EDITORIALS

Genetic Control of CancerMetastasis

Isaiah J. Fidler, * Robert Radinsky

Cancer is a complex of many types of malignant neoplasticgrowths, and each cancer of each organ exhibits numeroussubsets. Despite this heterogeneity, malignant neoplasms have auniform characteristic—the ability to invade host tissues andproduce metastases. The most fearsome aspect of cancer ismetastasis. This fear is well based. There have been greatimprovements in diagnosis, general patient care, surgical tech-niques, and local and systemic adjuvant therapies, but mostdeaths from cancer are still due to metastases that are resistant toconventional therapies. Continual empiricism is unlikely to pro-duce significant improvement. Rather, a better understanding ofthe metastatic process and the development of biologic heteroge-neity in neoplasms should provide a basis for new approaches tomore effective therapy. Identification of specific propertiesunique to metastatic cells is an obvious prerequisite.

The process of cancer metastasis consists of linked sequentialsteps. Many investigators (1-3) but not all (4) view this process asfavorable to the survival of subpopulations of metastatic cellspre-existent in the primary neoplasm (5). Metastasis begins withthe invasion of the surrounding normal stroma either by singletumor cells with increased motility or by a group of cells from theprimary tumor. Once the invading cells penetrate the vascular orlymphatic channels, they may grow there, or a single cell orclump(s) of cells may detach and be transported within thecirculatory system. Tumor emboli must survive the host's im-mune and nonimmune defenses and the turbulence of the circula-tion, arrest in the capillary bed of receptive organs, extravasateinto the organ parenchyma, proliferate, and establish a microme-tastasis. Growth of these small tumor lesions requires the devel-opment of a vascular supply and continuous evasion of hostdefense cells. When the metastases grow, they can shed tumorcells into the circulation and thus produce metastasis of me-tastases (7,2).

There is now wide acceptance that many malignant tumorscontain heterogeneous subpopulations of cells with differences insuch properties as growth rate, antigenic and immunogenicstatus, cell-surface receptors and products, response to cytotoxicagents, invasiveness, and metastatic potential. For production ofclinically relevant metastases, all of the steps outlined in figure 1must be completed. Failure to complete one or more steps of theprocess eliminates the cells. For this reason, the failure of mosttumor cells to produce a metastasis can be due to different single

or multiple deficiencies, such as inability to invade host stroma, ahigh degree of antigenicity, lack of aggregation, inability to arrestin the capillary bed, and above all, inability to grow in a distantorgan's parenchyma. Searching for a uniform factor that preventstumor cells from producing metastasis is therefore unproductive.

During the last decade, numerous laboratories have succeededin isolating cells with different metastatic properties from avariety of experimental rodent and human neoplasms. The avail-ability of these selected subpopulations with known quantitativeand/or qualitative differences in metastatic potential has facili-tated numerous studies on the biochemical characterization of the"metastatic phenotype" (3,6-8). With the new advances inmolecular biology, active research has been launched to identify(a) the genetic control of the metastatic phenotype and/or (b) thegenes responsible for regulating discrete steps of the metastaticprocess.

It is unlikely that a complex process such as cancer metastasisis regulated by one gene; there is much evidence (3,9-12) tosupport the concept that this process is regulated by the activationand/or deactivation of many specific genes. Each discrete step ofmetastasis (e.g., invasion or extravasation) is probably regulatedby transient or permanent changes at the DNA or RNA level indifferent genes. Moreover, the specific family of gene productsmay be different for each histologic tumor type. The use of clonalpopulations of tumor cells for analysis of genetic mechanisms ofmetastasis through comparative techniques is complicated by thefact that, over weeks or months of continuous culture, someclonal populations are unstable with regard to full metastaticcompetency (13,14). Although this problem can be circumventedby use of polyclonal populations of metastatic cells (13), repeatedin vivo assays to validate metastatic competency are mandatory.

Genetic control of the metastatic phenotype has been recentlyaddressed by the use of several related approaches. Investigatorshave performed DNA-mediated transfer experiments in an at-tempt to isolate genes involved in metastasis, but to date, thisapproach has not identified any "candidate genes." These exper-iments involve the transfer of the metastatic phenotype to tumor-igenic, nonmetastatic cells by transfection of high-molecular-weight DNA. The feasibility of using this technique to identifygenes that may regulate metastasis depends on two assumptions:

(/) Transfection and assays to select genes with metastatic

transfectants must be highly efficient.(2) Recipient tumorigenic, nonmetastatic cells must have a

single dominant gene capable of switching on the entiremetastatic process or a single gene that is complementaryto an essential gene and that either is not expressed or is notpresent in cells already expressing all other genes control-ling metastasis.

Received December 26, 1989; accepted December 26, 1989.Supported by Public Health Service grant CA-42107 from the National Cancer

Institute, National Institutes of Health, Department of Health and HumanServices; and by Postdoctoral Fellowship grant PF-3446 from the AmericanCancer Society.

I. J. Fidler, R. Radinsky, Department of Cell Biology, The University ofTexas, M. D. Anderson Cancer Center, Houston, TX.

Correspondence to: Isaiah J. Fidler, M. D., Department of Cell Biology (173),The University of Texas, M. D.Anderson Cancer Center, 1515 Holcombe Blvd.,Houston, TX 77030.

166 Journal of the National Cancer Institute

at University of D

ayton on August 12, 2014

http://jnci.oxfordjournals.org/D

ownloaded from

IMETASTASES

Figure 1. Sequential steps in the pathogenesis of cancer metastasis. To producemetastases, tumor cells must complete every step in the process. If they fail tocomplete even one step, they are eliminated. Each discrete step of metastasis islikely to be regulated by transient or permanent changes in DNA, RNA, orproteins of multiple genes. Since nonmetastatic cells fail to produce metastasisbecause of one or more deficiencies, induction of metastatic competence indifferent nonmetastatic cells may involve activation or deactivation of differentgenes. Examples are (/) metastasis-competent cell, (2) deficiency in invasion orextravasation step, (3) deficiency in aggregation, survival in circulation, andarrest, (4 and 5) multiple deficiencies, (6) high immunogenicity and antigenicity,and (7) deficiency in ability to grow at metastatic sites.

Transfection of a single gene has shown that some oncogeneclasses such as ras, src, myc, mos, raf, fes, fms, sis, and p53 caninduce metastatic competency in some recipient cells. Althoughthe mechanism of action of these various oncogenes is not clear,a multigene cascade presumably involving signal transductionpathways is implicated (9,10,15). Independent transfections withEl a, H-2k, or H-2D genes have produced a decrease in themetastatic competency of some highly metastatic cells (10,16).The effects of a transfected gene depend on the properties of thegene per se, the point of integration into the genome, the geneticbackground of the recipient cell, environmental influences, thestability of the transfection, and the host and assay systems used(9,11,15). Thus, a finding in one tumor system does not neces-sarily negate a finding in another tumor system. In any event,transfection of an oncogene(s) into appropriate cells may providea system for studying biochemical properties of metastatic cells,since this procedure can produce a large number of variants(3,10).

Differential ("plus-minus") screening of cDNA libraries con-structed from metastatic and nonmetastatic cells of the sametumor type has been used to identify candidate genes in metasta-sis. This technique relies on two premises:

(/) that phenotypic changes associated with metastasis reflectchanges in the synthesis and activity of many gene prod-ucts and

(2) that these products are regulated by transcriptional andposttranscriptional mechanisms manifesting themselves asquantitative changes in steady-state levels of numerousmessenger RNA (mRNA) transcripts.

The method allows a search for genes that have either prefer-ential or decreased expression in metastatic cells, compared withexpression in nonmetastatic counterparts. This technique has ledto the recent identification of mRNAs that are differentiallyexpressed in metastatic and nonmetastatic cells. Examples in-clude fibronectin (17), transin (18), murine calcium-binding

protein (19), reduced nicotinamide adenine dinucleotide(NADH) dehydrogenase subunit 5 (20), and human acidic ribo-somal phosphoprotein P2 (27). Other mRNAs identified correlatewith metastasis in murine melanoma (12,22) and murine largecell lymphoma (23). Unfortunately, the sensitivity of this tech-nique is limited: it cannot detect transcripts present at a levelbelow 0.01% in the mRNA population.

The more sensitive "subtractive hybridization" technique hasbeen used to construct cDNA libraries from metastatic andnonmetastatic cells. This method involves physical removal ofsequences common to both RNA populations under study andallows the isolation of mRNA transcripts constituting a percent-age of the mRNA population as low as 0.001% (24). Hence,subtractive hybridization coupled with differential screeningcould increase the chance of cloning competent genes that arerequired, but are not of themselves sufficient, to significantlyalter metastatic behavior (12).

In this issue of the Journal, Phillips and associates describetheir use of these techniques in the isolation and partial character-ization of one cDNA clone (pGM21) that is overexpressed in twodifferent metastatic DMBA8 rat mammary adenocarcinomamodel systems. Recently, other investigators using similar tech-niques isolated two other genes, WDNM1 (25) and WDNM2(26), which have decreased expression in metastatic cells, com-pared with that in nonmetastatic cells in the same DMBA8 ratmammary adenocarcinoma system. Phillips and colleagues cor-rectly point out that it is still unclear whether these gene(s) exertdirect or indirect effects on metastasis. Nevertheless, there ismuch evidence (12,17-26) to support the concept that a complexprocess like cancer metastasis is regulated by the activationand/or deactivation of multiple specific genes.

Several biologic criteria must be fulfilled to determine whetheran isolated (candidate) gene plays a direct role in regulating themetastatic phenotype. The gene of interest must be identified andits pattern of expression determined. Full-length cDNA clonesshould be isolated, incorporated into eukaryotic expression vec-tors, and transfected into the appropriate recipient cells. Inaddition, the phenotype should be assayed in a relevant modelsystem: orthotopic implantation into anatomically correct organs(27). Finally, the reintroduced gene must be intact and functional.

Genes that are transcriptionally more active or less active inmetastatic cells than their nonmetastatic counterparts could betested in selected populations of cells with known differences inmetastatic potential. These differences should be quantitative(i.e., high vs. low metastatic potential) and qualitative (i.e.,deficiencies in a single step or multiple steps in the process ofmetastasis). The introduction of anti-sense strand RNA mole-cules that selectively inhibit the activity of genes and block theproduction of specific proteins could be used to confirm the roleof a particular gene in regulation of one or more steps in themetastatic process.

The recent identification of novel genes that are associated withquantitative changes in the steady-state levels of mRNA inmetastatic versus nonmetastatic cells adds important evidence tosupport the concept that cancer metastasis is not a randomprocess; it is a highly regulated process that can now be studied onthe molecular level. This new knowledge should eventually leadto the design and implementation of more effective therapy forthis dreaded disease.

Vol. 82, No. 3, February 7, 1990 EDITORIALS 167

at University of D

ayton on August 12, 2014

http://jnci.oxfordjournals.org/D

ownloaded from

References(/) POSTE G, FIDLER IJ: The pathogenesis of cancer metastasis. Nature

283:139-146, 1980(2) FIDLER LI, BALCH CM: The biology of cancer metastasis and implications for

therapy. Curr Probl Surg 24:137-209, 1987(J) NicOLSON GL: Tumor cell instability, diversification, and progression to the

metastatic phenotype: From oncogene to oncofetal expression. Cancer Res47:1473-1487, 1987

(4) WEISS L: Principles of Metastasis. Orlando, FL: Academic Press, 1985(5) FIDLER LJ, KRIPKE ML: Metastasis results from pre-existing variant cells

within a malignant tumor. Science 197:893-895, 1977(6) FIDLER LI: The Ernst W. Bertner Memorial Award Lecture: The evolution of

biological heterogeneity in metastatic neoplasms. In Cancer Invasion andMetastasis: Biologic and Therapeutic Aspects (Nicolson G, Milas L, eds).New York: Raven Press, pp 5-26, 1984

(7) AUKERMAN SL, PRICE JE, FIDLER LI: Different deficiencies in the preventionof tumorigenic-low-metastatic murine K-1735b melanoma cells from pro-ducing metastases. JNCI 77:915-924, 1986

(8) NICOLSON GL: Cancer metastasis: Tumor cell and host organ propertiesimportant in metastasis to specific secondary sites. Biochim Biophys Acta948:175-224, 1988

(9) HART IR, GOODE NT, WILSON RE: Molecular aspects of the metastaticcascade. Biochim Biophys Acta 989:65-84, 1989

(10) LIOTTA LA: Gene products which play a role in cancer invasion andmetastasis. Breast Cancer Res Treat 11:113-124, 1988

(11) COLLARDJG, ROOSE, LA RIVIERE G, ETAL: Genetic analysis of invasion andmetastasis. Cancer Surveys 7:691-710, 1988

(12) STEEG PS, BEVILACQUA G, KOPPER L, ET AL: Evidence for a novel geneassociated with low tumor metastatic potential. J Natl Cancer Inst80:200-204, 1988

(13) POSTE G, DOLL J, FIDLER IJ: Interactions between clonal subpopulationsaffect the stability of the metastatic phenotype in polyclonal populations ofB16 melanoma cells. Proc Natl Acad Sci USA 78:6226-6230, 1981

(14) POSTE G, TZENC J, DOLL J, ET AL: Evolution of tumor cell heterogeneityduring progressive growth of individual lung metastases. Proc Natl Acad SciUSA 79:6574-6578, 1982

(15) CHAMBERS AF, TUCK AB: Oncogene transformation and the metastaticphenotype. Anticancer Res 8:861-871, 1988

(16) FELDMAN M, EISENBACH L: Genes controlling the metastatic phenotype.Cancer Surveys 7:555-572, 1988

(17) SCHALKEN JA, EBELING SB, ISAACS JT, ET AL: Down modulation offibronectin messenger RNA in metastasizing rat prostatic cancer cellsrevealed by differential hybridization analysis. Cancer Res 48:2042-2046,1988

(18) MATRISIAN LM, BOWDEN GT, KRIEG P, ET AL: The mRNA coding thesecreted protease transin is expressed more abundantly in malignant than inbenign tumors. Proc Natl Acad Sci USA 83:9413-9417, 1986

(19) EBRALIDZE A, TULCHINSKY E, GRIGORIAN M, ET AL: Isolation and character-ization of a gene specifically expressed in different metastatic cells andwhose deduced gene product has a high degree of homology to a Ca2+-binding protein family. Genes Dev 3:1086-1093, 1989

(20) LABICHE RA, YOSHIDA M, GALLICK GE, ET AL: Gene expression and tumorcell escape from host effector mechanisms in murine large cell lymphoma. JCell Biochem 36:393-403, 1988

(21) ELVIN P, KERR IB, MCARDLE CS, ET AL: Isolation and preliminarycharacterization of cDNA clones representing mRNA's associated withtumor progression and metastasis in colorectal cancer. Br J Cancer57:36-42, 1988

(22) ROSENGARD AM, KRUTZSCH HC, SHEAM A, ET AL: Reduced Nm23/Awdprotein in tumour metastasis and aberrant Drosophila development. Nature342:177-180, 1989

(23) NICOLSON GL, LABICHE RA, FRAZIER ML, ET AL: Differential expression of

metastasis-associated cell surface glycoproteins and mRNA in a murinelarge cell lymphoma. J Cell Biochem 31:305-312, 1986

(24) HEDRICK SM, COHEN DI, NIELSEN EA, ET AL: Isolation of cDNA clonesencoding T cell-specific membrane-associated proteins. Nature 308:149-153, 1984

(25) DEAR TN, RAMSHAW IA, KEFFORD RF: Differential expression of a novelgene, WDNM1, in nonmetastatic rat mammary adenocarcinoma cells.Cancer Res 48:5203-5209, 1988

(26) DEAR TN, MCDONALD DA, KEFFORD RF: Transcriptional down-regulationof a rat gene, WDNM2, in metastatic DMBA-8 cells. Cancer Res49:5323-5328, 1989

(27) FIDLER IJ: Rationale and methods for the use of nude mice to study thebiology and therapy of human cancer metastasis. Cancer Metastasis Rev5:29-49, 1986

Cancer Is Complex

Getting theInformationIs Simple

For the latest cancerinformation from theNational Cancer Institute'sCancer Information Service,write:

The National Cancer InstituteBuilding 31, Room 10A24Bethesda, Maryland20892-3100 U.S.A.

Or Call:1-800-4-CANCER*(1-800-422-6237)

*In Hawaii, on Oahu call524-1234.Neighbor islandscall collect.

168 Journal of the National Cancer Institute

at University of D

ayton on August 12, 2014

http://jnci.oxfordjournals.org/D

ownloaded from