Expression of HOXC6 was determined by quantitative RT-PCR in 45 prostate cancer and 13 benign tissues collected from prostatectomies. The majority of cancer tissues displayed—often grossly— elevated levels of HOXC6 mRNA resulting in an overall highly significant difference compared to benign tissues (Figure 1A). As reported by others, cancers with high HOXC6 expression had significantly higher T stage, had more often spread to lymph nodes and were assigned higher Gleason scores. Expression of MKI67 encoding the proliferation marker Ki67 was likewise enhanced in these cases (Mann-Whitney test: p = 0.004). Cancers with above median HOXC6 expression recurred significantly (log-rank p = 0.024) earlier than cancers with below median expression (Figure 1B). These data confirm previous reports on frequent HOXC6 overexpression in prostate cancer [7,10,14,15] and the association of increasing HOXC6 overexpression with adverse clinical parameters.

In the same set of samples, expression of DKK3, WIF1 and CNTN1 was observed to be significantly decreased (Figures 2A–C). Expression of each gene correlated inversely with that of HOXC6 in a statistically significant (each p < 0.001) manner (Figures 2D–F). Accordingly, expression of each target gene correlated significantly positively with that of each other, with Spearman rho coefficients between 0.4 and 0.6. Cases with lower than median expression of each target gene, CNTN1, DKK3 or WIF1, showed earlier recurrence, but the association was only significant at the p < 0.05 level for WIF1 (Figures 3A–C). In addition, low WIF1 expression was significantly associated with lymph node involvement (p = 0.036) and higher Gleason scores (p = 0.026), but not with tumor stage (pT2 vs. pT3). Expression of CNTN1 or DKK3 was not significantly associated with any histopathological parameter in our series.

Our measurements confirm the reported downregulation of the WNT factor antagonists DKK3 and WIF1 in prostate cancer [18,21-23,25,26]. In addition, our data hint at an association of stronger WIF1 downregulation with worse prognosis. Most importantly, our study demonstrates for the first time explicitly that expression of certain target genes is inversely correlated with overexpression of HOXC6. This is also the first report on CNTN1 in prostate cancer. The gene encodes a member of the contactin family which serves as a membrane receptor for chondroitin sulfate and regulates receptor tyrosine phosphatases. The function of contactin 1 has mainly been studied in neuronal and glial cells, where it regulates cell-cell and cell-substrate adhesion. Two studies in lung cancers and gliomas suggest that this function may also be relevant for cancer cell invasion and metastasis [27,28]. Investigations on the function of contactin 1 in prostate cancer and of the epigenetic regulation of its complex gene might therefore be rewarding.

Methylation of DKK3 and WIF1 promoter CpG-islands was analyzed by methylation-specific PCR as described in previous publications [21,29]. Methylation of DKK3 was observed in 16 of the 92 carcinoma samples, but not in benign controls. It was significantly more frequent in cases with lymph node involvement and significantly less frequent in cases with Gleason score < 6 (χ² test, p < 0.05). However, expression of DKK3 was not significantly different between carcinoma samples with or without hypermethylation. WIF1 methylation was more prevalent, being detectable in 31 of 92 carcinoma samples, but also in eight of 17 benign controls. The latter finding may relate to previous findings [18] suggesting that WIF1 downregulation may commence at early stages of prostate cancer. As for DKK3, WIF1 hypermethylation and the extent of downregulation of expression were not significantly related to each other. Because the bands obtained in the WIF1 MS-PCR assay with the methylated-specific primers were often weak (except in LNCaP cells used as a positive control), bisulfite sequencing was conducted across the region interrogated by the assay in prostate tissue samples and controls (Figure 4). The analyzed part of the WIF1 CpG-island was completely unmethylated in blood leukocytes, but was quite densely methylated in the prostate cancer cell line LNCaP which lacks WIF1 expression. In contrast, only occasional sites were methylated in the expressing 22Rv1 line. In all benign and carcinoma tissues, only patchy and weak methylation was found, as suggested by the results of the MS-PCR assay. Of note, we have previously shown that in this set of prostate cancer tissues GSTP1, EPB41L3 and several other genes are each hypermethylated at frequencies of 60%–80%, often displaying dense methylation [19,30]. Thus, DKK3 and WIF1 hypermethylation was much less widespread than downregulation and was often weak if it occurred.

High quality RNA was prepared from 13 normal prostate tissues from cancer-carrying prostates and 45 carcinomas from patients aged 59–74 years undergoing total prostatectomy as described [19]. According to the IUAC 2007 TNM classification, tumors were staged as pT2 in 20, pT3 in 23 and pT4 in 2 cases. A Gleason score of 7 was detected in 26 tumors, < 7 in 13 tumors and > 7 in 6 tumors. None of the patients had distant metastases, but 11 cancers had spread to local lymph nodes. High quality DNA was available from 92 cancer tissues encompassing the specimens used for RNA analysis. Of these, 43 were staged as pT2 and 49 as pT3 or pT4. Sixteen patients had lymph node metastases, but none distant metastases. Each 27 carcinomas were assigned a Gleason score > 7 or < 7 and 38 a score of 7. The median follow-up period was 98 months. The study was approved by the ethics committee of the Heinrich Heine University medical faculty.

Total RNA was isolated from preconfluent cells using the RNeasy ^® Mini Kit (Qiagen, Hilden, Germany). Two μg RNA were reversed transcribed using SuperscriptII (Invitrogen, Karlsruhe, Germany) with oligo-dT primers according to the manufacturer's protocol. Quantitative real-time PCR for CNTN1, DKK3, WIF1, and the reference gene TBP was performed using SYBR-Green reaction mix (Qiagen) with 0.4 μM of each primer in an ABI Prism 7900HT instrument (Applied Biosystems, Darmstadt, Germany) with the following conditions: activation at 95 °C for 15 minutes followed by 45 cycles of denaturation at 94 °C for 15 s, elongation at 72 °C for 30 s and measuring at 88 °C for 15 s. Amplification lasted 30 s with temperatures at 55 °C for CNTN1, 57 °C for DKK3 and WIF1, 61 °C for TBP. The quality of the amplification was assured by a melting curve established by incubation at 95 °C, 60 °C and 99 °C for 15 s each. HOXC6 expression was measured by Taqman assays Hs00171690 mL (Applied Biosystems). Duplicate measurements gave < 10% difference. For each gene, a standard curve was constructed using a reference cell line with high expression and expression in the samples was expressed relative to this standard. The same procedure was performed for the TBP control, to which the measurements were then adjusted.

High-quality DNA was extracted from tissues and prostate carcinoma cell lines as described [19]. The EZ DNA Methylation Kit (Zymo Research, Orange, CA, USA) was used for bisulfite conversion. PCRs were performed in a 50 μL reaction mixture consisting of 1 × buffer, 150 μM dNTPs, 15 pmol of each primer, 1 U Hotstar Taq polymerase, water and 2 μL bisulfite-converted DNA each. The following program was used: initial Taq activation at 94 °C for 15 min, followed denaturation at 95 °C for 30 s, annealing for 30 s, elongation at 72 °C for 45 s, with a final elongation for 10 min. MS-PCR amplification was performed for DKK3 unmethylated/methylated for 36/38 cycles with annealing at 61/65 °C, for WIF1 unmethylated/methylated MS-PCR amplification was performed for 34/36 cycles with annealing at 53 °C for both genes. Fully methylated and completely methylated controls were carried in each reaction as described [19]. For WIF1 bisulfite sequencing PCR was conducted for 35 cycles with annealing at 60 °C. PCR products were examined by agarose gel electrophoresis. For bisulfite sequencing PCR products were cloned into the E. coli TOPO 10 vector (TOPO TA Cloning Kit, Invitrogen). Four clones from each sample were sequenced using standard methods. Primer sequences are compiled in Table 1.

Statistical calculations were performed using SPSS 19.0.