Androgen receptor (AR) signals play a decisive role in regulating the growth and differentiation of both normal and cancerous prostate cells by triggering the regulation of target genes, in a process in which AR cofactors have critical functions. Because of the highly prostate-specific expression pattern of HOXB13, we studied the role of this homeodomain protein in prostate cells. Expression of HOXB13 was limited to AR-expressing prostate cells. Reporter transcription assay demonstrated that HOXB13 significantly suppressed hormone-mediated AR activity in a dose-responsive manner, and suppression was specific to AR with which HOXB13 physically interacts. Overexpression of HOXB13 further down-regulated the androgen-stimulated expression of prostate-specific antigen, and suppression of endogenous HOXB13 stimulated transactivation of AR. Functionally, HOXB13 suppressed growth of LNCaP prostate cancer cells, which could be counteracted by additional hormone-activated AR. On the other hand, the growth-suppressive function of HOXB13 in AR-negative CV-1 cells was not affected by AR. These results suggest that HOXB13 functions as an AR repressor to modulate the complex AR signaling and subsequent growth regulation of prostate cancer cells. In addition to the loss of HOXB13 expression, maintaining AR may be an important step for prostate cancer cells to tolerate the suppressor function of HOXB13. Altogether, our data present a novel mechanism for the HOXB13-mediated repression of AR signaling, which can be interpreted to a growth-suppressive event.
Prostate cancer is the most commonly diagnosed malignancy in men and poses a significant medical problem in the United States. Currently, androgen ablation (chemical or surgical) is the most effective therapy available for advanced prostate cancer, although androgen-independent cancers eventually develop and claim the life of patients. Understanding the complex regulation of androgen signaling is necessary to elucidate the androgen-independent growth of this advanced disease. Androgen receptor (AR) has a key regulatory function in the growth and differentiation of normal and cancerous prostate epithelial cells, triggering the regulation of target genes such as jun, fos, myc, cdks, fibroblast growth factor, and others. These AR-responsive genes are positively or negatively regulated by recruiting many cofactors into the hormone-receptor complex. Despite the fact that more than 100 steroid receptor cofactors are known, their mechanistic and functional association is not well understood. Only very limited information is available about the negative regulation of hormone-activated AR function.
AR transactivation is initiated upon binding to androgen. Then AR translocates into the nucleus and binds as a homodimer to the cognate DNA response elements on the promoter, where it further interacts with other transcriptional factors and activates downstream gene transcription such as prostate-specific antigen, kallikreins, and calreticulin. AR has distinct functional domains: a highly conserved DNA-binding domain comprising two zinc finger motifs, a COOH-terminal ligand-binding domain, and a poorly conserved NH2-terminal domain containing one or more transcriptional activation domains. Several cofactors associated with AR have been identified, such as SRC1, AIB1, TIF2, Smad3, BRCA1, CBP, ARAs, and NCoR (reviewed in ref. 1 ). All of these coregulators can interact with either the COOH- and/or NH2-terminal domains of AR and regulate AR transactivation. For example, the p160 coactivators SRC1, TIF2, and AIB1 play a central role in mediating transcriptional activation, binding directly to the COOH-terminal activation function domain-2 of nuclear receptors in a ligand-dependent manner. Furthermore, the COOH-terminal region of p160 can also interact with the NH2-terminal activation function domain-1 domain of some nuclear receptors (2, 3, 4) . On the other hand, NCoR, an AR corepressor, interacts directly with the AR ligand-binding domain and represses hormone-stimulated AR transcriptional activity (5) . These coactivators and corepressors seem to act as accelerators and/or brakes that modulate the transcriptional regulation of hormone-responsive target gene expression, further regulating cell growth. Thus, the discovery of new coactivators and corepressors will expand our knowledge of the mechanisms of steroid receptor action.
Hox homeobox genes encode a set of transcription factors that regulate axial regional specification during embryonic development and are expressed with temporal and spatial colineality. Therefore, Hox genes are expressed in a tissue-specific and frequently stage-related fashion. The mammalian Hox genes are clustered in four unlinked complexes in the genome (a b, c, and d). Each complex spans approximately 200 kb and contains 9 to 11 genes, composing 39 paralogs (reviewed in 6 ). The Hox-13 paralog is important to the development of the prostate. All Hox-13 genes except Hoxc13 are expressed in the prostate during embryonic development (7) . However, Hoxa13 and Hoxd13 are no longer expressed after the embryonic stage (8 , 9) . Hoxd13- and/or Hoxa13-deficient mice showed morphologic abnormalities in male accessory sex organs, including the seminal vesicle and prostate (8 , 10 , 11) . Hoxb13, the last identified vertebrate Hox gene, has been shown to be exclusively expressed in the caudal extent of the spinal cord, tail bud, and urogenital sinus in an androgen-independent manner (12 , 13) . Mice homozygous for Hoxb13 loss-of-function mutations showed overgrowth in all major structures derived from the tail bud (14) and malformation of ducts of the ventral prostate, including complete loss of secretory proteins (15) . Recently, we demonstrated that HOXB13 suppresses the growth of AR-negative PC3 prostate cancer cells by the down-regulation of T-cell factor 4 (TCF-4) expression (16) . In this study, we investigated the role of HOXB13 in AR-expressing prostate cancer cells, which may provide new insight into complex AR-signaling pathway mechanisms and a better understanding of its role in prostate carcinogenesis and the progression to androgen-independent prostate cancer.
MATERIALS AND METHODS
Plasmids and Reagents.
pFLAG-HOXB13, pIRES-HOXB13, and Ad-GFP-HOXB13 were described previously (16) . pGL-AREc-luc was also described previously (17) . Androgen response element core is known as the distal enhancer region of the prostate-specific antigen promoter (18) . pGL-ARE4-luc contains a synthetic TATA and four tandem copies of androgen response elements from the murine mammary tumor virus promoter in the context of the pGL3 backbone (Promega, Madison, WI). Estrogen response element–regulated luciferase vector, ERE2-Luc, was provided by Dr. Deborah Lannigan (University of Virginia Health System, Charlottsville, VA). Vectors encoding AIB1 and TIF2 were constructed in pAd1020SfidA (OD260, Inc.) under the control of cytomegalovirus (CMV) promoter. A human AR expression vector, pARO, was a gift from Dr. Jan Trapman (Erasmus University, Rotterdam, The Netherlands). pCMV-βgal was from Promega. Monoclonal antibodies to AR (generated against amino acids 299–315 of AR) and β-actin were from Santa Cruz Biotechnology (Santa Cruz, CA). For coimmunoprecipitation, polyclonal anti-AR antibodies (PG21, Upstate Biotechnology, Lake Placid, NY) were used. Monoclonal anti–prostate specific antigen antibodies were from Dako Cytomation (Carpinteria, CA). Anti-FLAG M5 and anti-c-myc antibodies were from Sigma (St. Louis, MO) and BD Biosciences (San Jose, CA), respectively. Synthetic testosterone, R1881, was from NEN Life Science (Boston, MA) and used at a final concentration of 10 nmol/L.
Human prostate cell lines, including P69, LNCaP, C4–2, CWR22RV, PC3 and DU145, and LoVo colon cancer cells, were routinely cultured in RPMI (Invitrogen, Carlsbad, CA) supplemented with 5% fetal bovine serum (FBS) at 37°C in an atmosphere containing 5% CO2 as described previously (19) . MDA PCa 2b prostate cancer cells were grown in BRFF-HPC1 medium (Athena Environmental Sciences, Inc., Baltimore, MD) with 20% FBS. CV-1 monkey kidney cells, MCF7A breast cancer cells, and HEK 293 human embryonic renal cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% FBS. All cultures were fed with fresh medium every 3 to 4 days.
Real-Time Reverse Transcription-Polymerase Chain Reaction.
Total RNA from normal human tissues was purchased from Clontech Laboratories (Palo Alto, CA). These RNA were quality-controlled and assured by the company. RNA from prostate, colon, and rectum was obtained from two other different sources to reduce sampling errors. RNA extraction from selected cultured cells and patient samples was performed as described previously (20) . Total RNA from PrEc, Pz-HPV7, LAPC-4, and MDA PCa 2a and 2b was generously provided by Dr. Jer-Tsong Hsieh (University of Texas Southwestern Medical Center, Dallas, TX). LNCaP RNA was used for standardizing the expression levels of HOXB13 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), an internal control. Fifty ng of total RNA from each sample were used to detect real-time reverse transcription-PCR products using Taqman probes and an ABI PRISM 7700 sequence detection system (PE Applied Biosystems, Foster City, CA). PCR cycling conditions were performed for all of the samples as follows: 30 minutes at 48°C for reverse transcription; 10 minutes at 95°C for AmpliTaq Gold activation; and 40 cycles for the melting (95°C, 15 seconds) and annealing/extension (60°C, 1 minute) steps. HOXB13 and GAPDH primers and probes for real-time reverse transcription-PCR were designed using the PRIMER Express program (PE Applied Biosystems). The sequences of the HOXB13 primers and probe are as follows: forward, 5′-ccgggaacctaccacgctat-3′; reverse, 5′-ttctccaggagcacccagagt-3′; probe, 5′-FAM-cagttacctggacgtgtctgtggtgca-TAMRA-3′. The sequences of the AR primers and probe are as follows: forward, 5′-aaggctatgaatgtcagccca-3′; reverse, 5′-cayygaggctagagagcaaggc-3′; probe, 5′-FAM-tgtgtgctggacacgacaacaacc-TAMRA-3′. The sequences of GAPDH primers and probe are as follows: forward, 5′-gaaggtgaaggtcggagtc-3′; reverse, 5′-gaagatggtgatgggatttc-3′; probe, 5′-VIC-caagcttcccgttctcagcc-TAMRA-3′. All real-time reverse transcription-PCR experiments were performed twice in duplicate in one 96-well plate. Using the comparative CT method (PE Applied Biosystems), resulting Ct values were converted to picogram quantities according to each standard curve. HOXB13 and AR quantity was then normalized by GAPDH and subtracted from no reverse transcriptase controls. This value was then averaged for each duplicate.
Standard reverse transcription-PCR was performed to verify the expression of HOXB13, AR, prostate specific antigen, and β-actin in xenograft models. RNA from prostate tumor cell xenograft models was generously provided by Dr. Wytske M. van Weerden (Erasmus University, Rotterdam, The Netherlands). The primer sequences for each gene are as follows: HOXB13fw, 5′-ccccactgagtttgccttctatc-3′; HOXB13rv, 5′-gcctcttgtccttggtgatgaac-3′; ARfw, 5′-ggatgaggaacagcaaccttcac-3′; ARrv, 5′-atggacaccgacactgccttacac-3′; prostate specific antigen fw, gcagcattgaaccagaggag; prostate specific antigen rv, agaactggggaggcttgagtc; β-actinfw, 5′-gcaccacaccttctacaatgagc-3′; β-actinrv, 5′-tagcacagcctggatagcaacg-3′.
In situ Hybridization and Immunocytochemistry.
For in situ hybridization, a segment of human HOXB13 cDNA (350 bp) was amplified using a set of promoter sequence–linked primers in reverse transcription-PCR. The T7 promoter was incorporated into the HOXB13fw primer, and the SP6 promoter was incorporated into the HOXB13fw primer. Labeled riboprobes were synthesized by in vitro transcription of PCR-amplified cDNA using digoxigenin-RNA labeling kit, SP6-T7 (Roche Diagnostics, Indianapolis, IN) as described by the manufacturer. In situ hybridization was performed as described previously (20) . For immunocytochemistry, human embryonic kidney 293 cells were plated onto two-well chamber slides, and pFLAG-HOXB13 was transfected. After 24 hours, cells were fixed and probed with anti-FLAG antibodies. Staining was developed using 3-amino, 9-ethyl-carbazole chromogen.
Approximately 1 × 105 cells were plated in a 24-well plate 16 hours before transfection. To see the hormone effect, cells were grown under 5% charcoal dextran–treated (CDT) FBS for 3 days before the transfection. Transfections were carried out using LipofectAMINE 2000 (Invitrogen) with 0.1 μg of reporter, 0.1 μg of test plasmid, and 20 ng of pCMV-βgal as described by the manufacturer’s protocol. Six hours after transfection, the cells were washed and fed with medium containing 5% CDT-FBS. If needed, cells were treated with either R1881 synthetic androgen or ethanol. After 36 hours, cells were washed with PBS, lysed with 100 μL of passive lysis buffer, and assayed for luciferase activity as relative light units using Luciferase assay system (Promega). β-Galactosidase activity was assessed for the normalization and transfection efficiency. Lysates (15 μL) were incubated in 150 μL of ONPG solution (1 mmol/L MgCl2, 0.8 mg/mL o-nitrophenyl-β-d-galactopyranoside, 0.1 mol/L sodium phosphate, and 45 μmol/L β-mercaptoethanol) at 37°C for 30 minutes. Reactions were stopped by the addition of 200 μL of 1 mol/L Na2CO3, and activity was measured as absorbance at 420 nm. Transfection experiments were done in triplicate and results are reported as mean ± SD.
Small Interfering RNA Transfection.
For RNAi assay, small interfering RNA corresponding to the nucleotide positions 837 to 855 of human HOXB13 (relative to stop codon) were synthesized by Dharmacon (Lafayette, CO). The selected sequences (5′-gaacagcgcuaccccuuaa-3′) were submitted to BLAST searches against the nonredundant protein database in National Center for Biotechnology Information to make sure that only the selected genes were targeted. Lamin A/C SMARTpool Reagent (Dharmacon) was used as a nonspecific control. LNCaP cells were cultured onto 24-well plate in RPMI containing CDT-FBS for 2 days before the transfection. Transfections were carried out using DOTAP (Roche Diagnostics) with 0.1 μg of reporter, 5 to 25 nmol/L small interfering RNA, and 20 ng of pCMV-βgal. After 48 hours, reporter activity was measured as described above. For rescue assays, two silent nucleotide changes were introduced into the region complementary to HOXB13 small interfering RNA. The region gaacagcgctaccccttaa was mutated to gaacagcgccacaccttaa, leaving the amino acid sequences unchanged. Mutations were accomplished by PCR using pFLAG-HOXB13 as a template. Primers used were 5′-aaggtgaagaacagcgccacacc-3′ and 5′-ggcgagaaccttcttctctttg-3′. Amplified fragments were ligated with Rapid ligation kit (Roche Diagnostics). Sequencing analysis was done to verify the mutations.
Western Blotting Assay.
Cells were grown up to 80% confluence in P60 culture dishes containing 5% FBS-T medium. Cells were then lysed in protein extraction buffer (1× Tris-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease inhibitors) followed by needle sonication to break up the ribonucleosome. Twenty μg of total cell lysates were loaded onto 10% Bis-Tris gel (Invitrogen) and separated using the Novex electroporation system (Invitrogen). After proteins were transferred to polyvinylidene difluoride membrane, primary antibodies were applied, followed by incubation with horseradish peroxidase–conjugated secondary antibodies. Blots were developed by the ECL detection system (Pierce, Rockford, IL).
LNCaP cells were grown under 5% CDT-FBS for 3 days and then infected with recombinant Ad-GFP-HOXB13 virus [5 multiplicity of infection (MOI)] and cultured in medium containing CDT-FBS. A 10 nmol/L final concentration of R1881 was added. Nuclear extracts were collected as described previously (21) . Nuclear extracts (100 μg) were preincubated with 4 μg of AR-PG21 antibodies (Upstate Biotechnology) for 3 hours at 4°C, followed by adding protein A/G agarose (Santa Cruz Biotechnology). To precipitate FLAG-HOXB13, nuclear extracts were directly mixed with EZView Red Anti-flag M2 Affinity Gel (Sigma). The mixture was incubated overnight at 4°C. Immunoprecipitates obtained by centrifugation were washed with Tris-buffered saline three times and eluted with 2× SDS loading buffer. Proteins were separated on 10% SDS-PAGE, and immunoblotting was followed by using either anti-AR antibodies (441; Santa Cruz Biotechnology) or anti-FLAG antibodies.
Colony formation assay was performed as described previously (16) . LNCaP cells were also monitored under the fluorescence microscope after virus infection to see the functional relation between HOXB13 and AR. In brief, LNCaP cells were infected with either Ad-GFP or Ad-GFP-HOXB13 at 2 MOI under the influence of androgen. For the coinfection study, LNCaP cells were grown under CDT-FBS for 3 days before the virus treatment. Recombinant AR virus (Ad-CMV-AR) was added to every well (1 MOI), together with either Ad-GFP or Ad-GFP-HOXB13 (1 MOI) as described above. At the same time, cells were treated with either R1881 (10 nmol/L) or ethanol. Infectivity and cell viability were carefully monitored by fluorescence microscopy for 10 days. Every 3 to 4 days, fresh medium was added. A 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) in vitro proliferation assay was also performed as described previously (16) . In brief, cells were plated on a 24-well plates and infected with Ad-CMV-AR (2 MOI) and either Ad-GFP-HOXB13 or Ad-GFP (2 MOI) followed by the addition of 10 nmol/L R1881. At five days after virus infection, cells were stained with 100 μL of 5 mg/mL MTT solution and incubated for 2 hours at 37°C. The reaction was stopped by adding 400 μL of extraction buffer [50% formamide and 10% SDS (pH 4.7)]. After overnight incubation at 37°C, the absorbance at 570 nm was measured using a Spectra Microplate Reader (Molecular Devices, Sunnyvale, CA). Densitometric values were analyzed with Student’s t test, using GraphPad Prism software (GraphPad, San Diego, CA).
Expression of HOXB13 Is Specific to Prostate and Colorectum.
Real-time reverse transcription-PCR was performed to determine the level of HOXB13 mRNA expression in various normal human tissues. No functional antibodies are available at this point. Fig. 1A ⇓ represents selective tissues that express substantial amounts of HOXB13 mRNA. HOXB13 was abundantly expressed in prostate (arbitrarily designated as 1-fold) and colorectal tissues (0.08–0.25-fold) compared with ovary and testis (≤0.005-fold). The other tissues tested (brain, cerebellum, spinal cord, lung, trachea, thymus, thyroid gland, heart, liver, mammary gland, pancreas, salivary gland, stomach, small intestine, adrenal gland, kidney, spleen, bone marrow, uterus, placenta, skin, and skeletal muscle) did not express a considerable amount of HOXB13 mRNA, below 0.005-fold compared with the prostate (data not shown). In mouse Hoxb13 study, Hoxb13 mRNA is confined to prostate, distal colon, and rectum (13) . To localize HOXB13 mRNA, in situ hybridization using HOXB13 riboprobe was performed in prostate tumors. HOXB13 RNA was exclusively localized to luminal epithelial cells of the prostate (Fig. 1B ⇓ , top panel), whereas sense strand of HOXB13 probe did not show signal (Fig. 1B ⇓ , middle panel). There were mixed populations of HOXB13-positive and -negative cells among tumor samples (data not shown). In addition, immunocytochemical staining of FLAG-HOXB13–transfected cells demonstrated that HOXB13 is predominantly localized at the nucleus (Fig. 1B ⇓ , bottom panel).
Expression of HOXB13 Is Limited in AR-Positive Prostate Cancer Cells.
The level of HOXB13 was determined in prostate cells using real-time reverse transcription-PCR analysis. RNAs from normal prostate were used for comparison. To reduce sampling error, RNA for normal prostate was acquired from three different sources. As shown in Fig. 2A ⇓ , HOXB13 was limitedly expressed in all prostate cancer cells compared with normal prostate, except that HOXB13 was highly expressed in MDA PCa 2b cells. MDA PCa 2b cells are developed from bone-metastasized androgen-independent tumors and have two mutations in ligand-binding site, and its growth is believed to be dependent on androgen-independent pathway (22, 23, 24) . Interestingly, all prostate cancer cell lines retaining AR (AR positive) expressed moderate to high levels of HOXB13 (LNCaP, C4–2, LAPC-4, and MDA PCa), whereas AR-negative cells had low to undetectable amounts of HOXB13. AR-negative cells include PC3 and DU145, most progressive (metastatic) and malignant cells, and immortalized normal prostate cells (P69, PrEc, and Pz-HPV7). Sequencing analysis showed that LNCaP and MDA PCa 2b cells did not retain any mutations in HOXB13 coding region. Using regular reverse transcription-PCR, prostate tumor cell xenograft models were evaluated for the expression of HOXB13 and AR. These models are known to be more genetically stable than cultured cells, have a more homogeneous population than patient samples, and include both androgen-dependent and -independent cells (25 , 26) . This system provides valuable information because the HOXB13 expression level was not altered in patient samples of prostate cancer compared with matching benign tissues due to the well-known heterogeneous population of prostate cancer (data not shown). Tumor cell xenograft models showed the same phenomenon of correlated expression pattern of HOXB13 and AR (Fig. 2B) ⇓ . Only PC-135 showed a different expression pattern, a finding that needs to be further investigated. Then we tested whether HOXB13 is controlled by androgen using the testosterone derivative R1881 in LNCaP cells. Cells (2 × 105) were plated in duplicate and grown for 48 hours with or without 10 nmol/L R1881. Total RNA was extracted and used for real-time reverse transcription-PCR analysis. As shown in Fig. 2C ⇓ (left panel), the addition of R1881 did not significantly alter the level of HOXB13 mRNA. This result is consistent with the mouse study (13) . Conversely, overexpression of HOXB13 did not regulate the expression of AR RNA (Fig. 2C ⇓ , right panel). These observations suggest that HOXB13 expression is regulated in androgen-independent manner, and its loss or decrease may be an important step for prostate tumor progression process.
HOXB13 Specifically Represses Hormone-Stimulated Androgen Receptor Transcriptional Activity.
To determine whether there was any mechanistic and functional association between HOXB13 and AR, transient transfection combined with reporter transcription analysis was performed. First, we used the androgen response element core region of the prostate-specific antigen promoter linked to luciferase (pGL-AREc-luc). HOXB13 repressed about 50% of androgen-activated AR activity on this prostate specific antigen promoter, whereas HOXB13 did not inhibit hormone-independent AR activity (Fig. 3A) ⇓ . Then we used four copies of androgen response elements cloned into a pGL-TATA-luc vector (Promega), pGL-ARE4-luc. As shown in Fig. 3B ⇓ , transfection of HOXB13 into AR-positive prostate cancer cells induced 3- to 4-fold suppression of androgen-activated AR transcriptional activity. Moreover, cotransfection of HOXB13 and AR into AR-negative cells caused greater inhibition activity (4–14-fold), notably more than a 10-fold decrease in PC3 prostate cancer cells (Fig. 3C) ⇓ . HOXB13-mediated inhibition of AR activity occurred in a dose-responsive manner (Fig. 3D) ⇓ in LNCaP cells, and estradiol-stimulated estrogen receptor activity was not suppressed by HOXB13 in MCF7A cells (Fig. 3E) ⇓ . Previously, we demonstrated that HOXB13 promoted the transcriptional activity of the other universal promoters (16) . These include SV40 and RSV promoters as well as general TATA, suggesting that HOXB13 may function as a general transcription factor in non-AR–mediated transactivation. To investigate whether forced expression of HOXB13 further down-regulates the expression of AR responsive genes, a HOXB13-expressing recombinant adenovirus (Ad-GFP-HOXB13) was used to overcome poor transfection efficiency in prostate cancer cells. This virus has shown high infectivity with minimal viral toxicity in a previous studies (16) . LNCaP cells were infected with either Ad-GFP-HOXB13 or Ad-GFP viruses in the presence or absence of androgen for 36 hours followed by whole-cell lysate extraction and Western blot analysis. As shown in Fig. 4A ⇓ , expression of androgen-activated prostate specific antigen was significantly decreased by the expression of HOXB13. Interestingly, overexpression of viral recombinant HOXB13 up-regulated the expression of AR protein without a known mechanism. To demonstrate whether the suppression of physiologic HOXB13 can promote the AR activity, we transfected HOXB13-specific small interfering RNA duplexes (Dharmacon) into LNCaP cells, in which HOXB13 is moderately expressed as shown in Fig. 2A ⇓ . Transfection of HOXB13 small interfering RNA significantly increased hormone-stimulated AR activity in a dose-responsive manner (5–25 nmol/L final concentration), whereas a maximum dose of Lamin A/C small interfering RNA did not affect AR activity (Fig. 4B) ⇓ . To show the specificity of small interfering RNA, mutant form of pFLAG-HOXB13 vector refractory to small interfering RNA was made by two silent third-base point mutations. As shown in Fig. 4C ⇓ , cotransfection of mutant HOXB13 almost completely negated the activity of HOXB13 small interfering RNA in ARE4-luc system.
HOXB13 Physically Interacts with Androgen Receptor.
To see whether there is any physical interaction between HOXB13 and AR and/or its accessory proteins, coimmunoprecipitation assay was performed. Nuclear extracts from LNCaP cells infected with Ad-GFP-HOXB13 were immunoprecipitated with either anti-Flag antibodies or AR-specific antibodies followed by Western blot analysis. As shown in Fig. 5 ⇓ , either Flag-HOXB13 or AR was detected by its counterpart antibodies, whereas no band was detected with normal immunoglobulin G control. However, HOXB13 does not possess the LXXLL or FXXLF motif, common motifs in AR-interacting proteins. These results suggest that HOXB13-mediated suppression of AR activity is accomplished through interacting with AR.
Androgen Receptor Activity Suppressed by HOXB13 Can Be Restored by Overexpression of Coactivators and Androgen Receptor.
To investigate whether HOXB13 can inhibit the synergistic function of p160 coactivators, knowing that HOXB13 binds to AR, LNCaP cells were cotransfected with AIB-1 or TIF-2 with pGL-ARE4-luc followed by reporter gene transcription assay as described above. Although HOXB13 significantly suppressed androgen-activated AR activity, synergistic activity induced by overexpression of either AIB-1 or TIF-2 was not completely abolished (Fig. 6A) ⇓ . This result suggests that HOXB13 may affect AR-mediated transcriptional activity without disturbing the coactivator function of p160 coactivators. We then tested whether the suppressive function of HOXB13 can be rescued by the overexpression of AR. CV-1 monkey kidney cells were cotransfected with HOXB13 and an increased amount of AR (Fig. 6B) ⇓ . The addition of AR partially restores the transcriptional activity of AR, suggesting that HOXB13 and AR may functionally counteract to manifest coactivator-mediated transcriptional activity.
Androgen-Activated Androgen Receptor Counteracts HOXB13-Mediated Growth Suppression.
To observe the growth-modulating function of HOXB13, a colony formation study combined with the transfection of HOXB13 was performed. HOXB13 expression vector (pIRES-HOXB13) was transfected into three different p60 dishes of LNCaP cells, followed by the selection of transfected cells in neomycin-containing medium. pIRES-neo vector was used as a control. Two weeks later, the numbers of surviving clones were observed under the microscope and counted. As shown in Fig. 7A ⇓ , overexpression of HOXB13 dramatically inhibited cell growth (9.7 ± 1.3 colonies) compared with controls (0.3 ± 0.3 colonies). In addition, HOXB13 seems to drive LNCaP cells into terminal differentiation status, characterized by large nuclei and cytoplasm. Additional analysis using viral recombinant HOXB13 demonstrated that HOXB13 significantly decreased the number of GFP-expressing cells at 5 days after virus infection (Fig. 7B) ⇓ . This result suggests that HOXB13 functions as a negative growth regulator of AR-expressing LNCaP cells.
To confirm whether the growth-suppressive function of HOXB13 is mediated by affecting the androgen-AR signal, AR was overexpressed via adenoviral vector in LNCaP cells in the presence of androgen. As demonstrated in Fig. 8A ⇓ , cells were infected with Ad-CMV-AR virus (1 MOI) combined with either Ad-GFP-HOXB13 or Ad-GFP virus (1 MOI). Cells were closely monitored under the fluorescence microscope. At day 1 after infection, most cells expressed GFP, and each group showed a similar number of GFP-expressing cells. At day 10, HOXB13-mediated growth suppression was significantly retarded by the addition of androgen-activated AR. To confirm the result that overexpression of AR can rescue HOXB13-mediated growth suppression under the influence of androgen, an in vitro MTT proliferation assay was performed after the cells (LNCaP and CV-1) were infected with a combination of Ad-CMV-AR and either Ad-GFP-HOXB13 or Ad-GFP virus (2 MOI for each virus). At a final concentration of 10 nmol/L, R1181 was added to all of the cells immediately after virus infection. At 5 days after virus infection, cells were stained with MTT and the O.D. was measured. As shown in Fig. 8B ⇓ , HOXB13 suppressed the growth of both LNCaP and CV-1 cells compared with the Ad-GFP–infected group. Growth-suppressive effect by HOXB13 was higher in CV-1 cells than in LNCaP cells, suggesting that endogenous AR may exert growth protection in LNCaP cells. Meanwhile, infection with AR virus also suppressed the growth of both LNCaP and CV-1 cells in the presence of hormone. AR is well known to have a biphasic effect depending on the activation status of AR, and overactivation of AR drives cells into an apoptotic process and cell death (27, 28, 29) . However, coinfection of AR virus in the presence of androgen counteracts the growth-suppressive effect of HOXB13 in LNCaP cells, but not in CV-1 cells. This suggests that HOXB13 and AR functionally counteract each other in AR-intact cells and that balance between these two proteins is important for cells to avoid abnormal proliferation. Previously, we have demonstrated that HOXB13 suppresses the growth of AR-negative cells, including CV-1, by affecting the β-catenin/TCF signaling pathway (16) . Therefore, overexpression of AR in androgen-insensitive CV-1 cells did not affect the growth-suppressive function of HOXB13, which may be accomplished by affecting β-catenin/TCF signaling rather than androgen-AR signaling.
An immense effort has been made to understand the regulation of complex androgen-AR signals. Nuclear receptor cofactors are believed to have a central role in this essential growth-regulating pathway. Although more than 100 coactivators are reported to stimulate steroid receptors, limited numbers of corepressors are known to confer negative signals for this pathway. The best-known corepressors are NCoR/SMRT, which are constitutive transcriptional repressors and bind to their cognate DNA-binding sites in the absence of ligand. Although the repressive function of NCoR/SMRT has been reported in some ligand-activated steroid receptors (30 , 31) , the mechanistic role of these universal nuclear receptor corepressors is not well understood in the regulation of steroid receptor–mediated activity, including AR. Recently, cyclin D1 and ARA67/PAT1 have been shown to function as AR repressors, mainly by interacting with the NH2-terminal region of AR (32 , 33) . The NH2-terminal enhancer of split (AES) is also known to repress AR activity (34) . Our data demonstrate that HOXB13 functions as a novel repressor for androgen-activated AR transcriptional activity and subsequently modulates the positive growth signal stimulated by AR. Expression of HOXB13 was highly specific to the prostate and HOXB13 did not suppress hormone-activated estrogen receptor transcriptional activity, suggesting that HOXB13 may function as an AR-specific repressor. Using cultured prostate cancer cells and xenograft models, we have also demonstrated that expression of HOXB13 is restricted to AR-expressing prostate cells and is not controlled by androgen. Altogether, HOXB13 may be a very important negative regulator of androgen-activated AR transactivation, which can modulate the suppression of androgen-AR–mediated growth stimulation.
Hox proteins are known to play pivotal roles in maintaining homeostasis in normal cell proliferation and differentiation. Despite intensive research, the mechanism of action of Hox proteins remains unclear. A paradigm that the Hox proteins function as transcription factors was set, merely due to the presence of the DNA-binding homeodomain (35) . However, large numbers of Hox proteins alone do not bind to target DNA with high affinity or specificity (36) , but require other homeodomain proteins to form cooperative DNA-binding complexes, such as Pbx/Exd and Meis/Prep/Htx (37) . Pbx proteins physically interact with the YPWM motif of the Hox proteins, but Hoxb13 does not have a typical YPWM motif and consequently does not interact with Pbx/Exd. A few studies have reported the involvement of Hox proteins in nuclear receptor regulation. For example, Hoxa5 transactivates the progesterone receptor promoter (38) . On the other hand, Hox proteins are also involved in the regulation of coactivator function without DNA binding. Shen et al. (39) reported that most Hox proteins, including the Hox-13 paralog, bind to CBP/p300 coactivators through their homeodomain. CBP/p300 does not form DNA-binding complexes with the Hox proteins, but instead prevents their binding to DNA, suppressing the transactivating function of Hox protein. Conversely, Hox proteins inhibit the activity of CBP, suggesting that Hox proteins may inhibit CBP histone acetyltransferase activity and thus function as repressors of gene transcription. In our study, the AR-suppressive function of HOXB13 was mediated through the interaction with AR although its association with androgen response element remains to be seen, suggesting that HOXB13 may have a transcription-regulating function as a nontranscription factor. To understand the mechanism on HOXB13-mediated suppression of AR activity, communication between HOXB13 and CBP/p300 needs to be clarified.
Recently, we demonstrated that HOXB13 functions as a growth suppressor of AR(−) and HOXB13(−) PC3 cells by the negative regulation of the β-catenin/TCF-4 signal, specifically down-regulating the expression of TCF-4 transcription factor (16) . As demonstrated by others (40 , 41) , we have observed that LNCaP cells did not respond well to β-catenin/TCF-4 signaling due to the low level of TCF-4 expression, suggesting that HOXB13-mediated growth suppression in LNCaP does not result from the suppression of β-catenin/TCF-4 signaling as we observed in PC3 cells. Both androgens and TCF-4 are highly expressed in prostate cancer and generally known to cause increased proliferation of prostate cancer cells (42 , 43) . Therefore, the growth-suppressive function of HOXB13 is thought to be accomplished through two different growth-modulating pathways, depending on the status of endogenous AR. Although there is increased evidence on the potential communication between AR and β-catenin/TCF signaling, such as the physical interaction of AR with β-catenin (44) or TCF-4 (40) , there is little possibility that HOXB13 functions as a mediator to control these two important signaling pathways for the following reasons: (1) AR transactivation is suppressed by overexpression of TCF-4 (activation of β-catenin; refs. 40 and 44 ), which HOXB13 negatively controls, and (2) HOXB13 physically interacted with neither β-catenin nor TCF-4 (data not shown). It is also possible that HOXB13 may use both signaling pathways to suppress the growth of AR-positive cells. Androgen-independent CWR22RV cells express moderate levels of AR and consequently manifest AR activity and β-catenin/TCF signaling activity and TCF-4 expression (40 , 41) . We have observed that HOXB13 regulates both signaling pathways in CWR22RV cells by reporter transcription assay (Fig. 3B ⇓ ; ref. 16 ). Therefore, the loss of AR or inactivation of AR signaling may result in more involvement of HOXB13 in the regulation of β-catenin/TCF signaling, such that cancer cells need to lose HOXB13 expression to acquire a growth advantage. This assumption suggests that the loss of HOXB13 may be an important step in the survival mechanism of androgen-independent prostate cancer cells. At the same time, AR counteracts the suppressor function of HOXB13, so that HOXB13 and AR need to coexist for prostate cancer cells to survive.
We are grateful to Dr. Chinghai Kao for critical discussions and advice and to Juan Antonio Jimenez for scientific editing. We also thank Drs. Jer-Tsong Hsieh and Wytske M. van Weerden for providing RNA from cultured prostate cancer cells and xenograft models, respectively.
Grant support: Indiana University Biomedical Research Fund grant 22-88-430 and Wendy Will Case Cancer Fund.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Chaeyong Jung, Department of Urology Indiana University, 1001 10th Street, OPW320, Indianapolis, IN 46202. Phone: 317-278-7617; Fax: 317-278-3432; E-mail:
- Received April 14, 2004.
- Revision received September 28, 2004.
- Accepted October 11, 2004.
- ©2004 American Association for Cancer Research.