S. S. Mandal, Department of Chemistry and Biochemistry, The University of Texas at Arlington, Arlington, TX 76019, USA Fax: +1 817 272 3808 Tel: +1 817 272 3804 E-mail: firstname.lastname@example.org
HOXC13, a homeobox-containing gene, is involved in hair development and human leukemia. The regulatory mechanism that drives HOXC13 expression is mostly unknown. Our studies have demonstrated that HOXC13 is transcriptionally activated by the steroid hormone estrogen (17β-estradiol; E2). The HOXC13 promoter contains several estrogen-response elements (EREs), including ERE1 and ERE2, which are close to the transcription start site, and are associated with E2-mediated activation of HOXC13. Knockdown of the estrogen receptors (ERs) ERα and ERβ suppressed E2-mediated activation of HOXC13. Similarly, knockdown of mixed lineage leukemia histone methylase (MLL)3 suppressed E2-induced activation of HOXC13. MLLs (MLL1–MLL4) were bound to the HOXC13 promoter in an E2-dependent manner. Knockdown of either ERα or ERβ affected the E2-dependent binding of MLLs (MLL1–MLL4) into HOXC13 EREs, suggesting critical roles of ERs in recruiting MLLs in the HOXC13 promoter. Overall, our studies have demonstrated that HOXC13 is transcriptionally regulated by E2 and MLLs, which, in coordination with ERα and ERβ, play critical roles in this process. Although MLLs are known to regulate HOX genes, the roles of MLLs in hormone-mediated regulation of HOX genes are unknown. Herein, we have demonstrated that MLLs are critical players in E2-dependent regulation of the HOX gene.
Homeobox-containing genes are key players in embryogenesis and development [1,2]. Misregulation of homeobox genes is associated with tumorigenesis. More than 200 homeobox-containing genes have been identified in vertebrates, and they have been classified into two major groups, class I and II. Class I homeobox-containing genes share a high degree of identity (more than 80%) and are called HOX genes. In humans, there are 39 different HOX genes, clustered into four different groups, called HOXA, HOXB, HOXC, and HOXD, located on chromosomes 7, 17, 12, and 2, respectively [1,2]. Each of these HOX genes plays critical roles in embryogenesis and organogenesis. The nature of a body structure depends on the specific combination of HOX gene products, and the expression of specific HOX genes varies at different stages of development. Therefore, proper regulation and maintenance of HOX genes are essential for normal physiological functions and growth.
HOXC13 is a critical gene involved in the regulation of the hair keratin gene cluster and alopecia [3–5]. Transgenic mice overexpressing HOXC13 in differentiating keratinocytes of hair follicles develop alopecia, accompanied by a progressive pathological skin condition that resembles ichthyosis [4,5]. HOXC13 mutant mice lack external hair, suggesting a critical role for the gene in hair development . HOXC13 has also been found to be a fusion partner of NUP98 in adult acute myeloid leukemia [6,7]. This protein also binds to the ETS family transcription factor PU.1 and affects the differentiation of murine erythroleukemia . Although HOX13 is critical player in hair development and disease, little is known about its own regulation. Steroid hormones are critical players in sexual differentiation. Steroid hormones such as estrogen (17β-estradiol; E2) and androgens are also linked with hair follicle growth and differences in hair patterning between males and females [9,10]. However, the molecular mechanism of the roles of steroid hormones in hair development is poorly understood. Herein, we have investigated whether HOXC13, a critical player in hair follicle development, is regulated by steroid hormones.
Mixed lineage leukemia histone methylases (MLLs) are human histone H3 lysine 4 (H3K4)-specific histone methyltransferases (HMTs) that play critical roles in gene activation. MLLs are key players in HOX gene regulation [11–22]. MLLs are also well known to be rearranged in acute lymphoblastic and myeloid leukemias [12,15]. In humans, there are several MLL families of proteins, such as MLL1, MLL2, MLL3, and MLL4. Each of them possesses H3K4-specific HMT activity and exists as a multiprotein complex with several common protein subunits [12,23,24]. Recently, we demonstrated that human CpG-binding protein interacts with MLL1, MLL2, and hSet1, and regulates the expression of MLL target HOX genes . Studies from our laboratory (and others) have demonstrated that MLLs are important players in cell cycle regulation and stress responses [25–33]. Knockdown of MLL1 resulted in cell cycle arrest at the G2/M phase .
Recent studies have demonstrated that several MLLs (MLL2, MLL3, and MLL4) act as coregulators for E2-mediated activation of E2-sensitive genes [12,35–38]. MLL2 interacts with E2 receptor (ER) in an E2-dependent manner, and regulates the activation of cathepsin D [35,38]. MLL3 and MLL4 regulate the E2-sensitive gene encoding liver X-receptor [36,39,40]. Although MLLs are recognized as major regulators of HOX genes during embryogenesis, they are not implicated in steroid hormone-mediated HOX gene regulation. Herein, we have investigated the roles of the MLL family of HMTs in E2-mediated regulation of HOXC13. Our results show that HOXC13 is transcriptionally regulated by E2, and that MLLs, in coordination with ERs, regulate E2-induced activation of HOXC13.
HOXC13 is transcriptionally regulated by E2
ERs are major players in E2-mediated regulation of E2-responsive genes [41,42]. In general, upon binding to E2, ERs are activated. The activated ERs bind to E2 response elements (EREs) present in the promoter of E2-responsive genes, leading to their transcriptional activation . In this work, before examining the E2-mediated regulation of HOXC13, we analyzed its promoter for the presence of any EREs. Our results demonstrated that the HOXC13 promoter contains six putative EREs (ERE1/2 sites) within −1 to −3000 bp upstream of the transcription start site (Fig. 1). All of the EREs show 100% homology with ERE1/2 sites (GGTCA or TGACC) but not with the consensus full ERE sequence (GGTCAnnnTGACC). The presence of these EREs in close proximity to the transcription start site indicated that HOXC13 might be potentially regulated by E2 via the involvement of ERs.
In order to examine whether HOXC13 is regulated by E2, we treated a steroidogenic human cell line (JAR cells, of choriocarcinoma placental origin, cultured in phenol-red free medium containing activated charcoal-treated fetal bovine serum) with different concentrations (1–1000 nm) of E2 for 8 h. The RNA was isolated from the E2-treated cells and analyzed by RT-PCR, using HOXC13-specific primers (Fig. 2; Table 1). Interestingly, our results demonstrated that HOXC13 was overexpressed upon exposure to E2 in a concentration-dependent manner (Fig. 2A,B). In comparison with the control, HOXC13 expression was four-fold to five-fold higher in the presence of 100 and 1000 nm E2 (Fig. 2A,B; compare lane 1 with lanes 5 and 6). As 100 nm was most effective, we analyzed HOXC13 expression using an E2 concentration range closer to 100 nm (20, 50, 100 and 250 nm), and found that 100 nm was the optimal concentration for the E2-mediated induction of HOXC13 (data not shown). The stimulation of HOXC13 expression upon exposure to E2 demonstrated that HOXC13 is transcriptionally regulated by E2. Time-dependence experiments demonstrated that HOXC13 activation was maximum after 6–8 h of E2 treatment (Fig. 2C,D; with 100 nm E2, lanes 4 and 5).
Table 1. Primers used for RT-PCR, ChIP and antisense oligonucleotide experiments.
Forward primer (5′- to 3′)
Reverse primer (5′- to 3′)
a Phosphorothioate antisense oligonucleotide.
ERs play a critical role in E2-induced HOXC13 expression
In order to examine the potential role of ERs in E2-induced activation of HOXC13, we knocked down ERα and ERβ separately, using specific antisense oligonucleotides, in JAR cells and exposed the cells to 100 nm E2 for an additional 8 h. A scramble antisense oligonucleotide (with no homology to ERs) was used as a negative control. Our results demonstrated that application of ERα or ERβ antisense oligonucleotide knocked down the respective genes efficiently, at both the mRNA and the protein level (Fig. 3A,B, lanes 4–6, and data not shown; the quantifications are shown in the respective bottom panels). After confirming effective knockdown, we analyzed the RNA from these ER knockdown and E2-treated cells for the expression levels of HOXC13 using RT-PCR. As seen in Fig. 3A,B, HOXC13 expression was increased upon exposure to E2 (compare lanes 1 and 2), and application of scramble antisense oligonucleotide did not have any significant effect on E2-mediated activation of HOXC13. Interestingly, upon knockdown of either ERα or ERβ, the E2-dependent activation of HOXC13 was suppressed almost to the basal level (Fig. 3 A,B, compare lanes 5 and 6 with lanes 1 and 2; quantifications are shown in the respective bottom panels). These results demonstrated that both ERα and ERβ are essential for E2-mediated transcriptional activation of HOXC13.
MLLs play critical roles in E2-induced HOXC13 expression
As MLLs are well known as master regulators of HOX genes, and several MLLs are implicated in E2 signaling, we examined whether any of the MLLs are involved in E2-dependent stimulation of HOXC13 expression. We knocked down different MLL genes (MLL1, MLL2, MLL3, and MLL4) separately by using specific phosphorothioate antisense oligonucleotides, and then exposed the cells to E2 (100 nm for 8 h). Before performing E2-related experiments, we examined the efficacies of different MLL (MLL1–MLL4)-specific antisense oligonucleotides and their most effective doses. The specific MLL knockdowns were confirmed by analyzing their respective gene expression at both the mRNA and protein levels (data not shown). On the basis of these initial experiments, we applied the specific concentration of each of the MLL antisense oligonucleotides that showed the most effective knockdown of the respective gene and then exposed the cells to E2 (100 nm for 8 h) in an MLL knockdown environment. In parallel, we also applied a scramble antisense oligonucleotide (no homology with any of the MLLs) as a negative control. As seen in Fig. 4A, upon application of MLL1 antisense oligonucleotide followed by exposure to E2, MLL1 was efficiently knocked down, whereas scramble antisense oligonucleotide had no significant effect on the level of MLL1 mRNA. Interestingly, upon downregulation of MLL1, E2-mediated upregulation of HOXC13 was slightly decreased (Fig. 4A, lane 3). Similar results were observed for MLL2 and MLL4 downregulation (Fig. 4B,D). The knockdown of MLL3 almost abolished the E2-mediated activation of HOXC13 (Fig. 4C). These results demonstrated that the MLL family of HMTs, especially MLL3, play critical roles in the E2-mediated regulation of HOXC13.
E2-induced recruitment of ERs and MLLs in the HOXC13 promoter
As the HOXC13 promoter contains several ERE1/2 regions within the first 3000 nucleotides upstream of the transcription start site, we analyzed the involvement of some of these EREs (ERE1–ERE4, located at −234, −1260, −1788 and −2000 bp upstream) by analyzing the in vivo binding of ERs and MLLs. We analyzed the in vivo binding of the different factors in the absence and presence of E2, using chromatin immunoprecipitation (ChIP) assays , using antibodies against ERs and MLLs. ChIP experiments were also performed in parallel with the use of antibody against actin as a nonspecific negative control. In brief, JAR cells were treated with 100 nm E2 for 6 h, and control and E2-treated cells were then subjected to ChIP analysis. The immunoprecipitated DNA fragments were PCR amplified using primers specific for ERE1, ERE2, ERE3 and ERE4 of the HOXC13 promoter. As seen in Fig. 5A, no significant binding of actin was observed in any of the EREs, irrespective of the absence and presence of E2. Binding of ERα and ERβ was increased in both ERE1 and ERE2 of the HOXC13 promoter (Fig. 5A, lanes 1–4). The levels of E2-induced binding of ERα and ERβ were higher in ERE2 than in ERE1. ERE3 and ERE4 were not sensitive to ER binding as a function of E2, probably because of their distance from the transcription start site, although some amount of constitutive binding was observed in both regions.
The binding profiles of different MLLs were interesting. First, although some amount of binding of MLL1 was observed in ERE3, no significant E2-dependent binding of any of the MLLs was observed in ERE3 and ERE4 (Fig. 5A, lanes 5–8). Significant amounts of constitutive binding of MLL1, MLL3 and MLL4 were observed in ERE1, even in the absence of E2 (Fig. 5A, lane 1). However, MLL2 binding to ERE1 was enhanced upon addition of E2 (Fig. 5A, lanes 1 and 2). Interestingly, binding of all of the MLLs (MLL1–MLL4) was greatly enhanced upon addition of E2 in ERE2 (Fig. 5A, lanes 3 and 4). These results demonstrated that ERE1 (−234 bp) and ERE2 (−1260 bp), which are close to the transcription start site, are mostly responsible for E2-dependent binding of ERs and MLLs and hence the regulation of HOXC13. ERE2 appeared to have more critical roles (sensitivity to E2) than the other EREs examined. ERE3 and ERE4, which are located far upstream (−1788 bp or further), were not sensitive to E2-dependent binding of any of the MLLs/ERs, indicating no significant roles of these EREs in HOXC13 activation (Fig. 5A).
To further confirm the E2-dependent binding of ERs and MLLs to the HOXC13 promoter, we analyzed their binding profiles in a time-dependent manner in ERE1 and ERE2 (Fig. 5B). In agreement with the above findings, binding of ERα and ERβ was increased in both ERE1 and ERE2 in the presence of E2. Interestingly, the kinetics of E2-dependent binding of ERα and ERβ to both ERE1 and ERE2 are different. The binding of ERα is very low in the absence of E2, and is significantly enhanced in the presence of E2 n both ERE1 and ERE2. However, in the case of ERβ, some constitutive binding was observed in ERE2 even in the absence of E2, and this binding was increased in the presence of E2 (Fig. 5A,B; compare 0 h and 6–8 h time points). These differences in the kinetic profiles of binding of ERα and ERβ suggest that they have distinct modes of action in regulating target gene activation. It is important to mention that, although it is poorly understood, the difference in the kinetics of binding of ERα and ERβ to the target gene promoters has been previously observed by other laboratories .
E2-dependent binding of MLLs (MLL1–MLL4) was primarily observed in ERE2 (Fig. 5B). Again, as seen above, MLL2 binding was observed in ERE1 as a function of E2 (Fig. 5B, left panel). The E2-dependent increase in binding of MLLs to the EREs were observed at as early 30 min post-E2 exposure, and increased with time, reaching a maximum at ∼ 6 h (Fig. 5B). The binding of MLL3 to ERE2 appeared to be most prominent, although E2-induced binding of other MLLs (MLL1, MLL2, and MLL3) was also significant (Fig. 5B). In addition, we also analyzed the status of RNA polymerase II (RNAPII) and H3K4-trimethylation level in ERE1 and ERE2. Our results demonstrated that in both ERE1 and ERE2, the levels of RNAPII and H3K4-trimethylation were increased in the presence of E2 (Fig. 5B). These results demonstrated that both ERE1 and ERE2 (especially ERE2) coordinate the binding of ER and MLL coregulators as well as RNAPII, and regulate the E2-mediated transcriptional activation of HOXC13. It is important to note that although ERE2 is located far upstream (1260 bp away from the transcription start site), we still observed significant transcription-dependent increases in RNAPII binding to these EREs. These observations suggest that there is probably a looping of the large promoter regions so that far upstream cis-elements could be placed closer to the promoter proximal sites and coordinate with RNAPII and other transcription factors during transcription initiation [45,46].
In addition, binding of some MLLs to certain EREs even prior to the addition of E2 suggests that this binding might be linked to the basal transcriptional regulation of the gene. Furthermore, we also observed that the recruitment of MLL2 is induced by E2 at both ERE1 and ERE2. However, the recruitment of other MLLs (i.e. MLL1, MLL3, and MLL4) at ERE1 is not induced by E2 (Fig. 5). These differences in recruitment profiles can be attributed to different possibilities. One of the possibilities is that, even if there is an ERE, it may not be responsive (not participating in the activation) all of the time, probably because of the presence of other EREs that are more appropriately positioned to coordinate with transcription factors and coactivators to initiate efficient transcription. The other possibility is that, in addition to ERE1/2 sites, other neighboring promoter elements coordinate with it, and that this ultimately drives the assembly of the MLLs and other coregulator complexes around the specific ERE.
Recruitment of MLLs to the HOXC13 EREs is mediated via ERs
ERs are well known to bind directly to the EREs of the E2-responsive genes via their DNA-binding domains. MLLs (MLL1–MLL4) also have DNA-binding domains that might be involved in direct binding of promoters. This binding may be critical for regulation of basal transcription of the target genes. On the other hand, MLLs might be recruited to the HOXC13 promoter via protein–protein interactions (direct or indirect) with ERs. Amino acid sequence analysis demonstrated that MLL1–MLL4 have LXXLL domains [also called nuclear receptor (NR) boxes], which are known to be involved in E2-dependent interactions with ERs . MLL1 has only one LXXLL domain, whereas MLL2, MLL3 and MLL4 have multiple LXXLL domains . In fact, MLL2, MLL3 and MLL4 have recently been shown to interact with ERs, and are involved in the E2-mediated activation of E2-responsive genes [12,35–38]. In the present study, we examined whether all of the MLLs that are involved in the E2-mediated activation of HOXC13 directly bind to the EREs, or whether they are recruited to EREs via interactions with ERs in an E2-dependent manner. To examine this, we knocked down both ERα and ERβ separately, exposed the cell to 100 nm E2 for 6 h, and analyzed the status of the binding of all the MLLs to ERE1 and ERE2 of the HOXC13 promoter (Fig. 6). As expected, our results demonstrated that binding of each of the MLLs (MLL1–MLL4) was increased in ERE2 of the HOXC13 promoter in the presence of E2 in the cells that were treated with scramble antisense oligonucleotide (Fig. 6, lanes 5 and 6). However, knockdown of either ERα or ERβ significantly decreased (or even abolished) the recruitment of MLLs, especially into ERE2 (Fig. 6, lanes 3 and 4, and 7 and 8). These results demonstrated that E2-induced binding of each of the MLLs to the HOXC13 promoter was mediated via interaction (direct or indirect via other MLL-interacting proteins) with ERα and ERβ.
The physical interactions of MLLs with ERs were further confirmed by using coimmunoprecipitation experiments. As MLL3 showed the most potent activity in E2-dependent HOXC13 regulation, we analyze the interaction of MLL3 with ERα and ERβ separately. In brief, JAR cells were treated with 100 nm E2 for 6 h. Nuclear extracts were prepared from these E2-treated and untreated cells, and were incubated with MLL3 antibody (bound to protein G agarose beads) overnight at 4 °C. Proteins bound to the MLL3-attached and control beads were analyzed by western blotting using antibodies specific for ERα, ERβ, and MLL3. Our results demonstrated that the interactions of both ERα and ERβ with MLL3 were increased in the presence of E2 (Fig. 5B). The direct physical interaction between MLL2 and ERα, MLL3 and ERα and MLL4 and ERα have been previously shown by other laboratories. Thus, our results, in agreement with other reported data, demonstrated that MLLs are recruited to the HOXC13 promoter via interactions (direct or indirect) with ERs.
HOX genes play major role in embryonic development, where they determine the anteroposterior body axis . HOX genes are also expressed in adult tissues, where they are necessary for functional differentiation . In general, HOX gene products act as transcription factors that regulate critical genes that are necessary for cell differentiation and development [1,2]. Despite their critical and well-characterized functions, the regulatory mechanisms that drive HOX gene expression are mostly unknown. Although the mechanism is unclear, several hormones have recently been shown to regulate HOX gene expression, and the endocrine regulation of HOX genes appears to allow the generation of structural and functional diversity in both developing and adult tissues .
HOXC13 is a homeobox-containing gene that plays critical roles in hair development. Hair follicle development, male-specific and female-specific hair patterning and sexual differentiation are strongly dependent on steroid hormones such as E2, progesterone, and androgens [3–5,10]. Herein, we have demonstrated that the HOXC13 gene is transcriptionally regulated by E2. ERα and ERβ are two major players in E2-dependent gene activation . Our studies demonstrated that antisense oligonucleotide-mediated knockdown of either ERα or ERβ downregulated the E2-mediated activation of HOXC13, indicating their critical roles in the process. ER-mediated regulation of E2-sensitive genes is a complicated process . In the presence of E2, ERs are activated and bind to the EREs of E2-responsive genes, eventually resulting in transcription activation . In addition to ERs, E2-mediated gene activation requires various other coregulators and coactivators that result in chromatin modification and remodeling [40,48]. Our results described herein demonstrated that MLLs and ERs play crucial roles in the E2-mediated regulation of HOXC13. Knockdown of MLLs (especially MLL3) suppressed the E2-mediated activation of HOXC13.
In general, ERs, along with various coregulators, are recruited to EREs present in the promoters of E2-responsive genes . Our sequence analysis demonstrated that the HOXC13 promoter contains at least six EREs within −3000 bp upstream of the transcription start site. In vivo binding analysis (ChIP) demonstrated that, in the presence of E2, ERs bind primarily to ERE1 (−234 bp) and ERE2 (−1260 bp), which are closer to the transcription start site. These results suggest that ERE1 and ERE2 of the HOXC13 promoter are primarily responsible for E2-mediated gene activation.
ChIP analysis also demonstrated that MLLs (MLL1–MLL4) were bound to the responsible EREs in an E2-dependent manner. Knockdown of ERα and ERβ downregulated the recruitment of MLLs into the HOXC13 EREs, demonstrating important roles of ER in recruiting MLLs into the HOXC13 promoter. Furthermore, our coimmunoprecipitation experiments demonstrated that MLL3 interacts with both ERα and ERβ in an E2-dependent manner. Consistent with our observations, MLL2, MLL3 and MLL4 have previously been shown to interact with ERα in an E2-dependent manner [12,35–38].
Importantly, there are so many MLLs (MLL1–MLL5) with similar enzymatic functions (H3K4-specific HMT activity), and they are probably involved in regulating different target genes. Because of the differences in promoter cis-elements and their organization, different genes require different activators and coactivators. On the basis of our knockdown experiments, MLL3 is the most important MLL coactivator for HOXC13 expression. However, we observed that other MLLs (MLL1, MLL2, and MLL4) are also involved in HOXC13 regulation, although with weaker effects (knockdown experiments) than MLL3. As MLL1, MLL2 and MLL4 are involved in E2-mediated HOXC13 expression, we expected (as observed; Fig. 5) them to bind to HOXC13 EREs as a function of E2. However, irrespective of the relative importance of the MLLs (MLL1–MLL4), ChIP analysis (Fig. 5) showed efficient E2-dependent binding of all the MLLs in ERE2. It should be noted that the ChIP assay does not provide a truly quantitative measurement in terms of activity of the enzyme, although it provides important information about relative binding efficiency. This might explain the difference in MLL binding profile (ChIP data) versus their activity in knockdown experiments.
Our studies demonstrated that, in addition to MLL2–MLL4, MLL1 is also recruited to ERE2 of the HOXC13 promoter in an E2-dependent manner. Amino acid sequence analysis demonstrated that each MLL (MLL1–MLL4) contains one or more LXXLL domains (NR boxes), which are known to interact with nuclear receptors (NRs) and mediate ligand-dependent gene activation . MLL1 contains one NR box, whereas MLL2–MLL4 contain several (three to four) NR boxes, indicating that each of the MLLs has the potential to interact with ERs and be involved in E2-mediated gene activation . Although further studies are needed to understand the detailed roles of different MLLs and their coordination with ERs, our studies have demonstrated that MLL1–MLL4 are involved in E2-mediated HOXC13 regulation. Furthermore, the E2-dependent increase in histone H3K4-trimethylation level suggested that some of the MLLs might be critical in regulating histone H3K4-methylation in the HOXC13 promoter, which is crucial for gene activation. Although MLLs are well known as major regulators of HOX genes, their roles in the endocrine regulation of HOX genes are unknown. Our results have demonstrated that MLLs play critical roles in the E2-dependent regulation of HOX gene expression. Steroid hormones have been linked with hair growth, sex differentiation and difference in hair patterning between males and females. Our studies provide a molecular link between steroid hormones and the regulation of HOXC13 that may have implications for our understanding of the mechanism of sex-specific hair development. In addition, our results have demonstrated that HOXC13 expression is induced by the steroid hormone E2 in JAR cells, which have a placental origin. Although, at this time, the role of HOX genes in placental function is not clear, this particular organ is critical in embryogenesis and fetal development. It is well known that the placenta produces several steroid hormones that are circulated maternally and to the fetus, and play critical roles in pregnancy and fetal growth . Significant amounts of these hormones remain in the placental tissue, and may regulate placental genes, development, and function. On the basis of our observations, we hypothesize that E2-mediated expression of HOXC13, and possibly various other HOX genes, may have crucial roles in placental function, and this aspect needs to be further investigated.
Cell culture, E2 treatment, and antisense oligonucleotide experiments
Human choriocarcinoma placenta (JAR) cells obtained from the ATCC were maintained in DMEM (Sigma, St Louis, MO, USA) supplemented with 10% fetal bovine serum, 2 mm l-glutamine and penicillin/streptomycin (100 units and 0.1 mg·mL−1, respectively) in a humidified CO2 incubator, as described previously [11,50,51]. Prior to E2 treatment, JAR cells were grown in phenol red-free DMEM-F12 (Sigma), containing 10% activated charcoal-stripped fetal bovine serum for at least three generations. The final round of the cells were grown up to 70% confluency and treated with different concentrations (0–1000 nm) of E2 for varying time periods. The cells were then harvested and subjected to either RNA and protein extraction or ChIP assay.
For treatment of JAR cells with antisense oligonucleotides, cells were grown up to 60% confluency in 60 mm culture plates and transfected with varying amounts (3–9 μg) of different antisense oligonucleotides. Briefly, cocktails of different amounts of antisense oligonucleotide and transfection reagents (ifect, MoleculA) were made in the presence of 300 μL of culture medium (without supplements) by incubating for 30 min, as instructed by the manufacturer. Cells were washed twice with supplement-free culture medium, and finally submerged in 1.7 mL of medium (without supplements). The antisense oligonucleotide/transfection reagent cocktail was applied to the cells and incubated for 7 h before the addition of 2 mL of culture medium with all supplements and 20% activated charcoal-stripped fetal bovine serum. The cells were then incubated for an additional 24 h before being treating with E2.
Preparation of RNA and protein extract
The cells harvested from culture plates were collected by centrifugation at 500 g for 5 min at 4 °C. The cells were then resuspended in diethyl pyrocarbonate (DEPC)-treated buffer A (20 mm Tris/HCl, pH 7.9, 1.5 mm MCl2, 10 mm KCl, 0.5 mm dithiothreitol, 0.2 mm phenylmethanesulfonyl fluoride) for 10 min on ice, and centrifuged at 3500 g for 5 min. The supernatant was subjected to phenol/chloroform extraction, followed by LiCl precipitation of cytoplasmic mRNA by incubating for 1 h at −80 °C. The mRNA was washed with DEPC-treated 70% ethanol, air dried, and resuspended in DEPC-treated water .
For preparation of protein extract, cells were incubated with whole cell extract buffer (50 mm Tris/HCI, pH 8.0, 150 mm NaCl, 5 mm EDTA, 0.05% NP-40, 0.2 mm phenylmethanesulfonyl fluoride, 1× protease inhibitors) for 20 min on ice, and centrifuged at 10 000 g for 10 min. The supernatant containing the whole cell protein extract was stored at −80 °C until further analysis.
RT-PCR and western blot analysis
The first cDNA was synthesized in a 25 μL reaction volume containing 500 ng of RNA, 2.4 μm oligo(dT) (Promega, Madison, WI, USA), 100 units of Moloney murine leukemia virus reverse transcriptase, 1× first-strand buffer (Promega), 100 μm each of dATP, dGTP, dCTP, and dTTP (Invitrogen, Carlsbad, CA, USA), 1 mm dithiothreitol, and 20 units of RNaseOut (Invitrogen). The cDNA was diluted to 100 μL, and 5 μL of the diluted cDNA was used for PCR performed with the gene-specific primer pairs described in Table 1. The PCR program consisted of 30 cycles of 94 °C for 30 s, 60 °C for 30 s, and 72 °C for 45 s, with a final extension at 72 °C for 5 min. Each of the experiments was repeated three times. The normality of the data was analyzed by using t-tests, and ANOVAs were performed at a 5% level of significance.
For western blot analysis, 25 μg of protein extract was subjected to SDS/PAGE and transferred to nitrocellulose membranes. The membranes were then probed with antibodies against MLL1 (Bethyl laboratory), MLL2 (Bethyl laboratory), MLL3 (Abgent, San Diego, CA, USA), MLL4 (Sigma), ERα (Santa Cruz Biotechnology, Santa Cruz, CA, USA), ERβ (Santa Cruz), and β-actin (Sigma), and developed using the alkaline phosphatase method.
ChIP assays were performed by using an EZ Chip chromatin immunoprecipitation kit (Upstate, Billerica, MA, USA), as described previously . In brief, cells were fixed in 4% formaldehyde, lysed, and sonicated to shear the chromatin. The fragmented chromatins were precleaned with protein G agarose and subjected to overnight immunoprecipitation with antibodies specific for ERα, ERβ, MLL1, MLL2, MLL3, and MLL4. Immunoprecipitated chromatins were washed and deproteinized, and DNA fragments were purified by phenol/chloroform extraction followed by precipitation overnight at −80 °C. The purified DNA fragments were then used as templates in PCR amplification of four EREs of the HOXC13 promoter, using the primer pairs listed in Table 1.
Coimmunoprecipitation of MLL–ER complexes
In order to confirm physical interaction of MLLs with ERα and ERβ, we performed coimmunoprecipitation from JAR cells in the absence and presence of E2. In brief, cells were treated with 100 nm E2 for 6 h, and harvested for preparation of nuclear extract. E2-treated and untreated nuclear extracts were incubated overnight at 4 °C with MLL3 antibodies bound to the protein G agarose beads. The beads were separated, and washed with buffer C (20 mm Tris/HCl, pH 7.9, 5 mm MgCl2, 420 mm KCl, 0.5 mm dithiothreitol, 0.2 mm phenylmethanesulfonyl) in the presence of 0.1% NP-40. The affinity-bound proteins were eluted from the beads using 0.2 m glycine (pH 2.9), and analyzed by western blot, using specific bodies, for the presence of ERα, ERβ, and MLL3. Western blots were developed using ECL-Plus (GE Healthcare, Piscataway, NJ, USA), and detected with a phosphorimager (Storm840).
We thank S. Mandal, B. P. Mishra and other laboratory members for helpful discussions. This work was supported in part by ARP (00365-0009-2006) and the American Heart Association (0765160Y).