Major Susceptibility Locus for Prostate Cancer on Chromosome 1 Suggested by a Genome-Wide Search

Despite its high prevalence, very little is known regarding genetic predisposition to prostate cancer. A genome-wide scan performed in 66 high-risk prostate cancer families has provided evidence of linkage to the long arm of chromosome 1 (1q24-25). Analysis of an additional set of 25 North American and Swedish families with markers in this region resulted in significant evidence of linkage in the combined set of 91 families. The data provide strong evidence of a major prostate cancer susceptibility locus on chromosome 1.

coupling constants from the 31 decoupled DQF-COSY and the 3D HMQC-TOCSY experiment with a short (20 ms) mixing time. For a gauche+ conformation, y was constrained to 55 + 30°(or +40°). The ribose sugar pucker was estimated from analysis of 3JH l'-H2' in the 31 P decoupled DQF-COSY spectrum.
Ribose conformation was restrained to C2.-endo (8 = 160 + 30°) or C3.-endo (6 = 85 + 300) when 3JH1 H2' was >8 Hz or <3 Hz, respectively. No restraints were used for riboses with mixed sugar conformations. Several paromomycin ring and 11 dihedral angles were restrained from analysis of the short mixing time TOCSY and DQF-COSY spectrum. 15. Structures were calculated using a simulated annealing protocol within the Insightil NMRArchitect package (Biosym Technologies, San Diego, CA). A randomized array of atoms corresponding to RNA and paromomycin was heated to 1000 K, and bonding, distance and dihedral restraints, and a repulsive quartic potential were gradually increased to full value over 40 ps of molecular dynamics. The molecules were then cooled during 10 ps to 300 K and subjected to a final energy minimization step that included an attractive Lennard-Jones potential. No electrostatic term was included in the target function. Using this protocol, 30% of the structures converged, as based on restraint violation energies, and 30 of them were collected to be further refined with the final set of restraints. There were differences of greater than 100 kcal * mol between converged and unconverged structures. During refinement, molecules were heated to 1000 K and subject to 30 ps of molecular dynamics following the same protocol as above. The molecules were then cooled during 10 ps to 300 K and subjected to a final energy minimization step that again included an attractive Lennard-Jones potential and no electrostatic term. Prostate cancer is the most common malignancy diagnosed in U.S. males, accounting for more than 40,000 deaths in this country annually (1). African Americans have the highest incidence and mortality rates of any population studied (2). Numerous studies have provided evidence for familial clustering of prostate cancer, indicating that family history is a major risk factor for this disease (3)(4)(5). Segregation analysis of familial prostate cancer suggests the existence of at least one dominant susceptibility locus and predicts that rare high-risk alleles at such loci account in the aggregate for 9% of all prostate cancers and more than 40% of early onset disease (6). Analyses of genetic alterations in pros-tate cancer have demonstrated frequent duplication of DNA sequences on the distal long arm of chromosome 8 (7), as well as loss of DNA sequences resulting in loss of heterozygosity (LOH) for the short arm of chromosome 8 and the long arm of chromosome 13 (8,9). Preliminary investigations by linkage analysis of these regions as well as sites of known tumor suppressor genes have not identified a susceptibility locus in prostate cancer (10,11  Finally, it is difficult to find extended pedigrees that are highly informative for linkage (in other words, that contain large numbers of affected family members) (12). In spite of these difficulties, we have undertaken a linkage analysis to search for evidence of loci contributing to risk for prostate cancer in a group of 79 North American and 12 Swedish pedigrees, each having at least three first-degree relatives affected with prostate cancer. These families were selected on the basis of the number of affected males from which samples could be obtained for typing, either as blood samples or archival specimens and the absence of evidence of bilineal inheritance (13). A summary of the characteristics of the families studied is given in Table 1. Overall, affected individuals in these families had an average age of diagnosis of 65, with a total of 34 males diagnosed before the age of 55.
To search for the location of high-risk alleles for prostate cancer, a genome-wide scan was performed in a subgroup of 66 North American families. A total of 341 dinucleotide repeat markers were analyzed in these pedigrees to complete a map with a marker density of 10 cM (14), requiring more than 130,000 genotypes. On average, 79% of our study group were heterozygous for each marker. For the parametric analysis of the genotype data, we used a model of dominant inheritance that includes a fixed phenocopy rate of 15% and the assumption that unaffected men over the age of 75 are not likely to be gene carriers (15). A plot of two-point lod (logarithm of the likelihood ratio for linkage) scores (16) for the genomewide scan (Z) is shown in Fig. 1. The highest lod score observed was 2.75 with marker DIS218, which maps to the distal long arm of chromosome 1 (lq24-25). As chromosome 1 showed the most significant evidence for linkage, additional markers in this region were typed in the original 66 families as well as in an additional group of 25 families, 12 of which were collected in Sweden (13). These analyses provided additional evidence for linkage in the lq24-25 region with a maximum two-point lod of 3.65 at recombination fraction 0 = 0.18 with marker DlS2883 ( Table 2).
As parametric analyses are model-dependent, we also used nonparametric analyses to further examine linkage data in this region (16). Nonparametric multipoint linkage (NPL) Z scores are given for this analysis in Table 2. Highly significant Pvalues were obtained for multiple markers, providing further evidence for linkage in this region. To determine the most likely location for the susceptibility locus, parametric multipoint analyses were performed with various combinations of markers in this region. Lod scores >4 were obtained, but did not allow unequivocal placement of the susceptibility locus due to apparent genetic heterogeneity. Significant evidence for locus heterogeneity (X2 = 8.11, P = 0.004) (16) was obtained by an admixture test with an estimate of 34% of the families being linked to the region. The maximum multipoint lod score with markers DlS2883, DJS 158, and DlS422 under the assumption of heterogeneity was 5.43, with the postulated susceptibility locus mapping close to D1S422 (Fig. 2). No clinical features appeared to distinguish families showing linkage to chromosome 1 from the nonlinked pedigrees.
The risk of prostate cancer in siblings of affected individuals is modified by the age of diagnosis (6). Subgrouping families by age of diagnosis, either by mean age within a family or by number of men diagnosed under age 55, provided little evidence that the families showing linkage to chromosome 1 had an earlier onset of prostate cancer than the unlinked families. However, because of difficulties in equating age of diagnosis with age of onset (17), further analysis will be necessary to support this conclusion.
Both African-American families analyzed in this study showed linkage to this region, yielding a combined lod score of 1.4. As there is evidence of linkage in Caucasian families in Sweden and North America as well, alterations in the 1q24-25 region may increase prostate cancer susceptibility in a variety of populations and ethnic backgrounds.
LOH studies have not previously implicated the chromosomal region lq24-25 in    (6), one can estimate (very roughly) that one in 500 may have an alteration in HPC 1. Because early diagnosis can be lifesaving in prostate cancer, the potential ability to identify individuals at genetically Table 2. Linkage results for susceptibility to prostate cancer and nine markers on chromosome 1 in 91 families. Z and 0 represent the maximum lod scores and recombination fractions, respectively. NPL Z scores are not directly comparable to parametric Z (LOD) scores. Therefore, significance levels are given for the NPL Z scores. For parameter (LOD) scores, a Z score of 3.0 corresponds to a signifiance level of a 0.0001.  (1995)] for multipoint linkage analysis. Multipoint analysis has the advantage of utilizing data from multiple linked markers to maximize the information in a given pedigree. Nonparametric multipoint analysis, which is robust even when the mode of inheritance is not known, was also performed, with GENEHUNTER [L. Kruglayk and E. S. Lander, Am. J. Hum. Genet. 57, 439 (1995)] to calculate normalized Z scores and associated P values. In all of the linkage analyses, allele frequencies for the markers were estimated from independent individuals in the families and unrelated individuals separately for the North American and Swedish families. CRIMAP [E. S. Lander and P. Green, Proc. Natl. Acad. Sci. U.S.A. 84, 2363 (1987)] was used to order the multiple markers on chromosome 1 using the genotype data from all pedigrees. The BUILD option of CRIMAP was first used to establish the order of markers with at least a likelihood ratio of 1000:1. The FLIP option was then used to calculate the likelihood of alternative marker orders by permuting adjacent loci (five flanking markers). The most likely order thus determined is the same as the published order (http: //cedar.soton.ac. uk/pub). The admixture test as implemented in HOMOG [J. Ott, Analysis of Human Genetic Linkage (Johns Hopkins Univ. Press, Baltimore, 1985), pp. 200-203] was used to test for genetic heterogeneity in the context of the two-point parametric analysis. 17. The evaluation of age as a variable is confounded because of the changing methods used to diagnose this disease, and increased interest in screening for this disease. For the years prior to the use of prostate-specific antigen (PSA), diagnosis of prostate cancer was often not made until men presented with advanced disease, whereas today most men are diagnosed younger and at an earlier stage. The RAC proteins have been implicated in the regulation of various fundamental cellular processes including actin cytoskeletal organization (I ), transcriptional activation (2), and cell proliferation (3)(4)(5). To identify the effector pathways that mediate the biological activities induced by RAC, we isolated mutant RAC proteins that could discriminate among the RAC targets PAK and PORI in the yeast two-hybrid system. PAK proteins are a family of highly conserved serine-threonine kinases that are activated by direct interaction with RACI (6). A role for PAK has been suggested in mediating RAC-induced activation of JNK and p38 mitogen-activated protein (MAP) kinase cascades (7). PORI interacts with RACI and appears to function in RACinduced membrane ruffling (8).
Libraries of vectors expressing mutant human RAC proteins fused to the LexA DNA binding domain (LBD) were created by polymerase chain reaction (PCR) mutagenesis (9) and screened for interaction with PAK3 and PORI. Two mutants con- T tTo whom correspondence should be addressed. tamning a single amino acid substitution in the RAC effector loop were identified. One mutant, RACVlZH4O, failed to bind PAK3 but did bind PORI, and another mutant, RACV12L37, bound PAK3 but not PORI (Table 1). Identical binding profiles were obtained for the interaction of these mutants with PAKI (10).
To investigate the biological activities of the RAC mutants, we first examined their abilities to stimulate PAK and activate the JNK pathway. COS-1 cells were cotransfected with either RACV 2, RACV 12H40 or RACVL2L37 expression plasmids and a plasmid encoding a Myc-tagged version of PAKI. PAKI activity was assayed in immunoprecipitates with myelin basic protein (MBP) as the substrate (11). Expression of RACV 2L37 resulted in stimulation of PAK activity, whereas expression of RACVI2H4C did not (Fig. 1, top). Thus, the activation of PAK by the RAC mutants is dependent on their ability to interact with PAK. To test for the ability of the RAC mutants to induce JNK activation, we cotransfected COS-1 cells with expression plasmids encoding RAC mutants and a plasmid encoding a FLAG-tagged version of JNKI. JNK activity was assayed with glutathione-S-transferase (GST) fused to c-Jun as the substrate (12). RACvl2H4o, which did not bind to or activate PAK, also did not stimulate JNK activity (Fig. 1, bottom). The RACV12L37 mutant, SCIENCE * VOL. 274 * 22 NOVEMBER 1996