The Rec protein of HERV-K(HML-2) upregulates androgen receptor activity by binding to the human small glutamine-rich tetratricopeptide repeat protein (hSGT)

Authors


Abstract

The expression of endogenous retroviruses of the HERV-K(HML-2) family is strongly upregulated in germ cell tumors and several other cancers. Although the accessory Rec protein of HERV-K(HML-2) has been shown to induce carcinoma in situ in transgenic mice, to increase the activity of c-myc and to interact with the androgen receptor (AR), whether or not Rec expression is indeed implicated causally in the initiation or progression of any human malignancies remains unclear. We used the yeast two-hybrid system involving the Rec protein of a recently integrated HERV-K(HML-2) element in an effort to identify potential Rec-related oncogenic mechanisms. This revealed the human small glutamine-rich tetratricopeptide repeat (TPR)-containing protein (hSGT) to be a cellular binding partner. The interaction of Rec with this known negative regulator of the AR was confirmed by coimmunoprecipitation, pull-down assays and colocalization studies. The interaction involves the TPR motif of hSGT and takes place in the cytoplasm and in the nucleoli. Using an AR-responsive promoter and gene we could demonstrate that Rec interference with hSGT resulted in an up to five-fold increase in the activity of AR. Furthermore, in AR positive cells, Rec was shown to act as transactivator by enhancing AR-mediated activation of the HERV-K(HML-2) LTR promoter. This is in line with previous observations of elevated HERV-K(HML-2) expression in steroid-regulated tissues. On the basis of our findings we propose a “vicious cycle” model of Rec-driven hyperactivation of the AR leading to increased cell proliferation, inhibition of apoptosis and eventually to tumor induction or promotion.

Retroviruses have been known for over a century to be associated with tumor diseases. Among human endogenous retroviruses (HERVs) the HERV-K(HML-2) subfamily is the most likely candidate for a retrovirus with oncogenic properties.1 This subfamily comprises the youngest and best preserved HERVs in the human genome. Although all known HERVs accumulated mutations and deletions after integration, several HERV-K(HML-2) members retained genes coding for intact and functionally active proteins, such as the HERV-K108 Env.2, 3

In general, HERV-K(HML-2) expression is suppressed in healthy tissues by epigenetic mechanisms.4 However, a low level of expression can be detected in some tissues such as testis cells, placenta and other steroid hormone-regulated tissues.5, 6 In malignant tissues such as germ cell tumors (GCT), melanomas or ovarian cancers the expression rate is considerably higher.5, 7–9 Breast cancer tissues, for example, were found to have elevated levels of HERV-K(HML-2) env RNA and gag RNA was detected in the peripheral blood cells of leukemia patients as well as in breast, prostate and ovarian cancers.5, 10–12

There are numerous cases in which HERV-K(HML-2) elements were shown to be involved in the malignant transformation of cells by recombination or downstream promotion.13, 14 Moreover, in recent years evidence has accumulated that suggests an association between the accessory HERV-K(HML-2) protein Rec with tumor development. The 14.5 kDa Rec protein is translated from doubly spliced transcripts and mediates the transport of unspliced or partially spliced HERV-K(HML-2) mRNAs from the nucleus into the cytoplasm.15 Rec expression might have a direct role in transforming cells into a pre-cancerous state. In 2005, Galli et al. showed that germ cell development in transgenic mice expressing Rec is disturbed and that some animals develop lesions similar to the predecessor lesions of classical seminomas in humans.16 However, the molecular mechanisms and pathways involved remain largely elusive.

The cellular interaction partners of Rec are also poorly defined, although some have recently been identified. Rec interacts with the promyelocytic leukemia zinc finger protein (PLZF) that is essential for the self-renewal of spermatogonial stem cells in mice and which appears to be also involved in certain human leukemias.17, 18 Transient coexpression of Rec or Np9, another HERV-K(HML-2) protein which is linked to tumorigenesis, with PLZF interferes with its function as a negative regulator of the c-myc proto-oncogene and leads to cell proliferation and apoptosis.18 Recently, Kaufmann et al. reported that Rec forms trimeric complexes with the testicular zinc-finger protein (TZFP) and the androgen receptor (AR) providing a further link to the steroid hormone system.19

In this report we describe the identification of a novel Rec-interacting partner: the human small glutamine-rich tetratricopeptide repeat-containing protein (hSGT) that controls mitotic processes and is a checkpoint protein during prometaphase.20 On the other hand, hSGT has also been described to control as a cochaperone the activity of the chaperones Hsp70 and Hsp90. Furthermore, the cochaperone is involved in the synthesis, assembly, maturation and regulation of activity of steroid receptors such as the androgen receptor (AR).21–23

We show here that the HERV-K(HML-2) Rec protein binds to hSGT and interferes with its role as a negative regulator of the AR, resulting in enhanced AR activity. Furthermore, we demonstrate that HERV-K(HML-2) elements profit from this enhanced AR activity, leading to a “vicious cycle” that could result in increased cell proliferation and inhibition of apoptosis, and eventually lead to tumorigenesis.

Material and Methods

Constructs and RT-PCRs

The pcDNA3.1_wtRec-V5 construct has been described previously.24 The codon-optimized Rec (coRec) was synthesized by fusion PCR using overlapping primers and labeled with a V5-tag (see sequence in Supporting Information Fig. S1). To generate pcDNA3_oricoRec-V5, the coRec-V5 sequence was aligned with 10 HERV-K(HML-2) sequences and postinsertional mutations were identified as previously described.25 Multisite mutagenesis (Stratagene) was applied to generate the T89A and D96E mutations using appropriate mutagenesis primers. For yeast two hybrid screens, wtRec and oricoRec were cloned in-frame to the LEX-A gene in pEG202 (Origene). Two-hybrid experiments were performed as already described elsewhere.26 OricoRec-V5 was expressed in E. coli BL21DE3 using a pET16b (Novagen) construct expressing oricoRec-V5 with an N-terminal His6-Tag. pET15_hSGT was a kind gift of Chung Wang (Academia Sinica, Taipei, Taiwan, ROC).27 Hsgt was cloned into the FLAG-tagged pCMV-Tag2B expression vector (Stratagene). For pull-down experiments and production of recombinant proteins hsgt was cloned either into pGex-5x-1 (Amersham Bioscience) or pET16b (Novagen). GST and Ni-NTA pull-down assays were performed as described previously.26

The pCMV-AR3.1 and ARR3-tk-Luc expression plasmids were a generous gift of Wayne Tilley (University of Adelaide).21, 28 The HERV-K(HML-2), KoRV and HIV-1 LTR reporter constructs have been described previously.24 For quantitative real time RT-PCRs we used the primer pair PSA_For: 5′GGCAAGCATTGAACCAGAGGAG and PSA_Rev: 5′GCA TGAACTTGGTCACCTTCT together with 5′-[6FAM]ATGA CGTGTGTGCGCAAGTTCACCC[BHQ1] as probe.

Transfection

Cells in 96- to 6-well dishes were transfected using Effectene or Polyfect Transfection Reagent (Qiagen) whereas transfections in 10-cm dishes were performed using calcium phosphate. For virus particle production Tera-1 cells were grown in 10-cm dishes and transfected with 24 μg oriHERV-K113, 8 μg AR expression construct, 8 μg Rec and 8 μg hSGT or empty control vector. Cells were cultured for 5 days in the presence of 0.1 nM DHT. Ultracentrifugation was performed as described elsewhere.25

Antibodies

Immunoblot and immunofluorescence analyses were carried out using mouse αV5 (Serotec), mouse αV5-HRP (Invitrogen), rabbit αFLAG (Sigma Aldrich), rabbit αGST (Sigma Aldrich) and αnucleolin (Abcam) as primary antibodies. For androgen receptor staining we used an αAR antibody (Epitomics). As secondary antibodies αmouseIgG-HRP (Sigma Aldrich), αrabbitIgG-HRP (Sigma Aldrich) and αrabbitIgG-Alexa647 (Invitrogen) were used. The αGag antibodies have been described elsewhere.29 Agarose covalently conjugated with αV5 antibody (αV5 sepharose, Sigma Aldrich) or with αFLAG antibody (αFLAG sepharose, Sigma Aldrich) was used for coimmunoprecipitations. Protein purification of His-tagged and GST-tagged proteins was performed using Ni-NTA Sepharose (Qiagen) or Glutathione-Sepharose 4B (GE Healthcare).

Microscopy

Confocal microscopy with a laser scanning microscope cLSM780 (Zeiss) was carried out on HEK 293T and LNCaP cells (200,000 cells per well) transfected for 24 h with oricoRec-pEGFP (100 ng/well) and hSGT-FLAG (100 ng/well). Cells were fixed with 2.5% PFA and stained as described previously.24

Transactivation assays

For AR transactivation assays LNCaP and DU145 cells were seeded in 24-well plates and transfected as indicated with 0.1 ng AR, 25 ng ARR3-tk-luc and up to 250 ng oricoRec-V5 and/or 50 ng hSGT-FLAG or the appropriate empty control vectors to maintain transfection conditions. About 2 ng pGL4.74 (renilla luciferase, Promega) was cotransfected for normalization and pGL-SV40 (Promega) was used in control experiments. HSGT titration experiments were performed in 96-well plates using 25 ng ARR3-tk-Luc, up to 50 ng Rec and hSGT expression plasmids or appropriate amounts of empty control vectors. Unless otherwise indicated, cells were grown in the presence of 0.1 nM DHT dissolved in ethanol. In experiments addressing the androgen responsiveness of retroviral LTRs the ARR3-tk-luc construct was replaced with 25 ng pGL3-LTR-Firefly constructs as described previously.24 All experiments were performed in 24-well plates and in the presence of 0.1 nM DHT unless stated otherwise. At 24 h post transfection cells were analyzed for luciferase activity using a dual-luciferase reporter assay system (Promega). Values given represent the means and standard deviations of six independent samples and each experiment was repeated at least three times with similar results. For statistical analysis we used the program GraphPad Prism 5 applying the one-way ANOVA analysis followed by Bonferroni multiple comparison test to estimate statistical significance of the different samples.

Cavidi RT activity assay

Supernatants of transfected cells were filtered through 45-μm membranes and assayed for RT activity using the CAVIDI HS-kit Mg2+ RT (Cavidi Tech AB, Uppsala, Sweden) according to the manufacturer's instructions. Each sample was measured in triplicate and values represent the means of six independent transfections.

Abbreviations

AR: androgen receptor; ARE: androgen responsive element; coRec: codon-optimized Rec; DHT: 5α-dihydrotestosterone; HERV-K: human endogenous retrovirus-K; HML-2: human MMTV-like 2; IB: immunoblot; IP: immunoprecipitation; MMTV: mouse mammary tumor virus; oricoRec: original codon-optimized Rec; PLZF: promyelocytic leukemia zinc finger protein; RLU: relative light units

Results

Reconstitution of the original HERV-K113 Rec sequence and identification of hSGT as a novel Rec interacting protein

Sequence analyses revealed that the codon usage of the gene coding for the HERV-K(HML-2) wtRec protein is not optimal for high expression in mammalian cells. Even CMV promoter driven expression of wtRec in transfected cells was low (Fig. 1a). The gene was therefore codon-optimized and this coRec gene was shown to be expressed at levels ∼10-fold higher compared to wtRec (Fig. 1a). Furthermore, the HERV-K113 Rec amino acid sequence was compared with those of other recently integrated HERV-K(HML-2) members to identify postinsertional mutations. The Rec sequences were surprisingly well preserved and only two amino acids in the second exon of the HERV-K113 ORF being exclusive for this element (and therefore most likely representing postinsertional mutations3, 25, 29, 30) could be identified (Fig. 1b). We reconstituted the sequence of the presumed original HERV-K113 Rec (albeit in a codon-optimized form) by introducing T89A and D96E substitutions into the coRec sequence. The resulting oricoRec amino acid sequence was therefore identical to the consensus Rec sequence (Fig. 1b). There was no apparent difference in the expression levels of coRec and oricoRec (Fig. 1c) and the oricoRec protein was located primarily in the nucleoli of the nucleus with an additional low diffuse cytoplasmic expression (Fig. 1d). A similar pattern of localization has been described previously for the wtRec protein of this particular provirus.24

Figure 1.

Codon optimization and reconstitution of the HERV-K(HML-2) Rec protein. (a) Immunoblot of HEK 293T cells transfected with wtRec-V5 or coRec-V5 expression plasmids. Monomers, dimers and trimers are indicated by arrows. (b) Alignment of amino acid sequences from 10 HERV-K(HML-2) members to identify postinsertional mutations. Two HERV-K113-exclusive postinsertional mutations (T89A and D96E) were reversed creating oricoRec. (c) Immunoblot of HEK 293T cells transfected with oricoRec-V5 and coRec-V5. (d) Analysis of the subcellular localization of oricoRec-GFP. The nucleoli were stained with an antibody directed against nucleolin. The nuclei were stained with DAPI.

To address the question of how the HERV-K(HML-2) Rec protein might contribute to tumor induction or progression, we performed yeast two-hybrid interaction studies to identify new host proteins that bind directly to Rec. Two versions of the HERV-K113 Rec protein were used (wtRec and oricoRec) and screened against a human spleen library. The spleen is highly enriched with lymphocytes, making the library suitable for screening as HERV-K(HML-2) expression is often enhanced in patients with leukemia or lymphomas.31 Fifteen of the 45 positive clones isolated had two open reading frames coding for hSGT: one spanning amino acids (aa) 34 to 237 and the second spanning aa 137 to 237 (Fig. 2a). These results imply that the interacting region is within the main functional domain of hSGT, the tetratricopeptide repeat-containing domain (TPR domain). In addition, staufen-1 has been identified as a potentially specific interaction partner of Rec (manuscript in preparation).

Figure 2.

HERV-K(HML-2) Rec and hSGT interact with each other in both the cytoplasm and the nucleus via the TPR domain of hSGT. (a) Schematic overview of sequences isolated after positive interaction in a yeast two-hybrid screen with Rec as bait. (b) Confirmation of interaction by coimmunoprecipitation. HEK 293T cells were transfected with either Rec-V5 and hSGT-FLAG or with hSGT-FLAG and empty pcDNA3 control vector. Coprecipitated hSGT was detected by an αFLAG antibody. (c) Colocalization studies of Rec-GFP and hSGT-FLAG stained with an αFLAG antibody and a secondary Alexa647 antibody. (d) Identification of the TPR motif of hSGT as region interacting with Rec by pull-down analysis. His6-tagged oricoRec-V5 and GST-tagged hSGT or TPR were incubated with Ni-NTA or αGST sepharose as indicated. (*unspecific band, ** unspecific cleavage product from GST-tagged hSGT and TPR).

Confirmation of the Rec-hSGT interaction in human cells

The interaction between HERV-K(HML-2) Rec and hSGT in mammalian cells (HEK 293T) was then confirmed by coimmunoprecipitation analysis and subcellular colocalization experiments using the oricoRec construct. Using αV5 agarose to bind the V5-tag at the Rec C-terminus it was possible to coprecipitate N-terminally FLAG-tagged hSGT (Fig. 2b). That hSGT precipitation depends on Rec-V5 expression was shown by the lack of FLAG-hSGT protein when the empty pcDNA3 control vector was cotransfected in place of Rec-V5 (Fig. 2b). Similar results were also obtained in the reverse experiment in which Rec-V5 was coimmunoprecipitated by FLAG-hSGT (data not shown). Cellular hSGT as well as overexpressed hSGT is located mainly in the cytoplasm with an additional diffuse nuclear signal (Fig. 2c). Coimmunofluorescence experiments with Rec and hSGT revealed a strong colocalization of Rec and hSGT in the cytoplasm as well as in the nucleus with a substantial enrichment in the nucleoli (Fig. 2c). The immunoprecipitation and subcellular localization experiments therefore provide strong evidence for the presence of Rec-hSGT complexes in both the cytoplasm and the nucleus, especially the nucleoli. The interaction with Rec appears to shift hSGT into the nucleus.

The central TPR motif of hSGT interacts with HERV-K(HML-2) Rec

The region of hSGT involved in the interaction with Rec was investigated using pull-down experiments with purified proteins expressed in E. coli. Both the full-length hSGT-GST protein and a deletion mutant (TPR-GST) comprising only the TPR domain of hSGT could be precipitated using Ni-NTA sepharose to bind the His6-Rec-V5-protein (Fig. 2d). In addition, the Rec-V5 protein was brought down by anti-GST sepharose in the presence of hSGT-GST and TPR-GST. This implies, together with results from the yeast two-hybrid screen that interaction occurs via the TPR motif of the cochaperone. A control experiment using a TPR deletion mutant of hSGT revealed that no other sites in the hSGT protein are involved in binding to Rec (Supporting Information Fig. S2).

HERV-K(HML-2) Rec inhibits the androgen receptor negative-regulator function of hSGT

hSGT functions as a cochaperone regulating the synthesis, maturation and activity of the androgen receptor (AR).21 The protein prevents the recruitment of the AR into the nucleus in the absence of ligands such as testosterone or its derivate 5α-dihydrotestosterone (DHT).21 In many tissues, activation of the androgen receptor results in an enhancement of proliferation and an inhibition of apoptosis—processes both well known to be deregulated in tumors.32, 33

We therefore investigated whether Rec has an influence on AR activity and whether this is mediated by binding to hSGT. A reporter construct comprising the probasin gene promoter containing an androgen-responsive element (ARE) in front of a firefly luciferase gene28 was used to measure AR activity. This AR-dependent reporter construct was transfected together with increasing amounts of Rec into LNCaP cells, a prostate cancer cell line expressing a functional AR. Increasing the amount of Rec transfected increased the probasin promoter activity by a factor of up to 3.5 compared to cells transfected with the empty control vector (Fig. 3a). In contrast, the AR-deficient prostate cancer cell line DU145 did not show enhancement of luciferase activity by Rec (Fig. 3b). Low ectopic expression of AR alone did not result in enhanced promoter activity, possibly because endogenous inhibitors such as hSGT keep the AR in an inactive state. Rec coexpression, however, lead to a fivefold enhancement of promoter function, presumably by competing with AR for binding to hSGT (Fig. 3b). These results support the hypothesis that Rec induced probasin promoter activation is mediated by the AR. Expression of an analogous luciferase construct carrying an androgen-unresponsive SV-40 promoter in place of the AR dependent probasin promoter was not enhanced in the presence of Rec, AR or Rec and AR. Instead, expressions of Rec lead to a slight decrease in the SV-40 promoter activity (Fig. 3c). The Rec-induced enhancement of AR activity therefore appears to depend on the AR-responsive element in the gene promoter.

Figure 3.

Rec modulates the activity of the androgen receptor (a) Modulation of AR basal activity in AR-positive LNCaP cells by expression of increasing amounts of Rec. Intrinsic AR activation was measured with a probasin promoter-driven luciferase reporter vector. (b) DU145 cells lacking a functional AR were transfected with 0.1 ng AR expression construct or empty control vector and with 50 ng Rec or empty pcDNA3. Luciferase activity from control cells expressing neither AR nor Rec was set at 100%. (c) Luciferase activation by Rec and AR is probasin promoter specific. LNCaP cells were transfected with the AR-independent SV40-luciferase reporter construct and cotransfected with AR, Rec or the appropriate empty control vectors. All values were calculated relative to those obtained from cells transfected with empty vector. (d) Immunofluorescence staining of AR in LNCaP cells cotransfected with Rec-GFP in the absence and presence of 0.1 nM DHT. p values are indicated by * (p < 0.05), ** (p < 0.01), *** (p < 0.001), and ns (not significant).

We also performed immunofluorescence studies to investigate whether Rec has influence on the subcellular distribution of the endogenous AR in LNCaP cells (Fig. 3d). In untreated cells, we find already most of the AR in the nucleus. Coexpression of Rec induces a shift of almost all of the remaining cytoplasmic AR into the nucleus, but only in cells treated with 0.1 nM DHT. Interestingly, colocalization of Rec and AR occurs mostly in the cytoplasm (and eventually in the nucleus) but not in the nucleoli where Rec and hSGT colocalize.

By transfecting LNCaP cells with the AR-responsive luciferase reporter construct and hSGT or an empty control vector, we could confirm the negative effect of hSGT on the AR activity (Fig. 4a) as previously reported.21 This effect was modulated by DHT, the activating ligand of the AR with 1 nM DHT almost completely abrogating suppression by hSGT (Fig. 4a). This indicates that hSGT only inhibits AR activity if it is not bound to its androgen ligand. We next analyzed the interplay of Rec and hSGT on AR activity. LNCaP cells were cotransfected with varying amounts of Rec and hSGT (Fig. 4b) and overexpression of hSGT was found to reduce the Rec induced enhancement of the probasin promoter. This effect was dependent on the amount of hSGT expressed in these cells. Additionally, we analyzed the effect of Rec and hSGT expression on the transcription of the strictly AR regulated PSA gene by quantitative real time RT PCR (Fig. 4c). A weak but no significant increase of PSA mRNA in the presence of Rec was observable in untreated cells. In contrast, enhancement reached more than twofold in cells treated with 0.1 nM DHT. On the other hand, coexpression of hSGT diminishes this enhancement.

Figure 4.

Rec dependence of the AR modulation by hSGT. (a) hSGT negatively regulates the AR. LNCaP cells were transfected as indicated with ARR3-tk-Luc, AR, hSGT or appropriate empty control vectors. Luciferase activity was determined in the presence of 1 nM DHT or ethanol as vehicle control. (b) Effect of Rec and hSGT on AR activity. LNCaP cells were transfected with ARR3-tk-Luc and varying amounts of Rec and hSGT. (c) Real time RT PCR quantification of PSA mRNA normalized against GAPDH mRNA. p values are indicated by * (p < 0.05), ** (p < 0.01), *** (p < 0.001), and ns (not significant).

HERV-K(HML-2) profits from the Rec-mediated enhancement of AR activity

Many retroviruses such as MMTV or XMRV have been shown to become transcriptionally stimulated by steroids, particularly by androgens.34, 35

We observed an increase in reverse transcriptase activity in the HERV-K(HML-2) expressing teratocarcinoma cell line GH in the presence of DHT (Fig. 5a). This effect was stimulated further by a factor of two when cotransfection with an AR expression construct was carried out. The simple expression of AR in the absence of DHT did not have an effect on RT activity.

Figure 5.

The activation of the AR by Rec results in an increase of HERV-K(HML-2) LTR activity. (a) The HERV-K(HML-2) expressing teratocarcinoma cell line GH was transfected with 0.1 ng AR or empty control vector and stimulated with 0.1 nM DHT. RT activity in supernatants was determined using the Cavidi RT assay. (b) Responsiveness of retroviral LTRs to Rec in AR regulated cells. LNCaP cells were transfected with luciferase reporter constructs driven by retroviral LTRs, with Rec or with empty pcDNA3. The activity in the absence of Rec was set as 100%. (c) Responsiveness of the HERV-K(HML-2) LTR to DHT and Rec in GH and HeLa cells transfected with the HERV-K113 LTR-luciferase reporter and pcDNA3.1_oricoRec-V5 (+) or empty pcDNA3.1 plasmid (−). (d) GH cells were transfected with HERV-K113 LTR luciferase reporter and varying amounts of Rec or empty control vector. (e) DU145 cells lacking a DHT-responsive AR were transfected with AR, LTR-luciferase reporter, 35 ng Rec or empty pcDNA3 and 35 ng hSGT or empty control vector. (f) Interplay of hSGT and Rec on AR activity influencing HERV-K(HML-2) LTR driven transcription. Tera-1 cells were cotransfected with the HERV-K113 LTR reporter and increasing amounts of Rec. 75 ng of hSGT were cotransfected in parallel experiments. (g) HERV-K(HML-2) particle production in Tera-1 cells in the presence of Rec and hSGT. HERV-K(HML-2) specific Gag-precursor bands in virus pellets were detected by immunoblots using an αp15 antibody.29 In addition, the same supernatants were analyzed for RT activity (lower panel). p values are indicated by * (p < 0.05), ** (p < 0.01), *** (p < 0.001), and ns (not significant).

We therefore investigated whether the HERV-K(HML-2) LTR is responsive to AR activity. Luciferase reporter constructs carrying the HERV-K10 and HERV-K113 LTRs were used, as well as control LTRs from KoRV and HIV-1 that have been shown to be functionally active.24 The HERV-K(HML-2) LTRs but not the LTRs from HIV-1 or KoRV were stimulated by Rec expression in LNCaP cells by a factor of two (Fig. 5b). Because Rec is often overexpressed in tumors of tissues that are regulated by androgens we further characterized the effects of Rec and AR activity on the HERV-K(HML-2) LTR in various AR responsive cell lines such as the teratocarcinoma cell lines GH and Tera-1 that express low levels of Rec from various endogenous proviruses. The results were compared with those from HeLa cells that clearly lack a functional AR.36 As expected for a cell line with a responsive AR, the addition of DHT resulted in a three-fold increase in HERV-K(HML-2) LTR activity in transfected GH cells. Cotransfection with additional Rec resulted in a further boost of the HERV-K(HML-2) LTR activity up to 10-fold higher than basal activity (Fig. 5c) and the degree of enhancement was dependent on the amount of Rec cotransfected (Fig. 5d). These effects were only measurable in the presence of the AR as cell lines such as HeLa that lack a functional AR did not show increased LTR activity in response to DHT treatment or Rec expression (Fig. 5c).

Inhibition of the stimulatory effect of Rec on the HERV-K(HML-2) LTR promoter was assessed in DU145 cells transiently transfected to express the AR. Similar to the activation of the probasin promoter, the LTR is also activated by Rec expression and down-regulated by ectopic hSGT expression. Cotransfection of Rec and hSGT neutralizes both effects (Fig. 5e). A similar outcome for Tera-1 cells expressing endogenous AR is shown in Figure 5f. Coexpression of hSGT significantly reduced the Rec-mediated LTR activation.

Enhancement of LTR activity by Rec and the inhibition by hSGT overexpression could also be demonstrated by the levels of particle production in AR expressing cells. Tera-1 cells were transfected with a molecular clone of the provirus oriHERV-K113 encoding ancestral viral proteins25, 31 (and Beimforde et al., in preparation) under the control of the same LTR sequence as that in the luciferase construct used in previous experiments. Tera-1 cells were cotransfected with various combinations of AR, Rec, hSGT plus empty vector (to standardize the amount of DNA). Virus particles from the supernatants were pelleted and analyzed by Western blot using a HERV-K(HML-2) specific antibody directed against the HERV-K(HML-2) Gag p15 protein.29 As shown in Figure 5g, virus particle production and RT activity were enhanced by Rec but coexpression of hSGT abrogated the enhancement.

Discussion

Using a yeast two-hybrid approach to identify Rec interacting cellular proteins, two different hsgt fragments, both comprising the central TPR motifs, were detected in 15 positive clones. Coimmunoprecipitation and pull-down assays verified that the interaction is not dependent on yeast specific factors. Pull-down experiments also confirmed that the central TPR domain containing three TPR motifs37 is essential for binding to HERV-K(HML-2) Rec.

hSGT is mainly located in the cytoplasm.20 Binding to HERV-K(HML-2) Rec, however, results in a significant displacement of hSGT into the nucleus. Along those lines it has been shown for the NS1 protein of parvovirus H-1, that binding can lead to an accumulation of hSGT in the nucleus and that this results in enhanced viral replication.38 However, for HIV-1, hSGT has been reported to interact with Gag and Vpu and to downregulate HIV-1 particle release probably by Vpu sequestration.39

Despite its role as a regulator of the heat shock proteins Hsp70 and Hsp90, little is known about the physiological function of hSGT. There are, however, some reports demonstrating that hSGT is a negative regulator of the androgen receptor.21, 22 Androgen activity is crucial for a multitude of physiological and developmental processes such as the male sexual development and maintenance or spermatogenesis.40 Furthermore, there are many recent reports showing an impact of the AR in female sexual regulatory processes.41

The androgen receptor is regulated by a protein complex consisting of chaperones and TPR-motif containing proteins. One of these is hSGT that binds to the AR hinge region via its TPR motif.21 HSGT helps the Hsp70/Hsp90 chaperon complex to fold and refold the AR and regulates the ATPase activity of Hsp70 that is necessary for chaperonic folding.42 As long as no ligand is present hSGT keeps the steroid receptor in a ligand responsive but inactive state.21 As soon as ligand (testosterone or its metabolite 5α-dihydrotestosterone) binds to the AR, hSGT is released from this complex, the AR moves to the nucleus and activates genes containing the androgen receptor responsive element (ARE).21

Our experiments demonstrate that binding of Rec to hSGT results in an enhancement of AR activity and that this effect is proportional to cellular Rec concentrations. AR responsive promoters are only activated in the presence of a functional AR suggesting a specific Rec-mediated stimulation of the steroid receptor. Coexpression of hSGT reduces this stimulation, implying a competitive effect in which Rec binds to hSGT and in turn reduces the amount and activity of hSGT in AR-chaperone complexes. The more hSGT is moved away from the AR, the more active AR becomes, with enhanced cell proliferation and reduced apoptosis. By this activity, Rec may therefore stimulate these processes and enhance tumor progression.

Interestingly, HERV-K(HML-2) replication itself profits from enhanced AR activity levels. Both LTR-luciferase reporter assays and particle production assays showed that Rec stimulates its own transcription in the presence of a functional AR and that hSGT has the opposite effect, downregulating HERV-K(HML-2) LTR activity. Nevertheless, it remains open whether or not the AR binds directly to the HERV-K(HML-2) LTR or whether other transcription factors are stimulated that result in the 2.5-fold increase in HERV-K(HML-2) LTR activity. An AR binding site in the HERV-K LTR has not been demonstrated yet. Various retroviral LTRs are known to be stimulated by androgens.34, 35, 43 Even so, it has long been known that androgen regulated tissues and cancers such as teratocarcinomas particularly tend to show HERV-K(HML-2) particle production.44 This study gives clear evidence for a direct role of AR action on HERV-K(HML-2) activity and vice versa. A close connection between AR activity and HERV-K(HML-2) transcription is also supported by a study by Kaufmann et al.19 in which a direct interaction between Rec with the testicular zinc finger protein (TZFP) and the androgen receptor resulting in enhanced AR activity was demonstrated.

One interpretation of our and previous data would suggest that HERV-K(HML-2) Rec expression could start a vicious circle in AR regulated cells (Fig. 6). For example, if Rec expression is stimulated in early prostate tumor cells this might perturb the balance of the AR/hSGT regulation machinery. Activity of the steroid receptor would be enhanced resulting in increased cell proliferation. As a secondary effect, HERV-K(HML-2) transcription would be stimulated which in turn would result in increased Rec levels, further upsetting regulation of the AR by hSGT. In that way, HERV-K(HML-2) might contribute to tumor progression in hormone regulated tissues. This model is supported by the fact that tissues such as breast, gonads, ovaries or prostates that are strongly regulated by steroid receptor action often show an increased HERV-K(HML-2) expression in malignant cells.5, 16 Various groups have shown that AR activity also plays an important role in the functional regulation of ovary and breast tissues.41, 45 Even melanocytes express (albeit at low quantities) AR and steroid hormones.46

Figure 6.

“Vicious circle” model of tumor progression driven by Rec-mediated AR stimulation based on a model by Buchanan et al.21 HSGT keeps the AR inactive until androgens bind to the steroid receptor. After ligand binding, hSGT is replaced by FK506-binding protein 4 (FKBP52). The AR complex, consisting of heat shock protein 90 (Hsp90), FKBP52, AR and the cochaperone p23, is translocated into the nucleus by (presumably) the interaction of the FKBP52 heavy chain (HC) and intermediate chain (IC) with dynein, resulting in activation of the transcription of genes carrying an ARE. In Rec-expressing cells, Rec binds to hSGT and decreases its negative regulatory effect on the AR. The AR is more active and transcription of AR-dependent genes is enhanced leading to cell proliferation and reduced apoptosis. In addition, the AR activates HERV-K(HML-2) LTRs in the genome causing HERV-K(HML-2) transcription and enhanced production of Rec. Elevated levels of Rec further reduce the negative effect of hSGT on the AR leading to even higher AR activity, increased cell proliferation, inhibition of apoptosis and eventually to tumorigenesis. NES: nuclear export signal, NLS: nuclear localization signal.

In conclusion, we could demonstrate that the HERV-K(HML-2) Rec protein elevates AR activity by binding to its negative regulator hSGT. AR activation enhances not only transcription of AR-dependent cellular genes but also HERV-K(HML-2) transcription and particle production. This could initiate and drive a vicious circle leading to cancer development or progression.

Acknowledgements

The authors are indebted to Ewelina Caspers and Chris Daubitz for excellent technical assistance and to Steve Norley for critical reading of the manuscript and inspiring discussions.

Ancillary