Estrogen Receptor Isoform-specific Regulation of the Retinoblastoma-binding Protein 1 (RBBP1) Gene

Estrogen (E2) is involved in mediating many important functions relevant to osteoblast biology through the actions of the estrogen receptors (ER) α and β. To further understand the mechanisms of ER-specific regulation, we used microarray and reverse transcription-PCR analyses of E2-treated U2OS-ERα or -ERβ cells and identified retinoblastoma-binding protein 1 (RBBP1) as a major E2-regulated gene. RBBP1 is a retinoblastoma cofactor involved in the control of osteoblastic proliferation. Although RBBP1 mRNA levels rapidly increased after 2 h of E2 treatment in both U2OS-ER-expressing lines, a sustained induction was only observed in U2OS-ERα cells. Examination of the RBBP1 genomic sequence revealed an ER response element and a Sp1 site located within the first intron. Chromatin immunoprecipitation analyses demonstrated that E2-dependent ERα binding to the intron 1 enhancer region was constitutive, whereas ERβ binding was transient, consistent with the mRNA time course. Interestingly, transient transfection and receptor mutational studies revealed that RBBP1 induction by ERα only requires the Sp1 site, whereas ERβ utilizes both the Sp1 and estrogen response elements binding sites for maximal E2-dependent activation. Stable U2OS transfectants containing a deletion of the ERα activation function 1 (AF1) resulted in a temporal mRNA induction profile similar to that of wild type ERβ. Further, overexpression and chromatin immunoprecipitation analyses also demonstrated that E2-dependent RBBP1 induction is SRC2-dependent for both ER isoforms. These results describe an E2-dependent, ER isoform-specific transcriptional activation of the RBBP1 gene, which in part, is explained by the differential activity of ER AF1 and enhancer element binding.


Estrogen (E2) is involved in mediating many important functions relevant to osteoblast biology through the actions of the estrogen receptors (ER) ␣ and ␤. To further understand the mechanisms of ER-specific regulation, we used microarray and reverse transcription-PCR analyses of E2-treated U2OS-ER␣ or -ER␤ cells and identified retinoblastoma-binding protein 1 (RBBP1) as a major E2-regulated gene. RBBP1 is a retinoblastoma cofactor involved in the control of osteoblastic proliferation. Although RBBP1 mRNA levels rapidly increased after 2 h of E2 treatment in both U2OS-ER-expressing lines, a sustained induction was only observed in U2OS-ER␣ cells. Examination of the RBBP1 genomic sequence revealed an ER response element and a Sp1 site located within the first intron. Chromatin immunoprecipitation analyses demonstrated that E2-dependent ER␣
binding to the intron 1 enhancer region was constitutive, whereas ER␤ binding was transient, consistent with the mRNA time course. Interestingly, transient transfection and receptor mutational studies revealed that RBBP1 induction by ER␣ only requires the Sp1 site, whereas ER␤ utilizes both the Sp1 and estrogen response elements binding sites for maximal E2-dependent activation. Stable U2OS transfectants containing a deletion of the ER␣ activation function 1 (AF1) resulted in a temporal mRNA induction profile similar to that of wild type ER␤. Further, overexpression and chromatin immunoprecipitation analyses also demonstrated that E2-dependent RBBP1 induction is SRC2-dependent for both ER isoforms. These results describe an E2-dependent, ER isoform-specific transcriptional activation of the RBBP1 gene, which in part, is explained by the differential activity of ER AF1 and enhancer element binding.
Recent data from our laboratory (26,27) and others (28,29) have demonstrated that significant differences exist in gene expression patterns, elicited by ER␣ and ER␤ homodimers and ER␣/ER␤ heterodimers, in a human osteosarcoma (U2OS) model system. Specifically, only ϳ20% of all genes regulated through either ER isoform are regulated by both ER isoforms. The mechanism(s) of ER isoform-specific gene regulation are unknown; however because of the low level of homology between the ER␣ and ER␤ A/B (AF1) domains, differences in transcriptional activation may be largely due to AF1 function (30,31).
In this study, we show that the RBBP1 gene, the protein product of which is involved in retinoblastoma-mediated suppression of cell proliferation, is differentially regulated by ER␣ and ER␤ in a U2OS model system. Using deletion and mutational analysis, we demonstrate that ER␣ utilizes a consensus Sp1 sequence, whereas ER␤ functions through both the Sp1 and a nearby ERE sequence. Deletion of the ER␣ AF1 in a stable U2OS transfectant cell line causes the temporal profile to mimic that of wild type ER␤. These data indicate that differen-tial transcriptional activation of RBBP1 by ER␣ and ER␤ and transcription factor binding site utilization are AF1-dependent.
Reverse Transcription PCR (RT-PCR) Analysis -The U2OS-ER␣ and -ER␤ cell lines were plated in 6-well cell culture dishes at a cell density of 50% and pretreated with 100 ng/ml Dox for 24 h in CS-FBS medium. The cells were then treated with 100 ng/ml Dox plus ethanol control or 10 nM E2 for 1, 2, 4, 6, 8, or 24 h. Mouse calvarial osteoblasts were plated and treated with E2 in the same manner as the U2OS-ER cells. The RNA was harvested using TRIzol reagent (Invitrogen). Four g of total RNA was heat-denatured at 68°C for 15 min in a reverse transcription reaction buffer (1ϫ first strand buffer (50 mM Tris-HCl, 75 mM KCl, 3 mM MgCl 2 ), 50 mM dithiothreitol, 1 M dNTPs, 500 ng of oligo(dT) primer). Following heat denaturation, 1 unit of mouse Moloney leukemia virus-reverse transcriptase (Invitrogen) was added, and the mixture was incubated at 37°C for 45 min followed by a 68°C incubation for an additional 15 min. The cDNA products were diluted to 50 l with PCR-grade water; 2 l was used in a PCR reaction using primers specific for human RBBP1 and ␤-actin; and products were separated on a 1.5% agarose gel as described previously (26). Real-time PCR was performed as described previously (32). Primers specific for RBBP1 were as follows: human RBBP1 forward, 5Ј-GCAAAAGCGTACCCCAAAG-CGAAC-3Ј; human RBBP1 reverse, 5Ј-TCCCGCATGAGAC-ACTTCCCTGTC-3Ј; mouse RBBP1 forward, 5Ј-GCAAAAG-CGAACCCCAAAGCGGAC-3Ј; mouse RBBP1 reverse, 5Ј-TCCTGCGTGAGACACTTCTCTGTC-3Ј. ␤-actin primers are described previously (26).
Production of the RBBP1 Intron 1 Enhancer Constructs-RBBP1 intron 1 enhancer constructs were produced by PCR amplification of the desired construct (see Fig. 4) using specific primers containing engineered XhoI restriction sites for cloning into the XhoI site of pGL3-promoter luciferase vector (Promega, Madison, WI). The PCR reactions were as described previously (32) using 100 ng of human genomic DNA as template. Platinum Pfx DNA polymerase (Invitrogen) was used to decrease amplification errors, and the resultant constructs were sequenced to ensure fidelity. The ERE and Sp1 mutations were introduced into the respective amplifying primers with the mutated sequence as seen in Fig. 5A.
Production of the DNA-binding ER Mutants-Construction of the ER␣ and ER␤ P-box mutation expression plasmids was performed using two-step PCR. Briefly, primers were designed to mutate Glu 203 and Gly 204 to alanines for ER␣ and Glu 167 and Gly 168 to alanines for ER␤. The ER␣ double mutation was as reported previously (34). Because ER␣ and ER␤ share 97% identify within the DNA-binding domain, the same mutation was made in ER␤. Resultant products were mixed and amplified using specific flanking primers containing an N-terminal FLAG epitope tag. The resulting PCR fragment was subcloned into the BamHI/XhoI and HindIII/XbaI sites of pcDNA4/TO for the ER␣ and ER␤ P-box mutations, respectively.

Development of the ER-⌬AF1 Expression Constructs and the U2OS-ER-⌬AF1 Cell Lines-Construction of the ER␣-⌬AF1
and ER␤-⌬AF1 expression plasmids was performed using PCR. Briefly, primers were designed to amplify amino acids 174 -595 of human ER␣ and amino acids 130 -530 of human ER␤. An in-frame N-terminal FLAG epitope was included in the forward primers for validation of protein expression (32). The resulting PCR fragment was subcloned into the BamHI/XhoI and HindIII/XbaI sites of pcDNA4/TO for ER␣-⌬AF1 and ER␤-⌬AF1, respectively. The generation of the U2OS-ER-⌬AF1 cell lines was performed using the T-REx system (Invitrogen) as described previously (26). The ER-⌬AF1-pcDNA4/TO vectors were linearized using a PvuI restriction digest and transfected into the U2OS-Tet R cells at a density of 50% using FuGENE 6 transfection reagent (Roche Diagnostics). 48 h later, the transfected cells were split and plated in an antibiotic-containing, selective medium. Individual cell clones were isolated and Western blot analysis performed to confirm expression of both ER␣-⌬AF1 and ER␤-⌬AF1. The resulting cell lines are termed U2OS-ER␣-⌬AF1 and -ER␤-⌬AF1.
Transient Transfection Analyses-U2OS-ER␣ and U2OS-ER␤ cells were plated at a density of 50% in 6-well plates. RBBP1 intron 1 enhancer constructs and ER mutant constructs were transiently transfected in triplicate using FuGENE 6 transfection reagent (Roche Diagnostics) at a concentration of 250 ng/well in CS-FBS-containing medium supplemented with 100 ng/ml Dox for ER induction. The next day, cells were washed twice with 1ϫ phosphate-buffered saline and treated with 100 ng/ml Dox and either ethanol control or 10 nM E2 in serum-free medium for an additional 24 h. Cells were harvested, and 1 l of protein extract was assayed using luciferase assay reagent (Pro-mega). For the SRC overexpression, 250 ng/well of SRC1 or SRC2 was cotransfected using the same procedure.
Western Blot Analysis-The U2OS-ER-⌬AF1 cell lines were plated at a density of 75% in 10-cm culture dishes and treated with 100 ng/ml Dox for 24 h. Total protein extracts were prepared in radioimmune precipitation assay buffer (1% (v/v) Nonidet P-40, 0.5% (v/v) sodium deoxycholate, 0.1% (v/v) sodium dodecyl sulfate, and 1ϫ protease inhibitor mixture (Roche Diagnostics) in 1ϫ phosphate-buffered saline) and the protein concentrations determined using a standard Bradford assay. ER-⌬AF1 protein levels were analyzed by Western blot using a FLAG-M2 antibody (Sigma) and visualized using enhanced chemiluminescence (Amersham Biosciences).

Regulation of RBBP1 by E2 in Osteoblastic Cells-Our previ-
ous microarray data demonstrated that RBBP1 was regulated 4.01-fold following 24 h of E2 treatment in the U2OS-ER␣ cell line, whereas no regulation was detected in the U2OS-ER␤ cell line (27). In contrast, at 2 h of E2 treatment, RBBP1 was regulated 3.74-and 2.55-fold in the U2OS-ER␣ and -ER␤ cell lines, respectively (data not shown). To further understand the complex kinetics of RBBP1 induction by both ER isoforms, we performed an E2 time course from 0 to 24 h of treatment and determined RBBP1 mRNA induction using conventional and real-time RT-PCR analyses. As shown in Fig. 1, RBBP1 expression is induced over control levels following 24 h of E2 treatment in the U2OS-ER␣ cell line. However, in the U2OS-ER␤ cell line, RBBP1 induction appears only transient, peaking at 2 h of E2 treatment and returning to near control levels by 8 -24 h, supporting the microarray expression data. To assess whether RBBP1 is regulated in nontransformed osteoblasts, a similar analysis was conducted using mouse calvarial osteoblasts, which are known to contain significant amounts of both ER isoforms. Fig. 1A demonstrates that RBBP1 induction by E2 is sustained, indicating that RBBP1 regulation by E2 is not specific to the U2OS-ER model and is observed in normal, nontransformed osteoblasts.
The RBBP1 Intron 1 Enhancer Region Contains an ERE and Binds ER␣ and ER␤-In an effort to understand the genetic components of E2 regulation of RBBP1 gene expression, the genomic sequence of the RBBP1 gene was scanned for potential ERE-like sequences. As depicted in Fig. 2, an ERE was found within intron 1 at nucleotide ϩ899 (relative to the transcriptional start site), which closely matched the pure ERE consensus sequence (GGTCAXXXTGACC). Interestingly, a Sp1 transcription factor binding site was also found at nucleotide ϩ949. The core Sp1 sequence and the extended sequence perfectly match the established Sp1 consensus, (G/T)GGGCGG(G/ A)(G/A)(C/T) (35).
ChIP analysis was conducted to determine binding of the ER isoforms in the context of chromatin structure and the extent of binding over a period of time to the RBBP1 intron 1 enhancer region. Fig. 3 demonstrates that significant binding, in vivo, of both ER␣ and ER␤ is observed following 1 h of E2 treatment. More specifically, ER␣ exhibits an E2-dependent binding to the RBBP1 intron 1 enhancer region throughout the 24-h time course, wherein ER␤ displays E2-dependent binding only at 1 h of E2 treatment. It is interesting to note that ER␤ binding is minimal at 2 h following E2 treatment and that the RBBP1 mRNA (see Fig. 1) does not completely decrease to near control levels until 8 h of E2 treatment, suggesting that mRNA stability is playing a role in these differential chronologies. These data are in general agreement with the RT-PCR data (Fig. 1), which suggest that the sustained increase in the expression of RBBP1 in U2OS-ER␣ cells and the temporary increase of RBBP1 expression in U2OS-ER␤ cells correlate with and are probably dependent on ER binding to the RBBP1 intron 1 enhancer region.

Analysis of the RBBP1 Intron 1 Enhancer Region Delineates Different Requirements for ER␣ and ER␤ Binding and Promoter
Activation-To determine the role of the ERE and neighboring trans-acting sites in E2 regulation of RBBP1 by the ER isoforms, various RBBP1 intron 1 enhancer constructs were PCR-amplified, cloned into a luciferase reporter vector, and transiently transfected in both U2OS-ER␣ and -ER␤ cells. Both ERs activate the RBBP1 ϩ412 to ϩ1140 construct, which contains the ERE and Sp1 sites (Fig. 4A). Loss of the ϩ950 to ϩ1140 sequences, including most of the Sp1 site, completely abolished ER␣-dependent activation, whereas only a slight effect was observed on ER␤-dependent activation. This suggests that the FIGURE 1. RBBP1 is regulated by E2 in human osteoblasts. A, U2OS-ER␣, U2OS-ER␤, and mouse calvarial osteoblasts (OBs) were treated for the time points indicated as described under "Experimental Procedures." RNA was harvested and subjected to RT-PCR analysis using primers specific for either human or mouse RBBP1 and ␤-actin. B, real-time RT-PCR was employed on the samples described in A with data expressed as mRNA abundance relative to ␤-actin. Asterisks denote significance at the p Ͻ 0.01 level (ANOVA) compared with control treatment (0 h). The experiment was repeated twice, and a representative experiment is shown.
Sp1 site within this sequence is critically important for ER␣-but not ER␤-dependent transcriptional activation.
To specifically characterize the effects of the ERE and Sp1 sites in E2-dependent activation in the same context, mutations of the ERE and Sp1 sites, alone or in combination, were created in the ϩ851 to ϩ1140 context and transiently transfected in the U2OS-ER cell lines. Fig. 4B demonstrates that mutation of the Sp1 site completely abolishes E2-dependent activation in the U2OS-ER␣ cell line, consistent with the deletion of the Sp1 site in Fig. 4A. Interestingly, no significant effect was observed with the ERE mutation in the U2OS-ER␣ cell line. Mutation of either the Sp1 or ERE decreased ER␤-specific activation by ϳ50% but still exhibited a statistically significant E2 induction, whereas mutation of both sites completely abolished E2-dependent activation in the U2OS-ER␤ cell line. These data suggest differential mechanisms for ER␣versus ER␤dependent activation, wherein ER␣ exclusively utilizes the Sp1 site and ER␤ utilizes both the Sp1 and ERE sites for maximal activation.
ER␣-dependent Activation of the RBBP1 Intron 1 Enhancer Region Is Independent on ERE Binding-Because the RBBP1 intron 1 enhancer region contains both an ERE and Sp1 element, we determined the relative contribution of these elements in the RBBP1 wild type ϩ851 to ϩ1140 context. Jakacka et al. (34) created a mutation in the mouse ER␣ DNA-binding domain, which inhibited ER binding to an ERE but did not affect signaling through non-ERE sites. We therefore created the same mutation in the human ER␣ and ER␤ DNA-binding domains (Fig. 5A) to generate constructs in which signaling through the RBBP1 ERE was eliminated yet still maintained the ability to signal through the RBBP1 Sp1 site. Fig. 5B demonstrates that both the ER␣ and ER␤ P-box mutations significantly inhibit activation through a consensus ERE element when compared with the wild type receptor. Interestingly, the ER␣ P-box mutation had no effect on transcriptional activation of the RBBP1 ϩ851 to ϩ1140 construct, whereas the ER␤ P-box mutation decreased activation by ϳ60% (Fig. 5C). This is in agreement with Fig. 4B demonstrating that activation of the RBBP1 intron 1 enhancer region through ER␣ is ERE-independent, whereas activation through ER␤ is at least partially ERE-dependent.
The ER␣ AF1 Domain Is Important for Sustained Activation of RBBP1 Intron 1 Enhancer Activity-To test the hypothesis that the ER A/B domain (AF1) was involved in mediating ER isoform-specific activation of the RBBP1 gene, N-terminal deletions of both ER isoforms were created (Fig. 6A). The ER␣-⌬AF1 construct, starting at Met 174 of ER␣, matches the previously described 46-kDa form of ER␣, which is expressed in normal osteoblasts (36,37). Therefore, wild type and ER-⌬AF1 constructs were transiently cotransfected with the RBBP1 ϩ851 to ϩ1140 reporter plasmids (Fig. 4B) in U2OS cells. The data presented in Fig. 6B closely resembles the transfection data from the U2OS-ER stable transfectants (Fig. 4B), wherein the activity of ER␣ is purely Sp1-dependent (ERE mutation has no effect), and maximal activity of ER␤ is dependent on both the ERE and Sp1 sequences. Coexpression of ER␣-⌬AF1 with the wild type RBBP1 ϩ851 to ϩ1140 construct (Fig. 6C) leads to an increased -fold activation by E2 over wild type ER␣ (4.96versus FIGURE 2. The RBBP1 intron 1 region contains both ERE and Sp1 binding sites. A, a schematic diagram of the RBBP1 intron 1 enhancer region (ϩ400 to ϩ1140, relative to the transcriptional start site) is depicted. B, the ERE (ϩ899) and Sp1 (ϩ949) sequences from the RBBP1 intron 1 enhancer is depicted with comparisons to the consensus binding sites.

FIGURE 3. ChIP analysis of the RBBP1 intron 1 enhancer region demonstrates ER binding.
A, ChIP assays were performed in U2OS-ER␣ and -ER␤ cell lines treated with 10 nM E2 for the indicated times using antibodies specific for either ER␣ or ER␤ as described under "Experimental Procedures." The resultant products were separated using agarose gel electrophoresis. B, realtime PCR analysis was employed on samples from three independent ChIP experiments. Asterisks denote significance at the p Ͻ 0.01 level (ANOVA) compared with control treatment (0 h) within each cell line.
1.99-fold), indicating that the ER␣ AF1 may play a suppressor role in E2-dependent activation of RBBP1. Interestingly, the ER␣-⌬AF1 construct is now sensitive to the RBBP1 ERE mutation as is normally observed with wild type ER␤ (Fig. 6, compare ER␤ WT (B) with ER␣-⌬AF1 (C)). No significant change was observed with the ER␤-⌬AF1 compared with wild type ER␤ with any of the RBBP1 constructs. These data suggest that the ER␣ AF1 is involved in mediating specific interactions with the RBBP1 Sp1 site and inhibition of the use of the RBBP1 ERE sequence because deletion of the ER␣ AF1 now allows signaling through the ERE.
Because the data in Fig. 6 suggest that the deletion of the ER␣ AF1 results in an "ER␤-like" pattern of E2-inducible expression, we determined the effect of ER␣-⌬AF1 on endogenous RBBP1 expression. Therefore, we constructed stable U2OS cells lines expressing either the ER␣or ER␤-⌬AF1 construct. As seen in Fig. 7A, the U2OS-ER␣-⌬AF1 and -ER␤-⌬AF1 cell lines express similar levels of ER to those observed in U2OS-ER wild type cell lines and correspond to the expected size (ϳ46 kDa). We next examined the temporal E2 induction profile of endogenous RBBP1 in the U2OS-ER␣-⌬AF1 and -ER␤-⌬AF1 cell lines. The U2OS-ER␣-⌬AF1 cell line exhibited a transient increase in RBBP1 expression peaking at 2 h of E2 treatment, whereas the U2OS-ER␣ cell line exhibited a sustained increase in RBBP1 expression (Fig. 7B) as observed previously in Fig. 1. Interestingly, the U2OS-ER␣-⌬AF1 expression pattern is similar to that of U2OS-ER␤. The U2OS-ER␤-⌬AF1 cell line exhibits an overall decreased level of RBBP1 expression at 2 h of E2 treatment with minor induction observed at later time points. Collectively, these data suggest that the ER␣ AF1 confers ER␣-specific patterns of regulation in both transient transfection analysis and endogenous RBBP1 expression.
SRC2 Preferentially Enhances E2-induced RBBP1 Intron 1 Enhancer Activity-To investigate the possible role of a differential SRC coactivator recruitment to explain the ER isoform-specific regulation of RBBP1, we overexpressed SRC1 and SRC2 with the ϩ851 to ϩ1140 construct in both U2OS-ER cell lines. SRC3 was not considered because a previous study demonstrated little or no SRC3 expression in osteoblasts (32). Fig. 8A demonstrates that SRC1 overexpression has no effect on E2-dependent transcriptional coactivation, whereas SRC2 has a significant effect in both U2OS-ER cell lines. In fact, SRC1 overexpression appears to inhibit the E2-dependent response to both ER isoforms. We repeated the experiment using the vitellogenin A2 ERE element to verify that our SRC overexpression was effective (38). Fig. 8B demonstrates that both SRC1 and SRC2 equally and effectively increase E2-dependent transcription with both ER isoforms with the vitellogenin A2 ERE. ChIP analysis was performed to determine SRC recruitment to the RBBP1 intron 1 enhancer region and demonstrated that no detectable SRC1 was recruited, whereas significant SRC2 protein was recruited in an E2-dependent manner (Fig. 8, C and D), which supports the transient transfection data (Fig. 8A). These results confirm that SRC2 is the preferred coactivator recruited to the RBBP1 intron 1 enhancer region. Interestingly, mutation of the RBBP1 ERE to the vitellogenin A2 consensus, in the RBBP1 ϩ851 to ϩ1140 context, does not FIGURE 4. Mutational analysis of the RBBP1 intron 1 enhancer region results in ER isoform-specific effects. A, U2OS-ER␣ and -ER␤ cells were transiently transfected in triplicate with the indicated RBBP1 intron 1 enhancer region constructs as described under "Experimental Procedures." -Fold regulation relative to ethanol control was determined for each of the constructs, which were assayed in at least three independent experiments. Individual errors were less than 10% of the mean for each experiment. The black bars represent the mean Ϯ S.D. of n number of trials. Asterisks denote significance at the p Ͻ 0.01 level (ANOVA) compared with the ϩ412 to ϩ1140 construct. B, cells were treated, transiently transfected, assayed, and analyzed with either ERE or Sp1 mutations in the ϩ851 to ϩ1140 context. Asterisks denote significance at the p Ͻ 0.01 level (ANOVA) compared with ϩ851 to ϩ1140 wild type construct. C, the wild type sequences of the ERE and Sp1 sites of the RBBP1 intron 1 enhancer region is depicted with the mutations listed in bold.
rescue the ability of SRC1 to function in RBBP1 activation (data not shown). This demonstrates that not only the ERE sequence but also the surrounding DNA context is important for the induction of RBBP1 by E2.

DISCUSSION
This paper presents evidence that ER isoform-specific transcriptional activation of the RBBP1 gene is because of the AF1 activity. We chose the RBBP1 gene because previous microarray and RT-PCR analyses demonstrated that RBBP1 was a highly E2-regulated gene, and ER␣ and ER␤ exhibited a differential induction pattern. Furthermore, RBBP1 is a known Rb cofactor and may have an involvement in the control of osteoblastic proliferation and differentiation. We demonstrate that the RBBP1 gene displays a unique regulatory response by which a sustained induction is observed with ER␣, whereas only a transient induction is seen with ER␤. This pattern was paralleled by ChIP assays demonstrating that differential ER occupancy of the RBBP1 intron 1 enhancer region was most likely responsible for the observed expression time course. Mutational analysis demonstrated different cis-acting requirements for activation of RBBP1 by the ER isoforms. ER␣ required the presence of the Sp1 site, whereas ER␤ was capable of E2 induction through either the ERE or Sp1 sites. Interestingly, deletion of the ER␣ AF1 domain elicited a temporal transcription profile similar to that of ER␤, suggesting that differences in the ER AF1 domains play a critical role in E2-dependent activation of the RBBP1 gene.
There are numerous reports describing the functional interactions of ER␣ and Sp1 in the E2-dependent regulation of gene Asterisks denote significance at the p Ͻ 0.01 level (ANOVA) compared with the wild type construct. The bars represent data from at least three independent experiments. C, identical to B, except it is for the RBBP1 ϩ851 to ϩ1140 reporter construct.  ). B, the ER␣ or ER␤ WT constructs were cotransfected with the indicated RBBP1 intron 1 enhancer region mutation constructs in U2OS cells, treated, and analyzed as in Fig. 4. Asterisks denote significance at the p Ͻ 0.01 level (ANOVA) compared with the wild type construct. The bars represent data from three independent experiments. C, idntical to B except with the ER␣-or ER␤-⌬AF1 expression constructs. SEPTEMBER 29, 2006 • VOLUME 281 • NUMBER 39 expression (39 -50). Furthermore, the N-terminal region of ER␣ is known to physically interact with the C-terminal zincfinger region of Sp1 (47,50,51). This functional and physical evidence suggests that DNA-bound Sp1 recruits the ER, which is necessary for transcriptional activation of numerous genes. Indeed, our data suggest that the Sp1 site is necessary and sufficient for ER␣-specific transcription of the RBBP1 intron 1 enhancer region. However, comparatively little is known concerning ER␤ and Sp1 interactions and their effects on transcriptional activity. Saville et al. (47) demonstrated that although both ER isoforms bind Sp1 in a ligand-independent manner, only the ER␣-Sp1 interaction is transcriptionally competent on a synthetic Sp1 reporter construct, whereas the ER␤-Sp1 interaction fails to activate transcription. In contrast, Salvatori et al. (52) demonstrated that ER␤-Sp1 interactions result in tran-scriptional activation of an epidermal growth factor receptor promoter construct. Our data using the RBBP1 intron 1 ERE enhancer domain demonstrate that ER␤ can support E2-inducible transcription through both the ERE and Sp1 sites. The apparent discrepancy in the literature of the effects ER␤-Sp1 interactions on transcription may be because of cell type and promoter/enhancer context specificity.

RBBP1 Is Regulated by Estrogen in Osteoblasts
Our laboratory has previously reported that in our U2OS-ER cell models, ER␣ and ER␤ homodimers as well as ER␣/ER␤ heterodimers appear to regulate distinct cadres of genes with only 20% of all E2-regulated genes being regulated by both ER isoforms (26,27). These data have been confirmed by other laboratories (28,29). Our laboratory reported a similar ER isoform specificity using the selective estrogen receptor modulator 4-hydroxytamoxifen and further showed that E2 and 4-hydroxytamoxifen regulated unique patterns of gene expression. The data presented in this paper begin to delineate the mechanism of these gene-specific inductions by the ER isoforms. Although ER isoform-specific interactions of coregulators cannot be ruled out, at least with respect to the RBBP1 gene, the FIGURE 7. Deletion of the ER␣ AF1 domain results in modulation of endogenous RBBP1 mRNA expression. A, stable U2OS cell lines expressing the ER␣-or ER␤-⌬AF1 expression construct were generated as described under "Experimental Procedures." Western blot analysis of the U2OS-ER␣, -ER␤, -ER␣-⌬AF1, and -ER␤-⌬AF1 was performed using an anti-FLAG antibody. The observed bands migrate at the expected molecular size. B and C, E2 treatment time courses were performed in the U2OS-ER␣, -ER␤, -ER␣-⌬AF1, and -ER␤-⌬AF1 cell lines as described in the legend to Fig. 1B. Real-time RT-PCR was employed with data expressed as mRNA abundance relative to ␤-actin. Asterisks denote significance at the p Ͻ 0.01 level (ANOVA) compared with control treatment (0 h) for each cell line. The bars represent data from three independent experiments. FIGURE 8. SRC2 preferentially activates and binds to the RBBP1 intron 1 enhancer region. A and B, U2OS-ER␣ and -ER␤ cells were transfected in triplicate with either the RBBP1 intron 1 enhancer ϩ851 to ϩ1140 construct or the consensus vitellogenin A2 ERE construct, alone or in combination with SRC1 or SRC2. Cells were treated and assayed for luciferase activity as described in the legend to Fig. 4. A single asterisk denotes significance at the p Ͻ 0.01 level (ANOVA) compared with the control treatment for each cell line and construct, whereas a double asterisk denotes significance at the p Ͻ 0.01 level (ANOVA) compared with the E2 treatment in the absence of SRC overexpression for each cell line and construct. C, ChIP assays were performed in U2OS-ER␣ and -ER␤ cell lines treated with 10 nM E2 for 2 h using antibodies specific for either SRC1 or SRC2 as described under "Experimental Procedures." D, real-time PCR analysis was employed on samples from three independent ChIP experiments. An asterisk denotes significance at the p Ͻ 0.05 level (ANOVA) compared with ethanol control. differential interactions of the ERs with the ERE enhancer domains/elements, the differential time course of ER occupancy on these elements, and the ER AF1 domain appear to play a large role in the ER isoform-specific regulation. Merot et al. (31) previously demonstrated that in well differentiated E2-sensitive tissues/cell lines, ER␣ activity is largely influenced by AF1. Our data support this model because deletion of ER␣ AF1 caused an ER␤-like temporal pattern in the E2-dependent activation of the RBBP1 gene.
Previous studies have reported the existence of a natural splice variant of ER␣ that lacks the first 173 amino acids of ER␣ (termed hER␣46) and is consequently identical to the ER␣-⌬AF1 receptor described in this study (36). hER␣46 is found to be an effective E2-inducible transcription factor and exhibits a strong inhibitory effect on wild type ER␣ (hER␣66) transactivation, possibly because of interference with coregulator binding to the hER␣46/hER␣66 heterodimer (36,53). Furthermore, hER␣46 form is demonstrated to be naturally expressed in human primary trabecular osteoblasts (37). Data from this study suggest that ER␣-⌬AF1 (e.g. hER␣46) is capable of RBBP1 activation in U2OS cells, however, in a manner reminiscent of wild type ER␤ and not wild type ER␣. Because the ER-⌬AF1 cell system in this paper expresses only ER-⌬AF1, future studies are needed to determine the interactions between ER-⌬AF1 mutants and wild type ERs in the same cell on RBBP1 regulation.
We previously reported decreased E2-dependent cell proliferation in the U2OS-ER␣ cell line, whereas no effect was observed in the U2OS-ER␤ cell line (26,27). The specific gene(s) involved in the suppression of osteoblastic proliferation are currently unknown. RBBP1, in concert with Rb, has been previously shown to decrease E2F1 promoter activity, to decrease cellular proliferation, and to be involved in transcriptional repression (54 -56). Interactions of Rb with Runx2, a major regulator of osteogenesis, result in decreased cellular proliferation and increased differentiation (57). Whether RBBP1 has a role, if any, in E2-dependent regulation of cellular proliferation or differentiation requires further investigation.
In conclusion, the current study describes RBBP1 as a novel E2-regulated gene in osteosarcoma cells that exhibit a differential response to the ER isoforms. The ER isoform-specific regulation of RBBP1 expression appears to be caused by the differential occupancy of the ER isoforms on the RBBP1 intron 1 ERE enhancer domain, which in turn appears to involve ER isoformspecific interactions with specific enhancer elements (e.g. ERE, Sp1). Furthermore, it appears that the ER␣ AF1 is involved in imparting ER␣-specific patterns to E2-dependent RBBP1 activation. Further experimentation is required to expand this hypothesis to other ER␣-regulated genes.