Five Loci, SLT1 to SLT5, Controlling the Susceptibility to Spontaneously Occurring Lung Cancer in Mice

Research Article

Daolong Wang and Ming You

Department of Surgery and the Siteman Cancer Center, Washington University School of Medicine, St. Louis, Missouri

Abstract between markers D6Mit54 and D6Mit304, is a major locus affecting A series of linkage studies was previously conducted to identify the inheritable predisposition to chemically induced lung tumor quantitative trait loci associated with chemically induced lung development in mice. Other pulmonary adenoma susceptibility loci tumors. However, little is known of genetic susceptibility to (i.e., Pas2, Pas3, and Pas4) have been shown to modulate the effect spontaneously occurring lung tumorigenesis (SLT) in mice. In of Pas1. Pas2 was mapped on mouse chromosome 17 between this study, we did a whole-genome linkage disequilibrium anal- D17Mit23 and D17Mit50; Pas3 on chromosome 19 in the region ysis for susceptibility to SLT in mice using f135,900 single- flanked by D19Mit42 and D19Mit19; and Pas4 on chromosome 9 nucleotide polymorphisms (SNPs) from the Roche and Genomic from D9Mit11 to D9Mit282. We reported a locus on mouse Institute of the Novartis Research Foundation SNP databases. chromosome 10 (D10Mit126) that modified the effect of Pas1 and A common set of 13 mouse strains was used, including 10 re- increased tumor multiplicities. Tripodis et al. (4) reported 30 sus- sistant strains (129X1/SvJ, AKR/J, C3H/HeJ, C57BL/6J, DBA/2J, ceptibility lung cancer (Sluc) loci using recombinant congenic inbred strains of mice and the lung tumor development in response NZB/BlnJ, CAST/EiJ, SPRET/EiJ, SM/J, and LP/J) and 3 susceptible strains (A/J, BALB/cJ, and NZW/LaCJ). Fisher exact test to chemical carcinogens. Moreover, several quantitative trait loci of was used to assess the association between individual SNPs and pulmonary adenoma resistance (Par) to chemical induction were susceptibility to SLT. Five regions, SLT1 to SLT5, were mapped also reported. Par1 locus was mapped to mouse chromosome 11, on chromosomes 6, 7, 8, 19, and X, respectively. SLT1 to SLT5 near the Rara locus, with a log of odds score of 5.3. Par2, located on showed a significant association with SLT under the empirical distal chromosome 18, accounts for 38% of the phenotypic variance threshold (P V 0.004) derived from permutation tests. SNP in the backcross population, and plays a major role in protection versus SNP association tests indicated that these SLT regions against lung tumor development. Resistance locus Par3, located on were unlikely to be caused by population substructure. Thus, chromosome 12, was considered to act synergistically with Par1 (5). SLT1 to SLT5 seem to be novel loci controlling the susceptibility Linkage disequilibrium is defined as the nonrandom association to spontaneously occurring lung cancer in mice. Our results of alleles at linked loci in the population. It is the basis of fine provide, for the first time, an insight into the genetic control of mapping of complex disease loci through whole-genome association spontaneously occurring lung tumorigenesis. (Cancer Res 2005; studies. In comparison with linkage mapping, linkage disequilibri- 65(18): 8158-65) um mapping can significantly increase the precision of locating a gene of interest (6). In humans, linkage disequilibrium has been successfully used to find the precise location of disease loci (7). Introduction Single-nucleotide polymorphisms (SNPs) are changes in a single Lung cancer is the leading cause of cancer mortality in both males base at a specific position in the genome. Over the past few years, and females in developed countries (1). Lung cancer development is a SNPs have been widely used for the identification of complex multistage process involving successive addition of genetic aberra- disease genes and for pharmacogenetic applications. Depending on tions (1). Accumulating evidence shows that genetic factors are in- where an SNP occurs, it might be responsible, in whole or in part, for volved in familial lung tumor development in humans. As a result, the phenotypic difference. With the increasing number of available identification and characterization of genes responsible for suscep- mouse SNPs, linkage disequilibrium mapping can be used to localize tibility or resistance to lung tumorigenesis will be of great interest in genomic regions responsible for mouse lung tumor susceptibility. cancer detection, prevention, and therapy. Given the phylogenetic In this study, we conducted a whole-genome linkage disequilib- proximity, common genetic changes were frequently found in both rium analysis for spontaneous lung tumorigenesis (SLT) suscepti- human and mouse lung tumors (e.g., p53, Kras2, and p16). Thus, the bility with two recently released SNP databases (containing a total mouse is a valuable model for studying tumor susceptibility, stages of f135,900 SNPs): the Roche mouse SNP database and the SNP of tumor development, and the interaction of genetic and environ- database of Genomics Institute of the Novartis Research Founda- mental factors that result in predisposition to neoplasia. tion. Our goal was to identify novel loci specifically predisposed to Linkage mapping of quantitative trait loci has revealed several spontaneously occurring SLT because little is known for the genetic regions on mouse chromosomes related to susceptibility and re- susceptibility to SLT in mice. sistance to chemically induced lung cancer (2, 3). Pulmonary adenoma susceptibility locus 1 (Pas1), located on chromosome 6 Materials and Methods Single-nucleotide polymorphism data and susceptibility to spontaneously occurring lung tumorigenesis. The SNP data were downloaded Requests for reprints: Ming You, Department of Surgery and The Alvin J. Siteman from the Roche database1 and the GNF database.2 As of the beginning of Cancer Center, Campus Box 8109, Washington University School of Medicine, 660 South Euclid Avenue, St. Louis, MO 63110. Phone: 314-362-8315; Fax: 314-362-8323; E-mail: [email protected]. I2005 American Association for Cancer Research. 1 http://mousesnp.roche.com doi:10.1158/0008-5472.CAN-05-1508 2 http://snp.gnf.org/GNF10K/files.html

Cancer Res 2005; 65: (18). September 15, 2005 8158 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Mapping of Novel Loci Affecting Spontaneous Lung Cancer this study, the Roche SNP database contained f125,000 SNPs on 18 inbred Table 1. Mouse strains and susceptibility mouse strains, which were from the regions of 2,175 genes across the ge- f nome. The GNF database contains 10,900 evenly distributed SNPs, which c were typed for 48 inbred mouse strains. The SNPs from the two data sets Strain Susceptibility* Incidences (%) Reference together cover the mouse genome at an average density of 18 kb per SNP. In each of the two data sets, some strains, such as A/J and A/HeJ, BALB/ 129X1/SvJ R 1 (9) cJ and BALB/cByJ, C57BL/10J, C57BL6J, C57BLKS/J, C57BR/cdJ, and C57L/J, AKR/J R 5 (8) C3H/HeJ R 5-10 (8) share very recent ancestors and have more than 99% of all the SNPs b identical among them. To avoid apparent genetic overrepresentation in the C57BL/6J R 1 Jackson MTB sample that can cause spurious linkage disequilibrium, we retained only CAST/EiJ R 0 Y. Wang and M. You, x one strain from each of the substrain groups. In addition, we had to exclude unpublished data. strains with unknown SLT susceptibility, which led to a common set of 13 DBA/2J R 5-10 (8) strains for both the Roche data and the GNF data. These strains include a LP/J R 4-5 (9) group of 10 strains (129X1/SvJ, AKR/J, C3H/HeJ, C57BL/6J, DBA/2J, NZB/ NZB/BlnJ R 2 (9) BlnJ, CAST/EiJ, SPRET/EiJ, SM/J, and LP/J) resistant to SLT, and a group of SM/J R 0 Y. Wang and M. You, 3 susceptible strains (A/J, BALB/cJ, and NZW/LaCJ). The classification of unpublished data. these strains was based on published observations on spontaneously SPRET/EiJ R 0 Y. Wang and M. You, occurring lung tumor incidence frequency (8, 9), the Mouse Tumor Biology unpublished data. Database of the Jackson laboratory,3 and our unpublished observations A/J S 90 (8) (Table 1). BALB/cJ S 40-50 (8) The SNP data were subjected to further scrutiny. This included the NZW/LaCJ S 24 Jackson MTB removal of SNPs with less than eight strains typed and SNPs with only one phenotypic category (either susceptible or resistant) after excluding missing data, which resulted in a final set of 110,245 SNPs (99,882 from *R stands for resistant (incidences V 10%) and S for susceptible the Roche data and 10,363 from the GNF data), and thus a reduced (incidences > 10%). c coverage density to 22 kb per SNP. In addition, the genomic positions (in Incidences were observed at 18 months of age in general. b base pair) of the SNPs for the two data sets were unified to the latest Mouse tumor biology database offered by the Jackson Laboratory National Center for Biotechnology Information (NCBI) mouse genome (http://www.jax.org). map (Build 33.1).4 The method was to first find all SNPs with an accession xBased on our unpublished observations. in the NCBI dbSNP database, and then use the NCBI positions of these SNPs to infer the positions of those having no accession in the NCBI database by simple interpolation but keeping their original SNP order unchanged. gained considerable attention (13). To assess the influence of population Statistical analysis. Fisher exact test was applied to assessing substructure in our analysis, we conducted Fisher exact tests for SNP versus associations between each SNP in the combined data set and SLT SNP associations between significant SNPs detected under the empirical susceptibility. To do the test, a 2 2 contingency table was created for threshold and all their respective independent SNPs (on all different counts of the four possible combinations between each SNP and SLT chromosomes). The P values from all such tests were then used to get a susceptibility. A two-sided P value for each SNP was obtained for testing baseline as a genomic control (13) for evaluating the reliability of the hypothesis of no association between the SNP and SLT susceptibility. All linkage disequilibria identified. Fisher exact tests were conducted using the R statistical package (10). Despite the fact that P values from Fisher tests by themselves are Results and Discussion probabilities for no associations, the overall false-positive rate for genome- wide tests is difficult to obtain. This is because there are no simple relations Determination of thresholds by permutation tests. To between tests for linked (dependent) SNPs, and the simple Bonferroni conduct genome-wide linkage disequilibrium tests at a given correction seems too stringent for our analysis. Therefore, to control the overall false-positive rate, it is necessary to determine the genome-wide false-positive rate that is inflated by multiple association appropriate threshold. In this study, we did permutation tests tests, we conducted permutation tests and evaluated the significances of for all 286 possible permuted samples. The permutation results SNP-SLT associations with an empirical threshold determined by permu- were first evaluated for the distribution of total numbers of SNPs tation tests (11). Each permutation was done by rearranging the phenotype detected at various nominal significance levels ranging from a = labels (susceptible or resistant) among all the mouse strains but keeping a their SNP data unchanged. In doing so, the true (if any) associations 0.1 to = 0.001. The number of SNPs detected with each between SNPs and SLT susceptibility would be wiped out. All 286 possible permuted sample at each of the significance levels was compared permuted samples were created each with rearranged phenotype data and with the total number of SNPs detected with the original data at were analyzed in the same way as for the original data. An empirical null the same significance level. Figure 1A shows the distribution of distribution was then generated for genome-wide minimum P values each total numbers of SNPs detected with all the permuted samples from the genome-wide Fisher tests with one permuted sample. Based on the that were larger than or equal to the corresponding total number empirical distribution, a critical P value for a given genome-wide false- of SNPs detected with the original data. It is clear from Fig. 1A positive rate was determined for testing the significance of each association. that with decreased a value (increased significance level), the Two significance levels were considered: one false-positive per genome-wide total number of SNPs detected with the original data declined, scan (suggestive level) and 0.05 false positive per genome-wide scan and so did the false-positive rate. When a was reduced to 0.006, (significant level; ref. 12). Population substructure (overrepresentation of certain genetic back- the total number of SNPs detected with the original data was grounds) may lead to spurious linkage disequilibrium and has recently reduced to 167, whereas only 3 of the 286 permuted samples (including the original data set itself) yielded the same or larger number of SNPs, a chance of about 1%. This false-positive rate V 3 http://www.jax.org remained until a = 0.004. At a 0.003, however, no SNPs could be 4 http://www.ncbi.nlm.nih.gov detected with either the original data or any permuted samples. www.aacrjournals.org 8159 Cancer Res 2005; 65: (18). September 15, 2005

The low overall false-positive rate at high significance levels indicates that associations detected at these significance levels likely showed true associations between the relevant SNPs and SLT susceptibility. Using the permutation results, an empirical distribution was created for genome-wide minimum P values, each from the analysis of one permuted sample. Due to the nature of the relatively small sample size, the distribution displayed an apparently discrete trend: all the minimum P values fell into six groups of distinct P values from 0.014 to 0.0035 (Fig. 1B). There were 216 permuted samples falling in the largest group with P value = 0.0035, implying that we would not be able to obtain a significant threshold for our association tests, according to the relevant definition mentioned above (12). However, we were able to get a suggestive threshold (one false-positive per genome-wide scan). Taking into account the distribution of total numbers of SNPs detected at different nominal thresholds (Fig. 1A) and possible higher false-positive rate resulting from the small sample size, we selected the relatively stringent threshold P = 0.004 as our suggestive significance level. Based on the empirical distribution, at this threshold, there would be an expectation of about 0.87 false positive per genome-wide scan. Genome-wide linkage disequilibrium mapping of mouse SLT loci. Genome-wide association tests were conducted with the original combined data set. P values generated from Fisher tests for associations between SLT susceptibility and all the SNPs in the combined data set were plotted against the unified genomic positions of the SNPs in Fig. 2A. With the empirical suggestive threshold (P = 0.004) previously determined, we identified five regions (SLT1-SLT5) on chromosomes 6, 7, 8, 19, and X, respectively. Each region has at least one SNP significantly associated with SLT (Table 2). These regions were extended to cover closely linked SNPs that were associated with SLT susceptibility with P < 0.05. SLT3 on chromosome 8 is 1.7 Mb, containing most of the SNPs (120) significant at the empirical threshold; in SLT5 (10.3 Mb) on chromosome X, seven significant SNPs were detected, each occupying a 1 or 2 Mb consecutive genomic region. Each of the other three SLT regions, ranging from 1.8 to 3.1 Mb, contains one SNP significant at the empirical threshold. Validation by a genomic control. To examine the likelihood of the detected SLT regions being true linkage disequilibrium, rather than spurious linkage disequilibrium resulting from population substructure, we conducted SNP versus SNP association tests. For each of the 130 SNPs significantly associated with SLT in the five SLT regions, we used Fisher test for its associations with all its respective independent SNPs. The independent SNPs were all those located on different chromosomes from that of the SNP under test, but not in any of the five SLT regions. The same empirical threshold (P = 0.004) was used for testing the significance of the association between each pair of SNPs, and the total number of significant associations was recorded. Our results showed that a total of 357 tests gave a significant P value (P V 0.004), which involved only 62 independent SNPs, a proportion of about 5.7 104 among all independent SNPs tested (Fig. 3). Moreover, most of these independent SNPs had significant associations with Figure 1. Empirical distributions derived from permutation tests. A, distribution less than 10 of the 130 SNPs from the SLT regions, and none of of total numbers of SNPs detected from all permuted samples at 16 successive nominal a. The numbers above or below the spots are the numbers of them had significant associations with more than 40 SNPs from the markers detected at a given nominal a with the original data. The proportion of SLT regions. This is in sharp contrast with the results for SLT permuted samples was calculated as the number of permuted samples that susceptibility, which had significant associations with all of the 130 yielded equal or more markers than that obtained from the original data divided by 286 (number of all possible permuted samples). B, frequency distribution SNPs. This result showed that the associations of SLT susceptibility of discrete genome-wide minimum P values. with the SNPs in the five linkage disequilibrium regions were less

Cancer Res 2005; 65: (18). September 15, 2005 8160 www.aacrjournals.org

Figure 2. Genome-wide linkage disequilibrium analysis for spontaneous lung cancer susceptibility using the Roche and GNF SNP data sets. A, genome-wide linkage disequilibrium analysis. Each point in the chromosome plots represents an SNP, and was plotted as log10(p) against its position in megabases. Vertical dashed side, P = 0.0032. Arrows, regions that were significant at the suggestive level. B, haplotypes around SNPs having significant associations with spontaneous lung cancer in the five linkage disequilibrium regions. Each row represents a mouse strain; each column stands for an ordered SNP. Strains above the horizontal line are resistant to spontaneous lung cancer; those below the line are susceptible. Boxed SNPs were significantly or nearly significant associated with lung cancer at the empirical threshold P = 0.004). Dots, missing observations. www.aacrjournals.org 8161 Cancer Res 2005; 65: (18). September 15, 2005

Table 2. Linkage disequilibrium regions detected for spontaneous lung cancer susceptibility

c b Linkage disequilibrium region* Chromosome Start (Mb) End (Mb) P No. genes

SLT1 6 83.4 85.5 0.0035 34 SLT2 7 3.9 5.7 0.0061x 30 SLT3 8 122.7 124 0.0035 28 SLT4 19 31.7 34.8 0.0035 34 SLT5 X 133.1 143.4 0.0035 86

*Linkage disequilibrium region with one or more SNPs having significant associations with spontaneous lung cancer. cGenomic positions of the linkage disequilibrium region, based on NCBI Map View (Build 33.1). bMinimum P value in the linkage disequilibrium region. P value V 0.004 represents suggestive linkage disequilibrium. xThis region was nearly significant under the empirical threshold. However, at P = 0.0061, the genome-wide false-positive rate was 0.9, still within the limit of suggestive significance level.

likely to be caused by the population substructure of the mouse Comparative genomic analysis in the SLT regions. Based on strains. In other words, these regions are likely the true linkage the NCBI genome map (Build 33.1), there are 34, 30, 28, 34, and 86 disequilibrium with the genes affecting the SLT susceptibility. genes in SLT1, SLT2, SLT3, SLT4, and SLT5, respectively (Table 2). To Haplotypes in the SLT regions. Figure 2B gives the haplotypes identify potential candidate genes in each of the regions, we first for each of the mouse strains in the five SLT regions. The SLT conducted comparative genomic analyses to obtain the genes that regions include varied numbers of SNPs: 11 in SLT1, SLT2, and were the most conserved. Using the web-based tool developed by SLT4; 1,293 in SLT3; and 32 in SLT5. The SNPs shown in Fig. 2B for Li et al. (14), we conducted three types of comparisons to identify SLT3 and SLT5 were selected to evenly cover the regions due to the genes homologous between mouse and other species: (a) com- high densities of SNPs. It is clear that in each of the five regions, the parisons of the proteome of mouse (Mus musculus) with those of susceptible strains have haplotypes very distinct from the resistant all 29 species available in the web-based program, including 14 strains. Significant associations between the SNPs and SLT in the multicellular organisms (Homo sapiens, Rattus norvegicus, Fugu regions often represented complete linkage disequilibrium. How- rubripes, etc.) and 15 unicellular (or unicellular-multicellular) ever, most of the 130 significant SNPs in the five regions were organisms (Chlamydomonas reinhardtii, Cryptococcus neoformans, specified as either intronic SNPs or ‘‘locus region’’ based on the etc.); (b) comparisons with all the 14 multicellular organisms only; NCBI dbSNP database; none of them were specified as potential (c) comparisons with all the 15 unicellular organisms only. Because functional changes (missense, nonsense, untranslated region, etc.). unicellular organisms are more primitive than multicellular Thus, we attempted to narrow the list of potential candidates using organisms, mouse genes homologous with those of unicellular comparative genomic analyses. organisms should be more conserved than those homologous with multicellular-organism genes and the latter more conserved than the genes having no homology. Results showed that the comparisons with all the 29 organisms yielded the smallest number (835) of homologous genes, whereas comparisons with multicellular organisms resulted in the most genes (4,170). The number of genes identified by comparisons with unicellular organisms (927) was a little more than that from comparisons with all organisms. In fact, all the genes from comparisons with all organisms were only a subset of the genes from comparisons with unicellular organisms, and most of the genes homologous with unicellular organisms were also homologous with multicellular organisms. We then compared the genes in each of the SLT regions with the mapped homologous genes from each type of comparison (824 genes with all-organisms homology, 4,088 genes with multicellular homology, and 913 genes with unicellular homology). The results are summarized in Table 3 and Fig. 4. Although SLT1 and SLT2 do not have genes with unicellular homology, there were five genes in SLT1 and one gene in SLT2 that are homologous with genes of multicellular organisms. SLT3 has three genes with both unicellular and multicellular homology, and also has one additional gene homologous with genes of Figure 3. Association tests between the 130 SNPs in the five linkage disequilibrium regions and their respective independent SNPs. The x axis is for multicellular organisms. SLT4 and SLT5 both have only one gene independent SNPs arranged in an arbitrary order. Dots, total number of homologous with genes from both unicellular and multicellular significant associations that an independent SNP had with the set of the 130 SNPs. The horizontal dashed line was drawn at y = 130, which is the number of organisms, but five (in SLT4) and seven (in SLT5) additional genes significant associations detected for spontaneous lung cancer. with multicellular homology. If the conservation of a genomic

Cancer Res 2005; 65: (18). September 15, 2005 8162 www.aacrjournals.org

Table 3. Homologous genes in the SLT regions based on comparative genomic analyses

c Region Gene symbol* Homologous with Description

All Multicellular Unicellular

SLT1 LOC381788 Yes Similar to electrogenic sodium bicarbonate cotransporter NBC4c C330016K18Rik Yes RIKEN cDNA C330016K18 gene Mobk1b Yes MOB1, Mps One Binder kinase activator–like 1B (yeast) D630030L16Rik Yes RIKEN cDNA D630030L16 gene Sec15l2 Yes SEC15-like 2 (S. cerevisiae) SLT2 Zik1 Yes Zinc finger protein interacting with K protein 1 SLT3 Afg3l1 Yes Yes Yes AFG3(ATPase family gene 3)–like 1 (yeast) Rhou Yes Yes Yes ras homologue gene family, member U Abcb10 Yes ATP-binding cassette, subfamily B (MDR/TAP), member 10 Taf 5l Yes Yes Yes TAF5-like RNA polymerase II, p300/CBP-associated factor (PCAF)–associated factor SLT4 Papss2 Yes 3V-Phosphoadenosine 5V-phosphosulfate synthase 2 Atad1 Yes Yes Yes ATPase family, AAA domain containing 1 Pten Yes Phosphatase and tensin homologue Acta2 Yes Actin, a2, smooth muscle, aorta Pank1 Yes Pantothenate kinase 1 Mphosph1 Yes M-phase phosphoprotein 1 SLT5 Pak3 Yes Yes Yes p21 (CDKN1A)-activated kinase 3 Alas2 Yes Aminolevulinic acid synthase 2, erythroid BC020354 Yes cDNA sequence BC020354 Ureb1 Yes Upstream regulatory element binding protein 1e Hadh2 Yes Hydroxyacyl-CoA dehydrogenase type II Smc1l1 Yes SMC (structural maintenance of chromosomes 1)-like 1 (S. cerevisiae) LOC245666 Yes Similar to KIAA0522 protein Ubqln2 Yes Ubiquilin 2

*This is based on NCBI mouse genome map Build 33.1. cAll stands for all 29 organisms, Multicellular for 14 multicellular organisms only, and Unicellular for 15 unicellular organisms only.

region can be determined by gene homology, SLT3 would be the X-linked mental retardation and may indirectly influence mitotic most conserved region. events in breast cancer cells (21). Other genes of interest in the Except for several predicted genes (such as LOC381788 in SLT1 SLT regions that are homologous only with multicellular and LOC245666 in SLT5), most of the 24 genes with varied organisms include Zik1 in SLT2 and Ureb1 in SLT5. Zik1 encodes conservation in the five regions are known genes. The five genes a zinc finger protein interacting with K protein 1, which was with unicellular homology are Rhou, Afg3l1, and Taf 5l in SLT3, found to be a transcriptional repressor (22). Ureb1 encodes Atad1 in SLT4, and Pak3 in SLT5 (Table 3). Rhou (also known as upstream regulatory element binding protein 1e. It was found to G28K or WRCH1) encodes a protein with homology to the human be overexpressed in human colorectal cancer (23). Another study cell division cycle protein 42/G25K (15). It was found to be suggested that optimal suppression of tumor suppressor p53 differentially expressed in primary kidney, colon, gastric, breast, transactivation requires tyrosine-phosphorylated Ureb1 (24). ovarian, and uterus cancer. Afg3l1 (also known as AFG3) encodes Known cancer genes in the SLT regions. The human homo- a protein that is targeted to the mitochondria (16), and is possibly logous regions for the SLT regions were indicated in Fig. 4. Studies related to disorders sharing features with mitochondrial disease on loss of heterozygosity (LOH) have shown that some of the syndromes, such as sensorineural deafness, diabetes, and regions entailed frequent loss in human lung cancer, such as 1q retinopathy. However, there was no published evidence indicating (SLT3; ref. 25), 2p (SLT1; ref. 26), and 10q (SLT4; ref. 27). In mice, its role in cancer. Taf5l encodes a protein that is a component of regions on chromosome 6 (SLT1) also incurred LOH in lung cancer the PCAF histone acetylase complex and structurally similar to development (28). These results suggest that the related genomic one of the histone-like TAFs, TAF5. The PCAF histone acetylase regions are genetically more fragile than other regions in lung complex plays a role in the regulation of transcription, cell cycle cancer development, and possibly harbor lung cancer suppressor progression, and differentiation (17). However, it is not clear genes. According to the list of known cancer genes provided on the whether this gene played a role in tumorigenesis in either Sanger Institute web site,5 four genes in three of the SLT regions are humans or mice. Atad1 in SLT4 is a member of the ATPase family known cancer genes in human. These genes include Fanca in SLT3, with quite disparate cellular activities (18). Pak3 belongs to the Pten and Tnfrsf6 in SLT4, and Lhfp in SLT5. Fanca is part of a family coding for p21-activated kinases that are involved in the control of cytoskeleton dynamics and cell cycle progression (19, 20). In humans, this gene was shown to be responsible for 5 http://www.sanger.ac.uk www.aacrjournals.org 8163 Cancer Res 2005; 65: (18). September 15, 2005

Figure 4. Physical maps of the five SLT regions. For each SLT region, the thick brown solid arrow (SLT arrow) indicates the SLT region and orientation. On the right side of the SLT arrow are mouse genes and their orientations (open arrows). Orange arrows, genes homologous with multicellular organisms; brown arrows, genes homologous with unicellular organisms. On the left side of the SLT arrow are homologous regions (dashed arrows), if any, and their orientations on the human genome. multiprotein nuclear Fanconi anemia complex with identical that Tnfrsf6 was involved in up-regulation of Fas/Fas ligand– function in cellular responses to DNA damage and germ cell sur- mediated apoptosis induced by gossypol in human alveolar lung vival (29, 30). In addition to the functions in development of cancer cell line (40). The Fas/Fas ligand system was also found to Fanconi anemia, Fanca has also been reported to be possibly be a critical regulator of perinatal alveolar epithelial type II cell involved in pathways for breast cancer, familial pancreatic cancer, apoptosis in mouse lungs (41). Thus, it is reasonable to consider and sporadic acute myeloid leukemia. Thus, it is logical to hypo- Tnfrsf6 as a promising candidate gene for genetic susceptibility to thesize that this gene may also play some roles in SLT suscepti- spontaneous lung cancer in mice. Lhfp (lipoma HMGIC fusion bility in mice. Pten is a tumor suppressor, coregulating oncogenic partner) acts as a translocation partner of HMGIC in a lipoma cell signaling pathways downstream of receptor tyrosine kinases with a t(12;13). It was found to be related to benign and low- (31–33). Many recent studies have shown that decreased expres- malignant lipomatous tumors (42). However, there were no sion of Pten was related to tumorigenesis of non–small-cell lung reports showing its relevance in the development of lung and/ cancer in humans (34–36). Gautam et al. (37) found that the or lung cancers. expression of Pten is an intermediate step for suppression of lung cancer metastasis induced by overexpression of gene RRM1 in both human and mouse cell lines. This strongly suggests that Pten Acknowledgments may also play an important role in inheritable spontaneous lung Received 5/3/2005; revised 6/21/2005; accepted 6/30/2005. cancer development of mice. Tnfrsf6 (also known as FAS) encodes Grant support: NIH grants CA099187, CA099147, ES012063, and ES013340. The costs of publication of this article were defrayed in part by the payment of page the cell surface receptor involved in apoptotic signal transmission charges. This article must therefore be hereby marked advertisement in accordance in many cell types, including cells of the immune system (38). with 18 U.S.C. Section 1734 solely to indicate this fact. We thank Dr. Tim Wiltshire for sharing the mouse SNP data. We are grateful to Tnfrsf6 is silenced in many tumor types, resulting in an inability several individuals in Chemoprevention Program of the Siteman Cancer Center for to respond to proapoptotic signals (39). A recent study showed making corrections to this article.

Cancer Res 2005; 65: (18). September 15, 2005 8164 www.aacrjournals.org

References 17. Ogryzko VV, Kotani T, Zhang X, et al. Histone-like 30. Wong JC, Alon N, McKerlie C, et al. Targeted TAFs within the PCAF histone acetylase complex. Cell disruption of exons 1 to 6 of the Fanconi Anemia group 1. Greenlee RT, Hill-Harmon MB, Murray T, Thun M. 1998;94:35–44. A gene leads to growth retardation, strain-specific Cancer statistics, 2001. CA Cancer J Clin 2001;51:15–36. 18. Hoyle J, Fisher EM. Genomic organization and microphthalmia, meiotic defects and primordial germ 2. Dragani TA. 10 years of mouse cancer modifier loci: mapping of the mouse P26s4 ATPase gene: a member cell hypoplasia. Hum Mol Genet 2003;12:2063–76. human relevance. Cancer Res 2003;63:3011–8. of the remarkably conserved AAA gene family. Genomics 31. AbounaderR,ReznikT,ColantuoniC,etal. 3. Demant P. Cancer susceptibility in the mouse: 1996;31:115–8. Regulation of c-Met-dependent gene expression by genetics, biology and implications for human cancer. 19. Faure S, Vigneron S, Doree M, Morin N. A member of PTEN. Oncogene 2004;23:9173–82. Nat Rev Genet 2003;4:721–34. the Ste20/PAK family of protein kinases is involved in 32. Steck PA, Pershouse MA, Jasser SA, et al. Identifica- 4. Tripodis N, Hart AA, Fijneman RJ, Demant P. both arrest of Xenopus oocytes at G2/prophase of the tion of a candidate tumour suppressor gene, MMAC1, at Complexity of lung cancer modifiers: mapping of thirty first meiotic cell cycle and in prevention of apoptosis. chromosome 10q23.3 that is mutated in multiple genes and twenty-five interactions in half of the mouse EMBO J 1997;16:5550–61. advanced cancers. Nat Genet 1997;15:356–62. genome. J Natl Cancer Inst 2001;93:1484–91. 20. Rudel T, Bokoch GM. Membrane and morphological 33. Gu J, Tamura M, Pankov R, et al. Shc and FAK 5. Pataer A, Nishimura M, Kamoto T, et al. Genetic changes in apoptotic cells regulated by caspase- differentially regulate cell motility and directionality resistance to urethan-induced pulmonary adenomas in mediated activation of PAK2. Science 1997;276:1571–4. modulated by PTEN. J Cell Biol 1999;146:389–403. SMXA recombinant inbred mouse strains. Cancer Res 21. Li F, Adam L, Vadlamudi RK, et al. p21-activated 34. Jimenez AI, Fernandez P, Dominguez O, Dopazo A, 1997;57:2904–8. kinase 1 interacts with and phosphorylates histone H3 Sanchez-Cespedes M. Growth and molecular profile of 6. Ardlie KG, Kruglyak L, Seielstad M. Patterns of linkage in breast cancer cells. EMBO Rep 2002;3:767–73. lung cancer cells expressing ectopic LKB1: down- disequilibrium in the human genome. Nat Rev Genet 22. Denisenko ON, O’Neill B, Ostrowski J, Van Seuningen I, regulation of the phosphatidylinositol 3V-phosphate 2002;3:299–309. Bomsztyk K. Zik1, a transcriptional repressor that kinase/PTEN pathway. Cancer Res 2003;63:1382–8. 7. Reich DE, Cargill M, Bolk S, et al. Linkage disequilib- interacts with the heterogeneous nuclear ribonucleopro- 35. Bepler G, Sharma S, Cantor A, et al. RRM1 and PTEN rium in the human genome. Nature 2001;411:199–204. tein particle K protein. J Biol Chem 1996;271:27701–6. as prognostic parameters for overall and disease-free 8. Malkinson AM. The genetic basis of susceptibility to 23. Yoon SY, Lee Y, Kim JH, et al. Overexpression of human survival in patients with non-small-cell lung cancer. lung tumors in mice. Toxicology 1989;54:241–71. UREB1 in colorectal cancer: HECT domain of human J Clin Oncol 2004;22:1878–85. 9. Drinkwater NR, Bennett LM. Genetic control of carci- UREB1 inhibits the activity of tumor suppressor p53 36. Ferraro B, Bepler G, Sharma S, Cantor A, Haura EB. nogenesis in experimental animals. Prog Exp Tumor Res protein. Biochem Biophys Res Commun 2005;326:7–17. EGR1 predicts PTEN and survival in patients with non- 1991;33:1–20. 24. Gu J, Dubner R, Fornace AJ Jr, Iadarola MJ. UREB1, a small-cell lung cancer. J Clin Oncol 2005;23:1921–6. 10. Ihaka R, Gentleman R. R : a language for data analysis tyrosine phosphorylated nuclear protein, inhibits p53 37. Gautam A, Li ZR, Bepler G. RRM1-induced metasta- and graphics. J Comput Graph Statist 1996;5:299–314. transactivation. Oncogene 1995;11:2175–8. sis suppression through PTEN-regulated pathways. 11. Churchill GA, Doerge RW. Empirical threshold values 25. Tsuchiya E, Nakamura Y, Weng SY, et al. Allelotype of Oncogene 2003;22:2135–42. for quantitative trait mapping. Genetics 1994;138:963–71. non-small cell lung carcinoma–comparison between 38. Krammer PH, Behrmann I, Daniel P, Dhein J, Debatin 12. Lander E, Kruglyak L. Genetic dissection of complex loss of heterozygosity in squamous cell carcinoma and KM. Regulation of apoptosis in the immune system. traits: guidelines for interpreting and reporting linkage adenocarcinoma. Cancer Res 1992;52:2478–81. Curr Opin Immunol 1994;6:279–89. results. Nat Genet 1995;11:241–7. 26. Yoshioka H, Takeuchi T, Matsuno Y, et al. Analysis of 39. Sibley K, Rollinson S, Allan JM, et al. Functional FAS 13. Devlin B, Roeder K. Genomic control for association loss of heterozygosity in small adenocarcinomas of the promoter polymorphisms are associated with increased studies. Biometrics 1999;55:997–1004. lung. Jpn J Clin Oncol 1998;28:240–4. risk of acute myeloid leukemia. Cancer Res 2003;63: 14. Li JB, Gerdes JM, Haycraft CJ, et al. Comparative 27. Dacic S, Sasatomi E, Swalsky PA, et al. Loss of 4327–30. genomics identifies a flagellar and basal body proteome heterozygosity patterns of sclerosing hemangioma of 40. Chang JS, Hsu YL, Kuo PL, Chiang LC, Lin CC. Up- that includes the BBS5 human disease gene. Cell 2004; the lung and bronchioloalveolar carcinoma indicate a regulation of Fas/Fas ligand-mediated apoptosis by 117:541–52. similar molecular pathogenesis. Arch Pathol Lab Med gossypol in an immortalized human alveolar lung 15. Daigo Y, Takayama I, Ponder BA, et al. Novel human, 2004;128:880–4. cancer cell line. Clin Exp Pharmacol Physiol 2004;31: mouse and xenopus genes encoding a member of the 28. Devereux TR, Holliday W, Anna C, et al. Map kinase 716–22. RAS superfamily of low-molecular-weight GTP-binding activation correlates with K-ras mutation and loss of 41. De Paepe ME, Mao Q, Embree-Ku M, Rubin LP, Luks proteins and its down-regulation in W/WV mouse heterozygosity on chromosome 6 in alveolar bronchiolar FI. Fas/FasL-mediated apoptosis in perinatal murine jejunum. J Gastroenterol Hepatol 2004;19:211–7. carcinomas from B6C3F1 mice exposed to vanadium lungs. Am J Physiol Lung Cell Mol Physiol 2004;287: 16. Kremmidiotis G, Gardner AE, Settasatian C, et al. pentoxide for 2 years. Carcinogenesis 2002;23:1737–43. L730–42. Molecular and functional analyses of the human and 29. Noll M, Battaile KP, Bateman R, et al. Fanconi anemia 42. Dahlen A, Debiec-Rychter M, Pedeutour F, et al. mouse genes encoding AFG3L1, a mitochondrial metal- group A and C double-mutant mice: functional evidence Clustering of deletions on chromosome 13 in benign loprotease homologous to the human spastic paraplegia for a multi-protein Fanconi anemia complex. Exp and low-malignant lipomatous tumors. Int J Cancer protein. Genomics 2001;76:58–65. Hematol 2002;30:679–88. 2003;103:616–23.

www.aacrjournals.org 8165 Cancer Res 2005; 65: (18). September 15, 2005

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Five Loci, SLT1 to SLT5, Controlling the Susceptibility to Spontaneously Occurring Lung Cancer in Mice

Daolong Wang and Ming You

Cancer Res 2005;65:8158-8165.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/18/8158

Cited articles This article cites 42 articles, 12 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/18/8158.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/18/8158.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/65/18/8158. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.