High Frequency of Interactions between Lung Cancer Susceptibility Genes in the Mouse: Mapping of SlucS to Slucl4l

Although several genes that cause monogenie familial cancer syndromes have been identified, susceptibility to sporadic cancer remains unresolved. Animal experiments have demonstrated multigenic control of tumor suscep tibility. Recently, we described four mouse lung cancer susceptibility (Slue) loci, the main effects of which are masked by their mutual interactions. Because such interactions can considerably affect the strategies for identifi cation of cancer susceptibility genes in humans, it is necessary to establish whether they are common or rare. Here, we report the mapping of 10 additional Simloci and show that 13 of the 14 Simloci are involved in one or more interactions, demonstrating that interactions of tumor susceptibility genes are frequent and that they probably form complex networks. Introduction Studies using inbred strains of mice demonstrate a multigenic basis for many traits, including susceptibility to cancer ( 1-4). Until recently, the approaches used to map loci involved in multigenic traits have been similar to those used to locate genes involved in monogenie traits. To increase the genetic resolution for mapping QTLs,4 we used RC strains, a system of mouse inbred strains in which the genetic complexity is reduced (5). Upon prenatal treatment with the carcinogen ENU, mice of the strain O20 were more susceptible to lung tumors than were mice of the strain B10.O20 (6). The O20-congenic-B10.O20/Dem (OcB) series of RC strains consists of 19 homozygous strains, each carrying a different random subset of —¿ 12.5% genes from the genome of the common donor strain B10.O20 on a background of 87.5% genes from the common background strain O20. In this way, the multigenic difference in lung tumor susceptibility between the parental strains O20 and B10.O20 was transformed into monooroligogenic differences (5). Because the size of lung tumors differs significantly between the RC strain OcB-9 and background strain O20 (7), we previously used (OcB-9 X O20)F2 hy brids for linkage analysis of lung tumor size. The data were evaluated using the MQM mapping statistical method, a multilocus method that is more powerful than single-QTL methods because, in the process of mapping one QTL, the main and interaction effects of other QTLs are taken into account (8-10). This analysis revealed four novel interacting susceptibility loci, Slucl-Sluc4 (11). The individual alÃ-elesof these loci are not intrinsically susceptible or resistant, but their effect is influenced by the genotype at the interacting locus. The allelic combinations of Slue I and Sluc2 corresponding to susceptibility are Slucl020fÃ-}20-Sluc2020/02° and siuel'"a02°/l"002u-Sluc2"!a°20""002", whereas the genotypes Slucl"2W"20-Sluc2ttl0°2â„¢"»°™ and Slucl"'" 02U/tt'°02u-Sluc202W02" Received 4/17/98; revised 8/17/98; accepted 9/18/98. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1This work was partially supported by European Commission Grant SCI-113 (to P. D.). 2 Present address: Department of Rheumatology, Allergy, and Immunology, Brigham and Women's Hospital. Smith Research Building. Room 628. I Jimmy Fund Way, Boston. MA 02115. *To whom requests for reprints should be addressed, at Division of Molecular Genetics, H-5. The Netherlands Cancer Institute. Plesmanlaan 121. 1066 CX Amsterdam, the Netherlands. 4 The abbreviations used are: QTL(s). quantitative trait locus (loci); RC. recombinant congenie; ENU. JV-ethyl-A'-nitrosourea: MQM. mulliple-QTL-model. are associated with resistance. A similar interaction was found between Sluc3 and Sluc4 and between the colon cancer susceptibility loci Scc4 and SccS (12). Due to these counteracting interactions, the main effect of the susceptibility loci is masked. We hypothesized that, if counteracting interactions are a common characteristic of lung cancer susceptibility, linkage analyses of crosses between O20 and RC strains with similar phenotypes could also result in mapping of multiple loci with relatively large effects. Hence, to map additional Slue loci and to establish whether interactions between them are common or exceptional, linkage analyses were performed for both lung tumor size and lung tumor number using three additional OcB strains. We analyzed F-, crosses between the common back ground strain O20 and each of the OcB-strains OcB-3, OcB-4, OcB-6, OcB-9 (tumor number only), and OcB-16, despite the fact that they do not differ significantly from strain O20 for these traits. We now report that this approach revealed the mapping of 10 additional Slue loci, Sluc5-Sluei4. Nine of these loci were found to be involved in one or more interactions, demonstrating that interactions between lung tumor susceptibility genes are frequent. Materials and Methods Mice and Carcinogen Treatment. The H2 congenie strain B10.O20/Dem (denoted B10.O20) carries the H2r: haplotype of strain O20/A (denoted O20) on the C57BL/10 background (N8). The mice used in this study received acidified drinking water and a standard laboratory diet (Hope Farms. Woerden, the Neth erlands) ad libitum. They were subjected to a strict light-dark regimen. F2-hybrid mice were obtained from crosses between the common background strain O20 and each of the OcB strains OcB-3, OcB-4. OcB-6. OcB-9. and OcB-16. The tumor number and tumor size (square root-log-transformed, respectively) for these strains were: O20, 0.78 ±0.09/3.27 ±0.14; OcB-3, 0.49 ±0.11/2.65 ±0.22: OcB-4. 0.97 ±0.11/3.02 ±0.14: OcB-6, 0.63 ±0.10/2.72 ±0.18; OcB-9, 0.68 ±0.11/ 2.31 ±0.18; and OcB-16.0.74 ±0.10/2.78 ±0.18. The tumor size of OcB-9 was significantly different from that of O20. F,-hybrid mice were mated overnight. At day 18 of gestation, pregnant F, mice were treated ¡.p. with 40 mg/kg body weight of ENU as described previously (11). The ENU-treated offspring was killed at 16 weeks of age, except for the (O20 X OcB-6)F2 cross, in which one group of mice was killed at 16 weeks and a second group, denoted (O20 X OcB-6)F2" wk,was killed at 35 weeks of age. The lungs were fixed in 40% (v/v) ethanol, 5% (v/v) acetic acid, 4% (v/v) formaldehyde, and 0.41% (m/v) NaCl and embedded in histowax. Phenotyping and Genotyping. The whole lungs were sectioned semiserially (5-fj.m sections at 100-/xm intervals (5-¿unsections at 100-f¿mintervals) and stained with H&E. Lung tumor number and lung tumor size were determined microscopically as described (11). DNA of F2-hybrid mice was isolated from their tails and genotyped with simple sequence length polymorphic markers (Mouse MapPairs; Research Genetics. Huntsville, AL; Réf. 11). Marker positions are based on Version 3.2 of the mouse genome database (see Appendix). Only limited sets of markers were used to analyze the (O20 X OcB-6)F2 and (O20 X OcB6)F'5 wkcrosses because the numbers of degrees of freedom available, defined by the sizes of these crosses, were not sufficient to allow the statistical analysis of all markers necessary to cover the segregating genomic segments and their interac tions. We choose to use restricted sets of data that included markers in those genomic regions that overlapped with segregating segments in any of the other 4794 MAPPING OF SLUCf TO SLUC14 Table 1 The numbers of F-, mice used in the various crosses anil the likelihood ratio tesi scores corresponding to the 5% significance thresholds for main effect ami for interaction effect of each particular linkage analysis 5% significance threshold In'" for: O20 and: c B -3OcB-3OcB-4OcB-4OcB-6OcB-6OcB-6. 35 wk'OcB-6. 35 wk'OcB-9OcB-9OcB-16OcB-16number or sizeNumberSizeNumberSizeNumberSizeNumberSizeNumberSizerfNumberSizemice"14113017915791816868220193188169Main interaction effect11.013.613.413.315.923.515.314.312.612. 5.6 9.1Effect1 .514.616.013.719.527.920.318.912.413.020.627.8 " Because mice without tumors cannot be included in the linkage analyses of lung tumor size, more mice are involved in the analyses of lung tumor number than of lung tumor size. ' In. likelihood ratio test scores. ' The cross (O20 X OcB-6)Fv15 wk (see "Materials and Methods"). d See Ref. 11. crosses (13. 14). as well as the Kras2 marker (7). which is known to be linked to lung tumor susceptibility (15-18). Statistical Analysis. Normal distributions were obtained by square root transformation of lung tumor numbers and '"'log transformation of lung tumor sizes. The transformed mean tumor sizes per mouse were weighted propor tionally to the number of tumors per mouse. Outlying values were excluded. MQM mapping was performed as described previously (11). Tumor number and tumor size are regressed on the markers in the segregating regions, on sex of the mice, and on interactions between pairs (marker-marker and markersex). This multilocus model contains parameters for additive and dominance effects per marker and for additive-by-additive effects per pair. Parameters for additive-by-dominance, dominance-by-additive, and dominance-by-dominance effects per pair were not be included so that the available degrees of freedom would not be exceeded. For each cross and for each phenotype (tumor number and tumor size), the likelihood ratio test scores corresponding to the 5% significance thresholds for main effect and interaction effect on a crossspecific basis were determined by permutation tests. In each of at least 1000 runs, the original phenotypic data were randomly coupled to the original set of marker data. Those crosses appearing to reveal significant linkage data were permutated 2000 times. Results and Discussion Most of the genomic segments that segregate in each of the F2 crosses have been identified (13, 14). The list of microsatellite mark ers used to type these segregating areas can be obtained from the authors. Because the various crosses involve different numbers of mice and markers, the likelihood ratio test values that represent the 5% significance threshold for each particular linkage analysis (Table 1) were established by permutation tests. The corrected significant (P < 0.05) linkage data for lung tumor number and lung tumor size are listed in Table 2, together with near-significant linkage data (P < 0.10) that confirm these loci. They include the 10 new lung cancer susceptibility loci Sluc5-Slucl4. the Kras2 gene/region (7, 15-18), and sex of the mice. The newly mapped Slue loci are located in regions of the mouse genome in which no lung tumor susceptibility genes have been mapped before, except for Slue? and Slucl 1. which are located in the same regions as Par4 and Pas4, respectively (17, 19). Nearly all Slue loci were detected as participants in pairwise interactions, in which the effect of one locus depends on the genotype of its interacting partner (Table 3. C-H). Significant main effects were only observed for Krax2 (Table 3A) and SlucI3 (Table 3B). The estimated effects of Slue loci as presented in Table 3 are calculated from the MQM model. When interpreting these data, one should take into account that the MQM model we used fitted interactions with additive-by-additive effects only (see "Ma terials and Methods"). Several of these interactions are completely counteracting, e.g., the interaction between Sluco and Sluc8: indi cations of main effects for Sluco and SlucS were not obtained (Table 3C). Other interactions are partially counteracting, e.g., the one between SlucS and Slucl2. Although no main effect could be observed for the Slucl2 locus, the SlucS locus does exhibit a Table 2 Cross-specific corrected significant (P < 0.05) and near-significant (P < O.IOl linkage data of the Slue loci and their interactions, affecting Inni; nunor muniwr or lung nantir rà ' Genomic positionMarker Chromosome no.SexSexSexSexDIMitISDIMÕ136DIMit36D2MÕI56'D2MU56D4MÕH58D6MÕU58D6MU218'Kras2D7Nds4D8Mit35D9Mit2D9MÕ12D9MU2D9MÕU2D9MÕ112DI IMil 15'D12Nds2D14MÃŒI120DI8MÃŒ117D19MU9'11122466678999991112141819Distance from acromene (cM)88929238386766174725917171755554060132047Interacting with 9 Œ112 18MÃŒII7D9MU2D12Nds2D9MÃŒ12D9MÃŒI12D19MÃŒ19D9MÃŒ12D7Nds4D8MU35DIlMitISD4MU158D6MÃŒ1I58D2MU56DIMU36SexSexD1MU36D6MÃŒ1218D1MÃŒII5SexD2MU56F-, cross between O20 and:OcBOc B -4OcB-3OcB-4OcB-6. 35 wk*OcB-4OcB-4OcB-9OcB-4OcB-9OcB-4OcB-9OcB-6OcB-9OcB-4OcB-4OcB-4OcB-4OcB-4OcB-4OcB-9OcB-6, 35 wk*OcB-6OcB-4OcB-9Influencing tumor size or numberSizeSizeSizeSizeSizeSizeSizeSizeSizeNumberSizeSizeNumberNumberSizeSizeSizeSizeSizeSizeSizeSizeNumberSizeSizeIn"25.83 3.891 .5411.2833.7812.3612.2115.3014.8015.2315.6313.4821.1815.2315.6314.8012.3611.2825.8312.2113.4833.7817.5713.8915.30Corrected P0.0040.0470.0980.0990.0120.0720.0740.0240.0390.0290.0320.0410.0120.0290.0320.0 90.0720.0990.0040.0740.0410.01 0.0300.0470.024S/mlocusSexSlncSSluc2SlucnSlue?SlucJKrau.SlufHStoiciSilicioSlue IISluc4Sliicl2SluclJSlm-14Slucl ' In. likelihood ratio test score. '' The cross (O20 X OcB-6)F,35 See Ref. 11. "k (see "Materials and Methods").