Human Genetics

, Volume 120, Issue 2, pp 187–192

Variants in the HEPSIN gene are associated with prostate cancer in men of European origin

Authors

  • Prodipto Pal
    • Department of Environmental Health, Center for Genome InformationUniversity of Cincinnati
  • Huifeng Xi
    • Department of Environmental Health, Center for Genome InformationUniversity of Cincinnati
  • Ritesh Kaushal
    • Department of Environmental Health, Center for Genome InformationUniversity of Cincinnati
  • Guangyun Sun
    • Department of Environmental Health, Center for Genome InformationUniversity of Cincinnati
  • Carol H. Jin
    • Department of PsychiatryWashington School of Medicine
  • Li Jin
    • Department of Environmental Health, Center for Genome InformationUniversity of Cincinnati
  • Brian K. Suarez
    • Department of PsychiatryWashington School of Medicine
  • William J. Catalona
    • Department of Urology, Robert H. Lurie Comprehensive Cancer Center, Feinberg School of MedicineNorthwestern University
    • Department of Environmental Health, Center for Genome InformationUniversity of Cincinnati
Original Investigation

DOI: 10.1007/s00439-006-0204-3

Cite this article as:
Pal, P., Xi, H., Kaushal, R. et al. Hum Genet (2006) 120: 187. doi:10.1007/s00439-006-0204-3

Abstract

There is considerable evidence that genetic factors are involved in prostate cancer susceptibility. We have studied the association of 11 single nucleotide polymorphisms (SNPs) in the HEPSIN gene (HPN) with prostate cancer in men of European ancestry. HPN is a likely candidate in prostate cancer susceptibility, as it encodes a transmembrane cell surface serum protease, which is overexpressed in prostate cancer; HPN is also located on 19q11–q13.2, where linkage is found with prostate cancer susceptibility. In this case-control association study (590 men with histologically verified prostate cancer and 576 unrelated controls, all of European descent), we find significant allele frequency differences between cases and controls at five SNPs that are located contiguously within the gene. A major 11-locus haplotype is significantly associated, which provides further support that HPN is a potentially important candidate gene involved in prostate cancer susceptibility. Association of one of the SNPs with Gleason score is also suggestive of a plausible role of HPN in tumor aggressiveness.

Introduction

Prostate cancer is the most commonly diagnosed visceral malignancy and the third- leading cause of death from cancer among men in the United States and in the western world (Jemal et al. 2006). There is considerable evidence that genetic factors are associated with the development and progression of prostate cancer, with an estimated risk of about 40% explained by hereditary factors (Lichtenstein et al. 2000). Linkage analyses have identified susceptibility loci on several chromosomes (Simard et al. 2002). Using whole genome linkage scans on affected brothers, we have identified regions on chromosomes 2q, 4q, 5q, 7q, 12p, 15q, 16p, 16q, and 19q that are associated with susceptibility to prostate cancer and/or aggressiveness to the disease (Suarez et al. 2000a, b; Witte et al. 2000, 2003; Goddard et al. 2001; Neville et al. 2003).

In this study, we report the association of single nucleotide polymorphisms (SNPs) on the HEPSIN gene (HPN) with susceptibility to prostate cancer among men of European descent. HPN, located on 19q11–13.2 is a 26 kb long gene with 13 exons and encodes a type II transmembrane cell surface serine protease that is over-expressed in prostate cancer (Xu et al. 2000; Dhanasekaran et al. 2001; Magee et al. 2001; Rhodes et al. 2002; Chen et al. 2003; Stephan et al. 2004). In our search for prostate cancer aggressiveness loci, we identified 19q12–q13 as a region harboring potential candidate genes for prostate cancer (Witte et al. 2000, 2003; Neville et al. 2003). The selection of HPN for this study is based on its location in this chromosomal region and our previous finding of the association of two HPN SNPs with prostate cancer (Burmester et al. 2004). In this work, we have tested the association of 11 SNPs distributed throughout the gene following a case-control design in 590 prostate cancer cases and 576 unrelated controls, both of Caucasian ancestry. Our results show significant association of five SNPs with prostate cancer. Although 19q was identified as a region of tumor aggression, these results indicate that HPN is also a plausible candidate for prostate cancer susceptibility. Additionally, association of one of the SNPs with Gleason score in our study is also suggestive of a role of HPN in tumor aggressiveness.

Materials and methods

Participants

The details of the enrollment of the cases (590 men of European descent with histologically confirmed prostate cancer recruited from 304 families) and controls (576 unrelated European men) were reported previously (Burmester et al. 2004; Suarez et al. 2005). Of the cases, 62 were single sibs, three first cousins, six half-sibs, and the remainder was derived from multiplex affected full sibships. The cases were ascertained from patients seen at the Washington University School of Medicine from 1989 to 2001 by staff urologists or were referred by other area urologists, or were participating in prostate cancer support groups or responded to published solicitations. Control subjects were recruited during the same time frame and were matched to the residential area of the cases, the St. Louis Metropolitan area. Control subjects were followed for several years as part of a long-term prostate cancer screening study in which men were screened at 6–12-month intervals with PSA blood tests and digital-rectal examination (DRE) of the prostate (Smith et al. 1997). The controls met the following four criteria: (a) were at least 65 years old, (b) never had registered a PSA level above 2.5 ng/ml, (c) never had DRE suspicious for prostate cancer, and (d) had no known family history of prostate cancer. The protocol for this study was approved by the Human Studies Committee of Washington University and the Institutional Review Board of the University of Cincinnati. Written informed consent was obtained from all participants.

DNA analysis

Twenty nanograms of genomic DNA were preamplified using the improved-primer extension preamplification (I-PEP; Dietmaier et al. 1999) method for whole genome amplification (WGA). This step was necessary to maximize the amount of DNA for genotyping purposes. High-fidelity Expand Template System kit for I-PEP was obtained from Roche Pharmaceuticals (Nutley, NJ, USA). The WGA protocols are validated for analysis of genetic markers in our laboratory (Jiang et al. 2005; Sun et al. 2005).

We analyzed 11 SNPs (Table 1, Fig. 1) spanning a region of ∼26 kb in HPN. SNPs were selected from the NCBI database (www.ncbi.nih.gov), based primarily on their locations in the gene, the validity status, and the heterozygosity. Of the selected markers, one is located at the 5′ untranslated region (UTR), two in the 3′ UTR, and the remaining eight are intronic SNPs. We sequenced the 5′ and the 3′ UTRs as well as all the exons and exon-intron boundaries in 16 randomly chosen samples (eight cases and eight controls) to identify new SNPs. However, we did not find any variant other than the ones that are already available in the NCBI dbSNP.
Table 1

HPN SNP locations, MAF and their association in prostate cancer cases and controls

SNP No (gene location)

rs numbera

NCBI location

Minor allele frequency

Likelihood ratio χ2

P

Cases (n)

Controls (n)

SNP1 (5′UTR)

rs10410046GA

40223642

0.012 (580)

0.018 (573)

1.477

0.224

SNP2 (INT 3)

Rs870379CT

40235596

0.356 (570)

0.332 (558)

1.137

0.286

SNP3 (INT 3)

Rs2451996GA

40236803

0.431 (551)

0.427 (549)

0.021

0.886

SNP4 (INT 8)

Rs1688043AG

40245181

0.146 (548)

0.153 (558)

0.197

0.657

SNP5 (INT 8)

Rs1350290CT

40245812

0.058 (543)

0.081 (543)

3.665

0.055

SNP6 (INT 8)

Rs1615767GA

40246607

0.057 (569)

0.082 (548)

4.743

0.029

SNP7 (INT 9)

Rs2305745AG

40248121

0.243 (561)

0.292 (547)

5.373

0.021

SNP8 (INT 9)

Rs2305746AG

40248140

0.055 (576)

0.078 (552)

3.929

0.047

SNP9 (INT 9)

Rs2305747CT

40248197

0.241 (552)

0.293 (523)

6.268

0.012

SNP10 (3′ UTR)

Rs1042328TG

40249100

0.327 (571)

0.306 (550)

0.892

0.345

SNP11 (3′ UTR)

rs1688029AG

40249280

0.319 (563)

0.309 (549)

0.211

0.646

aThe second nucleotide represents the minor allele

https://static-content.springer.com/image/art%3A10.1007%2Fs00439-006-0204-3/MediaObjects/439_2006_204_Fig1_HTML.gif
Fig. 1

Schematic representation of HEPSIN showing the relative locations of the 11 studied SNPs within the gene

The TaqMan™ (fluorogenic 5′ nuclease) assay was used for SNP genotyping. The primers and probes were obtained from Applied Biosystems (Foster City, CA, USA). PCR was conducted with both primers and probes added in ABI 9700 thermocycler, and the end-point results were scored using the ABI 7900HT Sequence Detection System. In each 384-well plate, two reference samples and two negative controls were included for quality control.

Statistical analysis

Allele frequencies were obtained by a maximum likelihood estimates using the USERM13 subroutine of MENDEL, which is specifically designed to take into account the relatedness of the samples to one another (Lange et al. 1988; Boehnke 1991). Allele frequencies in cases and controls were compared using a likelihood ratio test as described in Suarez et al. (2005). Conformity of genotype proportions to Hardy–Weinberg expectations was performed by a goodness-of-fit chi-square test statistic.

Pairwise linkage disequilibrium (LD) values, D′ and R2, were estimated using the HelixTree software (Golden Helix Inc., Bozeman, MT, USA) (Zaykin et al. 2002). Haplotypes were inferred using Phase, which implements Bayesian methods for estimating haplotypes from population data (Stephens et al. 2001). One sib was selected from each sibship, using a permutation-based randomized method, to represent an unrelated group of population sample. Differences in haplotype frequencies between cases and controls were tested using an r/c contingency test, and significance was empirically determined based on 10,000 permutations. Correction for multiple testing was performed using the permutation test by defining the smallest empirical significance level as our statistic of interest and assessing its significance level (Westfall and Young 1993).

Results

Table 1 gives details of the studied SNPs, their genomic locations, dbSNP ‘rs’ numbers, minor allele frequencies (MAF) in cases and controls and their associations with prostate cancer. The NCBI locations show the physical distances among the SNPs. With the exception of SNP1 (MAF = 0.012 in cases and 0.018 in controls), all other SNPs are highly polymorphic in both cases and controls (MAF > 0.05). None of the SNPs shows departure from Hardy–Weinberg expectations (P-values ranged from 0.06 to 0.97). Likelihood ratio tests show significant association of five contiguously located intronic SNPs (SNPs 5–9), with prostate cancer.

Figure. 2 shows the results of D′ and R2 analyses. We observe three separate blocks of strong LD (D′ > 0.98); the first between SNPs 2 and 3 (1,207 bp apart), the second among SNPs 5–9 (2,385 bp), and the third between SNPs 10 and 11 (180 bp). All of the significantly associated SNPs belong to the second LD block. SNP1 being marginally polymorphic in both cases and controls, its association with the other SNPs cannot be accounted for. SNP4 does not belong to any of these LD blocks.
https://static-content.springer.com/image/art%3A10.1007%2Fs00439-006-0204-3/MediaObjects/439_2006_204_Fig2_HTML.gif
Fig. 2

Linkage disequilibrium (LD) analysis of 11 SNPs in HEPSIN in the control population. Both D′ and R2 values are estimated with HelixTree software

We performed an analysis of association at the haplotype level, for which, as noted above, one sib was randomly selected from each sibship with prostate cancer to establish an unrelated population of cases. The 11 SNP markers resulted in 34 haplotypes, showing a moderate level of recombination within the gene region. Table 2 shows the distribution of haplotypes with frequency ≥1% in either cases or controls. The rare haplotypes are pooled together. One major haplotype, GCGATAAGTTG (17% in cases and 12% in controls) shows significant association (P=0.007) that remained significant after correcting for multiple testing (P=0.041).
Table 2

Distribution of major haplotypes in cases and controls and their association with prostate cancer

Haplotypes

Cases (n=606)

Controls (n=1152)

Chi-square

P

Multuple adjusted P

n

Frequency

n

Frequency

GTAATAAGTGA

311

0.513

551

0.478

1.936

0.168

0.830

GCGATAAGTTG

104

0.172

138

0.120

8.985

0.007

0.041

GCGATAGGCGA

58

0.096

129

0.112

1.106

0.297

0.984

GTGATAGGCTG

41

0.068

89

0.077

0.534

0.469

1.000

GCGGCGGACTG

27

0.045

68

0.059

1.627

0.206

0.937

GTAGTAAGTGA

33

0.054

81

0.070

1.647

0.204

0.937

ATAATAAGTGA

3

0.005

16

0.014

2.968

0.094

0.617

GCGACGGACTG

4

0.007

13

0.011

0.910

0.446

1.000

Rare haplotypes (<0.01)

25

0.041

67

0.058

   

Haplotypes based on SNPs 5–9

TAAGT

466

0.769

818

0.710

6.997

0.012

0.019

TAGGC

105

0.173

235

0.204

2.403

0.125

0.290

CGGAC

32

0.053

86

0.075

3.027

0.086

0.137

Rare haplotypes (<0.01)

3

0.005

13

0.011

0.040

  

Next we performed haplotype analysis based on the five markers (SNPs 5–9) that show significant allele frequency differences between cases and controls. The most common haplotype TAAGT (77% in cases and 71% in controls) is significantly overrepresented in cases (P=0.012), which remained significant after multiple correction (P=0.019). The LD analysis reveals SNPs 5, 6, and 8 to be in complete LD (D′=1 and R2>0.95), which renders two of these three SNPs redundant in information. Also, SNPs 7 and 9 are in complete LD (D′=1; R2=0.99). We selected SNPs 6 and 7 as the two informative markers from each LD group and inferred haplotypes; the most common haplotype, AA, is found to be significantly over-represented in cases (77% vs. 72%; P=0.034).

We had previously reported that 19q12–13, the region harboring HPN, is associated with prostate tumor aggressiveness (Witte et al. 2000, 2003; Neville et al. 2003). To test for association of the HPN variants with Gleason score as an index of tumor aggressiveness, we divided the cases into two groups with Gleason scores 2–6 and 7–9. It has been reported previously that patients with Gleason score 7 and above are at a higher risk of adverse outcomes from prostate cancer (Stephan et al. 2004; Albertsen et al. 2005). We analyzed the allele frequency differences at all 11 SNPs between the two groups noted above (Table 3). A highly significant association is observed at SNP4 (P<0.0005). None of the other SNPs shows significant allele frequency differences between the two groups.
Table 3

Allelic association of HPN SNPs in high (score 7–9) and low (score 2–6) gleason groups

SNP

Minor allele frequency

Likelihood ratio χ2

P

Low Gleason score 2–6 (n=363)

High Gleason score 7–9 (n=140)

SNP1

0.014

0.004

1.994

0.158

SNP2

0.328

0.384

2.313

0.128

SNP3

0.405

0.456

1.696

0.193

SNP4

0.163

0.076

11.967

<0.0005

SNP5

0.052

0.049

0.050

0.823

SNP6

0.049

0.053

0.039

0.842

SNP7

0.217

0.254

1.256

0.262

SNP8

0.049

0.052

0.035

0.852

SNP9

0.216

0.249

0.986

0.320

SNP10

0.318

0.361

1.391

0.238

SNP11

0.306

0.349

1.383

0.239

Discussion

HEPSIN gene is a potential candidate involved in prostate cancer susceptibility. It encodes a transmembrane serine protease that is over expressed in cancerous prostate tissue (Rhodes et al. 2002), particularly in the aggressive form of the cancer (Dhanasekaran et al. 2001; Stamey et al. 2001). HPN is located in 19q11–13.2, where linkage was found with prostate tumor aggressiveness (Witte et al. 2000, 2003; Neville et al. 2003). We had previously analyzed three HPN SNPs, two of which (rs1350290 and rs1688030) showed significant allele frequency differences between prostate cancer cases and controls of European origin (Burmester et al. 2004). To further saturate the HPN region, in this study, we have analyzed 11 SNPs, including two previously studied (rs1350290 and rs1042328) markers. The present results confirm the previously found association of rs1350290 (SNP5) as well as the lack of association of rs1042328 (SNP10). The second significantly associated SNP, rs1688030, in the previous study is located in intron 10. This SNP is in close proximity to SNP 9 (rs2305747), which shows significant association in the present study. Our finding of significant associations of five contiguously located SNPs (SNPs 5–9) that belong to a block of strong LD provides stronger support to the potential biological significance of HPN in prostate cancer susceptibility. These observations also suggest plausible roles of regulatory variants (all five SNPs are intronic) in disease risk. However, it needs to be explored whether these SNPs are in LD with functional variant(s).

Haplotype analyses show that a major haplotype spanning all 11 SNPs is significantly associated with prostate cancer susceptibility. Significant association is also observed for the most common haplotype consisting of the five SNPs (5–9) that show significant allele frequency differences between our cases and controls. As described in the results, SNPs 6 and 7 are likely to be informative markers within the 5-locus haplotype group. We find the haplotype ‘AA’ for these two SNPs to be significantly over-represented in cases. Interestingly, this two-SNP haplotype is embedded in the significantly associated 11-marker and 5-marker haplotypes (GCGATAAGTTG and TAAGT) described above. At this time, the functional role of this observed significance is unclear and requires further work. Nonetheless, these observations strengthen the role of HPN as a potentially important candidate gene involved in the etiology of prostate cancer.

An interesting observation is the significant association of SNP4 (rs1688043) with the Gleason scores, classified as low (2–6) and high (7–9). The data shows the frequency of the ‘major’ allele is significantly higher in patients with high-Gleason score (92.4%) compared to the patients with low-Gleason score (83.7%). While it is premature to draw a definitive conclusion based on a single marker association, it is plausible that the ‘minor’ allele at this SNP may act as a ‘protective allele’ against the aggressiveness of prostate cancer. However, we recognize that further work is necessary to establish this association.

HEPSIN is a membrane-associated serine protease that might activate growth factor precursors that stimulate prostate cancer cell proliferation. There is overwhelming evidence of over expression of HEPSIN in prostate cancer (Magee et al. 2001; Chen et al. 2003; Stephan et al. 2004). These studies underscore the importance of HEPSIN as a pharmacologic target for prostate cancer therapy. Our study reinforces that HPN as a DNA marker also holds prognostic value in prostate cancer. It is possible that genetic variants in HPN, together with the expression profiles at the RNA level, could be used for identifying susceptibility to prostate cancer and tumor aggressiveness. Also, since there is a promising experience in developing protease inhibitors for treatment of HIV, it is possible that specific inhibitors for HEPSIN could be developed for the treatment and perhaps even for the prevention of prostate cancer.

Acknowledgments

This study was supported by a grant from the Urological Research Foundation.

Copyright information

© Springer-Verlag 2006