Zinc Finger Homeodomain Factor Zfhx3 Is Essential for Mammary Lactogenic Differentiation by Maintaining Prolactin Signaling Activity*

The zinc finger homeobox 3 (ZFHX3, also named ATBF1 for AT motif binding factor 1) is a transcription factor that suppresses prostatic carcinogenesis and induces neuronal differentiation. It also interacts with estrogen receptor α to inhibit cell proliferation and regulate pubertal mammary gland development in mice. In the present study, we examined whether and how Zfhx3 regulates lactogenic differentiation in mouse mammary glands. At different stages of mammary gland development, Zfhx3 protein was expressed at varying levels, with the highest level at lactation. In the HC11 mouse mammary epithelial cell line, an in vitro model of lactogenesis, knockdown of Zfhx3 attenuated prolactin-induced β-casein expression and morphological changes, indicators of lactogenic differentiation. In mouse mammary tissue, knock-out of Zfhx3 interrupted lactogenesis, resulting in underdeveloped glands with much smaller and fewer alveoli, reduced β-casein expression, accumulation of large cytoplasmic lipid droplets in luminal cells after parturition, and failure in lactation. Mechanistically, Zfhx3 maintained the expression of Prlr (prolactin receptor) and Prlr-Jak2-Stat5 signaling activity, whereas knockdown and knock-out of Zfhx3 in HC11 cells and mammary tissues, respectively, decreased Prlr expression, Stat5 phosphorylation, and the expression of Prlr-Jak2-Stat5 target genes. These findings indicate that Zfhx3 plays an essential role in proper lactogenic development in mammary glands, at least in part by maintaining Prlr expression and Prlr-Jak2-Stat5 signaling activity.

The zinc finger homeobox 3 (ZFHX3, also named ATBF1 for AT motif binding factor 1) is a transcription factor that suppresses prostatic carcinogenesis and induces neuronal differentiation. It also interacts with estrogen receptor ␣ to inhibit cell proliferation and regulate pubertal mammary gland development in mice. In the present study, we examined whether and how Zfhx3 regulates lactogenic differentiation in mouse mammary glands. At different stages of mammary gland development, Zfhx3 protein was expressed at varying levels, with the highest level at lactation. In the HC11 mouse mammary epithelial cell line, an in vitro model of lactogenesis, knockdown of Zfhx3 attenuated prolactin-induced ␤-casein expression and morphological changes, indicators of lactogenic differentiation. In mouse mammary tissue, knock-out of Zfhx3 interrupted lactogenesis, resulting in underdeveloped glands with much smaller and fewer alveoli, reduced ␤-casein expression, accumulation of large cytoplasmic lipid droplets in luminal cells after parturition, and failure in lactation. Mechanistically, Zfhx3 maintained the expression of Prlr (prolactin receptor) and Prlr-Jak2-Stat5 signaling activity, whereas knockdown and knock-out of Zfhx3 in HC11 cells and mammary tissues, respectively, decreased Prlr expression, Stat5 phosphorylation, and the expression of Prlr-Jak2-Stat5 target genes. These findings indicate that Zfhx3 plays an essential role in proper lactogenic development in mammary glands, at least in part by maintaining Prlr expression and Prlr-Jak2-Stat5 signaling activity.
Development of the mammary gland occurs in connection with sexual development and reproduction and is regulated by multiple hormones (e.g. estrogen, progesterone, and prolactin) and growth factors (EGF, FGF, insulin-like growth factor, etc.). It can be divided into six distinct stages: embryonic, prepubertal, pubertal (the linear phase), pregnancy, lactation, and involution (the cyclic phase) (1,2). Unlike most other organs, 5 of the 6 stages of mammary gland development occur postnatally, providing an ideal model for studying genes in both normal development and neoplastic progression. Mammary glands retain plasticity for undergoing the cyclic phase, which implies the existence of specific molecules that are able to integrate a variety of signals from hormones and growth factors. For example, the STAT genes have been demonstrated to play roles in both hormone response and growth factor signaling (3). At present, a few factors have been identified and characterized for their roles in mammary gland development in response to hormonal signaling, but many more remain to be discovered (4,5).
The homeobox gene family contains the homeobox sequence that encodes for the homeodomain, a DNA-binding domain about 60 amino acids long. More than 200 homeodomain-containing proteins have been identified and characterized in a variety of species, most of which act as transcription factors in a wide range of critical activities during normal development and tumorigenesis (6). Several homeobox genes are expressed in mammary epithelial cells, and their functions as regulators of mammary gland development have been established using genetically modified mice (7,8). For example, loss of Msx2, Hoxc6, Hoxa9, and Pax2 in the mammary gland leads to a series of defects during mammary gland development, including failures in side branching and lobulo-alveolar development (8,9). Zinc finger homeobox 3 (ZFHX3), also named ATBF1 for AT motif binding factor 1, is a large transcription factor with 23 zinc finger and 4 homeodomains (10,11). Very few homeobox genes have more than 1 homeobox (12), and the existence of 4 homeoboxes in ZFHX3 suggests that it has a dynamic function in biological processes (13). For example, ZFHX3 is necessary for neuronal and myogenic differentiation in cell culture models (11, 14 -16), and deletion of Zfhx3 in mice causes developmental defects, interrupts epithelial homeostasis, and induces neoplastic morphology in mouse prostates (17)(18)(19). In mouse mammary glands, we previously demonstrated that Zfhx3 mRNA expression varies at different stages during development, reaching the highest level at lactation (20), and that Zfhx3 regulates pubertal mammary gland development (20). In addition, both estrogen and progesterone, two hormones essential for normal mammary gland development, induce or enhance the transcription of ZFHX3 in human and mouse mammary epithelial cells (21,22), although estrogen also causes protein degradation of ZFHX3 when too much estrogen is present (21). Taken together with the observation that deletion of Zfhx3 in mouse prostates alters the transcription level of Prlr (17), a key regulator of lactogenic differentiation in the mammary gland, we hypothesize that ZFHX3 is more relevant to lactogenic differentiation during mammary gland development.
In this study, we examined whether and how Zfhx3 regulates lactogenic development in the mammary gland using both a cell culture model and mouse model. In the HC11 mouse mammary epithelial cell line, where prolactin induces morphological and molecular changes indicative of lactogenic differentiation, knockdown of Zfhx3 expression attenuated the effect of prolactin. Consistently, deletion of Zfhx3 in mouse mammary glands prevented proper alveologenesis and lactogenic differentiation. Mechanistically, the effect of Zfhx3 was mediated at least in part by regulating the Prlr-Jak2-Stat5 signaling axis.

Experimental Procedures
Mice-Wild type C57BL/6 mice were purchased from the Academy of Military Medical Sciences (Beijing, China), and breeding was carried out following standard procedures for the collection of mammary tissues at different developmental stages. Breeding, genotyping, and preparation of mice with MMTV-Cre-mediated mammary-specific knock-out of Zfhx3 were previously described (20). All mice were fed with pathogen-free food and water and were closely monitored and humanely euthanized. At least four mice were used for each genotype at a time point for different analyses, including whole mount analysis, immunohistochemical staining, and real time PCR. All mouse experiments were approved by the Institutional Animal Care and Use Committee at College of Life Sciences of Nankai University.
Cell Culture-The human embryonic kidney cell line 293T and human breast cancer cell line MCF-7 were cultured in DMEM (Invitrogen) supplemented with 10% FBS. Human breast cancer cell lines MDA-MB-231 and T-47D were cultured in RPMI 1640 medium supplemented with 10% FBS. For prolactin treatment, T-47D cells were first incubated in serumfree medium overnight and then treated with 500 ng/ml recombinant human prolactin (R&D Systems) for indicated times.
In Vitro Lactogenic Differentiation of HC11 Cells-Mouse mammary epithelial cell line HC11 was maintained in RPMI 1640 medium (Invitrogen) supplemented with 10% FBS, 1ϫ GlutaMAX (Invitrogen), 5 g/ml insulin (Roche), and 10 ng/ml epidermal growth factor (EGF, R&D). To induce cell differentiation, HC11 cells were grown to confluence, incubated in medium deprived of EGF for 24 h, and then treated for 5 days with a combination of lactogenic hormones including 1 M dexamethasone (Sigma), 5 g/ml insulin, and 5 g/ml recom-binant mouse prolactin (R&D), referred to as DIP. 2 The mammospheres were photographed by phase contrast microscopy using a Olympus CKX41SF inverted microscope (Olympus, Tokyo, Japan) with a Canon EOS 600D digital camera (Canon, Tokyo, Japan).
Lentivirus Production and Cell Transduction-MISSION TM shRNAs targeting human ZFHX3 and mouse Zfhx3 in pLKO.1puro plasmid and empty vector control plasmid were purchased from Sigma. The insert sequences for these shRNAs are listed in Table 1. Lentiviral particles were produced in 293T cells by cotransfecting pLKO.1 with pMD2.G and psPAX2 plasmids using the FuGENE 6 transfection reagent (Promega, Beijing, China) according to the manufacturer's protocol. On the day after transfection, the medium was replaced, and the cells were incubated for an additional 2 days before viral supernatant was collected. Human T-47D or mouse HC11 cells were seeded in 6-well culture plates and grown to ϳ70% confluency, at which time the culture medium was replaced with fresh medium containing 8 g/ml of Polybrene (Sigma), and 1 ml of lentiviral solution was added dropwise. 6 -12 h after viral infection, the virus-containing medium was replaced with fresh medium again, and on the following day 2 g/ml of puromycin (Sigma) was added to the medium for 3-5 days to select cells stably expressing shRNAs. RNA was extracted, and protein lysate was prepared for analysis. Cells stably expressing the desired shRNAs were maintained in medium containing 1 g/ml of puromycin.
Reverse Transcription and Real Time PCR-Total RNA was used for cDNA synthesis with the Moloney murine leukemia virus reverse transcriptase system (Promega). RT-PCR was performed on the Mastercycler ep Realplex real time PCR system (Eppendorf, Hamburg, Germany) using the SYBR premix Ex Taq (TaKaRa Bio Inc., Tianjin, China). Human GAPDH or mouse ␤-actin gene was used as the internal control in real time PCR. Quantification of gene expression levels was determined using the comparative ‚‚CT method. Primer sequences used in real time PCR for each gene are listed in Table 2 and were from the PrimerBank (23) or the indicated references.
Immunoblotting-Cell lysate preparation and Western blotting were performed as previously described (20). Briefly, cells or tissues were lysed using radioimmune precipitation assay buffer supplemented with protease inhibitor (cOmplete, Roche) and phosphatase inhibitor (PhosStop, Roche). Protein concentrations were determined using a Pierce BCA protein , and molecular mass markers in kDa are indicated in applicable figure panels. The antibodies and dilution factors used in this study are listed in Table 3.

Whole Mount Preparation and Histological
Analysis of Mammary Glands-Whole mount analysis of mouse mammary glands was performed as previously described (20) except that whole mount glands were stained with carmine alum instead of hematoxylin. Briefly, the fourth abdominal mammary gland on the left was dissected at lactation day 2, placed between two glass slides and spread by placing weights on top of the slide for 10 min, followed by fixing in 3.7% formalin overnight. Tissues were hydrated, stained with carmine alum, and cleared with xylene. For histological analysis, formalin-fixed tissues were processed for sectioning, and stained with hematoxylin and eosin (H&E).
Immunohistochemistry-Immunohistochemical staining was performed on sections of formalin-fixed, paraffin-embedded tissues. Briefly, tissue sections were first deparaffinized and rehydrated following standard procedure. Antigen retrieval was carried out by heating in sodium citrate buffer or Tris-EDTA buffer using a pressure cooker for 3 min at full pressure. Sections were incubated with primary antibodies at 4°C overnight and then with the HRP solution and the DAB-chromogen. All sections were counterstained in hematoxylin, dehydrated in ethanol, cleared in xylene, and covered with a coverslip with Eukitt quick-hardening mounting medium (Sigma). Antibodies, dilution factors, and antigen retrieval buffers are listed in Table 3.
Statistical Analysis-Statistical analyses were performed using GraphPad Prism 5 software (GraphPad, San Diego, CA).
A two-tailed Student's t test was used to determine statistical differences between two groups, and one-way analysis of variance was used to compare three or more groups. p values less than 0.05 were considered statistically significant.

Expression of Zfhx3 Protein during Mammary Gland
Development-We previously measured Zfhx3 mRNA expression at different stages of mammary gland development and found that the Zfhx3 mRNA level varies at different stages, being the highest in late pregnancy and lactation (20). Here we determined Zfhx3 protein expression. Mammary glands were collected from mice at postnatal weeks 3, 6, and 9; pregnancy days 2 and 18; lactation days 1, 7, 14, and 20; and involution day 7. Western blotting demonstrated that the Zfhx3 protein level increased from weeks 3 to 9 after birth, further increased to an intermediate level during pregnancy, increased to and maintained a maximum level during lactation, and then dramatically reduced during involution (Fig. 1A). Even during lactation, the Zfhx3 protein level kept increasing, being higher at days 14 and 20 than at days 1 and 7 (Fig. 1A). This pattern of expression is almost identical to that for Zfhx3 mRNA expression (20).
We also performed IHC staining to detect Zfhx3 protein in the mammary gland at each developmental stage (Fig. 1B). We observed that Zfhx3 was expressed in both the epithelial cells and some types of stromal cells and was mainly localized in the nucleus. In a typical terminal end bud structure at postnatal week 3 (prepuberty) mammary gland, we could see two morphologically distinct cell types: the single-layered cap cells (outer layer, giving rise to basal cells) and multiple-layered body cells (centrally located, giving rise to luminal cells) (2). We observed that a significant portion of body cells in the terminal end bud were Zfhx3-negative (indicated by arrows in Fig. 1). During the pubertal and the early pregnancy stages when ductal morphogenesis occurs, we observed that some of the ductal luminal cells were also Zfhx3-negative (indicated by arrows in Fig. 1), but in contrast nearly all the mature alveolar luminal cells (indicated by arrowheads in Fig. 1) from the late pregnancy stage to the lactation stage were Zfhx3-positive, and at day 7 of involution Zfhx3 expression dramatically decreased. This pattern of expression strongly suggests that Zfhx3 may play a role in alveologenesis and lactogenic hormone-induced differentiation of mammary epithelial cells at lactation. Zfhx3 Is Necessary for Prolactin-induced Lactogenic Differentiation in HC11 Mammary Epithelial Cells in Culture-We first used the HC11 cell line model to test Zfhx3 function in lactogenic differentiation induced by lactogenic hormones. HC11 is a clone of the COMMA-1D mouse mammary epithelial cell line (24), derived from mammary tissue of BALB/c mice in the middle of pregnancy (25). Treatment with prolactin in the presence of insulin and glucocorticoid hormone dexamethasone (the DIP treatment) makes cultured HC11 cells produce milk protein ␤-casein, a marker of lactogenic differentiation (26), and form mammospheres in high density, making this cell line a widely used in vitro model of mammary gland differentiation in the context of prolactin and glucocorticoid stimulation (24).  (27), HC11 cells showed an expression level higher than T-47D cells but lower than MCF-7 cells, as detected by Western blotting (Fig. 2A). We then grew HC11 cells to confluence, incubated cells in EGF-depleted medium for 24 h, and treated them with a combination of lactogenic hormones including DIP to induce ␤-casein expression. Consistent with previous studies (28), an induction of ␤-casein was detected at both the mRNA and protein levels as expected (Fig. 2B, left and right panels). Meanwhile, hormone treatment did not alter Zfhx3 expression at either the mRNA or protein level (Fig. 2B, middle and right panels). Immunofluorescence staining  showed that Zfhx3 was localized in the nucleus regardless of hormone treatment (data not shown). These results validated the HC11 model for Zfhx3 study. We then knocked down Zfhx3 expression by lentivirus-mediated stable expression of shRNAs against Zfhx3 in HC11 cells. All four shRNAs induced significant knockdown of Zfhx3, as confirmed by real time PCR for Zfhx3 mRNA and Western blotting for its protein (Fig. 2C). Two of the shRNAs, shZfhx3-8 and shZfhx3-11, were more effective in Zfhx3 knockdown and thus were used for further experimentation. At day 5 of DIP treatment, ␤-casein mRNA (Fig. 2D, left panel) and protein (Fig. 2E) were significantly induced in the control shRNA group, whereas knockdown of Zfhx3 almost abolished the induction of ␤-casein (Fig. 2, D and E). Meanwhile, the connective tissue growth factor gene (Ctgf), which enhances lactogenic differentiation and only responds to dexamethasone but not to prolactin in its induction (29,30), did not show an expression change upon the knockdown of Zfhx3 (Fig. 2D, right panel). Parallel to ␤-casein induction, DIP-treated HC11 cells formed mammospheres during differentiation at high density in culture (Fig. 2F, left panel) as previously reported (31), but the sphere formation was interrupted by Zfhx3 knockdown (Fig.  2F, middle and right panel). These results suggest that Zfhx3 is necessary for DIP to induce lactogenic differentiation in HC11 cells, and the effect only involves the prolactin signaling, not the glucocorticoid signaling.
Deletion of Zfhx3 Attenuates Alveologenesis and Lactogenic Differentiation in Mouse Mammary Glands-To more directly test the necessity of Zfhx3 for lactogenic differentiation, we bred mice with floxed Zfhx3 alleles to MMTV-Cre mice in which the Cre recombinase is specifically expressed in mammary epithelial cells under the MMTV promoter, as described in our previous study (20). 9-week-old MMTV-Cre/Zfhx3 ϩ/ϩ , MMTV-Cre/Zfhx3 f/ϩ , and MMTV-Cre/Zfhx3 f/f female mice were mated and sacrificed at lactation day 2. Whole mount analysis showed that Zfhx3 deletion led to defects in alveologenesis, as indicated by reduced alveolar density and size (Fig.  3A). H&E staining of tissue sections demonstrated that Zfhx3 deletion caused the accumulation of large cytoplasmic lipid droplets in the luminal layer (Fig. 3B), which has been established as an indicator of failure in secretory activation in multiple lines of genetically engineered mice (32). We estimated the proportion of areas occupied by alveoli and the mean area of a single lumen in H&E-stained tissue sections using the ImageJ program. A significant decrease was detected in both parameters when Zfhx3 was deleted (Fig. 3, C and D).
We also measured the milk protein gene ␤-casein expression in these mice. The mRNA level of ␤-casein was significantly reduced by Zfhx3 deletion, as detected by real time PCR (Fig.  3E), consistent with the findings from HC11 cells. The protein expression of ␤-casein, as detected by IHC staining, was hardly detectable in some alveolar lumen or the mammary ducts, and the staining intensity was indistinguishable between wild type mice and those with Zfhx3 deletion (Fig. 3F), which could be due to quick emptying of the glands, as reported for IHC staining of the WAP milk protein in the Stat5a knock-out mammary glands at lactation day 1 (33). Lobulo-alveolar hypoplasia was evident in all four mice with homozygous deletion of Zfhx3.
Deletion of one Zfhx3 allele caused an intermediate abnormality in the size and density of alveoli, accumulation of cytoplasmic lipid droplets (Fig. 3B, middle panel). In addition, on day 2 after birth, although all litters from wild type mothers survived, most litters from mice with one Zfhx3 allele and all from Zfhx3-null mice died, likely because of dehydration. These findings indicate that Zfhx3 is required for proper development, maintenance, and function of structurally organized secretory lobulo-alveolar units in mammary glands.
Zfhx3 Is Required for Prolactin-induced Tyrosine Phosphorylation of Stat5-During lactogenic differentiation, prolactin binds to its receptor (PRLR), which then recruits the JAK2 kinase to phosphorylate STAT5 at Tyr 694/699 to transduce the signal into the nucleus, where lactogenic genes including milk protein ␤-casein are induced (34 -37). We therefore analyzed the phosphorylation status of Stat5 to determine whether Zfhx3 regulates alveologenesis and lactogenic differentiation by the canonical PRLR-JAK2-STAT5 signaling pathway.
In the HC11 cell system, treatment with prolactin for 7.5 min clearlyinducedStat5phosphorylationasexpected,andthephosphorylation was still detectable after 2.5 h of treatment and hardly seen after 24 h (Fig. 4A). When Zfhx3 was knocked down in HC11 cells, prolactin-induced Stat5 phosphorylation was dramatically decreased, whereas total Stat5 showed similar expression levels (Fig. 4B). In the T-47D human breast cancer cell line, which expresses a higher level of PRLR (38), prolactin treatment also significantly induced STAT5 phosphorylation as in HC11 cells (Fig. 4C). Stable knockdown of ZFHX3 expres- C, quantification of the total area filled by lactating alveoli in the H&E-stained sections using ImageJ. At least four mice were analyzed, and for each section, at least 20 images (ϫ20) were used to analyze the area occupied by alveoli. D, quantification of the mean area of a single alveoli in the mammary glands using ImageJ. A total of at least 1000 alveoli were analyzed per genotype. For C and D, the data are means Ϯ S.E. E, expression of ␤-casein by real time PCR; the data are means Ϯ S.E. F, IHC staining of ␤-casein at lactation in the mammary glands; arrows show empty alveolar lumen, and arrowheads show empty ducts. Statistical significance was calculated by two-tailed Student's t test. ***, p Ͻ 0.001. sion using shRNA in T-47D cells, which was confirmed by real time PCR and Western blotting (Fig. 4D), compromised the phosphorylation of STAT5 (Fig. 4C). Similar results were obtained using a different shRNA(s) in both HC11 and T-47D cells (data not shown). These results showed that Zfhx3 is necessary for STAT5 phosphorylation not only in the normal mammary epithelial cells but also in breast cancer cell lines, indicating that ZFHX3 protein may play a role in both normal development and tumorigenesis of the breast.
In mouse mammary tissues at lactation day 2 with different Zfhx3 deletion status, IHC staining demonstrated that although most epithelial cells had intense staining for phospho-Stat5 (Fig. 4E, left panel), homozygous deletion of Zfhx3 almost eliminated the staining (Fig. 4E, right panel). Deletion of one Zfhx3 allele showed an intermediate effect in both the p-Stat5 staining intensity and the number of p-Stat5-positive cells (Fig.  4E, middle panel). The IHC staining index (IHC score) for p-Stat5, which was produced by multiplying the percentage of positive cells (scored as 0 for negative, 1 for Ͻ10%, 2 for 11-50%, 3 for 50 -80%, and 4 for Ͼ80%) with the intensity of staining (0 for negative, 1 for weak, 2 for moderate, and 3 for strong), was significantly lower in Zfhx3-null tissue than in normal tissue (Fig. 4F). These observations suggest that Zfhx3 plays a role in the activation of Stat5 by prolactin-Prlr signaling.
Zfhx3 Plays a Role in the Expression of Prlr-In a previous work, Yan et al. (39) carried out ChIP-seq experiments using ChIP grade antibodies of more than 100 transcription factors, including ZFHX3. Reanalyzing their data using the ChIP-Enrich tool (40) with annotation of all peaks to the nearest transcription start sites, we found that a number of potential target genes of ZFHX3, including PRLR, played crucial roles in mammary gland development (data not shown). We therefore tested whether ZFHX3 regulates PRLR expression to modulate prolactin-PRLR signaling. We found that in mouse mammary glands at lactation day 2, deletion of Zfhx3 decreased the Prlr mRNA level, as detected by real time PCR (Fig. 5A). The decrease in Prlr mRNA level by Zfhx3 deletion was already significant at postnatal week 9 (data not shown), suggesting that Prlr down-regulation is independent of lactogenesis. We also detected Prlr protein in mouse mammary tissues by IHC staining and found that deletion of Zfhx3 also down-regulated Prlr protein at lactation day 2 (Fig. 5, B and C). Down-regulation of Prlr was also induced by the knockdown of Zfhx3 in HC11 cells at both the mRNA level (Fig. 5D) and the protein level (Fig. 5E). Consistent with the down-regulation of Prlr, the expression of two established players downstream of prolactin-Prlr signaling, Elf5 and Id2, was also down-regulated by Zfhx3 knock-out in mouse mammary glands at lactation day 2 (Fig. 5F), although the expression of two other prolactin signaling molecules, Lmo4 and Cebpb, was not altered by Zfhx3 deletion (Fig. 5F). Using the cBioPortal web tool (41,42), we analyzed the expression of ZFHX3 and PRLR mRNAs in 1100 breast cancer samples (TCGA, Provisional), and found a positive correlation between the expression of the two genes (Fig. 5G). These results indicate that ZFHX3 plays a crucial role in the expression of PRLR in mammary epithelial cells, which could mediate the effect of Zfhx3 on PRLR-JAK2-STAT5 signaling and mammary gland development.

Discussion
Postnatal mammary gland development is a multistage process involving a sequence of morphological and molecular changes (4,32). The transition from pregnancy to lactation activates the secretion of all the nutrients required for the newborns, including milk proteins, lactose, and milk fat. Like other stages of mammary gland development, alveologenesis and lactogenic differentiation are controlled by hormones such as progesterone and prolactin and key transcription factors (4,32). In this study, we performed histological and molecular analyses using in vitro and in vivo models to establish the Zfhx3 transcription factor as a crucial regulator of normal alveologenesis and lactogenic differentiation. We further demonstrated that FIGURE 4. Zfhx3 is necessary for prolactin-induced phosphorylation of STAT5. A, parental HC11 cells were treated with 5 g/ml prolactin in a time-dependent manner and Stat5 phosphorylation at tyrosine 694 (p-Y694-Stat5) was detected by Western blotting. B, mouse HC11 mammary epithelial cells with stable knockdown of Zfhx3 (shZfhx3-8) or control (shCtrl) were treated with insulin (I), insulin plus dexamethasone (DI), or DIP for 7.5 min or 2.5 h and then subjected to Western blotting to detect p-Y694-Stat5 and total Stat5. C and D, stable knockdown of ZFHX3 in human breast cancer cell line T-47D also reduced prolactin (Prl)-induced STAT5 phosphorylation, as detected by Western blotting. Knockdown efficiency was confirmed by RT-PCR and Western blotting (D). E and F, IHC staining of p-Y694-Stat5 in lactation day 2 mammary glands (E). IHC score is shown as means Ϯ S.E. (F) and was produced by multiplying the percentage of positive cells (scored as 0 for negative, 1 for Ͻ 10%, 2 for 11-50%, 3 for 50 -80%, and 4 for Ͼ 80%) with the intensity of staining (0 for negative, 1 for weak, 2 for moderate, and 3 for strong). Statistical significance was calculated by one-way analysis of variance followed by Bonferroni's test. *, p Ͻ 0.05; **, p Ͻ 0.01. All Western blots were quantified using the ImageJ program.
We first demonstrated that the Zfhx3 protein level varied at different stages of mammary gland development and reached a peak in late pregnancy and lactation (Fig. 1), which is consistent with the mRNA level of Zfhx3 reported previously (20). Based on transcriptomic analysis of non-human primate breast tissues at prepubertal, adolescent, adult luteal, pregnant, lactating, and postmenopausal stages, four distinct patterns of gene expression were noticed during reproductive life, including the juvenile pattern, adult pattern 1, adult pattern 2, and the lactational pattern (43). The pattern of Zfhx3 expression fits between the adult pattern 1 and the lactational pattern, and genes with either pattern of expression likely play a role in prolactin signaling (43). This appears to be the case for Zfhx3.
Our functional studies indicated the essential role of Zfhx3 in prolactin signaling and lactogenic differentiation. Prevention of Zfhx3 expression compromised alveologenesis and lactogenic differentiation. In the mouse HC11 cell line, a well established and widely used in vitro model of lactogenic differentiation, knockdown of Zfhx3 compromised prolactin-induced cell differentiation including ␤-casein expression (Fig. 2). More directly, when Zfhx3 was deleted in mouse mammary glands, the number and size of alveoli decreased, milk protein secretion was attenuated, and litters died after birth, likely because of dehydration (Fig. 3). Zfhx3 is essential for proper alveologenesis and lactogenic differentiation.
It has been established that lactation involves the expression of numerous genes and the secretion of their gene products in the milk (43,44). Our previous study demonstrated that dele- FIGURE 5. Zfhx3 is needed for the expression of Prlr in mouse mammary epithelial cells. A-C, deletion of Zfhx3 down-regulates Prlr expression, as detected by RT-PCR analysis for Prlr mRNA (total and long form, A) or by IHC staining for Prlr protein (B) at day 2 of lactation. C, IHC score is shown as means Ϯ S.E., statistical significance was calculated by one-way analysis of variance followed by Bonferroni's test; **, p Ͻ 0.01). D and E, detection of Prlr expression in HC11 cells with stable knockdown of Zfhx3 by RT-PCR (total and long form, D) or Western blotting (long form, E), band intensities in Western blot were quantified using the ImageJ program. F, Zfhx3-mediated gene regulation in mammary alveolar development. Gene expression analysis of mammary glands from MMTV-Cre/Zfhx3 ϩ/ϩ and MMTV-Cre/Zfhx3 f/f mice at day 2 of lactation. At least four mice per genotype were used; the data are means Ϯ S.E. Statistical significance was calculated by two-tailed Student's t test using GraphPad Prism 5; *, p Ͻ 0.05. G, the mRNA expression data of ZFHX3 and PRLR in breast invasive carcinoma (TCGA, Provisional) were obtained using cBioPortal. The case set contained 1100 tumor samples. Pearson or Spearman correlation coefficients (r) and two-tailed p value calculations to determine the significance of the correlations were calculated using GraphPad Prism 5. H, the 5Ј region of mouse Prlr and the location of five alternative first exons (represented by black boxes). E2, exon 2. tion of Zfhx3 in mouse prostates dysregulates a large number of genes, and those encoding for secretory and membrane proteins are the most affected gene groups (17). Using ␤-casein as a representative of milk proteins, we found that knockdown in HC11 cells and knock-out in mammary tissues indeed interrupted ␤-casein expression (Figs. 2 and 3), further indicating that Zfhx3 is necessary for lactogenic differentiation and the production of milk.
Mechanistically, our findings suggest that Zfhx3 regulates alveolar and lactogenic differentiation by maintaining prolactin-Prlr signaling activity, which is the single most potent regulator of alveologenesis and lactogenic differentiation (45,46). In this signaling pathway, binding of prolactin to PRLR activates JAK2 kinase, which then phosphorylates the STAT5 transcription factor, leading to the up-regulation of the Elf5 transcription factor and the expression of genes underlying alveolar and lactogenic differentiation (33,(47)(48)(49)(50)(51). We found that Zfhx3 was not only needed for the expression of the milk protein ␤-casein ( Figs. 2 and 3), an indicator of the PRLR-STAT5 signaling axis, it was also necessary for the phosphorylation of Stat5 in both mouse mammary tissues and cultured HC11 mouse mammary epithelial cells (Fig. 4). Furthermore, we found that Zfhx3 was required for the expression of Prlr. For example, deletion of Zfhx3 in mouse mammary tissues and knockdown of Zfhx3 in cultured HC11 mouse mammary epithelial cells significantly down-regulated Prlr expression at both the mRNA and protein levels (Fig. 5). Consistently, a significant correlation was detected between mRNA levels of ZFHX3 and PRLR in hundreds of human breast cancer samples (Fig. 5G). Therefore, maintaining PRLR expression and PRLR-JAK2-STAT5 signaling activity could be the major mechanism by which ZFHX3 regulates alveolar and lactogenic development of the mammary gland.
The promoter of PRLR is complicated, and so is its transcriptional regulation. For example, in addition to the abundantly expressed and widely conserved exon 1 among different species, the PRLR gene has multiple additional first exons because of alternative splicing in different tissues (52). Different tissues can have different 5Ј-regions of the Prlr gene (53). Human PRLR also has multiple first exons that employ distinct mechanisms for controlling PRLR transcription (54). In mice, five alternative first exons have been identified for Prlr, including E1 1 , E1 2 , E1 3 , E1 4 , and E1 5 (Fig. 5H) (55). Little is known about how the putative promoters upstream of these first exons drive Prlr transcription and what factors bind to the promoter sequences. We cloned the putative promoters of the five first exons and measured their promoter activities using dual luciferase reporter assay. Except for the E1 5 promoter, which showed a slight change in promoter activity, none of the other four putative promoters showed detectable changes when Zfhx3 was knocked down in HC11 cells (data not shown). It remains to be determined whether Zfhx3 directly binds to Prlr promoter, and if so where it binds. Porcine Prlr promoter, which is the best studied so far among different species for the potential use of Prlr as a marker for reproductive performance (56 -59), has 12 alternative first exons (52). Similar to porcine Prlr, the 5Ј region of mouse Prlr is also quite large, spanning more than 120 kb (Fig. 5H). It is thus possible that mouse Prlr has more alternative first exons. Furthermore, whereas STAT5 can bind to an interferon-␥-activated sequence (GAS) in the PRLR gene to regulate its transcription (60), it remains to be determined whether ZFHX3 interacts with STAT5 on the same promoter region. Currently we are in the process of examining how ZFHX3 regulates the transcription of PRLR in mammary epithelial cells.
Regulation of the PRLR-JAK2-STAT5 signaling by ZFHX3 could play an important role in human breast cancer, because the level of phosphorylated STAT5 gradually decreases during breast carcinogenesis, whereas the total expression level of STAT5 does not appear to change. The decrease of phosphorylated STAT5 is correlated with poorer patient survival and predicts both poor clinical outcome and increased risk of antiestrogen therapy failure (61)(62)(63)(64)(65). PRLR has been shown to suppress epithelial-mesenchymal transition and invasiveness of breast cancer cells (66), and PRLR expression is significantly down-regulated in invasive breast cancer, the down-regulation being associated with lymph node metastasis, lower tumor grade, and metastasis-free survival in breast cancer patients (67). Consistent with the findings of p-STAT5 and PRLR expression, the ZFHX3 mRNA level has also been shown to be frequently down-regulated in human breast cancer (68), which is likely mediated by hemizygous deletion (69), and again downregulation was correlated with a worse patient survival (68). Indeed, a tumor suppressor function has been established for ZFHX3 in prostate cancer, because mutation of ZFHX3 is frequent in human prostate cancer (70,71), and deletion of Zfhx3 in mouse prostates induced neoplastic lesions (17). These findings suggest that ZFHX3-mediated PRLR-JAK2-STAT5 activity plays a suppressive role in breast cancer development. Currently we are testing this hypothesis by investigating whether Zfhx3 deletion induces neoplastic lesions in mouse mammary glands.
In addition to Prlr, other molecules could also mediate the effect of Zfhx3 in alveolar and lactogenic development of the mammary gland. For example, deletion of Zfhx3 in mouse mammary glands led to the down-regulation of Elf5 and Id2 (Fig. 5F), both of which have established roles in mammary gland development. Id2 mediates the function of RANKL in the survival and proliferation of mammary epithelial cells (72,73), and deletion of Id2 leads to lactation defects in mice (74). The Elf5 transcription factor has a similar effect on mammary gland development as Zfhx3, and deletion of Elf5 leads to failure of Stat5 activation and defects in alveologenesis (75). Elf5 has also been demonstrated to regulate mammary gland stem/progenitor cell fate (76), although such a role has not been examined for Zfhx3. Both Elf5 and Zfhx3 are transcription factors, but whether they directly interact on gene promoters remains to be clarified. Recently Zfhx3 was also shown to modify circadian function by directly activating circadian genes Cry2, Per1, and Per2 (77). Interestingly, both Per1 and Per2 have been implicated in the development and differentiation of mouse mammary gland (78,79). Therefore, it is more likely that Zfhx3 regulates alveologenesis and lactogenic differentiation by interacting with or regulating the expression of other factors.
In summary, we found in this study that the Zfhx3 protein level varied during mammary gland development, reaching the highest level in late pregnancy and lactation. Knockdown of Zfhx3 in HC11 mammary epithelial cells and deletion of Zfhx3 in mouse mammary tissues attenuated prolactin-induced alveologenesis and lactogenic differentiation, as indicated by morphological and molecular alterations. Mechanistically, knockdown and knock-out of Zfhx3 decreased the expression level of Prlr, Elf5, Id2, and phosphorylated Stat5. These findings suggest that Zfhx3 plays an essential role in proper development of the mammary gland in pregnancy and lactation by maintaining the activity of the PRLR-JAK2-STAT5 signaling pathway.