12
Genetic identification of distinct loci controlling mammary tumor multiplicity, latency, and aggressiveness in the rat Xiaojiang Quan, 1 *  Jean-Franc ¸ois Laes, 1*à Daniel Stieber, 1 Miche `le Rivie `re, 1 Jose Russo, 2 Dirk Wedekind, 3 Wouter Coppieters, 4 Fre ´de ´ric Farnir, 4 Michel Georges, 4 Josiane Szpirer, 1 Claude Szpirer 1 1 Universite ´ Libre de Bruxelles, Institut de Biologie et de Me ´decine Mole ´culaires, Rue Profs Jeener & Brachet, 12, B-6041, Gosselies, Belgium 2 Breast Cancer Research Laboratory, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111, USA 3 Department of Animal Science, Medical School of Hannover, D-30623 Hannover, Germany 4 Department of Genetics, Faculty of Veterinary Medicine, University of Lie `ge (B43), B-4000 Lie `ge, Belgium Received: 19 September 2005 / Accepted: 9 January 2006 Abstract The rat is considered an excellent model for studying human breast cancer. Therefore, understanding the genetic basis of susceptibility to mammary cancer in this species is of great interest. Previous studies based on crosses involving the susceptible strain WF (crossed with the resistant strains COP or WKY) and focusing on tumor multiplicity as the susceptibility phenotype led to the identification of several loci that control chemically induced mammary cancer. The present study was aimed to determine whether other loci can be identified by analyzing crosses de- rived from another susceptible strain on the one hand, and by including phenotypes other than tumor multiplicity on the other hand. A backcross was generated between the susceptible SPRD-Cu3 strain and the resistant WKY strain. Female progeny were genotyped with microsatellite markers covering all rat autosomes, treated with a single dose of DMBA, and phenotyped with respect to tumor latency, tu- mor multiplicity, and tumor aggressiveness. Seven loci controlling mammary tumor development were detected. Different loci control tumor multiplicity, latency, and aggressiveness. While some of these loci colocalize with loci identified in crosses involving the susceptible strain WF, new loci have been uncovered, indicating that the use of distinct sus- ceptible and resistant strain pairs will help in establishing a comprehensive inventory of mam- mary cancer susceptibility loci. Introduction Breast cancer is a complex, multifactor disease that affects about 10% of women in industrialized coun- tries. One major risk factor for breast cancer is ge- netic predisposition (for reviews, see Martin and Weber 2000; Nathanson and Weber 2001). Two major breast cancer susceptibility genes (BRCA1, BRCA2) have been identified (Miki et al. 1994; Futreal et al. 1994; Wooster et al. 1995; Tavtigian et al. 1996), and 5%10% of all breast cancers can be explained by the inheritance of mutations in one of these two genes (Claus et al. 1996). However, a great deal remains to be understood with respect to the possible role of other genes involved in susceptibility to breast can- cer (Nathanson and Weber 2001; Claus et al. 1996, 1998; Ritchie et al. 2001; Pharoah et al. 2002). Some other susceptibility/modifier genes have been iden- tified and/or evaluated, mainly by association tests (Martin and Weber 2000; Nathanson and Weber 2001; The CHEK2-Breast Cancer Consortium 2002), but it is reasonable to assume that several additional genes remain to be discovered. Identifying genes that are not highly penetrant in human populations is a dif- ficult task (Dahlman et al. 2002), because of the ge- netic and environmental heterogeneity of most of the human population. Therefore, inbred animal models should provide an interesting tool for discovering genes that control susceptibility to breast cancer. *These authors equally contributed to work  Present address: KU Leuven, Laboratory of Neurogenetics, Department of Human Genetics, Flanders Interuniversity Insti- tute for Biotechnology, B-3000, Leuven, Belgium à Present address: Biovalle ´e, Rue Bolland, 8, B-6041, Gosselies (Charleroi), Belgium Correspondence to: Claude Szpirer; E-mail: [email protected] 310 DOI: 10.1007/s00335-005-0125-9 Volume 17, 310321 (2006) Ó Springer Science+Business Media, Inc. 2006

Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

  • Upload
    others

  • View
    5

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

Genetic identification of distinct loci controlling mammary tumormultiplicity, latency, and aggressiveness in the rat

Xiaojiang Quan,1*� Jean-Francois Laes,1*� Daniel Stieber,1 Michele Riviere,1 Jose Russo,2

Dirk Wedekind,3 Wouter Coppieters,4 Frederic Farnir,4 Michel Georges,4 Josiane Szpirer,1

Claude Szpirer1

1Universite Libre de Bruxelles, Institut de Biologie et de Medecine Moleculaires, Rue Profs Jeener & Brachet, 12, B-6041, Gosselies,Belgium2Breast Cancer Research Laboratory, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111, USA3Department of Animal Science, Medical School of Hannover, D-30623 Hannover, Germany4Department of Genetics, Faculty of Veterinary Medicine, University of Liege (B43), B-4000 Liege, Belgium

Received: 19 September 2005 / Accepted: 9 January 2006

Abstract

The rat is considered an excellent model for studyinghuman breast cancer. Therefore, understanding thegenetic basis of susceptibility to mammary cancer inthis species is of great interest. Previous studiesbased on crosses involving the susceptible strain WF(crossed with the resistant strains COP or WKY) andfocusing on tumor multiplicity as the susceptibilityphenotype led to the identification of several locithat control chemically induced mammary cancer.The present study was aimed to determine whetherother loci can be identified by analyzing crosses de-rived from another susceptible strain on the onehand, and by including phenotypes other than tumormultiplicity on the other hand. A backcross wasgenerated between the susceptible SPRD-Cu3 strainand the resistant WKY strain. Female progeny weregenotyped with microsatellite markers covering allrat autosomes, treated with a single dose of DMBA,and phenotyped with respect to tumor latency, tu-mor multiplicity, and tumor aggressiveness. Sevenloci controlling mammary tumor development weredetected. Different loci control tumor multiplicity,latency, and aggressiveness. While some of these locicolocalize with loci identified in crosses involvingthe susceptible strain WF, new loci have beenuncovered, indicating that the use of distinct sus-

ceptible and resistant strain pairs will help inestablishing a comprehensive inventory of mam-mary cancer susceptibility loci.

Introduction

Breast cancer is a complex, multifactor disease thataffects about 10% of women in industrialized coun-tries. One major risk factor for breast cancer is ge-netic predisposition (for reviews, see Martin andWeber 2000; Nathanson and Weber 2001). Two majorbreast cancer susceptibility genes (BRCA1, BRCA2)have been identified (Miki et al. 1994; Futreal et al.1994; Wooster et al. 1995; Tavtigian et al. 1996), and5%�10% of all breast cancers can be explained by theinheritance of mutations in one of these two genes(Claus et al. 1996). However, a great deal remains tobe understood with respect to the possible role ofother genes involved in susceptibility to breast can-cer (Nathanson and Weber 2001; Claus et al. 1996,1998; Ritchie et al. 2001; Pharoah et al. 2002). Someother susceptibility/modifier genes have been iden-tified and/or evaluated, mainly by association tests(Martin and Weber 2000; Nathanson and Weber 2001;The CHEK2-Breast Cancer Consortium 2002), but itis reasonable to assume that several additional genesremain to be discovered. Identifying genes that arenot highly penetrant in human populations is a dif-ficult task (Dahlman et al. 2002), because of the ge-netic and environmental heterogeneity of most of thehuman population. Therefore, inbred animal modelsshould provide an interesting tool for discoveringgenes that control susceptibility to breast cancer.

*These authors equally contributed to work�Present address: KU Leuven, Laboratory of Neurogenetics,Department of Human Genetics, Flanders Interuniversity Insti-tute for Biotechnology, B-3000, Leuven, Belgium�Present address: Biovallee, Rue Bolland, 8, B-6041, Gosselies(Charleroi), BelgiumCorrespondence to: Claude Szpirer; E-mail: [email protected]

310 DOI: 10.1007/s00335-005-0125-9 � Volume 17, 310�321 (2006) � � Springer Science+Business Media, Inc. 2006

Page 2: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

In this respect, the rat appears to be a very valu-able model. Indeed, the characteristics of rat mam-mary cancer are remarkably similar to those ofhuman breast cancer with respect to histopathology,hormone dependence, and lack of viral etiology(Gould 1993, 1995; Russo and Russo 1996a, 1996b;Shepel and Gould 1999). Several rat strains differstrikingly with regard to their susceptibility to bothspontaneous and induced mammary cancer (Shepeland Gould 1999; Isaacs 1986, 1988; Gould 1986).Interestingly, it has been shown that in one of theresistant strains (COP), resistance is a cell autono-mous rather than a systemic feature (Isaacs 1988;Gould 1986; Zhang et al. 1990) and that resistance isassociated with the loss of preneoplasic lesions ratherthan with the absence of such lesions (Korkola andArcher 1999). Gould et al. (1986) studied crosses be-tween COP (resistant) rats and WF (susceptible) ratsand showed that susceptibility to DMBA (7,12-dim-ethylbenz[a]anthracene)-induced mammary cancer iscontrolled by four quantitative trait loci (QTLs),named Mcs1 to Mcs4 (Mammary cancer susceptibil-ity) (Hsu et al. 1994; Shepel et al. 1998). Furthermore,by analyzing crosses between the same susceptiblestrain WF and another resistant strain WKY, thisteam identified five other loci (Mcs5 to Mcs8 andMcsm1) (Lan et al. 2001; Samuelson et al. 2003, 2005).These studies thus strongly suggest that mammarycancer resistance is controlled by different genes indifferent strains. This situation is similar to that ofuterine cancer susceptibility (Roshani et al. 2005) andof other quantitative traits such as hypertension orautoimmunity predisposition (Griffiths et al. 1999;Rapp 2000). On the other hand, rat strains also differin susceptibility to estrogen-induced mammary can-cer, and estrogen-induced mammary cancer suscep-tibility is not correlated with chemically inducedmammary cancer susceptibility. For instance, ACIrats are particularly susceptible to estrogen-inducedmammary cancer, though they are relatively resis-tant to chemically induced carcinogeneis (Isaacs1988; Shull et al. 1997). A cross between ACI andCOP rats recently led to the identification of twoQTLs that control this trait (Gould et al. 2004).

To identify new QTLs that control developmentof chemically induced mammary cancer, we chose toanalyze a cross involving a susceptible strain otherthan WF. We selected a SPRD inbred subline, SPRD-Cu3 (SPRD is an outbred strain), taking into accountthat SPRD rats have been shown to be susceptible tochemically induced mammary cancer (Isaacs 1986).A backcross was generated between SPRD-Cu3 ratsand WKY rats and we undertook an autosomal gen-ome scan to detect QTLs that influence threeparameters: number of mammary tumors (multi-

plicity), latency of tumor formation, and tumorgrowth rate.

Material and methods

Rat strains and genetic cross. The SPRD-Cu3 (acurly mutant of SPRD rats) (Greenhouse et al. 1990)and WKY/E56 (WKY) rats were obtained from theMedizinische Hochschule Hannover, Institut furVersuchstierkunde und Zentrales Tierlaboratorium(Hannover, Germany). SPRD-Cu3 females were ma-ted to WKY males, and F1 males were backcrossedwith SPRD-Cu3 females to generate females thatwere treated with DMBA.

Induction of mammary cancer and pheno-typing. At 53-58 days of age virgin females were in-jected intragastrically with a single dose of DMBA (65mg/kg, dissolved in sesame oil) (Gould 1986; Hugginset al. 1981). The treated animals were palpatedweekly to detect mammary tumors. The date ofdetection of the first tumor was recorded and the timebetween the date of DMBA treatment and the date ofdetection was defined as the latency. The diameter ofeach tumor was then measured once a week and theaverage change in diameter per week, measured dur-ing the last three weeks of tumor growth, was thetumor growth rate. Animals were killed 19 weeksafter DMBA treatment, except when a rapidly grow-ing tumor had developed. In that case, the affectedanimal was killed when the tumor had reached adiameter of about 5.0 cm or had caused an openwound, and the weekly tumor growth rate was cal-culated for the period of tumor growth, generallyshorter than three weeks. All animals were necrop-sied and the total number of tumors was counted andrecorded. Tumors were removed and fixed in 10%formalin.

Genotyping. DNA was obtained from tail clipsisolated at the age of about 25 days. Primer se-quences of microsatellite markers were obtainedfrom the MIT database at http://waldo.wi.mit.edu/rat/public/, except D2Uwm14 (Shepel et al. 1998).Markers that were polymorphic between the SPRD-Cu3 and WKY rats were selected to genotype the 187backcross females. PCR reactions were performed ina 25-ll final volume using 20 ng of genomic DNA, 25lM of each primer, 200 lM dNTPs, 0.5 unit of Taqpolymerase, and enzyme buffer. The PCR productswere separated on acrylamide gel. The DNA wasstained by SYBR Green I nucleic acid gel stain(Roche Diagnostics Belgium, Brussels, Belgium) andvisualized with an Image Master VDS (PharmaciaBiotech, Roosendaal, The Netherlands).

X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER 311

Page 3: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

Map construction. The microsatellite markermap was constructed using the TWOPOINT,BUILD, and CHROMPIC options of the CRIMAPpackage (Lander and Green 1987).

QTL mapping. QTL mapping was performedusing a previously described sum of rank-basednonparametric approach (Kruglyak and Lander1995a) implemented with HSQM software (Coppi-eters et al. 1998). Briefly, to measure the evidence infavor of a QTL at a given map position, we define thefollowing statistic for an (A · B) · A backcross:

Zw sð Þ ¼ Yw sð Þ� ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Yw

sð Þ2D Er

where

Yw sð Þ ¼Xn

i ¼ 1

n þ 1 � 2:rank ið Þ½ �

P gi;AA sð Þ gi;L ; gi;R

���� �

P gi;AB sð Þ gi;L ; gi;R

��� in which n is the number of progeny, rank(i) is therank by phenotype of progeny i, P[gi,AA(s)|gi,L,gi,R] isthe probability that progeny i has genotype AA atmap position (s) given its genotype at the left (gi,L)and right (gi,R) flanking markers, P[gi,AB(s)|gi,L,gi,R] isthe probability that progeny i has genotype AB atmap position (s) given its genotype at the left (gi,L)

and right (gi,R) flanking markers, and

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiY

wsð Þ2

D Eris the standard deviation of YW(s), expected under thenull hypothesis of no QTL over all possible sets ofgenotypes. Under the null hypothesis of no QTL, ZW

is shown to behave asymptotically as a standardnormal variable that reduces to a Wilcoxon rank-sum test at the marker positions. To facilitatecomparison with other methods, evidence in favor ofa QTL at a given map position was reported as

ZW sð Þ2.

2 ln 10ð Þ

which has the same distribution as the LOD score.

Information content mapping. Informationcontent along the marker map (Kruglyak and Lander1995c) was measured as

var P gi;AA sð Þ gi;L ; gi;R

��� �� P gi;AB sð Þ gi;L ; gi;R

��� ��

¼

Pni ¼ 1

P gi;AA sð Þ gi;L ; gi;R

��� �� P gi;AB sð Þ gi;L ; gi;R

��� �� 2

n � 1

where n is the total number of backcross offsprings.

LOD score thresholds for significant and sug-gestive QTLs. Significance thresholds were deter-mined by phenotype permutation following

Churchill and Doerge (1995). Phenotypic valueswere permutated 2000 times among backcross off-spring, and a whole-genome scan was performed oneach permutated data set. The highest LOD scoresobtained on each permutated data set were storedand ranked. The statistical significance of the LODscores obtained with the real data were then simplymeasured by comparison with this list and expressedas the probability (p-value) of obtaining such a largeor larger LOD score under the null hypothesis of noQTL. These p-values therefore account for the mul-tiple tests performed as a result of the analysis ofthe 118 markers composing our map. FollowingKruglyak and Lander (1995b), genome-wide p-valuesof 0.05 or less (= LOD score of 3.25 or more) wereconsidered significant evidence in favor of a QTL atthe given map position, while p-values of 0.64 (cor-responding to the occurrence of, on average, onesuch LOD score per permutation) or more wereconsidered suggestive evidence (= LOD score of1.76). Note that these thresholds are virtually iden-tical to those obtained by Kruglyak and Lander(1995b) using a theoretical approach (significant andsuggestive LOD scores of 3.3 and 1.9, respectively, ina backcross experiment).

Confidence intervals for the QTL posi-tion. Empirical confidence intervals for the positionof the QTL were calculated using a bootstrap re-sampling method according to Visscher et al. (1996).For chromosomes exhibiting a significant or sug-gestive QTL, we determined the most likely QTLposition for 1000 bootstrap samples. The confidenceinterval for the QTL location was determined as theshortest contiguous chromosome interval contain-ing 95% of these most likely positions.

Interactions between QTLs. Possible interac-tions between QTLs were tested using linear modelsthat included the individual QTL effect plus corre-sponding pairwise interactions. Microsatellitemarkers closest to the most likely position of theQTL were used to sort backcross individuals by QTLgenotype. Analyses were performed using the GLMprocedure of the SAS package (SAS Institute Inc.1991).

Results

Phenotyping: tumor multiplicity, latency, andaggressiveness in the N2 generation. Figure 1shows the tumor incidence and multiplicity in thetwo parental strains and in the (SPRD-Cu3 · WKY)F1 and SPRD-Cu3 · F1 backcross (N2) progenies. Ourobservations confirm that SPRD rats, and particu-

312 X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER

Page 4: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

larly the SPRD-Cu3 substrain rats, are susceptible tomammary cancer (Isaacs 1986): all treated females(10 animals) developed multiple mammary tumors(6-12 tumors per animal) within 8 weeks of DMBAtreatment. We also observed that the WKY-E56strain is resistant (no tumor in 10 treated females), asreported previously for WKY rats (Haag et al. 1992).In the F1 population, 34% of the treated rats (n = 32)developed mammary tumors; however, the numberof tumors per rat was fewer than 4. In the backcrossgeneration (n = 187), 95% of the animals developedmammary tumors, with the tumor multiplicityranging from 1 to 12.

The latency of tumor appearance (defined by thetime of detection of the first tumor) in the F1 popu-lation was more variable and sometimes longer thanin the SPRD-Cu3 parental strain. In the N2 popula-tion, the latency was more variable than in the F1

population, as illustrated in Figure 2. This featuresuggests that tumor latency itself is a polygenic trait.

The tumor growth rate was also variable, rangingfrom about 0.1 up to 5.0 cm in diameter/week

(Fig. 3). Some animals of the backcross generationdeveloped tumors reaching a diameter of 4.0-5.5 cmwithin 1-3 weeks after detection. These fast-growingN2 tumors often invaded the neighboring muscleand dermis, leading to ulcerations. Histologically,these tumors were classified as invasive adenocar-cinomas with a cribriform pattern (Fig. 4c, d). Giventhe invasiveness of these fast-growing tumors, wewill designate them as aggressive tumors. The lessaggressive (more slowly growing) tumors, which aremore typical of the tumors generally found in SPRDrats, were classified as papillary adenocarcinomastypes I and II (Fig. 4a, b) (Russo and Russo 1996a,2000). In this experiment, no aggressive tumors wereseen in the parental strain SPRD-Cu3. However, thisdifference is probably not significant. Indeed, wetested fewer animals in this group (10 SPRD-Cu3 vs.32 F1 and 187 N2 animals) and the most likelyexplanation is thus that the absence of aggressiveSPRD-Cu3 tumors simply reflects a statistical drift(in independent later experiments, aggressive tumorswere seen in SPRD-Cu3 females).

Fig. 1. Tumor incidence andmultiplicity in the two rat parentalstrains (n = 10 for each strain), andin the F1 (n = 32) and N2 (n = 187)generations. The histogram showsthe fraction of rats in each of thefive groups defined by the number oftumors indicated (from 0 to 10�12tumors per animal).

Fig. 2. Time course of DMBA-induced mammary tumors in thetwo rat parental strains and in the F1

and the N2 generations (samenumbers of rats as in Fig. 1).

X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER 313

Page 5: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

Linkage analysis and QTL genetic identifica-tion. One hundred eighty-seven female backcrossprogeny were genotyped for a panel of 118 autosomalmicrosatellites segregating in the backcross genera-tion. Two-point linkage analyses were first per-formed using TWOPOINT between all possiblemarker pairs to sort them by linkage group. Markerswere then ordered within linkage groups using theBUILD option of CRIMAP. All markers could beordered with odds versus all alternative orderssuperior to 1000:1. The resulting map covers a totalof 1745 cM (Kosambi) with an average marker

interval of 22.4 cM. Figure 5 illustrates the infor-mation content obtained across the genome. Itaverages 78%, ranging from 28% to 100%.

QTL mapping. Using a sum of rank-based non-parametric approach described in the Material andmethods section, we found significant evidence (z ‡3.25) for a QTL that affects tumor growth rate onChromosome 18 and suggestive evidence (z ‡ 1.76)for six additional QTLs: four that influence tumormultiplicity located on Chromosomes 1 (one at eachend of the chromosome), 2, and 5; one that influ-ences tumor latency on Chromosome 9; and one thatinfluences tumor growth rate on Chromosome 10(Figs. 5 and 6). The genome-wide significance levelsof the different QTLs are reported in Figure 5.

QTLs that control tumor number. Chromo-some 1 likely carries at least two distinct QTLs thatcontrol tumor number: one at the 1p telomeric end(peak at D1Rat246) associated with a LOD score of2.4, and the other at the 1q telomeric end in thevicinity of marker D1Rat88 associated with a LODscore of 3.0 (Fig. 6). These will be referred to asChromosome (Chr) 1p and 1q (suggestive) QTL,respectively. Supporting the hypothesis of two dis-tinct QTLs is the observation that for the Chr 1qQTL, the SPRD allele increases tumor number,while for the Chr1p QTL, it is unexpectedly theWKY allele that increases tumor number, i.e., a so-called ‘‘cryptic’’ allele. This allowed us to determine

Fig. 3. Growth rate of the DMBA-induced mammary tu-mors in the susceptible parental strain SPRD-Cu3 and inthe F1 and N2 generations (same numbers of rats as inFig.1).

Fig. 4. Histopathology of tumors. (a)Invasive papillary adenocarcinomatypes I and II, magnification: 4·. (b)Higher magnification (20·) of (a).(c)Invasive adenocarcinoma withcribriform pattern and comedonecrosis, magnification: 4·. (d)Higher magnification (20·) of (c).The staining was hematoxylin &eosin.

314 X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER

Page 6: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

confidence intervals separately for the two QTLs byaccounting for the sign of the QTL effect during thebootstrapping procedure. Unexpectedly, the confi-dence interval for the Chr 1q QTL, associated withthe highest LOD score, spans approximately 135 cM,while that for the Chr 1p QTL is confined to the veryend of the chromosome arm 1p. This could be be-cause additional QTLs with effects in the samedirection as the Chr 1q QTL are located on Chro-mosome 1 as suggested by the location score profileon this chromosome. One may expect that a single-QTL model applied to a chromosome harboring twoQTLs with opposing effects would tend to artifac-tually increase the distance between the two QTLs.To test this hypothesis we applied a two-QTL linearregression model to our data using QTL Express(Seaton et al. 2002). These results in essence con-firmed the results obtained with HSQM, i.e., thehighest test statistic (p = 0.00014) was obtainedwhen fitting one QTL at each end of the chromo-some (Fig. 7). Figure 6 (inset) shows the meannumber of tumors for BC rats sorted by D1Rat246 orD1Rat88 genotype. The effect of both QTLs can beseen to be of the order of one tumor per rat.

Suggestive evidence for two additional QTLsinfluencing tumor number was found on Chromo-somes 2 and 5 (Fig. 6). The most likely position ofthe QTL on Chromosome 2, associated with a LODscore of 2.6, coincides with marker D2Rat4 at posi-

tion 34 cM. However, the confidence interval for thisQTL spans as much as 140 cM on this chromosome.As for the Chr 1p QTL, it is the ‘‘cryptic’’ WKY alleleof this QTL that increases tumor number byapproximately one unit. The QTL on Chromosome 5yields a maximum LOD score of 2.5 at map position57 cM. Based on the available data, the confidenceinterval for the map position spans approximately 60cM. For this QTL, it is as expected, i.e., the SPRDallele increases the number of tumors by a littlemore than one unit. The LOD score profile onChromosome 5 is quite broad and trimodal, sug-gesting the possible occurrence of two distinct QTLsapproximately at positions 25 and 86 cM on thischromosome, which would jointly generate a ‘‘ghostQTL’’ in between the two under a single-QTLmodel. The same data therefore were analyzed byComposite Interval Mapping using QTL cartogra-pher. This analysis yielded a maximum signalaround position 57 cM as well (data not shown),which does not support the hypothesis of a ghostQTL. The broadness of the LOD score profile nev-ertheless suggests the occurrence of multiple QTLsinfluencing tumor number on Chromosome 5.

The joint effect of these four QTLs on tumormultiplicity was analyzed using a linear model (SASGLM procedure) that included all pairwise interac-tions in addition to the four individual effects. Thegenotype at the nearest marker locus was used to

Fig. 5. Information content and QTL detection. The continuous gray line (top) measures the information content of theutilized microsatellite marker map across the rat autosomes computed as described in Materials and methods. The otherline (bottom) illustrates the statistical evidence in favor of the presence of QTLs influencing tumor multiplicity (D),latency (O), and agressiveness (h) across the genome expressed as log(1/p), where p corresponds to the probability ofobserving the corresponding signal under the null hypothesis of no QTL; p was computed by permutations as described inMaterials and methods and accounts for the analysis of multiple markers. The thresholds associated with suggestive orsignificant QTLs are marked by the interrupted horizontal lines.

X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER 315

Page 7: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

trace individual QTLs. This model was shown toexplain 20% of the total trait variance in the back-cross generation. As expected, all four QTLs provedindividually significant in this analysis. In addition,two of the six interactions were significant at the 5%level: Chr 1q QTL * Chr 2 QTL (p = 0.034) and Chr 2QTL * Chr 5 QTL (p = 0.028). Table 1 summarizes

the least squares means for the corresponding com-posite genotypes. It can be seen that having eitherthe SPRD Chr 1q QTL allele or the WKY Chr 2 QTLallele is sufficient to increase tumor number by two,but that having both ‘‘risk’’ alleles does not have acumulative effect. For the Chr 2 QTL * Chr 5 QTLinteraction, one can see that both risk alleles need to

Fig. 6. LOD score profiles obtained on Chromosomes 1, 2, 5, 9, 10, and 18 for tumor multiplicity (D) (main peaks onChromosomes 1, 2, and 5), latency (O) (main peak on Chromosome 9), and aggressiveness (h) (main peaks on Chromo-somes 10 and 18). Marker positions (in cM) are given along the horizontal axes (q telomere on the right-hand side). Verticalbars represent the distribution of most likely QTL positions across 1000 bootstrap samples; horizontal bars correspond tothe 95% confidence interval for the QTL location computed by bootstrapping as described in Materials and methods. Foreach QTL, the mean phenotypic value (±two standard errors) of rats sorted by genotype for the marker yielding the highestLOD score is shown (insets); in the inset of Chromosome 9, the latency is expressed in weeks; in the inset of Chromo-somes 10 and 18, growth rate is expressed in cm/week. The microsatellite markers defining the map positions shown (incM) were Chromosome 1: D1Rat246 (0), D1Rat15 (11), D1Rat260 (22), D1Rat200 (33), D1Rat380 (48), D1Rat276 (65),D1Rat284 (71), D1Rat324 (104), D1Rat181 (131), and D1Rat88 (150); Chromosome 2: D2Uwm14 (0), D2Mit29 (17), D2Rat4(34), D2Rat14 (61), D2Rat104 (93), D2Rat40 (109), D2Rat60 (140), D2Rat62 (144), and D2Rat169 (173); Chromosome 5:D5Rat120 (0), D5Rat123 (7), D5Rat190 (12), D5Rat2 (25), D5Rat138 (38), D5Rat100 (44), D5Rat196 (53), D5Rat157 (63),D5Rat184 (75), D5Rat114 (83), and D5Rat39 (86); Chromosome 9: D9rat139 (0), D9Rat41 (7), D9Rat158 (21), D9Rat63 (45),D9Rat8 (72), D9Rat110 (86), and D9Rat102; Chromosome 10: D10Rat 47 (0), D10Rat71 (11), D10Rat116 (43), D10Rat19(74), D10Rat203 (88), and D10Rat97 (118); Chromosome 18: D18Rat114 (0), D18Rat65 (10), D18Rat102 (17), D18Rat55 (32),D18Rat14 (46), D18Rat45 (57), and D18Rat44 (72).

316 X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER

Page 8: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

be present jointly to have an effect on tumor num-ber. This suggests that neither of these ‘‘risk’’ allelescan trigger tumor development on its own, neither inthe SPRD strain nor in the WKY strain.

QTLs that control tumor latency. A distinctlocus associated with tumor latency was found atposition 17 cM of Chromosome 9 (Fig. 6), associatedwith a LOD score of 3. The confidence interval spansthe proximal 20 cM of this interval. Surprisingly, itis the WKY allele that reduces the latency byapproximately two weeks. This region was notassociated with any of the two other parameters,tumor multiplicity or aggressiveness.

QTLs that control tumor aggressiveness. Twoloci that control aggressiveness, measured by tumorgrowth rate, were detected on Chromosomes 10 and18 and associated with LOD scores of 3.0 and 3.3,respectively. The most likely position for the Chr 10QTL is 95 cM but with a confidence interval span-ning essentially the whole chromosome. For the Chr18 QTL, the most likely position is 40 cM with aconfidence interval from approximately 20 to 50 cM.For both these QTLs it is, as expected, the SPRDallele that increases growth rate by, respectively, 0.2and 0.3 cm/week. The combined effect of the twoloci on tumor aggressiveness was examined using alinear model that includes both QTLs (traced with

Table 1. Tumor number as a function of Mscmu2 * Mscmu3 and Mscmu3 * Mscmu4 genotypes

Chr 1q * Chr 2 QTLs Chr 2 * Chr 5 QTLs

Genotype LS means Std. error LS means Std. error

SPRD/SPRD*SPRD/SPRD 4.94 0.49 3.60 0.36WKY/SPRD*SPRD/SPRD 3.00 0.37 5.34 0.31SPRD/SPRD*WKY/SPRD 4.98 0.38 3.52 0.35WKY/SPRD*WKY/SPRD 4.80 0.39 3.49 0.33

LS = least squares.

Fig. 7. Analysis of the effect ofChromosome 1 on tumor numberusing a two-QTL model in a linearregression setting (Seaton et al.2002). The most significant signal[log(1/p) = 3.85] is obtained whenfitting one QTL at each end of thechromosome, thereby confirmingthe results of the single-QTLanalysis.

X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER 317

Page 9: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

the nearest microsatellites) as well as their interac-tion. This model explains 9% of the total trait vari-ance without any evidence for an interaction effectbetween the two loci.

Discussion

In this study we presented suggestive evidence forthe occurrence of seven QTLs that control mammarycancer development and segregate in a SPRD-Cu3 ·WKY backcross. Different biological properties of thetumors appeared controlled by distinct genetic loci,with four of the QTLs controlling tumor multiplicity(on Chromosomes 1, 2, and 5). Table 2 summarizesthese observations.

Two tumor multiplicity QTLs (both are sugges-tive QTLs) map to Chromosome 1, like Mcs3 (Shepelet al. 1998). However, Mcs3 resides in the middle ofthe chromosome, while the two Chr 1 QTLs wedetected map at the ends of the chromosome. Thethree loci are thus probably distinct.

The Chr 2 QTL maps close to the centromere ofthis chromosome, in the same region as Mcs1, asusceptibility locus defined previously in crossesinvolving two rat strains, WF (susceptible) and COP(resistant), different from those used in the presentwork (Hsu et al. 1994; Shepel et al. 1998). It isnoteworthy that on the basis of the distribution ofmammary tumors in different crosses involving thestrains WF, COP, and WKY, Haag et al. (1992) pre-viously concluded that the resistant strains COP andWKY carry one common resistant gene or tightlylinked genes. However, our analysis indicates thatthe WKY allele of the Chr 2 QTL is a susceptibilityallele (at least compared with the SPRD-Cu3 allele).Mcs1 and this QTL (at this stage, a suggestive QTL)might thus be distinct loci.

The fourth QTL we identified as controlling tu-mor multiplicity maps to Chromosome 5. Theinterval containing this QTL is relatively broad[from D5Rat190 (coordinates on the rat genomicsequnce: 24,978,084) to about D5Rat39 (coordinates:152,935,699)] (Ensembl Rat genome browser: http://www.ensembl.org/Rattus_norvegicus/), and mightin fact include more than one QTL. It is homologous

to the region of mouse Chromosome 4 that containstwo QTLs (Mmtg1 and Mmtg2) that modulate themass of transgene (the polyoma middle T gene)-in-duced mammary tumors (Le Voyer et al. 2000). TheChr 5 QTL region also overlaps the interval definingMcs5, a QTL that controls DMBA-induced mam-mary cancer susceptibility identified in a cross be-tween WF and WKY rats and recently shown tocontain three subloci (Mc5a, Mc5b, and Mcs5c) lo-cated between coordinates 61.2 Mb and 84.4 Mb (Lanet al. 2001; Samuelson et al. 2003, 2005). The locuswe identified might be identical to at least some ofthese subloci but might also include other subloci,outside the interval harboring the Mcs5 subloci. Thislocus also includes the region defining the QTLEmca1, which controls estrogen-induced mammarycancer susceptibility identified in a cross betweenACI and COP; Emca1 is located between D5Rat53(coordinates: 103,682,700) and D5Rat57 (coordinates155,130,915) (Gould et al. 2004). The region incommon with Emca1, Mcstm (between D5Rat53and D5Rat39), includes the cyclin inhibitor genesCdkn2a (coordinates: 108,910,763) and Cdkn2b,encoding p16/pArf and p15, respectively, and the Junoncogene (coordinates: 115,359,397). These threegenes are thus obvious candidate genes. This chro-mosome region also contains a transformation sup-pressor gene (Sai1) (Islam et al. 1989; Szpirer et al.1994; Helou et al. 2000).

The QTL that controls tumor latency, assignedto Chromosome 9, is a highly suggestive QTL thatdoes not colocalize with any of the other cancersusceptibility loci defined so far in the rat (Shepelet al. 1998; Lan et al. 2001; Gould et al. 2004;Tanuma et al. 1998; Kindler-Rohrborn et al. 1999;Ushijima et al. 2000), but it overlaps the intervaldefining Epdm9, a QTL that controls pituitary tumormass (Wendell and Gorski 1997). In the mouse,numerous tumor susceptibility QTLs have beengenetically defined (review: Balmain and Nagase1998), including a few QTLs that control tumorlatency (Yamada et al. 1994; Pataer et al 1996;Le Voyer et al. 2000; 2001). Interestingly, some ofthese QTLs control the latency of transgene-inducedmammary tumors (Le Voyer et al. 2000); these loci

Table 2. Summary of the QTLs detected in this work

Phenotype Chromosome and markers Allele and effect

Multiplicity 1p, peak at D1Rat246 WKY, increase (cryptic allele)1q, peak at D1Rat88 WKY, decrease2, peak at D2Rat4 WKY, increase (cryptic allele)5, broad peak, centered on D5Rat196/ D5Rat157 WKY, decrease

Latency 9, peak at D9Rat158 WKY, decrease (cryptic allele)Aggressiveness 10, between D10Rat116 and D10Rat71 WKY, decrease

18, peak close to D18Rat14 WKY, decrease

318 X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER

Page 10: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

have been assigned to mouse Chromosomes 7, 9, and15, in regions homologous to rat Chromosomes 1, 8,and 7 (Watanabe et al. 1999; Nilsson et al. 2001), andit is thus unlikely that any of these regions is themouse homolog of QTL we assigned to rat Chro-mosome 9.

Finally, this study disclosed the existence of twoQTLs that control tumor aggressiveness, which weassigned to rat Chromosomes 10 and 18 (highlysuggestive and significant QTLs, respectively). Locithat control metastasis in transgene-induced mousemammary tumors have been assigned to Chromo-somes 7, 9, 17, and 19 (Lancaster et al. 2005). RatChromosome 10 is partially homologous to mouseChromosome 17, but rat Chromosome 18 has nohomology to any of these mouse chromosomes(Nilsson et al. 2001). Interestingly, Gould et al.(2004) recently assigned a QTL that controls estro-gen-induced mammary cancer susceptibility to ratChromosome 18. This locus (Emca2) is associatedwith D18Rat 21 (coordinates: 44,464,649) and con-trols tumor latency and incidence. In our study, theChr 18 QTL controls aggressiveness and seems to belocated closer to the 18q telomere [betweenD18Rat17 (coordinates: 60,542,643) and D18Rat45(coordinates: 77,584,101)]. It thus seems that thesetwo QTLs are distinct.

Acknowledgments

This work was supported by the National Fund forScientific Medical Research (FRSM), the NationalFund for Scientific Research (FNRS, Televie), ‘‘FBAssurances,’’ and the ‘‘Federation Belge contre leCancer.’’ J.-F. Laes and D. Stieber were supported bya FRIA fellowship and C. Szpirer is a ResearchDirector of the FNRS (Belgium).

References

1. Balmain A, Nagase H (1998) Cancer resistance genes inmice: models for the study of tumour modifiers.Trends Genet 14, 139�144

2. Churchill GA, Doerge R (1995) Empirical values forquantitative trait mapping. Genetics 138, 963�971

3. Claus EB, Schildkraut JM, Thompson WD, Risch NJ(1996) The genetic attributable risk of breast andovarian cancer. Cancer 77, 2318�2324

4. Claus EB, Schildkraut J, Iversen ES Jr, Berry D,Parmigiani G (1998) Effect of BRCA1 and BRCA2 onthe association between breast cancer risk and familyhistory. J Natl Cancer Inst 90, 1824�1829

5. Coppieters W, Kvasz A, Farnir F, Arranz JJ, Grisart B,et al. (1998) A rank-based nonparametric method formapping quantitative trait loci in outbred half-sib

pedigrees: application to milk production in a grand-daughter design. Genetics 149, 1547�1555

6. Dahlman I, Eaves IA, Kosoy R, Morrison VA, Heward J,et al. (2002) Parameters for reliable results in geneticassociation studies in common diseases. Nat Genet 30,149�150

7. Futreal PA, Liu Q, Shattuck-Eidens D, Cochran C,Harshman K, et al. (1994) BRCA1 mutations in pri-mary breast and ovarian carcinomas. Science 266,120�122

8. Gould KA, Tochacek M, Schaffer BS, Reindl TM,Murrin CR, et al. (2004) Genetic determination ofsusceptibility to estrogen-induced mammary cancer inthe ACI rat: mapping of Emca1 and Emca2 to chro-mosomes 5 and 18. Genetics 168, 2113�2125

9. Gould MN (1986) Inheritance and site of expression ofgenes controlling susceptibility to mammary cancer inan inbred rat model. Cancer Res 46, 1199�1202

10. Gould MN (1993) Cellular and molecular aspects ofthe multistage progression of mammary carcinogene-sis in humans and rats.. Semin Cancer Biol 4, 161�169

11. Gould MN (1995) Rodent models for the study of eti-ology, prevention and treatment of breast cancer. Se-min Cancer Biol 6, 147�152

12. Greenhouse DD, Festing MFW, Hasan S, Chen AL(1990) Catalogue of inbred strains of rats. In HedrichHHJ, ed. Genetic Monitoring of Inbred Strains of Rats,(New York: Gustav Fischer Verlag Press), pp 410�486

13. Griffiths MM, Encinas JA, Remmers EF, Kuchroo VK,Wilder RL (1999) Mapping autoimmunity genes. CurrOpin Immunol 11, 689�700

14. Haag JD, Newton MA, Gould MN (1992) Mammarycarcinoma suppressor and susceptibility genes in theWistar�Kyoto rat. Carcinogenesis 13, 1933�1935

15. Helou K, Lu XC, Montelius-Alatalo K, Stahl F,Klinga-Levan K, et al. (2000) A dual-color FISHframework map for the characterization of the Sai1tumor suppression region on rat chromosome 5.Genes Chromosomes Cancer 27, 362�372

16. Hsu LC, Kennan WS, Shepel LA, Jacob HJ, Szpirer C,et al. (1994) Genetic identification of Mcs-1, a ratmammary carcinoma suppressor gene. Cancer Res 54,2765�2770

17. Huggins CB, Ueda N, Wiessler M (1981) N-Nitroso-N-methylurea elicits mammary cancer in resistant andsensitive rat strains. Proc Natl Acad Sci. USA 78,1185�1188

18. Isaacs JT (1986) Genetic control of resistance tochemically induced mammary adenocarcinogenesis inthe rat. Cancer Res 46, 3958�3963

19. Isaacs JT (1988) Inheritance of a genetic factor from theCopenhagen rat and the suppression of chemicallyinduced mammary adenocarcinogenesis. Cancer Res48, 2204�2213

20. Islam MQ, Szpirer J, Szpirer C, Islam K, Dasnoy JF,et al. (1989) A gene for the suppression of anchorageindependence is located in rat chromosome 5 bandsq22�23, and the rat alpha-interferon locus maps to thesame region. J Cell Sci 92, 147�162

X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER 319

Page 11: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

21. Kindler-Rohrborn A, Kolsch BU, Fischer C, Held S,Rajewsky MF (1999) Ethylnitrosourea-induced devel-opment of malignant schwannomas in the rat: twodistinct loci on chromosome 10 involved in tumorsusceptibility and oncogenesis. Cancer Res 59,1109�1114

22. Korkola JE, Archer MC (1999) Resistance to mammarytumorigenesis in Copenhagen rats is associated withthe loss of preneoplastic lesions. Carcinogenesis 20,221�227

23. Kruglyak L, Lander ES (1995a) A nonparametric ap-proach for mapping quantitative trait loci. Genetics139, 1421�1428

24. Kruglyak L, Lander ES (1995b) Genetic dissection ofcomplex traits: guidelines for interpreting and report-ing linkage results. Nat Genet 11, 241�247

25. Kruglyak L, Lander ES (1995c) Complete multipointsib-pair analysis of qualitative and quantitative traits.Am J Hum Genet 57, 439�454

26. Lan H, Kendziorski CM, Haag JD, Shepel LA, NewtonMA, et al. (2001) Genetic loci controlling breast cancersusceptibility in the Wistar�Kyoto rat. Genetics 157,331�339

27. Lancaster M, Rouse J, Hunter KW (2005) Modifiers ofmammary tumor progression and metastasis onmouse Chromosomes 7, 9, and 17. Mamm Genome 16,120�126

28. Lander ES, Green P (1987) Construction of multilocusgenetic linkage maps in humans.. Proc Natl Acad SciUSA 84, 2363�2367

29. Le Voyer T, Lu Z, Babb J, Lifsted T, Williams M, et al.(2000) An epistatic interaction controls the latency of atransgene-induced mammary tumor. Mamm Genome11, 883�889

30. Le Voyer T, Rouse J, Lu Z, Lifsted T, Williams M, et al.(2001) Three loci modify growth of a transgene-in-duced mammary tumor: suppression of proliferationassociated with decrease microvessel density. Ge-nomics 74, 253�261

31. Martin AM, Weber BL (2000) Genetic and hormonalrisk factors in breast cancer. J Natl Cancer Inst 92,1126�1135

32. Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA,Harshman K, et al. (1994) A strong candidate for thebreast and ovarian cancer susceptibility gene BRCA1.Science 26, 66�71

33. Nathanson KL, Weber BL (2001) ‘‘Other’’ breast cancersusceptibility genes: searching for more holy grail.Hum Mol Genet 10, 715�720

34. Nilsson S, Helou K, Walentinsson A, Szpirer C,Nerman O, et al. (2001) Rat�mouse and rat�hu-man comparative maps based on gene homologyand high-resolution zoo-FISH. Genomics 74,287�298

35. Pataer A, Kamoto T, Lu LM, Yamada Y, Hiai H (1996)Two dominant host resistance genes to pre-B lym-phoma in wild-derived inbred mouse strain MSM/Ms.Cancer Res 56, 3716�3720

36. Pharoah PDP, Antoniou A, Bobrow M, Zimmern RL,Easton DE, et al. (2002) Polygenic susceptibility to

breast cancer and implications for prevention. NatGenet 31, 33�36

37. Rapp JP (2000) Genetic analysis of inherited hyper-tension in the rat. Physiol Rev 80, 135�172

38. Ritchie MD, Hahn LW, Roodi N, Bailey LR, DupontWD, et al. (2001) Multifactor-dimensionality reduc-tion reveals high-order interactions among estrogen-metabolism genes in sporadic breast cancer. Am JHum Genet 69, 138�147

39. Roshani L, Mallon P, Sjostrand E, Wedekind D, SzpirerJ, et al. (2005) Genetic analysis of susceptibility toendometrial adenocarcinoma in the BDII rat model.Cancer Genet Cytogenet 158, 137�141

40. Russo IH, Russo J (1996a) Mammary gland neoplasia inlong-term rodent studies. Environ Health Perspect104, 938�967

41. Russo J, Russo IH (1996b) Experimentally inducedmammary tumors in rats. Breast Cancer Res Treat 39,7�20

42. Russo J, Russo IH (2000) Atlas and histologic classifi-cation of tumors of the rat mammary gland. J. Mam-mary Gland Biol Neoplasia 5, 187�200

43. Samuelson DJ, Haag JD, Lan H, Monson DM, ShultzMA, et al. (2003) Physical evidence of Mcs5, a QTLcontrolling mammary carcinoma susceptibility, incongenic rats. Carcinogenesis 24, 1455�1460

44. Samuelson DJ, Aperavich BA, Haag JD, Gould MN(2005) Fine mapping reveals multiple loci and a pos-sible epistatic interaction within the mammary car-cinoma susceptibility quantitative trait locus Mcs5.Cancer Res 65, 9637�9642

45. SAS Institute Inc. (1991) SAS System for Linear Mod-els, 3rd edn. (Cary, NC: SAS Institute Inc.)

46. Seaton G, Haley CS, Knott SA, Kearsey M, Visscher P(2002) QTL Express: mapping quantitative trait loci insimple and complex pedigrees. Bioinformatics 18,339�340

47. Shepel LA, Gould MN (1999) The genetic componentsof susceptibility to breast cancer in the rat. Prog ExpTumor Res 3, 158�169

48. Shepel LA, Lan H, Haag JD, Brasic GM, Gheen ME,et al. (1998) Genetic identification of multiple loci thatcontrol breast cancer susceptibility in the rat. Genetics149, 289�299

49. Shull JD, Spady TJ, Snyder MC, Johanssson SL, Pen-nington KL (1997) Ovary-intact, but not ovariecto-mized female ACI rats treated with 17beta-estradiolrapidly develop mammary carcinoma. Carcinogenesis18, 1595�1601

50. Szpirer C, Tissir F, Riviere M, Levan G, Szpirer J(1994) Assignment of the rat Jun family genes toChromosome 19 (Junb), Chromosome 5q31�33 (Jun),and Chromosome 16 (Jund). Mamm Genome 5,361�364

51. Tanuma J, Shisa H, Hiai H, Higashi S, Yamada Y, et al.(1998) Quantitative trait loci affecting 4-nitroquino-line 1-oxide-induced tongue carcinogenesis in the rat.Cancer Res 58, 1660�1664

52. Tavtigian SV, Simard J, Rommens J, Couch F, Shat-tuck-Eidens D, et al. (1996) The complete BRCA2 gene

320 X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER

Page 12: Genetic identification of distinct loci controlling mammary ...servnt.udi.fmv.ulg.ac.be/genmol/Department/Publications/Quan_2006.pdf · Genetic identification of distinct loci controlling

and mutations in chromosome 13q-linked kindreds.Nat Genet 12, 333�337

53. The CHEK2-Breast Cancer Consortium(2002) Low-penetrance susceptibility to breast cancer due toCHEK2(*)1100delC in noncarriers of BRCA1 orBRCA2 mutations. Nat Genet 31, 55�59

54. Ushijima T, Yamamoto M, Suzui M, Kuramoto T,Yoshida Y, et al. (2000) Chromosomal mapping ofgenes controlling development, histological grade,depth of invasion, and size of rat stomach carcinomas.Cancer Res 60, 1092�1096

55. Visscher PM, Thompson R, Haley CS (1996) Confi-dence intervals in QTL mapping by bootstrapping.Genetics 143, 1013�1020

56. Watanabe TK, Bihoreau MT, McCarthy LC, KiguwaSL, Hishigaki H, et al. (1999) A radiation hybrid map of

the rat genome containing 5,255 markers. Nat Genet22, 27�36

57. Wendell DL, Gorski J (1997) Quantitative trait loci forestrogen-dependent pituitary tumor growth in the rat.Mamm Genome 8, 823�829

58. Wooster R, Bignell G, Lancaster J, Swift S, Seal S, et al.(1995) Identification of the breast cancer susceptibilitygene BRCA2. Nature 378, 789�792

59. Yamada Y, Matsushiro H, Ogawa MS, Okamoto K,Nakakuki Y, et al. (1994) Genetic predisposition topre-B lymphomas in SL/Kh strain mice. Cancer Res 54,403�407

60. Zhang R, Haag JD, Gould MN (1990) Site ofexpression and biological function of the rat mam-mary carcinoma suppressor gene. Carcinogenesis 11,1765�1770

X. QUAN ET AL.: IDENTIFICATION OF LOCI CONTROLLING MAMMARY CANCER 321