HOXA5 regulates expression of the progesterone receptor.

The majority of breast carcinomas show reduced or no expression of the transcription factor, HOXA5. Recently, we have shown that HOXA5 is a potent transactivator of p53 in breast cells and thus may affect the response of breast cancer cells to DNA damage. To determine whether HOXA5 played a role in growth and homeostasis in breast cells, we studied its interaction with the progesterone receptor. The progesterone receptor (PR) belongs to the superfamily of nuclear receptors whose members co-ordinate morphogenesis of the mammary gland in response to binding to their cognate ligands. An increased expression of the endogenous PR gene was seen in MCF-7 cells following induced expression of an exogenously transfected HOXA5 gene. HOXA5, but not HOXB4, -B5, or -B7 activated the PR promoter in two breast cancer cell lines, MCF-7 and Hs578T. Deletion and mutation analysis of the promoter identified a single HOXA5-binding site required for transactivation of the PR gene by HOXA5. HOXA5 binds directly to this site in the PR promoter. Thus, HOXA5 may behave as a transcriptional regulator of multiple target genes, two among which are p53 and the progesterone receptor.

The majority of breast carcinomas show reduced or no expression of the transcription factor, HOXA5. Recently, we have shown that HOXA5 is a potent transactivator of p53 in breast cells and thus may affect the response of breast cancer cells to DNA damage. To determine whether HOXA5 played a role in growth and homeostasis in breast cells, we studied its interaction with the progesterone receptor. The progesterone receptor (PR) belongs to the superfamily of nuclear receptors whose members co-ordinate morphogenesis of the mammary gland in response to binding to their cognate ligands. An increased expression of the endogenous PR gene was seen in MCF-7 cells following induced expression of an exogenously transfected HOXA5 gene. HOXA5, but not HOXB4, -B5, or -B7 activated the PR promoter in two breast cancer cell lines, MCF-7 and Hs578T. Deletion and mutation analysis of the promoter identified a single HOXA5-binding site required for transactivation of the PR gene by HOXA5. HOXA5 binds directly to this site in the PR promoter. Thus, HOXA5 may behave as a transcriptional regulator of multiple target genes, two among which are p53 and the progesterone receptor.
The proper development of the embryonic body plan depends, in large part, on a family of genes called the HOX genes (1,2). Recently, it was reported that the loss of function of several genes belonging to HOX group 9 impaired proper development of mammary glands in mice during and after pregnancy, thereby leading to a strong deficit in milk production and, hence, an abnormal lactation capacity (3). Our own studies have shown that the expression of one member of this family, HOXA5, is undetectable in nearly 60% of breast cancers (4). One target of HOXA5 action could be the progesterone receptor. The progesterone receptor belongs to a superfamily of nuclear hormone receptors (5,6). Through its binding to progesterone, PR 1 is implicated in the control of proliferation, differentiation, and development of the breast and uterine tissues (5,6). In the breast, while estrogen transmits a proliferative signal, progesterone through its interaction with PR, functions as a modulator of estrogen action, leading to path-ways of differentiation (5,6). The most direct evidence for PR function in the mammary gland comes from studies with mice lacking the PR gene (7). The mammary glands of these mice show incomplete ductal branching and failure of lobulo-alveolar development (7). Few upstream regulators of PR gene expression have been identified (8,9). In this paper, we provide cellular and biochemical evidence that PR is directly regulated by HOXA5.

MATERIALS AND METHODS
Human HOX Recombinant Plasmids-The four HOX cDNAs from pB-SHOXA5 (John. F. Fuller, UCLA), pBSHOXB4 (C.-P. Chang, Stanford University), pBSHOXB5, and pBSHOXB7 (Corey Largman, UCSF), were subcloned into the KpnI and XbaI sites of the mammalian expression vector, pCDM8 (Invitrogen). The nature of the cloned fragments was confirmed by nucleotide sequencing.
Deletion Constructs of Progesterone Receptor Reporter Plasmids-The PR promoter construct was prepared by PCR amplification using the following primers (GenBank TM accession number X69068) PR forward primer 5Ј-ACCTTTCCTCTATCTGCCT-3Ј (nt 15-33) and PR reverse primer 5Ј-GCTTTTCTAACAACGCCTCC-3Ј (nt 1135-1114), cloned into TA cloning vector (Invitrogen, CA). The 1121-bp insert was excised from the TA cloning vector by XhoI-HindIII restriction enzymes and cloned into pGL2 basic luciferase reporter vector (Promega Corp., Madison, WI). The resulting clone was sequenced and is referred to as Ϫ1121-bp PRPLuc. To make deletion constructs, Ϫ1121-bp PRPLuc was digested with KpnI and XhoI to create a 3Ј and 5Ј overhang, respectively, at the multiple cloning site and subjected to exonuclease III nuclease digestion at 30°C. At 30-s intervals, a sample was transferred into the S1 nuclease mixture as per the manufacturer's instructions (Erase a base kit, Promega Corp.). Following ligation and transformation, the clones containing inserts of varying sizes of the promoter region were selected. The clones were designated as follows: Ϫ792-bp PRPLuc, Ϫ460-bp PRPLuc, Ϫ294-bp PRPLuc, Ϫ143-bp PRPLuc, Ϫ67-bp PRPLuc, and Ϫ48-bp PRPLuc. In addition, we generated a clone lacking the region with a canonical HOXA5-binding site located close to the transcription start site. This was done by digesting Ϫ1121-bp PRPLuc with AgeI and HindIII, followed by recessed end filling with Klenow enzyme followed by self-ligation. The resulting clone is referred to as ⌬-68 to Ϫ1-bp PRPLuc.
Site-directed Mutagenesis-The core HOXA5-binding site (nt Ϫ62 to Ϫ59) in the Ϫ1121-bp PRPLuc and Ϫ67-bp PRPLuc construct was mutated using the Altered Sites TM in vitro Mutagenesis System (Promega Corp.) according to the manufacturer's instructions. Mutations were confirmed by nucleotide sequencing.
DNA Transfection and Reporter Plasmid Assay-MCF7 and Hs578T breast cancer cell (ATCC, Manassas, VA) were maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum. Cells were plated at 5 ϫ 10 5 /60-mm dish. 24 h later, cells were transfected (Pan-Vera Corp.) with 2 g of the indicated reporter plasmid and/or 1 g of the expression plasmid pCMVHOXA5, -B4, -B5, or -B7. Transfection efficiency for each assay was assessed by cotransfection of 5 ng of SV40 Renilla luciferase plasmid DNA (Promega Corp.). Luciferase activities were assayed 24 h after transfection using the Dual Luciferase Assay kit (Promega Corp.). The PR promoter-firefly luciferase generated light output was normalized to the light output obtained with Renilla luciferase in each cell line.
The Inducible HOXA5 System-The two plasmid, ecdysone-inducible mammalian expression kit (Invitrogen), was used to generate the clones of MCF-7 cells containing an inducible HOXA5 gene. Briefly, the HOXA5 gene was cloned into the EcoRI site of the vector, pIND, generating the plasmid, pINDHOXA5. MCF-7 cells were transfected with 2 g each of recombinant pINDHOXA5 and pVgRXR (expresses * This work was supported by National Institutes of Health Grants RO1 C48943 (to S. S.) and T32 CA09630 (to M. V.). 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. § These two authors contributed equally to this work. ¶ Present address: Osaka City University Medical School, Osaka 545, Japan.
Gel-shift Assays-Nuclear extracts from SAOS2 cells transiently transfected with pCMV-HOXA5 were prepared, and gel-shift assays were performed as described by Raman et al. (10). The reaction was carried out in a 20-l final volume. Nonspecific binding was eliminated by incubating 2-5 g of extract in 20 mM HEPES-HCl, pH 7.9, 50 mM KCl, 1 mM EDTA, 10 mMMgCl 2 , 1 g of sonicated salmon sperm DNA (200 -300 bp), 10 g of bovine serum albumin, and 6% glycerol for 10 min at 4°C. [␥-32 P]ATP-labeled oligonucleotide probe (approximately 0.3 ng per reaction) was added and incubation continued for a further 15 min, loading buffer was added, and the DNA-protein complexes were separated from the unbound probe in nondenaturing 5% polyacrylamide gels by electrophoresis at 100 V for 3-4 h in 0.25 ϫ TBE (1ϫ: 0.089 M Tris base, 0.089 M boric acid, 0.002 M EDTA). The gels were dried and exposed to Kodak-RP film. The 30-mer oligonucleotides (nt Ϫ70 to nt Ϫ45) containing the canonical HOXA5-binding site within the PR promoter (5Ј-AGATCCTACCGGTAATTGG GGTAGGGAGGG-3Ј), as well as the mutated form (5Ј-AGATCCTACCGGTGGTTGGGG-TAGGGAGGG-3Ј), were synthesized and high performance liquid chromatography-purified. Double-stranded oligonucleotides were endlabeled with [␥-32 P]ATP. To ascertain specificity of binding, unlabeled competitor oligonucleotides were incubated with the protein extract prior to the addition of the labeled oligonucleotide. A supershift assay was also performed by incubating the DNA-protein complex with 2 g of rabbit polyclonal HOXA5 antiserum, AB2 (Babco) for 10 min on ice.

RESULTS AND DISCUSSION
In this paper, we investigated the hypothesis that HOXA5 plays a role in the development and homeostasis of the breast by regulating the levels of the hormone receptor, PR, which is crucial in its growth and differentiation (5). Seven putative HOX-binding sequences were found in the 1.1-kilobase pairs upstream regulatory sequence of the human progesterone receptor gene. The sequence similarity for the HOX-binding sites (Fig. 1, A and B) was determined by using the "sequence find" option within the DNA Strider program.
To investigate a possible regulation of PR by HOX proteins, a transient transfection assay was performed using the ERϩ/ PRϩ human breast cancer cell line, MCF-7. Cells were cotransfected with the reporter plasmid, Ϫ1121 bp PRLuc, together with expression plasmids encoding full-length human HOXA5. Strikingly, we found a 60 -66-fold increase in luciferase activity in MCF-7 cells. No such effect was observed when the reporter plasmid was cotransfected with plasmids encoding the paralogous gene, HOXB5, or an upstream effector of HOXA5 function, HOXB4, or another member of the HOX gene family, HOXB7 (Fig. 1C). None of the HOX proteins affected the renilla luciferase activity of the pRL-SV40-Luc control plasmid (data not shown). To extend these observations, we performed similar experiments in Hs578T cells, a human breast cancer cell line that is ERϪ/PRϪ. Transactivation of the PR promoter by HOXA5 was also observed here, and the level of activation was comparable with that of MCF-7 cells (Fig. 1C). In these cells, HOXB4 showed a low (up to 10-fold), but consistent, level of activation, but HOXB5 and HOXB7 were inactive. These results suggest that in two separate breast cancer cell lines, expression from the PR promoter is highly stimulated specifically by HOXA5, but not by three other members of the HOX gene family. Also, the ER/PR status of the cells did not influence the ability of HOXA5 to transactivate the cells, suggesting that factors responsible for the transactivation effect are present in both cell types.
We also varied the dose of HOXA5 plasmid DNA in experiments in MCF-7 cells. We found that the activation effect increased as the dose was increased from 0.25 to 1 g and then declined when 2 g was used (data not shown). 1 g of expression plasmid was therefore used in all subsequent experiments. A mutant of HOXA5 that introduced a stop codon leading to a C-terminal truncation was also tested in these experiments. This mutant, which lacked the HOXA5 DNA-binding site, was unable to activate expression from the PR promoter (data not shown). These results indicate that HOXA5 can activate transcription of the PR promoter in this system, and the transactivation is due to an interaction of the DNA binding domain of HOXA5 protein with the PR promoter.
To further define the sequence requirements for the transactivation function, deletion constructs of the PR promoter luciferase construct were tested in cotransfection assays with the full-length HOXA5 expression plasmids. Deletion of the Ϫ1121-bp PRPLuc construct to Ϫ792-bp PRPLuc construct, which removes one putative HOX-binding site, resulted in a 27% reduction in luciferase activity in response to HOXA5 in  Fig. 2A) and a 35% reduction in Luc activity in Hs578T cells ( Fig. 2A). A systematic deletion of the next two HOX-binding sites (Ϫ460-bp PRPLuc, Ϫ294-bp PRPLuc) resulted in only a slight decrease in luciferase activity in both the cell lines. A further deletion to Ϫ143 bp (Ϫ143-bp PRPLuc) stimulated luciferase activity 40 -45% more than the fulllength PR-Luc construct in MCF-7 cells but not in Hs578T cells. This increase in Luc activity in MCF-7 cells, but not in Hs578T, suggests the presence of a negative regulatory molecule in Hs578T cells, which is absent in MCF-7 cells, reflecting the genetic heterogeneity between the two cell lines. However, a further deletion to Ϫ67 bp resulted in a 2-fold increase in luciferase activity compared with the Ϫ143-bp PRPLuc construct in both MCF-7 and Hs578T cells. This stepwise analysis suggested that the site important for transactivation by HOXA5 is located between Ϫ67 and Ϫ1. In addition, it is quite likely that the removal of a binding site for negative regulatory molecules between Ϫ67 and Ϫ143 bp within the PR promoter enhanced the positive transcriptional ability of HOXA5. Cotransfection of HOXB4, -B5, or -B7 along with the Ϫ67-bp PRPLuc yielded results similar to those shown in Fig. 1A in MCF-7 and Hs578T cells (Fig. 2B). Further deletion of the last HOXA5-binding site (located at Ϫ62 bp), in constructing the Ϫ48-bp PRPLuc, resulted in a drastic decrease of luciferase activity in both cell lines. However, luciferase activity was not completely abolished, suggesting the presence of a putative transactivator binding sequence in this region. From the results obtained by the deletion analysis, it appeared that the 3Ј-most 67 bp, containing only one HOXA5-binding site and the putative transactivator binding sequence, was sufficient for the transactivation. Finally, the specific deletion of these sequences from the full-length PR-Luc construct completely abolished the ability of HOXA5 to transactivate PR-linked reporter activity, providing confirmation that the 3Ј-most HOXA5-binding site is essential for the activity of PR.

MCF-7 cells (
The HOX-binding site that resides at the 3Ј end of promoter region contains the core HOX motif, TAAT. To determine whether the "TAAT" core-containing site in the PR promoter is a bona fide HOXA5-binding site, electrophoretic mobility shift assays were performed. We tested the binding of oligonucleotides corresponding to this site to HOXA5 protein present in nuclear extracts of SAOS2 cells transfected with pCM-VHOXA5. Western analysis of cell extracts using a HOXA5specific antibody showed the overexpression of the 40 -42-kDa HOX protein in HOXA5-transfected MCF-7 cells (data not shown). Equal amounts of protein from cells transfected with vector DNA and pCMVHOXA5 were used. Nuclear extracts from vector-transfected and HOXA5-transfected MCF-7 cells were incubated with a 32 P-end-labeled 30-mer oligonucleotide probe containing the HOXA5-binding site (Fig. 3A). HOXA5 produced a protein-DNA complex efficiently with the probe containing the wild type sequence (TAAT) (lane 2). Addition of 50-fold molar excess of unlabeled probe resulted in inhibition of this binding (lane 3), while a heterologous probe of a random sequence did not mediate competitive inhibition of protein/ DNA binding (lane 4). Furthermore, when cell extracts were mixed with an oligonucleotide probe that carries two mutations (TGGT) in the core binding site, no protein/DNA complex formation was observed (lane 5). Finally, HOXA5 antibodies (lane 6) caused a supershift of the bound HOXA5 protein-oligonucleotide complex. These results, combined with detailed mutational analysis, indicate that the core sequence TAAT is necessary for this binding. These results clearly show that the TAAT-containing sequence present in the Ϫ67-bp PR promoter is indeed a HOX-binding sequence.
In a final test to confirm that the site lost in the Ϫ48-bp promoter construct is the HOXA5-binding site, we tested the in vivo effects of the same mutations in the core binding site that had abolished DNA-protein complex formation in cell extracts. Transient transfection assays were performed using two Ϫ67-bp PRPLuc reporter constructs, MUT-1 containing two mutations (TGGT) or MUT-2 carrying one mutation (TAGT, MUT-2), in the core-binding site. The results showed that transactivation of the reporter gene was strikingly reduced with both the mutant promoters (Fig. 3B). In addition, we mutated the 3Ј-most HOXA5-binding site within the context of the whole PR promoter construct by converting the TAAT to TAGT. The mutated construct had a 6 -9-fold lower luciferase activity when cotransfected with HOXA5 into MCF-7 and Hs578T cells, compared with the wild type 1211-bp PR promoter construct (Fig. 3C). Thus, by a number of separate and complementary tests, we have shown that transactivation of the Ϫ1121-bp PR promoter-Luc construct is mediated by direct binding of HOXA5 to a single site, at Ϫ62 bp, in this promoter.
The above results suggest that HOXA5 is a transactivator of the PR gene. To test this possibility, we generated stable MCF-7 cultures that could express the HOXA5 gene under the control of an ecdysone-inducible system. Within 3 h after induction of HOXA5 expression by the ecdysone analog, Ponasterone A (Pon A), levels of the endogenous 116-kDa PR-B isoform were induced up to 3-fold (Fig. 4A), when compared with the control vector. The induction of the PR-B isoform, although In summary, we have described an effect of HOXA5 upon the protein levels of PR-B isoform. HOXA5 is a member of a class of proteins known principally for their role in pattern formation during development. Also, an oncogenic function for both murine and human HOX proteins, by overexpression or by untimely expression, has been well established (11,12). While the oncogenic activity of overexpressed Hox genes has substantial experimental support, these observations and our data suggest additional functions for HOXA5. Our recent studies (4) indicate that HOXA5 is a potent transcriptional activator of p53 in breast cells. Breast cancer cell lines and primary breast cancer specimens displayed a co-ordinate loss of p53 and HOXA5 mRNA and protein expression. We proposed that reduced or lack of expression of HOXA5 could lead to loss of p53 expression, providing a mechanism, other than by mutation, for loss of function of this important tumor suppressor during the development of breast cancer. Based upon the results of the study presented here, we raise the possibility that HOX genes play an important role in normal breast development by regulating the expression of the PR gene and that its loss in breast cancer could influence the transcription and thereby the expression of several key genes important for normal differentiation.
In our experiments, HOXA5 is a strong positive regulator of PR. We have yet to define the biological significance of the relationship described here. The major modulators of PR concentration are the ovarian hormone estrogen and progesterone itself (13). However, even though estrogen levels are biphasic in the mammary gland during menstrual cycle in normal women, the concentrations of PR are uniform, suggesting that its synthesis is not under the exclusive control of estrogen in breast tissues (14). Similar results were obtained using estrogen receptor-␣ knock-out mice. In the ER null-homozygous mice, PR mRNA was nevertheless detected, suggesting the existence of both estrogen-dependent and -independent gene regulation (15). In this study we have identified HOXA5 as a novel upstream regulator of PR gene expression in normal breast cells and presented evidence for the presence of an additional factor, other than estrogen, that controls PR expres-

FIG. 4. HOXA5 up-regulates PR levels in breast cancer cells.
Western blot analysis of cell extracts at different times following addition of Pon A to cultures of MCF7-HOX-2. Antibodies used were HOXA5 (HOXA5-2, Babco), PR (C-19, Santa Cruz), and actin (AC-15, Sigma). HOXA5 is frequently seen as a 40 -42-kDa doublet; a third band is seen after Pon A induction of the transfected HOXA5 gene in MCF-7-HOX2 cells. The multiple bands may be due to post-translational modifications or alternate splicing. The smaller molecular weight bands seen in the blot probed for PR protein may represent degraded PR protein products. sion in breast cancer. Similar to its well known role in body patterning during embryonic development as an "master regulator of gene action," it is possible that HOXA5 has multiple roles in breast development and that loss of HOXA5 will have a major impact upon the action of multiple genes important in homeostasis.