1 (GibcoBRL, Gaithersburg,MD), 10% FCS, and 10% dimethyl sulphoxide. Organoids were cultured on plastic in CDM3culture media for primary breast epithelial cell outgrowth and subculture. Primary epithelialoutgrowth was infected with an LXSN retroviral construct containing the human papilloma virustype 16 E7 gene (LXSN16E7). HMEC were incubated with LXSN16E7 in CDM3 plus 4 Ag/mlpolybrene (Sigma) for 24 h. Viral supernatant was aspirated and HMIEC were cultured in virus-freeCDM3 for 48 h prior to selection in CDM3 containing 50 jig/ml Geneticin (Gibco BRL). Earlypassage HMEC (either passage one or two) and HMEC containing LXSN16E7 (HMEC+E7) werecultured in a rBM, Matrigel (Becton Dickinson, Bedford, MA), as previously described. Briefly, 2.5x 105 HMEC were resuspended as single cells in 300 R1 of 10 mg/ml Matrigel per well and platedonto Nunc 4-well multidishes coated with 100 g1 Matrigel. Matrigel cultures were overlaid with 500[tl CDM3.

15

Table 1.Image ID Clone definition

1 358433 Human retinoid X receptor-gamma mRNA, complete cds2 295401 ESTs3 1088345 S100 calcium-binding protein A7 (psoriasin 1)4 276282 ESTs5 269017 Human 0-linked GIcNAc transferase mRNA, complete cds6 545239 Neutrophil Gelatinase-Associated Lipocalin Precursor7 882141 DNA G/T mismatch-binding protein8 283063 MHC class II DQ-beta associated with DR2, DOwi protein9 156431 Ciliary neurotrophic factor receptor10 770435 Transcription factor p6511 70046612 277134 ESTs13 275272 ESTs14 273039 ESTs15 283618 ESTs16 23240 SM22-alpha homolog17 627104 Human alpha-tubulin mRNA, complete cds18 33934 ESTs19 610187 Proteasome Component 0920 592947 Autoantigen PM-SOL21 46743 ESTs, Highly similar to RAS-related protein RAP-1 B22 23904 ESTs23 264369 ESTs, Highly similar to GLUCOSYLTRANSFERASE ALG824 200531 ESTs25 126783 ESTs26 201891 ESTs27 265684 ESTs28 261971 Metallopeptidase 1 (33 kD)29 129503 ESTs

The genes listed in this table showed 3x fold or more variation in expression levels in a comparisonbetween primary human mammary cells, with and without induced P3-catenin.

Figure 1.This figure shows a Northern Blot of four Vector 131-5 131-10 15"-3 151-9different cell lines transfected with the + + ÷ " - +

LINX P3-catenin construct and the vectorcontrol. In absence of Doxycilin (indicated <by -) O3-catenin is over-expressed resultingin a three fold up-regulation Psoriasin.

16

Figure 2.a. Conditionally inducible vectors. -C LINX LINXTo test the efficacy of the tetracycline regulatedE dN131 dN151retroviral vector system, dominant-negative b- -U - ---0catenin genes were cloned into pRetro Off, .. + - + DOXpRetro On, and LINX. Following packaging,transfection into primary human mammary pRetro ON -ww b t o.-- endogeneous

epithelial cells and two weeks of selection, totalcell extracts were subjected to Western analysis pRetro OFF .. ,.o - -*-- endogeneoususing anti-C-terminal j3-catenin antibody. As -- exogeneousshown here, cells infected with pRetro ONconstruct did not induce dominant 13-catenin at 6W" 4- exogeneousall. Cells infected with pRetro OFF induceddominant 03-catenin in response to DOXremoval, however the amount induced (shown as exogeneous) were much less than that of endogeneousprotein. On the other hand, cells infected with LINX vector construct produced great response in bothinducibility and amount.

b. LEFI-Luciferase Reporter Assay.Shown here are relative LEFl-reporter activities indicating thatinduced dominant f3-catenin is in fact functional in regard to <L..

activating genes containing LEFM responsive element. - .....

0

CL.

a------------------------ --

U_

-ar-

(J

0

4)

Control dNl3l

LI + DOX (uninduced)-DOX (induced)

17

GENES, CHROMOSOMES & CANCER 13:278-284 (1995)

Loss of Chromosome 17 Loci in Prostate Cancer Detectedby Polymerase Chain Reaction Quantitation ofAllelic Markers

Arthur R. Brothman, Michael R. Steele, Briana J. Williams, Emma Jones, Shannon Odelberg, Hans M. Albertsen,Lynn B. Jorde, L Ralph Rohr, and Robert A. Stephenson

Departments of Human Genetics (A.R.B., S.O., H.M.A., L.B.J.), Pediatrics (A.R.B., MR.S., B.J.W., E.J.), Pathology (L.R.R.), and Urology(R.A.S.), University of Utah School of Medicine, Salt Lake City, Utah

Using a polymerase chain reaction/microsatellite marker system, we demonstrated that 6 of 22 (27%) clinical stage B (early)primary prostate tumors showed loss of heterozygosity at one or more of five loci on chromosome 17. The sensitivity of thisstudy was increased by use of a Phosphorlmager and statistical analysis of replicate tumor-normal DNA pairs. Two patientsshowed tumor-specific interstitial loss at a locus in close proximity to the familial breast cancer gene BRCAI. These findingssuggest that genes on the proximal long arm of chromosome 17 play a pivotal role in the early development of at least a subsetof prostatic tumors. Genes Chromosom Cancer 13:278-284 (1995). © 1995 Wiley-Liss, Inc.

INTRODUCTION tions in the TP53 gene have also been observed in

Prostate cancer is the most common cancer in a rare subset of advanced prostate tumors (Effert et

males in the United States, with an estimated al., 1992; Bookstein et al., 1993). Recently, Brew-

200,000 new cases in 1994 (Boring et al., 1994), yet ster and colleagues (1994) have observed loss at the

the etiology of this disease is still poorly under- nm23-H1 locus on 17q in one late-stage prostate

stood. Genetic and cytogenetic changes have been tumor,; however, no reports have indicated loss of

associated with the disease, but no consistent cel- regions on the long arm of chromosome 17 in early-lular abnormality has been observed. The recent stage tumors.use of fluorescence in situ hybridization (FISH) The susceptibility gene for familial breast and

techniques with chromosome-specific probes has ovarian cancer, BRCA1, is located on the proximal

suggested that chromosome 17 (i.e., the whole long arm of chromosome 17 (Miki et al., 1994) and

chromosome) is frequently lost in prostate tumors has been theorized to be involved in prostate can-

(Brothman et al., 1992, 1994; Jones et al., 1994); cer (Arason et al. 1993; Ford et al., 1994). There-hence, we sought to evaluate primary tumor tissue fore, we chose to study markers near BRCA1 in ourfor molecular abnormalities of this chromosome. initial screen for loss of heterozygosity (LOH) on

Several molecular studies have been published chromosome 17, to our knowledge, previous stud-in which certain chromosomal regions are lost in ies included only markers on distal 17q.prostatic tumors, suggesting that loss of a tumor The considerable cellular heterogeneity within

suppressor function may be involved in this dis- the prostate gland creates difficulties in allelotyp-ease. Loss of loci on the short arm of chromosome ing of these tumors. Some investigators have used

8 is, to date, the most frequently observed loss in microdissection of apparent tumor tissue for DNA

prostate tumors (Bergerheim et al., 1991; Kunimi preparation (Bova et al., 1993; Macoska et al.,

et al., 1991; Bova et al., 1993; Macoska et al., 1993; MacGrogan et al., 1994), whereas others

1993). A putative tumor suppressor gene in this have limited their analysis to gross pathologic char-

region has been associated with other solid tumors, acterization prior to DNA extraction (Bergerheimincluding hepatocellular, colorectal, and lung can- et al., 1991). Even with the most careful microdis-

cers (Emi et al., 1992, 1993). Chromosomal loss sections, however, some proportion ofcontaminat-has also been observed in prostate cancer at sites ing normal epithelial cells or of fibroblastic stromal

including 5q, 10p, 10q, 16q, 17p, and 18q (Carter cells remains. We have approached the problem of

et al., 1990; Bergerheim et al., 1991; Kunimi et al., Received November 29, 1994; accepted March 31, 1995.

1991; Brewster et al., 1994; Latil et al., 1994).Macoska and colleagues (1992) have shown that Address reprint requests to Arthur R. Brothman, PhD, Depart-there is loss of short-arm material from chromo- ments of Pediatrics and Human Genetics, 1C204 University ofUtah

Medical Center, University of Utah School of Medicine, Salt Lake

some 17 in metastatic prostate tumors, and muta- City, UT 84132, U.S.A.

© 1995 Wiley-Uss, Inc.

LOSS OF CHROMOSOME 17 LOCI IN PROSTATE CANCER 279

cellular heterogeneity in our specimens by relying Development and Mapping Group: UT7on the gross histopathologic evaluation of adjacent (D17S752), UT573 (Dl7S902), and UT752tumor specimens and by analyzing multiple repli- (D17S907) located on the long arm of chromosomecations of polymerase chain reaction (PCR) prod- 17 (Utah et al., in press), and UT751 (D17S906)ucts with a Molecular Dynamics Phosphorlmager. and UT5265 (D17S 1149) located on the short armUsing this technique, we have observed tumor- of chromosome 17 (Albertsen et al., 1994). In ad-specific losses of regions on chromosome 17 in 6 of dition, a polymorphic marker for the short arm of22 (27%) of the patients evaluated. Furthermore, chromosome 8 at the LPL locus (Zuliani andwe have confirmed these results by using a novel Hobbs, 1990) was analyzed for each tumor-normalFISH approach for deletion detection with se- DNA pair for information on an independent chro-lected single-copy P1 clones serving as probes mosome and as a test for efficiency of our PCR(Williams et al., 1995). technique in comparison to frequencies of loss ob-

served at this locus by others.MATERIALS AND METHODS

Specimens PCROne member of a specific primer set was labeled

Twenty-two primary prostate tumor specimens at the 5' end with [3 2P]gamma ATP (New Englandobtained from radical prostatectomies at the Uni- Nuclear-Dupont) by use of polynucleotide kinaseversity of Utah School of Medicine were preparedfor DNA and single-cell preparations. A central (Boehringer Mannheim). The PCR was optimizedsecton f a of ndulrit wasobtine at and performed with a Techne PHC-3 thermal cv-suetioy an aregion ofcnt secti was evained at cler with a mixture of each primer set (1:100 radio-surgery, and an adjacent section was evaluated his-actively labeled to unlabeled primer), dNTPstopathologically as described previously (Jones et (Pharmacia), 0.001 U/a.d Taq polymerase (Boeh-

al., 1994). All specimens were from patients who ringer Mannheim), 2 mM spermidine, and PCRhad early (clinical stage B) cancer at evaluation, buffer ( i0 mM Tris-H2 I, pH 8.8 , 40 M C I

Tumor specimens were coded and separated for . Cl, pH 8.8, 40 mM NaCI

various other studies in this laboratory, containing'specific concentrations of Mg2+ rangingvrosohrongoing suisnthslbroy, from 1 to 2 maM).

and portions were frozen in liquid nitrogen for sub-

sequent DNA extraction. A peripheral blood sam- Analysis of PCR Productpie was obtained prior to surgery from each patient After amplification of the PCR product, the sam-for direct preparation of constitutional DNA and pies were run on an acrylamide denaturing gel con-for immortalization to a lymphoblastoid line. taining 7% acrylamide [38:1 acrylamide:bisacryla-

DNA Extraction mide (BioRad), 32% formamide (BRL), 5.6 M

Frozen tissue was ground with a micropestle, urea, and Tris-borate-EDTA buffer] and exposedincubated in a lysis buffer (10 mM Tris, pH 8.0, overnight at -80'C in an X-ray film holder with

100 mM NaCI,. 1 mM EDTA, 1% SDS, and 0.2 intensifying screen and Amersham Hyperfilm MP.

mg/mi proteinase K) at 37°C overnight, phenol- Images were both evaluated visually and analyzed

chloroform/isoamyl alcohol extracted, and ethanol with a Molecular Dynamics Phosphorlmager. After

precipitated prior to resuspension in 10 mM Tris/ exposure to X-ray film, the gel was placed against

0.1 mM EDTA buffer. DNA was extracted from a Phosphorlmage screen for 20 minutes to 16 hour,

peripheral blood that was first treated with Triton depending on signal intensity. A digital image was

X lysis buffer (0.32 M sucrose, 10 mM Tris-HCI, obtained and was then analyzed by the drawing of

pH 7.5, 5 mM MgCI,, I% Triton X-100), followed boundaries around each allelic band, subtraction

by treatment of pelleted nuclei with a buffer con- of local background, and integration of the volumetaining 1% SDS/0.075 Ni NaCI/0.024 M EDTA, of each band. The value of the top band was alwaystainng 1 SD/0.05 MNa~l0.04 M DTA taken as a percentage of the bottom band.pH 8.0, and 0.25 mg/ml proteinase K overnight at37°C. DNA was isolated the following day by phe- Statistical Analysisnol and chloroform/isoamyl alcohol extraction andethanol precipitation prior to resuspension in Tris- At least four replicates (separate aliquots of theEDTA buffer (Bell et al., 1981). same DNA isolates) of each tumor and the corre-

sponding normal sample were run in parallel on theMicrosatellite Primer Sets Phosphorlmager, and differences were then eval-

Primer sets for the following markers were gen- uated for significance by use of the nonparametricerated through the Utah Genome Center Marker Mann-Whitney U test. This is considered a robust

280 BROTHMAN ET AL.

TABLE I. Control Mixing Experimenta

95 90 85 80 75 70 65 60 55 50 45 40 35 30 25 20 15 10 5

100 + + + + + + + + + + + + + + + + + + +95 .-.. + + + + + + + + + + + + +90 . ..- + + + + + + + + + + + + +85 - - + + + + + + + + + + + + + +80 - - + + + + + + + + + + + + +75 . . . . . + + + + + + + + +

70 + + - + + + + + + + + - + +65 - - - + + + + + + + + +60 - - + + + + + + + + +55 - + + + + + + + + +

50 - + + + + + + + +45 + + + + + + + +40 - + + + + + +35 + - + + + +30 - + + + +25 + + + +20 + + +15 - +10 +

'Comparison of percent-mixed groups of normal (two-allele) DNA to increasing percentages of the tumor (one-allele) DNAs. Values represent percentnormal DNA: + and - signify difference or no difference, respectively, by the Mann-Whitney U test for eight replicates when Phosphorlmager data atthe percentages shown on the x and y coordinates of this table were compared.

test and is preferable to a parametric Student's t ogous to "tumor" DNA in the experiment). Thetest, because it does not assume a normal distribu- patient whom we chose for lymphoblastoid DNAtion. Four replicates were used for all initial eval- contained an allele at UT573 identical to that ofuations; if overlapping values were detected for cell line MH-22.6. These DNAs were combined toallele ratios between tumor and normal specimens, simulate normal DNA contamination of tumorthen additional replicates were evaluated applying DNA at 5% increments from 0 to 100%. UT573the Mann-Whitney U test to all accumulated val- primers were used for PCR amplification and sub-ues. We performed a power analysis to determine sequent Phosphorlmager analysis as describedthe probability of incorrectly rejecting the null hy- above; eight replicates (from separate PCR reac-pothesis of no difference (presence of LOH, type I tions) were tested for each mixture.error) and the probability of incorrectly acceptingthe null hypothesis of no difference (no LOH, typeII error). If the type I error is fixed at 0.05 (i.e., the RESULTSsignificance level for deciding that an allele hasbeen lost), then the Mann-Whitney U test yieldstype II error levels of 0.11 when eight replicates are The results of the simulated "tumor-normal"used and 0.35 when four replicates are used. cell contamination areshown in Table 1. Whereas

we were able to detect allelic imbalance in thisControlled experiment for LOH experiment with 95% contaminating normal DNA,

To ascertain the sensitivity of the Phosphorlm- the experiment in Table 1 was designed to repli-ager quantitation, we performed a controlled mix- cate other unforeseen variables such as differentialing experiment by using DNA from a lymphoblas- allelic intensities. This was done by extending ourtoid line from a patient who was informative for analyses to decreasing amounts of tumor DNAmarker UT573 (analogous to "normal" DNA) and compared to less than 100% "control" or normalDNA from cell line MH-22.6 (Coriell Institute for DNA. Although allelic imbalance was generally de-Medical Research, Camden, NJ). This human/ro- tected in high amounts of normal DNA, the lowestdent somatic cell hybrid line contains a single copy observed imbalance was seen at 60% contaminat-of human chromosome 17 and, thus, shows only ing normal (75% value, which did not detect allelicone allele when primers for UT573 are used (anal- loss until at least 45% of normal DNA was present;

LOSS OF CHROMOSOME 17 LOCI IN PROSTATE CANCER 281

T N T N T N T N T N T N T N

a b

T N T N T N T N T N T N T N T N

Fgure 1. Representative autoradio-graphs of replicates showing LOH in

`6• PCR products at different loci. The var-ious markers and corresponding speci-mens (UCAP number) are as follows, a:

C -LPL (UCAP 14). Arrows indicate allelicbands. b: UT751 (UCAP 18). c: UT752

d (UCAP 24). d: UT573 (UCAP 24).

TABLE 2. Summary of Informativeness and LOH Data for imbalance study. If there was any overlap in valuesMarkers Used where ratio differences were suspected (seen for

UCAP 24 with the marker UT752 in Table 3), thenwe ran additional PCR amplifications and gels toconfirm or refute our initial observations.

LPL 8p22 13 of 22 (59) 5 of 13 (39)UT7 17 q21.3 14 of 22 (64) I of 14 (7) A summary ofthesedata is shown in Table 4 andUT573 17q21.2 16 of 22 (73) 4 of 16 (25) is depicted in the partial map of chromosome 17 inUT752 17q1 1.2-12 18 of 22 (82) 4 of 18 (22) Figure 2. Values in Table 4 represent mean allelicUT751 17 pt2-13 19 of 22 (86) 3 of 19 (16) ratios determined by Phosphorlmager analysis.UT5265 17p12-13 15 of 22 (68) 4 of 15 (27) Hence, in general, the closer these values are to

100%, the more likely it is that tumor and normalvalues were identical. It is important to remember,however, that the statistic used (Mann-Whitney U

45%/75% = 0.6). Overall, these results confirm test) also takes into account the possibility of over-our prediction that we can detect loss in a highly lap in values (see the example discussed above).

contaminated background. The relatively conser- Because we have chosen a significance level ofP P-vative Mann-Whitney U test criteria were used for 0.05, we are 95% confident that LOH detectedthe following analysis of prostate tumors. represents true loss. Values that appear to deviate

from 100% and were not significant (e.g., UCAP 5LOH at marker UT5265) represent mean values, but a

Tumor-specific allelic losses on chromosome 17 wider range of PhosphorImager data deemed thesewere observed in 6 of 22 prostate cancer specimens loci insignificant.(27%) with the markers used in this study. A sum- It can be seen in Figure 2 and Table 4 that twomarv of informativeness (frequency of heterozygos- patients showed interstitial loss on 17q (UCAP 24ity) of the markers used and the LOH observed is and UCAP 30), one patient showed an apparent lossshown in Table 2. Representative autoradiographs of the entire chromosome 17 (UCAP 18), one pa-of replicates showing varying degrees of loss at par- tient showed only short-arm loss (UCAP 17), andticular loci for individual patients are presented in one patient showed loss at both distal ends of chro-Figure 1, with corresponding Phosphorlmager data mosome 17 (UCAP 29). Loss of UT573 was alsoin Table 3. The numerical data from the Phosphor- seen in UCAP 28, whereas the most centromericImager analysis were chosen because they provide long-arm marker was retained. Because the distalan objective measurement for allelic loss and an marker, UT7, was uninformative for that specimen,increased sensitivity over visual detection, which we have not considered this to be an interstitial loss.can be seen when data from Figure 1 and Table 3 For a positive control, primers at an indepcndentare compared. When all data were evaluated, loss chromosome 8 locus were tested in a similar man-of the large (upper band) or small (lower band) ner. Loss at the LPL locus on chromosome 8 wasallele was seen, which is expected in any allelic observed in five cases (39% of informative cases)

282 BROTHMAN ET AL.

TABLE 3. Representative Phosphorlmager Allele Ratios for CorrespondingPhotographs in Figure Ia

Locus (Specimen) Allele ratio, tumor Allele ratio, normal

LPL (UCAP 14) 27.9/25.4/32.2 67.8/70.1/71.4UT751 (UCAP 18) 30.1/31.6/34.0/37.1 51.0/54.4/60.8/50.3UT752 (UCAP 24) 91.2/105.8/106.3/82.4 103/I 14.2/104.9/I 12 .8 b

UTS73 (UCAP 24) 47.3/51.2/54.2/53.8 92.9/101.9/100.6/96.3

'Corresponding numbers in each series represent adjacent lanes in Figure I.bBecause of the overlapping values in this series, additional replicates were evaluated, and the

statistical power was thus increased (final analyses for UCAP 24, see Table 4).

TABLE 4. Primary Prostate Tumor/Normal Phosphorlmager Data and Significant Loss at FiveChromosome 17 (UT) Markers and One Chromosome 8 (LPL) Marker

Microsatellite markers

Specimen UT7 UT573 UT752 UT751 UT5265 LPL

UCAP 2 N N 96 84 N 99UCAP 3 94 N 99 N 88 NUCAP 4 89 N 93 N 99 NUCAP 5 94 98 N 99 60 NUCAP 6 97 93 92 93 66 96UCAP 14 89 92 99 94 96 47a

UCAP 16 N 99 85 79 N NUCAP 17 N 94 N .43a 57a 573UCAP 18 N 873 75a 61a 79a N

UCAP 19 99 98 98 99 99 NUCAP 20 91 95 93 97 86 NUCAP 21 88 99 99 99 83 94UCAP23 97 98 88 67 N 99UCAP 24 96 54a 82r 98 N 52a

UCAP 25 89 95 98 95 N NUCAP 26 N 92 93 96 93 94UCAP 27 N N N 83 93 76UCAP 28 N 54a 96 49a 42a 50aUCAP 29 903 N 96 98 69' NUCAP 30 95 83a N 96 94 99UCAP 31 100 94 92 99 N 95UCAP 33 N N 90 N N 631

"LOH. Numbers represent percentages of the normal vs. tumor allele ratios for the mean values for eachinformative set of replicates [(allelic tumor ratios/allelic normal ratios) x 100]. Significance values (P !S 0.05)show that tumor allele ratios differ from normal allele ratios using the Mann-Whitney U test. In general, thecloser a number is to 100%, the closer all replicate Phosphorlmager values were between tumor and normalpairs. N, noninformative.

including UCAP 18, UCAP 24, and UCAP 28, ated with prostate carcinogenesis. Oshimura andwhich showed loss of loci on chromosome 17 as well. Sandberg (1975) noted isochromosomes for 17q in

a metastatic prostate tumor as early as 1975. In ourDISCUSSION laboratory, FISH analysis has demonstrated whole

The results of this study suggest that deletions chromosome 17 loss in both cultured and archivalor LOH on the long arm of chromosome 17 may be prostate cancer specimens (Brothman et al., 1992,involved in some early-stage prostate tumors. Al- 1994; Jones et al., 1994).though allelic loss on 17q has not been reported There have been observations of an increasedpreviously in early-stage prostate tumors, previous incidence of prostate cancer in families that showcytogenetic and molecular cytogenetic data indi- linkage for breast cancer. Arason and colleaguescated that changes in chromosome 17 were associ- (1993) proposed that breast cancer genes may pre-

LOSS OF CHROMOSOME 17 LOCI IN PROSTATE CANCER 283

17 18 24 28 29 30 centromere show deletions. Loss of one entire

1 - Tu5265 4 1 • chromosome in this sample was also observed byr1,12 FISH studies with pericentromeric probes (Jones

111uT751 4"1.2 et al., 1994) and a combination of pericentromeric

2- u and PI probes (Williams et al., submitted). Whole-21.2,,

22 chromosome loss is an alternative mechanism that23 uT573 would reveal the presence of a tumor suppressor24 . T7

25T _,7 locus. UCAP 28 and UCAP 29 both yielded un-17 usual results, in that there appears to be interstitial

retention of chromosome 17 material (see Fig. 2).Figure 2. Schematic showing the location of the polymorphic mark- Te sen t wo chses osug e th at ea co p e r rg e-

ers (to the right of the ideogram) and regions lost in the six prostate These two cases suggest that a complex rearrange-tumors (UCAP 17, 18, 24, 28, 29, and 30). Solid circles represent allelic ment including the region near 17q21 has takenloss, open circles represent allelic retention, and triangles indicate that amarker was uninformative (homozygous at the locus studied). place; alternatively, two or more deletional events

may have occurred.dispose individuals to prostate cancer, based on a The use of the Phosphorlmager, in addition tostudy of seven Icelandic families and association of the evaluation of replicate specimens, has proved17q12-q23-linked markers. Furthermore, carriers invaluable in overcoming the problem of contami-of mutations in a gene predisposing to breast and nating normal or stromal cells in our LOH study ofovarian cancer (BRCAI) have recently been shown prostate cancer. Microdissection of tumor tissue isto have an increased risk of colon and prostate can- a feasible alternative to multiple replicate analyses;cer (Ford et al., 1994). Markers that we chose for yet, nontumor cells may persist in considerablethis initial study included those that mapped near amounts and cannot be eradicated completely.the BRCA1 locus, because we sought to determine Bookstein and coworkers (1993) have very recentlywhether this locus was deleted in prostate cancers. taken a similar approach to the analysis of prostateIn light of the recent isolation of the BRCA1 gene tumor specimens with Phosphorlmager-analyzed(Miki et al., 1994), we have now determined that data of marke'rs on chromosome 8 but have exam-the most frequently lost marker, UT573, is within ined a single measurement of each tumor/normal1 Mb of BRCA1. Clearly, additional markers, in- DNA pair for each specimen (Cher et al., 1994;cluding any that span BRCA1, must be evaluated. MacGrogan et al., 1994). An advantage of limiting

Our findings suggest that a locus near BRCA1 may analyses to a single pair of DNAs is that many morebe involved in early prostate cancer progression. markers can be readily evaluated, but this may beParticularly interesting are specimens UCAP 24 and at the expense of sensitivity. The use of replicateUCAP 30, which show interstitial deletions on the samples increases the statistical power for evalua-long arm of chromosome 17 (Fig. 2) in a region in tion and helps eliminate artifacts. The frequencyclose proximity to BRCA1 (Miki et al., 1994; of loss we observed at LPL (39% of informativeWilliams et al., submitted). UCAP 28 also shows cases) is comparable to that detected by MacGro-specific loss on 17q at UT573, but we have not yet gan et al. (1994), where 43% of informative casesdetermined whether this is interstitial, because the showed LOH when a combination of three LPLdistal long-arm marker (UT7) was uninformative for polymorphic sites was used . It should also bethis specimen. This specimen also showed loss of noted that, although a limited number of markersboth short-arm markers. Likewise, both UCAP 17 has been evaluated at other chromosomal sites, weand UCAP 29 showed specific loss of chromosome have not yet detected a high frequency of LOH on17 short-arm material. Therefore, we suggest that chromosomes other than 8 or 17 in prostate tumorsother regions on chromosome 17, including portions using these methods.of the short arm, may also be involved in prostate Our method appears highly reliable and sensi-cancer. Previously, mutations in the TP53 gene on tive, having the ability to detect LOH in tumor17p have been observed in advanced-stage prostate cells with a relatively high background of normaltumors (Macoska et al., 1992; Latil et al., 1994). cells. The mixing experiment (Table 1), in whichWhether there are abnormalities in TP53 or there is normal DNA from one patient was added to a by-involvement of a different locus on 17p in our spec- brid cell line containing one copy of chromosomeimens remains to be seen. 17 with an allele identical to that seen in the nor-

One specimen in the current study, UCAP 18, mal control, presents information about both theappears to have lost an entire chromosome 17, be- sensitivity and the limitations of the multiple-rep-cause all informative markers on both sides of the licate PCR approach. Our data suggest that allelic

284 BROTHMAN ET AL.

losses are detectable well above the traditionally (1993) p53 is mutated in a subset of advanced-stage prostate can-cers. Cancer Res 53:3369-3373.accepted 50% level of contaminating normal cells Boring CC, Squires TS, Tong T, Montgomery S (1994) Cancer

(Table 1). Our analysis shows that this system has statistics. CA J Clin 44:7-26.Bova GS, Carter BS, Bussemakers MJG, Emi NM, Fujiwara Y,high sensitivity for LOH detection with only a Natasha K, Jacobs SC, Robinson JC, Epstein JI, Walsh PC, Isaacs

moderate probability of missing LOH where it ac- WB (1993) Homozygous deletion and frequent allelic loss ofchro-mosome 8p22 loci in human prostate cancer. Cancer Res 53:

tually may be present. Thus, our detection rate of 3869-3873.

27% of primary tumors with loss of chromosome 17 Brewster SF, Browne S, Brown KW (1994) Somatic allelic loss at theDCC, APC, nm23-HI and p53 tumor suppressor gene loci in

sequences is conservative, human prostatic carcinoma. J Urol 151:1073-1077.

Indeed, our parallel studies, in which we used Brothman AR, Patel AM, Pcehl DXI, Schellhammer PF (1992)Analysis of prostatic tumor cultures using fluorescence in situFISH with selected bacteriophage P1 probes hybridization (FISH). Cancer Genet Cvtogenet 62:180-185.

flanking the regions on 17q that showed loss, have Brothman AR, Watson MJ, Zhu XL, Williams BJ, Rohr LR (1994)Evaluation of 20 archival prostate tumor specimens by fluores-

confirmed that true loss exists in each specimen in cence in situ hybridization (FISH). Cancer Genet Cytogenet 75:which we detected LOH by PCR (Williams et al., 40-44.

Carter BS, Weing CM, Ward WS, Treiger BF, Aalders TW,submitted). Hence, by using two independent Schalken JA, Epstein JI, Isaacs WB (1990) Allelic loss ofchromo-

methods, we have indicated that regional loss on somes 16q and 10q in human prostate cancer. Proc Natl Acad SciUSA 87:8751-8755.17q occurs in prostate tumors, supporting the view Cher ML, MacGrogan D, Bookstein R, Brown JA, Jenkins RB,

that a tumor suppressor gene is present here. Jensen RH (1994) Comparative genomic hybridization, allelic im-balance, and fluorescence in situ hybridization on chromosome 8The initial cellular processes that affect progres- in prostate cancer. Genes Chromosom Cancer 11:153-162.

sion of prostate cancer are proving to be complex Effert PJ, Neubauer A, Walther PJ, Liu ET (1992) Alterations ofthe p53 gene are associated with the progression of human pros-and may include multiple genetic events. Markers tate carcinoma. J Urol 147:789-793.

at additional chromosomal sites must be examined, Emi M, Yoshiyuki F, Nakajima T, Tsuchiya E, Tsuda H, HirohashiS, Maeda Y, Tsuruta K, Mivaki M, Nakamura Y (1992) Frequent

so that the relationship between various genetic loss of heterozygosity for loci on chromosome 8p in hepatocellular

events in the progression of prostate tumorigenesis carcinoma, colorectal cancer and lung cancer. Cancer Res 52:can e dterine. Th reatielvsmal se oftu- 5368-5372.

can be determined. The relatively small set of tu- Emi M, Yoshiyuki F, Ohata H, Tsuda H, Hirohashi S, Koike M,

mors examined in this study showed that approxi- Miyaki M, Mondne M, Nakamura Y (1993) Allelic loss at chro-mosome band 8p2l.3-p22 is associated with progression of hepa-

mately one-fourth of early prostate tumors have toceflular carcinoma. Genes Chromosom Cancer 7:152-157.

loss of at least some portion of chromosome 17, a Ford D, Easton DF, Bishop DT, Narod SA, Goldgar DE (1994)Risks of cancer in BRCAI mutation carriers. Lancet 343:692-695.

frequency that compares favorably to observed Jones E, Zhu XL, Rohr LR, Stephenson RA\, Brothman AR (1994)

losses on 10q and 16q (Carter et al., 1990). Our Aneusomy ofchromosomes 7 and 17 detected by FISH in prostatecancer and the effects of in vitro selection. Genes Chromosomfindings suggest that chromosome 17 is likely to Cancer 11:163-170.

play a significant role in this disease. Kunimi K, Bergerheim USR, Larsson IL. Ekman P, Collins VP(1991) Allelotyping of human prostatic adenocarcinoma. Genom-ics 11:530-535.ACKNOWLEDGMENTS Latil A, Baron JC, Cussenot 0, Fournier G, Soussi T, Bacron-

The authors thank Drs. Rosie Plaetke and Ray Gibod L, LeDuc A. Rovesse J, Lidereau R (1994) Genetic alter-ations in localized prostatic cancer: Identification of a common

White for their helpful discussions during the region of deletion on chromosome arm 18q. Genes ChromosomCancer 11:119-125.

course of these studies. We also appreciate the ef- MacGrogan D, Levy A, Bostwick D, Wagner M, Wells D, Book-forts of individuals within the Utah Genome Cen- stein R (1994) Loss of chromosome 8p loci in prostate cancer:

Mapping by quantitative allelic imbalance. Genes Chromosomter Marker Development and Mapping Group. Cancer 10: 151-159.This study was supported by NIH grants R01- Macoska JA, Powell IJ, Sakr W, Lane MA (1992) Loss of the 17p

chromosomal region in a metastatic carcinoma of the prostate. JCA46269, MO1-0RR00064, and NCI CCSG 5P30 Urol 147:1142-1146.

CA42014-08. M/lacoska JA, Micale MA, Sakr WA. Benson PD, Wolman SR (1993)Extensive genetic alterations in prostate cancer revealed by dual

REFERENCES PCR and FISH analysis. Genes Chromosom Cancer 8:88-97.Miki Y, Swensen J, Shattuck-Eldens D, Futreal PA, Harshman K,

Albertsen H, Plaetke R, Ballard L, Fujimoto E, Connolly J, Tavtigian S, Qingyun L, Cochran C, Bennett LM, Ding W, BellLawrence E, Rodriguez P, Robertson M, Bradley P, Miliner B, R, Rosenthal J, Hussey C, Tran T, McClure M, Frye C, HattierFuhrman D, Marks A, Sargent R, Cartwright P, Matsunami N, T, Phelps R, Haugen-Strano A, Katcher H, Yakumo K, GholamiWhite R (1994) Genetic mapping of the BRCAI region on chro- Z, Shaffer D, Stone S, Bayer S, Wray C, Bogden R, Dayananth P,mosome 1

7q

21. Am J Hum Genet 54:516-525. Ward J, Tonin P, Narod S, Bristow PK, Norris FH, Helvering H,

Arason A, Barkardottir RB, Egilsson V (1993) Linkage analysis of Morrison P, Rosteck P, Lai %I. Barrett JC, Lewis C, Neuhausenchromosome 1

7q markers and breast-ovarian cancer in Icelandic S, Cannon-Albright L, Goldgar D, Wiseman R, Kamb A, Skol-

families, and possible relationship to prostatic cancer. Am J Hum nick MH (1994) A strong candidate for the breast and ovarianGenet 52:711-717. cancer susceptibility gene BRCAI. Science 266:66-71.

Bell GI, Karam JH, Rutter WJ (1981) Polymorphic DNA region Oshimura M, Sandberg AA (1975) Isochromosome 17 in prostateadjacent to the 5' end of the human insulin gene. Proc Natl Acad cancer. J Urol 114:249-250.Sci USA 78:5759-5763. Utah NMarker Development Group (in press) A collection of ordered

Bergerheim USR, Kunimi K, Collins VP, Ekman P (1991) Deletion tetra-nucleotide repeat markers froth the human genome. Am Jmapping of chromosomes 8, 10, and 16 in human prostatic carci- Hum Genet.noma. Genes Chromosom Cancer 3:215-220. Zuliani G, Hobbs HH (1990) Tetranucleotide repeat polymorphism

Bookstein R, MacGrogan D, Milsenbeck SG, Sharkey F, Allred DC in the LPL gene. Nucleic Acids Res 18:4958.

Glý,ýomiOcs 28, 25-31 (1995)

A Gene (DLG2) Located at 17q12-q21 Encodes a New Homologueof the Drosophila Tumor Suppressor dig-A

SYLVIE MAZOYER, 1 SIMON A. GAYTHER, MARIA A. NAGAI, 2 SIMON A. SMITH,* ALISON DUNNING,ELIZABETH J. VAN RENSBURG, 3 HANS ALBERTSEN,* RAY WHITE,* AND BRUCE A. J. PONDER

CRC Human Cancer Genetics Research Group, Laboratories Block, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QP,United Kingdom; and *Eccles Institute of Human Genetics and Howard Hughes Medical Institute,

University of Utah, Salt Lake City, Utah 84112

Received December 13, 1994; accepted April 17, 1995

gene encoding a protein that contains a 90-amino-acidWe have isolated a novel cDNA that maps distal to repeat domain of unknown function, a SH3 (src homol-

BRCA1 at 17q12-q21. The total sequence predicts a ogy region 3) motif, and a guanylate kinase domainprotein of 576 amino acids with three conserved re- and thus belongs to the discs-large family of proteins.gions: a 90-amino-acid repeat domain, a SH3 (src ho- This family comprises the Drosophila discs-large tumormology region 3) motif, and a guanylate kinase do- suppressor product, dlg-A, localized to the septatejunc-main. These conserved regions are shared among tions in various epithelia (Woods and Bryant, 1991),members of the discs-large family of proteins that in- the human p55 membrane protein expressed in eryth-clude human p55, a membrane protein expressed in rocytes (Ruff et al., 1991), PSD-95/SAP90, a presynap-erythrocytes, rat PSD-95/SAP90, a synapse protein ex- tic junction protein from rat brain (Kistner et al., 1993),pressed in brain, Drosophila dlg-A, a septate junction and human and mouse ZO-1 and dog ZO-2, two tightprotein expressed in various epithelia, and human and junction proteins (Willot et al., 1993; Itoh et al., 1993;mouse ZO-1 and canine ZO-2, two tight junction pro-

teins. dlg-A has been shown to act as a tumor suppres- Jesaitis and Goodenough, 1994). As a result of this

sor, and the other members may all be involved in sig- homology, we naried this new gene DLG2.nal transduction through specialized membrane do- Allelic loss on chromosome arm 17q has been re-mains with highly organized cytoskeletons and thus ported in sporadic breast and ovarian tumors (Jacobsare potential tumor suppressors. Since allelic loss has et al., 1993; Saito et al., 1993; Nagai et al., 1994), andbeen reported in the 17q12-q21 region in breast and detailed maps delimiting the area of losses have beenovarian cancer and it appears that BRCA1 is not the constructed to investigate whether the breast and ovar-target of the losses, we looked for somatic alterations ian cancer susceptibility gene BRCA1 located inin DLG2 in sporadic breast tumors. No evidence 17q12-q21 is involved in the genesis of sporadic asfor mutation was found, making it unlikely that DLG2 well as inherited tumors, as is usually the case withis involved in sporadic breast cancer. © 1995 Academic genes predisposing to cancer. BRCA1 has been recentlyP•rss, Inc. cloned (Miki et al., 1994). Contrary to expectations, no

somatic mutations were identified in 44 sporadic breastand ovarian tumors studied, suggesting that BRCA1

INTRODUCTION is not involved in noninherited forms of the disease(Futreal et al., 1994). However, because LOH are fre-

While screening human cDNA libraries with re- quent in or adjacent to the BRCA1 region (Nagai et al.,agents used to generate a 4-cM physical map in 17q12- 1994; Cropp et al., 1994; Futreal et al., 1994), one orq21 (Albertsen et al., 1994), we have identified a new even more tumor suppressor genes may be located in

the same region (Ponder, 1994; Vogelstein and Kinzler,Sequence data from this article have been deposited with the 1994). As the DLG2 gene was a good candidate, we

EMBL Data Library under Accession No. X82895. looked for rearrangements in DLG2 in DNA from spo-'To whom correspondence should be addressed. Telephone: 44 radic breast and ovarian cancers and for mutations in

1223 336923. Fax: 44 1223 336902. E-mail: smazoyer@hgmp.mrc.a- a set of sporadic breast tumors for which genomic DNAc~uk. c. uk.was also available and that showed LOH in 17q.2 Present address: Disciplina de Oncologia, Departamento de Ra-diologia da Faculdade de Medicina da Universidade de Sao Paulo,Silo Paulo, Brazil. MATERIALS AND METHODS

:'Present address: Department of Human Genetics and Develop-nfental Biology, Medical Faculty of the University of Pretoria, Preto- DLG2 cDNA isolation and sequencing. A cDNA clone, 38B1/1,

South Africa. containing a 3.1-kb insert was isolated using a combination of 3 P1

250888-7543/95 $12.00

Copyright © 1995 by Academic Press, Inc.All rights of reproduction in any form reserved.

26 MAZOYER ET AL.

phages, 146B3, 750G12, and 92E12 when screening a fetal brain ized between D17S78 and 17HSD on chromosome ban1dcDNA library (Stratagene, Cat. No. 936206). Prior to further analy- 17q12-q21 (Albertsen et al., 1994). Comparison of thesis, this cDNA was shown to map back to 17q12-q21 when hybrid-ized to a filter containing P1 phages and YACs DNA covering the sequence of this cDNA clone, called 38B1/l, to thoseBRCAI region (see Fig. 2, Albertsen et al., 1994). eDNA sequence in the GenBank/EMBL database failed to reveal anywas then determined using a Taq DyeDeoxy Terminator Cycle se- convincing homology to known genes. However, thequencing kit on a automated sequencer ABI 373A (Applied Biosys- amino acid sequence deduced from the longest Opentems). The BLAST algorithm was used to look for homologies or reading frame showed a strong homology with p5 5 (dis.identities between the sequence identified and other sequences fromdatabases (GenBank/EMBL at the nucleotide level. Swissprot/NBRF cussed below), which is a major palmitoylated men,.at the amino acid level). While a polyadenylation signal and the brane protein of human erythrocytes (Ruff et al., 1991),beginning of a poly(A) tail were present in 38B1/1, the 5' end of the and to a lesser extent with a Drosophila tumor suppres.cDNA appeared to be missing. We then used a 5'-RACE-Ready testis sor protein, dlg-A (Woods and Bryant, 1991) and withcDNA kit (Clontech) to isolate additional 5' region. The PCR frag- a wide range of guanylate kinases (GK), the best homol.ment generated using this kit was cloned in a TA cloning vector(InVitrogen) and sequenced as described. Exon-intron boundaries ogy being with pig GK. This gene was called DLG2.were localized as follows: primers distributed evenly along the cDNA Since the initiation codon was not present in 38B1/1,sequence were used to amplify fragments from genomic DNA as well the 5' end of DLG2 was then pursued using the 5'as from the 38B1/1 clone. Whenever a difference in size was observed RACE technique, which allowed us to identify an addi.between genomic DNA and cDNA, the genomic fragment was sub- tional 357 bp. The nucleotide sequence and predictedcloned in a TA cloning vector and sequenced and intron-exon bound-aries were determined by comparison with the cDNA sequence. Be- amino acid sequence of the full-length cDNA are showncause we were unable to amplify from genomic DNA in the very 5' in Fig. 1. The first methionine codon is at nucleotideend of the coding sequence, certainly due to the presence of a big 88 and reveals an open reading frame of 1731 bp. Ifintron, the structure of this part of the gene has not been completely this ATG represents the initiation codon as suggestedresolved. by the presence of a purine 3 bp downstream, the open

Northern and Southern blots. A multiple human tissue Northern byathepresene of a prn3bpdnsram, te openand a zoo blot were purchased from Clontech and hybridized as rec- reading frame codes for a 576-amino-acid protein Of aommended. Southern filters were made from 5 pg EcoRI- or Rsal- predicted molecular weight of 64.5 kDa. The presence Idigested DNA purified from paired 28 blood/sporadic breast tumors of an in-frame stop codon upstream from the initiationand 33 blood/sporadic ovarian tumors. Hybridization and washes codon further supports that the sequence coding for thewere performed using standard protocols. Probes were labeled using N-terminal end of the protein has been isolated. The,the random-primed labeling protocol. 3' untranslated region is 1651 bp long and contains a

Sequence comparison analysis. Protein sequences were comparedby the PILEUP and GAP algorithms using the UK Human Genome consensus polyadenylation signal, AATAAA, followed'Mapping Project computing services, by the beginning of a poly(A) tail 15 bp downstream,)

Mutation screening. Paired primary breast tumors and normal Two RNA species of approximately 3.8 and 4.8 kb weretissue DNA samples were obtained from patients at the A. C. Ca- revealed on a Northern blot (Fig. 2a); the total sequencemargo Hospital, Sao Paulo, Brazil, and were fully described else- isolated is consistent with the size of the smallerwhere (Nagai et al., 1994).

One hundred nanograms of DNA was amplified in a 50-pl reaction mRNA. Attempts to isolate further sequence in the 5'using the following primers (see Fig. 4): 9F, GAAGAGACCCGGGAC- end using the 5' RACE technique was unsuccessful.ACTG; 9R, ATTCTCAATCCCCCCACC; 10F, ATTCTGACTTGG- Alternative splicing was observed in the 5' end (dataAGCAATTGG; 10R, GACAGCCAGGAGGCTCAG; IF, AGCCTT- not shown) but the splices identified cannot account foiCTGCTGACCCTTC; 1R, CCCTCCCATGCACCATAC; 2F. GTAT- the 4.8-kb mRNA.GGTGCATGGGAGGG; 11R, AAGGCATGCCATGTTAGAGG; 11F,AGCCTCCCTGAGAGCTCC; 3R, GAGCCCTGCTCCTTGTCC; 4F,GAGGCACTGTAACCTGCCC; 4R, AGGCAGCAGAGAGGACATTG; Northern Blot and Zoo Blot Analysis5F, TCCTAGGACAGACATGGGGA; 5R, CCACCCTAGGCAGCT-ATCAG; 6F, CTTCTGACAGTGGGGGTGAG; 6R, TCAGGTTCT- To assess the pattern of expression of dlg2 mRNA inGGGTTTCAACA. Exons 7 and 8 were amplified together because of normal tissue, the 38B p/I insert was used as a probetheir small size and because of the intron between them, whereas nthe other exons were amplified separately, in a Northern blot assay (Fig. 2a). Two mRNA of 4.8

PCR products were diluted with an equal volume of 95% for- and 3.8 kb were detected in prostate, testis, ovary,mamide and denatured at 100°C for 10 min before quenching on small intestine, and colon but not in spleen, thymus,ice. Prior to loading and polyacrylamide gel electrophoresis, samples and peripheral blood leukocytes. The level of expres'Iwere retained on ice for 10 min to allow a proportion of the single-strand DNA to reanneal, thereby encouraging the formation of any sion of these mRNAs is higher in testis than in the

potential double-stranded heteroduplexes. Electrophoresis was per- other tissues, while hybridization with an actin probeformed in nondenaturing 0.5X MDE polyacrylamide gels (J. T. indicated that each lane contained equal amounts OfBaker) using a Protean II vertical gel apparatus (Bio-Rad). Twenty- RNA (data not shown). The DLG2 gene was also foundcentimeter gels were run at 200 V for 8-12 h and the SSCP condi- to be expressed in breast epithelium and placenta whentions optimized for temperatures between 4 and 15-C. Polyacryl- Iamide gels were stained with 0.1% silver nitrate solution for instan- cDNA made from total RNA extracted from these two,taneous visualization of results. tissues was used as template in PCR experiments (dati

RESULTS not shown).Zoo blot analysis using the 38B1/1 insert as a probL

Isolation of DLG2 demonstrated sequence conservation in monkey, ratJWe isolated a 3.1-kb cDNA from a human fetal brain mouse, dog, and cow, but not rabbit, chicken, or yeasý

cDNA library using a mixture of three P1 phages local- (Fig. 2b).

CLONING OF A NEW GENE LOCALIZED AT 17q12-q21 27

GGAGCGC CCGCIGcTGC TGGAGCCGcc CGGAGcTAGG cGcCFCCCCGG GGCGAGGG AGAcGNTrCA GAGCCCTTGC CICCTTCACC 87

ATG CCG GTT GCC GCC ACC AAC TCT GAA ACT GCC ATG CAG CMA G`C CTG GAC MAC TiT GGA TCC Cit CCC AGT GCC ACG GGG OCT GCA GAG 177Met pro Val Ala ala thr asn ser glu thr Ala met gin gin Val leu asp agn let, gly ser let, pro ser Ala thr gly Ala Ala glu 30

VCTG GAC CiT ATC TTC CNT CGA GGC ATT ATG GMA AGT CCC ATA GrA AGA ICC CTG GCC MAG GTG ATA ATG GTA 'rIG TGG TTT ATG CAG CAG 267let, asp let, lie phe let, arg gly ile met glu ser pro lie val arg ser let, Ala lye val lie met val let, trp phe net gln gin 60

VMAT GTC TTT G`T CCI' ATG AAA TAC ATG CIrG AAA TAC TTN GGG GCC CAT GAG AGG CMt GAG GAG ACG MAG CTG GAG GCC GTG AGA GAC MAC 357asn val phe Val pro met lye tyr met leu lye tyr phe gly Ala his git, erg leu giu glu thr lye ieu giu ala val erg asp asn 90

MIC CIG GAG CIG GTG CAG GAG ATC CMG CGG GAC CiT GCG CAG CiT GCI' GAG GAG AGC AGC ACA GCC GCC GAG Cit GCC CAC ATC Cit GAG 447asn ieu giu let, Val gin glu lie let, erg asp let, Ala gin leu ala glu gin ser ear thr ala Aia git, ieu ala his le let, gin 120

VGAG CCC CAC TTC CAG TCC CrC CiT GAG ACG CAC GAC TCI' GTG GCC TCA MAG ACC TAT GAG ACA CCA CCC CCC AGC CCT GGC CiT GAC CC`T 537giu pro his phe gin ser leu leu giu thr his asp ser vai ala ser lys tin- tyr glu thr pro pro pro ser pro gly iou asp pro 150

VACG 'TT AGC MAC CAG CCT GTA CCT CCC GAT GCT GTG3 CCC ATG GTG GGC ATC CGC AAG ACA GCC GGA GMA CAT Cit GGT GTA ACG TTC CGC 627thr phe ser aen gin pro val pro pro asp ala val erg met Val gly lie erg iye thr aia gly glu his let, gly Val thr phe erg 180

GTrG GAG GGC GGC GAG CT G GTG ATC GCG CGC ATT Cit CAT GGG GGC ATG GTG OCT CAr CMA GGC Cit CTG CAT GTG GGT GAC ATC Alt MAG 717val giu gly gly giu let, valilie ala erg lie let, his giy gly met Val ala gin gin gly let, leu his Val gly asp lie iie lye 210

GAG GIG MAC GGG CAG CCA GTG GGC AGT GAC CCC CGC GCA Cit CAG GAG CTC CI'G CCC MAT GCC AGT GGC ACT GTC ATC CTC MAG ATC CTG 807git, val aen gly gin pro Val gly ser asp pro erg ala leu gin glu ieu let, erg asn ala ser gly ser Val lie leu lye le let, 240

VCCC MAC TAC GAG GAG CCC CAT CIG CCC CGC GAG GTA 'NT GIG MAA TGT CAC TTT GAC TAT GAC CCG GCC CGA GAC ACC CTC ATC CCC TCC 897pro asn tyr gin giu pro his ieu pro erg gin val phe val lye eye his pile asp tyr asp pro ala erg asp ser let, le pro eye 270

MAG GMA GCA GGC Cit CGC TTC MAC GCC GGG GAC itT CTC GAG ATC GTA MAC GAG GAT GAT GCC AAC TGG TGG GAG GCA TCC CAT GTC GMA 987lys giu ala gly ieu arg phe pen ala gly asp leu let, gin lie Val asn gin asp asp ala. aen trp trp gin ala eye his val git, 300

OGG GCC AGT GCT GGG CTC ATT CCC AGC GAG Cit CiT GAG GAG MAG CGG MAA CCA TNT GTC MAG API GAC CTG GAG CTG ACA CCA MAC TCA 1077gly gly ser ala gly let, lie pro ser gin let, let, git, giu lye erg lye ala phe val lye erg asp let, glu leu thr pro aen scr 330

GGG ACC CIA TCC GCC ACC CIT TCA GGA MG MAA AAG MG CGA ATG ATG TAT TTG ACC ACC MAG MAT GGA GAG 'NT GAC CGT CAT GAG CTG 1167gly thr let, eye gly ser leu ser gly lye lye lys lye erg met met tyr leu thr thr lye Ann Ala glu phe asp erg his giu ieu 360

mCC AN' TAT GAG GAG GIG GCC ccC ATG CCC CCG TIC CGC CGG MAA ACC CTG GTA Cit ATT GGG GC1P CAG GGC GIG GGA CGG CCC ACC CiT 1257leu lie tyr giu glu val ala erg met pro pro phe erg "ra lye thr Ie 1a 1-1 4i- ply glp gjn gl Vl -1 nv mra pro 5c e,390

MAG MAC MG CTC Alt ATG TGG GAT CCA GAT CGC TAT GGC ACC ACG GTG CCC TAC ACC TCC CGG CGG CCG MAA GAC TGA GAG CGG GMA GGT 1347lye pen lye let, ile mnet trnpn ag ro aso ýar tvr ply thr tin- ycl nro tythnerponp rleas5rcia ipy 420

GAG GCI TAC AGC TN' GTG TCC CGT GGG GAG ArG GAG GCI' GAC CIC CGT GCT GGG CGC TAC CTG GAG CAT GGC GM 'TAC GAG GGC MAC CTG 1437pin alv t-va or the -Al -j- Mr0 oly alPtpy i o a r icpyPotrlt 1,1l ly it, r ci, ly &1v n 1e- 450

TAT GCC ACA CGT AN' GAC TCC Alt CGG GGC GTG CCC GCr GCI' GGG MG GTG ICC GTG C~T GAT GTC MAC CCC GAG GCG GTG MAG GTG CIA 1527F~vr aly thlr .r. 4ie Ae 41a -i crc ply Vei vol cIc ale ply Vys ye1 -y 1 y -leg, c -1 yAl n nrQ pin Aie Vel 1y-cii 1 480

OGA ACG GCC GAG TT' GTC CCI' TAC GTG GTG 'NC Alt GAG GCC CCA GAC TTC GAG ACC Cit CGG. GCC ATG MAC AGG GCI' GCG C~T GAG AGT 1617Argin-rh Ala cl u rhe Val oro tvr yal yAl the lieA plj Ala oro asn nhe p1,, thi eaccamtar r i i e l 510

VGGA ATA TCC ACC MAG CAG CTC ACG GAG CCG GAC Cit AGA CGG ACA GTG GAG GAG ACC ACC CGC AlT GAG CGG PlC TAC GGG CAC TAC TTT 1707ply lie ner tin- lye aln let, thin alu Ala asp 1-t Ara prc tin- Val pn~i pluj -c -c vr l i r tyr ply his tvin Mhe 540

GAC CiT TGC Cit GTC MAT AGC MAC Cit GAG AGO ACC TTC CGC GAG CTCCGAG ACA GCC ATG GAG MAG CIA CGG ACA GAG CCC GAG TGG GTG 570Ago I-t cye leyietpreni, i, r h h r-1,lt i rraenc ilelt r thin giu pro gin trp val 1797

=CI GTC AGC TGG GiT TAC TGA 1818pro yal ser trp val tyr OPA 576

GCCTGTTCAC CiTG~TCCNG GCrGACTCCG TGTTGMAACC CAGAACCTG;A ATCCATCCCC CTCCCIGACCT GTGACCCCCr GCCACMATCC TTAGCCCCCA 1918TATCiTGCTG TCCI'TGGGTA ACAGCI'CCCA GCAGCCTrA AGTCI'GGCN'T CArCACAGAG CCOTGCACI'G CCGGGAGGT GGGCATI'CAT GGGGTACCTT 2018GTGCCCAGGT GCI'GCCCACT CCTGATGCCC ATTGGTCACC AGATATcrcT GAGGGCCAAG CrATGCCCAG GMATGTCI'CA GAGTCACCT CATMATGGTC 2118ACAGAGAAGA AGTGMAACAC TGCTTTGGGA CCACATGGTC AGTAGGCACA CitCCC=G CCACCCCI'CC CCAGTCACCA GTTCrCrCTC GGACiTGCCA 2218C-ACCCACCCC ATICCIGGAC TCC~TCCACC TC~TACCI'Cr GTGTCGGAGG AACAGGCCTT GGGCTG7TITC CGTGTGACCA GGGGAATGTG TGGC0CýCCG~ 2318GCAGCCAGGC AGGCCCGGGT GGTGGTGCCA GCCI'GGTGCC ATCI'TGAAGG CMGAGGA

trT GAGAGTGAGA CCCAGTGGCC ACAGCI'GCAG ACCACTGCAG 2418

CI'=cGCTC CN'TGGAAAG GGACAGGGTC GCAGGGCAGA TGCI'GCICGG TCCN'CCCI'C ATCCACAGCT ITCCACTIGCC GAAGTTrCrC CAGATITCTC 2518CAATGTGTCC TGACAGGTCA GCCCI'GCI'CC CCAGAGGGCC AGGCiTGGG GPGCAGTGG GTTCAGCCCCA GGTAGGOGCA GGATGGAGGG CrGAGCCCCCG 2618TGACMACCTG CIGTI'ACCAA CI'GAGAGCC CCMAGCTCI'C CATGGCCCC AOCAGGCACA GCCTrGAGC1' CTATGTCCTT GAC'TTGGC CANTGTT N 2718TCTGCIAGC CAGG'TCCAGG TAGCCCACTT GCATCAGGCC TGCI'GGGTTG GAGGGGCTMA GGAGGAGTCC AGAGGGGACC TTGCGAGCCI' GGGCTTGMAG 2818AGACAGTGCC CICCArGAGG TTCCrGACAC ACMACTCCAG AGGCGCCATr TACACTGTAG TCTGTACMAC CrGTGO'NCC ACGrGCATGT TCGGCACCTG 2918

TCI'GI'GCCC TGGCACCAG3G TTGiTTGTGT GTGCGTGTCC AC~rGCGTGT GTGTGTGTiGT GrTCCAGGT TAGTTTGGG AGGAACCAAA GGGTITTG'N' 3018TI'GGAGGTCA CI'CMIIGGGG CCCCITCrG GGGGTTCCCC ATGAGCCCI'C AN'TCN'ATA ATACCCI'GAT CCCGACIC MAAGCCCI'GG TCCT'N'CCG 3118ATGTC='C' CCN'CICTTA TTGITCC=CCT ACCCTAAATG CCCCCTGCC ATMACITGGG GAGGGCAGTT TI'GTAMAATA GGAGAC='C TTI'AAGAAAG 3218MATGCrG;TCC TAGATGTACT TGGCCATCTC ATCCTTGATr ATTCT~r=C TTCCTTCCGG PGGGGAGCCTG TCCCAGAGG GGACMACCIG TGACACCCI'G 3318AGTCGAMACC CN'GTGCC1C CCAGTTCTTC CAAGTGCrCA ACTAGTCI'T GCI'GCAGCGT CAGCCAAACC TGGCCCCIGA ACCACitTGT GCCCATTTCC 3418TAGGGM4GGG GMGCGAGMAT AAACAGMATA TTTATTACM MAA 3461

FIG. 1. Nucleotide sequence of DLG2. The nucleotide sequence of DLG2 is presented along with the predicted amino acid sequence.Tb.. Positions of 11 introns are indicated with arrowheads above the sequence, the DHR domain is underlined with crosses, the SH3 domainis double-underlined, and the guanylate domain is single-underlined. Poly(A) signal is underlined with asterisks. The first nucleotide ofclone 38B 1/1 is indicated with a dot.

28 MAZOYER ET AL.

a b dga bDHR3 SH3 GK

S0 00

>1 .) )-p55'&= 2-• - . 40 66 42

9.5 H 9R4 - DHR1 08R2 (63X85) (68)

765 - 6.5 26 dig-A4. - . 26 45 31

(48) (68) (58)I I I I ////IPSD-95/SAP90

' 28 37 302.4

(49) (70) (58)2.3_ - I I I zo-122 27 2 8

FIG. 2. (a) Northern analysis ofDLG2. A human multiple-tissue (51)(56) (s.)

Northern blot was probed with the 38B1/1 insert. Two mRNA species I -IM Z0-228 23 24

of 4.8 and 3.8 kb are detected in prostate, testis, ovary, small intes- (57)(60) (54)

tine, and colon but not in spleen, thymus, and peripheral blood leuko-cytes. Size standards are in kilobases. (b) Conservation analysis of FIG. 3. Schematic comparison of structural domains of dlg2 withDLG2. An EcoRI zoo blot was probed with the 30B1/1 insert. Frag- those of p55, dlg-A, PSD-95/SAP90, ZO1, and Z02 (DHR, 90-aa inter-ments are detected in monkey, rat, mouse, dog, and cow, but not in nal repeat; SH3, src homology 3; GK, guanylate kinase domain).rabbit, chicken, or yeast. Size standards are in kilobases. Percentage identity and percentage similarity (in parentheses) for

each domain were calculated using the GAP algorithm.

dlg2 Belongs to the Discs-Large Family of Proteins larity for dlg2, 28% identity, 48% similarity for p55),

The predicted amino acid sequence of dlg2 is homolo- while it shows 40% identity (63% similarity) betweengous to the human erythrocyte membrane protein p55 dlg2 and p55. The N-terminal amino acid sequence of(Ruff et al., 1991) (42% identity, 64% similarity), the ZO-2 has not yet been deduced, and although only onerat brain synapses protein PSD-95/SAP90 (Kistner et repeat, DHR3, has been found so far, two more domainsal., 1993) (27% identity, 53% similarity), the Drosoph- may be identified, as suggested by the high degree ofila dlg-A tumor suppressor (Woods and Bryant, 1991) homology between ZO-1 and ZO-2 (Jesaitis and(26% identity, 48% similarity), the dog tight-junction Goodenough, 1994). The SH3 domain of dlg2 is moreprotein ZO-2 (Jesaitis and Goodenough, 1994) (22% homologous to the SH3 domain of p55 than to that ofidentity, 50% similarity) and the human tight-junction dlg-A, PSD-95/SAP90, ZO-1, and ZO-2 (Fig. 3). Also,ZO-1 (Willott et al., 1993) (20% identity, 48% similar- the dlg2 guanylate kinase domain shows closer homol-ity) (Table 1). The sequence homologies of these family ogy to that of p55 than that of the other members of themembers are clustered in three domains (Fig. 3): a 90- discs-large family. Furthermore, while three of eightamino-acid internal repeat region called DHR (discs- amino acids that are thought to bind the ATP phos-large homology region) (Bryant et al., 1993), a SH3 phate donor (Stehle and Schulz, 1990) are deleted indomain (src homology 3) which was first identified in the , 'anylate kinase domain of dlg-A, PSD-95/SAP90,the noncatalytic region of the src family of protein tyro- ZO-1, and ZO-2 (Koonin et al., 1992; Willott et al.,sine kinases (Musacchio et al., 1992), and now found 1993), both dlg2 and p55 retain these residues. Never-in multiple signaling and cytoskeletal proteins (Mayer theless, this motif (GxxGxGKS/T) does show substitu-and Baltimore, 1993), and a guanylate kinase domain. tions in highly conserved residues in both proteins.

While ZO-1, PSD-95/SAP90, and dlg-A have three These might affect enzyme activity (GxxGxGRR for dlg-DHR domains, DHR1, DHR2, and DHR3, p55, and dlg2 1; GxxGxRS for p55); however, the Lys to Arg substitu-contain only one repeat (Fig. 3). Although this repeat tion has been observed in several putative ATPasesshows greater homology to the third repeat of dlg-A, (Gorbalenya and Koonin, 1990).this homology is not very high (26% identity, 48% simi-

Mutation Analysis of DLG2

TABLE 1 To determine whether DLG2 is the target for the

Discs-Large Family Member Comparisons allele losses frequently reported in the BRCA1 regionin sporadic breast and ovarian cancer (Jacobs et al.,

PSD-95/ 1993; Saito et al., 1993; Nagai et al., 1994; Cropp et al.,p55 SAP90 dlg-A ZO-1 ZO-2 1994), we first hybridized the major part of dlg2 cDNA

with a set of Southern filters containing DNA purified42 (64) 27 (53) 26 (48) 20 (48) 22 (50) dlg2 from blood/tumor pairs from 28 sporadic breast and

27 (55) 28 (52) 24 (47) 20 (46) p55

58 (76) 31(55) 17 (41) PSD-95/ 33 sporadic ovarian tumors. No rearrangements were

SAP9o found. We then performed a more sensitive mutation26 (47) 20 (42) dlg-A analysis by screening genomic and sporadic breast tu-

52 (69) ZO-1 mor DNA from 19 patients using single-strand confor-

Note. Percentage identity and percentage similarity (in brackets) mation polymorphism (SSCP) and heteroduplex analy-were calculated for each pair throughout the protein using the GAP ses. This set of samples was chosen on the basis ofalgorithm. The highest scores are shown in bold type. displaying losses on 17q, as has been described else-

, CLONING OF A NEW GENE LOCALIZED AT 17q12-q21 29

1 2 3 4 5 6 7 8 9 10 11 12

9F/9R 1OF/1OR IF/1R 2F/11R 11F/3R 4F/4R SF/SR 6F/6R(274) .;. (333) - (230) -.,- (387) - (279) = (309) - (230) (283)

DHR SH3 Guanylate Kinasedomain domain domain

FIG. 4. Schematic structure of the DLG2 gene. Intron-exon structure is shown, as well as primers used for the mutation analyses. Thesi/e of each exon is given in the box, while the sizes of the PCR fragments are given in parentheses; the sizes of the introns are not toscale. The conserved domains are shown below.

where (Nagai et al., 1994), for D17S855, which is located showed a strong homology at the amino acid level within an intron of the BRCA1 gene (Futreal et al., 1994) an erythroid membrane protein, p55 (Ruffet al., 1991),and is close to DLG2. Exons covering 78% of the coding and to a lesser extent with a Drosophila septate junc-sequence including the DHR, SH3, and guanylate kinase tion protein, dlg-A (Woods and Bryant, 1991), a ratdomains were amplified using primers located in adjacent presynaptic protein, PSD-95/SAP90 (Kistner et al.,introns (Fig. 4). Single-strand variants were observed in 1993), and two tight junction proteins, human (Willott1() tumors, some of which displayed abnormal fragments et al., 1993) and mouse ZO-1 (Itoh et al., 1993) andin more than one exon, but because these fragments were canine ZO-2 (Jesaitis and Goodenough, 1994). This pro-present in germ-line DNA as well, they were assumed to tein, which we named dlg2, contains 576 amino acidsrepresent polymorphisms. A total of five different vari- and displays all of the features shared by proteins thatants were found in four different exons (representative belong to the discs-large family. These include a 90-SSCP and heteroduplex analyses are shown in Fig. 5). amino-acid motif of unknown function called DHR, anEight individuals were heterozygous in exon 4, three in SH3 domain, and a guanylate kinase domain. The DHRexon 11, and one in exon 12. In exon 5, two different domain, also present in nitric oxide synthetase (Cho ettypes of variants were found, three individuals being het- al., 1992), a human cytosolic tyrosine phosphatase (Gucrozygous for the first variation and four for the second et al., 1991), and the cytoplasmic domain of a humanofiw. However, no aberrant bands were observed in the brain transmenfbrane protein (Duclos et al., 1993)tumor samples only and not in blood. Both the Southern might be involved in directing membrane association.and the SSCP/heteroduplex analyses therefore suggest The SH3 domain has been shown to be involved in thethat DLG2 is not mutated in sporadic breast cancers. targeting of proteins to specific subcellular locations

DISCUSSION (Bar-Sagi et al., 1993). It is also possible that certain

One of the clones identified from screening a human SH3 domains participate in the control of cytoskeletalfetal brain library, which mapped to 17q12-q21, organization. The yeast guanylate kinase (GK) cata-

lyzes the conversion of GMP to GDP by transferring aAR9 112L 110L 11L phosphate group from ATP. Although the GK domainsTa T N T N T N b IOL ~1 L of dlg2 and p55 retain the conserved amino acids

T N T N thought to interact with ATP and thus are more likely

to exert an enzymatic activity than those of ZO-1, ZO-' ~ ~ 64" r't:"& A 's s[U 2, dlg-A, and PSD-95/SAP90, direct evidence for such] an activity is needed. An alternative function suggested

, , for these GK domains comes from the observation thatthe structure of yeast GK is similar to that of smallGTP-binding proteins (Stehle and Schulz, 1990). Thus,members of the discs-large family of proteins could be

2SG IsG involved in signaling pathways through interactionT N with G-protein-binding proteins and not by altering

I ýOww WGDP/GTP ratios.The dlg-A, ZO-1, ZO-2, and PSD-95/SAP90 proteins

if •are localized to regions of cell-cell contact, and p55 iscopurified during the isolation of dematin, an actin-

ao•,.lIw 0i u• W bundling protein of the erythrocyte membrane cy-

toskeleton (Ruff et al., 1991); dlg2 might then beFIG. 5. Mutation analysis ofDLG2 in breast tumors. Representa- localized at the membrane of the cells in which it is

tive pattern of single-strand (SS) and double-stranded hetero- (HeD) expressed, possibly more specifically in specializedand homo- (HoD) duplexes for paired tumor (T) and normal (N) tissue membrane domains with a highly organized cyto-Of individuals for (a) exon 4-samples 112L and I10L show SS vari- skeleton.ants; (b) axon 5-sample 110L demonstrates variation in both SSand HeD patterns; and (c) exon 11-sample 2SG demonstrates SS Thus far, only two tissues have been tested for thevariation, expression of both dlg2 and p55: while p55 alone is

30 MAZOYER ET AL.

expressed in red cells, both are expressed in placenta zler, 1994). Since DLG2 was a good candidate, we in-(Metzenberg and Gitschier, 1992). The coexpression of vestigated its involvement in sporadic breast cancer.these two very closely related proteins implies a specific Although a total of five different variants were foundfunction for the first 81 amino acids of dlg2, the only in four different exons when performing a mutationpart of this protein not homologous to p55. It should analysis, which demonstrates the reliability of the tech-be noted that all discs-large protein encoding genes niques used, no variations in migration of PCR frag-reported to date, except ZO-1 and ZO-2, give rise to at ments generated from tumor DNA alone were ob-least two mRNAs both expressed in the same tissues, served. Although it could be argued that mutationsalthough at different levels (Woods and Bryant, 1991; were missed, we believe that the combination of SSCPRuffet al., 1991; Kistner et al., 1993). Nothing is known and heteroduplex analyses is sensitive enough to con-about how these two mRNA species are generated. In clude that DLG2 is not mutated in breast tumors. How-the case of PSD795/SAP90, the larger mRNA species ever, the involvement of DLG2 in other cancers as adisappears during development (Kistner et al., 1993). tumor suppressor warrants further attention.Only one of the corresponding sequences is availablefor each of these four genes, and it would be interesting ACKNOWLEDGMENTSto know whether these isoforms show any homology toor affect the specificity of each of these proteins. It is We thank P. Russel, P. Harrington, P. Holik, and J. Stevens for

also noteworthy that all of these genes are subject to assistance and I. Mika6lian for helpful suggestions. This work was

alternative splicing, supported by the Cancer Research Campaign (S.A.G., A.D., and

The first member of this protein family to be cloned B.A.J.P.). S.M. is supported by an EC fellowship, S.A.S. is supportedby an EMBO fellowship, and E.J.v.R. is the holder of the Lady Cade

is also the one whose function is best elucidated. Reces- Memorial fellowship funded jointly by the CRC and National Cancer

sive lethal mutations in the Drosophila dig gene pro- Association of South Africa. B.A.J.P. is a CRC Gibb fellow.

duce a neoplastic overgrowth phenotype in imaginal Note added in proof Because a new human gene called hdlg (HGM

discs, due to an absence of proliferation control, apical- symbol DLG1), which codes for a protein that belongs to the discs-

basal cell polarity, cell adhesion, and ability of cells large family, was published after this article was submitted (R. A.

to differentiate (Woods and Bryant, 1991). Mutation Lue, S. M. Marfatia, D. Branton, and A. H. Chishti (1994). Cloningand characterization of hdlg: The human homologue of the Drosoph-

analysis indicates that both the GK domain function ila discs large tumor suppressor binds to protein 4.1. Proc. Natl.

and the amino-terminal end of the gene product are Acad. Sci. USA 91: 9818-9822), the gene that we report here has

important for its tumor suppressing function (Woods been called DGL2.

and Bryant, 1991). The GK domain of dlg-A, even if notfunctional as such, is potentially involved in guanine REFERENCESnucleotide metabolism or regulation, which suggests arelationship with signal transduction mechanisms Albertsen, H. M., Smith, S. A., Mazoyer. S., Fujimoto, E., Stevens,

known to control cell proliferation. Because of their J., Williams, B., Rodriguez, P., Cropp, C. S., Slijepcevic, P., Carl-

homology to dlg-A, all of the members of the discs-large son, M., Robertson, M., Bradley, P., Lawrence, E., Harrington, T.,Mei Sheng, Z., Hoopes, R., Sternberg, N., Brothman, A., Callahan,

family of proteins are considered to be potential tumor R., Ponder, B. A. J., and White, R. (1994). A physical map and

suppressors. In the case of p55, this hypothesis has candidate genes in the BRCA1 region on chromosome 17q12-21.

been strengthened by the finding that p55 interacts Nature Genet. 7: 472-479.

with a 30-kDa domain located in the N-terminal end Bar-Sagi, D., Rotin, D., Batzer, A., Mandiyan, V., and Schlessinger,

of protein 4.1, a domain that is also present in the NF2 J. (1993). SH3 domains direct cellular localization of signaling mol-

(neurofibromatosis 2) tumor suppressor gene (Marfatia ecules. Cell 74: 83-91.

et al., 1994). Furthermore, the junctional plaque pro- Bryant, P. J., Watson, K. L., Justice, R. W., and Woods, D. F. (1993).Tumor suppressor genes encoding proteins required for cell inter-

teins, which include dlg-A, PSP95-SAP90, ZO-1, and actions and signal transduction in Drosophila. Dev. Suppl. 239-ZO-2, are thought to be directly involved in the regula- 249.

tion of the organization of the cytoskeleton, cell adhe- Cho, K.-O., Hunt, C. A., and Kennedy, M. B. (1992). The rat brain

sion, and cell motility and thus are considered to be postsynaptic density fraction contains a homolog of the Drosophila

potential tumor suppressors (Tsukita et al., 1993). discs-large tumor suppressor protein. Neuron 9: 929-942.

The DLG2 gene is located in a region showing LOH Cropp, C. S., Nevanlinna, H. A., Pyrhonen, S., Stenman, U.-H., Sal-

in about 25% of breast sporadic cancers (Nagai et al., mikangas, P., Albertsen, H., White, R., and Callahan, R. (1994).Evidence for involvement of BRCAI in sporadic breast carcinomas.

1994; Cropp et al., 1994; Futreal et al., 1994). Until it Cancer Res. 54: 2548-2551.

was cloned, BRCA1, the breast and ovary cancer sus- Duclos, F., Boeschert, U., Sirugo, G., Mandel, J. L., Hen, R., and

ceptibility gene located on 17q, was thought to be the Koenig, M. (1993). Gene in the region of the Friedreich ataxia

target for these losses. However, analysis of 32 sporadic locus encodes a putative transmembrane protein expressed in the

breast and 12 sporadic ovarian tumors displaying LOH nervous system. Proc. Natl. Acad. Sci. USA 90: 109-113.

for a marker located in an intron of BRCA1 failed to Futreal, P. A., Liu, Q., Shattuck-Eidens, D., Cochran, C., Harshman,

reveal any somatic mutations in BRCA1 (Futreal et al., K., Tavtigian, S., Bennett, L. M., Haugen-Strano, A., Swensen, J.,Miki, Y., Eddington, K., McClure, M., Frye, C., Weaver-Feldhaus,

1994). This has raised the possibility that a second gene J., Ding, W., Gholami, Z., Soderkvist, P., Terry, L., Jhanwar, S.,located nearby might be involved in sporadic breast Berchuck, A., Iglehart, J. D., Marks, J., Ballinger, D. G., Barrett,

and ovarian tumors (Ponder, 1994; Vogelstein and Kin- J. C., Skolnick, M., Kamb, A., and Wiseman, R. (1994). BRCAI

CLONING OF A NEW GENE LOCALIZED AT 17q12-q21 31

mutations in primary breast and ovarian carcinomas. Science 266: man, K., Tavtigian, S., Liu, Q., Cochran, C., Bennett, L. M., Ding,

120-122. W., Bell, R., Rosenthal, J., Hussey, C., Tran, T., McClure, M., Frye,

Gorbalenya, A. E., and Koonin, E. V. (1990). Superfamily of UvrA- C., Hattier, T., Phelps, R., Haugen-Strano, A., Katcher, H., Ya-

related NTP-binding proteins: Implications for rational classifica- kumo, K., Gholami, Z., Shaffer, D., Stone, S., Bayer, S., Wray, C.,

tion of recombination/repair system. J. Mol. Biol. 213: 583-591. Bogden, R., Dayananth, P., Ward, J., Tonin, P., Narod, S., Bristow,P. K., Norris, F. H., Helvering, L., Morrison, P., Rosteck, P., Lai,Gu, M. X., York, J. D., Warshawsky, I., and Majerus, P. W. (1991). M., Barrett, J. C., Lewis, C., Neuhausen, S., Cannon-Aibright, L.,

Identification, cloning, and expression of a cytosolic megakaryocyte GoMdgar, D., Wiseman, R., Kamb, A., and Skolnick, M. H. (1994).

el flt pr( Tiosine-Phosphatase with sequence homology to cytoskel- A strong candidate for the breast and ovarian cancer susceptibilityeti1I protein 4.1. Proc. Natl. Acad. Sci. USA 88: 5867-5871.geeBC1Scne26:6-.

gene BRCA1. Science 266: 66-71.Itoh. M., Nagafuchi, A., Yonemura, S., Kitani-Yasuda, T., Tsukita, Musacchio, A., Gibson, T., Lehto, V.-P., and Saraste, M. (1992).

S., and Tsukita, S. (1993). The 220-kD protein colocalizing with SH3-An abundant protein domain in search of a function. FEBScadherins in non-epithelial cells is identical to ZO-1, a tight junc- Lett. 307: 55-61.tion-associated protein in epithelial cells: cDNA cloning and immu- Nagai, M. A., Yamamoto, L., Salaorni, S., Pacheco, M. M., Brentani,noelectron microscopy. J. Cell Biol. 121: 491-502. M. M., Barbosa, E. M., Brentani, R. R., Mazoyer, S., Smith, S. A.,

Jacobs, I. J., Smith, S. A., Wiseman, R. W., Futreal, P. A., Harrington, Ponder, B. A. J., and Mulligan, L. M. (1994). Detailed deletionT., Osborne, R. J., Leech, V., Molyneux, A., Berchuck, A., Ponder, mapping of chromosome segment 17q12-21 in sporadic breast tu-B. A. J., and Bast, R. C. (1993). A deletion unit on chromosome mours. Genes Chrom. Cancer 11: 58-62.17q in epithelial ovarian tumors distal to the familial breast/ovar- Ponder, B. A. J. (1994). Searches begin and end. Nature 371: 279.ian cancer locus. Cancer Res. 53: 1218-1221. Ruff, P., Speicher, D. W., and Husain-Chishti, A. (1991). Molecular

.(JsýIis, L. A., and Goodenough, D. A. (1994). Molecular character- identification of a major palmitoylated erythrocyte membrane pro-ization and tissue distribution of ZO-2, a tight junction protein tein containing the src homology 3 motif. Proc. Natl. Acad. Sci.homologous to ZO-1 and the Drosophila discs-large tumor suppres- USA 88: 6595-6599.sor protein. J. Cell Biol. 124: 949-961. Saito, H., Inazawa, J., Saito, S., Kasumi, F., Koi, S., Sagae, S., Kudo,

Kistner, U., Wenzel, B. M., Veh, R. W., Cases-Langhoff, C., Garner, R., Saito, J., Noda, K., and Nakamura, Y. (1993). Detailed deletionA. M., Appeltauer, U., Voss, B., Gundelfinger, E. D., and Garner, mapping of chromosome 17q in ovarian and breast cancer: 2 cMC. C. (1993). SAP90, a rat presynaptic protein related to the prod- region on 17q21.3 often and commonly deleted in tumors. Canceruct of the Drosophila tumor suppressor gene dlg-A. J. Biol. Chem. Res. 53: 3382-3385.268: 4580-4583. Stehle, T., and Schulz, G. E' (1990). Three-dimensional structure of

Koonin, E. V., Woods, D. F., and Bryant, P. J. (1992). dlg-R proteins: the complex of guanylate with its substrate GMP. J. Mol. Biol.

Modified guanylate kinases. Nature Genet. 2: 256-257. 211: 249-254.

Marfatia, S. M., Lue, R. A., Branton, D., and Chishti, A. H. (1994). Tsukita, S., Itoh, M., Nagafuchi, A., Yonemura, S., and Tsukita, S.Inz vitro binding studies suggest a membrane-associated complex (1993). Submembranous junctional plaque proteins include poten-

between erythroid p55, protein 4.1, and glycophorin C. J. Biol. tial tumor suppressor molecules. J. Cell Biol. 123: 1049-1053.

Chem. 269: 8631-8634. Vogelstein, B., and Kinzler, K. W. (1994). Has the breast cancer genebeen found? Cell 79: 1-3.

Mayer, B. J., and Baltimore, D. (1993). Signaling through SH2 and Willott, E., Balda, M. S., Fanning, A. S., Jameson, B., Van Itallie,C., and Anderson, J. M. (1993). The tight junction protein ZO-1 is

Metzenberg, A. B., and Gitschier, J. (1992). The gene encoding the homologous to the Drosophila discs-large tumor suppressor proteinpalmitoylated erythrocyte membrane protein, p55, originates at of septate junctions. Proc. Natl. Acad. Sci. USA 90: 7834-7838.the CpG island 3' to the factor VIII gene. Hum. Mol. Genet. 1: 97- Woods, D. F., and Bryant, P. J. (1991). The discs-large tumor sup-101. pressor gene of Drosophila encodes a guanylate kinase homolog

Miki, Y., Swensen, J., Shattuck-Eidens, D., Futreal, P. A., Harsh- localized at septate junctions. Cell 66: 451-464.

GENOMICS 31, 145-150 (1996)ARTICLE NO. 0025

Isolation of a Gene (DLG3) Encoding a Second Member of theDiscs-Large Family on Chromosome 17q12-q21

SIMON A. SMITH,' PILAR HOLIK, JEFF STEVENS, SYLVIE MAZOYER,* ROBERTA MELIS,BRIANA WILLIAMS, RAY WHITE, AND HANS ALBERTSEN

Huntsman Cancer Institute, University of Utah, Salt Lake City, Utah 84112; and * CRC Human Cancer Genetics Research Group,University of Cambridge, Addenbrooke's Hospital, Cambridge CB2 2QQ, United Kingdom

Received June 8, 1995; accepted October 30, 1995

homology motif 3 (SH3); and a C-terminal domain withThe discs-large family is a collection of proteins that homology to guanylate kinase (GK). The founding

have a common structural organization and are member of this group of proteins is the gene productthought to be involved in signal transduction and me- of the Drosophila lethal (1) discs large (dlg) 1 locus,diating protein-protein interactions at the cyto- which was originally identified through the character-plasmic surface of the cell membrane. The defining ization of recessive lethal lesions that lead to overproli-member of this group of proteins is the gene product feration of the imaginal disc epithelium and death ofof the Drosophila lethal (1) discs large (dig) 1 locus, the fly in the larval stage (Stewart et al., 1972; Woodswhich was originally identified by the analysis of re- and Bryant, 1989, 1991). At all stages of development,cessive lethal mutants. Germline mutations in dig re- the dlg protein appears to be a component of the septatesult in loss of apical-basolateral polarity, disruption junctions (Woods and Bryant, 1991), which are foundof normal cell-cell adhesion, and neoplastic over- in (od a nd brant 1991),thich a ndgrowth of the imaginal disc epithelium. We have iso- in the apical-lateral membrane of epithelial cells andlated and characterized a novel human gene, DLG3, are thought to be functionally equivalent to vertebrate

that encodes a new member of the discs-large family tight junctions (Noirot-Timothee and Noirot, 1980).

of proteins. The putative DLG3 gene product has a mo- Germline mutations in dig result in loss of apical-baso-

lecular weight of 66 kDa and contains a discs-large lateral polarity, disruption of normal cell-cell adhe-

homologous region, a src oncogene homology motif 3, sion, and neoplastic overgrowth of the imaginal discand a domain with homology to guanylate kinase. The epithelium (Stewart et al., 1972; Woods and Bryant,DLG3 gene is located on chromosome 17, in the same 1989), suggesting that the function of the wildtype dlgsegment, 17q12-q21, as the related gene, DLG2. The protein is concerned with maintaining normal epithe-products of the DLG2 and DLG3 genes show 36% iden- lial structure and possibly with regulating cell divisiontity and 58% similarity to each other, and both show and differentiation (Woods and Bryant, 1991). Othernearly 60% sequence similarity to p55, an erythroid members of the discs-large family include ZO-1 andphosphoprotein that is a component of the red cell ZO-2, which are mammalian tight junction proteinsmembrane. We suggest that p55, DLG2, and DLG3 are (Willott et al., 1993; Jesaitis and Goodenough, 1994);closely related members of a gene family, whose pro- SAP90/PSD-95, a rat synapse-associated protein (Chotein products have a common structural organization et al., 1992; Kistner et al., 1993); p55, an erythroidand probably a similar function. © 1996 Academic Press, Inc. phosphoprotein that is a component of the red cell

membrane (Ruff et al., 1991); and hdlg, a recently dis-covered human homolog of Drosophila dlg (Lue et al.,

INTRODUCTION 1994).

The DHR motif is approximately 90 amino acids longThe discs-large family is an expanding group of pro- and appears to be involved in forming protein-protein

teins that each contain three distinct structural do- interactions. In vitro binding studies have demon-mains: an N-terminal segment comprising one or more strated that hdlg binds to protein 4.1 (Lue et al., 1994),discs-large homologous regions (DHRs); a src oncogene the defining member of a group of proteins that in-

cludes talin (Rees et al., 1990) and ezrin (Gould et al.,The DNA sequence presented in this article has been deposited 1989), which are thought to couple the cytoskeleton to

with the EMBLJGenBank Data Libraries under Accession No. the cell membrane (Marfatia et aI., 1994). The bindingU37707.ofpoen41thdgwsmpetotostsohl:

'To whom correspondence should be addressed at Huntsman Can- of protein 4.1 to hdlg was mapped to two sites on hdlg:cer Institute, University of Utah, Building 533, Suite 7410, Salt Lake a domain containing all three DHR segments and an

City, Utah 84112. Telephone: (801) 585-6224. Fax: (801) 585-3833. insertion domain (13) found between the SH3 and GK

1450888-7543/96 $12.00

Copyright © 1996 by Academic Press, Inc.All rights of reproduction in any form reserved.

146 SMITH ET AL.

domains in a subset of hdlg isoforms (Lue et al., 1994). described by the manufacturer. A 1.8-kb EcoRI restriction DNA frag-In human erythrocytes, spectrin binds to the C-termi- ment excised from the 40F1 cDNA was used as the probe.

nal end of protein 4.1, while at the N-terminus, protein Database searches and sequence comparisons. To identify homol-ogies with nucleic acid and protein sequences in the GenBank and

4.1 forms a ternary complex with p55 and glycophorin EMBL databases, we used the BLAST algorithm (Altschul et al.,

C (a transmembrane protein) that links the spectrin 1990). Sequences were aligned and compared using the Wisconsin

network to the cell membrane (Marfatia et al., 1994). Sequence Analysis Package Version 8.0.

However, it is not clear what the function of p55 is inthis process, and the binding sites in p55 have not been RESULTS AND DISCUSSIONprecisely mapped. The SH3 domain comprises about60 amino acids and was originally identified in the non- A novel cDNA clone (40F1) derived from human chro-catalytic region of the src family of nonreceptor protein mosome segment 17q12-q21 was isolated from a fetaltyrosine kinases (Pawson, 1988). SH3 domains have brain cDNA library and used to identify further over-now been found in a wide variety of proteins, including lapping clones. The combined DNA sequence from sev-signal-transducing proteins and membrane-associated eral clones revealed a single long open reading framecytoskeletal elements, and are thought to be involved (ORF) that extended from the most 5' end of the se-in mediating protein-protein interactions by binding quence to a stop codon located over 1700 bp down-to specific target sequences (Koch et al., 1991; Mayer stream. The cDNA contained 900 bp of 3'-nontrans-and Eck, 1995). The GK domain is homologous to the lated sequence and terminated with a poly(A) tail.complete amino acid sequence of yeast guanylate ki- Since the DNA sequence did not contain a candidatenase (Berger et al., 1989), an enzyme that catalyzes the initiation codon, the 5' end of the cDNA was pursuedtransfer of phosphate from ATP to GMP, forming GDP, using the RACE technique, which extended the 5' se-although at the moment, it is unclear whether any of quence for a further 100 bp and revealed a probablethe dig-related proteins have guanylate kinase activity, start codon. The completed ORF extended over 1755

Recently, we reported the isolation and characteriza- bp (Fig. 1), encoding a protein of 585 amino acids withtion of a gene, DLG2, that encodes another novel mem- a predicted molecular weight of 66 kDa. The expectedber of the discs-large family of proteins (Mazoyer et al., gene product represents a new member of the discs-1995). The predicted DLG2 gene product contains a large family of proteins (see below), and the newly iso-single DHR motif and shows greatest homology to p55. lated gene has been designated DLG3.Here, we describe the isolation of a second novel human The molecular characterization of DLG3 included angene, DLG3, located close to DLG2 on chromosome seg- analysis of the genomic structure of the gene. Intron-ment 17q12-q21, which is predicted to encode a pro- exon boundaries were identified as points of disconti-tein that is nearly 60% homologous to the p55 and nuity between the cDNA sequence and portions of geno-DLG2 gene products. We suggest that p55, DLG2, and mic sequence. Fragments of human DNA containingDLG3 are three members of a family of genes that en- potential boundaries were generated either by PCR us-code proteins of similar size and structural organiza- ing primer pairs located within adjacent coding exonstion that are probably involved in coupling the cytoskel- or by subcloning fragments from P1 phage. The codingeton to the cell membrane. sequence was found to be distributed between 15 exons,

the boundaries of which are indicated in Fig. 1. The 9

MATERIALS AND METHODS 3'-most exons were found to span a genomic intervalof about 16 kb (based on the combined sizes of genomic

DLG3 cDNA isolation and DNA sequencing. Genomic DNA clones clones and PCR products); the 3 5-most exons were

that form part of a physical contig surrounding the BRCA1 locus contained within a 1200-bp region; and the genomic

(Albertsen et al., 1994) were used to screen a fetal brain cDNA library extent of exons 4, 5, and 6 was not clearly defined.to identify transcribed sequences in the interval 17q12-q21. Over Comparison of DLG3 with sequences present in the100 cDNAs were isolated, representing transcripts of known and EMBL and GenBank databases revealed significant ho-unknown genes, several of which have been reported elsewhere (Al- mology, at the amino acid level, to several members ofbertsen et al., 1994; Mazoyer et al., 1995; Smith et al., 1995). A clone,40F1, encoded part of the transcript of another novel gene, which the discs-large family of proteins. No homologies to anyhas not been previously described, and was used to isolate further known genes were observed at the nucleotide level, al-overlapping cDNAs. Inspection of the combined DNA sequence from though identity to three anonymous genomic DNA se-several clones did not reveal a candidate initiation codon, suggesting quences (T27219, T27220, and T27221) was noted (seethat the extreme 5' end of the cDNA remained to be isolated. RACE below). The greatest protein homology observed was(Frohman et al., 1988) experiments extended the sequence in the 5'direction for a further 100 bp and revealed a candidate ATG located to human erythrocyte p55, with which it shared 33%close to the 5' end of the sequence. Careful rescreening of cDNA identity and 56% similarity. Alignment of the predictedlibraries failed to identify any additional clones, and thus we sought amino acid sequence of the DLG3 protein with otherto isolate further sequence information from genomic DNA directly. members of the discs-large family using the PILE-UPAnalysis of subclones generated from the P1 phage, 124D3, yielded program revealed that DLG3 contains a single DHR,a further 500 bp of 5' sequence that confirmed the presence of the in-frame ATG and identified stop codons in all three frames upstream. an SH3 motif, and a guanylate kinase domain (Fig. 2).

Analysis of DLG3 gene expression. A blot of RNAs extracted from The DHR in DLG3 is, however, not well conserved: itdifferent tissues was purchased from Clontech and hybridized as shares only 30% identity to the third repeat in dlg, and

27 54 81 108AGC TTC CAC 0CC OCT OTT TCT TCA TCA ATA AAA TOG GAA CAT TCO CAG TCA TTr CTC AGG ACT TTO TOC AGA CAA AAA GTT TAT CAT AGA AAT CAC CCC ACA CAA TOO

135 162 189 216TTG OCA CAT COA OGA CAC TCA OCA CAT CTT AGO TTT ATT TCC T'rC CTC TOT OAA AAA AGT AAO CAG AGT TAT TTG TTC CCA TCC TCT CTO TCC ACA 0CC TOO I'CT CAG

243 270 297 + 324AGA ACA CAG TAO ATO TOT AGT CAG TTT GTT CAA TAA ATA TOC TTO ATG GAG CTT CAA TAC CCA CCT CCA CCC CCT TCT NCC AGA ATC TOO AGO GAG AGO TCG GGA GOT.

351 4 378 405 432GAC AAC 0CC AGC ATO CCA GTG OVA TCG GAG GAC TCT GOT TTO CAT GAA ACC CTG 0CC OTG CTG ACC TCC CAG CTC AGA CCT GAC TCC AAC CAC AAG GAG GAG ATG 000

bMT Pro Val Lou Ber Ulu ASP Bar Ply Lou Rio Ulu Thr Lou Ala Lou Lou Thz Sex Pin Lou Arp Pro Asp Bar Asp His Lye Ulu Ulu Not Ply

459 4 486 513 540TTC CTG AGO OAT OAT TTC APT GAA AAA AGO OTC AOT TAC TTA AVG AAG ATT CAT GAO AAG CTT COC TAT TAT OAA AGO CAA AOT OCA ACC CCA OTT CTO CAC AOC OCTPh. Lou A~rq Asp Asp Ph. gar Ulu L4's Bar Lou Bar Tyr Lou Nat Lys I1. His Ulu Lys Lou Asp Tyr Tyr Ulu Asp Oln Bar Pro Thr Pro Val Lou His Bar Ala

4 567 594 621 648OVO 0CC CTC OCT GAG GAC OVO ATO GAG GAO TTG CAUG0CC GCC TCC GTG CAC AGT OAT GAG AGO GAG CTG CTC CAG CTG CTO TCC ACC CCO CAC CTG AGO 0CT OTO CTCVol Ala Lou Ala Ulu Asp Val Nat Ulu Ulu Lou Oln A1La Ala Bar Val His Bar "sp Glu Aspg Ulu Lou Lou aln Lou Lou Bar Thrh Pro His Lou Asp Ala Val Lou

675 702 729 756ATG OVA CAT GAO ACG OTT 0CC CAG AAG AAT TNr GAO CCC OTT CTC CCO CCV CTG CCV GAO AAT ATC OAT GAO OAT TTT OAT GAO OAA TCO OTO AAG ATC OTC COO TTOHot Val Him Asp Thr Vol Als. Oin Lys mAs Ph., Asp Pro Val LOU Pro Pro Lou Pro Asp Aso 21. Asp Ulu Asp Ph. Asp Ulu Ulu gar Val LYS r o s o

4 783 810 837 9OTO AAO AAO AAG GAA CCC CTO GOT 0CC ACC AVC COO COG GAC GAG CAC VCA 000 OCT OTT OTO OTO 0CC AOO ATC ATO CGA OGA 000 GCA GCA GAC AOG AGO GGC OTO

891 918 U972GTC CAC OTT GOA GAT GAG CTC OGA OAA GTO A.AC GOG ATC OCA OTO CT0 CAC AAG COO CCC GAC GAG ATO AGC CAG ATT CTO 0CC CGO TCC CAP OOA TCC ATC ACC OVA

999 U1053 1080AAA ATC ATO CCA 0CC ACC CGO GAG OAA OAT COO TTA AAO GAO AGC AAO OVO TTC ATO 000 0CC CTC TTC CAC TAC AAO OCT COG GAO GAO COO 0CC ATC COT TOO CAGLys I1. Ile Pro Ala mxr Gln Ulu (31u Asp Arp Lou Lys Ulu Bar Lys Vol Ph. Mat Arg Ala Lou Ph. Hi. Tyr Amn Pro ASP Plu Asp Ara Ala Il. Pro Cy. Pl1

1107 1134 1161 1188GAG 000 000 CTO COO TTO CGO CGC AGO CAG OTC Cr0 GAG OTO OTO AGO CAG GAO GAO COO ACO TOO TOO CAG 000 AAG OGA OTO 000 GAO ACC AAC OTT OGA 0CC GOSUlu Ala Gly Lou Pro Ph. Gln Avg Asp Gln Val Lou Plu Vol Val Bar Oln Alt) Asp Pro Thr Trp Tro Gin Ala LYS Asp Val Gly ASP mxr Aso Lou Asp Ala Gly

a 1242 1269 4 1296OTO ATC CCC TOO AAG 000 TTO CAG GAO AGA OGA OTA AGO TAO COO AGA GCC GCO 000 ACC OVO CCO AGC CCC CAG AGC OVO AGO AAO CCC CCC TAT OAT CAG COT TOTLou Ile Pro Bar Lys Ply Phe Gln Glu AXq Asp Lou Bar Tyr Asp Asp Ala Ala Ply mxr Lou Pro Bar Pro Gln gar Lou Arg Lys Pro Pro Tyr Asp Gln Pro Cys

U1323 U1377 1404GAO AAA GAO ACC TOT GAO TOT GAG 000 TAO OTO AAA 000 CAO TAT OTO GOT GOT OTT COO AGO AGC TTC COO OTO 000 TOT AGO GAG AGA OTO GOT GOC TOG CAG GAAAsp Lys Plu mxr Cya AsD Cy. Plu Ply Tyr LOU Lys Ply His Tyr Val Ala Ply Lou Asp Asp Bar Ph* Asp Lou Ply Cy. sP P~glu Asp Lou Ply Ply Bar Oln Plu

1431 1458 1485OOA AAG ATO TOO TOC GGA OCT GAG TOT COG GAO Cr0 OTO ACT TAO GAA GAG GTO 000 AGO TAO CAA CAC CAG 000 GOA GAG COG CCC COO OTO OVO OTT OTO ATO 000Ply Lys met gar Bar Ply Ala Plu Bar Pro Plu Lou Lou mxr Tyr Plu Ulu Val Ala Asp Tyr Plo Hi. Plu pro Ply Plu A~rp Pro Ara Lou Vol Vol Lou I1 aPly

1539 1566 a 1620TOT OTO OGA GCC OOA Cr0 CAC GAG Cr0 AAO CAA AAG OTO OTO OCT GAG AAC OCA CAG CAC TTT 000 OTO OCT OTT OCA CAT ACC ACC AGO COO OGA AAO AGO OAT GAOBar Lou Gly Ala Ara Lou His Olu LoU LYS Pln Lv. Vol Vol Ala Ulu Ann Pro Pln Hi. Ph* Ply Vol Ala Vol Pro Hi. Thr Thr ArsP Pro Ara jLV Br His Ulu

1647 1674 U 1701 1728AAG OAA OGA OTO GAA TAT CAC TTT GTG TOT AAG CAA GCA TTT GAG 000 GAO TTA OAT CAC AAC AAO TTO Cr0 OAA CAT OGG GAA TAT AAG OAA AAT Cr0 TAT OGA ACC

1755 1782 1809 1836A00 Cr0 GAO GCC ATT CAG OCT OTT ATO 000 AAA AAO AAA OTT TOT TTO GTO OAT OTO GAO OCA OAA OCA Cr0 AAA OAA OTO AGO ACC TCA OAA TTT AAA COO TAT ATT

1863 1890 1917 1944ATA TTT OTA AAO OCT OCA ATT CAO OAA AAA AGA AAA ACC OCA COT ATO TOO OCA GOT TOT GAO GAO ACA 00A 000 OCA TTT OAT GAG CAG CAG CAA GAG ATO 000 OCT21. Ph* Val LYS pro Ala Ile Gin aim LY. asP ILYA Thr Pro Pro met 3aroAaOv , App -M- -1- Ala Pr Ph S l l l l l a l l

1971 1998 2025 2052TCT 000 000 TTO ATA GAO 000 OAT TAO 000 CAC Cr0 OTA GAO 000 OTO Cr0 OVO AAO GAO OAT OTO CAG OOT 000 TAO AGC CAG CTC AAA OTO OTO TTA GAG AAO Cr0Bar Ala Ala Ph. I1. Azg Aspj Hi, Ty? Ply His Lou Val Asp Ala Vola o l LIPuApLuPoPyAaTBrPoL L.Val Vol Lou Plu Lys Lou

2079 2106 2133 2160AGO AAG GAO ACT CAO TOO OVA COT OTT AGT TOO OTO AGO TAA OTT TAT 000 AGA ACA TCC AAG Cr0 GAO 000 ACC TTG AAO ATC ATO TAO TOO AGA Cr0 OCr CAT ?1Tgar Lys Asp Thr Hi. Trp Val Pro Val Bar Tsp Vol Asp STOP

2187 2214 2241 2268ACO ATO AAG OAA TOT CAA GCG CAG AGA 000 AGA GAA TTO TOO ACA AAT TOO ATO ATO GAG AAO ACT ATA AGT 000 AAO TCT TOT TTO TTO VTO OTT TTV OTO TOT TOT

2295 2322 2349 2376TTT TCA OTO CAC Mr TTT OGA TOA VGA TTT GAA AGO 000 ATA TCA GAA AAO AAO ACA TTT CAT TTA TTA AAG TAT CAO AGO CAA OCT GAO COT OAT TOT TTG TAO CAA

2403 2430 2457 2484AOT TAA OVA 0CC ACT OTO TTT TOT 000 TOO TAO TOO TTA ATT TAT ACA OVA Cr0 ATT 000 AGA ATO TTT AAG OTT TTT AAA CAT AOT GAO OCT TAO TAG TTT TTT TOO

2511 2538 2565 2592AAG OVA ACT TOT TTT ATO CAO 000 OAT TTT ACA TOT AAC TGA AGT TOO OCT OTO TTO AAO CAO TAA AAO OTT OAT OTT AAO OTT TTT TTT OAA OTO OTT 000C TOG TAA

2619 .2646 2673 2700TAO AAA AGO GOT TOT OTG COT ATT TTA AAA TAO TGA ATA TAO OTA AAT VTT CrC TOO AArGOCT GAO GCA CAC TTO ACO ATO AAO ATO AAT TAO TOT ACT ATO OTO TAO

2727 2754 2781 2808TOO AGV GOT 0CC TTO AGO GAO TOO AGO AAT OTA AGO TTG COT TTC COO TTT OVA AAT ACC OTO AGA TVO OVA ACA TOG AOC OCA TOO TTT OTT VGA In- OTT CTA TTC

2835 .2862 2889 2916CAT OCA TTG TOO OTT TTG TTA Cr0 ACA OTT 000 TTO OTO OTA 000 AOT COO TOC CAT GAO ATO ATA 000 OTT CCC ATT OTO OTA OAT OTT 000 AAA OCA OAT GAO TOT

2943 2970 2997COO TOT CAA AAC TAT GOC TAO OTO ACT OTA AAO OAT TTO TOT CAA GAA TA.A AAO TAT OTA GAO CCA GAO TOT 000 COT AAA AAA AAA AAA AAA "AA

FIG. 1. Nucleotide sequence and predicted protein translation of the DLG3 gene. The dashed line indicates the position of the DHR, thesolid line indicates the SiE3 domain, and the double underline indicates the GK domain. The vertical arrows in the coding sequence indicate thepositions of exonic boundaries, and the black diamond identifies the point 5' from where the sequence was derived from genomic DNA.

148 SMITH ET AL.

1 t1ood1g2 MPVAATNSETAMQQVLDNLG. SLPSETGAAELDLMFLRGIMESPIVRSLAKVIMVLWFMQQNVFVPM. KYMLKYFGAHERLEETKLEAVRDNNLELVQEIp55 MTLKASEGESG ........ G.SMHT .............................................................. ALSDLYLE .....

d1g3 MPVLS, . EDSGLHETLALLT. SQLRPDSNHKEEMGFLRDDFSEKSLSYLMKIHEKLRYYERQSPTPV. LH .............. SAVALAEDVMEEL...dlgl_drome . . . SASASVIASNNTISNTTVTTVTATATASNDSSKLPPSLGANSSISISNSNSNSNSNNINNINSINNNNSSSSSTTATVAAATPTAASAAAAAASSPP

101 -- DHR --d1g2 LRDLAQLAEQSSTAAELAHILQEPHFQSLLETHDSVASKTYETPPPSPGLDPTFSNQPVPPDAVRMVGIRKTAGEHLGVTFRVEG . VIARILHGGMp55 ..... HLLQKRSRPEAVSHPL ............ NTVTEDMYTNGSPAPGSPAQVKGQEVRK.. VRLIQFEKVTEEPMGITLKLNE. KQSCTVARILHGGM

dlg3 ..... QAASVHSDERELLQLLSTPHLRAVLMVHDTVAQKNFD.. PVLPPLPDNI. DEDFDEESVKIVRLVK. NKEPLGATIRRDEHSGAVVVARIMRGGAdlgldrome ANSFYNNASMPALPVESNQTNNRSQSPQPRQPGSRYASTNV. LAAVPPGTPRAVSTEDITREP.RTITIQK.GPQGLGFNIVGGEDGQGIYVSFILAGGP

GLGF201 300

dlg2 VAQQGLLHVGDIIKEVNGQPVGSDP.RALQELLRNASGSVIL .... KILPtNYQ. EPHLPR..Q ...................... VFVKCHFDYDPARDSLIp55 IHRQGSLHVGDEILEINGTNrrNHSVDQLQKAM!ETKGMISL .... KVIPN.Q. QSRLPALQ ...................... MFMMRAQFDYDPKKDNLI

dlg3 ADRSGLVHVGDELREVNGIAVLHKRPDEISQILAQSQGSITL .... KI IPATQEEDRLKESK ...................... VFMRALFHYNPREDRAIdigl drome ADLGSELKRGDQLLSVNNVNLTHATHEEAAQALKTSGGVVTLLAQYRPEEYNRFEAR IQELKQQAALGAGGSGTLLRTTQKRSLYVRALFDYDPNRDDGL

301 <-- SH3 -9 400d1g2 PCKE.AGLRFNAGDLLQIVNQDDANWWQACHVEGG .... SAGLIPSQLLEEI-K .... AFV ........ KRDLELTP ...... NSGTLCG..SL .......p55 PCKEAGLKFATGDIIQIINKDDSNWWQG.RVEGSS.KESAGLIPSPELQEWRV .... ASM ........ AQS... .AP ...... SEAPSCS..PF .......

dlg3 PCQEAGLPFQRRQVLEVVSQDDPTWWQAKRV. GDT. NLRAGLIPSKGFQERRL .... SYRRAAGTLPSPQSLRKPPYDQPCDKETCDCE.. GYLKGHYVAdlgidrome PSR.. GLPFKHGDILHVTNASDDEWWQARRVLGDNEDEQIGIVPSKRRWEPRMRARDRSVKFQGHAAANNNLDKQSTLDRKKKNFTFSRKFPFMKSRDEK

401 500d1g2 SGKKKKRMMY ........ LT ...... TKNAEFDRHE.. LLIYEEVARM ... PPFRRKTLVLIGAQGVGRRSLKNKLIMWDPDRYGTTVPYTSRRPKDSERp55 GKKKKYKDKY ........ LA ...... KHSSIFDQLD..VVSYEEVVRL. .. PAFKRKTLVLIGASGVGRSHIKNALLSQNPEKFVYPVPYTTRPPRKSEEd1g3 GLRRSFRLGCRERLGGSQEG ...... KMSSGAESPE.. LLTYEEVARYQHQPGERPRLVVLIGSLGARLHELKQKRVAENPQHFGVAVPHTTRPRKSHEK

dlgl_drome NEDGSDQEPNGVVSSTSEIDINNVNNNQSNEPQPSEENVLSYEAVQRLSI..NYTRPVIIL .. . GPLKDRrNDDLISEYPDKFGSCVPHTTRPKREYEV

GxxGxGK TT~xxRxxEx

501 -- GK --d1g2 EGQGYSFV. SRGEMEADVRAGRYLEHGEYEGNLYGTRI DSIRGWAAG KVCVLDVNPQAVKVLRTAEFVPYVVFXEAPDFETLRAMMNRAALESGISTKQL

p55 DGKEYHFI. STEEMTRNISANEFLEFGSYQGNMFGTKFETVHQIHKQNKIAILDIEPQTLKIVRTAELSPFIVFIAPTDQGT ................. Qd1g3 EGVEYHFV. SKQAFEADLHHNKFLEHGEYKENLYGTSLEAIQAVMAKNKVCLVDVEPEALKQLRTSEFKPYIIFVKPAIQEKRKTPPMSP. .ACEDTAAP

dlgl_drome DGRDYHFVSSREQMERDIQNHLFIEAGQYNDNLYGTSVASVREVAEKGKHCILDVSGNAIKRLQVAQLYPVAVFIKPKSVDSVMEMNR .......... RMxGxxY

659dlg2 TEADLRRTVEESSRIQRGYGHYFDLCLVNSNLERTFRELQTAMxEKLRTEPQWVPVSWVYp55 TEA. LQQLQKDSEAIRSQYAnYFDLSLVNNGVDETLKKLQEAFDQACSS PQWVPVSWVY

dlg3 FDEQQQEMAASAAFIDRHYGHLVDAVLVKEDLQGAYSQLKVVLEKLSKDTHWVPVSWVRdlgldrome TEEQAKKTYERAIKMEQEFOEYFTGVVQGDTIEEIYSKVKSMIWSQSGPTIWVPSKESL

FIG. 2. Comparison of the predicted amino acid sequences of the p55, DLG2, DLG3, and Drosophila dig gene products. The putativeamino acid sequences are complete with the exception of dig, for which the N-tenninal 325 residues were deleted for clarity. The sequenceswere compared using the PILEUP program (Wisconsin Sequence Analysis Package Version 8.0) with a gap weight of 3.6 and a lengthweight of 0.30. The shading indicates the locations of the DHR, SH3, and GK domains. The sequence "GLGF" indicates the position of thismotif in Drosophila dlg (see text); GxxGxGK and TTRxxRxxExxGxxY are consensus sequences (where x is any amino acid) marking thepositions, in the GK domain, of the "anion hole" and the "GMP-binding motif," respectively.

the sequence motif, GLGF (Cho et al., 1992), which is the inclusion of DLG3 as a new member of the discs-conserved in dig, hdlg, and PSD-95, is absent in DLG3. large family seems reasonable based upon the func-The N-terminal region of the DLG3 protein most re- tional domains that it contains, which are common tosembles p55; both proteins contain only one DHR all discs-large proteins.(whereas the other members contain three), and the Analysis of DLG3 gene expression revealed that theDHR in both proteins is poorly conserved compared transcript is present in all tissues tested, except forwith dlg. In contrast, the SH3 motif appears to be peripheral blood leukocytes (Fig. 3). At least four tis-highly conserved between all members of the discs- sue-specific transcripts were identified, ranging fromlarge family, including the DLG3 gene product. Two about 2 to 4.5 kb. Thymus contained approximatelyregions of the guanylate kinase domain deserve close equal proportions of two transcripts of 3.5 and 4.5 kb,inspection, the "anion hole" and the "GMP-binding site" while spleen, prostate, ovary, small intestine, and colon(Koonin et al., 1992; Willott et al., 1993). The anion contained an additional (though frequently poorly ex-hole comprises the signature GxxGxGKST, which is pressed) 2.5-kb transcript. In testis only, a 2-kb mRNAwell conserved in p55, but contains a deletion in dlg, was identified (in addition to the three other tran-hdlg, and PSD-95. In DLG3, the spacing is correct, but scripts) and was the most abundantly expressed tran-the GKST residues are not retained, and thus it is un- script in this tissue. Since the largest transcript identi-clear whether the putative DLG3 protein is a kinase. fled was about 4.5 kb, and the length of the cDNA fromThe GMP-binding site contains the consensus se- the presumed ATG start codon to the poly(A) tail isquence, TTRxxRxxExxGxxY, which is conserved in all only 2650 bp, it is not certain that the complete codingdlg-related proteins with the exception of ZO-1 and ZO- sequence has been identified. RACE experiments and2 and is present in DLG3, although the second arginine rescreening of libraries did not extend the cDNA se-residue is replaced by a lysine, as it is in p55. Thus, quence further in the 5' direction, and thus the length

A SECOND DISCS-LARGE GENE (DLG3) ON 17q12-q2l 149

with respect to each other is not known. Finally, theV. observation that three genomic sequences present in

.- the EMBL and GenBank databases (EMBL Accession"• E . •-Nos. T27219, T27220, and T27221), which are also de-"•: • .• "= "rived from 17q12-q21, share identity with small re-

gions of the DLG3 cDNA sequence supports the local-ization of the DLG3 gene on chromosome 17.

Comparison of the predicted DLG3 protein with that-' of DLG2 showed that they contain similar structural

"domains: both proteins contain a single DHR, an SH3- imotif, and a guanylate kinase domain containing a

highly conserved GMP-binding site. Overall, the aminoacid sequences of the two proteins share 36% identityand 58% similarity, which is close to that observed forboth DLG2 and DLG3 compared individually with p55.The relatedness of DLG2, DLG3, and p55 suggests acommon functionality, possibly as "linker proteins" im-portant in coupling the cytoskeleton to the cell mem-brane.

Another feature of the DLG2, DLG3, and hdlg pro-teins (though not p55) is that they are each predicted

FIG. 3. Analysis of the expression of the DLG3 gene in different to start with the same three amino acids, MPV. Whiletissues. A blot of RNAs extracted from various tissues was hybridized the function of the MPV signature is not known andwith a radiolabeled DNA fragment from the 40F1 cDNA that con- the sequence match is unlikely to be coincidental, thetained most of the coding sequence of DLG3. No bands were identifiedin the lane containing RNA from peripheral leukocytes; hybridization presence of this motif at the start of the DLG3 proteinof the same blot with an actin probe indicated that each lane was suggests that the translational start site of the DLG3loaded with an equivalent amount of mRNA. gene has been correctly identified. The presumed ATG

start codon also occurs within the consensus Kozakof the 5'-untranslated region (UTR) is not known. In environment, GCC(A/G)CCATGG, which is found atany case, it is unlikely that the 5' UTR would account the translational start site of most vertebrate genesfor the discrepancy between the size of the transcript (Kozak, 1991). At the C-terminus, the p55 and DLG2and the length of the coding sequence. Other possibilit- proteins end with the sequence WVPVSWVY, while theies include alternative 3' polyadenylation sites that in- DLG3 protein terminates with a near-identical se-crease the length of the 3' UTR, or the ATG codon that quence in which the tyrosine residue is replaced by anwe identified as the translational start of the gene may arginine. Again, the function of this C-terminal motifin fact be downstream of the true initiator, although is not known, but its presence in the predicted DLG3this is unlikely (see below). Another explanation is that protein strongly indicates that the extreme 3' end ofother coding exons that are not expressed in fetal brain the coding sequence has been correctly obtained.exist, and thus were spliced out of the cDNA clones Analysis of the nucleotide sequences of the p55,that we had isolated. This explanation is all the more DLG2, and DLG3 genes using the BESTFIT programplausible given the observation that most tissues tested (Wisconsin Sequence Analysis Package Version 8.0),contain several different DLG3 transcripts, although which introduces gaps into the sequences to obtain op-it is not known which variants are expressed in fetal timal alignment, reveals that they show approximatelybrain. 60% identity to one another. The sequence homology

Recently, we reported another gene located on chro- among p55, DLG2, and DLG3 suggests that they belongmosome 17, DLG2, that also encodes a protein belong- to a "gene family," although it is not clear whethering to the discs-large family (EMBL Accession No. other family members exist and how the genes evolved.X82895; Mazoyer et al., 1995). The DLG2 and DLG3 However, the close physical proximity of DLG2 andgenes have both been mapped to the 280-kb YAC, DLG3 (p55 is located on the X chromosome (Metzen-853B3, whose location on chromosome 17q has been berg and Gitschier, 1992)) suggests that the two genesconfirmed by fluorescence in situ hybridization (FISH) evolved by a duplication event.(Albertsen et al., 1994). P1 clones containing the DLG2 Recent experiments indicate that the p55 gene prod-and DLG3 genes have also been isolated and mapped uct is involved in coupling the cytoskeleton to the redto 17q by FISH, although attempts at isolating clones cell membrane by forming a ternary complex with pro-that contain all or part of both genes have been unsuc- tein 4.1 and glycophorin C (Marfatia et al., 1994). Pro-cessful. The colocalization of DLG2 and DLG3 on tein 4.1 also binds to spectrin, a key component of the853B3 certainly suggests that they are located close intracellular protein network that underlies the eryth-together on the chromosome, although the precise dis- rocyte membrane. It is possible that the DLG2 andtance between them and the orientation of the genes DLG3 gene products have a function similar to p55 in

150 SMITH ET AL.

helping to maintain the integrity of the cell membrane Koonin, E. V., Woods, D. F., and Bryant, P. J. (1992). dlg-R proteins:

and its links with associated cytoplasmic proteins. Modified guanylate kinases. Nature Genet. 2: 256-257.Kozak, M. (1991). An analysis of vertebrate mRNA sequences: Inti-

mations of translational control. J. Cell Biol. 115: 887-903.ACKNOWLEDGMENTS Lue, R., Marfatia, S. M., Branton, D., and Chishti, A. H. (1994).

Cloning and characterisation of hdlg: The human homolog of theWe thank Margaret Robertson and Elizabeth Lawrence for DNA Drosophila discs-large tumour suppressor binds to protein 4.1.

sequencing and Ed Meenen for synthesizing oligonucleotides. S.A.S. Proc. Natl. Acad. Sci. USA 91: 9818-9822.is supported by an EMBO fellowship. This work was supported in Marfatia, S. M., Lue, R. A., Branton, D., and Chishti, A. H. (1994).part by Grant DAMD17-94-J-4129 from the Department of Defence In vitro binding studies suggest a membrane-associated complexto R.W. between erythroid p55, protein 4.1 and glycophorin C. J. Biol.

Chem. 269: 8631-8634.

REFERENCES Mayer, B. J., and Eck, M. J. (1995). SH3 domains. Minding your p'sand q's. Curr. Biol. 5: 364-367.

Mazoyer, S. M., Gayther, S. A., Nagai, M. A., Smith, S. A., Dunning,Albertsen, H. M., Smith, S. A., Mazoyer, S., Fujimoto, E., Stevens, A., van Rensburg, E. J., Albertsen, H., White, R., and Ponder,

J., Williams, B., Rodriguez, P., Cropp, C. S., Slijepcevic , .,Carl- B.A. J. (1995). A gene (DLG2) located at 17q12-q21 encodes a newson, M., Robertson, M., Bradley, P, Lawrence, , Harrington, T., homologue of the Drosophila tumor suppressor dlg-A. Genomics 28:Sheng, Z. M., Hoopes, R., Sternberg, N., Brothman, A., Callahan, 25-31.

R., Ponder, B. A. J., and White, R. (1994). A physical map and

candidate genes in the BRCA1 region on chromosome 17q12-21. Metzenberg, A. B., and Gitschier, J. (1992). The gene encoding the

Nature Genet. 7: 472-479. palmitoylated erythrocyte membrane protein, p55, originates atthe CpG island 3' to the factor VIII gene. Hum. Mol. Genet. 1: 97-

Altschul, S. F., Gish, W., Miller, W., Myers, E. W., and Lipman, 101.D. J. (1990). Basic local alignment search tool. J. Mol. Biol. 215: Noirot-Timothee, C., and Noirot, C. (1980). Septate and scalariform403-410.NortToteCadNioC(18)Settansclifm junctions in arthropods. Int. Rev. Cytol. 63: 97-140.

Berger, A., Schiltz, E., and Schulz, G. E. (1989). Guanylate kinase Pawson, T. (1988). Non-catalytic domains of cytoplasmic protein-from Saccharomyces cerevisiae. Isolation and characterisation, tyrosine kinases: Regulatory elements in signal transduction. On-crystallization and preliminary x-ray analysis, amino acid se- cogene 3: 491-495.quence and comparison with adenylate kinases. Eur. J. Biochem. cgn :4145184: 433-443. Rees, D. J. G., Ades, S. E.,' Singer, S. J., and Hines, R. 0. (1990).rat: brai. Sequence and domain structure of talin. Nature (London) 347:

Cho, K-O., Hunt, C. A., and Kennedy, M. B. (1992). The rat brain 685-689.postsynaptic density fraction contains a homolog of the Drosophiladiscs-large tumour suppressor protein. Neuron 9: 929-942. Ruff, P., Speicher, D. W., and Chishti, A. H. (1991). Molecular identi-

fication of a major palmitoylated erythrocyte membrane proteinFrohman, M. A., Dush, M. K., and Martin, G. R. (1988). Rapid produc- containing the src homology 3 motif. Proc. Natl. Acad. Sci. USA

tion of full length cDNAs from rare transcripts: Amplification using 88: 6595-6599.a single gene-specific oligonucleotide primer. Proc. Natl. Acad. Sci. Smith, S. A., Holik, P., Stevens, J., Melis, R., White, R., and Al-USA 85: 8998-9002. bertsen, H. (1995). Isolation and mapping of a gene encoding a

Gould, K L., Bretscher, A., Esch, F. S., and Hunter, T. (1989). cDNA novel human ADP-ribosylation factor on chromosome 17q12-q21.cloning and sequencing of the protein-tyrosine kinase substrate, Genomics 28: 113-115.ezrin, reveals homology to band 4.1. EMBO J. 8: 4133-4142. Stewart, M., Murphy, C., and Fristrom, J. (1972). The recovery and

Jesaitis, L. A., and Goodenough, D. A. (1994). Molecular characteri- preliminary characterisation of X chromosome mutants affectingsation and tissue distribution of ZO-2, a tight junction protein imaginal discs of Drosophila melanogaster. Dev. Biol. 27: 71-83.homologous to ZO-1 and the Drosophila discs-large tumour sup- Willott, E., Balda, M. S., Fanning, A. S., Jameson, B., Van Itallie,pressor protein. J. Cell Biol. 124: 949-961. C., and Anderson, J. M. (1993). The tight junction protein ZO-1

Kistner, U., Wenzel, B. M., Veh, R. W., Cases-Langhoff, C., Garner, is homologous to the Drosophila discs-large tumour suppressorA. M., Appeltauer, U., Voss, B., Gundelfinger, E. D., and Garner, proteins of septate junctions. Proc. Natl. Acad. Sci. USA 90: 7834-C. C. (1993). SAP90, a rat presynaptic protein related to the prod- 7838.uct of the Drosophila tumour suppressor gene dig-A. J. Biol. Chem. Woods, D. F., and Bryant, P. J. (1989). Molecular cloning of the lethal268: 4580-4583. (1) discs-large (1) oncogene of Drosophila. Dev. Biol. 134: 222-235.

Koch, C. A., Anderson, D., Moran, M. F., Ellis, C., and Pawson, T. Woods, D. F., and Bryant, P. J. (1991). The discs-large tumour sup-(1991). SH2 and SH3 domains: Elements that control interactions pressor gene of Drosophila encodes a guanylate kinase homologof cytoplasmic signaling proteins. Science 252: 668-674. localised at septate junctions. Cell 66: 451-464.

SHORT COMMUNICATION

Isolation and Mapping of a Gene Encoding a Novel HumanADP-Ribosylation Factor on Chromosome 17q12-q21

SiMON A. SMITH, 1 PILAR R. HOLIK, JEFF STEVENS, ROBERTA MELIS, RAY WHITE, AND HANS ALBERTSEN

Huntsman Cancer Institute, University of Utah, Building 533, Suite 7410, Salt Lake City, Utah

Received December 12, 1994; accepted April 19, 1995

represents a distinct ARL isoform that maps to chromo-A gene encoding a low-molecular-weight GTP-bind- some 17q12-q21.

ing protein was isolated from a retinal cDNA library A retinal cDNA library (Stratagene 937202) wasand mapped to human chromosome 17q12-q21. Corn- screened by hybridization with P1 phage clones locatedpa Ason of the predicted protein with the protein data- on 17q12-q21. One of several positive clones thatbases revealed striking homology to the family of ADP- mapped back to chromosome 17q, 16RB1, contained anribosylation factors (ARFs), which are thought to be insert of 1376 bp and had a short poly(A) tail at one end.involved in membrane trafficking and protein secre- Sequence analysis revealed a single long open readingtion. The greatest homology observed was with the rat frame of 603 bp that was predicted to encode a proteinARF-like 4 protein (ARL4), with which it shared 58% that has a molecular weight of 22.3 kDa (Fig. 1). Com-identity, while the more highly conserved human parison of the nucleotide sequence of the 16RB1 cloneARF1 and ARF3 proteins each shared 46% identity. In- with the n d sequence ui the BLASTspection of the predicted new protein showed that it with the EMBL and NCBI databases using the BLASTcontained each of the six conserved motifs that are algorithm revealed identity to the partial sequence ofrequired for guanine nucleotide binding and hydroly- a human cDNA clone, c-29a01 (Accession Nos. Z40585sis, and thus it is probably a novel ARF isoform. We and Z44776), while the predicted amino acid sequencehave designated the new protein and its correspond- revealed striking homology to ARF and ARF-like pro-ing gene ARF4L. © 1995 Academic Press, Inc. teins from several different species. The greatest pro-

tein homology observed was with rat ARL4, which

showed 58% identity and 78% similarity (Fig. 2). InADP-ribosylation factors (ARFs) are a subfamily of contrast, the highly conserved ARF proteins were less

low-molecular-weight GTP-binding proteins that are homologous to the predicted new protein: human ARF1thought to be involved in membrane trafficking and and ARF3 showed 46% identity to the new isoform.protein secretion (2, 4, 6). Each protein has a predicted Since the predicted new protein meets the criteria formolecular weight of approximately 20 kDa and con- belonging to the ARF family of GTP-binding proteinstains six distinct domains that are required for nucleo- (see below) and the protein appears to represent a pre-tide binding and hydrolysis (7). Three domains (PM1- viously unknown isoform, it has been designated the3) are required for phosphate/magnesium binding, name ARF4L.while the other three domains (G1-3) are required for Inspection of the predicted amino acid sequence ofguanine nucleotide binding. The genes for at least six the ARF4L protein reveals that it contains each of thehuman ARF isoforms have been identified (3); compari- motifs required for guanine nucleotide binding and hy-son of the deduced amino acid sequences indicates a drolysis (Fig. 2). While the motifs themselves matchhigh degree of sequence conservation. Mouse homologs the consensus very closely (with the exception of G3,o" 'he human genes have also been isolated, and two which appears most like the human ARL2 isoform), therat genes, encoding ARF-like (ARL) proteins ARL1 and spacing between the PM2 and the PM3 domains inARL4, have been identified (5). The predicted ARL1 ARF4L is especially noteworthy, since it is similar toand ARL4 proteins also contain the six conserved mo- that in rat ARL4: both of these proteins contain antifs required for GTP-binding and hydrolysis, but share extra five amino acids that are absent from the otheronly 55 and 41% identity to ARF1, respectively. Here, ARF and ARF-like proteins. In common with the otherwe report the cloning of a novel human ARF gene that members of the ARF subfamily, ARF4L contains a gly-is most homologous to rat ARL4, but that probably cine at position 2 and does not contain any cysteine

residues in the C-terminus of the protein, which are

Scquence data from this article have been deposited with the common in other GTP-binding proteins. Thus, the in-EMBL/GenBank Data Libraries under Accession No. L38490. clusion of the ARF4L protein as a member of the ARF

'To whom correspondence should be addressed. Telephone: (801) subfamily seems reasonable based upon its alignment585-6224. Fax: (801) 585-3833. with known members of this family.

113 GENOMICS 28, 113-115 (1995)0888-7543/95 $12.00

Copyright © 1995 by Academic Press, Inc.All rights of reproduction in any form reserved.

114 SHORT COMMUNICATION

The precise functions of the ARF and ARF-like pro- rARL4 MGNGLSDCTSILSSLPSFSFXVILGLDCAGKTTVLYR.QF!8PNVTV 49S L P FQ. H-11-+GLD AG;KT,+LYRI,.F EFV +V

teins are unknown. Several studies have indicated that hA R4 L MGNHLTEMAPTASSFLPHFQALHVVVIGLDSAGKTSLLYRLKFrEPVQSV 50- -+-(GLD÷AGKT- LYýLK E -+

some of the ARF genes show tissue-specific expression. hARF1 MGNIFANLFKGLFGPEMRILMVGLDAAGKTTILYrLKLGEIVTTI 46

For instance, ARF5 is expressed predominantly in P1 G1

brain, and ARF4 is expressed in adipose tissue but notrARL4 PTKGFNTEKIKVTLGNSKTVTFHFWDVGGQEKLRPLWKSYTRCTDGIVFV 99

PTKGFNTEFK:V LG S- CTF WDVGCQEKLRPLWrSY R TDG+VFV

hARF4 L PTKGFNTEKIRVPLGGSRGITFQVWDVGGQKLR PLWRSYNRRTDGLVFV 10028 P55i rIVoVc~rPu90 LPgc cagg ttct 9999 ctcPT GFN E , , 1÷F VWDVCGQýKRPLWR Y" + T GL.FV

gca cog gog 181 ctg cgg 800 tnt 9c ggg ocg gct 9cg gtg tt- ac egg gga hAPFl PTIGFNVETVZY ----- KNISFTVIDVGGODKIRPLWRHYFQNTQGLbPV 91

82 109 PM2 PM3aag gct cga gga gag cgc ggc tca cga gag ata acE cag ctg tgc tcc ctg gaa

136 163cnt tca alt tca agg cct ccc tgc Ctc tac tag gcg cct tag Ctc act ATG GGG rARL4 VDSVDVERMEEARTELHKITRISENQGVPVLCVANKQDLRNSLSLSEI8K 149

MET GCy VD- + LR.EOAK 5LH-r R SrNQCVPV1/LA1KQD ÷LS +E-EK

190 217 hAPF4L VDAAEAER3.LEAXVELHRISRASDNQGVpVLVLANK1ODOPGALSAAEVEK 150AAC CAC TTG ACT GAG ATG GCG CCC ACT GCC TCC TCC TTC TTG CCC CAC TC CACA- V ER, EA, E' R+ + LV ANKQD P A+-AAE+Ash His Leu Thr Gl Met Ala Pro Thr Ala Ser Set Phe Leu Pro His Phe Gin hARFl V)SN RERVNiAREELMRMLARDELRDAVLLVFANKODLPNAMNAAEXTD 141

G2244 271

GCC CTG CAT GTC GTG GTC ATT GGG CTG GAC TCT GCT GGA AAG ACC TCC CTC CTTAla Leu His Val Val Val le Gly Leu Asp Ser Ala Gly Lys Thr Ser Leu Leu

rAOL4 LLAMGELS SS TPW38LQPTCA I I GGLKZGLEKLHDMS I KPRDKMQQKKK 199

298 325 LA, EL+- T H.Q A. G GL-CLE-L.-MILKRK R KK+TAC CGS CTC AAG TTC AAG GAG TTT GTC CAG ACT GTC CCC ACC AAA GGC TTC AAC hARF4L RLAVRELAAATLTHVQGCSAVDGLGLQQGLERLYEMILKRKXAARGGKJIR 200Tyr Arg Leu Lys Phe Lys Glu Phe Val Gin Set Val Pro Thr Lys Giy Phe Asn +L + L +-Q A G GL C•. L . +K

hARF1 KLGLHSL-RHRNWYIOQATCATSGDGLYEGLDWLSNQLRNOK 181352 379

ACC GAG AAG ATC CGG GTG CCC CTC GGG GGA TCG CGT GGC ATC ACC TTC CAA GTC G3Thr Glu Lys Ile Arg Val Pro Leu Gly Gly Ser Arg GCy 1le Thr Phe Gln Val

rARL4 R 200456 433

TGG GAC GTC GGG GGG CAG GAG AAG CTG CCA CCA CTG TGG CGC TCT TAT AAC CGC C

Trp Asp Val Gly Gly Gin Glu Lys Leu Arg Pro Leu Trp Arg Sec Tyr Asn Arg hARF4L R 201

46C 487 FIG. 2. Comparison of the predicted amino acid sequence of theCoG ACA GAC CGT CTA GTG TTT T GTG Ga AC GCT GCG GAG GCT GAG CGG CTT GAArg Thr Asp Gly Leu Val Phe Val Val Asp Ala Ala Glu Ala Glu Arg Len Glu human ARF4L protein with rat ARL4 and human ARF1. The de-

514 541 duced amino acid sequence of ARF4L was aligned with ARL4 andGM GCC MG CCC GAG CCC CAC CCA ACC ACC CCC GCC TCC GAC MC CAC CCC CCC ARF1 with the help of the BLAST algorithm. Hyphens representClu Ala Lys Val Glu Leu His Arg Ile Ser Arg Ala Ser Asp Asn Gin Gly Val gaps introduced into the sequences for optimal alignment; + indi-

568 ------------- cates a conservative substitution. The consensus motifs presumed toCCA GTS CTG GTG CTC GCC AAC AAG CAG CAC CAG CCC GGG GCA CTG AGC GCT GCTPro Val Leu Val Leu Ala Asn Lys Gin Asp Gin Pro Gly Ala Leu Sen Ala Ala be involyed in phosphate/magnesium binding (PM1-3) and guanine

-S2- - 622 649 nucleotide binding (G1-3) are underlined in the ARF1 sequence.GAG GTC GAG AAG ACG CCC GCA GTC CGA GAG CTA GCA GCC GCC ACT CTC ACT CATGiu Val Glu Lys Arg Leu Ala Val Arg GIu Leu Ala Ala Ala Thr Leu Thr His

676 ----- 82AR--- in brain, heart, or muscle (5). The rat ARL4 gene, whichCCC CAA CCC CCC 0CC CCC CCC GAC CCC CCC CCC CCC CAG COG CCC CCC GAG CCCVal Gin Gly Cys Sen Ala Val Asp COY Len Gly Len Gin GIn Gly •e• Giu Org shows greatest homology to ARF4L, is not expressed------------- 730 757 in undifferentiated fibroblasts, but is abundant in dif-CTC TAT GAG ATC ACC CTC AAG AGG AAG AAG GCA CCT CGG GGT GGC AAG AAG AGALeunTyr Gin Met SIl Len Lys Org Lys Lys Ala Ala Aro Gly COY Lyn Lys A• g ferentiated cells with an adipocyte-like phenotype (5).

784 811 This may suggest that the function of ARL4 is relatedCCC TA• ... ag .E .C tC t c cSTtP a a t ct .... ac to the differentiation state of fibroblasts, although theOrg SCGP

838 ARL4 gene aiso ato be in heart,tgg aca oca ggg tOO gac cag cct gtg acc tct cag tca gac tgo ggt gca gga appears expressed brain,

892 9i9 and muscle. At the moment, therefore, it is difficult tocct gtc cac ctc nat gaa gga gag agg agc atg ggg tgt ccc gtt ttg gtg ccao46 aot speculate on the function of the human ARF4L protein

can tgo oot g00 Oat 000 aoa too at otc ttt oca tat Etc tct cat Ccl tc. without further insight into the functions of the other1oo0 1027 ARF and ARF-like proteins.tgg aga ag0 ggg Cgc tgc agg act gtg gag acg taa atg taa act gtg act ata

1054 1081Cnt cga ccc tot tyc tta ttt ttc ttc tct ggc taa aaa tt tta att gga tgt

1108 1135 ACKNOWLEDGMENTSgtt tgg ggs cgg ggg gat gga agt gac ttg gag aat gtg ttt ggg atg aaa taa

1162 L189eta tCt ccc ctt cct cto tcc ccc aac tgg gga otc tcc cca ogc tgc ttt tct We are grateful to Margaret Robertson and Elizabeth Lawrence for

1216 1243 DNA sequencing, and Ed Meenen for synthesizing oligonucleotides.agg aat acE agt cac ata gtt ttt att ttt gtg tct g-g aaa gtg cca aga acc S.A.S. is supported by an EMBO fellowship. This work was supported

nct Ccc nac alt lgt aoa tcC g lO3

a tot tot L297 by Grant DAMD17-94-J-4129 from the Department of Defense1324 1351 to R.W.

tat tat tot tat taa cta ttt ttt aoc att tgc ctg tag g~t att aaa gac tga

taa ctg tag ctc tta gag gag gag a REFERENCES

FIG. 1. Nucleotide sequence and predicted protein translation ofthe human ARF4L gene. ARF4L was initially localized on chromo- 1. Albertsen, H. M., Smith, S. A., Mazoyer, S., Fujimoto, E., Ste-some 17 by hybridization ofa cDNA probe to YACs 19DC6 and 300C2, vens, J., Williams, B., Rodriguez, P., Cropp, C. S., Slijepcevic,which form part of a physical contig on 17q12-q21 (1). The localiza- P., Carlson, M. N., Robertson, M., Bradley, P., Lawrence, E.,tion of ARF4L was confirmed by PCR (using primers S2AF and Harrington, T., MeiSheng, Z., Hoopes, R., Sternberg, N., Broth-S2AR), which demonstrated that ARF4L was specifically amplified man, A., Callahan, R., Ponder, B. A. J., and White, R. (1994).from genomic clones located on 17q12-q21. Amplification of total A physical map and candidate genes in the BRCA1 region onhuman DNA using the same primer pair produced a single band, chromosome 17q12-21. Nature Genet. 7: 472-479.which demonstrates the specificity of the primers for the ARF4L 2. Balch, W. E., Kahn, R. A., and Schwaninger, R. (1992). ADP-locus. ribosylation factor is required for vesicular trafficking between

SHORT COMMUNICATION 115

the endoplasmic reticulum and the cis-Golgi compartment. J. 5. Schurmann, A., Breiner, M., Becker, W., Huppertz, C., Kainu-Biol. Chem. 267: 13053-13061. lainen, H., Kentrup, H., and Joost, H.-G. (1994). Cloning of two

3. Clark, J., Moore, L., Krasinkas, A., Way, J., Battey, J., Tamkun, novel ADP-ribosylation factor-like proteins and characteriza-J., and Khan, R. A. (1993). Selective amplification of additional tion of their differential expression in 3T3-L1 cells. J. Biol.members of the ADP-ribosylation factor (ARF) family: Cloning Chem. 269: 15683-15688.of additional human and Drosophila ARF-like genes. Proc. Natl. 6. Serafini, T., Orci, L., Amherdt, M., Brunner, M., Kahn, R. A.,Acad. Sci. USA 90: 8952-8956. and Rothman, J. E. (1991). ADP-ribosylation factor is a subunit

4. Kahn, R. A., Randazzo, P., Serafini, T., Weiss, 0., Rulka, C., of the coat of Golgi-derived COP-coated vesicles: A novel roleClark, J., Amherdt, M., Roller, P., Orci, L., and Rothman, J. E. for a GTP-binding protein. Cell 67: 239-253.(1992). The amino terminus of ADP-ribosylation factor (ARF)is a critical determinant of ARF activities and is a potent and 7. Valencia, A., Chardin, P., Wittinghofer, A., and Sauder, C.specific inhibitor of protein transport. J. Biol. Chem. 267: (1991). The RAS protein family: Evolutionary tree and role of13039-13046. conserved amino acids. Biochemistry 30: 4637-4648.

Am. J. Hum. Genet. 54:516-525, 1994

Genetic Mapping of the BRCA I Region on Chromosome 17q21Hans Albertsen,* Rosemarie Plaetke,* Linda Ballard,* Esther Fujimoto,* Judith Connolly,*Elizabeth Lawrence,* Pilar Rodriguez,* Margaret Robertson,*'t Paige Bradley,* Bruce Milner,*David Fuhrman,* Andy Marks,* Robert Sargent,* Peter Cartwright,*Nori Matsunami,*'t and Ray White*'t

"Eccles Institute of Human Genetics and tHoward Hughes Medical Institute, University of Utah, Salt Lake City

Summary

Chromosome 17q21 harbors a gene (BRCA1) associated with a hereditary form of breast cancer. As a step towardidentification of this gene itself we developed a number of simple-sequence-repeat (SSR) markers for chromosome17 and constructed a high-resolution genetic map of a 40-cM region around 17q21. As part of this effort wecaptured genotypes from five of the markers by using an ABI sequencing instrument and stored them in a locallydeveloped database, as a step toward automated genotyping. In addition, YACs that physically link some of theSSR markers were identified. The results provided by this study should facilitate physical mapping of theBRCA1 region and isolation of the BRCA1 gene.

Introduction fall within this region or flank it closely on either side.Breast cancer is an often fatal neoplastic disease of To facilita'te efforts to identify the BRCA1 gene itself,

mammary tissue that causes '-50,000 deaths/year in we have integrated these new markers with previously

the United States alone. Most cases are sporadic, with known markers to form a highly resolved genetic map

no apparent genetic lineage. However, a hereditary Thi reg of the long arm of cform of the disease, observed in approximately 5% of The new SSR markers were developed as part of aall cases and characterized by early age at onset, has comprehensive effort to generate large numbers ofbeen genetically linked to marker loci on chromosome highly informative DNA markers with which a high-res-17q21 (Hall et al. 1990). A summary of 13 published olution genetic map of the entire human genome mightreports located the cancer-predisposing locus, BRCA1, be constructed. Such loci are easily typed by the PCR,

within the interval defined by D17S250 (mfdl5) and using unique primers flanking each variable-repeat re-

D17S588 (42D6)(Easton et al. 1993). Several authors, gion (Saiki et al. 1988; Weber and May 1989; Orita et

however, had suggested a more narrow localization de- al. 1990). That markers of this type are abundant and

fined proximally by THRA1 and distally by D175579 highly informative has been shown by several studies

(mfdl88) (Bowcock et al. 1993; Devilee et al. 1993; describing di-, tri-, and tetranucleotide repeats (Litt and

Simard et al. 1993; Smith et al. 1993). The genetic dis- Luty 1989; Economou et al. 1990; Weissenbach et al.

tance separating these two markers is -5 cM (O'Con- 1992; Melis et al. 1993). As part of an effort to auto-

nell et al. 1993). In the study reported here, we found mate SSR genotyping, five of our new markers in the

supporting evidence for this estimate and identified 25 BRCA1 region were fluorescently labeled, gel sepa-

simple-sequence-repeat (SSR) markers, many of which rated, and analyzed on an automated AB1373A DNAsequencer (Ziegle et al. 1992). Specialized software wasdeveloped to facilitate capture and storage of allelic

Received July 27, 1993; accepted for publication October 25, information.1993. In addition to providing genetic information, the

Address for correspondence and reprints: Dr. Hans Albertsen, Ec- new SSR markers can serve as anchor points for a physi-cles Institute of Human Genetics, Building 533, Room 2100, Salt cal map of the BRCA1 region. By screening the CEPHLake City, UT 84112.@ 1994 by The American Society of Human Genetics. All rights reserved, library of YACs (Albertsen et al. 1990), we have identi-0002-9297/94/5403-0015$02.00 fled several YACs that provide evidence for physical

516

Mapping the BRCA1 Region on Chromosome 17 q2I 517

linkage among some of the SSR markers on the genetic systems, Foster City, CA). After lyophilization, themap reported here. primers were resuspended in 300 1 I of TE_4 (10 mM

Tris-CI pH 7.8, 0.1 mM EDTA pH 8.0). Primer concen-

Material and Methods trations were determined by densitometry at 260 nm,and working stocks were prepared at 25 pmol/pi..

Preparation and Screening of Sau3A-digested Genomic

M l3 Library Primer Selection and Development

Genomic DNA from lymphocytes of a human male PCR primers were developed using a locally devel-was digested with Sau3Al, partially filled in with oped computer program, OLIGO (J.-M. Lalouel and T.Klenow large fragment and dGTP/dATP, and size frac- Eisner, personal communication), on the basis of ge-tionated on a 1.2% agarose gel. DNA fragments of nomic sequence flanking the repeats. This program al-400-900 bp were recovered from the gel and ligated lows sequence comparisons between oligonucleotidewith M13mpl8 vector that had been digested with Sail primers and Alu consensus sequences, to detect homol-and partially filled in with dTTP/dCTP. The library ogies and to exclude primers with homologies above awas plated on Luria-Bertani-medium plates, transferred user-defined threshold. PCR conditions were opti-to nylon membranes and UV-cross-linked, and hybrid- mized with respect to MgCI2 concentration (1.0, 1.25,ized sequentially with end-labeled oligonucleotide 1.5, 2.0, 3.0, and 4.0 mM) and annealing temperatures.probes (dG-dT)10 , (dA 3-dT)6, (dA 3-dG) 6 , and (dA-dG- The size of each PCR product was determined from thedA-dT)6. Hybridization was carried out for 3-4 h at sequence and verified, first through gel electrophoresis65'C in 10% polyethylene glycol, 7% SDS, and 1.5 in 5% NuSieve 3:1 agarose (FMC BioProducts, Rock-X SSPE (NaCI-NaHPO4-EDTA buffer). Membranes land, ME) and later through acrylamide gel electropho-were washed in 6 X SSC and 0.1% SDS at 65°C. resis. The primer sequences and their characteristics are

listed in table 1 for each of the 25 new SSR markers onSequencing of Positive Clones chromosome 1-7.

Positively hybridizing plaques were directly pickedinto 100 -p1 PCR cocktails containing 10 mM Tris-HCI Radioactive GenotypingpH 8.8, 40 mM NaCI, 1.5 mM MgCI2 , 5 pmol of each The primer of each pair that showed the least homol-vector primer (A-TGT AAA ACG ACG GCC AGT ogy to Alu was labeled with 32P in a kinase reaction.CGC CAG GGT TTT CCC AGT CAC GAC; and B1- PCR was carried out for a total of 30 cycles. Denatur-CAG GAA ACA GCT ATG ACC AGC GGA TAA ation was done at 95°C for 6 min in the first cycle andCAA TTT CAC ACA GGA); 2.5 ltmol each of dNTPs, for 10 s in each of the subsequent 29 cycles; annealingand 2 U of Taq DNA polymerase (Boehringer-Mann- was done for 10 s at the temperature specific to each setheim, Indianapolis). The reaction mixtures were heated of primers; and extension was done at 72'C for 20 s.at 94'C for 6 min, and the PCR was performed on a The reactions were carried out in a 96-well TechneGeneAmp PCR System 9600 (Perkin-Elmer) for 25 cy- MW-2 thermal cycler (Techne, Cambridge). PCR prod-cles as follows: 95°C for 20 s (denaturation), 60'C for ucts were mixed with a formamide dye solution (98%20 s (annealing), and 72°C for 20 s (extension). The formamide, 0.1% xylene cyanol, 0.1% bromophenolamplified inserts were purified with a Centricon-100 blue, and 10 mM EDTA) and heated to 94°C for 4 min.microconcentrater (Amicon, Danvers, MA), and the se- A 3-p.l aliquot of each sample was electrophoresedquencing reactions were carried out on ABI Catalyst through a 7% denaturing polyacrylamide sequencing(Applied Biosystems, Foster City, CA) by the dideoxy gel with 32% of deionized formamide, 5.6 M urea, andchain-termination method using fluorescently labeled 1 X TBE (Tris-boric acid-EDTA buffer). After electro-M13 sequencing primers: -21M13-TGT AAA ACG phoresis, the gels were exposed to X-ray film withoutACGGCCAGT;andM13RPI-CAGGAAACAGCT drying, at -70'C for 12-24 h, with intensifyingATG AC. The sequences were collected on an screens.AB1383A sequencer (Applied Biosystems, FosterCity, CA). Fluorescent Genoty ping

As with radioactive genotyping, the primer showingPrimer Synthesis the least homology to Alu was labeled, albeit with a

Primers were synthesized in 40-nmol reactions by us- fluorochrome, for detection on an AB1373A sequenc-ing an AB1394 DNA/RNA synthesizer (Applied Bio- ing instrument. Of the three different dyes, the blue dye

Mapping the BRCA1 Region on Chromosome 17q21 519

(FAM) can be incorporated directly onto the 5' end of Results and Discussionan oligonucleotide on the DNA synthesizer; labeling To supplement the existing archive of SSR markers,with the yellow and green dyes (TAMRA and JOE) we have developed >2,000 genomic-sequence-taggedrequires synthesis of oligonucleotides with aminolink 2 we base pedo>2,000 g enomic-equen e-(400808, ABI) attached 5', and subsequent addition of markers based predominantly on tetranucleotide re-the0808, appro attahedy eandsubster. Ur aceddytisn r- peats, as this type of repeat is, in general, highly infor-the appropriate dye-NHS ester. Unreacted dye is re- mative and tends to show less susceptibility to PCRmoved by means of a PD-10 column (17-0851-01, Phar- atifecandstends to show le suceptibi to PCmacia) (for technical details, see Genescan 672 Software artifacts such as laddering than the dinucleotide (CA)nUser's Manual, Appendix D, ABI). Unlabeled oligonu- repeats (Litt and Luty 1989; Tautz 1989). To augment

ase re removlAppendix on apuri Unatn cridge, the marker density on chromosome 17 specifically, acleotides areflow-sorted cosmid library (a gift from Dr. L. Deaven,(400771, ABI; User Bulletin 51, ABI). fLos Alamos National Laboratory) was subcloned into

To establish working conditions for each primer th

pair, PCR was performed on a few samples of DNA the M13 sequencing vector and screened for the pres-

from the CEPH reference panel in a GeneAmp PCR ence of selected di- and tetranucleotide repeats; appro-System 9600. Reaction volumes of 100 gIl contained priate SSR loci were sequenced for development of400 ng of template, 200 Rai M each dNTP, 0.5 jiM each primers. Approximately 80 new SSR markers for chro-

mosome 17 were obtained in this way (data not shown).primer, 2.7 U of Taq DNA polymerase, 0.24 mM sper- We have now mapped 25 of those markers, by geneticmidine, 10 mM Tris-HCI pH 8.7, and appropriate con- linkage analysis, to a 40-cM region surrounding thecentrations (1.0-4.0 mM) of MgCI2 . Annealing temper- BRCA1 locus.atures were estimated on the basis of primer sequenceand were adjusted where necessary. The fluorescentPCR products were also tested for optimal signal on the L clatien o N e the e rs to lt e in i nR esequencing instrument, by analyzing aliquots taken at Strategies to reduce the effort involved in linkage14, 17, 20, and 23 cycles to determine the number of analyses, by reducing the number of genotypes required

cycles necessary to observe a specific product with for map construction, are being implemented in ourminimal formation of spurious product. Following the laboratory. One of these strategies begins by genotyp-establishment of working conditions, the PCR reac- ing each new marker in only four CEPH pedigrees (884,tions were proportionally scaled down to 25 gl and 1331, 1332, and 1362) and comparing two-point lodwere used for genotyping individuals in the CEPH scores obtained with selected loci, to determine an ap-panel. proximate chromosomal location; the marker is then

genotyped on a panel of CEPH individuals with knownLinkage Analysis meiotic breakpoints in that region. An initial rough lo-

All linkage analyses described in this paper were per- calization of each new SSR marker to the BRCA1 re-

formed using four programs from the LINKAGE pack- gion was based on information from the four selectedage (version 5.1): CFACTOR, CLODSCORE, CILINK, CEPH pedigrees. The power of a two-point lod score

and CMAP (Lathrop et al. 1984). analysis in these pedigrees, when testing SSR markersagainst the Genethon markers (Weissenbach et al.

Linkage Data 1992), depends on the degree of informativeness ofLinkage data used in this study were derived partly both markers in the comparison, as well as on their

from O'Connell et al. (1993), partly from the CEPH genetic distance. For two completely informativedatabase (version 6), and partly from genotypic analysis markers <5 cM apart, the lod scores can approach 20.of new SSR markers that were labeled with 32 p or with a However, in light of the fact that markers in the com-fluorochrome. parisons rarely reveal complete genetic information,

most SSR markers are assigned to particular chromo-Identification of YAC Clones somal regions with lod scores of 6-15 and recombina-

To initiate the development of physical representa- tion fractions of .05-.20.tion from the BRCA1 region, the CEPH library ofYACs was screened by PCR according to a protocol Selection of Markers for the BRCAI Mapdescribed by Green and Olson (1990). Some of the The initial stage of the genetic analysis of markersYACs isolated in this way identified close physical link- from the BRCA1 region was in part based on resultsage for several of the SSR markers described here. published elsewhere (O'Connell et al. 1993), which had

File Zoom Pedisree-IPC Mi~sc

UT232-1 246 2422825.26?6I24 6

15Lock Ino 12L6.13I ~ ~ A 1321 11252.I1

r.' Lock 2nf F A21294

1 133 2-14 IAl 2ý50.j1

VS Lock FinfoI A 2252.6O

1332-3 jAlFox .1

IS Lock Inf ýo 1A2 [2764' _____________________

* 1332-6 Al [260.4-

CS Lock 2[Q6

* 1332-7 g ~p

r, Lock Ino A 26 t8

1332- Al 252.1-

M~ Lock Linfo A2 260. _ _ _ _ _ _ _ _ _ _ _ _ _

1332-1 Al 260.9]0S Lock info__ Q2I261

IS Lock , =Info R 2P i [2___6___0.__6

1 - Lock E o A2 2ý6ý041

oad'rn9 1332-15 from~ lanie 2 CVBlue in Y0 lBO.C9-32.v l.res.Loair 1332-16 from lane 22/Blue in YOIB9.C9-32.qjl.res.

Load i rn9 1332-17 from lane 19/Blue in YO1BQ9.C9-32.vi.res.

Figure 2 Data traces from fluorescent allelotyping of UT224 labeled with the FAM flUorochrorne, as an alternative to traditionalvisualization of SSR markcrs with 121). Each blue trace represents the PCR-amplification product of a single individual who is identified bypedigree number and individual number (e.g., 1332-13) to the left of each trace. Also indicated at left are allele sizes (in base pairs) observed foreach individumal. The pedigrec structure shown to the right is added here to aid interpretation of the traces. The split peaks I bp apart, seen ineach allele, could be interpreted as an effect of the Taq DNA polymerase, which is known to unspecifically add a single A nucleotide to the 3'terminus of an extension product (Clark 1988).

Mapping the BRCA1 Region on Chromosome 17q 2 1 523

NFl

0.047 2.33+03 ! TD17S83

DT

0.040 6.12+04 Tco )

0.018 1.58+07D17S511 o ,T

0.021 1.01+13 T I -0.019 1.57+06 cmUT224 -T

0.023 1.96+06 T T

D17S250 -- TT0.019 2.53+11 0 :UT205 - _L. a 00- T + 00•

UT394 0.011 1.20+10 LTJ- -U"UT539 -- - T' T• •oLO~

U53 0.017 1.06+12 Ho to T- "no.017S183 0.012 6.91+03 ,c : .:D17S509" 0.010 4.60+04 "D c I L

D17S508 0.007 3.00+06 T Tw±TO T0.021 1

D17S507 * m1,LO tJ

:3_ X0.063 1.64+12 T ,U D

NME1 I I

UT7 0.035 3.39+04

I0.045 1.96+19

UT217

0.039 8.66+09

D17S74

Figure 3 Upper right, Karyogram of chromosome 17, with a vertical line to indicate the approximate coverage of the map developed inthis study. Left, All markers that could be linearly ordered with odds >1,000:1, starting, at the top, with the most centromeric marker, NFL.D17S74 marks the map boundary on the telomeric side. Recombination frequencies between neighboring markers are indicated on the left sideof the map, and the odds against inversion in each interval are on the right. The markers that could not be placed in a single interval on the mapare placed within confidence intervals of 1,000:1 odds and are illustrated by vertical lines. The two pairs of loci that were haplotyped (see text)are indicated by single and double asterisks (* and **). Marker UT159 is drawn with an arrow pointing up to indicate that the confidenceinterval extends beyond NFl. Marker UT401 is drawn in two intervals connected by a dashed line to indicate a noncontinuous confidenceinterval; however, the location of UT401 around D17S250 is favored with >40:1 odds over the other location. The location of THRAI, shownby its confidence interval, is based on the genetic information; however, since this location is in disagreement with observations made withYACs and pulsed-field gel electrophoresis, its physical location is indicated by a plus sign (+). PPY could not be placed into an unambiguousconfidence interval with 1,000:1 odds and is therefore shown on the map by a dashed line to indicate its most likely location.

analyzed for errors and, where possible, were tested by YACs; data not shown); yet the genetic analysis ex-secondary typings (see O'Connell et al. 1993). Inconsis- cluded UT205 from the 1,000:1-odds interval oftency between the genetic and the physical map loca- THRA1. The physical location of THRA1 in the imme-tions of THRA1 suggested that the genetic data were diate vicinity of the UT205 locus is supported by obser-probably incorrect, in that this gene had been physically vations using pulsed-field gel electrophoresis (Lemonslinked to UT205 on YAC 44F8 (280 kb) (and on other et al. 1990). We attempted to place another marker,

Mapping the BRCA1 Region on Chromosome 17q2l 525

breast-ovarian cancer gene (BRCA1) on chromosome Wright E, Skolnick M, et al (1988) A mapped set of DNA17q21. Am J Hum Genet 52:718-722 markers for human chromosome 17. Genomics 2:302-309

Clark JM (1988) Novel non-templated nucleotide addition O'Connell P, Albertsen H, Matsunami N, Taylor T, Hundleyreactions catalyzed by procaryotic and eucaryotic DNA JE, Johnson-Pais TL, Reus B, et al (1994) A radiation hy-polymerases. Nucleic Acids Res 16:9677-9686 brid map of the BRCA1 region. Am J Hum Genet 54:526-

Cropp C, Liderau R, Campbell G, Champene HM-H, Calla- 534 (in this issue)han R (1990) Loss of hererozygosity on chromosome 17 O'Connell P, Plaetke R, Matsunami N, Odelberg S, Jorde L,and 18 in breast carcinoma: two new regions identified. Chance P, Leppert M, et al (1993) An extended geneticProc Natl Acad Sci USA 87:7737-7741 linkage map and an "index" map for human chromosome

Devilee P, Cornelis RS, Bootsma A, Bardoel A, van Vlier M, 17. Genomics 15:38-47van Leeuwen 1, Cleton FJ, et al (1993) Linkage to markers Orita M, Suzuki Y, Sekiya T, Hayashi K (1990) Rapid andfor the chromosome region 17q1 2 -q21 in 13 Dutch breast sensitive detection of point mutations and DNA polymor-

cancer kindreds. Am J Hum Genet 52:730-735 phisms using the polymerase chain reaction. GenomicsEaston DF, Bishop DT, Ford D, Crockford GP, The Breast 5:874-879

Cancer Linkage Consortium (1993) Genetic linkage analy- Saiki RF, Gelfand DH, Stoffel S, Scharf SJ, Higuchi R, Hornsis in familial breast and ovarian cancer: results from 214 GT, Erlich HA (1988) Primer directed enzymatic amplifica-

families. Am J Hum Genet 52:678-701 tion of DNA with a thermostable DNA polymerase.

Economou EP, Bergen AW, Warren AC, Antonarakis SE Science 239:487-491

(1990) The polyadenylate tract of Alu repetitive elements is Sheer D, Sheppard DM, le Beau M, Rowley JD, San Roman

polymorphic in the human genome. Proc Nat] Acad Sci C, Solomon E (1985) Localization of the oncogene c-erbAl

USA 87:2951-2954 immediately proximal to the acute promyelocytic leukemia

Green ED, Olson MV (1990) Systematic screening of yeast breakpoint on chromosome 17. Ann Hum Genet 49:167-

artificial-chromosome libraries by use of the polymerase 171

chain reaction. Proc Natl Acad Sci USA 87:1213-1217 Simard J, Feunteun J, Lenoir G, Tonin P, Norman T, The VL,

Green P, Helms C, Cartinhour S, Weiffenbach B, Stephens K, Vivier A, et al (1993) Genetic mapping of the breast-ovar-

Keith T, Bowden DW, et al (1987) Towards a primary ge- ian cancer syndrome to a small interval on chromosome

netic linkage map of the human genome: a progress report 17q1 2-2 -1: exclusion of candidate genes EDH17B2 and

on linkage studies with more than 300 new RFLPs. Cyto- RARA. Hum Mol Genet 2:1193-1199

genet Cell Genet 46:623 Smith SA, Easton DF, Ford D, Peto J, Anderson K, Averill D,

Hall JM, Lee MK, Newman B, Morrow JE, Anderson LA, Stratton M, et al (1993) Genetic heterogeneity and localiza-

Huey B, King M-C (1990) Linkage of early-onset familial tion of a familial breast-ovarian cancer gene on chromo-

breast cancer to chromosome 17q21. Science 250:1684- some 17 q12-q21. Am J Hum Genet 52:767-776

1689 Takeuchi T, Gumucio D, Eddy R, Meisler M, Minch C,

Joslyn G, Carlson M, Thliveris A, Albertsen H, Gelbert L, Dixon J, Yamada T, et al (1985) Assignment of the related

Samowitz W, Groden J, et al (1991) Identification of dele- pancreatic polypeptide (PPY) and neuropeptide Y (NPY)

nmutations and three new genes at the familial polypo- genes to regions on human chromosomes 17 and 7. Cyto-tioclus anl 66:601f613 genet Cell Genet 40:759sis locus. Cell 66:601-613 Tautz D (1989) Hypervariability of simple sequences as a gen-

Lathrop GM, Lalouel J-M, Ott J (1984) Strategies for multilo- eral source for polymorphic DNA markers. Nucleic Acids

cus linkage analysis in humans. Proc Natl Acad Sci USA Res 17:6463-6471

8 1:3443-3446 Rs1:4367Varesco L, Caligo MA, Simi P, Black DM, Nardini V, Casario

Lemons RS, Eilender D, Waldman RA, Rebentisch M, Frej L, Rocchi M, et al (1992) The NM23 gene maps to humanA-K, Ledbetter DH, Willman C, et al (1990) Cloning and chromosome band 17q22 and shows a restriction fragmentcharacterization of the t(15;17) translocation breakpoint length polymorphism with Bglll. Genes Chromosomregion in acute promyelocytic leukemia. Genes Chromo- Cancer 4:84-88som Cancer 2:79-87 Weber JL, May PE (1989) Abundant class of human DNA

Litt M, Luty JA (1989) A hypervariable microsatellite re- polymorphisms which can be typed using the polymerasevealed by in vitro amplification of a dinucleotide repeat chain reaction. Am J Hum Genet 44:388-396within the cardiac muscle actin gene. Am J Hum Genet Weissenbach J, Gyapay G, Dib C, Vignal A, Morissette J,44:397-401 Millasseau P, Vaysseix G, et a! (1992) A second-generation

Melis R, Bradley P, Eisner T, Robertson M, Lawrence E, linkage map of the human genome. Nature 359:794-801Gerken S, Albertsen H, et al (1993) Polymorphic SSR (sim- Ziegle JS, Su Y, Corcoran KP, Nie L, Mayrand E, Hoff LB,pIe-sequence-repeat) markers for chromosome 20. Genom- McBride LJ, et al (1992) Application of automated DNAics 16:56-62 sizing technology for genotyping microsatellite loci. Geno-

Nakamura Y, Lathrop M, O'Connell P, Leppert M, Barker D, mics 14:1026-1031

arrzcte

A physical map and candidategenes in the BRCA I region onchromosome 17q12-21

H. M. Albertsen', S. A. Smith', S. Mazover, E. Fuiimoto', J. Stevens', B. Williams;,P. Rodriguez', C. S. Cropp3, P. Shiepcevic:, M. Carlson', M. Robertson', P. Bradley',E. Lawrence', T. Harrington2 , Z. Mei Sheng", R. Hoopes', N. Sternberg', A. Brothman',R. Callahan', B. A. J. Ponder' & Ray White'

We have constructed a physical map of a 4 cM region on chromosome 17q12-21 thatcontains the hereditary breast and ovarian cancer gene BRCA 1. The map comprises acontig of 137 overlapping yeast artificial chromosomes and P1 clones, onto which wehave placed 112 PCR markers. We have localized more than 20 genes on this map, tenof which had not been mapped to the region previously, and have isolated 30 cDNAclones representing partial sequences of as yet unidentified genes. Two genes that liewithin a narrow region defined by meiotic breakpoints in BRCA 1 patients have beensequenced in breast cancer patients without revealing any deleterious mutations. Thesenew reagents should facilitate the identification of BRCA 1.

'Eccresinnuttof' At the end of 1990, a susceptibility gene for hereditarv known to flank BRCAI. Several clones were identifiedHuman Getics breast and ovarian cancer. BRCA 1, was assigned by genetic that formed small islands of overlapping DNA fragnents.an Howard linkage to the long arm of chromosome 17 (ref. 1) About With the map seeded in this way, the Wirslds were extendedHughae Medica

umr.. UnMe,'arv 5% of all breast cancers appear to be hereditary. Using and adiacent islands bndged bywalking. Thiswasdonebv

of U•a• Si L meiotic breakpoints, a consortium of investigators cloning and sequencing the extremities of YAC and PICity, Umh a42. subsequently refined the region containing BRCAI to a inserts from which we could develop new sequence-USA -12 centiMorgan (cM) (sex averaged) interval flanked tagged-site ISTS) markers, We used the STS markers to'CRC Human proximally by D17S250 and distally by D17S588 (ref. 2). isolatenewclonestherebvextendiLg the pnvslcalcoverageCancer Gentrics Other recombinauon events in BRCA I families narrowed of the region. In the process of assembling the conug. weResea.'rch Group.DRawcihGr of the location ofBRCA 1 further to a region of less than 4 cM were careful to test the chromosomal ongin of each newlvPDalnovy, defined by the thyroid receptor gene A I (THRA 1) (ref. 3) developed pair of primers, to avoid introducing errors in

Unrwmrvor' and D17S579(ref. 4). In addition to its role in familial the map as a consequence of chimaeric YAC clones. AsCambindge. Tennis breast and ovarian cancer, BRCA I is probably involved in Fig. I (column 4) indicates, 46 of 127 primer pairs (37%)Court Road. non-familial forms of the disease, a concept which is developed from YAC ends were cumaernc and thereforeCambndge supported bytheobservatnon that 17q is a frequent site of not useful as map reagents. In all. a total of 112 PCRC321QP, UK iossoftheterozygosity (LOH) in sporadic breast and ovarian markers were mapped, of which 92 were developed in our'Laiatiry otTumor Immuno•,rv turnours. laboratory and 20 were obtained from other sources. Ofand Bioloo', High-density maps of polymorphic markers and genes the 104 YAC clones we analysed. 20 had been identified inNational Cancer in chromosome 17q12-21 have been constructed using parallel from the CEPH YAC library by other researchlamnire. NIH. genetic', radiation hybri•" and FISH' data for the purpose groups who made the mapping information available viaBhesa. Marviand of localizing BRCAI more precisely. Using the available World Wide Web (WWW) database servers, and nine20892. USA markers and genes, we have developed a physical map of clones were isolated from the St. Louis YAC library and'Thte Du PontTe rck the region. In turn, we have used the physical map reagents made available by the Michigan Genome Center. A total

Phanrnreuncai to screen cDNA libraries. In this way we have identified of 33 PI clones were isolated from the Du Pont PI phageCompany, several new genes in the minimal region of 1-2 cM, that is library. A chart indicating which genomic clones wereWilmington. nowthoughttocontainBRCA1.Thesegenesarecurrently positive for which markers is shown in Fig. I.Delaware 19M- beinginvesugatedascandidates forthe BRCAl geneitself. Several regions of the contig appear to contain YACs0328. USA which show evidence of rearrangement or internai

Cortspondece Conthrctlon of the physical map deletion. For instance, three clones (251H5, 883E6 andshowadbe aarvad Todevelopphysicalcoverageofthe BRCAI region, large- 963B10) indicate a link between D17S579 and PPY, andto H.M.A inseveast artificial chromosome (YAC) "I- and PI" this link is supported by pulsed-field gel electrophoresis

libraries were mitiallv screened witn a small collection of (PFGE) experiments". However. we have not been able topolymorphic and non-polymorphic PCR-based markers resolve this link unambiguously, most likely because of

article

STS markers usea in mop constructlon

ii

so 7 0I r 287 .. 93041. . .

2 Y ' 3510 I I f'r ' 5921• • ,+ • , .I Yý 41o on! ,sf l *c-. ,i

j y] Oil 2271•2 , , ° ,• ° °5 ' Y /I all 1 7 1 .•1

Y 01' ~2 I3 7ASI. .. .

7'1 270 =IV A790ilI 7 713644

'0 I [ 0/il 316881 * , *,

!2 'Y1 D IOi 486

3 Y' SM.(111 Al"I464 Y, 3II 137(31 . . .. . . . .S Y '0 0o o 1329I1 . .. . . . . .

Y6 7 3M0 ,O0, ,247E21

Y, 3.1111011' 7' 3/0, r,0X10i

3 7' /2 ''1 1, O-..I7i &W 0I 41C2I

:2 PI 2. 72 • 010•23 -A; 17971

24 P 0 V6 4671

26 7 I M 1 12r 156. 21

'7 '2902/ 2n 210&I24 1 2420/ 4031435I

V 1 I1h021 2116451H ,

30 v 2W20'I206 .. .. 16061!1 ' .5W 0/1,, 237F 1

12 '270 O0 2V 9309•

13,6 V /0 370195

17 P 0 t 21. C61)a Y 310 (VI 4 16,61

37 Y 43 ,0/1 279CAl

IA Y,' 290 '1,/l2 ,' 41 H I I I197Y74W 2/2 1/DC6I:'.40 7'ac 'Ili 74441 Y LI 2- 201214 7 42 IY/I 29 1GO 5~43 Y I 0101 236H71,44 7 Y ! 260 I .14 71 ..

4S Y7' 120012n2' '64C11I146 6' 3.42 '4603A7 7 P 110 5D 1~2144 .8 I P121 ,

/9 Y7 W 0(2. A167161

50 Y' IM W,0 5176A12151 vi 3W0 0/ '73671 IQ3 Y 320 O• 7 27AI21

23 Y = 420111$_417 2n' 16787156 Y 460/2 22OE(51 . *,

546 2/0ý W42157 P, • 0/I 51641 '

AI 7Y 2010/1I 171631~6 719 i 300/ 30nGIII71 7 7./I3@1 Y$4i'4111 , .,i" .1.....5t Y 71. tlt .l306H I I, I* * * '

,I Y X20,0/ f 26F31. . ..

4/ Y 27/950/I !.4AC11'd Y ,•5,,(Wl 1.,1 A I I *i*W4 7 42.0/I •1.SAI i;*

46.7 2/OMOM 21/, i4 Y, 4,60/I' =2i

571l A,'* W 0/0 OG2141'Y! 320 0/I 135(121

59 Y1 310M~. M~IT3'0 17 S ~212 r 14G31 *I 1 6' OM. 124621

'3 P 30 27145

'74 P I / 3CF31 I75 1 P I 'V. 4,6G(A15171 30 2,2: 22TA101.

Fig. 1 (This and next page) YAC and P 1 clones in the BRCA 1 region. a and b, Slightly overlapping charts of YAC and P1 clones in the intervalbetween 0 1 7S579 and ERBB2. The map indicates wh~ich clones, indicated in the address column. are posMtVe for the STS markers.;)clymorpnisnms and genes indicated across the top of the Fig. The STS markers, genes and polymorphisms used to constiruct Fig. 2 areindicated in bold type. Markcers that are present in a clone are indicated with a +.. Clones from the ICI library can be identified by two letterstociether within their names. Clones from the St. Louis library have a single letter prefix. The sizes of Pi clones were not determined as they are

article

internal deletion within YACs in this region. Similarly, positive YAC recovered and that YAC was approximatelygenetic evidence supports a close link between D1I7S579 80 kb in size. The small size of this YAC was consistentand D17S509(ref. 6), but it has not been possible to with it being considerably shortened bvdderieon.In severalconfirm this link with certainty in the YACs. For instance, regions of the contig it was only possible to bridge twoDI 7S579 and D I 7S509 identified the same coordinates in adiacent SITS markers by isolating P1I clones. For instance,the YAC library, 259D2, but it was not possible to recover PFGE experiments indicated that 283F1l-5 and 22 1F11-5a single YAC clone positive for both markers even upon are in close proximity; however. neither the CEPH nor thetesting 40-50 colonies from that plate address. Similar ICI YAC libraries contained clones bearing both markers.evidence for instability in YAC clones was observed in In contrast, a P1I clone, I OC7, identified with 221 F1I 1- 5other regions as weil. Forexampie. one address, 99G6, was was positive with both markers thereby bridging the gap.positive for the markers UT62, 361H I 1- 3. 104H3-3 and The fragment of DNA represented by 10C7, which was283FII-5. Only upon screening with 283F1l-5 was a apparently absent from the YAC libraries, appeared tobe

STS markers iLed in moo Corsstruction

68 MO . 1* 12!2

773 : awi 0 7. vUII .. ... ..

73 P . .2, 12AlF2 .

74 P W10 306

7S. P W 0.76, V 32nfl 221AI0 .

77 01 6 .,..

179 Y'6602/2 18116,1

SI 6 , 210011' 170111 .......

82 V It121 101GI21&2 IV' 172, ~ A I I84 P; W W/212

as P . 0)0( "W4I58'' W212 37CiI ..67 V INGJ0/V. 78121,.. , .

U V'X 3~0/1 =WIli89 W

92' I Y2 2W3IliI

97 P (W) 011 G71

69 P . 0/0' 122A!

ig 0 01 1FII1101 V 2 0/1, XIdS2

1()21 V IM 0/I 361141

I 1 0/I. 10" ll701 Y' O7011 72111

Ica, P Cu/V I0371

I1K I dW P S2I1121' 0a01 ISIIII3

IS1 V 012, dl l ~iI1134 0/0 161101

lid V 1W1013 ImAit

11 25 YV 2It2 17

129 Y I 1/ lo w, 1

127 V ~2*2 117521

126 V I40/ 1 f 76D1

17 Y OW 72n I F3 2SI I

121 V IWIC/VI AIll1291V Y1 'a& I

123''301 2 0 261O

132 1 Y' 44FIl I I i1

1341 Y

137 '1 lM 69d"1

P IA 28 fF1.1 8C3.i AC-4 yCIj. ATPCL CNP . KI(aPB

iek•o r e C eifomr fle

oa 7s ST OPZ 3 DI 75I13PPY UT MJ•N2 O 17 . /HSDTSIA'€'IU4 I;q4• I/TI? DI7I;24 Uta r D ooS.1 FSC IUT i3'

017IS06

he- * BRCA 1-- .-

Fig. 2 Contig of overlapping YAC (darl blue) and P1 (light blue) clones from the BRCA I region, showing approximate locations of 39 geneticmarkers (in black type) and genes (in red). Black arrows at th~e bottom of the map indicate meiotic bounlaries thdat incrementally nave helpedto rehne the BRCAI lOCuS. The map is drawn with a minimal set of reagents, composed of 41 YACs and 11 P1 clones, that cover this interval:detailed information about these and additional clones can be found in Fig. 1. The map is estimated to span a physical distance of • 3.5 Mb asfudged by the accumulated size of non-overlapping clones. At the top of this Figure, genes and cDNAs that were identIfied and/or preciselymalpoed in the present study are shown in red, with the direction of transcription indicatedl, where known. Circles indicate the three P1 clonesmaooed by FISH in the expenment shOWn in Fig. 3, in their resoective colours. A previous candidate for the BRAC1 gene, MDC. was idlentifiedby Emi et a/.' 0 and localized between 017S579 and PRY.

amenable to cloning in the P1 vector, of the YACs contain deletions, and only about 100 pointsAnewlyidentifiedpolymorphism ,TT8175, recognizing along the length of the map have been tested, we cannot

locus D17S1136, was developed from the end-clone of exclude the possibility, that other small deletions existYAC 308B3, which contained a nearly perfect (A•G),• which remain undetected. Overall, the map establishesrepeat. Primers designed from this locus revealed four the likely linear order ofa large coLlection of PCR markersalleles and genetic analysis in a subset of CEPH pedigrees over a -4 megabases (Mb) region.confirmed regional localization by showing tight linkageto UT394 and D17S800. Vertfic~aton of the map by in situ hybridization

A map of selected overlapping clones from the 4 cM To help verify our physical map, we analysed several YACregion between D17S579 and THRA) is shown in Fig. 2. and P1 clonesbyinsiruhybridization.Fournon-chimaericThe information on the order of the markers is dernved YACs (A79D1, 397F9, 16CF4 and 17D1 I) that werefrom the overlapping P1 and YAC clones (Fig. 1). The ascertained with D17S579, THRAl, UT394 and UT8region betweer. PPY and D17S579, which appears to respectively, were each tested on 20 metaphase spreadscontain YACs that have suffered deletions or and shown to map to the proximal region chromosomerearrangements, has been included in our map. A small 17q. Dual-colour hvbridization of the YACs furtherbreak in the contig occurred between THRA I and RAJRx. supported the relative ordersuggested by PCR experimentsNo YACs could be isolated that linked THRA 2 to the Idata not shown). In another series of experiments, thecontig, but a link between THRAI and RARA has been relatnveorderofselectedtripletsofeightPlclones(110D12,demonstrated by PFGE16. Since a significant proportion l0Hi 1, 92E1 I, 26F4, 124F2, 108B1 1, 50H1 and 59H9)

was determined by hybridization to interphasepreparations, and found to be in agreement with the

Table F Results of duel-colour P1 hbr0 lizatlon physical map (Table 1). An example of one such

P1 tnplet Signal between Signal Signal outside experiment is shown in Fig. 3, which confirmed that clonesupenmposeo I0HiI is located between 110D12 (ascertained with

(%) (%) (%) D 17S509) and 26F4.

110D12-10H11-26F4 81 7 12110D12-9E!-124F2 65 22 13 Identification of genes 1n the BRCAI region10H11-2g2 oA-50H1 65 15 20 Crucial to the identification of BRCAt is the isolation of26E4-10(c11-59H9 79 16 4 eDNA clones from the minmatl region that is thought to26E4-5to1-59H9 90 6 5 contain the disease locus. The minimal region has been108B11-ou hi-s9H9 92 a c systematically narrowed by the identificaton of critical

recombmnation events in BRCA I famitie 0 : :he most recentTwo of the P1 clones were labelled with a red fluorocprome observations suggest a location for BRaCAo proximal toand a third (underined), was labelled with a green D17i78wref. 17) and distal to D7S776 tref. 18), a geneticfluorocnrome. For each experment 100 interpease distance of 1-2 cM. Usint YAC and P1 clones from thiscnromosomnes were scored. The Table indicates thepercentage of chromosomes in which tne green-labelled region, we have located and determined the nuc1eotidecione hybn tized between the two red labelled clones sequence of five genes (Table 2 and Fig. we and havelcolumn 2), supenmposed on one of the red-labelled clones identified atleasts30othercDNAclones forwhich wehavetcolumn 3), or outside of the interval defined by the two rec partialseouenceinformationp lnoneseriesotexperimentio

labelleda clones (column 4) YACs 727A12 and i9DCa were isoiated from the

Other genes on the physical mapWe placed a number of known genes which had beer,

mapped genetically on 17q12-21 accurately on ourphysical map (Fig. 2) by designing STS primers from thepublished sequences. For example, the 2'-3'-cyclicnucleotide Y-phosphohydrolase (CNP) gene', which hadpreviously been mapped in the 6-cM interval betweenTHRAI and nerve growth factor receptor (NGFR), has

now been localized between D1 75776 and GAS. Similarly,the insulin-like growth factor-binding protein 4 (IGFBP4)gen" 4 which had been mapped on 17q 12-21, was localizedbetween D17S857 and RARA.

Other genes were localized through comparisons of thesequences of the YAC end clones with published sequencedata. Such comparisons revealed that the 5' end of YAC417D9 had identity to part of the human DNAtopoisomerase 11 (TOP2) mRNA-5 (GenBank accessionno. 104088). The sequence identity extended for 79

Fig. 3 Fluorescent in situ hybridization of P1 clones In this basepairs (bp) and diverged at position 4304 in the J04088

example, three P1 clones (110D12, 1OHl1 and 26F4), from sequence, suggesting an intron/exon boundary at thisthe BRCA1 region on the proximal long arm of position. The orientation of the coding sequence of thechromosome 17 were mapped relative to each other. This TOP2 gene in the YAC indicates that transcription of thisexperiment performed on F6606 skin fibroblast cells gene is toward the centromere. Similarly, the 5' end ofarrested at interphase, confirmed that 1 OHi 1 (green) is YAC 263C 11 had 50 bp of identity to the human ATPlocated between 110D12 and 26F4 (both red). citrate lyase (ATPCL) gene"° (GenBank accession no.

X64330), suggesting that ATPCL is also located within theBRCA I region. Divergence of the two sequences at position

endogenousyeastchromosomesbyPFGE, radio-labelled, 1556 in the X64330 sequence suggested that an exon/and used as hybridization probes to screen a fetal-brain' intron boundaryislocated at this position. The orientationcDNA library. YAC 727A12 identified several positive of the coding sequence in the YAC end-clone indicatedclones, one of which, IGI-1, had sequence identity with that transcription of this gene also is toward theIA1.3B, which was recently reported by Campbell and centromere. The 5' end of 883E6 showed 98% identitycolleagues'9 .YAC 19DC6 identified more than 20 positive over its entire length to 398 bp of the 3' untranslatedclones, although most of these clones appeared to be region ofthepeptideYY(PYY) gene27(GenBankaccessionderived from the chimaeric ends of the YAC. As many of no. L25648), indicating that PYY is located close to thethe YACs in the critical region had chimaeric ends and pancreatic polypeptide Y (PPY) gene. PYY and PPY areYAC probes proved troublesome to radiolabel with members of the small-peptide hormone family. Thesufficient specific activity, we used the P 1 clones as probes direction of transcription of PYY is also toward thewhenever possible. For example, P1 clones 51D3 and centromere.Finally, theDNAsequenceofthe Yendclone27H5 identified 15 cDNA clones, representing three from YAC 211 FI I showed 85% sequence similarity to thedifferent genes. The nucleotide sequence of one of these sheep high-sulphur keratin genes B2A and B2D.clones, BCI-16, showed strong homology to the Rab5 Translation of the end clone sequence revealed an 81%family of genes"2 , and the translated sequence of another similarity on the amino acid level to the human keratinclone, BC3- 1, showedsequence identityto the Ki antigen". gene HSHB2AI (Genbank accession no. X63337),A third clone, BC 1-6, had partial homology to the yeast suggesting that we have identified a previously unknowntranscriptional factor, GCN5 (ref. 22), although the start keratin gene, or a pseudogene, that belongs to the keratincodon in the human gene has not yet been localized with gene cluster located on 17q. This gene, if expressed, iscertainty. Work is in progress to map precisely and to transcribed towards the centromere.determine the complete nucleotide sequence of all clonesthat have been isolated. Mutation screening in candidate genes

The coding regions of two of the newly identified genes,

Table 2 Genes identified between DI 7S579 and THRA 1BC I-16 and the Ki antigen, were screened for mutationsTable__2_GenesidentifiedbetweenD17, ___79_andTHRA_1 _ by direct sequencing ofcDNA from 24 unrelated patients.

Homology to: Ofthese patients, 11 camefrom confirmedBRCA] familiesClone name Genbank no. Gene name Clone ID % identity and 13 were selected on the basis of an unusually high

incidence of breast and/or ovarian cancer among their883E6-5 (STS) L1 8244 PYY 125648 98% first-degree relatives. The analysis revealed several silent1G1-1 - 1A1.3B X76952 100% basepair substitutions in both genes, but no missense or1F1-1 not submitted' rat L21 A33295 92% frame-shift mutations that could characterize either geneBC3-1 V1 1292 Ki antigen A60537 100% as BRCAI.BC1-6 not submitted, GCN5 X68628 41%BC1-16 V11293 dog Rab5c S38625 89%263C1 1-5 (STS) L18219 ATPCL X64330 100% Discussion221 F1 1-3 (STS) L18207 HSHB2A_7 X63337 81% We have described the construction ofa physical map on

chromosome 17q, between THRAI proximally and'Submission pending completion of sequence. D17S579distally, that encompasses the BRCAi locus. The

article

map comprises 137 overlappingYAC and P1 clones onto clones that map in the minimal tegion, awaitwhich we have ordered 112 PCR markers. We estimate characterization and testing in the patniot samples.that our map spans -4 Mb of DNA. The cDNAs and genomic clones we idescribe here are

In many places, attempts to construct a physical contig important resources for continued charavaerization andwith YACs highlighted the need for an alternative cloning isolation of candidate genes, and should facilitate thesystem. Several parts of the map were not represented in identification of BRCAI in the near futut'e_YACs, as a result of small deletions or rearrangementswhich presumably occurred during construction of the MethodologyYAC library. Other YACs showed evidence of instability, Libraries. YAC clones were isolated from three ikbwaries: the CEPHlosing specific markers upon replica-plating. In most library; the CEPH mega-YAC library" and the tO library". Localcases, however, these problems were overcome by copies of the CEPH libraries were screenedussmg a PCR-basedsupplementing the YAC map with clones from a P1 strategy". YACs from the ICI library wewtatitained through alibrary. For exampie, it was not possible to obtain a YAC screening service supported by the UK Hurmnaw Genome Mapping

Proiect. A small number of clones obtained frowa- te St. Louis YACclone that was positive for both 283F 1-5 and 221F 1-5 library" had been identified at the Michigan Genmae Center. A PIfrom any of three YAC libraries, although both markers phage library (DMPC-HFF#I Series B) contamsusg three humanwere presenton asingle 80 kb Pl clone. These experiments haploid genome equivalents was obtained %Urum Du Pont Merckdemonstrated that investigators should be cautious in Pharmaceutical Company". This library was reeved in the formatinterpreting mapping results, and whenever possible or 125, 96-well microtitre dishes. with each wvll containing 12employ more than a single source of DNA fragments. different clones. The ?1 library was arrayed onto, aine Hybond N+Since the original observation of King and colleagues (Amersham)filtersusinga BeckmanHigh DenrttReplicangTool.

The cells were grown and lvsed following smauktd procedures andthat a breastandovarian cancer susceptibility gene, BRCA i, the DNA was bound to the filters by UV cro€uA uking. DNA probeswas linked to D17S74 (ref. 1), the localization of BRCAI for screening the PIl Library were obtained b* P.Rin which a radio-on 17q has been incrementally refined by several groups nucleotide, a-'5 P-dCTPwasincorporatedinto wectionproduct.on the basis of meiotic recombination events in BRCAI- Posiuvep Iaddresseswere identifiedthroughapnrmwrvhvbndization.

linked families"-.- 1 6'2 9 . The most recent observations Singie positiveciones were identified bya stscondhfer hybridization

suggest a localization distal to D17S776 and proximal to of an mnoculum from theaddress. Pi DNAvassolsted as described".

D17S78, a distance of 1-2 cM, which is completely DNAprepahratios.^aroseplugsofyyeastcelhaormainingtotalYACcontained within the physical contig described here. DNA (final density of 1.3x 10* cells mi" ) were- p repared using a

Analysis of loss of heterozygosity (LOH) in breast modified protocol". The plugs were rinsed-in Tx•. (10 mMTRIS-tumours defined a different minimal region than the one HCI. pH 7.8, 5 mM EDTA, pH 8.0) and ahwnettther subjected to

defined by meiotic breakpoints. Since BRCA I is widely pulsed-field gel electrophoresis. digested iut •sincuon enzymesthought to be a tumour suppressor gene"', the andend-cloned. orstoredat4'C.

identification of the smallest region ofdeletion in sporadic DNA labelling. DNA from YACs, P I sand cDNA4-lones was labelledbreast and ovarian tumours has received much attention for hybridizationexpertmentsusingoneoft'rx-wetandardtechniques.indefiningnewboundariesforthelocalizationofBRCA1. For in stru hybridization P1 clones were l~itinylated by nick-From an analysis of 130 breast tumours, Cropp er a!." translation as descrbedld. For cDNA screenus* the probes werehave demonstrated a minimal region of LOH centered radio-labelled with a-UP-dCTP using random-oAo pnmers'. Foraround D17S846, which is approximately 0.5 Mb P1 library screening DNA probes were ladadied by directly

centromeric of D 17S776. Given the high likelihood that incorporating 15P-dCTP by PCR into thvRtLeRt-,mduct.

the recombination event placing BRCAI telomenc to YAC analysis. Yeast chromosomes prepared i•-agarose plugs wereD17S776 is genuine, the simplest explanation for these size-separatedon 1%SeaKemagarosegeisinO.5'TBEusingaCHEFfindings is that at least two separate loci on 17q 12-21 are DRIL Mapper (BioRad). Gels were blotted oatwo Biotrans Nylonimportant in breast cancer development: BRCAI and a membranesICN)orHvBond-N(Amersiamtessmgalkalineblotting.

and UV cross-linked as directed by the manmuacturer. Filters weresecond, more proximal, locus defined by LOH in tumours. hybridized and washed according to standiatd procedures. thenWe have mapped a total of seven genes in the minimal exposed to Kodak XAR film. The sizes of inawirdual clones were

BRCAI region defined by meiotic breakpoints. At least determined by comparson to their relative pwations among theoneofthese genes, EDH17B2( 17-HSD), has beenexcluded natural yeast chromosomes.as BRCAI by sequencing the gene in a large number ofpatients who are members of linked families'",-'. Another YAC end-cloning. Inverse PCR and end'bc:vne sequencing wasgene, 1A1.3B, was excluded in one study!, but probably performed as described'*. Also. a new prsnu was developed

specifically for use with Rsal-digested subtateA(sa.-YAC3LU CAG-warrants further investigation. Two of the remaining GAA ACA GCT ATC ACC GGA AGA ACG ANG GAA GGA GC).genes, BCI-16 and Ki, have been excluded as candidates The inverse PCR technique was ad.rtett ,Itin" the P1 vectorin thisstudy. BCI-16shows 86% homologyatthe amino- (pAdl0sacBIl) using te following modidications: 100-200 ng PIacid level to the Ras-related gene RAB5. RAB5 is a GTP- DNA was digested with either Alul, Hhal, ailpll, NlaIlIl. Rsal orbinding protein that appears to be located in epithelial Taqlphenol-extracted, andethanol-precipttateW.LvophilizedDNA

cells at the cytoplasmic surface of the plasma membrane". pellets were resuspended in 10 pil of TE,•5 • and Mcae DNA was ligatedat a concentration of 2 gjg ml-1. The higationoianons were diluted to

Ki encodes the Ki antigen, which is a hi[hly conserved 0.4 UZ Ml" and ! ul oi each wis used a- ,he temt',ljte for PCRnuclearproteinoriginailydetectedwithseratfrom patients ampiification of end-clones. The Pl3Rup t(,'GT AAA ACG ACGwith the autoimmune disease, systemic lupus GCCAGTGGCCGCTAATACGACTCAC'IA)andPi3Lrp(CAGerythematosus (SLE) . Wehave determined the complete GAA ACA GCT ATC ACC GCA ATA TAG TCC TAC AAT GTC)coding sequences ofboth genes in samples from 24 patients primers were used with templates digested wift Alul. Hhai. Hpall,and have found no mutations. The other genes (BC1-6, N/aIlIor Taql:P ISRrp(CAGGAAACA C-CTATC ACCGGATCG-1and have fn no tvetbenscrenedfo mutations. TheothergenAAA CGG CAG ATC GCA Iand Pl5Lup (TGTAAA ACG ACG GCCIFI - I and RN2) have not vet been screened for mutations AGTTAA'I G GCC GTC GAC ATTTAG) wereused when templatesin patient samples, and thus remain untested candidates were digested with Nlalll and Rsal. For Alu-,tetor PCR the DNAfor BRCAI. In addition, some 30 other partial cDNA fromtwoagarosepiugswasheatedat 68Cin.400Wdistiedwater.

Aliquots of 1-3 ml (-5 ng) were used in PCR with one vector prnmer the two homoiogues. basepair differences between the homologous(YAK3R or YAKSL)M and one of the AMu.pnmer"P. YAC DNA was transcripts can be observed as two overlapping peaks.also amplified using the Alu primer alone and analysed adiacent tothe correspondingAlu-vectorproducts ona2% agrose geL Unique Development of PCR-based markers. New PCR primers wereAlu-vector fragments were excised from the gel and prepared for developed from the DNA sequence derived from the extremies ofDNA sequencing. large-insert clones using a locally developed computer algorithm,

OLIGO. The localization of a new STS was verified using a panel ofEnd-donesequedng. End-done PCRproducts from eitherinverse somatic cell-hybrids (Coriell Laboratories). An STS found to lie onPCR or alu-vector PCR were purified with Centricon-100 micro- chromosome 17 was further testedagainst 5 ngofetchoftheisolatedconcentrators (Anicon) according to manufacturer's directions. YACandP1 clones to identify regions ofoverlap. New STSs developedSince the PCR primers were designed with M 13 universal primer (- from end-cones that were not positive for othere stingcones were21M13) and M13 reverse primer (MI3RPI) sequencing tails, the used to screen the libraries to identify additional dones. All PCRend-clones could be sequenced on an automated sequencer ABI primers and conditions are available electronically by anonymous373A (Applied Biosystems) without cloning, ftp (see below).

In situ hybridization. Interphase nuclei were obtained from the DNA sequence comparison. To identify homologies or identitiescultured human skin fibroblast cell line F-6606, prepared for FISH, between sequence we obtained and other sequences, we used theand hybridized as describede with the following modifications. The BLAST algorithm" via the internet This allowed us to compare ourhybridization signal from PI DNA was detected using either an sequence on both the DNA and protein levels, with all sequence dataFITC-anti-digoxigenin antibody (Boehringer-Mannheim) for stored at NCBI. Based on these results, sequences were selected anddigoidgenin-labelled probes, or sreptavidin-Cy3 (Jackson Immuno retreved for further analysis from NBCI using the E-mad retrieveResearch Laboratories, Inc.) for biotin-labeiled probes, and the server.nuclei were counter-stained with DAPI. Fluorescence was visualizedusingan Olympus BXSOepifiuorescence microscope equipped with Electronic Iformation access. Detailed information about PCRfilter sets specific for FITC, DAPI and Cy3 (Chroma Technology). markers and genomic clones is available from databases connectedImages were collected and processed using a cooled charge-coupled to the internet Some YAC information is accessile on World Widedevice (CCD) and software specially designed for FISH analysis Web(WWW)'databaseserversatMichigan HumanGenomeCenter(BDS, Inc.), The P1 clones were labelled by nick translation with (http://mendeLhgp.umich.edu/Home.html) and Baylor College ofeitherdigoxsgenin lI-dUTP(Boehringer Mannheim)or biotin 14- Medicine Genome Center (http://gc.bcm.tmc.edu:8088/). DNAdATP (Gibco BRL); the order of the P1 clones was determined by primer sequences and PCR conditions are available from GDB.hybridizing two l cloneslabelled in red together with one Pl clone Sequence information relating to clone extrenmites and cDNAs islabelled in preen, and scoring the position of the green site relative to available from GenBank. The sequence, geneand priner informaoonthe two red sites in 100 or more interphase nuclei. The percentage of may aiso be obtained directly from the authors via anonymous ftp tonuclei in which the green signal was observed either between or coronamed.utah.edu (128.110.23 1. 1).outside the two red signals was recorded.

cDNA screening. A commercially available fetal-brain cDNA library(Stratagene, no. 936206) was screened with isolated YAC or P1 Acl aowedgemetclones. Phage plating, filter lifts and clone purification were all We acbnowledge the CEPH for provding PCR pools for YAC libraryperformed according to standard procedures. screening and a collection of chromosome 17-specific YACs, the UK

Human Geiome Mapping Proiea for assutance •ith screening the ICIMutation beening. To screen for mutations, we designed oligo YAC library and the Michigan Genome Cente for prosding YACsprimerswithsequencingtails foreachgeneofinterest.Asatemplate from the St. Louis YAC library. We are grawful to Ed Menen forfor mutation detection, RT-cDNA was prepared from RNA isolated synvhewng oligonudeotide primers, Dave Fuhrman and Rob Sargenrfrom lymphoblas cultures from selected patients using the RNA for computer support and Ruth Foltz *or ediming the manuscript ThisPCR kitfrom Perkin Elmer. The PCRproductswere gel-purified and work was supported by NIH grants ROI-HGO0367-04 and ROl.subsequently prepared for either Taq cycle sequencing or T7 solid HGO0339. S.A.S. is supported by an EMBO fellowship and S.M. by anphase sequencing on the AB1373A sequencing automat Because the ECfellowship. R. W. is an Invesngaror of the Howard Hugher MedicalRNA recovered from the cell culture is a mixture of transcripts from Institute.

Received 4 April; accepted 2 June 1994.

aracte

1 . HalIUJM. et al. Uinkage Of early-onset familial breast cancer to chromosome systemic lupus erytnemratosus. Chrt. axp. Immunot. 79,.200-214 (1990).1 7g21. Science 250, 1 684-1689 11990). 22. Georgakooouios. T. & Thireos, G. Two distinct yeast transcriptional activators

2. Easton, D.F. at al. Genetic linkage analysis in familial breast and ovarian require the function of the GCN5 protein to promote normal levels ofcancer. results from 214 families. Am. J. hum. Genet. 52. 678-701 (1993). transcription. EMBO J. 11. 4145-4152 11992).

3. Bowcock, A.M. efal TH-RAI and 017S183 flank an interval of c4CM for the 23. Sprinkie, T.J... Koir. R.E.. Fain. PD.. Stoming. l.A. S Whitney. .1.5breast-ovarian cancer gene (8RCA 1) on chromosome 17Q2 1. Am. J. htum. Chromosomial mapping of tne numan CNP gene using a meioticGanet. 52, 718-722 (199M). crossover DNA panel. PCR, ana aillee-speciflc probes. Genomics 16.

4. Chamberlain, J.S.. eal. BACA1 maps proximal to 0l 7S579 on chromosome 542-5 (1993).17q21 by genetic analysis. Am. J. trum. Genete. 52. 792-798 (1993). 24. Tonin, P. et a). The human insulin-like growth factor-bindting protein 4 gene

5, Anderson, L.A. at al. High-density, genetic mao of tne BRCA1 region of maps to chromosome region 17o12-021.l and is close to the gene forcnromosome-l7g12--o21. Genomricst 17. 618-623 (1993). nereditary breast/ovarian cancer. Genomics 18. 414-417 (1993).

6 Albertsen, H. oatla. Genetic mapping of the BRCA1 region on cnromosome 25. Tsai-Pffugfelder. M. eta). Cloning arid sequencing of cONA encoding human1

7q21 Am. J. numi. Genet. 54, 516-525 (1994). DNA topoisomerase l1and locai~zation of the gene to chromosome region

7. Abel. K.J. eta). A ramation nylond mao of t~he8RCAt region 01 chromosome- 1 7g21 -22. Proc. natrn. Acad. Sci. U.S.A. 815, 7177-7181)(1988).t7g1 2--o21. Genorrscs 17. 632-641 (1993) 26. Eishoubjagy. N. etal. Cloning and expression of anhuman ATP-citrate lyase

8, O'Connell, P. etul. A radiation hybrid map of the BRCA I region. Am. J. hum cDNA. Eur. J. Biocfiem. 20.4. 491-499 (1992).Genet. 54. 526-5.34 (1994). 27. Kohnn, K. et a. Cloning and structural determination of numan beptide rYY

9. Plaiter W.L. et a). Multicolor FISH mapping with Aiu-PCR-amoiifiea YAC cDNA and gene. aliochem. Siopnyys. Acre 1173. 345-349 (1993).clone DNA determines the order of markers in the 8PCA1 region on 28. Keiseil. D P.. Black, D.M., Bishop. D.T. &Spurr, N.K. Genetic analysis of thecnromosome 17gt2-g21. Genomico; 17. 624--631 (1993). 8RCAI region in a large breast/ovarian family: refinement of the minimal

10. Albertsen, H.M. et al. Constniction and Characterization of a human yeast region containing BRCAI1 Hum. molec. Genet. 2. 1823-1828 (1993).artificial Chromosome library containing seven haploid genome equivalents. 29. Smith, S.A. at al. Localization of the breast-ovanan cancer susceptibilityProc. netrt. AcedS. Sci. U.S.A. 87. 4256-4260)(1990). gone (BRCAt1)on 17q 12-21 to an interval of s 1 CMA.Gentes Chrom. Cancer.

11, Chumakov. f.M. et al. Isolation of chromosome 21 -specific yeast aitificial 10. 71-76)(1994).chromosomes from atotal human genomelibrar-yNature Gentet. 1, 222-225 30. Smith. S.A., Easton. OF., Evans D.G.R. & Ponoer. 8.A.J. Allele losseshi the(1992) region t7q 12-21 in familial breast and ovarian cancer invoive the wild-type

12. Anand, R.. Riley. J.H.. Smith, J,C. 6 Markham, A.F. A 3.5 genome equivalent chromosome. Nature Genet. 2. 125-131 (1992),multi access YAC library: construction. charactensation. screening ano 31. Croop. C.S. at al. Evidence for involvement of EIPCA I in sporadic breaststorage. Nuci. Acids Res. 18. 1951-1955(1t990) carcinomas. Cancer Res. 54. 2548-2551 (1994).

13. Burke. 0.7.. Carlo. G.F. &Olson, M.V. Cloning of large segments of exogenous 32. Green, 8.0. &Olson, M.V. Systematic screening of yeast artificial -chromosomeDNA into yeast by means of artificial Chromosome vectors. Science 236, libraries oy use of the polymerase chain reaction. Proc. hatm. Acaed. Sci.806-812 (1987). U.S.A. 87, 1213-1217 (19901.

14. Shepherd. N.S. at al. Preparation and screening of an arrayed human 33. Riley, J. at al. A hovel, rapid method for the isolation of terminal sequencesgenomic libraryqgenerated wifh the Pi Cloning system. Proc. haen. Acad. Sc,. from yeast artificial chromosome (YAC) clones. Nucl. Acids Res. 18. 2887-U.S.A. 91. 2620-2633 (1994) 2890 (1990).

15 Xu, W., Rider. S.H, Tanner, M. &Solomon, E.Mapping of umanchromosome 34. Trask, 8.J.. Massa, H., iKenwicK. S. & Gifscmier. J, Mapping of human7 using a combination of pulsed field gel electroonoresis and Chromosome chromosome X028 by Two-coior fluorescencen sintu hybridization of DNA

mediated gene transfer. Cytogertet. cell) Ganet. 51 till1-i1112 (1989) sequences to interphase Cell nud~ei. Am. J. hum. Genet 48. 1-15 (1991),.16. Lemons, P.S. eta). Cloning and characterization of the tll 5:17) transiocation 35. Feinberg, A.P.,& Vogeistein, B. A technique for racioiabeiing DNA restniction

breakpoint region in acute promyeiocytic leukemia. Genes Chrom. Cartcer2. endonucleese fragments to high specific activity. Anal. Bioctlem. 132. 6-1379-87 (1990). (1984).

17. Simard, J1. at al. Genetic mapping of the breast-ovanian cancer syndrome to 36. Joslyn. G. at al. Identification of deletion mutations and three new genes ata small interval on Chromosome 17g12-21: exclusion of candidate genes . thalfamulial polyposis locus. Cell 66. 601-613 (1991).EDH1 792 and RARA. Hum. mocic Genet. 2. 1193-1199)(1993) 37. NelsOn,.DL. etal. Aiu Dolymerase chain reaction: a method for rabid isolation

18. Gologar, D.E. et al. A latge kihored with 17o-linked breast and ovarian of human-specific sequences from complexDNA sources. Proc. natn, Acaa.cancer. genetic. phienotypic, and genealogicalianalysis. J. hatr. Cencerlrtst. Sc,. U S.A. 86, 6686-6690 (1989)88. 200-209(ý1994,1. 38. Tagle. D A. & Collins, F.S. An optimized Aiu-PCP pnimer pair for human-

19. Campbell, I.G. tal. A novel gene encoding a 8-pox protein within the BRCA1 aoecific amplification of YACs and somatic celi hyoinds. Hum. mo/ac. Genet.region at 17q2l.1.t Hum. molec. Ganet. 3, 589--594 (1994). 1. 121-122 1992).

20. Chearier. P.. Parton, P.G.. Heun, H.P.. Simons. K. & Zonial. M. Localization 39. AtASchul, SYF., Gish, W.. Miller. W.. Myers, E.W. & Lipmhan. 0.1. Basic localof low moiecular weight GTP binding proteins to exocytic and endocytic alignment search tool. J. molac. Siof. 215. 403-410 (1990)compartments Cell 82, 317-329 (1990). 40 Emi. M. etal. A novel metaiioorotease/dysintegenn-uike gene at 17021 .3 is

21. Nikaido T. et a) Cloning and nucleotide sequence of cDNA for K, antigen, a somatically rearranged in two Primary breast cancers. Nature Geneat. 5. 151 -highly conserved nuclear protein detected with sera from patients with 157 (1993).

479

GENOMICS 33, 207-213 (1996).ARTICLE NO. 0185

Sequence, Genomic Structure, and ChromosomalAssignment of Human DOC-2

HANS M. ALBERTSEN,*'t'1 SIMON A. SMITH,t ROBERTA MELISt BRIANA WILLIAMS,*PILAR HOLIKt JEFF STEVENSt AND RAY WHITE*'T

* Eccles Institute of Human Genetics and t Huntsman Cancer Institute, University of Utah, Salt Lake City, Utah 84112

Received August 21, 1995; accepted December 21, 1995

suppressors will be identified in various epithelial cellDOC-2 is a human gene originally identified as a 767- types. One candidate for such a role, DOC-2, was iden-

bp cDNA fragment isolated from normal ovarian epithe- tified by means of the differential display techniquelial cells by differential display against ovarian carci- (Liang and Pardee, 1992), and DOC-2 was shown tonoma cells. We have now determined the complete be expressed in all normal ovarian epithelial cells butcDNA sequence of the 3.2-kb DOC-2 transcript and lo- significantly down-regulated or absent in all of a seriescalized the gene to chromosome 5. A 12.5-kb genomic of ovarian carcinoma cell lines tested (Mok et al., 1994).fragment at the 5'-end of DOC-2 has also been se- The biological function of DOC-2 remains unclear,quenced, revealing the intron-exon structure of the but the predicted protein sequence shows some degreefirst eight exons (788 bases) of the DOC-2 gene. Transla- of homology to that of proteins from other species. Attion of the DOC-2 cDNA predicts a hydrophobic protein the amino-terminal end of DOC-2, a 140 amino acidof 770 amino acid residues with a molecular weight of segment shares homology with a recently described82.5 kDa. Comparison of the DNA and amino acid se- seget shae hom ology mwit a reenl d r dquences of DOC-2 to publicly accessible sequence data- phosphotyrosine interaction domain (PID) (Bork andbases revealed 83% identity to p96, a murine protein of Margolis, 1995).. Recently published results indicatesimilar size, thought to be a mitogen-responsive phos- that the mouse homologue of DOC-2, p96, is phosphory-phoprotein. In addition, about 45% identity was ob- lated on serine residues (but not on tyrosine residues)served between the first 140 N-terminal residues of in a pattern that appears to be linked to the cell cycle:DOC-2 and the Caenorhabditas elegans MllO.5 andDro- a minimal degree of phosphorylation occurs in the G1sophila melanogaster Dab genes. © 1996 Academic Press, Inc. stage and rapidly increases following mitogenic stimu-

lation with CSF-1 (Xu et al., 1995).Originally, only a 767-bp fragment of the DOC-2

INTRODUCTION cDNA was identified. Here, we present the complete-3200-bp sequence of DOC-2, its chromosomal loca-

Genes that show differential expressionbetween nor- tion, and a 12.3-kb contiguous genomic sequence har-mal and tumor tissues are likely to function either di- boring the first eight exons coding for the N-terminalrectly in growth regulation or cellular differentiation third of the DOC-2 protein. We also demonstrate thator indirectly as a response to changes in the cellular DOC-2 is expressed in a wide variety of tissues.environment. Those genes that directly influencegrowth or differentiation can be grouped into two sepa- MATERIALS AND METHODSrate classes, the first type functioning as negative regu-lators (the prominent group known as tumor suppres- Isolation of human DOC-2 cDNA clones. cDNA clones were iso-

geners) athe endthe sond funtowni tmor stmultes- lated from two commercially available cDNA Lambda ZAP phagesor genes) and the second functioning to stimulate libraries: a brain-stem library and a fetal-retina library (Stratagene,growth and differentiation. Several of the tumor sup- San Diego, CA, Catalog Nos. 936206 and 937202). Each library was

pressor genes characterized to date, including RB (Lee plated at a density of approximately 25,000 pfu per 150-mm petri

et al., 1987), APC (Groden et al., 1991), and BRCA1 dish on Escherichia coli strain LE392. Phage plaques were lifted

(Miki et al., 1994), control growth in epithelial cells; on Biotrans nylon membranes (ICN) or Hybond-N (Amersham) andscreened according to standard procedures. DNA probes for hybrid-

accumulated evidence suggests that additional growth ization were radioactively labeled with [a-3 P]dCTP using a Prime-It II DNA labeling kit (Stratagene, Catalog No. 300385). Filters were

Sequence data from this article have been deposited with the washed and exposed to HyperFilm (Amersham) overnight at -701C.EMBL/GenBank Data Libraries under Accession Nos. U39050, Plasmids were excised from the phage according to the manufactur-U41096, and U41111. er's recommendations.

1To whom correspondence should be addressed. Telephone: (801) 585- DNA sequencing. Bluescript plasmids or PCR products designed6178. Fax: (801) 585-3833. E-mail: Hans.Albertsen@genetics.utah.edu. with M13 universal primer (-21M13) and M13 reverse primer

207 0888-7543/96 $18.00Copyright © 1996 by Academic Press, Inc.

All rights of reproduction in any form reserved.

208 ALBERTSEN ET AL.

(M13RP1) sequencing tails were sequenced using the dye-primer RESULTS AND DISCUSSIONtechnique on an ABI 373A automated sequencer (Applied Biosys-tems). Alternatively, plasmids were sequenced on the ABI 373A in-strument using a dye-terminator technique. Isolation of cDNA Clones from the DOC-2 Locus

DNA and protein sequence analysis. DNA sequences were mergedand edited using the Intelligenetics suite of programs. Searches for During a search for expressed genes in the BRCA1protein alignments and motifs were performed using the Wisconsin region, several novel transcripts were isolated usingSequence Analysis Package (Version 8). To identify homologies or genomic clones to screen cDNA libraries (Albertsen etidentities between the DOC-2 sequence and other sequences, we used al., 1994). One of the cDNA clones isolated in this man-the BLAST algorithm (Altschul et al., 1990) through the Internet.This procedure allowed for sequence comparisons on both the DNA ner, 40F1, was found to be chimeric; 542 bp at one endand protein levels, against all sequences stored at NCBI. On the of the insert derived from the DOC-2 gene, and thebasis of these results, we retrieved selected sequences from NCBI remaining 1888 bp derived from the DLG3 gene (Smithfor further analysis using the E-mail retrieve server. et al., 1995). The predicted protein translation of the

RNA preparation and first-strand cDNA synthesis. RNAwaspre- chimeric clone 40F1 showed a single open readingpared from three different solid tissues as well as cultured cells from frame that included the correct orientations and framestwo epithelial cell lines, using the TRIzol reagent (Gibco BRL, Cata-log No. 15596-026) according to a protocol provided by the manufac- from both genes. Rescreening of the cDNA librariesturer. First-strand cDNAs were synthesized from a RNA template with 40F1 resulted in the isolation of several newusing the SuperScript II RT kit (Gibco BRL, Catalog No. 18064-014). cDNA clones, most of which were transcripts of the

DNA primers, primer design, and PCR amplification. DNA prim- DLG3 gene. However, a 2726-bp clone, 1RA1, whichers were designed to be 18-21 bases longwith calculated melting points corresponded uniquely to the DOC-2 gene, was alsobetween 56 and 60'C. Tailed sequencing primers deviated from this isolated.pattern by additionally having the 18-base-long universal forward orreverse primer sequences attached at the 5'-end of each locus-specificprimer. Bases from the sequencing tails were ignored for the purpose of Sequence Analysis of 1RA1determining PCR annealing temperatures. The oligonucleotide primerswere synthesized on an ABI394 DNAIRNA synthesizer (Applied Biosys- The complete sequence of 1RA1 was determined bytems), lyophilized, and resuspended in H 2 0 to a concentration of 20 pM. first sequencing the two extremities of the clone usingPrimers used for direct sequencing by means of the chain terminatortechnique were purified on OPC columns (Applied Biosystems) follow- the standard universal forward and reverse sequencinging the manufacturer's recommendations. In general, PCR amplifica- primers; then, using custom-made primers, the inter-tion was performed in a Perkin-Elmer 9600 ThermoCycler in a PCR nal sequence of 1RA1 was completed following severalbuffer with MgC12 to a final concentration of 1.5 maM, under these rounds of sequence walking. The consensus cDNA se-standard thermocycling conditions: quence derived from combining 1RA1 and the origi-

Initial denaturation: 94°C for 120 s nally published partial sequence (GenBank Accession'Denaturation: 94°C for 20 s No. L16886) extended for 2960 bp, with the L16886

30X Annealing: Lower T, of the two primers for 20 s DNA sequence extending 241 bases further 5' thanLExtension: 72°C for 40s 1RA1. Analysis of this sequence revealed an open read-

PCR products were visualized in standard 150-ml agarose gels con- ing frame extending from the extreme 5'-end of thetaining 1X TBE, 0.8-2.0% SeaKem (FMC BioProducts), and 1 M1 sequence for 2163 bp, followed by 792 bp of 3'-untrans-ethidium bromide. lated sequence and a poly(A) tail. Of four candidate

In situ hybridization. Metaphase chromosomes were prepared s eque nce an d a atY\ofrom peripheral blood from a normal male and hybridized with the initiation methionine codons located at the 5'-end ofP1 clone, 35G12, using the general procedures described by Lichter the cDNA, none were in the consensus environmentet al. (1991). The hybridization signal from the biotinylated P1 DNA described for translational initiation (Kozak, 1991), norwas detected using streptavidin-Cy3 (Jackson Immuno Research were any in-frame stop codons present in the 5'-end ofLaboratories, Inc.), and the chromosomes were counterstained with the predicted reading frame. This suggested that theDAPI (4'-6-diamidino-2-phenylindole). Fluorescence was visualizedusing an Olympus BX50 epifluorescence microscope equipped with 5'-end of the DOC-2 cDNA remained to be found.filter sets specific for DAPI and Cy3 (Chroma Technology). Imageswere collected and processed using a cooled charge-coupled device Isolation and Sequencing of Genomic Clonesand software specially designed for FISH analysis (Vysis, Inc.). Containing the 5'-End of DOC-2

Somatic cell hybrid analysis. The NIGMS human/rodent somaticcell hybrid mapping panel 2 and the regional mapping panel forchromosome 5 (Coriell Cell Repository) were tested with the DNA To isolate genomic clones containing the 5'-end ofoligo primer pair doc2-B (AAATTT=GGAGAGTCTAGAGC) and DOC-2, the 1RA1 cDNA was used to screen a P1 phagedoc2-C (GAATACGCTTGGTTCGTCC). Each reaction contained 50 library (Shepherd et al., 1994). Two clones, 35G12 andng of template DNA and 12.5 pmol of each primer in a 25-Al reaction 47E3, were obtained, from which restriction fragmentsvolume. The PCR buffer contained MgCl2 to a final concentration of generated by the enzymes EcoRI, HindIII, and PstI2.0 mM, and the amplification was carried out under the followingthermocycling conditions: were cloned into pBluescriptII. Four subclones derived

from 47E3 were selected by hybridization with a radio-Initial denaturation: 94'C for 120 s labeled fragment from the 5'-end of the DOC-2 cDNA

-Denaturation: 94°C for 30 s and sequenced as described above. The sequences we30x Annealing: 52°C for 30 s

LExtension: 72°C for 40 s obtained fell into two contigs of 2073 and 12518 bpFinal extension: 72'C for 120 s respectively.

GENOMIC CHARACTERIZATION OF HUMAN DOC-2 209

Identification of the 5'-End of DOC-2 cDNA and tral portion and toward the carboxy terminal of theDetection of Eight Intron-Exon Boundaries protein. The region between basepairs 1503 and

1594 in the human sequence appeared to be shiftedSince attempts at isolating other cDNA clones that by a single nucleotide, resulting in a local shift of

could extend the DOC-2 cDNA sequence further 5' were reading frame between DOC-2 and p96. To verify theunsuccessful, we sought to identify the 5'-end of DOC- sequence of the p96 gene, total RNA extracted from2 by comparing the genomic DNA sequence with that of a differentiated and an undifferentiated mouse stemmurine p96 (GenBank Accession No. U18869), which cell line was reverse-transcribed and used as tem-we believe is the homologue of DOC-2 (see below). This plate in a RT-PCR reaction with a pair of primersanalysis revealed a single region in the human genomic specific to the mouse (the RNA was generously do-sequence that was highly homologous to the DNA se- nated by Dr. Suzi Mansour). The DNA sequence ofquence of the first exon of p96. To verify that the homolo- the resulting PCR product indicated that two errorsgous sequence indeed corresponded to the 5'-end of the had been made during the original sequencing ofDOC-2 cDNA, we developed four PCR primers (doc2-5A, p96. This result has been confirmed independently,GTCTAATACAGCAGGAGAAGG; -5B, CTAAACTAT- and the DNA sequence of p96 has been correctedGCCTACAGGTGTC; -5C, GGGATCGCCTGGTGTCAC- (Dr. Charles 0. Rock, St. Jude Children's ResearchCAA; and -5D, GTCCACTGGTACTGAGGTTTG) lo- Hospital, Memphis TN, pers. comm., Oct. 24, 1995).cated in the genomic sequence of DOC-2, upstream of Translation of the DOC-2 cDNA predicts a hydropho-the predicted translational initiation site. The primers bic protein of 770 amino acid residues with a molecu-were sequentially positioned, with doc2-5A being far- lar weight of 82.5 kDa sharing 81% identity with thethest away from the predicted translational initiation gene product of p96, which is predicted to encode asite. Each of the PCR primers was tested separately protein of 766 amino acid residues with a molecularwith a second primer (40Irp, TTGTCCCTGAGACCG- weight of 82.7 kDa.ACCA) located in the third exon of DOC-2, using first- Only a single methionine is present in the first exonstrand cDNA prepared from small intestine, arm skin, of DOC-2, and its location is identical to the predictedfallopian tube, PPC1 (a prostate cancer cell line), and translational start site of p96. Only 1 of the 31 aminoSF15-2 (a derivative of PPC1 exhibiting reduced malig- acids encoded by the first exon of DOC-2 varies withnant growth characteristics as a result of introduction the mouse sequence. However, the DNA sequences im-of normal human chromosome 17q (Murakami et al., mediately upstream of the translational start sites1995)) as template. No products were observed with show extensive divergence. Also, the sequences rele-doc2-5A, while doc2-5B showed amplification using tem- vant to translational initiation show some degree ofplate prepared from fallopian tube only. Both doc2-5C variation between the two species, and each conformsand doc2-5D gave rise to PCR products from each of the only moderately well to the consensus sequence pro-templates tested (data not shown). Sequence analysis of posed by Kozak (1991). Clone 1RA1 has a poly(A) tail atseveral of these PCR products confirmed the high degree its 3 '-end, but the presumptive polyadenylation signalof homology between the 5'-end of DOC-2 and that of approximately 15 bases upstream of this poly(A) tailp96 and extended the DOC-2 ORF by an additional 146 (at position 3185) is somewhat degenerate (AACAGAbp (the locations of doc2-5B, -5C, -5D, and 401rp are as opposed to AAUAAA), although it retains the correctindicated in Fig. 1). The full-length DOC-2 cDNA se- composition of five purines and one pyrimidine. Thequence, along with the genomic sequences from the 5'- possibility that this is not an artifact of 1RA1 is sup-end of the DOC-2 gene, are available from GenBank ported by several expressed sequence tags (EST) (e.g.,under Accession No. U39050. Further sequence compar- R37515) that share the same polyadenylation site.ison between the cDNA sequence and the genomic se- However, investigation of several other retrieved ESTquence of DOC-2 enabled us to determine the genomic sequences (e.g., R63200) indicates that an additionalstructure of the amino terminal third of the gene, includ- polyadenylation site (at positions 2787) may also being the exact positions of the first eight intron-exon used (see Fig. 1).boundaries (see Fig. 1). Evidence indicates that at least one variant of DOC-

2 mRNA exists, probably as an alternative splice form.Homology of DOC-2 with Murine p96 and Discovery When we compared the retinal cDNA clone, 1RA1, to

of Alternative Splice Forms the ovarian clone (L16886) identified by differentialdisplay, we noted the absence of a 60-bp exon between

Comparison of the predicted amino acid sequence bases 729 and 789 in the cDNA sequence. Comparisonof the DOC-2 protein with sequences present in the of this presumed splice variant with the databases re-NCBI database revealed very high homology with vealed that the same splice variant of the mouse p96murine p96 (U18869). The homology with p96 was gene exists, under the name p93, almost certainly con-strongest at the amino terminal of DOC-2 corre- firming the variant as an alternative splice form (seesponding to the phosphotyrosine domain (Bork and Fig. 1).Margolis, 1995), but appeared to weaken in the cen- A second sequence variation was also detected as a

210 AIBERTSEN ET AL.

0 00 0 0 0 0 0 0 0 0 0 0 a 0 0 0 a 0 0 0 0 0 0 0N 4 N ' 4 4 w' 0 w4 0 N '0 4 0 0 N .0 40 4 0 N

0 N N N Nv

a. r~ E- 00 C 0100 0 0 0o 0 14 <14

U >0 u1-0U 0 C 4 E- W 0 0C <E .a 0 E. w cd 0 0 .w - o c

0. >. (Do 4. 04 0w U 0. u m. -CE-4 U

<.4 0v u- -0 04. 0 44 Go 1'- -9 01 CO 05 4<0A a>L .4 04.

u C: c 14, 0E . 0 4 u 1. m 0 u .4. <0 (4- ý w, L uu0 u 4

0w ý>. 00 M.0 4 - - .-4 4 u 0 4. uo .j a 4 1

0 0 4. 1-4.01-0 < . 0 0 ~ ~. 00060. t 0ý 0 .

V)0 0 0 0 ' So uS4.0 L

<~~~ <4- 0. . 0 .

00 0En <.4 00. 00-C a 02 0< 40

0. 0C4 4 0.2 00 00 0< 00 6> 00 :0 1- 40c (0 00j 0 144 1

0 41 'C 10,4 U0 14 U0 so <0 00 a >O U4 1- V0x

0 a u. 0 0<Z~ ~~ 0<U 0 0 ~ ~ 4 ~ 0 0 ou

.. 0 0- 00 ~0 0 0 <> U I0 o0 j0 a.4

~0 - 0 0 < 0 0> 00. 0w .2 <d 4 <

:30 00 0 00 00 -C 0* 09 0 17C 04w 4 4 2ýv C(1<40 1 4 0.4 <4 41 14 U 04 U 0C 0- 0 4. 8 1

u -- < 0 < 0 0 00 0 1-0u U m 0. < I- 0 ,. 0< a. m U uu 0.2 [4 0 V ( 0< 00 W < 0> 0Q .U C4. 04W t0 Ww1 0 0.

.2 00. 00 11 0> 00 0. 0 0. <1 <4 00. 00 00. 0ý

t0 c U t.4 <0 00 0 0. E4~- In 00 4 0) 1400 0

r- E-.w 0 E- 0 1-u< 19 El2 00~ o'4C :1 t2 00 §. ~0> 00 0 u0 0o U) u N0 0,

01 0p E0 U- w <>1w ' 0

E4. 0.u w Q 0 00 .4C 00. 0 e0 < 0 w04.00

-C 4 Z0 000 <1) 0 04 04 <'A P. 4. A. ul 0oa1 -.4 00 <- 00 00 <0 0< 0.ýIl : 4

wo 00 14 00 <0 <0 0 >14 4 0

U. C4 In <.0 0> 04. 04 u.- g0 u. E0< 0<<*.

.4< <2 00 00. 0 W0 00.4 0 o 0c <. 04. -- 0 00 1- 1-4 0 0 00 <. < 4 0> 0 <> 01 41 L, 0 1

0. <0 N- 0.2 00 14. F. 0 10 1- ~ . ~4~ 4 < O ~0 1 00 0< 0>00 0

0 00 Do 00 00 00 04 0 14'0 1

d-. L0 to4 u- w E. w< . 1 u 0 8 0V~~~~1 0 lE 4 jE

>iA . 4 04 0.0 > 40 <-4 4 V) 00 u4 04.u ) m u V) 1- u

00 0. 0< 00 00 <0 00 00. 0 < L0

<4. 04:U . 0, 00 0 0 <0 0 E0. W: o0: 0> t 4) c:1 u c C

0<1 aw. 00' . w0 00 0ý 0.2 1. <0 00U< 20 14 0. 4. a4 <4 14>) Qd >4 0 <0 4

.E0 >>0 En u 4 .

V 0 1 - ~ 4 - .~ 0 8 >00 ~0 0) <4 ý4 4~u .4 0. <00. 0 00 0 0 0> 0 <1 0< > 00 04j o 0 o .0 00 <0' U 0 < - 0 14 <. < 4 "

<4 < w < u < 0.0 4 a-4 C 'A

<4. <w o40 m0 a).1E.. 014>u<0 0<4E <. 0 a0 0 0c 0~ ~ 0 o ~ o ~~ 0. < 4 0 <

0 ~ ~ ~ ~ 1 0A u w N4 U.10 4 00 -0 0-04 0- 04 0480~ 01oc 40 H 0 x 00 00.: (A 0- 0 140 00 00 0> 0000 0< 0 0. 00 0 40. <

0 < 1- <4 0 < 1-0I0. 14. 0. 0 14>40 1- 0 0 0 10, 40 0- <0 ILI) 0- 14. 04 U0 0:)- 0 04 r4 N. 0 0t 0) 0ý( t

00 00 0. u i0 -> 0 0 .4 0 t<0 u ; 00 <4 04. <4 0w <. 00u 14<0 1. 00 00 0 1 0<0.0 0. 0.0 <0 1- <.4 04. u. 00 04 U M: <0

0. 00 <1 <14 < 0< 0> 00 00 :. < 0 00 0 r 8 0 < (1 C6 0L 't u

00.~~~~ 0 4- - 0 10 0 0 0 < 0 . 0 0 0 . 0.2 0< . 0t

1-0 E< 00 0E- 14 10 14. 14C- 0 <0 1- 0 140<

Q) E, 0 ý 0< 0< 00 00-4 mw 0 0 >0 wo s.o (7 <0 0. 0l 14 000

,c 4. 0< 0. a4 r E 0 r4 .

L,.4 00 <0 < 0> 00 00 < 4 0< <1 0S<0. 0.O < 0 < 0 0

90 .3 1) 01 00 L) 04 . u 0 00< 4

0> 'i 04u < 40 0 .0 <-u,~ 1

0> .4. U. 0. 00 0> E 0> 4. o 4 E 0 0 0 0 u l

04. .0 EC < 4 ~0 U4 M U. $400 02 01 0 . 4. 0 0 . 0 0 < 00 0. 0 0 0 1 0 .c 0 0 u4 0 0 0

<.u 0 > D 00 .4 <00 04 . <40 04 00 0'o 0 >0 <.- 04 0.m4.00 v. <0 w< <2 <V2 .< v < 00. < 14. C< w0 1-1 14 o 00 0. u<

00 00. <0- 00 0 < 0 0>40> ~~~4 .~4 4 00 <- 14I 0 0.d0 14 0 . 0>- 8 4. 12

0 U 0.0 ., <0. a4 00 0a, <4 < 0- 0 .3 U 0- 04 0 a ~ 0 0. 04 1 14 1- 14

~~~C Ic 02 002 <E4 04 4 0 4 4 ~ ~ 0. 0

0 t4 X: 04 -4 ot 4> U a. 4 0 >

GENOMIC CHARACTERIZATION OF HUMAN DOC-2 211

1 100

MI10.5 MEMXTGRSCA PKGADHDFTL LSHSIPHHHL TILFGLTFVP KSFVILLVFF YCSLETMAQK SDISVETANA TSGKPNPPSP KSRLAXLKRT KKASNASSDPDab ............. .......... .......... ........ MV KSLV ...... . AKLST..AS SNLSLA... S TFGGGSGAAE ETNYAKHR ... ....... NDP

DOC-.2 ........... ........... ........... .......... ........... ......... M SN.EVET.SA TNGQPDQQAA PK..APSKKE KKKGPEKTDE

101 200

M110.5 F. . RFQNNG ISYKGKL'GE QDVDKARGDA MCAEAMRTAK SII.. .KAAG AI!KTR. .ITL QINIDGIKVL DEKSGAVLHN FPVSRTSFIA RDSSDARArG

Dab G... .RFFGDG VQFKAKLIGI LEV. .ARPEV IGC.ARRRCK ISK.. .WHPG GWRAQAAITI HVTIDGLRLR D9KTGDSLYH HPVHKISFIA QDMTDSRAFG

DOC-2 YLLAa7KGDG VKYKAKLIGI DDVPDARGDK MSQDSMMXLK GMAARGRSQG QHKQR. .WV NISLSGIKII DEKTGVIEHE HPVNKISFIA RDVTDNRAFG

201 300

M110.5 LVYGEPGGKY KFYGIKTAQA ADQAVLAIRD MFQVVFEMKK KQIEQVKQQQ IQDGG ......... AEI. SSKKEGGVA VADLLDLESE LQQIERG...

Dab YIFGSPDSGH RFFGIKTDKA ASQVVLAMRD LFQVVFELKK KEIZ.MARQQ IQGKSLHDHS SQLASL.... SSLKSSGLG GMGL .................

DOC-2 YVCGG.EGQH QFFAIKTGQQ AEPLVVDLKD LFQVIYNVKK KEEE... KKK IEEASKAVEN GSEALMILDD QTNKLKSGVD QMDLFGDMST PPDLNSPTES

FIG. 2. Homologies between the 140 N-terminal amino acids of DOC-2 and genes from two widely divergent species, D. melanogaster(Dab gene) and C. elegans (M110.5 gene). Identities, highlighted in boldface, were found using the computer algorithm PILEUP with theparameters GapWeight set to 2.0 and GapLengthWeight set to 0.1.

result of the database searches. In this case we found tion of the DOC-2 mRNA using RT-PCR, DOC-2 mRNAthree ESTs (Accession Nos. R80479, R81944, R63199) is known to be present in normal ovarian epithelialthat differed from IRA1 in the section between base- tissue (Mok et al., 1994). We also can deduce that DOC-pairs 2580 and 2598 in the 3'-untranslated region of 2 must be expressed in cells from brain stem and fetalDOC-2. The differences were in all cases consistent retina, since we isolated cDNA clones from librarieswith an 18-bp inversion. We found a 5-bp palindrome made from these tissues. We have thus determinedflanking the inversion on both sides, which could be that DOC-2 is expressed in at least seven different hu-responsible for the inversion (see Fig. 1). We also found man tissues. In mouse, p96 expression has been foundevidence for a 70-bp duplication in the genomic se- in differentiated and an undifferentiated mouse stemquence during the search for exon boundaries. This cell line (this study) and in mouse macrophage andduplication, which includes part of the alternatively brain cells (Xu et al., 1995). Combined, these observa-spliced exon 8, spans the region between bases 10890 tions give the impression that DOC-2 and its murineand 10942 and was inserted in the same orientation homologue, p96, are expressed in a tissue-independentbetween bases 10693 and 10745, approximately 160 manner.bases upstream of its original location (data not Chromosomal Mapping of DOC-2shown). Eleven percent variation between the dupli- We were able to determine the genomic location ofcated sequences (8 of 70 nucleotides) suggests that the DOC-2 to chromosome 5p13 by fluorescence in situ hy-event is quite ancient. bridization using the P1 clone 35G12 as hybridization

probe against metaphase chromosomes from a normalDOC-2 Protein Homology to Other Genes male (see Fig. 3). The map location of DOC-2 was inde-

Two other proteins in the sequence databases pos- pendently confirmed by analyzing Coriell human/ro-sessed significant homology to the amino terminal of dent somatic cell hybrid mapping panel 2 in conjunc-DOC-2. Caenorhabditis elegans gene M110.5 (Z49968) tion with the Coriell regional mapping panel of chromo-showed the highest degree of identity, with 47% in the some 5 (NIGMS, Camden, NJ). Using the oligo primerspresumptive PID, followed by the Drosophila melano- doc2-B and doc2-C, only cell lines containing the hu-

gaster disabled (Dab) gene (L08845), with 43% identity man chromosome 5p13 region provided PCR-positivein the same region. Figure 2 shows the result of the template (data not shown). No amplification was ex-

multisequence alignments of the amino-terminal ends pected from the rodent homologue, since the primers

of M110.5, Dab, and DOC-2. used in the experiment were located in intronic se-quence.

DOC-2 Tissue Specificity CONCLUSION

In addition to the three normal tissues and the two In summary, the full-length sequence of the humanprostate cancer cell lines where we detected transcrip- DOC-2 gene has been ascertained, together with its

FIG. 1. Nucleotide sequence and predicted protein translation of the human DOC-2 gene. The first eight exon boundaries are indicatedwith vertical arrows. The location of the oligo primers (doc2-5B (only the last three bases), doc2-5C, doc2-5D, and 40Irp) used to verify the5'-end DNA sequence of DOC-2 are shown with bold underlines of the DNA sequence. Two other regions in the translated part of the geneare underlined: the location of the 140 amino acid segment with the phosphotyrosine interaction domain (PID) (solid line under the aminoacid sequence), and the location of the alternatively spliced form of DOC-2 (dotted line under the amino acid sequence). Three underlinedsequences in the 3 '-untranslated region of DOC-2 indicate the region of inversion and the two functional polyadenylation sites. The smallestpossible inversion is indicated in boldface with the flanking 5-bp palindromes indicated with the underline.

212 ALBERTSEN ET AL.

;,j *

b

FIG. 3. Localization of a DOC-2-specific P1 clone, 35G12, using fluorescence in situ hybridization (FISH). (a) Illustration of a pseudocol-ored image of the localization of DOC-2 to band 5p13 of DAPI-counterstained normal male metaphase chromosomes. (b) Illustration of thesame metaphase with pseudo-Giemsa-banded chromosomes.

chromosomal location and the genomic structure at the son, M., Robertson, M., Bradley, P., Lawrence, E., Sheng, Z. M.,

5'-end of the gene. Sequence analysis on the DNA and Hoopes, R., Stemnberg, N., Brothman, A., Callahan, R., Ponder,B. A. J., and White, R. (1994). A physical map and candidate genes

protein levels revealed that DOC-2 is the human homo- in the BRCA1 region. Nature Genet. 7: 472-479.

logue of the murine mitogen-responsive phosphopro- Altschul, S. F., Gish, W., Miller, W., Myers, E. W., and Lipman,

tein, p96, with 81% amino acid identity. The sequence D. J. (1990). Basic local alignment search tool. J. Mol. Biol. 215:

comparisons also revealed that a 140 amino acid seg- 403-410.

ment at the amino-terminal end of the gene, which in Bork, P., and Margolis, B. (1995). A phosphotyrosine interaction do-

a separate study has been characterized as a potential main. Cell 80: 693-694 [Letter to the Editor].

PID, shares greater than 40% identity with genes iso- Groden, J., Thliveris, A., Samowitz, W., Carlson, M., Gelbert, L.,

lated from distant species like C. elegans and D. mela- Albertsen, H., Joslyn, G., Stevens, J., Spirio, L., Robertson, M.,Sargent, L., Krapcho, K., Wolff, E., Burt, R., Hughes, J. P., War-

nogaster. Given the apparently ubiquitous expression rington, J., McPherson, J., Wasmuth, J., Le Paslier, D., Ab-

pattern of DOC-2 in all cell types tested, together with derrahim, H., Cohen, D., Leppert, M., and White, R. (1991). Identi-

the potential of p96 for phosphorylation at serine resi- fication and characterization of the familial adenomatous polyposis

dues and the presence of the PID, we can hypothesize coli gene. Cell 66: 589-600.

that DOC-2 is a tissue-independent component of cellu- Kozak, M. (1991). An analysis of vertebrate mRNA sequences: Initia-

lar signal transduction. tion of translational control. J. Cell Biol. 115: 887-903.

Lee, W. H., Bookstein, R., Hong, F.,Young, L. J., Shew, J. Y., and Lee,E. Y. (1987). Human retinoblastoma susceptibility gene: Cloning,

ACKNOWLEDGMENTS identification, and sequence. Science 235: 1394-1399.

We gratefully acknowledge Dr. Suzi Mansour for providing mouse Lichter, P., Chang, C-J. C., Call, K., Hermanson, G., Evans, G., Hous-

embryonic stem-cell RNA, Margaret Robertson and Elizabeth Law- man, D., and Ward, D. C. (1991). High resolution mapping of hu-man chromosome 11 by in situ hybridization with cosmid clones.

rence from the DNA Sequencing Facility, Ed Meenen for synthesizing Science 247: 64-69.

DNA oligonucleotides, and Ruth Foltz for editing the manuscript.

Vysis generously provided the FISH analysis software. B.J.W. is Liang, P., and Pardee, A. B. (1992). Differential display of eukaryotic

supported by a fellowship from the American Foundation for Urologic messenger RNA by means of the polymerase chain reaction. Sci-

Disease, and S.A.S. is supported by an EMBO fellowship. This work ence 257: 967-971.

was supported by Grant DAM17-94-J-4129 from the Department of Mild, Y., et al. (1994). A strong candidate gene for the breast and

Defense to R.W. ovarian cancer susceptibility gene BRCA1. Science 266: 66-71.

Mok, S. C., Wong, K-K., Chan, R. K. W., Lau, C. C., Tsao, S-W.,

REFERENCES Knapp, R., and Berkowitz, R. S. (1994). Molecular cloning of differ-entially expressed genes in human epithelial ovarian cancer. Gyne-

Albertsen, H. M., Smith, S. A., Mazoyer, S., Fujimoto, E., Stevens, col. Oncol. 52: 247-252.

J., Williams, B., Rodriguez, P., Cropp, C. S., Slijepcevic, P., Carl- Murakami, Y. S., Brothman, A. R., Leach, R. J., and White, R. L.

GENOMIC CHARACTERIZATION OF HUMAN DOC-2 213

(1995). Suppression of malignant phenotype in a human prostate Smith, S. A., Holik, P. R., Stevens, J., Mazoyer, S., Melis, R., Wil-

cancer cell line by fragments of normal chromosome 17q. Cancer liams, B., White, R., and Albertsen, H. (1996). Isolation of a geneRes. 55: 3389-3394. encoding a second member of the disc-large family on chromosome

Shepherd, N. S., Pfrongner, B. D., Coulee, J. N., Ackerman, S. L., 17q12-q21. Genomics 31: 145-150.

Vaidyanathan, G., Sauer, R. H., Balkenhol, T. C., and Sternberg, Xu, X.-X., Yang, W., Jackowski, S., and Rock, C. 0. (1995). CloningN. (1994). Preparation and screening of an arrayedhuman genomic of a novel phosphoprotein regulated by colony-stimulating factorlibrary generated with the P1 cloning system. Proc. Natl. Acad. 1 shares a domain with the Drosophila disabled gene product. J.Sci. USA 91: 2629-2633. Biol. Chem. 270: 14184-14191.

Am. J. Hum. Genet. 56:484-499, 1995

A Strategy for Constructing High-Resolution Genetic Maps of theHuman Genome: A Genetic Map of Chromosome 17p, Orderedwith Meiotic Breakpoint-Mapping Panels

Steven C. Gerken,' Hans Albertsen,' Tami Eisner,' Linda Ballard,' Pilar Holik,' Elizabeth Lawrence,'Mary Moore,' Xuyun Zhao,' and Ray White"2

'Department of Human Genetics and 2Huntsman Cancer Institute, University of Utah, Salt Lake City

Summary leles. Haplotype construction (Kelsell et al. 1994) and link-

Genetic linkage analyses with genotypic data obtained age disequilibrium studies (Bowcock et al. 1994; Jorde etfrom four CEPH reference families initially assigned 24 al. 1994) in affected pedigrees can lead to high-resolutionnew PCR-based markers to chromosome 17 and located genetic maps over regions of a few centimorgans, but such

studies are often limited by the number of informativethe markers at specific intervals of an existing genetic map meioses available. Advances in the positional cloning ofof chromosome 17p. Each marker was additionally geno- nes have betyped with an ordered set of obligate, phase-known recom- with cloned inserts of human genomic DNA (Burke et al.binant chromosomes. The breakpoint-mapping panels for wt lndisrso ua eoi N Bree leinantchrfami onsimes.Thed oftwoparentsone s hanl nonre 1987). However, the lack of high-resolution genetic mapseach family consisted of two parents, one sib with a nonre- of the entire human genome means that new markers andgate recombinant chromosomes. The relative order of new maps must be developed for each region surroundinggatrkecombinant detromined sorting Tegrelativonrtern o a locus of interest, before the minimal physical region canofnewmarkers w ande orermined anhorting mgrkersgationd bbe determined for molecular analysis. A genetic map of the

human genome, at a resolution the same as or higher thanimizing double-recombination events. Consistency of seg- that of current physical maps would facilitate integrationregation patterns with multiple flanking loci constituted of the physical and genetic maps and would acceleratesupport for order. A genetic map of chromosome 1 7 p was cloning of the minimal region of the genome necessary tocompleted with 39 markers in 23 clusters, with an average isolate any gene of The physical map constructedspace of 3 cM between clusters. The collection of informa-tive genotypes was highly efficient, requiring fivefold fewer by Cohen et al. (1993) provided the first large-scale effortgenotypes than would be collected with all the CEPH fai- to integrate a genetic map of the human genome with thelies. Given the availability of large numbers of highly in or- physical map. Since some 10,000 genetic markers, -4,500

aGiven C-ed mvailabilityoflark enumbers, me c b p inof which are PCR based, have been reported to the Genomemative PCR-based markers, meiotic breakpoint mapping Data Base, it would appear that sufficient numbers ofshould facilitate construction of a human genomic map markers are available to develop a 1-cM genetic map span-with 1-cM resolution.makraravialtodvlpa1cgetcmpsan

ning the 4,000-4,900 cM of the human genome (NIH!CEPH Collaborative Mapping Group 1992; Gyapay et al.1994). The need for a high-resolution genetic map of the

Introduction human genome will become increasingly urgent as geneticlinkages are discovered for loci associated with the numer-

Recently published genetic linkage maps of human chro- ous human genetic disorders whose etiology is still un-mosomes are the result of an international effort to map known (McKusick 1992).and sequence the human genome (NIH/CEPH Collabora- Multipoint linkage analysis with genotypic data collectedtive Mapping Group 1992; Gyapay et al. 1994). These from the CEPH reference families is a commonly usedmaps provide the information necessary for general linkage method for building genetic maps. To satisfy the statisticalstudies, including localizing genes with disease-causing al- requirements of linkage analysis, genotyping of an exten-

sive collection of individuals is necessary. An alternativeapproach to building genetic maps is meiotic breakpoint

Received July 26, 1994; accepted for publication November 17, 1994. mapping, which utilizes known meiotic recombinant chro-Address for reprints and correspondence: Steven C. Gerken, Ph.D., mosomes to determine marker order on much smaller sam-

Department of Human Genetics, University of Utah, Salt Lake City, UT pie sizes (White et al. 1985; Fain et al. 1989; Weber et al.84112

40 1995 by The American Society of Human Genetics. All rights reserved. 1991; Maestri et al. 1992; Sprinkle et al. 1993; Attwood0002-9297/9S/5602-0018$02.oo et al. 1994). The use of recombinant chromosomes as map-

484

Gerken et al.: Meiotic Breakpoint Map of I 7 p ,485

ping panels to construct high-resolution genetic maps has total of 39 markers, were ordered for the 65-cM region ofbecome routine for mapping murine loci and disease-caus- chromosome 17 p, using mapping panels that representeding genes in humans (European Backcross Collaborative a fivefold reduction in the number of genotypes collected.Group 1994; Kelsell et al. 1994). In this approach, chromo- Material and Methodssornes with meiotic breakpoints known to reside betweenpreviously mapped loci are selected for genorvping. Segre- Genetic Markers

gation patterns of a test marker are compared with the Twenty-four new PCR-based genetic markers, isolatedpatterns of anchor markers that flank the breakpoints, in a for chromosome 17 p, as part of a program to developmanner that minimizes the number of double-recombinant large numbers of PCR-based probes that detect humanchromosomes. Segregation patterns of multiple test mark- polymorphic loci, and three previously published PCR-ers located in the region allow for the ordering of the test based probes (2KZ14-B4, 2KZ22-B10, and UT172) thatmarkers relative both to each other and to the known detected dinucleotide-repeat polymorphisms (table 1) (Utahanchor markers. The rationale is that, although typing of Genome Center, personal communication) were used inall members of the CEPH panel would provide genotypic this study'. The new PCR-based probes were developeddata for all of the meioses necessary to order new markers from the DNA sequences of clones containing microsatel-and for an estimate of the recombination fractions, using lite repeats detected by hybridization to end-labeled oligo-only the informative recombinant chromosomes localized nucleotide probes d(CA)20, d(AAAG)20, d(AAAT)20, orto the region of the test locus constitutes a more efficient d(AGC)20 (Melis et al. 1993). Twenty-six cloned DNAapproach to ordering new markers. fragments from chromosome 17 that were used as molecu-

Human chromosome 17 consists of 92 Mb of DNA, lar probes for detecting RFLPs or VNTRs were derived-3% of the human genome. This chromosome is rich in from previously published data (table 1).

loci involved in cancer; for example, it harbors the tumor- Genotyping and Data Collectionsuppressor genes p53, NF1, and BRCA1, and loss of bet- Genotypes for the IFLP and VNTR probes were ob-erozygosity of other loci on chromosome 17p has been t ype s fo r the data and Vntr obe were obobserved in breast carcinomas and uterine cancers (Isomura rained from the data used to construct the genetic map of

et al. 1994; Jones et al. 1994). Numerous other disease chromosome 17.(O'Connell et al. 1993). Autoradiographsof the originaml Southern blots were available for review for

alleles have been mapped to this chromosome, including these markers, since all of the genotypes had been deter-

those responsible for the neurological diseases Charcot- mined in our laboratory. Four of the CEPH reference fami-

Marie-Tooth type IA (CMT1A) and hereditary neuropa- lies (1331, 1332, 1362, and 884) were genotyped with

thy with liability to pressure palsies (Fain 1992; Chanceand Pleasure 1993). More than 563 probes that detect every PCR-based marker except UT18 and UT49, whichandPoleasurpce 1993). More than563prof whitha 227aet were typed in families 1347, 1362, 1416, 1454, and 1463.polymorphic loci on chromosome 17, of which 227 are Genotyping for the PCR-based markers was performed asPCR based, are available. Almost 300 of these markers described by Melis et al. (1993).have been typed in the CEPH families (CEPH database, DNA for the mapping panels was aliquoted into masterversion 6), and, on the basis of these data, several genetic DNA trays, which were then dispensed in a Beckman Bio-linkage maps of chromosome 17 have been developed mek 1000 pipetting robotic workstation to minimize sam-(O'Connell et al. 1993; Gyapay et al. 1994). pIe errors. Genotypes were collected for each marker, with

The large number of available genetic markers and access the mapping panel for which the most likely location ofto a genetic map composed of markers typed in 59 CEPH the test marker had been determined (fig. 1). For those locifamilies made chromosome 17p an excellent choice for the test maker hadien dtried (i 1)corithos oiconstructing a high-resolution genetic map of a whole chro- whose most likely location showed no recombination with

I* an index marker, genotypes were collected with the panelmosome arm, using recombinant-chromosome mapping for the most distal interval. If the marker was nor orderedpanels. In this report, we describe a two-tiered approach within this interval, genotypes were collected with the panel

to high-resolution m apping that reduces the num ber of for the n tervaltgeike yy esnt erva l.

genotypes necessary to order a new marker with respect to f mages of th e intera l.

an existing map. Likelihood analysis was used to assign Images of ste a nd were digitized to anZ1 image-capture system and were transferred to a computer-

new markers to chromosome 17 and to determine the most asitdgnypgssemAlgnopswreeemndassisted genotyping system. All genotypes were determinedlikely intervals (defined by flanking index markers) for their independently by two individuals. Once genotypic datalocations on the index map. Using two new computer pro- were verified, they were transferred to the Utah relationalgrams, RBUILD and CSORT (Eisner et al. 1995 [in this database for storage.issue]), we constructed mapping panels of recombinantchromosomes and ordered new markers on the basis of Meiotic Recombinant-Chromosome Panels

analysis of the genotypic data collected from these mapping RBUILD, the computer program for selecting break-panels. Twenty-three distinct clusters, which comprised a point-mapping panels, uses pairwise linkage analysis of

00ý (7 mN a, ON N 0 0 C 0 N " N

'0* ~ ~ ~ ~ ~ ~ ~ ~ ~ r cN 'C N'o N " " C 0 0 00 O 0 0 N ' N O

N~ ~~~~~ N'N -C1 N 00 ' N

C, 0 - 8 ' C ' 0 N '

N4 ON -0 N - - ' N ' ON 'N " "N 'N cm ON, ON N 0 N'

~ N - '0 N - N 0 '0 '0~ 00<

"N~~~ "N "Ný <'

N O N N O N

< -u~u00 "N 0 N 0 0 " 0 0 'N "N 0 0 0 0 'C 00 '

U U < '0U

U H F- < uU U U U U < U U < U U < U, <, t UU

H <<U U < UU

U <U U - U <U U <uFU -,U U u u U. 0. .UH~~~~~ U UH~ U~U H U

U <<~< U~*HU H << Up.U U<<

uu-. ý<UUu us ý < U UUýUUU<UU UUU U~r << ~'UHu U<U~ <tUUU

<H<Uuu<< U<U<UH<U u- U < <k u '02uu ~ u u-ý IUUU u<U u -U t ~ < 0 '

*uU< <H<uu<< U<<,<uý<~U<<U u<HU.)U <<U<UUU<~ <-UU U <UU <U U uuýu

ýUU<UUU uu~u<U<<U< HUUUUUUUUFFýuU Uh'. -UUU<UU<< u v< _u _

v-~o 1< U < CD _M aare) U < <U U <U U aH a\tU In~U InU UH <U m n C)

a,~~~~~- Cc \ON a\0 CN a , N O , a, a

"m N0 14 NN " N " C ' C ' N 'C 'C ' N N 0 - N '0 ' o < O

ON ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ L ON? N NN " ' 'N ' ' ' ' N:NN " N~<

ON O ON ON N ON ON N O ON N O ON ON ' 'C 00 0 ON ~ O0 'Z B

ON'" 0 'o -,4

N N 0 -NN C00 ONO A N * 04 O0'*'0 u

0' N 0 - N 'C 0 N N 0 O ON - 0 04O8O

Gerken et aI.: Meiotic Breakpoint Map of 17p , 489

markers whose order is known, relative to the breakpoints, Table 2allows for orientation of the test markers, relative to theanchor markers. Pairwise Analysis of Linkage Between 24 UTAH Markers and

Genethon Markers: Lod Score Z_,_ at RecombinationUnder a model of complete positive interference, the Fraction Theta

phase of each marker for each recombinant chromosometyped was determined by comparing the pattern of inheri- UTAHtance of the recombinant chromosome with that of the Marker CEPH Marker Theta Z-..

nonrecombinant chromosome. This pair of chromosomes UT18 D17S849 .0010 4.8is referred to as a "sibship pair." Sibship pairs were ana- UT20 D17S796 .056 12.8lyzed to determine whether the parental origin of each UT25 D17S783 .001 13.5

test marker was the same, different, or unknown, when UT39 D17S796 .001 15.6

compared with the nonrecombinant chromosome. The UT49 D17S796 .001 5.4

summary of comparisons for all markers for each recombi- UT65 D17S786 .043 9.4UT6D17S798 .001 6.0

nant chromosome (i.e., a "haplotype" of the parental con- UT72 D17S786 .024 10.6

tributions to each of these chromosomes) is referred to as UT137 D17S804 .001 6.3

the "chromosome composite" (Eisner et al. 1995). UT146 D17S798 .001 12.3

The two markers for which the greatest number of sib- UT158 D17S786 .047 8.2

ship pairs are informative form the first interval of the UT159 D17S783 .001 6.6UT184 D17S796 .037 9.5

breakpoint map. The recombinant sorting program, UT222 D17S786 .050 8.3

CSORT, tests the next-most-informative marker in each UT225 D17S796 .012 20.8interval of the map, including the ends of the map, to UT263 D17S783 .001 12.6

determine whether a double-recombination event would be UT263 D17S805 .001 12.6UT269 D17S849 .001 12.6required for the marker to be located in this interval. The UT403 D17S786 .040 11.7

test marker is placed in the interval of the map where no UT405 D17S799 .026 9.4double recombinant is required to explain the data. This UT751 D17S796 .001 19.8process is iterated until all markers are tested. Markers that UT1860 D17S783 .001 12.6

are excluded from all intervals are included in a secondary UT1985 D17S796 .001 7.2UJT5265 D17S796 .001 13.8map. Next, the chromosome composites are shifted to or- -T5265 _ D17S796 _ .001 _ 13.8

der the breakpoints from right to left to provide an arrayof chromosome breakpoints whose order has been deter-mined by the recombinant-sorting algorithm, significant identity to the Utah microsatellites were per-

The foregoing technique detects possible genotypic errors formed with the GCG program QUICKSEARCH, and re-in the data for a test marker when an anchor marker is sults were displayed with the program QUICKSHOW.ordered incorrectly or excluded from the breakpoint map.Autoradiographs for the test marker are then checked for Resultserrors. If placing a test marker has caused an anchor markerto be excluded or incorrectly ordered, it is removed from Genetic Markersthe breakpoint map and the analysis is restarted. Markers Twenty-four new PCR-based probes were assigned tothat cannot be ordered within the breakpoint map but chromosome 17 (table 1) on the basis of results obtainedwhose location scores show high likelihoods for inclusion by two-point linkage analysis (table 2). The microsatellitewithin an interval are reordered with only the anchor mark- repeats detected by these probes included 10 d(GT), 8ers for the interval. Segregation patterns that exclude the d(AAAG), 3 d(AAAT), and other motifs (table 1). DNAmarker from certain locations and that minimize the num- sequences of the primer pairs, the primers to end-label forber of double or multiple recombinants determine the loca- visualization by autoradiography, and the optimized condi-tion of the test marker. tions for PCR are reported in table 1. Accession numbers

are provided to facilitate retrieval of the DNA sequence ofSequence Analysis of Sequence-Tagged Sites (STS) each STS from the GenBank sequence database.

DNA-sequence databases representing all STS-con- Estimates of heterozygosity for the polymorphic locitaining microsatellite repeats from the Utah, G~n&hon, ranged from .49 to .91, with an average heterozygosity ofCHLC, and Marshfield collections were constructed using .68. Heterozygosities for the dinucleotide repeat-con-the GCG (1991) program DATASET. DNA sequences for taining loci averaged .65; and those for tetranucleotide re-the G~n&hon, Marshfield, and CHLC markers were re- peat-containing loci d(AAAT) and d(AAAG) averaged .59trieved from the GenBank sequence database (version 78) and .84, respectively. The autoradiographic images ob-or the CHLC server. DNA sequence analyses for STS with tained with PCR probes for the loci containing tri- and

" , Table 3

Meiotic Recombinant-Chromosome Mapping Panels

Chromosomes ChromosomesCEPH Family' Selected' CEPH Familya Selected'

Obligate recombinant and nonrecombinant 1357 Nonrecombinant ......... *14 paternalchromosomes for the interval 144D6-LB17.8: 1357 Nonrecombinant ......... 4 maternal

13281 Recombinant ........... *3 paternal 1362 Recombinant ............... *4 paternal13281 Nonrecombinant ...... *4 paternal 1362 Recombinant ............... *5 paternal13291 Recombinant ........... *9 paternal 1362 Nonrecombinant ......... *3 paternal13291 Nonrecombinant ...... *8 paternal 1413 Recombinant ............... '9 paternal13293 Recombinant ........... 3 paternal 1413 Recombinant ............... "14 paternal13293 Recombinant ........... *7 paternal 1413 Nonrecombinant ......... "12 paternal13293 Recombinant ........... *9 paternal 1416 Recombinant ............... 4 paternal13293 Nonrecombinant ...... *8 paternal 1416 Recombinant ............... 6 paternal13294 Recombinant ........... 6 paternal 1416 Recombinant ............... 10 paternal13294 Recombinant ........... 6 maternal 1416 Nonrecombinant ......... 8 paternal13294 Nonrecombinant ...... 3 paternal 1418 Nonrecombinant ......... *9 paternal

13294 Nonrecombinant ...... 3 maternal 1418 Nonrecombinant ......... *8 paternal1331 Recombinant ........... *6 paternal 1421 Recombinant ............... 7 maternal1331 Recombinant ........... *7 paternal 1421 Recombinant ............... *12 paternal1331 Recombinant ........... *16 paternal 1421 Nonrecombinant ......... "11 paternal1331 Nonrecombinant ...... *3 paternal 1421 Nonrecombinant ......... 3 maternal1332 Recombinant ........... *8 paternal 1423 Recombinant ............... *3 paternal1332 Nonrecombinant ...... *6 paternal 1423 Recombinant ............... *9 paternal1333 Recombinant ........... *8 maternal 1423 Nonrecombinant ......... *8 paternal1333 Recombinant ........... 10 paternal 66 Recombinant ................ 3 maternal1333 Recombinant ........... 10 maternal 66 Recombinant ............... 4 maternal1333 Nonrecombinant ...... 3 paternal 66 Recombinant ............... 6 paternal1333 Nonrecombinant ...... *3 maternal 66 Recombinant ............... 8 paternal1340 Recombinant ........... *3 paternal 66 Nonrecombinant ......... 3 paternal1340 Recombinant ........... 4 paternal 66 Nonrecombinant ......... 5 maternal1340 Recombinant ........... *7 paternal 104 Recombinant ............... 4 paternal1340 Recombinant ........... *8 paternal 104 Recombinant ............... 8 paternal1340 Nonrecombinant ...... *13 paternal 104 Recombinant ............... 10 paternal1341 Recombinant ........... *3 paternal 104 Nonrecombinant ......... S paternal1341 Recombinant ........... *8 paternal 1444 Recombinant ............... *9 paternal

1341 Nonrecombinant ...... *4 paternal 1444 Recombinant ............... *10 paternal1346 Recombinant ........... *7 paternal 1444 Nonrecombinant ......... *8 paternal1346 Nonrecombinant ...... *8 paternal 1447 Recombinant ............... *6 maternal1347 Recombinant ........... *3 maternal 1447 Recombinant ............... 7 paternal1347 Recombinant ........... *10 maternal 1447 Recombinant ............... *8 maternal1347 Nonrecombinant ...... *4 maternal 1447 Nonrecombinant ......... 3 paternal1349 Recombinant ........... *3 paternal 1447 Nonrecombinant ......... *7 maternal1349 Recombinant ........... 4 paternal 1454 Recombinant ............... *4 paternal1349 Recombinant ........... 6 maternal 1454 Nonrecombinant ......... *7 paternal1349 Nonrecombinant ...... *8 paternal 1458 Recombinant ............... 5 paternal1349 Nonrecombinant ...... 5 maternal 1458 Nonrecombinant ......... 3 paternal1350 Recombinant ........... *7 paternal 1459 Recombinant ............... *3 paternal1350 Nonrecombinant ...... *5 paternal 1459 Recombinant ............... *5 paternal1353 Recombinant ........... 15 paternal 1459 Recombinant ............... 7 maternal1353 Nonrecombinant ...... 5 paternal 1459 Nonrecombinant ......... *4 paternal

1354 Recombinant ........... *3 maternal 1459 Nonrecombinant ......... 4 maternal1354 Recombinant ........... *4 paternal 1463 Recombinant ............... *4 paternal1354 Nonrecombinant ...... *3 paternal 1463 Recombinant ............... *7 paternal1354 Nonrecombinant ...... *4 maternal 1463 Recombinant ............... *10 paternal1355 Recombinant ........... 10 paternal 1463 Recombinant ............... *12 paternal1355 Recombinant .................... 13 paternal1355 Nonrecombinant ...... 3 paternal

(continued)

Table 3 (continued) I I I

Chromosomes ChromosomesCEPH Family' Selected' CEPH Family' Selected'

Obligate recombinant and nonrecombinant 1400 Recombinant .................. 5 maternalchromosomes for the interval LB17.8-1C6G1: 1400 Nonrecombinant ............. 3 maternal(continued) 1427 Recombinant .................. *5 maternal

1358 Nonrecombinant ....... *9 maternal 1427 Recombinant .................. *6 maternal1362 Recombinant ............. *12 paternal 1427 Recombinant .................. *8 maternal1362 Nonrecombinant ....... *5 paternal 1427 Nonrecombinant ............. *7 maternal1377 Recombinant ............. 3 maternal 1582 Recombinant .................. *3 paternal1377 Nonrecombinant ....... 4 maternal 1582 Recombinant .................. *3 maternal1408 Recombinant ............. 4 maternal 1582 Recombinant .................. "4 maternal1408 Recombinant ............. 8 maternal 1582 Recombinant .................. "6 paternal1408 Nonrecombinant ....... 6 maternal 1582 Recombinant .................. *9 paternal1413 Recombinant ............. *7 paternal 1582 Nonrecombinant ............. "4 paternal1413 Recombinant ............. *10 paternal 1582 Nonrecombinant ............. *6 maternal

Obligate recombinant and nonrecombinantchromosomes for the interval IC6G1-fHu39.3:

13281 Recombinant ............. *4 maternal 1358 Nonrecombinant ............. "3 maternal13281 Nonrecombinant ....... *5 maternal 1362 Recombinant .................. *3 maternal13291 Recombinant ............. *6 maternal 1362 Nonrecombinant ............. *5 maternal13291 Recombinant ............. *8 maternal 1377 Recombinant .................. 8 paternal13291 Nonrecombinant ....... "3 maternal 1377 Recombinant .................. *9 maternal13294 Recombinant ............. *3 maternal 1377 Recombinant .................. *14 maternal13294 Nonrecombinant ....... *4 maternal 1377 Nonrecombinint ............. *3 maternal

1333 Recombinant ............. *6 maternal 1416 Recombinant .................. *4 maternal1333 Recombinant ............. *7 maternal 1416 Recombinant .................. "5 maternal1333 Nonrecombinant ....... *3 maternal 1416 Nonrecombinant ............. *3 maternal1340 Recombinant ............. *8 maternal 1418 Recombinant .................. *3 maternal1340 Nonrecombinant ....... *3 maternal 1418 Nonrecombinant ............. *6 maternal1344 Recombinant ............. *9 maternal 1421 Recombinant .................. *5 maternal1344 Recombinant ............. 10 paternal 1421 Nonrecombinant ............. *3 maternal1344 Recombinant ............. 11 maternal 1423 Recombinant .................. *3 maternal1344 Nonrecombinant ....... 3 paternal 1423 Nonrecombinant ............. *4 maternal1344 Nonrecombinant ....... *3 maternal 1424 Recombinant .................. *4 maternal1347 Recombinant ............. *7 maternal 1424 Recombinant .................. *8 maternal1347 Recombinant ............. *11 maternal 1424 Nonrecombinant ............. *5 maternal1347 Nonrecombinant ....... *3 maternal 104 Recombinant .................. *8 maternal1353 Recombinant ............. *4 maternal 104 Recombinant .................. *9 maternal1353 Nonrecombinant ....... *5 maternal 104 Nonrecombinant ............. *5 maternal1354 Recombinant ............. *9 maternal 1447 Recombinant .................. *3 maternal1354 Recombinant ............. *13 maternal 1447 Nonrecombinant ............. *4 maternal1354 Nonrecombinant ....... *3 maternal 1454 Recombinant .................. 7 paternal1356 Recombinant ............. *10 maternal 1454 Nonrecombinant ............. 3 paternal1356 Recombinant ............. "15 maternal 1456 Recombinant .................. *10 maternal1356 Nonrecombinant ....... *6 maternal 1456 Nonrecombinant ............. "4 maternal1358 Recombinant ............. *4 maternal 1458 Recombinant .................. *8 maternal1358 Recombinant ............. *5 maternal 1458 Nonrecombinant ............. *4 maternal1358 Recombinant ............. *9 paternal 1427 Recombinant .................. "4 maternal1358 Recombinant ............. *9 maternal 1427 Nonrecombinant ............. *9 maternal1358 Recombinant ............. *10 maternal 1477 Recombinant .................. *7 paternal1358 Nonrecombinant ....... *3 paternal 1477 Nonrecombinant ............. *3 paternal

"Recombinant" indicates that the specified sibling contains an obligate recombinant chromosome inherited from the specified parent; and

"Nonrecombinant" indicates that the specified sibling contains a nonrecombinant chromosome inherited from the specified parent."In specified sibling. An asterisk (") indicates an individual typed with PCR-based test markers.

493

W8-16ZEI

UJ6-toI

UtOI-99E______do -." I ____________________________

W9--EEE

dC-1t, I

Ui6-t" 1Ui6-ggtlIU16-89t -

UWt,-ZggI'W9-LrItI

09-991 1

UIS-Lztlf

wg-otcI

U19-OIM

U1L-ICEtWL-t6ZEI ___

-il-Et I

d6-T6gE =

dg-ltEI z=*

00 8 0L4 00

d - Ilii I I - III 0too -E E

dL 0Lg~

d6-L*d Clo *~

*Lo, iICS CrCC

Gerken et al.: Meiotic Breakpoint Map of 17 p ' 497,

one obligate recombinant; the multipoint map did not re- The achievement of a genetic map of chromosomeport the relative order of these two loci (Fain 1992). The 17p, with 39 markers organized into 23 clusters withorders of all markers flanking the CMT1A locus that were an average spacing of 3 cM between clusters, providescommon to both maps were also identical. For loci flanking a level of resolution that can be useful in either thethe NF1 locus, O'Connell et al. (1989) reported 1% recom- construction of or comparison with the physical map.bination between p3.6 and pHHH202, but we observed When the physical and genetic maps are integrated atno recombination between these loci. This discrepancy re- such a high resolution, a genetic map can be scaled inflected the fact that the recombinant in kindred 1333, indi- megabases, and it will be possible to compare recombi-vidual 9, which was observed in the study by O'Connell nation fractions with physical distances within the ge-et al. (1989), was not detected by the anchor markers used nome. For regions with unusual recombination featuresto identify obligate recombinants in this study. UT 172 was (i.e., high recombination fractions over small physicalreported to be located 1.5 Mb proximal to NF1 (Shannon distances), the boundaries of the recombination eventset al. 1994). Our recombinant panel also located UT172 will also have been mapped in the CEPH families.proximal to NF1 (table 5). The clusters of loci where the markers are not sepa-

An assumption of complete positive interference in the rated by recombination events could be due to the factsmall genetic intervals analyzed here is supported by recent that the appropriate recombinant chromosomes neces-evidence of interference in human meioses (Kwiatkowski sary to order these loci were not genotyped or that theet al. 1993; Weber et al. 1993). Interference minimizes recombination events are not randomly distributedmultiple-recombination events and makes the breakpoint- throughout this arm of chromosome 17p. Differences inpanel approach to ordering genetic loci more feasible sex-specific recombination at various regions of human(Copeland and Jenkins 1991). Observed double-recombi- and murine chromosomes have been reported (O'Con-nant events in our analysis of breakpoints on chromosome nell et al. 1987; Shiroishi et al. 1991). Determination of17p were most likely due to genotypic errors or to new the distribution of the chromosome 17p recombinationundetected mutations that converted alleles to those already sites, with probes for large numbers of polymorphic loci,occurring in the family. Errors in the collection and inter- will be necessary to test whether the distribution of re-pretation of genotypic data occur at a frequency of -1% combination events in the CEPH chromosomes are ran-in the CEPH database (Buetow 1991), while estimates of dom or clustered.meiotic gene conversions for STRP without recombination The isolation by different laboratories, of the samehave been estimated at 1:3,300 meioses (Weber et al. 1993). microsatellite repeat containing loci reflects the fact thatWhile the observed "double recombinants" are likely to be the number of polymorphic loci in the human genomegenotyping errors, they may also provide evidence of events is finite. Although the number of dinucleotide repeat-such as gene conversion. Although we were careful to mini- containing loci in the human genome is an estimatedmize the occurrence of errors in genotypic data, we would several hundred thousand, bias introduced by the use ofneed to retype the family members in question, using DNA biological cloning vectors, the size selection of inserts,from primary sources such as blood, to better understand and the length of microsatellite repeats influence thethe nature of the "double-recombinant" chromosomes. population of loci that will be isolated. DNA sequence

This approach to construction of high-resolution genetic analysis should be performed to determine whethermaps decreased the collection of genotypic data from 899 newly isolated loci detected by STS are identical to ex-individuals for the extended reference families to -180 isting loci. This method offers a new tool for the geneti-individuals for the present study. In the past, meiotic re- cist to determine whether two genetic loci are identicalcombinant panels have been used only for determining the and represents another approach to determiningorder of a few genetic loci, especially those with disease- whether errors exist in genotypic data.causing alleles. The approach and tools presented here Our observation that UT72 (D17S755) is identical inallow for a large number of loci and meiotic recombinant sequence to the G~nfthon marker D17S786 serves tochromosomes to be ordered in seconds of computer time anchor the two genetic maps of chromosome 17p at thisand is similar to that used to construct high-resolution locus. UT72, which is <28 cM from the most distallinkage maps of the murine genome (Guenet and Brown marker in this study, 144-D6, is reported to be 19 cM1993). The incorporation of "anchor markers," for which from the most distal marker on the Gýnfthon map. Se-order has already been well established, provides a level quence-known loci should expedite the integration ofof confidence that significant errors have not been intro- the physical and genetic maps, a notion proposed whenduced into the map. Simultaneous evaluation of large num- the STS was adopted as the "common language" of thebers of genetic markers typed in the meiotic chromosome- human genome program (Olsen et al. 1989).mapping panels is a logical use of the computational capac- If the genetic loci detected by the newly mappedity of the computer. probes were uniformly distributed and fully informative,

"0 Gerken et al.: Meiotic Breakpoint Map of 17p 499

O'Connell P, Plaetke R, Matsunami N, Odelberg S, Jorde L, Vanagaite L, Savitsky K, Rotman G, Ziv Y, Gerken SC, WhiteChance P, Leppert M, et al (1993) An extended genetic linkage R, Weissenbach J, et al (1994) Physical localization of microsa-map and an "index" map for human chromosome 17. Geno- tellite markers at the ataxia-telangiectasia locus at 11q22-23.mics 15:38-47 Genomics 22:231-233

Olsen M, Hood L, Cantor C, Botstein D (1989) A common Weber JL, Polymeropolous MH, May PE, Kwitek AE, Xiao H,language for the physical map of the human genome. Science McPherson JD, Wasmuth JJ (1991) Mapping of human chro-245:1434-1435 mosome 5 microsatellite DNA polymorphisms. Genomics

Shannon KM, O'Connell P, Martin GA, Paderanga D, Olson K, 11:695-700Dinndorf P, McCormick F (1994) Loss of the normal NF1 Weber JL, Wang Z, Hansen K, Stephenson M, Kappel C, Salzmanallele from the bone-marrow of children with type 1 neurofi- S, Wilkie PJ, et al (1993) Evidence for human meiotic recombi-bromatosis and malignant myeloid disorders. N Engl J Med nation interference obtained through construction of a short330:597-601 tandem repeat-polymorphism linkage map of chromosome 19.

Shiroishi T, Sagai T, Hanzawa N, Gotoh H, Moriwaki K (1991) Am J Hum Genet 53:1079-1095Genetic control of sex-dependent meiotic recombination in the White R, Lalouel J-M (1987) Interference and mapping function.major histocompatibility complex of the mouse. EMBO J In: Harris G, Hirschhorn K (eds) Advances in human genetics,10:681-686

Sprinkle TJ, Kouri RE, Fain PD, Stoming TA, Whitney JB III vol 16 LenuM, New Yr, pp 121i228(199) Cromsoml mppig ofthehumn CP gne sin White R, Leppert M, Bishop DT, Barker D, Berkowitz J, Brown(1993) Chrom osom al m apping of the hum an CN P gene usingC ,C l a nPe al( 9 5 C o s r ci n fl nk g m ps w ta meiotic crossover DNA panel, PCR and allele-specific probes. C, Callahan P, et al (1985) Construction of linkage maps withGenomics 16:542-545 DNA markers for human chromosomes. Nature 313:101-105

Inhibition of DNA methyltransferase stimulates theexpression of signal transducer and activator oftranscription 1, 2, and 3 genes in colon tumor cellsAdam R. Karpf, Peter W. Peterson, Joseph T. Rawlins, Brian K. Dailey, Qian Yang, Hans Albertsen, and David A. Jones*

The Huntsman Cancer Institute, University of Utah, Salt Lake City, UT 84112

Communicated by Raymond L. White, University of Utah, Salt Lake City, UT, September 23, 1999 (received for review May 5, 1999)

Inhibitors of DNA methyltransferase, typified by 5-aza-2'-deoxy- representational difference analysis, restriction landmark ge-cytidine (5-Aza-CdR), induce the expression of genes transcription- nome scanning, methylation-sensitive, arbitrarily primed PCR,ally down-regulated by de novo methylation in tumor cells. We and methylated DNA-binding protein affinity chromatographyutilized gene expression microarrays to examine the effects of (21-24). Another strategy, gene expression microarrays, is par-S-Aza-CdR treatment in HT29 colon adenocarcinoma cells. This ticularly suited for identifying candidate, methylation-silencedanalysis revealed the induction of a set of genes that implicated IFN genes and for assessing the downstream, cellular consequencessignaling in the HT29 cellular response to S-Aza-CdR. Subsequent of reactivating these genes. Microarray technology permits theinvestigations revealed that the induction of this gene set corre- systematic examination of thousands of gene expression changeslates with the induction of signal transducer and activator of simultaneously and has been used to follow the transcriptionaltranscription (STAT) 1, 2, and 3 genes and their activation by changes that accompany disease development and cellular re-endogenous IFN-a. These observations implicate the induction of sponses to environmental stimuli (25-29).the IFN-response pathway as a major cellular response to 5-Aza- In view of the clinical interest in 5-Aza-CdR and our incom-CdR and suggests that the expression of STATs 1, 2, and 3 can be plete understanding of the cellular consequences of inhibitingregulated by DNA methylation. Consistent with STAT's limiting cell DNA MeTase, we have utilized gene expression microarrays toresponsiveness to IFN, we found that 5-Aza-CdR treatment sensi- probe the effects of treating colon tumor cells with 5-Aza-CdR.tized HT29 cells to growth inhibition by exogenous IFN-m2a, Here, we show that 5-Aza-CdR inhibits the growth of HT29indicating that 5-Aza-CdR should be investigated as a potentiator colon carcinoma cells and that this growth inhibition parallelsof IFN responsiveness in certain IFN-resistant tumors. the transcriptional induction of IFN-responsive genes. Subse-

quent analysi s revealed induction of signal transducers andD NA cytosine methyltransferase I (DNA MeTase) recog- activators of transcription 1, 2, and 3 (STATs 1, 2, and 3),

nizes hemimethylated CpG dinucleotides in mammalian elements central to IFN signaling. Given the established growth-DNA and catalyzes the transfer of methyl groups to cytosine inhibitory properties of IFNs, our data offer a new model forresidues in newly synthesized DNA (1). The methylation of understanding the cellular consequences of inhibiting DNAcytosines within CpG islands located in core promoter regions MeTase.can negatively regulate the transcription of the adjacent genes.The basis for this negative regulation may involve recruitment of Materials and Methodshistone deacetylases to methylated CpG islands (1). Holliday Cell Culture and Drug Treatments. HT29 adenocarcinoma cellsfirst suggested a relationship between abnormal DNA methyl- (American Type Culture Collection) were cultured at 37°C in 5%ation and cancer (2). Subsequently, a number of methylation- CO 2 by using McCoy's medium supplemented with 10% FBSsilenced tumor suppressor genes, including p16Ink4a, retinoblas- (GIBCO). For treatments with 5-Aza-CdR, cells were exposedtoma, estrogen receptor, hMLH1, and E-cadherin, have been to 500 nM 5-Aza-CdR (Sigma) 24 hr after passage in completeidentified in cancer cells in vitro and in vivo (3-8). It is becoming culture medium. Control cultures were treated in parallel withclear that epigenetic processes constitute a significant factor in vehicle (PBS). Twenty-four hours after drug addition, culturethe formation of cancer (9). In this regard, DNA methylation medium was replaced with drug-free medium. Control andabnormalities have been implicated in colon cancers in both 5-Aza-CdR-treated cells were subcultured at equal densities atmouse and human tumor model systems (6, 10-12). 1 and 5 days after the initial treatment, and proliferation was

5-Aza-2'-deoxycytidine (5-Aza-CdR) inhibits DNA methyl- measured at the subsequent time point by using a Coulteration and often is used in vitro to induce the reexpression of genes counter.putatively silenced by promoter methylation (8). 5-Aza-CdR is In other experiments, HT29 cells (control or pretreated withsubstituted for cytosine during replication and is recognized by 500 nM 5-Aza-CdR) were exposed to human recombinantDNA MeTase (13). Attempted transfer of methyl groups to IFN-a2a (a gift from Roche) at 1 X 105 units/ml or human5-Aza-CdR, however, covalently traps the enzyme to newly recombinant IFN-y (GIBCO) at 5 X 102 units/mi. RNA wassynthesized DNA (14, 15). This sequestration ultimately depletes harvested for microarray expression analysis at 10, 24, and 96 hr.cellular stores of DNA MeTase and results in widespread IFN concentrations were established by measuring growth in-genomic hypomethylation. Clinical trials have demonstrated hibition in HT29 cells after treatment and approximating thepromise in the use of 5-Aza-CdR (decitabine) for treating IC515 for each IFN type (data not shown).leukemia, and current trials are evaluating 5-Aza-CdR in thetreatment of lung and prostate cancers (16-19). It is plausiblethat the antitumor activity of 5-Aza-CdR results from the Abbreviations: S-Aza-CdR, 5-aza-2'-deoxycytidine; DNA MeTase, DNA cytosine methyl-induction of methylation-regulated tumor-suppressive pathways. transferase; STAT, signal transducer and activator of transcription; EST, expressed sequence

tag.h iTowhom reprint requests should be addressed at: University of Utah, 2000 Circleof Hope,new insights into tumor development and may reveal the poten- Salt Lake City, UT 84112. E-mail: david.jones@hci.utah.edu.tial for inhibiting DNA methylation as a cancer treatment (20). The publication costs of this article were defrayed in part by page charge payment. ThisIn this regard, a number of strategies have been used to uncover article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C.methylation-regulated genes, including candidate gene analysis, §1734 solely to indicate this fact.

PNAS I November 23, 1999 I vol. 96 I no. 24 1 14007-14012

Construction of Microarrays. The cDNA clones on the microarray 100were obtained from Research Genetics (Huntsville, AL) and AGenome Systems (St. Louis). Transformants were grown over- 75night at 37°C in 96-well microtiter dishes containing 0.2 ml/well 2Terrific Broth supplemented with ampicillin. Cultures were 0 00transferred to a Millipore multiscreen, 96-well glass-fiber filtra- 2-tion plate (MAFB NOB), and growth medium was voided.Twenty-five microliters of 25 mM Tris.HC1, pH 8/10 mM 0EDTA/50 ti1 of 0.2 M NaOH/1% SDS/160 Al of 0.7 M 1.0 5.0 9.0potassium acetate, pH 4.8/5.3 M guanidine hydrochloride was days post-treaflretadded to each well of the glass filtration plate. Cell lysates weredrawn through the glass filters under vacuum, and filter-bound 10-9 10-1 10-7 5x10-7 M 5-Aza-CdRDNA was washed four times with 200 l.l of 80% ethanol. PlasmidDNAs were eluted by centrifugation after the addition of 65 Al Bof distilled H20. Samples were collected in a 96-well microtiter DNA MeTasedish during centrifugation.

PCR amplifications (30 cycles, 52°C annealing) were per-formed in 100-tl reaction volumes in a 96-well format by using2 Al of purified plasmid as template and vector-specific primers vehicle 3 5 9 day(typically T7 and T3). PCR products were combined with 200 LI Cof binding solution (150 mM potassium acetate, pH 4.8/5.3 Mguanidine hydrochloride) in a Millipore multiscreen glass-fiber p16

filtration plate. Vacuum was applied to void the binding solution,and bound PCR product was washed four times with 200 Al of80% ethanol. Products were eluted in 65 jil of distilled H 20. Fig. 1. 5-Aza-CdR inhibits HT29 cell proliferation and sequesters DNAPCR product size ranged from 300 bp to 2.0 kb, with 1.0 kb as MeTasel.(A)HT29cellsweretreatedwithvehicleor500nM5-Aza-CdRfor24a typical length. DNA was prepared for spotting by diluting the hr. After this treatment, the drug was removed and cell proliferation waspurified PCR products in DMSO at a final concentration of measured by directly counting cells at the indicated time points (see Materials20-45 ng/pAI. and Methods). Data are presented as mean count ± 1 SD, (n = 3). (B) HT29 cells

Microarray slides were produced by using a Generation III were treated with the indicated concentrations of 5-Aza-CdR for 24 hr, and

Microarray Spotter (Molecular Dynamics). Each microarray the presence of DNA MeTase I (200 kDa) was assessed in nuclear proteincontained 4,608 minimally redundant cDNAs spotted in duplic extracts by immunoblotting. Sequestration of DNA MeTase I by 500 nM

5-Aza-CdR continued for 4 days after treatment (data not shown) (C) Thecate on 3-aminoproply-trimethoxy silane-coated (Sigma) slides expression of p16 in HT29 cells at time points after treatment with 500 nMand UV crosslinked in a Stratalinker (Stratagene). 5-Aza-CdR was measured by Northern blot analysis.

Generation of Microarray Probes, Microarray Hybridizations, andScanning. Total RNA was isolated by using Trizol reagent macia). Reverse transcription-PCR of type I (cx, P) and II (y)(GIBCO) and poly(A) RNA was selected by using an Oligotex IFN genes was carried out on cDNAs prepared from vehicle-Kit (Qiagen). First-strand cDNA probes were generated by treated and 500 nM 5-Aza-CdR-treated HT29 cells 9 days afterincorporation of Cy3-dCTP or Cy5-dCTP (Amersham Pharma- treatment. The primers used for amplification of IFN-a arecia) during reverse transcription of purified mRNA (1 Acg) with within the coding region and are capable of amplifying eachSuperScript II (GIBCO). After synthesis, RNA/cDNA hybrids member of the IFN-a gene cluster.were denatured and the mRNA was hydrolyzed with NaOH.Single-stranded cDNA probes were transferred to a Millipore Cell Fractionations and Western Blotting. Nuclear and cytoplasmicglass-fiber filtration plate containing two volumes of 150 mM fractions were prepared as described previously (30). Proteinpotassium acetate, pH 4.8, and 5.3 M guanidine hydrochloride, extracts (50 Ag) were fractionated through 10% SDS/PAGE gelsThe mixture was voided by vacuum, and bound cDNA was (Novex) and blotted onto poly(vinylidene difluoride) mem-washed four times with 80% ethanol. Probes were eluted by the branes (Amersham Pharmacia). Antibody to DNA methyltrans-addition of 50 Al of distilled H20, recovered by vacuum con- ferase I was a kind gift from Moshe Szyf (McGill University,centration, and reconstituted in 30 Al of 5 X SSC/0.1% SDS/0.1 Montreal, Canada). STATs 1, 2, and 3 antibodies were purchasedJ.g/ml salmon sperm DNA/50% formamide. After denaturation from Transduction Laboratories (Lexington, KY). Final proteinat 94°C, the hybridization mixture was deposited onto an arrayed detection employed a horseradish peroxidase-conjugated goatslide under a coverslip.

Hybridizations were performed overnight at 42°C in a humid- anti-mouse secondary antibody (GIBCO) and chemilumines-ified chamber. After hybridization, slides were washed for 10 mi cence (NEN Renaissance).in 1X SSC/0.2% SDS and then for 20 min in 0.lx SSC/0.2% ResultsSDS. Slides were rinsed briefly in distilled water and dried withcompressed air, and the fluorescent hybridization signatures 5-Aza-CdR Treatment Inhibits the Proliferation of HT29 Cells. HT29were captured by using the "Avalanche" dual-laser confocal colon adenocarcinoma cells are p53- and APC-deficient andscanner (Molecular Dynamics). Fluorescent intensities were mismatch repair-proficient. Treatment of these cells with 500 nMquantified by using ARRAYVISION 4.0 (Imaging Research, St. 5-Aza-CdR for 24 hr caused a time-dependent, 3-fold inhibitionCatherine's, ON, Canada). of proliferation (Fig. 1A). As determined by flow cytometric

analysis of propidium iodide-stained cells, apoptosis failed toNorthern Blotting and Reverse Transcription-PCR. Five micrograms account for the reduced cell numbers in response to 5-Aza-CdR.of total RNA was fractionated through formaldehyde- Rather, growth inhibition was characterized by an increasedcontaining agarose gels and transferred onto nylon membranes proportion of cells in G, (data not shown). Treatment with(Amersham Pharmacia). Hybridizations with 32P-labeled probes 5-Aza-CdR depleted HT29 cells of soluble, nuclear DNAwere carried out by using Rapid-hyb buffer (Amersham Phar- MeTase I as determined by Western analyses of nuclear protein

14008 I www.pnas.org Karpf et al.

extracts (Fig. 1B). This depletion corresponded with the reex-pression of a known methylation-silenced gene, p16 (8) (Fig. IC). AThe kinetics of growth inhibition, depletion of DNA MeTase I,and the reactivation of p16 were consistent with the mechanisticproperties of 5-Aza-CdR and verified our HT29 cell modelsystem. Although the induction of p16 may contribute to thegrowth inhibition seen in response to 5-Aza-CdR (31), wehypothesized that the genomewide nature of 5-Aza-CdR-induced hypomethylation was likely to affect other growthinhibitory pathways.

5-Aza-CdR-Treatment Induces the Expression of IFN-Responsive Genesin HT29 Cells. To investigate the molecular mechanisms involvedin 5-Aza-CdR-induced growth inhibition in HT29 cells, we 10constructed and utilized high-density cDNA microarrays toanalyze gene expression changes coincident with 5-Aza-CdR Btreatment. Our array was composed of 4,608 randomly selected,minimally redundant cDNAs from the Unigene set (32). LabeledcDNA probes were prepared from vehicle-treated and 500 nM "5-Aza-CdR-treated HT29 cells 9 days after the initial drugexposure, a time that coincided with maximal growth inhibition W(Fig. 1A). First-strand cDNAs were reverse-transcribed frommRNA samples in the presence of Cy-3dCTP (vehicle-treated) 0. .or Cy-5dCTP (5-Aza-CdR-treated). After labeling, the two 0 0 2000 3000 4000probes were hybridized simultaneously to the microarray slide Array Element Number(Fig. 2A). Subsequent analysis revealed up-regulation of 19genes by greater than 2 SD above the mean expression ratio forthe entire gene set (Fig. 2B). We confirmed the induction of vehicle 5 9 vehicle 5 9 day

these genes with Northern analyses (Fig. 2C) and their identityby DNA sequencing (Table 1). We noted that 10 of 19 genes C 11.5 kD Midkineinduced by 5-Aza-CdR were established IFN-response genes(Table 1) (27, 34-36). Because IFNs are established cell growth 17.5 kD ........ * Mx-linhibitors (37-39), the stimulation of IFN-responsive genes in 17.5 .D,÷5-Aza-CdR-treated HT29 cells presented an attractive hypoth- 56 kD gapdhesis to explain the coincident growth inhibition (Fig. 1A).

To determine whether these 10 genes are regulated by IFN in Fig. 2. Microarray analysis of gene expression changes in HT29 cells afterHT29 cells, and to assess whether the other 9 genes also are 5-Aza-CdR treatment. (A) A cDNA microarray containing 4,608 target genesresponsive to IFN, we conducted microarray experiments on was constructed from a set of minimally redundant expressed sequence tagsHT29 cells treated for 10, 24, or 96 hr with either IFN-a or IFN-Y (ESTs). The microarray was hybridized with cDNAs prepared from vehicle-(Fig. 3). Interestingly, each of the 19 genes regulated by 5-Aza- treated [Cy3-dCTP-labeled (green)] and 500 nM 5-Aza-CdR-treated HT29 cellsCdR were also induced by either IFN-c or IFN- , (Table 1). [CyS-dCTP(red)]9daysaftertreatment.Tworepresentative12X32genegrids

(of 12) aredisplayed. (B) The fluorescent signal from the hybridized microarrayComparison of the induced genes revealed a significantly greater slide was detected, quantified, and plotted as a ratio (Cy-5 signal/Cy-3 signal)overlap between 5-Aza-CdR- and IFN-cr-induced genes than for each array element. The average expression ratiofor all genes on the arraybetween 5-Aza-CdR- and IFN-y-induced genes (17/19 vs. 12/19 was normalized to 1.0 and had a SD of 0.177. The black line indicates a trendgenes, respectively), line 2 SD above the mean expression ratio for all genes on the microarray. The

small, blue diamonds are genes below this cutoff; the large, red diamonds are5-Aza-CdR Treatment Induces the Nuclear Accumulation and the genes above the cutoff. (C) Microarray expression data were confirmed by

Expression of STATs 1, 2, and 3. A simple explanation for the Northern blot analysis. Induction of five representative transcripts (see Table1) 5 and 9 days after 5-Aza-CdR treatment is shown, along with glyceralde-

activation of IFN-responsive genes by 5-Aza-CdR is that the drug hyde-3-phosphate dehydrogenase (gapdh), an RNA-loading control. 11.5 kD,stimulated the synthesis and release of IFNs. Consistent with this IFN-inducible protein 27; 17.5 kD, IFN-induced 17-kDa protein; 56 kD, IFN-possibility is the observation that the expression of IFN--y can be induced protein 56.regulated by DNA methylation (40-42). We, therefore, mea-sured the mRNA levels for IFNs-a, -/3, and --y in HT29 aftertreatment with 500 nM 5-Aza-CdR. Reverse transcription-PCR caused us to investigate the IFN-signaling pathway to account foranalysis detected only IFN-a, and its mRNA level remained the induction of IFN-responsive genes by 5-Aza-CdR. Zunchanged after 5-Aza-CdR treatment (Fig. 4A). We also were Because STAT transcription factors are effectors of IFNunable to detect any increase in IFN-a (or the presence of signaling (43), we next examined whether 5-Aza-CdR treatmentIFN-y) by Western blot analysis of protein extracts from 5-Aza- caused them to accumulate in the nuclei of HT29 cells. ToCdR-treated HT29 cells at 2, 5, or 7 days after treatment (data address this, we performed Western blot analyses on fraction-CRrte shown).Thelse obseati2,5,orn7 daysifter treatmen(dat leveated HT29 cells by using antibodies specific for STATs 1, 2, 3, 4,not shown). These observations eliminated increased levels of 5, and 6. We observed a time-dependent increase of STATs 1, 2,IFNs as an explanation for the induction of IFN-responsive genes and 3 in the nuclei of HT29 cells after treatment with 500 nMby 5-Aza-CdR. In addition, the transfer of medium harvested 5-Aza-CdR (Fig. 4B). In contrast, STATs 4, 5, and 6 did notfrom 5-Aza-CdR-treated cells at 9 days after treatment onto accumulate in the nuclei after 5-Aza-CdR treatment (data notcontrol HT29 cell cultures did not inhibit cell growth. This shown). We also noted an increase in the total cellular levels ofindicates that the growth inhibition observed in 5-Aza-CdR- STATs 1, 2, and 3 after 5-Aza-CdR treatment (Fig. 4B). Thistreated cells does not result from an increase in secreted IFN novel observation raised the possibility that inhibition of DNAprotein or of other growth inhibitory cytokines. These data MeTase induced the expression of STATs 1, 2, and 3.

Karpf et al. PNAS I November 23, 1999 I vol. 96 I no. 24 1 14009

Table 1. Genes up-regulated by 5-Aza-CdR treatment of HT29 A vehicle 9 daycells*

Unigene IFN-a "" • , -2 .

number or Regulation Regulation gapdh W-- M5-Aza-CdR-induced gene t image ID by IFN-al by IFN--y

Human mRNA for Stac Hs. 56045 + + vehicle 5 9 day

IFN-induced protein 56 Hs. 20315 + BIFNa-inducible protein 27 Hs. 2867 + +IFN-induced 17-kDa protein liD 149319 + + STAT I 80kdEST Hs. 6166 + STAT 2EST Hs. 165240 + +Myxovirus resistance gene 2 Hs. 926 + - STAT 3 w0kdPurinergic receptor P2Y5 Hs. 189999 + + 'N:. .,EST Hs. 109309 + vehicle 5 9 day

CpG island DNA fragment No match§ + +TGF-p3 superfamily member MIC-1 Hs. 116577 + + C STAT Ic • 4.4 kbIFN-induced protein IFI-6-16 Hs. 21205 + - STAT Il ,.,,EST Hs. 47783 + +

MHC class I Hs. 77961 + + .i

Midkine Hs. 82045 - + STAT 2 40-o• 4.4 kb

Myxovirus resistance gene 1 Hs. 76391 + - _

2'-5'-Oligoadenylate synthetase 3 No matchl + + STAT 3 2.37 kbNuclear antigen SP100 Hs. 77617 +IFN-inducible protein 10 Hs. 2248 - + 28S

Bold type indicates previously identified lFN-responsive genes.

*Genes induced in HT29 cells by treatment with 500 nM 5-Aza-CdR at day 9 186thatwere up-regulated bygreaterthan 2 SD abovethe mean expression ratiofor all genes on the microarray (see Fig. 28), listed in descending order. Fig. 4. 5-Aza-CdR treatment activates STATs 1, 2, and 3 in HT29 cells. (A) The

TThe identity of each gene was verified by DNA sequencing and BLAST expression level of IFN-a in HT29 cells before and after 500 nM S-Aza-CdRanalysis (33). treatment was measured by reverse transcription-PCR along with gapdh to

$The expression of the genes induced by 5-Aza-CdR (column 1) was measured confirm equivalent cDNA input. (B) STAT transcription factor levels wereby microarray analysis after treatment of HT29 cells with 1 x 105 units/mi measured by Western blotting. Cytoplasmic (C) and nuclear (N) cell extractsIFN-a2a or 5 x 102 units/ml IFN-yfor 10, 24, and 96 hr. Expression was scored were prepared from HT29 cells after treatment with vehicle or 500 nMas induced if the gene was up-regulated at any time point. 5-Aza-CdR. A poly(vinylidene difluoride) membrane harboring the protein

§Identical to accession numbers Z61029 and Z61030 (100% identity over 115 extracts was probed sequentially with mAbs specific to STATs 1, 2, and 3. Inbases) in the nonredundant Genbank database, each case, the antibodies recognized proteins of the appropriate molecular

lildentical to accession number NM006187 (100% identity over 366 bases) in weight for each STAT. Molecular mass markers are indicated. (C) The expres-the nonredundant Genbank database. sion of STAT 1, 2, and 3 genes was measured by Northern blotting. RNA was

isolated from HT29 cells after treatment with vehicle or 500 nM S-Aza-CdR.The locations of molecular mass markers are indicated. Ethidium bromide

Because cDNAs corresponding to STATs 1, 2, and 3 were staining confirmed equal RNA loading (28S, 18S rRNAs).not on the microarray, we next performed Northern blotanalyses on RNAs from 5-Aza-CdR-treated HT29 cells byusing probes specific for STATs 1, 2, and 3. Fig. 4C illustrates 3 in the response of HT29 cells to 5-Aza-CdR. We also found

the time-dependent up-regulation of STATs 1, 2, and 3 mRNA that the STAT genes were expressed above control levels for

levels after 5-Aza-CdR treatment. This induction correlated at least 17 days (5 cell passages) after treatment with 5-Aza-

temporally with growth inhibition in response to 5-Aza-CdR CdR (data not shown).

and implicates the transcriptional activation ofSTATs 1,2, and 5-Aza-CdR Treatment Sensitizes HT29 Cells to Exogenous IFN-a2a. The

above data suggest that STATs 1, 2, and 3 limit the response ofHT29 cells to IFNs. With this in mind, we hypothesized that

5-Aza-CdR IFN-cc2a IFN-y 5-Aza-CdR treatment could potentiate the response of HT29

cells to IFN-a. To test this hypothesis, we exposed control and5-Aza-CdR-treated HT29 cells to various concentrations ofIFN-c2a and measured growth rates. We found that 5-Aza-CdRincreased the responsiveness of HT29 cells to growth inhibitionmediated by IFN-a2a (Fig. 5). This effect corresponded to atleast a 5-fold increase in the potency (IC.() of 2 X 105 units

Fig. 3. Microarray expression profiling of 5-Aza-CdR and IFN-treated HT29 IFN/ml for control cells vs. IC5j( of 4 X i04 units IFN/ml forcells. HT29 cells were treated with 500 nM 5-Aza-CdR, 1 x 105 units/ml 5-Aza-CdR-treated cells) of IFN-a for inhibiting HT29 cellIFN-a2a, or 5 x 102 units/ml IFN-3,. RNA was harvested 9 days (5-Aza-CdR) growth. It is important to note that the increased responsivenessor 4 days (96 hr) (IFN-a or --y) after treatment and used to generate probes was observed despite the high level of growth inhibition elicitedfor microarray analysis. Shown in the figure is a representative section of by 5-Aza-CdR treatment alone (Fig. IA).

the microarray after hybridization with Cy-S-labeled cDNAs from 5-Aza-CdR-, IFN-a-, or IFN--y-treated cells and Cy-3-labeled cDNAs from control Discussioncells. Four genes up-regulated by 5-Aza-CdR treatment are on the displayedgrid. They are IFN-a-inducible protein 6-16 (row 4, column 9), expressed Transcriptional silencing of tumor-suppressor genes by CpGsequence tag (EST) Hs.109309 (8, 7), EST Hs.165240 (9, 14), and human methylation may contribute to the development of humanmRNA for Stac (9, 16). carcinomas. A model wherein methylation-induced gene silenc-

14010 I www.pnas.org Karpf et al.

100- HI29 vehicle pointed us to the up-regulation of STATs 1, 2, and 3, which are

ýHT29 5-Aza-CdR required to mediate the growth-inhibitory effects of IFN-y(STAT 1) and IFN-a (STATs 1, 2, and 3) (43, 46, 47).

80- The up-regulation of STATs in response to 5-Aza-CdR maybe0 •explained in at least two ways. One explanation is that STAT

2 60- genes are directly silenced by de novo methylation in tumor cells.0 In support of this model, the 5' regions of STAT 1, 2, and 3

cDNAs contain likely CpG island targets for methylation (48)40 (see also GenBank accession no. L29277 for STAT 3). A second

explanation is that induction of STATs 1, 2, and 3 by 5-Aza-CdR20 is the result of the epigenetic activation of another, upstream

regulator of STAT expression. We do not believe it is likely that1 2 3 4 5 the stimulation of STATs and the IFN-induced gene set is due

IFN-c2a (LOgl0 units/ml) to nonspecific cellular toxicity or growth arrest because microar-ray experiments performed in our laboratory with agents such as

TNF, TRAIL, FasL, and TGF-P3 have not revealed the inductionFig. 5. 5-Aza-CdR treatment increases the responsiveness of HT29 cells to of a similar gene set as that seen with 5-Aza-CdR and IFN-a and,growth inhibition mediated by exogenous IFN-a2A. HT29 cells were treatedwith 500 nM 5-Aza-CdR or vehicle (PBS). Ten days after removal of the drug, accunsfr express on oS ateisrunliel

triplicate wells were treated with a concentration curve of IFN-a2a. Four days a t the sm e up-reatio ofpteseignes also rts inltheirlater, cell proliferation was measured by using a Coulter counter. Percentage that the simple up-regulation of these genes also results in theirof control growth was calculated by dividing the mean cell count at each IFN activation and nuclear accumulation. Rather, our data support aconcentration by the mean cell count of untreated control cells (either HT29 scenario in which endogenous IFN-a is responsible for activatingor 5-Aza-CdR-treated HT29 cells, respectively). Data are presented as mean ± STATs 1, 2, and 3. Several lines of evidence support this1 SD, (n = 3). Similar results were obtained in four independent experiments, explanation. First, our analysis indicates the presence of IFN-a

in control and 5-Aza-CdR-treated HT29 cells whereas IFN-P3and -,y were undetectable under either condition. Second, our

ing accompanies tumor development raises the potential for microarray analysis showed substantial overlap in genes induceddrug-induced reactivation of methylation-silenced tumor- by 5-Aza-CdR and those induced by the direct addition of IFN-a.suppressor genes as a therapeutic strategy. In this context, Finally, our observation that STATs 1, 2, and 3 each accumulatedpharmacological inhibition of DNA MeTase by 5-Aza-CdR in HT29 cell nuclei follows a number of studies demonstratinginhibits the growth of bladder, colon, and melanoma tumor cell that IFN-a stimulation leads to activation of STAT1/2 orlines, whereas control human fibroblasts are unaffected (31). STAT1/3 heterodimers (43, 49, 50).Also, consistent with this model, a number of methylation-silenced Our observation that 5-Aza-CdR stimulates expression oftumor-suppressor genes have been identified by candidate gene STATs 1, 2, and 3 holds important clinical implications. First, theapproaches in tumor cells (3-5, 7, 11, 12, 44, 45). Among these, expression of STAT 1 in certain metastatic melanoma andBender et al. have demonstrated induction of p1 6 in a number gastric adenocarcinoma cell lines is greatly depressed and cor-of tumor cells that are growth-inhibited after 5-Aza-CdR treat- relates with a reduced level of responsiveness of these tumors toment and that this induction correlates with the methylation IFN-a (51-53). In the clinic, metastatic melanomas often fail tostatus of the p16 promoter (31). However, it is reasonable to respond to IFN-ca (54). Dampening of the IFN-response pathwayassume that the pharmacology of 5-Aza-CdR extends beyond by methylation silencing of STATs or other signaling compo-p16-mediated growth arrest in that tumor cells in which p16 is nents could account for lack of responsiveness of certain mel-not induced by 5-Aza-CdR are also growth-inhibited (31). anomas to IFN-a. Further, the activation of STATs by 5-Aza-

Our observation that 5-Aza-CdR inhibits HT29 cell growth CdR treatment raises the possibility that this drug could sensitizeparallels the results seen in other tumor cell lines (Fig. 1) (31) resistant tumor cells to IFN. As an initial test of this hypothesis,and validated them as a model system for microarray expression we examined the sensitivity of HT29 cells to the growth-analysis. However, the results of our microarray analysis led to inhibitory effects of IFN-a before and after treatment witha new hypothesis for explaining the growth-inhibitory properties 5-Aza-CdR. We saw that 5-Aza-CdR treatment increased theof 5-Aza-CdR in tumor cells in vitro and, perhaps, the efficacy of responsiveness of HT29 cells to IFN-a-mediated growth inhibi-this compound in vivo. Our data indicate that STAT 1, 2, and 3 tion This result offers a plausible new line of investigation on theexpression is induced by 5-Aza-CdR, that these proteins accu- combination of 5-Aza-CdR and IFNs for the treatment of certainmulate in the nucleus of 5-Aza-CdR-treated cells, and that these IFN-resistant tumors.phenomenon parallel 5-Aza-CdR-induced growth inhibition. In conclusion, our work shows the value of microarrayThese data suggest that the presence of STAT proteins in tumor expression analyses in analyzing the mechanistic actions ofcells can dictate responsiveness to certain chemotherapeutics pharmaceutical agents. Two previous studies have utilizedand raise the possibility that STATs 1, 2, and 3 are methylation- microarrays to examine the specificity of drug actions in yeast.silenced tumor suppressors. Gray et al. examined the transcriptional perturbations elicited

Our microarray approach started with an unbiased look at by structural analogs of cyclin-dependent kinases inhibitorsHT29 cell responses to 5-Aza-CdR and led us, indirectly, to the (28). In another study, Marton et al. compared the transcrip-IFN-signaling pathway as a potential tumor-suppressive path- tional profiles resulting from cyclosporin A and FK506 treat-way. e saw that the genes responding most robustly to 5-Aza-CdR ment of yeast mutant strains defective in calcineurin andtreatment in HT29 cells were also responsive to IFN treatment. This immunophilin genes (29). These studies illustrate that microar-suggested the activation of the IFN-signaling pathway as a major rays can be used to examine drug-target specificity and po-cellular response to 5-Aza-CdR. The induction of IFN-responsive tential secondary drug effects. We have extended these ap-genes presents an attractive hypothesis for explaining 5-Aza- proaches by presenting a microarray-based evaluation of aCdR-mediated growth inhibition in that IFNs are established clinically relevant compound in a human cell line. Althoughgrowth-inhibitory cytokines (37, 39). However, it was unlikely the explicit mechanistic basis for inhibition of DNA MeTase bythat each of these IFN-responsive genes were regulated directly 5-Aza-CdR is known, our study provides new, testable hypoth-by promoter methylation. As an alternative, the microarray data eses that may explain the consequences of inhibiting DNA

Karpf etal. PNAS I November 23, 1999 1 vol.96 1 no. 24 I 14011

methylation in a clinical setting. Further pharmacological We thank Ann Jones for cell culture expertise. We thank the DNAstudies that utilize microarrays are likely to reveal new lines of Sequencing Facility, DNA/Peptide Synthesis Facility, and Flow Cytom-investigation, both in vitro and in vivo, that more focused etry Facilities at the University of Utah. This work was supported hyexperimental approaches may overlook, grants from the Huntsman Cancer Foundation.

1. Nan, X., Ng, H. H., Johnson, C. A., Laherty, C. D., Turner, B. M., Eisenman, C. J1., Stoughton, R., Burchard, J., Slade, D., Dai, H., et al. (1998) Nat. Med. 4,R. N. & Bird, A. (1998) Nature (London) 393, 386-389. 1293-1301.

2. Holliday, R. (1979) Br. 1. Cancer 40, 513-522. 30. Wang, P., Wu, P., Siegel, M. L., Egan, R. W. & Billah, M. M. (1995)1. Biol.3. Issa, J. P., Ottaviano, Y. L., Celano, P., Hamilton, S. R., Davidson, N. E. & Chem. 270, 9558-9563,

Baylin, S. B. (1994) Nat. Genet. 7, 536-540. 31. Bender, C. M., Pao, M. M. & Jones, P. A. (1998) Cancer Res. 58, 95-101.4. Ottaviano, Y. L., Issa, J. P., Parl, F. F., Smith, H. S., Baylin, S. B. & Davidson, 32. Miller, G. S. & Fuchs, R. (1997) Comput. AppI. Biosci. 13, 81-87.

N. E. (1994) C~ancer Res. 54, 2552-2555. 33. Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z., Miller, W.5. Yoshiura, K., Kanai, Y., Ochiai, A., Shimoyama, Y., Sugimura, T. & Hirohashi. & Lipman, D. J. (1997) Nucleic Acids Res. 25, 3389-3402.

S. (1995) Prnc. Nail. Acad. Sci. USA 92, 7416-7419. 34. Stadler, M., Chelbi.Alix, M. K., Koken, M. H., Venlurini, L., Lee, C.. Saib, A.,6. Herman, J. G., Umar, A., Polyak, K., Graff, J. R., Ahuja. N., Issa. J. P.. Quignon, F., Pelicano, L., Guillemin, M. C., Schindler, C., et al. (1995)

Markowitz, S., Willson, J. K., Hamilton, S. R., Kinzler, K. W., et at. (1998) Proc. Oncogene 11, 2565-2573.Nail. Acad. Sci. USA 95, 6870-6875. 35. Rasmussen, U. B., Wolf, C.. Mattei, M. G., Chenard, M. P., Bellocq, J. P.,

7. Gonzalez-Zulueta, M., Bender, C. M., Yang, A. S., Nguyen, T., Beart. R. W., Chambon, P.. Rio. M. C. & Basset, P. (1993) Cancer Res. 53, 4096-4101.Van Tornout, J. M. & Jones, P. A. (1995) Cancer Res. 55, 4531-4535. 36. Luster, A. D., ihanwar, S. C., Chaganli, R. S., Keraey, J. H. & Ravetch, J. V.

8. Bender, C. M., Zingg, J. M. & Jones, P. A. (1998) Pliarm. Ret. 15, 175-187. (1987) Proc. Nail. Acad. Sci. USA 84, 2868-2871.9. Bird, A. P. (1996) Cancer Surv. 28, 87-101. 37. Pfeffer, L. M., Dinarello, C. A., Herberman, R. B., Williams, B. R., Borden,

10. Laird, P. W., Jackson-Grusby, L., Fazeli, A., Dickinson. S. L., Jung, W. E., Li, E. C., Bordens, R., Walter, M. R., Nagabhusltan, T. L., Trotla, P. P. & Pestka,E., Weinberg, R. A. & Jaenisch, R. (1995) Cell 81, 197-205. S. (1998) Cancer Res. 58, 2489-2499.

11. Veigl, M. L., Kasturi, L., Dlechnowicz, J., Ma, A. H., Lutterhaugh. J. D., 38. Stark, G. R., Kerr. 1. M., Williams. B. R., Silverman. R. H. & Schreiber, R. D.Periyasamy, S., Li, G. M., Drummond, J., Modrich, P. L_, Sedwick. Wv. D., et (1998) Annu. Rev. Biocliem. 67, 227-264.al. (1998) Proc. Nail. Acad. Sci. USA 95, 8698-8702. 39. Nathan, C. (1992) in Inflammation: Basic P'rinciples and Clinical Correlates, eds.

12. Cote, S., Sinnett, D. & Momparler, R. L. (1998) Anticancer Drugs 9, 743-750. Gallin, J. 1. & Goldstein, L. M. (Raven, New York), 2nd Ed., pp. 265-289.13. Momparler, R. L. (1985) Pharmacol. Thter. 30, 287-299. 40. Fukunaga, R., Matsuyama, M., Okamura, H., Nagata, K., Nagata, S. & Sokawa,14. Juttermaun, R., Li, E. & Jaenisch, R. (1994) Proc. Nail. Acad. Sci. USA 91, Y. (1986) Nucleic Acids Re~s. 14, 4421-4436.

11797-11801. 41. Young, H. A., Ghosh, P., Ye, J., Lederer, J., Lichtman, A., Gerard, J. R., Penix,15. Ferguson, A. T., Vertino, P. M., Spitzner, J. R., Baylin. S. B., Muller, M. T. & L., Wilson, C. B., Melvin, A. J., McGurn, M. E., ei al. (1994)1J. Immunol. 153,

Davidson, N. E. (1997)1J. Bilo. Chiem. 272, 32260-32266. 3603-3610.16. Kantatjian, H. M., O'Brien, S. M., Eatey, E., Giralt, S., Beran, M., Rios, M. B., 42. Mikovits, J. A., Young, H. A., Vertino, P., Issa, J. P., Pitha, P. M., Turcoski-

Keating, M., de Vos, D. & Talpaz, M. (1997) Leukemia It, Suppl 1, S35-S36. Corrales, S., Taub, D. D., Pesrow, C. L., Baylin, S. B. & Ruscetti, F. W. (1998)17. Momparler, R. L., Rivard, G. E. & Gyger, M. (1985) Pharrnacol. Ther. 30, Msol. Cell. Biol. 18, 5166-5177.

277-286. 43. Darnell, J. E.; Jr. (1997) Scientce 277, 1630-1635.18. Momparler, R. L., Bouffard, D. Y., Momparler, L. F., Dionne, J., Belanger, K. 44. Belinaky, S. A., Nikula, K. J., Palmisano, W. A., Michels, R., Saccomanno, G.,

& Ayoub, J. (1997) Anticancer Drugs 8, 358-368. Gabrielson, E., Baylin, S. B. & Herman, J. G. (1998) Proc. Nall. Acad. Set. USA19. Thibault, A., Figg, W. D., Bergan, R. C., Lush, R. M., Myers, C. E., Tompkins, 95, 11891-11896.

A.. Reed, E. & Samid, D. (1998) Tumori 84, 87-89. 45. Lapidus, R. G., Ferguson, A. T., Ottaviano, Y. L., Panl, F. F., Smith, H. S.,20. Jones, P. A. & Laird, P. W. (1999) Nat. Genet. 21, 163-167. Weitzman, S. A., Baylin, S. B., lass., J. P. J. & Davidson, N. E. (1996) Clin.21. Ushijima, T., Morimura, K., Hosoys, Y., Okonogi, H., Tatcmatsu, M., Cancer Res. 2, 805-810.

Sugimura, T. & Nagao, M. (1997) Proc. Nail. Acad. Set. USA 94, 2284- 46. Bromberg, J. F., Horvaths, C. M., Wen, Z., Schreiber, R. D. & Darntell, J. E..2289. Jr. (1996) Proc. Nail. Acad. Sci. USA 93, 7673-7678.

22. Hatada, L., Sugama, T. & Mukai, T. (1993) Nucleic Acids Res. 21, 5577- 47. Ihle, J. N., Thierfelder, W., Teglund, S., Stravapodis, D., Wang, D., Feng, 3. &5582. Parganas, E. (1998) Ann. N. Y. Acad. Sct. 865, 1-9.

23. Gonzalgo, M. L., Liang, G., Spruck, C. H., III, Zingg, J. M., Rideout, Wv. M., 48. Yan, R., Qureshi, S., Zhong, Z., Wen, Z. & Darnell, J. E., Jr. (1995) NucleicIII, & Jones. P. A. (1997) Cancer Res. 57, 594-599. Acids Res. 23, 459-463.

24. Shiraishi, M., Chuu, Y. H. & Sekiya, T. (1999) Proc. Nail. Acad. Sci. USA 96, 49. Yang, C. H., Murti, A. & Pfeffer, L. M. (1998) Proc. Nail. Acad. Set. USA 95,2913-2918. 5568-5572.

25. Schena, M., Shalon, D., Davis, R. W. & Brown, P. 0. (1995) Science 270, 50. Stancato, L. F., David, M., Carter-Su, C., Lamner, A. C. & Pratt, W. B. (1996)467-470. J. Biol. Chem. 271, 4134-4137.

26. Cho, R. J., Campbell, M. J., Winzeler, E. A., Steinmetz, L., Conway, A., 51. Abril, E., Real. L. M., Serrano, A., Jimenez, P., Garcia, A., Canton, J., Trigo,Wodicka. L., Wolfsberg, T. G., Gabrielian, A. E., Landsman, D., Lockhart, D. J. I., Garrido, F. & Ruiz-Cabello, F. (1998) Canscer Immunol. Immunoiher. 47,& Davis, R. W. (1998) Mol. Cell 2, 65-73. 113-120.

27. Der, S. D., Zhou, A., Williams, B. R. & Silverman, R. H. (1998) Proc. Nail. 52. Sun, W. H., Pabon, C.. Alsayed. Y., Huang, P. P., Jandeska, S., Uddin. S.,Acad. Set. USA 95, 15623-15628. Platanias, L. C. & Rosen, S. T. (1998) Blsosd 91, 570-576.

28. Gray, N. S., Wodicka. L., Thunnissen, A. M., Norman, T. C., Kwon. S., 53. Wong, L. H., Krauer, K. G., Hatzinisiriou. I., Estcourt, M. J., Hersey, P., Tam,Espinoza, F. H., Morgan, D. 0., Barnes, G., LeClerc, S., Meijer, L., etal. (1998) N. D., Edmondson, S., Devenish, R. 3. & Ralph, S. . (1997)1J. Biol. Chaem. 272,Science 281, 533-538. 28779-28785.

29. Marton, M. J., DeRisi, J. L., Bennett, H. A., Iyer, V. R., Meyer, M. R., Roberts, 54. Punt, C. J. (1998) Melanoma Res. 8, 95-104.

14012 1 www.pnas.org Karpf et At

[CANCER RESEARCH 59, 6042-6045, December 15, 19991

" Advances in Brief

E7-transduced Human Breast Epithelial Cells Show Partial Differentiation in

Three-dimensional Culture'

Kimberly M. Spancake,2 Christine B. Anderson, Valerie M. Weaver, Norisada Matsunami, Mina J. Bissell, andRaymond L. White

Huntsman Cancer Institute, University of Utah, Salt Lake City, Utah 84112 [K. M. S., C. B. A., N. M.. R. L. W.], and Lawrence Berkeley National Laboratory, Berkeley, California

94720 [V. M. W., M. J. B.]

Abstract In addition to the cellular inactivation of pRB by cycins and cyclin-dependent kinases, several DNA tumor virus proteins abrogate pRB

Disruption of the retinoblastoma (RB) tumor suppressor pathway is a activity by binding to and functionally sequestering pRB. The HPV16 E7common and important event in breast carcinogenesis. To examine the protein not only inactivates pRB function by binding but also enhancesrole of the retinoblastoma protein (pRB) in this process, we createdhuman mammary epithelial cells (HMEC) deficient for pRB by infecting degradation of pRB through ubiquitin-dependent proteolysis (6).

primary outgrowth from breast organoids with the human papillomavirus HPV16 E7 was shown previously to immortalize early passage HMEC in

type 16 (HPV16) E7 gene. HPV16 E7 binds to and inactivates pRB and culture (7), suggesting a critical role in growth regulation for pRB in vivo.

also causes a significant down-regulation of the protein. Culturing normal Disruption of such fundamental growth control mechanisms should generate-1HMEC in a reconstituted basement membrane (rBM) provides a correct cell populations predisposed for transformation. To create a pRB-deficientenvironment and signaling cues for the formation of differentiated, acini- cell line, we infected primary breast epithelial cell outgrowth with a retroviruslike structures. When cultured in this rBM, HMEC+E7 were found to expressing the HPV16 E7 gene (HMEC+E7). In this study, only earlyrespond morphologically as normal HMEC and form acinar structures. In passage, G418-resistant, precrisis HMEC+E7 cells were examined. To char-contrast to normal HMEC, many of the cells within the HMEC+E7 acterize the impact of this pRB deficiency, we used a sensitive three-structures were not growth arrested, as determined by a 5-bromo-2'-deoxyuridine incorporation assay. pRB deficiency did not affect polariza-tion of these structures, as indicated by the normal localization of the spheroid structures that resemble breast acini (8). Culturing human, luminal

cell-cell adhesion marker E-cadherin and the basal deposition of a colla- breast epithelial cells in a rBM provides an environment that supports not

gen IV membrane. However, in HMEC+E7 acini, we were unable to only organogenesis but also formation of endogenous BM necessary for

detect by immunofluorescence microscopy the milk protein lactoferrin or differentiation (9). This culture system can also be used to distinguish be-cytokeratin 19, both markers of differentiation expressed in the normal tween normal and cancer cells because of the inability of the transformedHMEC structures. These data suggest that loss of RB in vivo would cells to organize into acinar structures (10).compromise differentiation, predisposing these cells to future tumor- We examined the effect of pRB deficiency on structure formation,promoting actions. cell-cell interactions, and differentiation by culturing the cells in a

rBM. Our data demonstrate that even in the absence of functionalIntroduction pRB, human breast epithelial cells organize and form morphologically

The tumor suppressor gene RB3 can play a significant role in breast typical acini-like structures. Furthermore, in these pRB-deficient

carcinogenesis. RB has been shown to be inactivated in 19% of human structures, proper localization of E-cadherin and deposition of a

breast tumors and 25% of human breast carcinoma cell lines (1, 2). collagen IV BM, both markers of organogenesis, were found. How-

Other members of the RB regulatory network have also been reported ever, unlike normal HMEC, the structures formed by HMEC+E7 do

as having aberrant expression in a significant number of breast car- not exit the cell cycle and are deficient for the expression of cyto-

cinomas. For example, cyclin Dl, which through regulation'of the keratin 19 and lactoferrin, two proteins associated with the differen-

cyclin-dependent kinase Cdk4 phosphorylates and inactivates pRB, is tiation of luminal breast epithelial cells. These data suggest that pRB

overexpressed in 35% of breast tumors (3). The p16 protein, also a is not necessary for the transmission of signals from the extracellular

regulator of pRB activity through inhibition of Cdk4, is absent in 40% matrix that convey information necessary for structure formation, but

of breast tumors, as determined by immunohistochemical analysis (4). pRB does appear to be necessary for cell cycle withdrawal and

Additionally, restoration of pRB expression in human breast cancer expression of some proteins normally present in the differentiated

cell lines lacking functional pRB has been shown to cause a reduction luminal epithelial cell. Our data suggest that the initial loss of RB in

in the cells' tumorigenicity (5). These data suggest that disruption of vivo would not predispose breast epithelial cells to an apoptotic or

the RB regulatory network is a common and important step in breast immediate transformed fate, but rather loss of pRB activity would

carcinogenesis. contribute to a less differentiated cellular state that, in turn, mayeventually lead to a malignant phenotype.

Received 8/26/99; accepted 11I//99.The costs of publication of this article were defrayed in part by the payment of page Materials and Methods

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact. Cell Culture and Retroviral Infection. Surgical discard material from

This work was supported in part by the United States Army Medical Research and reduction mammoplasties was minced with opposing scalpels, placed in digestion

Materiel Command Grant DAMD-17-94-J-4154 and the Huntsman Cancer Institute.I To whom requests for reprints should be addressed, at Huntsman Cancer Institute, University buffer, and incubated in spinner flasks at 37°C until stroma dissolved (-3-5 h).

of Utah, 2000 Circle of Hope, Building 555, Room 567B, Salt Lake City, Utah 84112. Phone: Digestion buffer contained I unit/mil Collagenase D (Roche Molecular Biochemi-

(801) 581-6401; Fax: (801) 585-0374; E-mail: kimberly.spancake@hci.utah.edu. cals, Indianapolis, IN), 2.4 units/ml dispase (Roche Molecular Biochemicals), and' The abbreviations used are: RB, retinoblastoma; pRB, retinoblastoma protein; 6.25 units/ml DNase (Sigma Chemical Co., St. Louis, MO) in Dulbecco's PBS.

HMEC, human mammary epithelial cells; HPVl6, human papillomavirus type 16; rBM.reconstituted basement membrane; DAPI, 4,6-diamidino-2-phenylindole; BrdUrd, 5-bro- The digested material plus 10% FCS was centrifuged at 800 rpm for 10 min. The

mo-2'-deoxyuridine. resulting pellet was resuspended in wash buffer (11), and the organoids were

6042

RB-DEFICIENT BREAST (ELLS FAIl. TO FULLY DIFFERENTIATE

HMEC HMEC+E7 prior to freezing the culture without fixative. Cryosections were fixed in 2%paraformaldehyde in PBS for 10 rin, followed by a 30-rain incubation in 2 MHCI. all performed at room temperature. Staining proceeded as above uisinganti-BrdUrd (mouse IgGI. BMC93I ) 1:10 and sheep anti mouse inmmntuIto-

kDa globulin-FITC 1:10 (Bochringer MNanheim. Indianapolis. IN). Slides werescored visually (-200 cells/experiment) for BrdUrd-labelcd nuclei, and indi-ces were calculated by expressing this number as a percentage of the total

- 220 nuclei scored (10).

Results

pRB-- - 7 HMEC expressing HPV16 E7 Show a Decrease in pRB Ex-

pression. Primary epithelial outgrowtlt from normal human or-

ganoids was infected with a retroviral construct expressing the- 66 HPVI6 E7 gene. HMEC+E7 were selected by culturing the cells

in CDM3 growth medium containing Geneticin. To determine the- 46effect of HPVI6 E7 expression on pRB, HMEC and HMEC+E7were examtned by Western imnmunohlot analysis. Fig. I shows the

Fig. I. Western blot analysis of pRB in normal and HPV tb E7-expressing breast

epithelial cells. Protein lysatcs from HMEC and HMEC+E7 were analvycd by SDS- expression level of' pRB in the normal, parental U-MEC whenPAGE and imninunoblotting with a monoclonal anti-pRB antibody, pRB is present in the cultured on plastic. In contrast. HMEC+ E7 showed a significantnormal HMEC. bnt the level of expression is redneed in HMEC+E. decrease in pRB. This down-regulationi of pRB is in agreement

with previous results demonstrating the ubiquitin-dependent pro-separated from stromal and blood cells by sequential sedimentation at I X g. teolysis of pRB in the presence of HPV 16 E7 (6).

Organoids were cultured immediately or frozen in DMEM: Nutrient Mixture F-12F HMEC+E7 Form Acinar Structures When Cultured in Extra-(Ham) 1:1 (Life Technologies, Inc., Gaithersburg, MD), 10% FCS, and 10%DMSO. Organoids were cultured on plastic in CDM3 culture medium (12) forprimary breast epithelial cell outgrowth and subculture. Primary epithelial out- HMEC and HMEC+E7 were cultured in rBM. By 10 days in clture,

"growth was infected with an LXSN retroviral construct containing the HPV16 E7gene (LXSNI6E7; Ref. 13). HMEC were incubated with LXSNI6E7 in CDM3plus 4 Acg/ml Polybrene (Sigma) for 24 h. Viral supematant was aspirated. andHMEC were cultured in virus-free CDM3 for 48 h prior to selection in CDM3containing 50 4g/ml Geneticin (Life Technologies, Inc.). Early-passage HMECI HMEC(either passage one or two) and HMEC containing LXSN 16E7 (HMEC-+ E7) werecultured in a rBM, Matrigel (Becton Dickinson, Bedford. MA), as describedpreviously (10). Briefly, 2.5 X 105 HMEC were resuspended as single cells in 300A.l of 10 mg/ml Matrigel per well and plated onto Nunc four-well multidishescoated with 100 ,. of Matrigel. Matrigel cultures were overlaid with 500 tl ofCDM3.

Western Blot Analysis. HMEC and HMEC+E7 cultured in polystyreneflasks were scraped in lysis buffer (PBS containing 0.1% Triton X-100, 0.1%NP40, 0.2 mg/mI Pefablock, 0.01 mg/ml aprotinin. 0.01 mg/mI pepstatin. 0.01 HMEC+E7mg/ml leupeptin. 20 mM sodium fluoride, and I mm sodium orthovanadate)and sonicated for 30 s on ice. HMEC and HMEC+E7 cultured in rBM for 10days were'liberated from the Matrigel by incubation with dispase (BectonDickinson) for I h at 37:C. The acini were then washed with PBS plus 5 mmEDTA, resuspended in lysis buffer, homogenized for 10 s, and sonicated for E30 s on ice. Protein lysates from cells cultured on plastic or in Matrigel were 80-boiled in SDS-PAGE sample buffer for 3 min, and 50 Ag of total protein wereresolved on 4-12% gradlent acrylamide Tris-Glycine gels (Novex, San Diego. 70-CA) by SDS-PAGE. Proteins were transferred to nitrocellulose membrane 60(Schleicher & Schuell, Dassel, Germany). Immunoblot analysis was performedas described previously 114). Antibodies were used in the following concen- 50-trations: RB (mouse IgGI. G3-245) 1:400 (PharMingen. San Diego. CA),cytokeratin 19 (mouse IgGI, RCK 108) 1:100 (DAKO), and horseradish = 40-

()peroxidase-goat antimouse (Sigma) 1:30,000. .I

Immunofluorescence. Ten-day HMEC and HMEC+E7 rBM cultures _. 30-were fixed, frozen, and cut into 5-Am sections as described previously (15). 20Antibodies were used in the following concentrations: E-cadherin (mouse v TIgG2a, 36) 1:100 (Transduction Laboratories. Lexington. KY), collagen IV • 10(mouse IgGI, CIV 22) 1:50 (DAKO, Carpinteria. CA), cytokeratin 18 (Imotuse IIgGI. CY-90) 1:800 (Sigma), lactoferrin (rabbit sera) 1:50 (Zymed, San HMEC HMEC+E7Francisco. CA), cytokeratin 19 (mouse Ig G1. RCK 108) I:100 (DAKO), goat Fig. 2. HMEC-E7 cultured it extracellular mitrix form acinar otucinres that are sotantimouse [gG I-FITC 1:200 (Southern Biotechnolo,!,v .-,, Bcits irmino- Fg..HM- E7ctreiaetaelurmtixfrainrsctr ttaeno

nc ... growth arrested. Breast cells were cturited tn a rBJ tfoT 10 daVs. A and (', phase contrast

ham, AL), Alexa 488 goat antimouse IgG (H+L), F(ab')z fragment conjugate images of representative HMEC (A) and HMEC+E7 (C) structures. 8 and 1), frozen1:1000 (Molecular Probes, Eugene. OR). and goat antirabbit-Texas Red 1:200 sections of HMEC (B) and HNIEC+E7 tD acini stained with DAPI. HMEC+E7 cultured(Accurate Chemical and Scientific, Westbury, NY). Nuclei were counter- in rBIM fornted structures torphologicallv sitilar to HIIEC, although the nuclei appearstained with 100 ng/ml DAPI (Sigma)otar TO-PRO-3 1:750 (Molecular Probes). larger and slightly less organized. Percentagee of BrdUrd labeling in HMEC and

s HMEC+E7 structures is expressed as the BrdUrd labeling index from three separateProliferation Assay. rBM cultures were prepared as above with the fol- experitients (E: har., SD). ApproximatelN 90% of HNtEC were growth arrested as

lowing exceptions. Ten j.u BrdUrd was added to the culture medium 12 h compared with HMEC+E7 that continued to synthesize DNA. Bar. 20 ,urn.

6043

RB-DEFICIENT BREAST CELLS FAIL TO FULLY DIFFERENTIATE

r I HMEC HMEC+E7 the expression of cytokeratin 19 (Fig. 4F). These immunofluorescence0 data were confirmed by Western immunoblot analysis of HMEC and

HMEC+E7 acinar structures, which showed barely detectable levelsof cytokeratin 19 in HMEC+E7 (Fig. 4G).

E-Cadherin Discussion

HMEC+E7 are deficient for pRB, a key protein governing cellcycle regulation and differentiationm thus, formation of even a partially"differentiated" structure when cultured in rBM was unexpected.Further investigation showed that although these pRB-deficient cellswere capable of organizing into morphologically normal acini, the

Collagen IV nuclei of these structures appeared to be larger and slightly less

HMEC HMEC+E7Fig. 3. Localization of E-cadherin and collagen IV in HMEC and HMEC+E7 using

immunofluorescence microscopy. , and B, frozen sections of HNIEC and HMEC+E7acini, respectively, were stained with a monoclonal anti-E-cadherin antibody, Alexa 488goat antimouse lgG (H+L), F(ab'). fragment conjugate, and the nuclei were counter-stained with TO-PRO-3. ConfocaI fluorescence microscopy showed E-cadherin localiza-tion at points of cell-cell contact (4 and B, arrows). C and D, frozen sections of HMEC Cytokeratin 18and HMEC+E7 acin, respectively. were stained with a monoclonal anti-collaen IVantibody and goat antimouse IgGI-FITC. Collagen IV was localized in a continuous BM(arrowheads) at the basal surface of HMEC (C) and HMEC +E7 (D) structures. Bar, 20 ctit.

normal HMEC organized into differentiated acini -55 pxm in diam-eter (Fig. 2A). A 5-Am frozen section of a representative structurestained with DAPI revealed a ring of basally located nuclei (Fig. 2B). LactoferrinHMEC+E7 also formed morphologically normal-appearing acinarstructures when cultured in rBM (Fig. 2C). The HMEC+E7 aciniwere similar in size to the parental structures, although DAPI stainingrevealed somewhat larger and slightly less organized nuclei (Fig. 2D).Additionally, HMEC acini had growth arrested, as determined byminor BrdUrd incorporation similar to levels reported previously (15);however, a significant increase in BrdUrd labeling was observed inHMEC+E7 (Fig. 2E). Cytokeratin 19

Normal Localization of E-cadherin and Type IV Collagen inHMEC+E7 Acinar Structures. HMEC+E7 were examined for theexpression of proteins shown previously to have distinctive localiza-tion patterns characteristic of organogenesis in normal HMEC acini(15). HMEC and HMEC+E7 rBM cultures were fixed, frozen, cutinto 5-Atm sections, and stained with the indicated primary antibodies Gand corresponding secondary antibodies. In normal HMEC acinar HMEC HMEC+E7

structures, the cell-cell adhesion protein E-cadherin was localized to kDapoints of cell-cell contact (Fig. 3A). This same pattern of E-cadherin -66expression at points of cell-cell contact was present in the HMEC+E7structures (Fig. 3B). Collagen IV was basally deposited in a contin-uous BM surrounding the normal HMEC structures when cultured in -46rBM (Fig. 3C). HMEC+E7 acini also deposited a basal collagen Cytokeratin 19--MIV-containing BM (Fig. 3D).

Lactoferrin and Cytokeratin 19 Are Not Expressed in Struc- -30turally Differentiated HMEC+E7. The proteins lactoferrin and cy-tokeratin 19 are markers for differentiation in luminal breast epithelialcells (16-18). HMEC and HMEC+E7 were prepared as described -21above and examined for the expression of lactoferrin and cytokeratin -1419. HMEC and HMEC+E7 acinar structures were identified by thepresence of the luminal marker cytokeratin 18 (Fig. 4A and B, respec- Fig. 4. Expression of cytokeratin 18, lactoferrin. and cytokeratin 19 in HMEC and

tively) and the absence of the myoepithelial marker cytokeratin 14 HMEC+E7 acinar structures using immunofluorescence microscopy and Western immu-

(data not shown). Lactoferrin is an iron-binding milk protein that was noblot analysis. A and B, frozen sections of HMEC and HMEC+E7, respectively, were

expressed in the normal HMEC acini (Fig. 4C). In contrast, immu- stained with a monoclonal cytokeratin IS antibody and goat antimouse IgGI-FITC toconfirm that structures were of luminal origin. Frozen sections of HNIEC (C and E) and

nofluorescence analysis of HMEC+E7 structures demonstrated the HNIEC+E7 (D and F) were stained for lactoferrin IC and D), cytokeratin 19 (E and Fl,

absence of lactoferrin (Fig. 4D). As expected, structures derived from and the nuclei were visualized with a DAPI counterstain (C and M). Protein lysates fromHMEC and HMEC+E7 acinar structures were analyzed by SDS-PAGE and ittutUnoblot-normal HMEC expressed the luminal differentiation marker cyto- ting with a monoclonal anti-cytokeratin 19 antibody (G). HMEC+E7 acinar structures

keratin 19 (Fig. 4E). HMEC+E7 acini, however, were deficient for were deficient for both lactoferrin and cytokeratin 19 expression. Bar, 20 ýcnt.

6044

RB DEFICIENT It3RrP`\;T CFLI 17 \1 TO Pt LLY DIFFERSINTTIATF

organized. Additionally, HMEC+E7 acini contained many cells that Acknowledgmentshad not growth arrested, unlike the normal HMEC structures. lmmu-nofluorescence analysts of the HMEC+E7 structures revealed normal We are grateful to Dr. Kristi L. Neufeld for useful comments on thenoflorecene aalyis f te H EC+E stuctresrevale nomal manuscript and Diana Lira for assistance with figure preparation.patterns of E-cadherin and collagen IV localization, indicating that theHMEC +E7 were forming normal cell-cell contacts and were properly References

polarized. HMEC+E7 acini expressing the luminal marker cytokera- 1. Varley. J. M., Armour, J., Swallow, 1. E.. Jeffrevs, A. I., Ponder. B. A., T'Ang, A.,

tin IS lacked proteins typically expressed in the differentiated luminal Fung, Y. K., Brammar, W. J., and Walker. R. A. The retinoblastonia gene is

breast epithelial cell. Neither the milk protein lactoferrin nor the frequently altered, leading to loss of expression in primary breast tunmours. Oncogene.4: 725-729. 1989.

luminal differentiation marker cytokeratin 19 were found in the 2. T'Ang, A., Varlev. 1. M.. Chakraborty. S.. Murphree, A. L.. and Fung, Y. K.HMEC+E7 acinar structures. Structural rearrangement of the retinoblastoma gene in human breast carcinoma.

pRB has a significant role in the differentiation of many different ~ Science (Washington DC). 242: 263-266. 1988.3. Gillent. C., Fantd, V.. Smuith. R.. Fisher. C., Bariek, J., Dickson, C., Barnes, D., and

cell types, including muscle, neuronal. and erythroid cells (19). Typ- Peters. G. Amplification and overexpression of cyclin DI in breast cancer detected by" immnohistochemical staining. Cancer Res., 54: 1812-1817. 1994.

ically, an early event in differentiation is the activation of pRB by 4. Geradts, J., and Wilson, P. A. High frequency of aberrant pl6(INK4A) expression indephosphorylation, allowing the cell to arrest in G,-G,. During mus- human breast cancer. Am. J. Pathol.. 149: 15-20. 1996.

cle cell differentiation, hypophosphorylated pRB associates with the 5. Wang. N. P.. To, H.. Lee. W. H., and Lee, E. Y. Tumor suppressor activity of RB andp53 genes in human breast carcinoma cells. Oncogenc, 8: 279-288, 1993.

transcription factors MyoD and myogenin. preventing phosphorla- 6. Boer. S. N., Wazer, D. E., and Band, V. E7 protein of human papillotna virus-16

tion of pRB and the consequent reemergence into the cell cycle (20). induces degradation of retinoblastonta protein through the ubiquitin-proteasome path-pRB has not been shown previously to have a role in directing the ~ vway. Cancer Res., 56: 4620-4624, 1996.

7. Wazer, D. E,, Liu, X. L.. Chu, Q., Gao, Q., and Band, V. Imnmortalization of distinctdifferentiation of breast epithelial cells. Because pRB is down-regu- human mammary epithelial cell types by human papilloma virus 16 E6 or E7. Proc.

lated in HMEC+E7, our data suggest that pRB is not necessary during Natl. Acad. Sci. USA, 92: 3687-3691. 1995.C o n o 8. Li, M. L.. Aggeler. J.. Farson. D. A., Hatier. C.. Hassell, I., and Bissell, M, J.the organization and structure formation of HN/EC when cultured in initteoareottudbsnet tntanadiscmpetsncsingeC, Influence of a reconstituted basemnent mnembrane and its components on casein gene

a three-dimensional rBM. However, the absence of pRB does impact expression and secretion in mouse mammary epithielial cells. Proc. Nail. Acad. Sci.

the expression of some proteins normally associated with the differ- USA, 84: 136-140, 1987.9. Bissell, NM. J.. Weaver, V. M., Lelievre, S. A., Wang, F.. Petersen, 0. W., and

entiated phenotype. Interestingly, a similar observation was made Schmeichel. K. L. Tissue structure, nuclear organization, and gene expression in

during MyoD-induced skeletal muscle differentiation in which cells normal and malignant breast. Cancer Res., 59: 1757s-1764s, 1999.10. Petersen. 0. W., Ronnov-Jessen, L., Howlett, A. R., and Bissell, M. J. Interaction with

showed attenuated expression of a late differentiation marker, myosin basement membrane serves to rapidly distinguish growth and differentiation pattern

heavy chain (21). The deficiency of differentiation markers in the of normal and malignant human breast epithelial cells. Proc. Natl. Acad. Sci. USA,HMEC+E7 structures could be attributable to the role of pRB in 89: 9064-9068. 1992.11s Taylor-Papadimitriou. J., and Stampfer, M, R. Culture of human mammary epithelial

transcriptional regulation outside its classically described interaction cells. In: R. I. Freshney (ed.), Culture of Epithelial Cells. pp. 108-133. New York:with E2F (reviewed in Ref. 19). Recently, pRB was shown to interact John Wiley & Sons. Inc., 1992.

12. Petersen. 0. W., and van Deurs. B. Preservation of defined phenotypic traits indirectly with the transcription factor SPl and increase transcription of short-term cultured human breast carcinoma derived epithelial cells. Cancer Res., 47:

the dihydrofolate reductase gene via the SP1 binding site (22). Se- 856-866. 1987.quence analysis of human lactoferrin and mouse cytokeratin 19 has 13. Halbern. C. L.. Demers, G. W.. and Galloway. D. A. The E7 gene of human

papillomavirus type 16 is sufficient for inmmortalization of human epithelial cells.revealed SPI sites in the promoter regions of both these genes (23, J. Virol., 65: 473-478, 1991.

24). Taken together, these data suggest a possible role for pRB in 14. Neufeld. K. L., and White. R. L. Nuclear and cytoplasmic localizations of theadenomatous polyposis coli protein. Proc. Natl. Acad. Sci. USA, 94: 3034-3039,

transcriptionally regulating lactoferrin and cytokeratin 19, a control 1997.

that is abolished by HPV16 E7. 15. Weaver, V. M., Petersen, 0. W., Wang, F., Larabell. C. A., Briand, P., Damsky, C.,HPV16 E7 protein can also interact with the pRB family members and Bissell. M. J. Reversion of the malignant phenotype of human breast cells in

three-dimensional culture and in vivo by integrin blocking antibodies. J. Cell Biol.,p107 and p13 0 . However, neither p107 nor p130 have been shown to 137: 231-245, 1997.

be targeted for ubiquitin-dependent proteolysis by HPV16 E7, as has 16. Campbell. T., Skilton, R. A., Coombes, R. C.. Shousha, S.. Graham. M. D,, andLuqmani, Y. A. Isolation of a lactoferrin cDNA clone and its expression in human

been established for pRB (25). Additionally, to date no human tumors breast cancer. Br. J. Cancer, 65: 19-26, 1992.

have been identified that contain mutations in p107 or p13 0 (reviewed 17. Close. M. J., Howlett. A. R.. Roskelley, C. D., Desprez. P. Y., Bailey, N., Rowning,

in Ref. 26). Consequently, the roles of p107 and p 130 in transforma- B., Teng, C. T.. Stampfer, M. R., and Yaswen, P. Lactoferrmn expression in mammaryepithelial cells is mediated by changes in cell shape and actin cytoskeleton. J. Cell

tion remain unclear, and inactivation of these proteins could also be Sci., 110: 2861-2871, 1997.

contributing to the phenotype observed in the HMEC+E7 structures. 18. Ronnov-Jessen, L.. Petersen, 0. W., and Bissell, M. J. Cellular changes involved inconversion of normal to malignant breast: importance of the stromal reaction. Physiol.

Our data demonstrate that breast epithelial cells with down-regulated Rev., 76: 69-125, 1996.pRB retain the ability to respond to structure-forming signaling cues 19. Herwig, S.. and Strauss, M. The retinoblastoma protein: a master regulator of cell

from a rB however, these acinar structures are not growth arrested 2cycle, differentiation and apoptosis. Eur. J. Biochem., 246: 581-601, 1997.G0, a u. W., Schneider, J. W., Condorelli. G., Kaushal. S., Mahdavi, V., and Nadal-

and do not express some of the proteins normally associated with Ginard. B. Interaction of myogenic factors and the retinoblastoma protein mediates

differentiation. These data suggest that mutation of RB alone in vivo muscle cell commitment and differentiation. Cell, 72: 309-324, 1993,21. Novitch. B. G.. Mulligan, G. J.. Jacks, T., and Lassar. A. B. Skeletal muscle cells

in human breast epithelial cells would not cause transformation but lacking the rctinoblastoma protein display defects in muscle gene expression and

rather create a lesser differentiated cellular state. This modified phe- accumulate in S and G2 phases of the cell cycle. J. Cell Biol., 135: 441-456, 1996.notype. in conjunction with additional mutations. growth factor and 22. No,. V.. Alemtan.v C.. Chasin, L. A., and Ciudad, C. J. Retinoblastomna protein

associates with SPI and activates the hamster dihydrofolate reductase promoter.hormonal modulations, and extracellular matrix perturbations, would Ontcogene, 16: 1931-1938, 1998.

likely result in malignancy. 23. Teng, C. T.. Liu, Y., Yang. N., Walmer. D.. and Panella. T, Differential molecularmechanism of the estrogen action that regulates lactoferrin gene in human and mouse.

Finally, it should be noted that the retinoblastoma protein has a Mol. Endocrinot., 6: 1969-1981, 1992.

complex set of functions within cells, and finding conditions that 24, Lussier, M.. Filion. M , Compton, J. G.. Nadeau. J. H.. Lapointe. L., and Royal, A.

allow separation of these functions has been challenging. Our obser- The mouse kcratin 19-enTcToding gene: sequence, structure and chrortmosotmal assign-wment. Gene (Amst.). 95: 203-213, 1990.vations indicate that culture in a three-dimensional matrix affords a 25. Berezutskaya. E., Yu. B.. Morozov, A., Raychaudhuri, P., and Bagchi, S. Differential

new operational definition of pRB activities that distinguishes those regulation of the pocket domains of the retinoblastorna family proteins by the HPVi6functions involved in polarity and initiation of differentiation from 26. -7 oncoprotein. Cell Growth Differ., 8: 1277-1286, 1997.

6. Mulligan, G., and Jacks. T. The retinoblastoma gene family: cousins with overlappingthose functions that define a fully differentiated cell. interests. Trends Genet.. 14: 223-229. 1998.

6045

BibliographyMazoyer S, Gayther SA, Nagai MA, Smith SA, Dunning A, van Rensburg EJ, Albertsen H, White

R, & Ponder BA. A gene (DLG2) located at 17q12-q21 encodes a new homologue of theDrosophila tumor suppressor dIg-A. Genomics. 1995 Jul 1;28(l):25-31.

Smith SA, Holik PR, Stevens J, Melis R, White R, & Albertsen H. Isolation and mapping of a geneencoding a novel human ADP-ribosylation factor on chromosome 17q12-q21. Genomics. 1995Jul 1;28(1):113-5.

Brothman AR, Steele MR, Williams BJ, Jones E, Odelberg S, Albertsen HM, Jorde LB, Rohr LR,& Stephenson RA. Loss of chromosome 17 loci in prostate cancer detected by polymerase chainreaction quantitation of allelic markers. Genes Chromosomes Cancer. 1995 Aug;13(4):278-84.

Smith SA, Holik P, Stevens J, Mazoyer S, Melis R, Williams B, White R, & Albertsen H. Isolationof a gene (DLG3) encoding a second member of the discs-large family on chromosome 17q12-q21. Genomics. 1996 Jan 15;31(2):145-50.

Albertsen HM, Smith SA, Melis R, Williams B, Holik P, Stevens J, & White R. Sequence, genomicstructure, and chromosomal assignment of human DOC-2. Genomics. 1996 Apr 15;33(2):207-13.

Karpf AR, Peterson PW, Rawlins JT, Dalley BK, Yang Q, Albertsen H, Jones DA. Inhibition ofDNA methyltransferase stimulates the expression of signal transducer and activator oftranscription 1, 2, and 3 genes in colon tumor cells. Proc Natl Acad Sci U S A. 1999 Nov23;96(24): 14007-12.

Spancake KM, Anderson CB, Weaver VM, Matsunami N, Bissell MJ, & White RL. E7-transducedhuman breast epithelial cells show partial differentiation in three-dimensional culture. CancerRes. 1999 Dec 15;59(24):6042-5.

List of Salaried Personnel:

Name: Position: Percent contribution to salary:Hans Albertsen, Research Instructor, 67%Jeff Stevens, Research Associate, 100%Ray White, P.I., Professor, 10%

18