Experiment
To explore the consequences of germline variation in an aggressive form of prostate cancer and to identify susceptibility genes, an F2 cross of C57BL/6-Tg(TRAMP)824Ng/J (TRAMP) and NOD/ShiLtJ (NOD), which is highly susceptible to aggressive disease development, was used. The TRAMP mouse model is a model of aggressive neuroendocrine prostate cancer which develops extensive tumorigenesis and metastasis by 30 weeks of age. QTL mapping was performed using 228 transgene positive (C57BL/6-Tg(TRAMP)824Ng/J x NOD/ShiLtJ)F2 intercrossed male mice.
QTL were mapped in the F2 males using 666 informative SNPs. Analyses were performed in J/qtl using a single locus model of inheritance. QTL were considered statistically significant when genome wide p<0.05.
Four metastasis-related traits were identified on Chrs 1, 11, and 13.
QTL Dmfsq1 (distant metastasis free survival QTL 1) mapped to Chromosome 1, LOD=3.93 at 35.0 cM, p=0.042, within a 2 LOD confidence interval between 40,760,231-95,290,730 bp.
QTL Dmfsq2 (distant metastasis free survival QTL 2) mapped to Chromosome 11, LOD=3.97 at 30.9 cM, p=0.039, within a 2 LOD confidence interval between 41.325.431-69,191,538.
QTL Lnmbq1 (lymph node metastasis burden QTL 1) mapped to Chromosome 13, LOD=4.69 at 22.1 cM, p=0.011, within a 2 LOD confidence interval between 4,829,663-46,774,063.
QTL Lsmcq1 (liver surface metastasis count QTL 1) mapped to Chromosome 11, LOD=4.01 at 8.6 cM, p=0.037, within a 2 LOD confidence interval between 11,062,569-35,356,130 bp.
Five tumor related traits were identified on Chrs 2, 4, 8, 13 and 17.
QTL Ptbq1 (primary tumor burden QTL 1) mapped to Chromosome 13, LOD=4.86 at 18.7 cM, p=0.007, within a 2 LOD confidence interval between 4,758,113-60,501,553.
QTL Svtbq1 (seminal vesicle tumor burden QTL 1) mapped to Chromosome 2, LOD=5.01 at 84.4 cM, p=0.005, within a 2 LOD confidence interval between 146,404,042-165,979,416 bp.
QTL Svtbq2 (seminal vesicle tumor burden QTL 2) mapped to Chromosome 4, LOD=5.24 at 7.6 cM, p=0.003, within a 2 LOD confidence interval between 5,191,558-53,264,210 bp.
QTL Svtbq3 (seminal vesicle tumor burden QTL 3) mapped to Chromosome 8, LOD=4.22 at 52.8 cM, p=0.022, within a 2 LOD interval between 83,633,294-111,798,566 bp.
QTL Svtbq4 (seminal vesicle tumor burden QTL 4) mapped to Chromosome 17, LOD=5.20 at 11.1 cM, p=0.004, within a 2 LOD interval between 3,499,649-36,093,828 bp.
Two QTL were evident for age of death on Chrs 7 and 8:
QTL Agedq1 (age of death QTL 1) mapped to Chromosome 7, LOD=4.35 at 76.4 cM, p<0.001, within a 2 LOD interval between 122,268,816-144,131,415 bp.
QTL Agedq2 (age of death QTL 2) mapped to Chromosome 8, LOD=4.65 at 50.8 cM, p<0.001, within a 2 LOD interval between 87,425,863-111,798,566 bp.
To identify candidate genes microarray analysis was performed to analyze patterns of global gene expression in all available F2 prostate tumors (n=126). Expression QTL mapping was performed using Matrix eQTL. The Benjamini-Hochberg false discovery rates (FDR) were calculated, with an FDR <0.05 used as the threshold for significant eQTL.
Expression levels of all transcripts within the boundaries of the 11 aggressive disease QTL were correlated with the QTL trait to increase the stringency for candidate gene identification. High priority candidates were those processing both high levels of expression-trait correlation and a proximal expression QTL. Thirty five aggressive prostate tumorigenesis candidate genes were identified each of which exhibited a statistically significant proximal eQTL and correlation between transcript expression and the trait of interest. See Table 3.
For QTL Agedq1 an eQTL at rs13479522 (128,129,547 bp), expressed transcript NM_021334, FDR=0.045, was correlated with increased expression with the B6 allele of candidate gene Itgax. The human homolog is ITGAX.
For QTL Agedq2 an eQTL at rs13479871 (84,946,610 bp), expressed transcript ENSMUST00000109736, FDR=0.046, was correlated with increased expression with the NOD/ShiLtJ allele of candidate gene Rnaseh2a. The human homolog is RNASEH2A.
For QTL Dmfsq2 an eQTL at rs3711357 (61,505,144), expressed transcript ENSMUST00000102657, FDR=0.021, was correlated with increased expression with the NOD/ShiLtJ allele of candidate gene B9d1. The human homolog is B9d1.
Several eQTL and candidate gene were detected for Ptbq1. An eQTL at 56,288,643 bp, expressed transcript NM_019568, FDR=0.001 correlated with increased expression with the NOD/ShiLtJ allele of candidate gene Cxcl14. The human homolog is CXCL14.
Several eQTL and candidate genes were detected for Svtbq1. An eQTL at rs6247960 (153,345,845 bp), expressed transcript ENSMUSG000109790, FDR=0.000, was correlated with increase expression with the B6 allele of candidate gene Asxl1. Another eQTL at rs6247961 (153,345,845 bp), FDR=0.000, correlated with increased expression with the NOD/ShiLtJ allele of gene Asxl1. The human homolog is ASXL1.
Several eQTL and candidate genes were also detected for Svtbq2. An eQTL at rs3698283 (42,629,332 bp), expressed transcript NM_011888, FDR=6.85E-05, was correlated with increased expression with the NOD/ShiLtJ allele of candidate gene Cdl19. The human homolog is CCL19.
For QTL Lsmcq1 an eQTL at rs3023251, expressed transcript NR_035454, FDR=0.023, was correlated with an increase in expression with the B6 allele of genome feature Mir1933.
Two concurrent approaches were then used to investigate the roll of the 35 candidate genes in aggressive forms of human prostate cancer. Logistic regression analysis in two human prostate gene expression datasets revealed expression levels of 5 genes (CXCL14, ITGAX, LPCAT2, RNASEH2A, ZNF322) positively correlated with aggressive postate cancer and two (CCL19 and HIST1H1A) were protective for aggressive prostate cancer.
A strong level of interdependency between tumor related traits and traits commonly associated with survival in human prostate cancer (age at death, distant metastasis free survival, presence or absence of lymph node metastasis, lymph node metastasis burden) was clear in the F2 mapping population. The data also clearly demonstrated that mice with greater seminal vesicle tumor burden, and thus a lower primary tumor burden, were less prone to more aggressive disease. The germline polymorphisms driving lower seminal vesicle
tumor burden and higher primary tumor burden may be associated with a predisposition to more aggressive disease.
Analysis Tools