Optimal vitamin D status been proposed to prevent prostate carcinogenesis  and this anticancer activity is most likely mediated through VDR-dependent changes in the prostate transcriptome . By applying microarray technology to the immortalized but non-tumorigenic human prostate epithelial cell line RWPE1, we have identified a number of mechanisms by which vitamin D may influence the early stages of prostate carcinogenesis. Our data show that 1,25(OH)2D influences many pathways relevant to prostate carcinogenesis and they underscore the critical role of this molecule in the maintenance of prostate epithelial development, function, and turnover.
Several studies have shown that 1,25(OH)2D treatment causes cell cycle arrest and growth suppression of primary prostate epithelial cell lines or prostate cancer cell lines. While this has been viewed as the major anticancer effect for 1,25(OH)2D, the mechanism accounting for this effect is not known with certainty. Liu et al. showed that the cyclin-dependent kinase (CDK) inhibitor p21 was strongly induced by 1,25(OH)2D treatment in the monocytic cell line HL-60 and they identified a functional Vitamin D Response Element (VDRE) in the p21 promoter . However, 1,25(OH)2D does not increase p21 transcript level in LNCaP cells  and our data show that the impact of 1,25(OH)2D on p21 mRNA level is modest (1.32-fold at 6 h). Another CDK inhibitor, Wee1, is induced by 1,25(OH)2D in cultured keratinocytes leading to G2/M arrest  and this is also modestly induced by 1,25(OH)2D in RWPE1 cells (1.4-fold at 6 h). We examined other transcripts related to cell cycle control in our study but most of these were suppressed only at the later time points: e.g. GAS6, ETS1, CDK6, cyclin B2, cyclin A, CDC25C, and CDC27. This suggests that they are not primary effects of 1,25(OH)2D action.
In contrast, our microarray data suggest that disruption of Wnt-signaling may be an alternative mechanism for 1,25(OH)2D-mediated growth arrest. A number of Wnt pathways and genesets were reduced by 1,25(OH)2D treatment by 6 h, e.g. the geneset containing genes with LEF1/TCF4 binding motifs in their promoters that includes classical Wnt target genes like c-myc, cyclin D, PPARδ (see tables 4, 5 and 6). This is consistent with a model developed for colonocytes where VDR directly interacts with β-catenin to disrupt transcriptional events that normally increase cell proliferation [24, 25]. 1,25(OH)2D treatment also induced E-cadherin mRNA 1.7-fold in RWPE1 cells at 6 h. E-cadherin antagonizes Wnt signaling by inducing translocation of β-catenin to the plasma membrane (see figure in Additional file 10 for a summary of the transcript-level changes occurring in Wnt signaling). Disruption of Wnt/β catenin signaling could be a means whereby vitamin D treatment amplifies its impact on biology. In support of this model, network analysis in Metacore identified a gene network with the β catenin/TCF gene target c-myc at its center. This network connects the suppression of c-myc expression to a large number of other transcripts that were differentially expressed by 1,25(OH)2D treatment (see Additional file 11). However, while our data shows a consistent, early suppression of Wnt/β-catenin signaling, careful experimental evaluation of this hypothesis in prostate epithelial cells is needed.
Our data also show that transcripts for the Notch ligands, JAG1, JAG2, and DLL1 were suppressed by 1,25(OH)2D treatment in RWPE1 cells (each reduced by -2 fold at 6 or 24 h). NOTCH1 and JAG1 are proposed as markers of normal prostate stem cells and they are necessary for fate determination of the proliferating stem cell . In addition, expression of JAG1 protein is increased in metastatic prostate cancer  and prostate cancer cells suggesting Notch1 and JAG1 mediated signaling may enhance carcinogenesis [26, 28]. Collectively, these observations suggest that suppression of Notch or its ligands by 1,25(OH)2D could be associated with cancer protection.
In addition to modulating cell grown, vitamin D has been proposed to inhibit the development of the tumor vasculature that is required for the progression of solid tumors [29–32] by suppressing expression of Vascular Endothelial Growth Factor (VEGF) family members, the major pro-angiogenic cytokines in normal prostate epithelial cells . The impact of 1,25(OH)2D on VEGF gene regulation has been confusing. In mouse embryo fibroblasts and human vascular smooth muscle cells 1,25(OH)2D induces VEGFA expression through a VDRE in its promoter , yet 1,25(OH)2D can also suppress VEGF-induced vasculogenesis in cultured endothelial cells and in nude mice implanted with MCF-7 breast cancer cells . We found that 1,25(OH)2D treatment suppressed VEGFC mRNA levels at all time points (-2.1 to -1.6 fold). Higher expression of VEGFC occurs after NKX3.1 loss in prostate cancer and is correlated with lymph node metastasis of prostate cancer . VEGF promotes angiogenesis by binding to and activating the receptors KDR, FLT1, and NRP1 and we found that 1,25(OH)2D significantly suppressed KDR and NRP1 expression. Suppressing the activation of these receptors reduces tumor angiogenesis and promotion in the Dunning cell carcinoma model . Finally, VEGF signaling can be suppressed by competitive binding of semaphorins to NRP1. Semaphorins induce apoptosis, inhibit growth of lung and breast tumor cells , and modulate invasion and adhesion of prostate cancer cells . In our study, 1,25(OH)2D induced expression of several semaphorin isoforms including SEMA3B, 3F, and 6D (19.4-, 2.5-, and 18-fold, respectively at 6 h). Collectively our array data show that 1,25(OH)2D induces an anti-angiogenic transcript profile in RWPE1 cells.
While our discussion has focused on the modulation of classical anti-cancer effects, our array analysis also revealed other potential mechanisms for vitamin D mediated cancer prevention. For example, oxidative stress-induced damage of DNA and other cellular components are implicated in cancer . These effects can be prevented by induction of antioxidant defense or DNA repair mechanisms that subsequently reduce the biological impact of reactive oxygen species. In our study 1,25(OH)2D influenced the expression of genes related to these events (see figure in Additional file 12) and our observations are consistent with a previous microarray study in SCC25 cells that showed EB1089-regulated induction of transcripts whose gene products are involved in antioxidant (e.g. thioredoxin reductase 1, TXNRD1) and DNA repair processes (e.g. GADD45α) . There is some evidence that 1,25(OH)2D directly regulates transcripts controlling these functions. Glucose-6-phosphate dehydrogenase (G6PD) is an enzyme involved in maintaining cellular glutathione levels and its mRNA was significantly induced at all time points following 1,25(OH)2D treatment in RWPE1 cells (3.4-6.8 fold). Recently, Bao et al.  showed that G6PD expression is controlled by 1,25(OH)2D in prostate epithelial cells through a VDRE located in the first intron of the gene and that the induction of G6PD by 1,25(OH)2D protected RWPE1 cells against H2O2-induced apoptosis. It is also possible that vitamin D-mediated protection from pro-oxidant stress is indirect due to the induction of nuclear factor (erythroid-derived 2)-like 2 (NFE2L2), a transcription factor that controls expression of genes for many antioxidant enzyme systems . NFE2L2 expression is down-regulated in prostate cancer and suppression of NFE2L2 promotes prostate tumor development in TRAMP mice . Consistent with a role for NFE2L2 in vitamin D-mediated cancer prevention, a number of NFE2L2 target genes were increased in RWPE1 cells after 1,25(OH)2D treatment, e.g. GPX3, HMOX1, AKR1C2, and TXNRD1.
Several studies have shown that vitamin D is anti-inflammatory and our data are consistent with these findings. In the immune system 1,25(OH)2D promotes immunotolerance and immunosuppression by altering the differentiation and function of tolerogenic dendritic cells , suppressing NFkB signaling necessary for T helper cell activation , and increasing the activity of regulatory T cells necessary for immunosuppression . These actions would be expected to protect tissues from pro-inflammatory stresses that cause prostatitis  and promote prostate carcinogenesis . However, many cells outside of the traditional immune system have the capacity to respond to and produce immuno-modulatory factors and we found that in RWPE1 cells vitamin D-treatment regulated a large number of transcripts for proteins controlling immune function. In fact, induction of CD14 was one of the most strongly 1,25(OH)2D up-regulated transcripts in RWPE1 cells. Prostate epithelial cells are thought to be early sensors of infection and CD14 and toll-like receptor 4 (TLR4) production in these cells contributes to protection from Chlamydia infection . While a role for vitamin D-induced CD14 or TLR4 induction in the regulation of prostate infection/inflammation has not been studied directly, others have identified CD14 as crucial factor for vitamin D induced expression of the antimicrobial peptide cathelicidin in human keratinocytes .
Another point where vitamin D may inhibit inflammatory processes is through suppression of cytokine signaling and production. Consistent with this, we found that 1,25(OH)2D suppressed several components of JAK-STAT signaling in RWPE1 cells including JAK1, STAT1, n-myc and STAT interactor (NMI), STAT2, and STAT3. JAK-STAT signaling is required for the pro-proliferative effects of many cytokines including the pro-inflammatory cytokines IL6, IL12 and IFNγ . In addition, transcripts for many cytokine receptor transcripts (i.e. upstream regulators of JAK-STAT signaling) and cytokines (i.e. downstream targets of JAK-STAT signaling) were suppressed by 1,25(OH)2D-treatment (see figure in Additional file 13). Disruption of JAK-STAT signaling could be a means whereby vitamin D treatment amplifies its impact on the prostate epithelial cells. In support of this model, network analysis of the data from the 48 h timepoint identified a gene network with STAT1, STAT3, and the transcription factor PU.1 at three interacting centers. This network connects the suppression of STAT1 and STAT3 expression to a large number of other transcripts related to immunoregulation that were differentially expressed by 1,25(OH)2D treatment (see Additional file 14). Many changes in immune or cytokine signaling pathways occur only at or after 24 h of treatment but this reflects a clear anti-inflammatory role for 1,25(OH)2D in prostate epithelial cells that is consistent with findings by others in Jurkat T cells  and Th1 immune cells . In addition, Nonn et al. had previously shown that in normal prostate epithelial cells, 1,25(OH)2D inhibits TNFα-induced IL-6 production through a mechanism that requires direct transcriptional regulation of the MAPK phosphatase 5 gene (DUSP-10, increased 9.9-fold at 6 h in our analysis) . Finally, our data suggest that NFkB signaling is modulated by 1,25(OH)2D treatment; i.e. upregulation of IkB (NFKBIA) expression and suppression of RELB mRNA levels. These observations are consistent with data showing that 1,25(OH)2D suppressed secretion of IL-8 by interfering with NFκB signaling in RWPE1 cells  and that it enhanced radiosensitivity of prostate cancer cells by selectively suppressing radiation-mediated RELB activation in prostate cancer cell lines .
Prostaglandin signaling is a final pro-inflammatory pathway that has been identified as vitamin D regulated by others. Krishnan et al.  found that the message for the prostaglandin inactivating enzyme 15-PGDH was significantly increased and the mRNA levels for COX2, an enzyme that drives production of PGE2 levels, was significantly reduced by 1,25(OH)2D in LNCaP cells. Moreno et al.  subsequently showed that 1,25(OH)2D reduced the mRNA levels for two prostaglandin receptors (EP2, FP) in prostate cancer cells. Since the prostanoid pathway is a critical component in acute inflammation that may contribute to the development of prostate cancer , this suggested that vitamin D-mediated chemoprevention involves disruption of prostaglandin signaling. However, our RWPE1 data is not consistent with this hypothesis. In contrast to the earlier studies in LNCaP cells, we observed induction of COX2 by 1,25(OH)2D in RWPE1 cells (6.3-fold at 6 h) and neither 15-PGDH nor prostaglandin receptor mRNA levels were altered.
The functional analysis of transcript-levels changes induced by 1,25(OH)2D reveals how the biology of prostate epithelial cells is changed by the hormone but microarray studies cannot differentiate between transcripts that are differentially regulated due to direct, VDR-mediated transcriptional activation and those that are secondary effects following the primary transcriptional events. As such, we can only infer the direct VDR gene targets based upon our data and their relationship to other studies. Unfortunately, there is very little information regarding the effect of 1,25(OH)2D on the prostate epithelial cell transcriptome to draw upon from earlier studies. Using a spotted cDNA microarray with no sample replicates, Peehl et al. [18, 22] identified 48 transcripts as 1,25(OH)2D regulated in primary normal prostate epithelial cells and 52 transcripts in primary cultures of prostate cancer cells. Twenty-one of the differentially regulated transcripts from the normal prostate epithelial cells and 28 of the transcripts from the primary prostate cancer cells were also differentially regulated by 1,25(OH)2D in RWPE1 cells. Still, the only overlap between the three lists was CYP24, DUSP10, AKAP12, P2RY2, BMP6, TGFB2, and TXNRD1. In contrast, of the 22 transcripts identified by Krishnan et al. [18, 22] as differentially regulated in 1,25(OH)2D-treated LNCaP cells only IGFBP3, ABCA1 and FKBP5 were regulated in the same direction in RWPE1 cells. This suggests that the response of RWPE1 cells is more similar to that of primary cultures of human prostate epithelial cells.
Of the genes identified in the three different prostate array studies only CYP24 has been identified as a direct target for 1,25(OH)2D. Our ChIP examination of the AKAP12 and CYP26B1 promoters revealed significant VDR binding to putative VDREs in those promoters too, but given the large number of transcript level changes we observed, we expect that many more primary VDR target genes exist in the prostate epithelial cells. Before our analysis, the most comprehensive array-based analysis of 1,25(OH)2D action and putative VDR target genes was conducted by Wang et al. . Using a bioinformatics approach they identified putative VDREs in the promoters of genes whose transcripts were differentially regulated by treatment with EB1089 in SSC25 cells (12 h in presence of cycloheximide). We compared their list of differentially expressed transcripts to our list of transcripts regulated by 1,25(OH)2D at 6 h in RWPE1 cells based on the assumption that the early time point is less likely to contain transcripts that are regulated as a secondary consequence of primary vitamin D-mediated transcription events. This analysis identified 414 transcripts that were vitamin D regulated in both cell lines. 267 of these had a putative VDRE (see Additional file 15), including 16 of the 21 transcripts regulated by 1,25(OH)2D in both RWPE1 and primary prostate epithelial cells . This suggests there may be a much larger number of direct VDR target genes than has been suggested by earlier research. Future studies using either ChIP-chip or ChIP-sequencing  will be necessary to validate whether these 267 transcripts are truly direct vitamin D target genes.