Key Roles for MED1 Lxxll Motifs in Pubertal Mammary Gland Development and Luminal-Cell Differentiation

Key roles for MED1 LxxLL motifs in pubertal mammary gland development and luminal-cell differentiation

Pingping Jianga,b,1, Qiuping Hua,1, Mitsuhiro Itoc,1,3, Sara Meyera, Susan Waltza, Sohaib Khana, Robert G. Roederc,2, and Xiaoting Zhanga,2

aDepartment of Cancer and Cell Biology, College of Medicine, University of Cincinnati, 3125 Eden Avenue, OH 45267; bCollege of Life Sciences, Zhejiang University, 388 Yuhangtang Road, Hangzhou 310058, China; and cLaboratory of Biochemistry and Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10065

Contributed by Robert G. Roeder, February 16, 2010 (sent for review January 18, 2010)

Mediator recently has emerged as a central player in the direct (7–9). In general, individual Mediator subunits interact transduction of signals from transcription factors to the general specifically with their corresponding transcription factors and de- transcriptional machinery. In the case of nuclear receptors, in vitro letions of these Mediator subunits often affect expression primar- studies have shown that the transcriptional coactivator function of ily of target genes and pathways controlled by their corresponding the Mediator involves direct ligand-dependent interactions of the transcription factor(s) (9). In the case of nuclear receptors, the MED1 subunit, through its two classical LxxLL motifs, with the re- Mediator interactions are ligand- and AF-2-dependent and ceptor AF2 domain. However, despite the strong in vitro evidence, mediated through the LxxLL motifs in the MED1 (a.k.a. there currently is little information regarding in vivo functions of TRAP220/PBP/DRIP205) subunit (10–13). Importantly, in rela- the LxxLL motifs either in MED1 or in other coactivators. Toward tion to the present study, in vitro assays with intact Mediator com- this end, we have generated MED1 LxxLL motif-mutant knockin plexes containing mutated MED1 LxxLL motifs have established mice. Interestingly, these mice are both viable and fertile and do direct roles for the LxxLL motifs in strong (ligand-dependent)

not exhibit any apparent gross abnormalities. However, they do nuclear receptor-Mediator interactions and in Mediator-depen- BIOCHEMISTRY exhibit severe defects in pubertal mammary gland development. dent transcription by nuclear receptors (14, 15). Consistent with this phenotype, as well as loss of the strong li- LxxLL motifs were initially discovered in SRC1 and CREB- gand-dependent estrogen receptor (ER)α-Mediator interaction, ex- binding protein/p300 (16) and later reported in many other pression of a number of known ERα-regulated genes was down- nuclear receptor coactivators including MED1 and SRC/p160 regulated in MED1-mutant mammary epithelial cells and could and PGC-1 family members (4, 17). Crystal structures of the no longer respond to estrogen stimulation. Related, estrogen-sti- agonist-bound nuclear receptor AF-2 domains with LxxLL motifs mulated mammary duct growth in MED1-mutant mice was also have provided considerable mechanistic insights into their inter- greatly diminished. Finally, additional studies show that MED1 is actions (17–20). Although the LxxLL motif itself is sufficient to differentially expressed in different types of mammary epithelial mediate nuclear receptor interactions, sequences flanking the cells and that its LxxLL motifs play a role in mammary luminal LxxLL motifs play key roles in determining receptor binding epithelial cell differentiation and progenitor/stem cell determina- specificity (19, 20) and allow division of LxxLL motifs into several tion. Our results establish a key nuclear receptor- and cell-specific classes according to their flanking sequences (reviewed in ref. 17). in vivo role for MED1 LxxLL motifs, through Mediator-ERα interac- The fact that diverse coactivators such as SRC/p160, MED1, and tions, in mammary gland development. PGC-1 have overlapping interaction sites on the AF-2 domain has led to the concept of cofactor exchange during transcriptional MED1/TRAP220 ∣ estrogen receptor ∣ progenitor/stem cell activation (4, 21). Interestingly, and related, recent studies have identified AF-2 domain mutations that result either in selective uclear receptors are a family of transcription factors that, in loss of TRβ binding to SRC/p160 relative to PGC-1α (22) or in Nresponse to small lipophilic ligands, specifically regulate selective loss of ERα binding to SRC/p160 family members rela- expression of target genes in development, differentiation, meta- tive to MED1 (23). These findings have led to the idea that these bolism, homeostasis, and reproduction (1, 2). These nuclear motifs could be targets for the development of reagents that se- receptors are expressed in a developmental stage- and tissue- lectively disrupt nuclear receptor interactions with certain coac- specific manner and play important roles, not only in normal phy- tivator complexes to achieve tissue- and gene-specific therapeutic siology, but also in pathological conditions (1, 2). Nuclear recep- interventions (6). tors have characteristic domains that include an N-terminal AF-1 Although it has been shown that the LxxLL motifs of MED1 domain that mediates ligand-independent transcription, a central are required for its interactions with nuclear receptors and for DNA-binding domain that binds to regulatory elements in cog- nuclear receptor-mediated transcription in vitro, there is no avail- nate target genes, and a C-terminal ligand-binding domain with able information regarding the functional significance of MED1 an associated (ligand-induced) AF-2 domain that mediates ligand-dependent activation of target genes (1, 2). Author contributions: P.J., Q.H., M.I., R.G.R., and X.Z. designed research; P.J., Q.H., S.M., The ultimate action of nuclear receptors following binding to M.I., and X.Z. performed research; P.J., Q.H., S.W., S.K., M.I., R.G.R., and X.Z. analyzed data; target genes is to enhance the recruitment and/or function of the and R.G.R. and X.Z. wrote the paper. general transcription machinery, including RNA polymerase II The authors declare no conflict of interest. and general transcription factors, on cognate core promoter 1P.J., Q.H., and M.I. contributed equally to this work. elements (3). Interactions of the AF-1 and AF-2 domains with 2To whom correspondence may be addressed: E-mail: [email protected] or roeder@ various transcription coactivators are required for this process rockefeller.edu. – (4 6). Among these diverse coactivators, Mediator, a large 3Present address: Laboratory of Hematology, Kobe University Graduate school of Health (∼1.6 MDa) protein complex composed of at least 25–30 distinct Sciences, Kobe 654-0142, Japan subunits, has emerged as the main bridge for direct communica- This article contains supporting information online at www.pnas.org/cgi/content/full/ tion between transcriptional activators and RNA polymerase II 1001814107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1001814107 PNAS Early Edition ∣ 1of6 Downloaded by guest on September 26, 2021 LxxLL motifs in vivo. Toward this end, we have generated MED1 impaired in Med1KI∕KI mice relative to wild-type mice. Similar LxxLL motif-mutant knockin mice and, surprisingly, found that defects were observed throughout pubertal mammary gland these mice are apparently fertile and grossly normal. However, development at different ages (Fig. S2). they do exhibit profound defects in virgin mammary gland devel- To determine if the mammary gland defects observed in KI∕KI opment. Consistent with this observation, we observed significant Med1 mice resulted from disrupted cell proliferation, we impairments both in ERα-dependent gene expression in mam- performed BrdU incorporation assays (25). Seven- to eight- KI∕KI mary epithelial cells and in estrogen-stimulated mammary ductal week-old wild-type and Med1 age-matched female mice growth. Moreover, we found that MED1 is differentially ex- were injected intraperitoneally with BrdU 2 h prior to sacrifice. pressed in mammary epithelial cells and plays a role in luminal Mammary glands were harvested, fixed, and then subjected to lineage determination. BrdU staining. Ten random areas of 100 cells in each sample were selected and counted to estimate the percentage of total epithe- Results lial cells that were BrdU-positive (Fig. 1 D and E). As expected, Med1KI∕KI Mice Exhibit Profound Defects in Pubertal Mammary Gland about 20% of mammary epithelial cells stained positive for BrdU Development. in wild-type mice. In contrast, there were about 4 times fewer To study the role of the MED1 LxxLL motifs in a KI∕KI physiological context, we generated MED1 LxxLL motif-mutant BrdU-positive mammary epithelial cells in the Med1 mice. 1KI∕KI These data support the idea that the observed mammary gland knockin mice (Med ). The point mutations that were intro- KI∕KI duced result in conversion of both MED1 LxxLL motifs to LxxAA defects in Med1 mice are caused, at least in part, by motifs (Fig. 1A and Fig. S1), which previously was shown to decreased cell proliferation. disrupt strong Mediator-nuclear receptor interactions in vitro KI∕KI α (14, 15). These mutations had no effect on the expression level Med1 Mice Show Impaired ER Target Gene Expression in of the MED1 protein (Fig. S1). Surprisingly, in contrast to Mammary Epithelial Cells. Estrogen is the dominant hormone pro- the embryonic lethality of a total MED1 knockout (24), moting mammary epithelial cell proliferation at the stage of Med1KI∕KI mice were viable, fertile, and grossly normal. How- mammary gland development that we studied. Thus, we reasoned ever, they did exhibit profound defects in mammary gland devel- that the mutations in the MED1 LxxLL motifs exerted their ef- opment. In these studies, mammary glands of 8-week-old fects by disrupting the estrogen signaling pathway, either by in- KI∕KI fluencing the production of estrogen or by directly affecting Med1 and wild-type virgin mice were isolated, fixed in Car- ERα-mediated transcription. To discriminate between these noy’s solution, and then stained with Carmine. As shown morpho- possibilities, we first examined the blood estrogen levels of 8- logically in Fig. 1B and quantitated in Fig. 1C, mammary gland week-old adult mice by ELISA (Cayman). We found that the ductal growth and branch morphogenesis were significantly MED1 LxxLL motif mutations did not affect the production of estrogen (Fig. 2A). To determine whether mutations disrupted ERα interactions with Mediator, immobilized GSTand GST-ERα (ligand-binding domain) fusion proteins were independently incubated with nuclear extracts from either wild-type or Med1KI∕KI mouse embryo fibroblast cells. As expected (26), GST-ERα (ligand-binding domain), but not GST alone, bound Mediator from wild-type nuclear extract in a ligand (estrogen)-dependent manner (Fig. 2B). However, it failed to bind Mediator from Med1KI∕KI nuclear extract, even in the presence of estrogen. As a control, a GST-VP16 activation domain fusion protein, which interacts with the MED17 subunit of Mediator, interacted equally well with Mediator in extracts from wild-type or mutant mice. These data confirm that the strong ligand-dependent ERα-Mediator interaction is effectively and selectively disrupted by the LxxLL to LxxAA mutations. To determine whether expression of ERα target genes was affected in the mammary glands of Med1KI∕KI mice, mammary epithelial cells were first isolated from mammary glands of 7- to 8-week-old Med1KI∕KI mice and control wild-type mice. Total RNA was isolated and subjected to semiquantitative real-time PCR analyses following reverse transcription. We examined expression of a number of known ERα target genes, including WNT1-inducible signaling pathway protein 2 (WISP2), diamine oxidase (Dio), cyclin D1, pS2, and progesterone receptor (PR) (Fig. 2C). We found that expression of most of these ERα target genes (pS2, cyclin D, Dio, WISP2) was significantly reduced in Med1KI∕KI mammary epithelial cells. Interestingly, the expression of another ERα target gene, PR, was not significantly altered, Fig. 1. MED1 LxxLL mutations impair mammary gland development. (A) Dia- which is consistent with our finding that these mammary epithe- gram indicating the four L to A point mutations introduced into the two lial cells were still able to differentiate during pregnancy (Fig. S3). MED1 LxxLL motifs in the mutant MED1 knockin mice. (B) Mammary glands Furthermore, we found that the expression levels of both ERα from 8-week-old wild-type and MED1 mutant mice. Glands were fixed and and PGC-1α, another ERα coactivator (27), also were not chan- subjected to whole-mount staining. (C) Mammary duct lengths. The distances ged. These data are consistent with the idea that MED1 LxxLL of mammary ducts from the lymph node were measured and quantified in motifs play an important in vivo role in ERα-mediated transcrip- five individual mice. (D) BrdU immunohistochemial staining of mammary gland sections from 7- to 8-week-old wild-type and MED1 mutant virgin mice. tion in the mammary gland. (E) Percentage of BrdU-positive staining mammary epithelial cells from wild-type and MED1 mutant mice are shown (N ¼ 5). Scale bar, 1 mm (B) Estrogen-Induced Expression of ERα Target Genes Is Blocked KI∕KI and 50 μm(D). in Med1 Mammary Epithelial Cells. We next carried out

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1001814107 Jiang et al. Downloaded by guest on September 26, 2021 Fig. 2. MED1 mutations abolish the ligand-dependent ERα-Mediator inter- Fig. 3. MED1 LxxLL motif mutations block estrogen-dependent gene expres- action and ERα target gene expression. (A) Estrogen levels in sera from 8- sion and estrogen-stimulated mammary duct growth. (A) Gene expression week-old wild-type (black bar) and Med1KI∕KI (gray bar) mice (n ¼ 5). (B) levels. Wild-type mammary epithelial cells were isolated from wild-type ERα-Mediator interactions. GST-pulldown assays employed equivalent and Med1KI∕KI mice and treated with vehicle or 10 nM estrogen for the in- amounts of purified GST, GST-ERα(LBD) and GST-VP16 proteins and nuclear dicated time. Total RNAs were isolated and expression of the indicated ERα extracts from wild-type and mutant mouse embryonic fibroblasts as sources target genes was measured by real-time PCR as above (Value ¼ Mean STD; of Mediator. (C)ERα target gene expression. Real-time PCR was carried out P < 0.05). (B) Mammary duct growth. 3-week-old wild-type and Med1KI∕KI after reverse transcription of total RNA from mammary epithelial cells of mice were ovariectomized and implanted with slow-release E2 pellets for Med1KI∕KI and control wild-type mice. Relative expression levels of the indi- 21 d. Whole-mount Carmine stainings of inguinal glands from these mice BIOCHEMISTRY cated ERα target genes (normalized to 18S rRNA content) are shown are shown. (Value ¼ Mean STD; P < 0.05). failed estrogen response and that MED1 LxxLL motifs play experiments to determine whether the impaired expression of the critical in vivo roles in mediating estrogen functions in the mam- KI∕KI ERα target genes in Med1 mice was caused by a disruption mary gland. in the response to estrogen stimulation. Mammary epithelial cells KI∕KI again were isolated from both wild-type and Med1 mice and MED1, Like ERα, Is Highly Expressed in Luminal Epithelial Cells. Of the then grown in DMEM/F12 phenol red-free medium with two major types of mammary epithelial cells, luminal and basal charcoal-stripped serum for 2 d. The cells then were treated with cells, ERα is expressed mainly in the luminal epithelial cells. To vehicle or estrogen for different durations (0, 5, 20, 120 min), better understand the role of MED1 in mammary gland develop- after which they were collected. Total RNAs were isolated and ment, we determined the pattern of MED1 expression in these reverse transcribed for real-time PCR analyses to assess expres- cells. Mammary gland tissue sections from 8-week-old mice were sion of representative ERα target genes (pS2, cyclin D, and Dio). prepared as above and subjected to coimmunofluorescence stain- Estrogen-dependent expression of these genes was observed in ing using anti-ERα and anti-MED1 antibodies. Alexa 555- and wild-type mammary epithelial cells, as expected, but abrogated 488-conjugated secondary goat-anti-mouse and goat-anti-rabbit in Med1KI∕KI mammary epithelial cells (Fig. 3A). These data antibodies were used to detect ERα and MED1, respectively. are consistent with the notion that impaired expression of these As expected, we observed a high level of ERα staining in the genes in Med1KI∕KI mammary epithelial cells is caused by nuclei of mammary luminal epithelial cells but not in the basal disrupted ERα-mediated transcription. cells (Fig. 4A, Top). Interestingly, the staining pattern for MED1 was very similar to that of ERα—almost undetectable Med1KI∕KI Mammary Ducts Fail to Respond to Estrogen-Stimulated in basal epithelial cells and high in luminal epithelial cells, with Growth in Vivo. To determine, in vivo, whether the observed de- the strong ERα overlap clearly evident in the merged images. To fects in mammary gland ductal growth and branch morphogenesis further confirm this observation, we carried out coimmuno- KI∕KI in Med1 mice are due to an impaired response to estrogen, fluorescence staining experiments with antibodies against KI∕KI 21-day-old wild-type and Med1 littermates were first MED1 and the basal epithelial cell marker CK14. The results ovariectomized and then implanted with slow-release pellets con- clearly show that MED1 and CK14 have a mutually exclusive taining estradiol (E2) or placebo (Innovative Research of expression pattern (Fig. 4A, Bottom). Coimmunostaining with America). After 21 d, inguinal glands were excised and processed antibodies against MED1 and the luminal epithelial marker for whole-mount staining, while uteri were also collected as po- CK18 further confirmed preferential expression of MED1 in lu- sitive controls for the efficacy of estrogen treatment. We found minal epithelial cells (Fig. S5). Interestingly, and in contrast to that uterus weights in both control and Med1KI∕KI mice were in- the differential expression of MED1, the core Mediator subunit creased by more than 10-fold, indicating successful E2 treatments MED30 (indicative of the Mediator complex) is expressed in both (Fig. S4A). We also found that placebo treatment did not stimu- luminal and basal mammary epithelial cells at very comparable late mammary ductal growth (Fig. S4B). Importantly, whereas a levels (Fig. S6). These data indicate that MED1 is differentially significant induction of mammary ductal growth by E2 treatment expressed in the different types of mammary epithelial cells, with was observed in wild-type mice, the growth of mammary ducts high expression in the luminal mammary epithelial cells in the was greatly inhibited in the Med1KI∕KI mice despite E2 treatment normal mammary gland. (Fig. 3B). These data further support the idea that the mammary Regarding the issue of cell-specific MED1 expression and gland developmental defects in Med1KI∕KI mice are due to a function, the differentiation of mammary epithelial cells during

Jiang et al. PNAS Early Edition ∣ 3of6 Downloaded by guest on September 26, 2021 Fig. 4. MED1 is differentially expressed in mammary epithelial cells and plays a role in luminal epithelial cell differentiation. (A) Mammary glands from wild- type and MED1 mutant mice were isolated, fixed, and subjected to coimmunofluorescent staining with antibodies against MED1 and ERα (Top) or MED1 and basal epithelial marker CK14 (Bottom). (B) Mammary gland sections of wild-type and Med1KI∕KI mice were subjected to coimmunofluorescent staining using antibodies against the cytokeratin markers CK14 and CK18 (luminal epithelial marker). Scale bar, 50 μm. (C) Mammary epithelial cells from wild-type and Med1KI∕KI mice were stained with the indicated PE- or FITC-conjugated antibodies and analyzed by FACS.

pregnancy involves, in addition to ERα (28), other nuclear mice. To confirm these findings, we examined the expression receptors (e.g., ERβ and PR) that also have been shown to inter- of another well-known mammary stem/progenitor cell marker, act with MED1 in vitro (29, 30). Since we found that MED1 stem cell antigen-1 (Sca1) (32), in the mammary epithelial cells is expressed in mammary epithelial cells during pregnancy of these mice, as it has been shown previously that the (Fig. S7A), our observation that Med1KI∕KI mice exhibit a severe CD24þSca1− population is enriched in luminal progenitor cells. phenotype in mammary gland pubertal development, but not dur- Consistent with the above observations, we found a significant ing pregnancy, reflects differential (cell- and receptor-specific) increase in the CD24þSca1− mammary epithelial cell population MED1 LxxLL motif requirements for ERα,ERβ, and PR func- from the mammary glands of Med1KI∕KI mice (Fig. 4C). Taken tion in vivo. This is also consistent with the lack of any apparent together, these results indicate that MED1 LxxLL motifs play phenotype in Med1KI∕KI mice in other steroid responsive tissues critical roles in luminal progenitor cell determination. (e.g. uterus) that also express MED1 (Fig. S7B). Discussion MED1 LxxLL Motif Mutations Affect Mammary Epithelial Cell Lineage It is well known that mammary gland development is under Determination. In view of our finding that MED1 is not expressed precise temporal control by specific hormonal and signaling path- universally in all mammary epithelial cells but, rather, is ways during each developmental stage (reviewed in refs. 33, 34). expressed mainly in luminal epithelial cells, we carried out experi- The estrogen signaling pathway has been shown to be the key ments to determine whether MED1 LxxLL motif mutations have pathway involved in ductal growth and branch morphogenesis an impact on the lineage determination of mammary epithelial in the stages from prepuberty to puberty (33, 34). We previously cells. Immunofluorescent staining was carried out using anti- have shown that MED1 is required for optimal ERα-dependent bodies against luminal (CK18) and basal (CK14) epithelial mar- transcription in vitro (26, 35). In this study, we provide in vivo kers, as described above, on paraffin-embedded 8-week-old adult evidence that the MED1 LxxLL motifs play important roles in mammary gland sections. As expected, mammary epithelial cells ERα-mediated functions in early mammary gland development. in wild-type mice were well differentiated into distinct luminal We found that mammary ductal growth and branch morphogen- KI∕KI (CK18þ) and basal (CK14þ) epithelial cells. In contrast, esis are significantly impaired in virgin Med1 mice relative to KI∕KI Med1KI∕KI mice contained a significant population of CK14 wild-type mice. Consistent with this observation, Med1 and CK18 double-positive progenitor cells (Fig. 4B, merged mammary epithelial cells show a significant reduction in the ex- images). These data suggest an important role for MED1 in mam- pression of a number of known ERα target genes. Furthermore, mary luminal epithelial cell lineage determination. the expression of these genes is no longer responsive to estrogen It has been reported that the cell-surface markers CD24 and stimulation in isolated mammary epithelial cells. Importantly, we CD29 can be used to distinguish mammary progenitors with provide further evidence showing that exogenous estrogen-stimu- luminal-cell fate (CD29lowCD24þ) (31). We therefore examined lated mammary duct growth in ovariectomized MED1 mutant CD24 and CD29 expression profiles to determine the specific cell mice is also blocked in vivo. populations that are affected by the MED1 LxxLL-motif muta- Interestingly, whereas Mediator is generally thought to be es- tions. Wild-type and Med1KI∕KI mammary epithelial cells were sential for all RNA polymerase II-transcribed genes, Med1KI∕KI isolated and subjected to FACS analyses after staining with anti- mice did not exhibit some defects (e.g., in uterine development, bodies against these markers. As shown in Fig. 4C, we found a fertility, or estrogen production) that are observed in ERα null significant increase in CD29lowCD24þ mammary epithelial cell mice (25). This lack of detectable effects on other tissues could populations from Med1KI∕KI mice relative to wild-type mice. reflect a previously unrecognized tissue-specific requirement for These data suggest that MED1 LxxLL motifs play a role in MED1 and its LxxLL motifs in mediating ERα function, similar controlling the luminal progenitor cell population in pubertal to what has been previously shown for other nuclear receptor

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1001814107 Jiang et al. Downloaded by guest on September 26, 2021 cofactors such as SRC/p160 family members (reviewed in ref. 21). of Mediator complexes with unique subunit compositions (35). This tissue-specific role of MED1 LxxLL motifs could be caused Importantly, our finding that Mediator core subunit MED30 is by the differential expression of other compensatory coactivators. comparably expressed in both basal and luminal mammary Thus, it has been reported that ERα shows distinct responses to epithelial cells indicates that the absence of MED1 in basal antiestrogens in different (mammary vs. uterine) tissues because epithelial cells is not likely due to an overall decrease in the of variations in coactivator expression levels (36). Furthermore, it Mediator levels in these cells. With regard to the issue of recently was found that Mediator can be recruited to ERα target MED1-related Mediator heterogeneity and estrogen signaling, genes by other coactivators such as CCAR1 (37) and to other it is relevant to note that breast cancer has currently been cate- nuclear receptor-promoter complexes by PGC-1α (38), which gorized into different subtypes [e.g. luminal type (A and B), basal also interacts directly with ERα (27). We also have found that type, and triple negative] according to their distinct signature the MED1 LxxLL motifs, but not MED1 itself, is dispensable gene expression profiles and altered signaling pathways (40). for PPARγ-mediated adipogenesis (15). Therefore, the presence Importantly, these different subtypes of breast cancer show dif- or high level expression of certain other ERα-interacting ferent prognoses and require different treatment strategies (41). cofactors in these estrogen-responsive tissues could bypass the Interestingly, MED1 has been found to be overexpressed and am- requirement for the MED1 LxxLL motifs. We also note that plified in more than 40–50% of human breast cancers (42), which the mammary gland developmental defects in Med1KI∕KI mice, was further confirmed by a number of genome-wide microarray although profound, are not as severe as those in ERα null mice. analyses of human breast cancer patient samples (Oncomine). Consistent with this observation, we have already observed that Therefore, future determination of the role of MED1 in breast not all ERα target genes are affected in the mammary epithelial tumorigenesis and its expression in these different subtypes of cells of these mice. Therefore, even if the same sets of genes are human breast cancer may have important prognostic or thera- affected in tissues that do not exhibit severe defects, this could be peutic implications. because those genes do not play the same important roles in Materials and Methods those tissues. KI∕KI KI∕KI Our current studies have focused on the role of the MED1 Generation and Maintenance of Med1 Mice. To generate Med1 mice, LxxLL motifs in mammary gland development because defects the linearized targeting vector was electroporated into E14 embryonic stem cells. Positive clones were selected, confirmed by Southern blot analyses and in this tissue were the first and most dramatic that were observed 1KI∕KI α microinjected into blastocysts of C57BL6/J mice. These mice were extensively in Med mice. Although linked to ER in this tissue, MED1 backcrossed (10 times) to a C57BL6/J background. All animal were housed in

also has been reported to interact with a number of other nuclear AAALAC (Association for Assessment and Accreditation of Laboratory Animal BIOCHEMISTRY receptors [e.g. ERβ, PR, VDR (vitamin D receptor), PPARγ Care)-approved facilities at the University of Cincinnati and at Rockefeller Uni- (peroxisome proliferator-activated receptor γ), TR (thyroid re- versity. All procedures were approved by Institutional Animal Care and Use ceptor), AhR (aryl hydrocarbon receptor), and GR (glucocorti- Committees, and in accordance with the National Institutes of Health coid receptor)] in vitro through its LxxLL motifs. However, these guidelines outlined in the Guide for Care and Use of Laboratory Animals. nuclear receptors are not likely to contribute in a significant way to the specific mammary gland developmental defects observed Isolation of Mouse Mammary Epithelial Cells. Mammary glands (thoracic and inguinal glands) were excised, minced, and digested for 2 h at 37 °C in DMEM/ in Med1KI∕KI mice, as corresponding receptor deletions do not F12 medium containing 0.1% collagenase. The digest was centrifuged at result in comparably severe phenotypes in pubertal mammary 1,000 rpm for 5 min and the pellet was resuspended and digested with gland development. It is notable, however, that whereas 2 U∕ml DNase I. Mammary epithelial cells were further purified by repeated KI∕KI Med1 mice exhibit a deficiency in early (pubertal) mammary washes with phosphate buffered saline followed by brief centrifugation gland development, they show normal mammary epithelial differ- (10 s) at low speed (900 rpm). Pooled cell pellets containing epithelial orga- entiation during pregnancy—a process that involves ERβ and PR, noids were filtered through a 400-mm polyester mesh (PE400; Clarcon, War- as well as ERα (28–30). Our data, altogether, are thus indicative rington, UK). of both nuclear receptor- and tissue-specific functions of the MED1 LxxLL motifs—similar conceptually to what has been de- Estrogen Measurement and E2-Stimulated Responses. Blood was drawn from monstrated in vivo for one of the two LxxLL motifs in coactivator the heart of wild-type and MED1 mutant mice immediately after sacrifice. Serum E2 (17-βestradiol) levels were measured by an Estradiol EIA kit accord- NCoA6 (39). However, we cannot exclude the possibility that the ing to the manufacturer’s protocol (Caymen). To examine E2-stimulated functions of these steroid receptors are affected by MED1 LxxLL mammary gland growth, 21-day-old virgin females were ovariectomized motif mutations in other tissues where they might normally func- and implanted with 21-day slow-release pellets (0.025 mg, Innovative tion, especially since some phenotypic abnormalities might be ex- Research of America). After 21 d, inguinal glands were excised for whole- hibited only during certain stress conditions that elicit demands mount staining with carmine (Sigma). for higher levels of nuclear receptor function. Indeed, we have found that mice containing muscle-specific deletions of MED1 Fluorescence Immunohistochemistry. Inguinal glands were fixed and show normal glucose and insulin levels but significant embedded in paraffin blocks as described above. Following heat-induced alterations in both glucose tolerance and insulin sensitivity. antigen retrieval in citrate buffer, tissue sections were incubated with pri- mary antibodies against MED1, ERα (clone 6F11, Novocastra), CK14 (clone Therefore, it will be particularly interesting to determine whether LL002, Thermo Scientific), or CK18 (Fitzgerald) overnight at 4 °C. Alexa Fluor MED1 LxxLL motifs play a role in eliciting these responses under 488- or 555-conjugated goat anti-rabbit or anti-mouse secondary antibodies these stress conditions. These observations further emphasize the were purchased from Invitrogen. The images were visualized and captured importance of studying the role of these cofactors in vivo, in using an Axioplan Imaging 2e fluorescence microscope (Zeiss). specific tissues, and under the appropriate conditions. Another significant finding of this study is that MED1 is not FACS Analysis. Mammary epithelial cells were isolated from the thoracic and only differentially expressed in different types of mammary inguinal mammary glands, as above, followed by a disassociation step with epithelial cells but also plays a role in mammary luminal epithe- trypsin. Cell suspensions were incubated at room temperature for 1 hr with the following antibodies: PE-CD24 (1∶200, BD Biosciences), CD29 (1∶200, AbD lial cell lineage determination. Having been considered a more 1∶200 general transcriptional cofactor, MED1 was thought to exist in Serotec), and FITC-Sca-1 ( , BD Biosciences). Cells were sorted using a FACSCanto II system (BD Biosciences) and the data were analyzed with all Mediator complexes and to be uniformly expressed. However, FACSDiva 6.1.1 software. our previous biochemical analyses provided evidence both for an otherwise intact Mediator complex in Med1-∕- cells (14) and for ACKNOWLEDGMENTS. We thank Drs. Shuk-Mei Ho (University of Cincinnati), compositional heterogeneity within Mediator complexes, as Glendon Zinser (U. of Cincinnati), Robert Glazer (Georgetown University) and evidenced by the existence of MED1 in only a subpopulation their lab members for reagents and helpful discussions. X.Z. was supported at

Jiang et al. PNAS Early Edition ∣ 5of6 Downloaded by guest on September 26, 2021 The Rockefeller University by postdoctoral fellowship awards from the Susan a Ride Cincinnati Award (to X.Z), in part by Public Health Service Grant G. Komen Breast Cancer Foundation and the Breast Cancer Alliance. This DK P30 DK078392, and by National Institutes of Health Grants CA100002 study was supported by University of Cincinnati Cancer Center grants and (to S.W.) and DK071900 (to R.G.R.).

1. Mangelsdorf DJ, et al. (1995) The nuclear receptor superfamily: The second decade. 23. Acevedo ML, et al. (2004) Selective recognition of distinct classes of coactivators by a Cell 83:835–839. ligand-inducible activation domain. Mol Cell 13:725–738. 2. Tsai MJ, O’Malley BW (1994) Molecular mechanisms of action of steroid/thyroid recep- 24. Ito M, et al. (2000) Involvement of the TRAP220 component of the TRAP/SMCC coac- tor superfamily members. Annu Rev Biochem 63:451–486. tivator complex in embryonic development and thyroid hormone action. Mol Cell 3. Roeder RG (2003) Lasker Basic Medical Research Award. The eukaryotic transcriptional 5:683–693. machinery: Complexities and mechanisms unforeseen. Nat Med 9:1239–1244. 25. Couse JF, Korach KS (1999) Estrogen receptor null mice: What have we learned and 4. Glass CK, Rosenfeld MG (2000) The coregulator exchange in transcriptional functions where will they lead us?. Endocr Rev 20:358–417. of nuclear receptors. Genes Dev 14:121–141. 26. Kang YK, Guermah M, Yuan CX, Roeder RG (2002) The TRAP/Mediator coactivator 5. McKenna NJ, O’Malley BW (2002) Combinatorial control of gene expression by nuclear complex interacts directly with estrogen receptors alpha and beta through the receptors and coregulators. Cell 108:465–474. TRAP220 subunit and directly enhances estrogen receptor function in vitro. Proc Natl 6. Hall JM, McDonnell DP (2005) Coregulators in nuclear estrogen receptor action: From Acad Sci USA 99:2642–2647. – concept to therapeutic targeting. Mol Interv 5:343 357. 27. Tcherepanova I, et al. (2000) Modulation of estrogen receptor-alpha transcriptional 7. Malik S, Roeder RG (2005) Dynamic regulation of pol II transcription by the mamma- activity by the coactivator PGC-1. J Biol Chem 275:16302–16308. – lian Mediator complex. Trends Biochem Sci 30:256 263. 28. Feng Y, Manka D, Wagner KU, Khan SA (2007) Estrogen receptor-alpha expression in 8. Kornberg RD (2005) Mediator and the mechanism of transcriptional activation. Trends the mammary epithelium is required for ductal and alveolar morphogenesis in mice. – Biochem Sci 30:235 239. Proc Natl Acad Sci USA 104:14718–14723. 9. Blazek E, Mittler G, Meisterernst M (2005) The mediator of RNA polymerase II. Chro- 29. Zhao C, Dahlman-Wright K, Gustafsson JA (2008) Estrogen receptor beta: An overview – mosoma 113:399 408. and update. Nucl Recept Signal 6:1–10. 10. Fondell JD, Ge H, Roeder RG (1996) Ligand induction of a transcriptionally active thyr- 30. Conneely OM, Jericevic BM, Lydon JP (2003) Progesterone receptors in mammary oid hormone receptor coactivator complex. Proc Natl Acad Sci USA 93:8329–8333. gland development and tumorigenesis. J Mammary Gland Biol 8:205–214. 11. Yuan CX, et al. (1998) The TRAP220 component of a thyroid hormone receptor- 31. Shackleton M, et al. (2006) Generation of a functional mammary gland from a single associated protein (TRAP) coactivator complex interacts directly with nuclear receptors stem cell. Nature 439:84–88. in a ligand-dependent fashion. Proc Natl Acad Sci USA 95:7939–7944. 32. Woodward WA, Chen MS, Behbod F, Rosen JM (2005) On mammary stem cells. J Cell Sci 12. Ren Y, et al. (2000) Specific structural motifs determine TRAP220 interactions with 118:3585–3594. nuclear hormone receptors. Mol Cell Biol 20:5433–5446. 33. Hennighausen L, Robinson GW (2001) Signaling pathways in mammary gland 13. Burakov D, et al. (2000) Functional interactions between the estrogen receptor and development. Dev Cell 1:467–475. DRIP205, a subunit of the heteromeric DRIP coactivator complex. J Biol Chem 34. Sternlicht MD (2006) Key stages in mammary gland development: The cues that 275:20928–20934. regulate ductal branching morphogenesis. Breast Cancer Res 8:201–211. 14. Malik S, et al. (2004) Structural and functional organization of TRAP220, the TRAP/ 35. Zhang X, et al. (2005) MED1/TRAP220 exists predominantly in a TRAP/Mediator sub- mediator subunit that is targeted by nuclear receptors. Mol Cell Biol 24:8244–8254. 15. Ge K, et al. (2008) Alternative mechanisms by which mediator subunit MED1/TRAP220 population enriched in RNA polymerase II and is required for ER-mediated transcrip- – regulates peroxisome proliferator-activated receptor gamma-stimulated adipogenesis tion. Mol Cell 19:89 100. and target gene expression. Mol Cell Biol 28:1081–1091. 36. Shang Y, Brown M (2002) Molecular determinants for the tissue specificity of SERMs. – 16. Heery DM, Kalkhoven E, Hoare S, Parker MG (1997) A signature motif in transcrip- Science 295:2465 2468. tional co-activators mediates binding to nuclear receptors. Nature 387:733–736. 37. Kim JH, et al. (2008) CCAR1, a key regulator of mediator complex recruitment to – 17. Savkur RS, Burris TP (2004) The coactivator LXXLL nuclear receptor recognition motif. nuclear receptor transcription complexes. Mol Cell 31:510 519. J Pept Res 63:207–212. 38. Chen W, Yang Q, Roeder RG (2009) Dynamic interactions and cooperative functions of 18. Shiau AK, et al. (1998) The structural basis of estrogen receptor/coactivator recogni- PGC-1alpha and MED1 in TRalpha-mediated activation of the brown-fat-specific UCP-1 tion and the antagonism of this interaction by tamoxifen. Cell 95:927–937. gene. Mol Cell 35:755–768. 19. McInerney EM, et al. (1998) Determinants of coactivator LXXLL motif specificity in nu- 39. Li Q, Chu MJ, Xu J (2007) Tissue- and nuclear receptor-specific function of the C-term- clear receptor transcriptional activation. Genes Dev 12:3357–3368. inal LXXLL motif of coactivator NCoA6/AIB3 in mice. Mol Cell Biol 27:8073–8086. 20. Darimont BD, et al. (1998) Structure and specificity of nuclear receptor-coactivator in- 40. Chung CH, Bernard PS, Perou CM (2002) Molecular portraits and the family tree of teractions. Genes Dev 12:3343–3356. cancer. Nat Genet 32(Suppl):533–540. 21. Xu J, Wu RC, O’Malley BW (2009) Normal and cancer-related functions of the p160 41. O’Shaughnessy JA (2006) Molecular signatures predict outcomes of breast cancer. steroid receptor co-activator (SRC) family. Nat Rev Cancer 9:615–630. N Engl J Med 355:615–617. 22. Wu Y, Delerive P, Chin WW, Burris TP (2002) Requirement of helix 1 and the AF-2 do- 42. Zhu Y, et al. (1999) Amplification and overexpression of peroxisome proliferator-acti- main of the thyroid hormone receptor for coactivation by PGC-1. J Biol Chem vated receptor binding protein (PBP/PPARBP) gene in breast cancer. Proc Natl Acad Sci 277:8898–8905. USA 96:10848–10853.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1001814107 Jiang et al. Downloaded by guest on September 26, 2021