1α,25-Dihydroxyvitamin D₃ Inhibits Growth of VCaP Prostate Cancer Cells Despite Inducing the Growth-Promoting TMPRSS2:ERG Gene Fusion

Abstract

Vitamin D receptor (VDR) agonists have been shown to reduce the growth of several prostate cancer cell lines. However, the effects of VDR activation have not been examined in the presence of the recently identified androgen-regulated TMPRSS2:ERG gene fusions, which occur in a high percentage of prostate cancers and play a role in growth and invasiveness. In a previous microarray study, we found that VDR activation induces TMPRSS2 expression in LNCaP prostate cancer cells. Here we show that the natural VDR agonist 1α,25-dihydroxyvitamin D₃ and its synthetic analog EB1089 increase expression of TMPRSS2:ERG mRNA in VCaP prostate cancer cells; this results in increased ETS-related gene (ERG) protein expression and ERG activity as demonstrated by an increase in the ERG target gene CACNA1D. In VCaP cells, we were not able to prevent EB1089-mediated TMPRSS2:ERG induction with an androgen receptor antagonist, Casodex, although in LNCaP cells, as reported for some other common androgen receptor and VDR target genes, Casodex reduces EB1089-mediated induction of TMPRSS2. However, despite inducing the fusion gene, VDR agonists reduce VCaP cell growth and expression of the ERG target gene c-Myc, a critical factor in VDR-mediated growth inhibition. Thus, the beneficial effects of VDR agonist treatment override some of the negative effects of ERG induction, although others remain to be tested.

1α,25-Dihydroxyvitamin D₃ induces expression of the growth-promoting TMPRSS2:ERG gene fusion in VCaP prostate cancer cells, but inhibits cell growth and reduces c-Myc expression.

Prostate cancer is the most common noncutaneous cancer and the second most frequent cause of death from cancer in American men. Although prostate cancer typically is androgen dependent, androgen ablation therapy for metastatic disease usually is effective only for a relatively short time (2–3 yr) before the tumors become resistant. Thus, there is a great deal of interest in chemopreventive and therapeutic agents. Among the candidates are vitamin D receptor (VDR) agonists. Some epidemiological studies show that low levels of sunlight (the main source of vitamin D) and low levels of circulating vitamin D metabolites are correlated with higher risk for prostate cancer (1,2,3), although other studies have failed to find a correlation (4). The biologically active metabolite of vitamin D, 1α,25-dihydroxyvitamin D₃ (1,25D) and its less calcemic, more potent analog EB1089 (seocalcitol) (5) are ligands for VDR and reduce the growth of several prostate cancer cell lines in vitro (6,7,8) as well as tumor growth in vivo (9,10). 1,25D often acts as a differentiating agent in normal tissues, and it is likely that there are multiple targets of VDR that contribute to inhibition of tumor growth. In many cell lines, 1,25D treatment causes cells to arrest in the G₁ phase of the cell cycle (8,11,12,13). Although 1,25D regulates a number of proteins that control the G₁/S transition, our data suggest that one of the key actions of 1,25D is the down-regulation of c-Myc, which reduces levels of E2F (14) as well as the activity of cyclin-dependent kinases (15).

There have been no reports of the effects of VDR activity in prostate cancer cells overexpressing erythroblast transformation-specific (ETS) transcription factors as a result of somatic genomic translocations or deletions. These recently identified genomic rearrangements, in which coding regions of ETS factors [especially ETS variant 1 (ETV1) and ETS-related gene (ERG)] are placed under transcriptional control of the 5′ regulatory region of TMPRSS2, occur in up to 79% of prostate cancers (16). TMPRSS2 is an androgen-regulated serine protease and is highly expressed in normal and neoplastic prostate tissues (17,18). Its regulation thereby leads to overexpression and androgen-dependent regulation of ETS factors when fusion between the two genes occurs. ERG is a protooncogene that can transform NIH3T3 cells and induce tumorigenesis in mice (19). It is the most frequently overexpressed oncogene in prostate cancer (20).

Despite the frequency of ETS factor rearrangements in prostate cancer, they are not found in the most commonly studied prostate cancer cell lines. However, the androgen-dependent VCaP cell line contains a copy of a fusion of TMPRSS2 with ERG (TMPRSS2:ERG) as well as a normal copy of each gene (21). Depletion of ERG in VCaP cells reduces motility and invasiveness (22,23); some investigators also have reported reduced cell growth in vitro (23,24) and reduced tumor growth in vivo (23,24) as well as reduced c-Myc expression (24) when ERG is depleted. Overexpression of ERG in benign prostate cells increases invasiveness (22), and ERG expression in transgenic mice causes prostatic intraepithelial neoplasia (22,25). The frequency of these translocations and their biological characteristics suggest that they likely are an important factor in prostate cancer.

A microarray done in our lab (Yepuru, M., manuscript in preparation) shows that TMPRSS2 is up-regulated by EB1089 in the LNCaP prostate cancer cell line, raising the questions of whether activation of VDR also increases TMPRSS2:ERG expression and what the net effect of VDR action is in cells expressing this fusion gene. We report that VDR activation does induce the TMPRSS2:ERG gene, as well as ERG protein, but that despite this, it reduces VCaP cell growth and down-regulates c-Myc. However, another ERG target gene, CACNA1D, is induced by VDR, suggesting that inhibition of ERG targets is selective.

Materials and Methods

Cell culture and reagents

VCaP prostate cancer cells were obtained from Dr. Kenneth Pienta (University of Michigan, Ann Arbor, MI) and maintained in DMEM and 10% fetal calf serum (FCS). LNCaP and PC-3 prostate cancer cells were purchased from American Type Culture Collection (Manassas, VA). LNCaP cells were cultured in RPMI 1640 medium and 10% FCS. PC-3 cells were maintained in DME/F12 medium and 10% FCS. LAPC-4 cells (26) were obtained from Dr. Charles Sawyers (University of California, Los Angeles, Los Angeles, CA) and maintained in Iscove’s modified Dulbecco’s medium and 10% FCS. Dulbecco’s PBS, Hank’s buffered saline solution, and all cell culture media were purchased from Invitrogen (Carlsbad, CA). EB1089 was provided by LEO Pharmaceuticals (Ballerup, Denmark). 1,25D was obtained from Solvay Pharmaceuticals (Weesp, The Netherlands), methyltrienolone (R1881) was purchased from PerkinElmer Life Sciences, Inc. (Waltham, MA), and Casodex was purchased from LKT Laboratories, Inc. (St. Paul, MN). All other reagents were reagent grade unless otherwise indicated.

Quantitative RT-PCR

RNA was isolated using Trizol reagent (Invitrogen). cDNA was prepared from 300 ng mRNA with Superscript III reverse transcriptase (Invitrogen) and used for detection of E2F1, E2F2, E2F3, +72-bp ERG, −72-bp ERG, and ERG with SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). For detection of CYP24, TMPRSS2, TMPRSS2:ERG, IGFBP3, c-Myc, and CACNA1D, 200 ng mRNA template was used with the TaqMan One-Step PCR Master Mix (Applied Biosystems). All target genes were normalized to 18S (4319413E-0402011; Applied Biosystems) using TaqMan One-Step PCR Master Mix (for RNA templates) or TaqMan Universal Master Mix (for cDNA templates). See Table 1 for all primer and probe sequences.

Table 1.

Quantitative real-time primers and probes

mRNA	Sequence 5′→3′	NCBI accession no.
CYP24	CCCAGCGGCTGGAGATC	NM_000782 (variant 1),
	CCGTAGCCTTCTTTGCGG	NM_001128915 (variant 2)
	AACCGTGGAAGGCCTATCGCGACT
TMPRSS2	AGAATCGGTGTGTTCGCCTC	NM_001135099.1 (variant 1),
	CTCGTTCCAGTCGTCTTGGC	NM_005656.3 (variant 2)
	ACCAAACTTCATCCTTCAGGTGTACTCATCTCAGAG
TMPRSS2:ERG	CTGGAGCGCGGCAGGAA	NM_005656.3 (TMPRSS2 Variant 2),
TMPRSS2 exon 1, ERG exon 2	CCGTAGGCACACTCAAACAACGA	NM_182918.3 (ERG variant 1)
	TTATCAGTTGTGAGTGAGGAC
IGFBP3	GACAGAATATGGTCCCTGCCG	NM_001013398.1 (variant 1),
	TTGGAAGGGCGACACTGCT	NM_000598.4 (variant 2)
	ACACACTGAATCACCTGAAGTTCCTCAATGTGCT
c-Myc	AGCTGCTTAGACGCTGGATTTT	NM_002467.3
	GTTCCTGTTGGTGAAGCTAACGT
	AGCCTCCCGCGACGATGCC
E2F1	TCCAAGAACCACATCCAGTG	NM_005225.2
	CTGGGTCAACCCCTCAAG
E2F2	TGAAGGAGCTGATGAACACG	NM_004091.2
	TTAAAGTTGCCAACAGCACG
E2F3	ATATCCCTAAACCCGCTTCC	NM_001949.3
	TGGTCCTCAGTCTGCTGTAAGA
+72 bp	CTCTCACATCTCCACTAC	NM_182918.3 (variant 1),
	CTGGCCTAGTTGTAATTCTTTGC	NM_001136154.1 (variant 3),
		NM_001136155.1 (variant 4)
−72 bp	ACATTTGACTTCAGATGATGTTGATAAA	NM_004449.4 (variant 2)
	GGGCTCATATGGTAAATCTGTGTTT
ERG (exons 3 and 4)	CACGAACGAGCGCAGAGTT	All mentioned ± 72-bp variants
	ACTGCCGCACATGGTCTGTA
	TCGTGCCAGCAGATCCTACGCTATGG
CACNA1D	AGCCATCTCAAAATCCAAACTCA	NM_000720.2 (variant 1),
	CACGGCGGCCCTACATCT	NM_001128840.1 (variant 2),
	TGGCGTCGCTGGAACCGATTCA	NM_001128839.1 (variant 3)

Sets are listed as forward primer, reverse primer, and FAM/TAMRA probe. 

Western blot analysis

Cells were washed in PBS, and protein extracts were prepared by three freeze/thaw cycles in TESH lysis buffer [0.01 m Tris, 1 mm EDTA, 12 mm monothioglycerol (pH 7.7), and protease inhibitor cocktail] for detection of c-Myc and ERG or in 250 mm Tris (pH 7.4) and 0.4 m NaCl with protease inhibitors for detection of VDR. Lysates were run on SDS-PAGE and transferred to nitrocellulose membranes. Membranes were blocked for 1 h at room temperature with 1% milk-Tris-buffered saline with Tween 20 [TBST: 10 mm Tris (pH 7.5), 0.15 m NaCl, 1% Tween 20]. Membranes were then incubated with primary antibodies at 4 C overnight. All subsequent incubations and washes were done at room temperature. Primary rat monoclonal VDR antibody (MA1-710, Affinity Bioreagents, Golden, CO) was diluted 1:4000 in 1% milk-TBST. Primary rabbit polyclonal Erg-1/2/3 antibody (sc-354; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) was diluted 1:500 in 1% milk-TBST. Primary mouse monoclonal c-Myc antibody clone 9E-10 (11 667 149 001; Roche Applied Science, Indianapolis, IN) was diluted 1:80 in 1% milk-TBST. Primary mouse monoclonal actin antibody (Millipore, Billerica, MA) was diluted 1:10,000 in 1% milk-TBST. After primary antibody incubation, blots were washed three times with 1% milk-TBST. ERG blots were then incubated with horseradish peroxidase-conjugated donkey antirabbit IgG (Amersham Pharmacia Biotech, Piscataway, NJ) diluted 1:30,000 in TBST for 1 h. Actin blots were incubated with horseradish peroxidase-conjugated antimouse IgG for 1 h in TBST. VDR blots were incubated with a secondary antibody, rabbit antirat IgG (Zymed Laboratories, San Francisco, CA), and c-Myc blots were incubated with rabbit antimouse IgG (Zymed Laboratories) diluted 1:5000 in 1% milk-TBST for 1 h. After secondary antibody incubation, VDR and c-Myc blots were washed three times in 1% milk-TBST and then incubated for 1 h with horseradish peroxidase-conjugated donkey antirabbit IgG diluted 1:30,000 in TBST for 1 h. After horseradish peroxidase incubation, all blots were washed three times with TBST. Proteins were detected with ECL+ (GE Healthcare, Piscataway, NJ).

Cell growth analysis

VCaP cells were plated in 12-well plates at a density of 100,000 cells per well. Every 3 d, the medium was changed, and cells were treated with fresh vehicle (0.1% ethanol), EB1089, or 1,25D. Cells were washed with Hank’s buffered salt solution and then harvested with 0.25% trypsin. Cells were counted using a Coulter (Hialeah, FL) particle counter Z1.

Statistics

PASW Statistics 17.0 software was used to analyze statistical significance (P < 0.05). The unpaired t test was used to compare the means of two groups. One-way ANOVA was used to compare multiple means to a control group. Results are represented as means with error bars indicating sem. Unless otherwise indicated, all experiments were done a minimum of three times, with one representative experiment shown.

Results

VCaP cells express a functional VDR

To examine VDR action in the VCaP cell line, it was necessary to first determine whether VCaP cells express a functional VDR. The cells were treated with VDR agonist EB1089, and VDR expression and transcriptional activity were measured. VDR is expressed (Fig. 1A), and as has been reported for some other cell lines (27,28,29), agonist treatment increased expression. Moreover, EB1089 increased expression of the well characterized VDR target gene CYP24, indicating that the VDR is active (Fig. 1B). Thus, if VDR induces TMPRSS2, then it is likely that the TMPRSS2:ERG fusion also is induced by VDR in VCaP cells.

Figure 1.

VDR induces TMPRSS2:ERG in VCaP cells. A, VCaP cells were treated with ethanol (vehicle control) (E) or 10 nm EB1089 (EB) for 48 h. Cell extracts were prepared, and VDR and actin were detected by Western blotting. B, VCaP cells were treated with ethanol (E) or 10 nm EB1089 (EB) for 24 h, RNA was extracted, and CYP24 expression levels were measured by qPCR and normalized to 18S RNA (*, P = 0.005, significant effect of EB1089 vs. ethanol). C, PC-3 cells were treated as in B, and TMPRSS2 levels were measured by qPCR (*, P < 0.02, significant effect of EB1089 vs. ethanol). D, VCaP cells were treated with ethanol (E), 10 nm EB1089 (EB), or 10 nm R1881 (R) for 24 h. TMPRSS2 and TMPRSS2:ERG RNA levels were measured by qPCR (*, P ≤ 0.001, significant effect of EB1089 and R1881 vs. ethanol). E, VCaP cells were treated with ethanol (E), 10 nm EB1089 (EB), 10 nm 1,25D (10 D), or 100 nm 1,25D (100 D) for 24 h. TMPRSS2:ERG RNA levels were measured by qPCR (*, P < 0.01, significant effect of treatment vs. EtOH). F, VCaP cells were treated as in E, and ERG and actin protein were detected by Western blotting. G, VCaP cells were treated with ethanol (EtOH) or 10 nm EB1089 for the indicated times. TMPRSS2 and TMPRSS2:ERG levels were measured by qPCR (*, P < 0.025, significant effect of EB1089 vs. EtOH). Bars, sem.

VDR agonists increase TMPRSS2:ERG expression

To verify the EB1089-mediated TMPRSS2 induction seen in our microarray, prostate cancer cells were treated with EB1089 and TMPRSS2 levels were measured. EB1089 induced TMPRSS2 in androgen receptor (AR)-negative PC-3 cells (Fig. 1C) as well as in AR-positive LNCaP cells (Fig. 2A). To evaluate VDR-mediated regulation of TMPRSS2:ERG, expression levels of TMPRSS2 and TMPRSS2:ERG RNA in VCaP cells were examined after EB1089 treatment. EB1089 increased both TMPRSS2 and TMPRSS2:ERG to about the same extent as the synthetic androgen, R1881 (Fig. 1D). As expected, the natural VDR ligand, 1,25D, also increased TMPRSS2:ERG levels (Fig. 1E). In addition to increasing TMPRSS2:ERG mRNA expression, EB1089 and 1,25D also increased ERG protein expression, as detected by Western blotting (Fig. 1F). A time course of EB1089-mediated TMPRSS2 and TMPRSS2:ERG induction reflects a time-dependent increase in gene expression, beginning within a few hours of treatment (Fig. 1G).

Figure 2.

Casodex inhibits EB1089-dependent TMPRSS2 induction in LNCaP cells but not in VCaP cells. A, LNCaP cells were treated for 24 h with ethanol, 10 nm EB1089, 10 μm Casodex, or both EB1089 and Casodex. RNA was extracted, and IGFBP3 and TMPRSS2 expression levels were measured by qPCR and normalized to 18S RNA (*, P < 0.05, significant effect of Casodex plus EB1089 vs. EB1089 alone). This experiment was performed twice with EB1089 and once with 1,25D. Similar results were obtained with both compounds. B, VCaP cells were treated for 24 h with ethanol, 10 nm EB1089, 0.5 nm R1881, 10 μm Casodex, or the indicated combinations. TMPRSS2 and TMPRSS2:ERG levels were measured by qPCR (*, P < 0.025, significant effect of Casodex plus R1881 vs. R1881 alone). C, LNCaP cells were treated for 24 h with ethanol, 10 nm EB1089, 10 μm Casodex, or both EB1089 and Casodex, and CYP24 levels were measured by qPCR (*, P < 3 × 10⁻⁹, significant effect of Casodex plus EB1089 vs. EB1089 alone). D, VCaP cells were treated and CYP24 levels were measured as in C (*, P = 0.002, significant effect of Casodex plus EB1089 vs. EB1089 alone). E, VCaP cells were cultured in 10% charcoal-stripped FCS and treated for 24 h with ethanol or 10 nm EB1089. TMPRSS2:ERG levels were measured by qPCR (*, P = 0.004, significant effect of EB1089 vs. ethanol). Bars, sem.

Antagonist-bound AR does not prevent VDR-mediated transcription of TMPRSS2:ERG

Although many of the effects of VDR activation are beneficial, presumably induction of TMPRSS2:ERG is not, and we therefore wanted to determine whether we could prevent it. Our lab and others have shown that for some genes that are regulated by both AR and VDR, treating LNCaP prostate cancer cells with an AR antagonist can reduce VDR-mediated transcription (30). Thus, it is possible that an AR antagonist could prevent EB1089-mediated induction of AR/VDR-regulated TMPRSS2:ERG. The IGFBP3 gene has been reported to be induced by both VDR and AR, and binding sites for both receptors have been identified in the promoter (31,32). In LNCaP cells, the AR antagonist Casodex reduced EB1089-mediated IGFBP3 induction (Fig. 2A). Casodex also reduced EB1089-mediated TMPRSS2 induction in LNCaP cells (Fig. 2A). However, although Casodex inhibited R1881-dependent induction of TMPRSS2 and TMPRSS2:ERG in VCaP cells (Fig. 2B), it had no effect on EB1089-dependent induction (Fig. 2B), suggesting that this activity of Casodex is limited to specific cell lines. As a control to demonstrate that Casodex is not a universal inhibitor of VDR action in LNCaP cells, we measured expression of CYP24, a well-characterized VDR target gene. Surprisingly, Casodex strongly stimulated EB1089- dependent expression of CYP24 in LNCaP cells (Fig. 2C) and had a modest stimulatory effect in VCaP cells (Fig. 2D), but VDR activity clearly was not inhibited in either case.

VDR and AR also can cooperate to induce common target genes (30,31,33). To determine whether EB1089-mediated TMPRSS2:ERG induction occurs in the absence of androgens, gene regulation was examined in androgen-depleted charcoal-stripped serum. Under these conditions, EB1089 induced TMPRSS2:ERG (Fig. 2E). EB1089 also induced TMPRSS2 in AR-negative PC-3 prostate cancer cells (Fig. 1C). Therefore, EB1089 can induce TMPRSS2:ERG in an androgen-depleted environment.

VDR agonists inhibit growth of VCaP cells

VDR activation typically is growth inhibitory in prostate cancer cell lines that do not contain the TMPRSS2:ERG rearrangement. However, ERG has been shown to contribute to in vitro and in vivo growth of VCaP cells by two groups (23,24). To evaluate the effects of EB1089 on proliferation of VCaP cells, the cells were treated with EB1089 or 1,25D, and cell number was measured as a function of time of treatment. The VDR agonists reduced the rate of VCaP cell growth (Fig. 3A), although not nearly as effectively as in some lines such as the LNCaP/C4-2 lineage of prostate cancer cells (34). Our lab has shown that reduction of c-Myc is a major factor in VDR agonist-mediated growth inhibition of C4-2 cells (14), and c-Myc is a primary target of ERG in VCaP cells (24). Therefore, we measured the effects of EB1089 on c-Myc expression in VCaP cells and found that EB1089 reduced both c-Myc RNA and protein levels (Fig. 3, B and C). One of the activities of c-Myc in stimulating growth is up-regulation of E2F transcription factors, which are necessary for progression to S phase of the cell cycle (35). The EB1089-dependent down-regulation of c-Myc is sufficient to reduce expression of E2Fs 1, 2, and 3 in VCaP cells (Fig. 3D).

Figure 3.

VDR agonists reduce VCaP cell growth and expression of cMyc and E2F. A, Cells were treated with ethanol (EtOH), 10 nm EB1089, 10 nm 1,25D, or 100 nm 1,25D for the indicated times and then counted with a Coulter counter (*, P < 0.001, significant effect of treatment vs. EtOH). B, Cells were treated with EtOH or 10 nm EB1089, RNA was extracted at 24 and 48 h, and c-Myc RNA levels measured by qPCR (*, P = 0.001, significant effect of EB1089 vs. EtOH). C, Cells were treated with EtOH or 10 nm EB1089 for 72 h, and cell lysates were prepared and analyzed for c-Myc and actin protein by Western blotting. D, At 48 h after EtOH or 10 nm EB1089 treatment, expression levels of E2Fs 1, 2, and 3 were measured by qPCR (*, P < 0.05, significant effect of EB1089 vs. EtOH). Bars, sem.

This finding suggests several possibilities. First, there may be so much ERG expressed in medium containing FCS (which contains endogenous androgens) that the additional increase in ERG in response to VDR activation does not alter ERG target gene expression, and other VDR actions overcome the growth-stimulatory effects of overexpressed ERG. Second, although we have shown that VDR agonists induce expression of ERG, they may be inducing a splice variant that does not affect cell growth. Third, despite induction of ERG, other VDR actions dominantly counteract the increase in ERG expression.

EB1089 up-regulates some ERG target genes

To examine whether the naturally overexpressed levels of ERG are saturating, such that increased expression of ERG by EB1089 causes no increase in total ERG activity, the expression of several ERG target genes was evaluated. Whereas some ERG target genes, such as KCNS3 (potassium voltage-gated channel, delayed-rectifier, subfamily S, member 3) and PLA1A (phospholipase A1 member A) (22,23) were variably regulated by EB1089 treatment of VCaP cells (data not shown), the ERG target gene CACNA1D (calcium channel, voltage-dependent, L type, α-1D subunit) (22,23) was consistently and significantly up-regulated (Fig. 4). This gene was not induced by EB1089 in LAPC-4 cells, which do not express TMPRSS2:ERG (Fig. 4). This suggests that EB1089-dependent induction of ERG can cause an increase in ERG transcriptional activity.

Figure 4.

EB1089 treatment causes induction of an ERG target gene in VCaP cells. VCaP and LAPC-4 cells were treated with ethanol (EtOH) or 10 nm EB1089 for 48 h. After RNA extraction, CACNA1D levels were measured by qPCR (*, P = 0.008, significant effect of EB1089 vs. EtOH). Bars, sem.

VDR does not preferentially induce the +72-bp or −72-bp isoform of ERG

ERG coding exons are heterogeneous, and some splice variants have varying levels of effects on prostate cell behavior. In particular, the presence of a variably expressed 72-bp exon causes increased growth, motility, and invasiveness (23). To determine whether VDR preferentially induces either the +72-bp or −72-bp variant of ERG, quantitative RT-PCR (qPCR) analysis was done using primers that would specifically include or exclude this exon. EB1089 up-regulated the two isoforms to similar degrees (Fig. 5, A and B), indicating that VDR does not alter splicing to cause preferential expression of the less growth stimulatory form.

Figure 5.

EB1089 does not preferentially induce the ERG isoform that either contains or excludes the 72-bp exon. A, VCaP cells were treated with ethanol (E) or 10 nm EB1089 (EB) for 24 h, RNA extracted, and ERG isoforms measured by qPCR (*, P < 0.02, significant effect of EB1089 vs. ethanol). B, Shown is the fold induction of the isoforms relative to the level in ethanol (EtOH)-treated cells. Bars, sem.

Discussion

Studies in the last few years have shown that a high proportion of prostate cancers contain genomic rearrangements that link the promoter of the TMPRSS2 gene to a portion of the coding region of an oncogenic ETS factor, most commonly ERG (22,36). The resulting overexpressed protein stimulates cell growth and invasiveness in vitro and tumor growth in vivo (22,23,24). Despite the frequency of these rearrangements, none of the most commonly used prostate cancer cell lines express the gene fusions, and conclusions regarding the response of prostate cancer to various treatments, including VDR agonists, have been made from studies of cells and animal models lacking the rearrangements. In the course of identifying VDR target genes in prostate cancer cell lines that are growth-inhibited by VDR agonists, we found that one target gene is TMPRSS2. We asked whether the TMPRSS2:ERG fusion gene might also be induced and what the response of cells expressing TMPRSS2:ERG would be to VDR agonist treatment. Using the VCaP cell line, which contains a copy of the TMPRSS2:ERG rearrangement, we found that VDR agonists 1,25D and EB1089 increase expression of TMPRSS2:ERG mRNA and ERG protein. The levels of TMPRSS2:ERG are not saturating, as demonstrated by the induction of the ERG target gene CACNA1D by EB1089. However CACNA1D is not induced as much as ERG, and other target genes show modest to no induction. When VCaP cells are grown in FCS, which contains endogenous androgens, ERG levels are approximately 2000-fold higher than in LNCaP cells, which do not express TMPRSS2:ERG (16). Therefore, the level of ERG in VCaP cells may be close to optimal for ERG target gene expression. It is possible that higher levels of ERG are required for optimal induction of CACNA1D, although we cannot exclude that other factors limit further induction of some ERG target genes.

We found that VDR agonist treatment reduces the growth of VCaP cells despite increasing expression of TMPRSS2:ERG. Sun et al. (24) found that ERG is required for expression of c-Myc in VCaP cells. We had shown previously that VDR agonists decrease c-Myc in other prostate cancer cell lines (14,37) and that decreasing c-Myc expression in C4-2 cells using small interfering RNA is sufficient to mimic VDR-mediated growth inhibition (14). Although EB1089 induces ERG, it reduces c-Myc levels in VCaP cells (Fig. 3). VDR has been reported to regulate c-Myc transcription, elongation rate, mRNA stability, and protein stability in various cell types (14,38,39,40,41). One or more of these mechanisms likely is responsible for the reduction of c-Myc in VCaP cells. Among the possible mechanisms for regulation of c-Myc transcription is altered growth factor signaling. TGFβ is known to inhibit c-Myc expression, and 1,25D and EB1089 induce TGFβ in some prostate and breast cancer cells, contributing to VDR agonist-mediated growth inhibition (42,43). Another possible mechanism is induction of a competing ETS factor that is unable to induce c-Myc. In our unpublished microarrays, we found that EB1089 induces PDEF (prostate-derived ETS factor). Overexpression of PDEF in colon cancer cells inhibits growth, causes a G₁ arrest, and reduces migration (44). It will be important, in the future, to determine the mechanisms by which VDR reduces c-Myc expression in the TMPRSS2:ERG-expressing VCaP cells because the pathways may not be active in all TMPRSS2:ERG-expressing prostate cancers.

In VCaP cells, one possible explanation for the ability of VDR agonists to reduce growth and c-Myc expression was that they could alter splicing of TMPRSS2:ERG to yield a preferential increase in the −72-bp form of ERG, with a reduction in the +72-bp form, which appears to be more growth stimulatory (23) and thus may be more effective in inducing c-Myc. However, the −72-bp form is not preferentially induced by EB1089, and there is no reduction in the more aggressive form (+72 bp), so this does not contribute to the ability of VDR agonists to override the growth-stimulatory effects of ERG overexpression. It is important to note that VCaP cells were less growth inhibited relative to some other prostate cancer cell lines (34,37). Whereas in LNCaP and C4-2 cells, after three to four doublings, 1,25D reduces growth up to 90% and the cells stop increasing in number (34), in VCaP cells, 1,25D reduces growth only up to 50% after the same number of doublings and the cell number continues to increase. Therefore, there may be other factors, including ERG, in the VCaP cell line that reduce its ability to be growth inhibited by VDR agonists.

One of the main activities of ERG is regulation of cell motility/invasiveness. ERG overexpression in normal prostate cells increases invasion and motility (22,23), and ERG levels correlate with expression levels of genes involved in the plasminogen activator pathway, which plays a role in invasion (22,25). Previous studies show that VDR agonists reduce the motility and invasiveness of some prostate cancer cell lines (45,46,47). In these lines, 1,25D acts to inhibit the matrix metalloproteinase protease pathway and cathespin pathway and not the plasminogen activator pathway (46), so it will be important to determine the effects of VDR agonists on motility and invasiveness in the context of fusion gene expression.

In addition to factors that affect growth in vitro, there are other factors that contribute to growth in vivo including those that induce angiogenesis. There is some evidence that VDR agonists can reduce angiogenesis both by decreasing the expression of angiogenic factors in tumor cells (48) as well as by direct effects on tumor endothelial cells (49). VCaP cells form tumors in mice, and when ERG is knocked down, tumors grow at a slower rate (23). Transgenic expression of ERG in prostate induces prostatic intraepithelial neoplasia in mice (22). Whether EB1089 treatment can overcome the effects of increased ERG expression in vivo remains to be determined.

There is evidence that in some prostate cancer cell lines, there is functional cross talk between the AR and VDR to enhance their activities (30,33) on genes regulated by both receptors. 1,25D and R1881 cooperatively cause maximal induction of the AR/VDR-regulated IGFBP3 promoter (31), and the AR antagonist Casodex reduces the ability of 1,25D to induce expression of the AR/VDR-regulated AS3 gene (30) and to inhibit growth of LNCaP cells (30,33). We found that Casodex reduces EB1089-dependent induction of the IGFBP3 and TMPRSS2 genes in LNCaP cells. However, although Casodex reduced R1881-mediated TMPRSS2 and TMPRSS2:ERG induction in VCaP cells, it was ineffective in reducing EB1089-mediated induction. We assume that the TMPRSS2 regulatory region is the same in both cell lines [two AR binding sites, one strong site ∼13.5 kb upstream of the transcriptional start site and a weaker site in the promoter region (50), and unidentified VDR-responsive elements], but there may be other cellular factors contributing to the specificity of this effect. In LNCaP cells, Casodex does not reduce VDR-mediated transcription of a VDR-responsive reporter that does not contain AR binding sites (30), and Casodex does not reduce VDR-dependent induction of CYP24 in our study. In fact, we see a super-induction of CYP24 by the combination of Casodex and EB1089 in the LNCaP cells and a more modest increase in induction in VCaP cells. Lou et al. (51) have reported that the AR agonist dihydrotestosterone reduces VDR-mediated CYP24 expression in LNCaP cells, although the mechanism is not known. Although they observed no effect of Casodex on VDR-mediated expression, the study was performed in medium depleted of androgens, whereas our study used FCS containing endogenous androgens. It is possible that the addition of Casodex in our study relieved some of the repression caused by androgens in the FCS, thereby causing the super-induction of CYP24 that we observe.

In summary, our studies show that VDR induces TMPRSS2 and the TMRPSS2:ERG fusion gene and that neither androgen nor AR are required for induction. Although EB1089 and 1,25D are modestly growth inhibitory in VCaP cells, the effects of VDR agonists on other ERG-dependent activities remain to be determined. In particular, the consequences of VDR activation under conditions of androgen depletion, where VDR-mediated induction of the TMPRSS2-regulated ETS factors may play a greater role, should be examined.