Abstract
Linkage to a prostate cancer susceptibility locus was recently reported on chromosome 16q23. We now report a region exhibiting a high frequency of allelic imbalance (AI) corresponding to this locus in tumors from 51 men diagnosed with prostate cancer using the same linked markers. The highest frequency of AI was found at markers D16S3096(45%) and D16S516 (53%) that map to chromosome 16q23.2. In addition, 19 of the 51 (37%) prostate tumors showed interstitial AI involving one or both of these markers. This result strongly suggests that a candidate prostate cancer tumor suppressor gene maps between markers D16S3096 and D16S516. We estimate that the distance between these markers is approximately 118 kb using a Stanford radiation hybrid panel. We observed a positive association with family history (P = 0.048) when comparing those men showing interstitial AI at markers D16S3096 and/or D16S516 with those without any imbalance at these two markers. Taken together, these data suggest that we have precisely localized a region of chromosome 16q23.2 that may harbor a prostate cancer tumor suppressor gene implicated in the development of non-familial and possibly familial forms of prostate cancer.
INTRODUCTION
Prostate cancer is the second leading cause of cancer-related deaths among men in the United States (1) . Unfortunately, the molecular pathogenesis of this disease remains poorly understood. Epidemiological data suggest a strong familial component, and it has been estimated that 9% of all prostate cancers occurring by age 85 years are the result of a hereditary predisposition (2) . This frequency rises to 43% of early-onset (age 55 years or younger) prostate cancer cases (2) . A hereditary prostate cancer susceptibility locus termed HPC1 was mapped to chromosome 1q24-q25 using a genome-wide scan of families that each had three or more affected first-degree members (3) . Further analysis and expansion of these data suggested that linkage to HPC1 may be restricted to families with early-onset prostate cancer (4 , 5) . There is also some suggestion that linkage to this region may be positively affected by the inclusion of African-American prostate cancer families (6) . Whereas a modest familial association with this region has been supported by two independent studies (6 , 7) , three other studies have been unable to confirm linkage (8, 9, 10) . These data suggest that HPC1 may account for some, but not all, hereditary prostate cancer cases.
Another locus on chromosome 1q42.2-q43 has recently been reported to show linkage in French and German prostate cancer families (10) , but this has not been confirmed in independent studies (11 , 12) . Linkage to a third prostate cancer susceptibility locus on chromosome 1 at 1p36 that may also be associated with increased risk of brain cancer has recently been reported (13) . In addition, a locus on chromosome X has also been reported (14) . There have not yet been any independent reports to confirm the latter associations. This ambiguity in linkage may reflect considerable heterogeneity in hereditary prostate cancer, with individual loci accounting for only a small fraction of the hereditary prostate cancer population.
In a recent study of 504 brothers with prostate cancer from 230 multiplex sibships involving two authors of this study (J. S. W. and W. J. C.), a positive linkage to chromosomes 2q, 12p, 15q, 16p, and 16q was reported (15) . The strongest association was at chromosome 16q23. Frequent deletions of 16q have been reported in prostate cancer, and the literature suggests that this region may harbor at least three tumor suppressor loci (16, 17, 18, 19, 20, 21, 22, 23, 24, 25) . LOH 3 on 16q has also been reported in other cancers, including breast cancer (26) , Wilms’ tumors (27) , and hepatocellular carcinoma (28) . These data imply that one or more genes on 16q23 may be implicated in the development of prostate cancer and other malignancies. To date, no tumor suppressor genes have been identified from this region.
In this study, we sought to provide further evidence for a prostate cancer susceptibility gene on chromosome 16q23 and to further delineate this region using AI approaches. We now report a high frequency of AI in prostate tumors within the region on chromosome 16q23 corresponding to that identified in our sibling pair studies. We have further narrowed this region to approximately 118 kb by identifying tumors with interstitial loss. Our studies provide strong evidence that this region harbors a prostate cancer tumor suppressor gene that may be inactivated in a significant number of familial and nonfamilial forms of prostate cancers.
MATERIALS AND METHODS
Patient Selection and Tissue Evaluation.
A series of 55 prostate cancer patients who underwent radical prostatectomy between the years of 1991 and 1998 at the Cleveland Clinic Foundation was identified through the Cleveland Clinic Foundation’s Tumor Registry. Clinical characteristics, including Gleason grade, PSA, and tumor-node-metastasis (TNM) stage, and other potentially important factors, such as age at diagnosis, were obtained from medical records and are presented in Table 1 ⇓ . All tumors were graded according to the Gleason system (29) . The tumor stage was determined after review of the microscopic sections from the surgical specimen (30) . The mean age at the time of diagnosis was 61 years (age range, 47–73 years). The study design was approved by the Institutional Review Board of the Cleveland Clinic Foundation.
Selected clinical parameters for the 51 prostate cancer patients in the study
Tissue Microdissection and DNA Extraction.
For each patient, a paraffin-embedded tumor tissue block was chosen such that both normal and tumor tissue were present. Three 5-μm-thick unstained slides were prepared from each block. A consecutive slide was stained with H&E to assign normal and tumor areas. One pathologist (H. L.) assessed all of the cases. Areas of normal and cancer tissue were microdissected with the aid of the outlined H&E-stained slide, as described previously (31 , 32) . Four patients were removed from the study because the archived tissue did not yield DNA of sufficient quality for PCR amplification. The total number of patients that were included in our genetic analysis was 51.
After microdissection, DNA was extracted using the QiaAmp Tissue Kit (Qiagen, Valencia, CA). The final elution was performed in 100 μl of Tris buffer (pH 9).
Marker Information and Radiation Mapping.
Seven microsatellite markers were used in the study. Five of these markers (D16S3049, D16S3096, D16S516, D16S504, and D16S3040) showed significant linkage in our prostate cancer sibling pair study (15) . Sequence information for the seven microsatellite markers used in the present study was obtained from the Genome Database web site. 4 Oligonucleotide primers were synthesized by Genosys Biotechnologies (The Woodlands, TX). An optimal PCR annealing temperature was determined for each microsatellite marker.
The chromosomal order of the microsatellite markers used in the study was confirmed using the Stanford high-resolution TNG3 radiation hybrid panel (Research Genetics, Huntsville, AL). The panel, consisting of DNA from 90 human lymphoblastoid-derived human:hamster hybrids, was screened using each of the seven microsatellite markers. The PCR reactions were performed using a PCR thermal cycler (Ericomp, San Diego, CA). Each 15-μl reaction contained 2 μl of eluted DNA, 1.25 mm of each deoxynucleotide triphosphate, 0.5μ m of each primer, 0.75 unit of Taq DNA polymerase (Life Technologies, Inc., Rockville, MD), 67 mm Tris-HCl (pH 8.8), 6.7 mm magnesium chloride, 16.6 mm ammonium sulfate, 10 mm β-mercaptoethanol, 10% DMSO, and 1× Redi Load (Research Genetics). The step cycle file was comprised of a 5-min denaturation at 94°C and 35 cycles of 94°C for 45 s, annealing at the appropriate temperature for 1 min, and extension at 72°C for 1 min, followed by a final extension of 72°C for 7 min. Ten μl of each PCR reaction were run on a 6% nondenaturing polyacrylamide gel and visualized by ethidium bromide staining. The PCR result of the presence or absence of each marker was recorded for each hybrid, analyzed using the Map Manager QT program (33) , and used to determine the overall chromosomal order and the approximate distance between markers.
Allelic Imbalance Studies.
For the AI studies, separate PCR reactions were performed using DNA from microdissected normal and tumor tissue. A fluorophore was included at the 5′ end of each forward primer for detection on an ABI Prism 373 XL DNA Sequencer. The fluorophore choice, 6-carboxyfluorescein, tetrachloro-6-carboxyfluorescein, or 4,7,2′,4′,5′,7′-hexachloro-6-carboxyfluorescein, was made to allow for multiplexing of the PCR products for loading. PCR conditions were as described above, but without Redi Load. To determine the appropriate dilution for the genetic analysis, 2 μl of each PCR reaction were run on a 6% nondenaturing polyacrylamide gel and visualized by ethidium bromide staining. The PCR products were diluted in water, with PCR products from a maximum of four markers from one patient multiplexed for loading. One μl of each multiplexed solution was combined with formamide gel loading dye containing 1.5 mm EDTA and 3 mg/ml dextran blue and a 350 base pair 6-carboxytetramethylrhodamine size standard (Perkin-Elmer, Foster City, CA). The sample mixtures were denatured at 95°C for 5 min and immediately put on ice before loading onto a 6% denaturing polyacrylamide gel. Gels were run for a minimum of 7 h at 30 W with ABI Collection software (Perkin-Elmer). ABI GeneScan software version 3.1 (Perkin-Elmer) was used to process the runs. The software package Genotyper version 2.1 (Perkin-Elmer) was used to analyze the data for AI. Alleles were quantitated by peak heights (as recommended by Perkin-Elmer Corp.). The ratio of allele 1 to allele 2 for the tumor was divided by the ratio of allele 1 to allele 2 for the normal sample of each patient. A ratio less than 0.75 or greater than 1.33 was assigned as AI, as described previously (32) .
Statistical Analysis.
Basic descriptive statistics were used to quantify the evidence for the presence of a tumor suppressor gene at the markers of interest. In addition, contingency table analyses were used to evaluate whether AI at each marker correlated with clinical parameters (e.g., family history of prostate cancer). Test statistics were calculated using Fisher’s exact test because some cell counts were small. All analyses were undertaken with the statistical software SAS (SAS Institute Inc., Cary, NC).
RESULTS
Radiation Hybrid Mapping of 16q Microsatellite Markers.
The order of the seven microsatellite markers used in this study was analyzed using the Stanford high-resolution TNG3 radiation hybrid panel. All of the markers had been mapped and ordered previously by Genethon (34 , 35) . We found the mapping order of these markers to agree with the Genethon map, with the exception of marker D16S3144. According to our radiation hybrid mapping data, marker D16S3144 lies distal to marker D16S504. Table 2 ⇓ shows the chromosomal order of the markers based on analysis with Map Manager QT. An intermarker distance could not be obtained for the most proximal markers used in the study (D16S518 and D163049), suggesting that they were too far apart to map with this panel. The estimated physical distance between the markers is shown in Table 2 ⇓ . Markers D16S3096 and D16S516,which were subsequently shown to exhibit the highest frequency of AI, were shown to lie within approximately 118 kb, based on our radiation hybrid mapping results.
Radiation hybrid mapping of seven microsatellite markers on 16q
Allelic Imbalance of 16q Markers.
Fifty-one prostate cancer samples were screened for AI using a panel of seven microsatellite markers on chromosome 16q23. The overall frequency of AI with each marker is shown in Table 3 ⇓ . The markers that showed the highest frequency of AI were D16S3096 (45%) and D16S516 (53%). The most distal marker, D16S3040, showed the lowest frequency (14%) of AI, and this may reflect background instability. Twenty-eight of the 51 (55%) samples studied showed AI involving D16S3096and/or D16S516, and interstitial AI involving one or both of these two markers was seen in 19 of the 51 (37%) samples. We define interstitial AI in this study as prostate tumors that show genomic deletions within the region defined by the five internal markers used in the study (D16S3049, D16S3096, D16S516, D16S504, and D16S3144). The patterns of AI for these 19 samples are shown in Fig. 1 ⇓ . Representative histograms for two of the samples showing interstitial imbalance within this region are shown in Fig. 2 ⇓ .
Specific loss on 16q involving D16S3096 and/or D16S516 in 19 primary prostate tumors. Summary of samples that showed interstitial AI. Sample numbers are listed across the top; markers are listed on the left. □, informative samples with no AI; ▪, informative samples with AI; ovals, noninformative (homozygous) samples; dashes, no result. The maximum area of AI is boxed for each sample.
Examples of histograms from AI studies on normal (N)/tumor (T) prostate cancer pairs. A, sample 3-104. The top panel shows no AI for marker D16S3096 (T:N ratio = 0.854), the middle panel shows AI for marker D16S516 (T:N ratio = 0.517), and the bottom panel shows no AI for marker D16S504 (T:N ratio = 1.189). B, sample 7-433. The top panel shows no AI for marker D16S3096 (T:N ratio = 0.856), the middle panel shows AI for marker D16S516 (T:N ratio = 1.496), and the bottom panel shows no AI for marker D16S504 (T:N ratio = 0.911). X axis, size (in bp); Y axis, fluorescence intensity (in arbitrary units); arrows, the alleles.
AI at chromosome 16q23 loci in 51 primary prostate tumors
Clinical Associations.
Potential clinical associations with the presence of AI at each marker were assessed. Clinical parameters examined included age, family history, surgical Gleason grade, PSA at diagnosis, and pathological T stage. No noteworthy associations were identified when all 51 samples were examined. However, we also examined any associations between those samples that showed interstitial AI involving markers D16S3096and D16S516 (from Fig. 1 ⇓ ) compared with seven samples that showed no AI at either marker. This analysis was performed to eliminate any confounding due to adjacent regions of AI. Here, we observed a significant association between family history and the presence of AI at markers D16S3096 and D16S516(P = 0.048). Eight of the 19 samples (42%) exhibiting AI at either marker had a positive family history of prostate cancer in at least one first-degree relative. In contrast, none of the seven samples without AI at either marker had a family history of prostate cancer. If samples 5-905, 7-451, and 7-485 are removed from the analysis because they were not informative at one or both markers, this relationship did not change (P = 0.048).
DISCUSSION
A goal of this study was to provide further evidence for a prostate cancer susceptibility locus on chromosome 16q23. Our previous genome-wide genetic linkage analysis of 504 sibling brothers affected with prostate cancer showed significant linkage to five genomic regions, with the strongest association on chromosome 16q23 (15) . On 16q23, five consecutive markers (D16S3049, D16S3096, D16S516, D16S504, and D16S3040) showed significant linkage, covering a distance of approximately 7.5 cM. The strongest association was with marker D16S3096 (15) . We argued that the identification of increased AI in prostate tumors using the same markers as those used in the linkage studies would provide further support for a prostate cancer susceptibility gene on chromosome 16q23. Our analysis of 51 primary prostate tumors not only revealed a high AI within this region, but also helped to localize the candidate region to approximately 118 kb.
The highest frequency of AI was found with markers D16S3096(45%) and D16S516 (53%) that map to chromosome 16q23.2. Previous studies that examined AI at 16q23.2 in prostate tumors reported frequencies that varied between 23% and 56% (17 , 18 , 20 , 21 , 25) . Studies that used samples with greater than 50% tumor involvement (similar to our study) reported 48–56% AI in this region (18 , 21 , 25) , consistent with our findings. Furthermore, we found that 37% (19 of 51) of prostate tumors showed interstitial AI involving markers D16S3096and/or D16S516. In all of these tumors, deletions were restricted between the most distal and proximal markers used in the study. These data suggest that we have precisely localized a candidate prostate cancer tumor suppressor gene between markers D16S3096 and D16S516. We estimate the distance between markers D16S3096 and D16S516 to be approximately 118 kb using the Stanford high-resolution TNG3 radiation hybrid panel. This estimate of distance between these two markers is supported by the fact that we have subsequently identified four bacterial artificial chromosome clones that are positive for both markers. 5
At least three loci on chromosome 16q have been reported to be involved or deleted in prostate tumors, including 16q23 (17 , 18 , 25) . In a study of 59 prostate tumors, Latil et al. (4) reported an AI frequency of approximately 50% with markers D16S518 and D16S507 on 16q23.2, a region they estimated to be 10 cM. This region encompasses markers D16S3096 and D16S516 that define our region of AI. A commonly deleted region on 16q23–24 defined by markers D16S515 and D16S516 was also reported in prostate tumors by LOH studies (25) . This region also encompasses markers D16S3096 and D16S516 but may extend more proximal to these markers, where another region of AI in prostate tumors has been mapped. A third study of LOH in prostate tumors also implicates 16q23.2–24.1, but once again, the region defined extends distal to 16q23.2 and may encompass a region of AI implicated in prostate cancer metastasis (17) . Therefore, we have precisely mapped a region of AI in prostate tumors that may have been reported previously in independent studies.
We did not identify noteworthy associations between any of the clinical parameters examined and the presence or absence of AI at any marker when all of the samples were included in the analysis. However, we did identify a statistically significant association between AI at markers D16S3096 and D16S516 and family history of prostate cancer when comparing those samples that showed interstitial AI involving these two markers with those without any evidence for AI with these markers. These restricted analyses were undertaken to eliminate potential confounding due to possible overlapping genomic deletions corresponding to adjacent distal and proximal prostate cancer loci. In conjunction with our independent sibling pair linkage analyses that identified this region (and marker D16S3096 in particular), these findings strongly suggest that this region may harbor a prostate cancer tumor suppressor gene involved in both nonfamilial and hereditary forms of prostate cancer.
If independent studies support our findings, this would suggest that a gene in this region may be analogous to the gene APC, which is the familial adenomatous polyposis colon cancer susceptibility gene (36) , as well as being implicated in the development of the majority of colorectal cancers (37) . LOH corresponding to chromosome 16q23 has also been reported in breast cancer and other cancers (26, 27, 28) ; therefore, this locus may harbor a tumor suppressor gene that is inactivated in many different cancers. No tumor suppressor genes have been reported to be mapped to the region that we have identified. Interestingly, two recent publications report that this region contains a fragile site, FRA16D (38 , 39) . Fragile sites are thought to be more prone to breaks; therefore, a neighboring tumor suppressor gene may be deleted as a result of such breakage.
In summary, we have identified a high frequency of AI in prostate tumors on chromosome 16q23.2 localized to two markers, D16S3096and D16S516. This locus corresponds to a region that we identified previously in our prostate cancer sibling pair linkage analyses (15) . Taken together, these data suggest that we have precisely localized a region of chromosome 16q23.2 that harbors a prostate cancer tumor suppressor gene implicated in the development of both nonfamilial and familial forms of prostate cancer cases.
Acknowledgments
We thank Linda Webster for patient identification and Alicia Paullin for sample accrual.
Footnotes
-
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.
-
↵1 Supported in part by Grant DAMD17-98-1-8589 from the United States Army and by grants from the General Motors Foundation and the Urologic Research Foundation.
-
↵2 To whom requests for reprints should be addressed, at Department of Cancer Biology, ND50, Lerner Research Institute, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195. Phone: (216) 445-9754; E-mail: caseyg{at}ccf.org
-
↵3 The abbreviations used are: LOH, loss of heterozygosity; PSA, prostate-specific antigen; AI, allelic imbalance.
-
↵4 http://gdbwww.gdb.org.
-
↵5 P. L. Paris and G. Casey, unpublished data.
- Received December 22, 1999.
- Accepted May 15, 2000.
- ©2000 American Association for Cancer Research.
References
Volume 60, Issue 13
