STAT3 and MCL-1 associate to cause a mesenchymal epithelial transition

Successful implantation, and thereby pregnancy, is a hormone-driven process. Several essential cytokines and growth factors regulate the steroidal actions of hormones in order to prepare the endometrium for implantation. STAT3, a versatile member of the family known as ‘signal transducers and activators of transcription’ (STAT), mediates the axial responses of cytokines. STAT3 is involved in normal cellular responses, as well as oncogenesis (Takeda and Akira, 2000). In resting cells, STATs, including STAT3, are localized to the cytoplasm. Upon activation by cytokines, STATs are known to be phosphorylated by Janus kinases, which leads to formation of homo- or heterodimers through interactions between the Src homology 2 (SH2) domains and phosphorylated tyrosine residues. STAT3 then rapidly translocates to the nucleus, where it increases the expression of target genes. STAT3 plays an essential role in interleukin (IL)-9-induced expression of primary response genes, such as c-Myc and Cited2 (Zhu et al., 1997). Increased phosphorylation of STAT3 is known to be associated with elevated expression of potential downstream targets of STAT3 – these include the genes encoding apoptosis inhibitors [survivin, MCL-1, HSP27 (also known as HSPB1), adrenomedullin and Bcl-xL], cell cycle regulators (cyclin D1, cyclin-dependent kinase inhibitor 1, c-Fos, MAP2K5 and c-Myc) and inducers of tumor angiogenesis [vascular endothelial growth factor, COX-2 (also known as PTGS2) and the matrix metalloproteases (MMP)-2, MMP-10 and MMP-1] in invasive breast cancer tissues (Hsieh et al., 2005; Sinibaldi et al., 2001). The pleiotropic response of STAT3 has been attributed to its ability to act downstream of receptors for a number of IL-6-family cytokines, including IL-6, IL-11, ciliary neuorotrophic factor, oncostatin M, leukemia inhibitory factor (LIF) and interferon family members (IL-10, IFN-γ and IFN-α) (Heinrich et al., 1998; Kisseleva et al., 2002). Furthermore, STAT3 has been shown to mediate transcriptional responses to granulocyte colony-stimulating factor, leptin (Akira, 1997), receptor tyrosine kinases (epidermal growth factor, colony stimulating factor 1 and platelet-derived growth factor) and IL-2 family members (IL-2, IL-7 and IL-15) (Zhong et al., 1994).

Members of the IL-6 family of cytokines are expressed during implantation in the mouse (Bhatt et al., 1991) and human (Charnock-Jones et al., 1994; Nachtigall et al., 1996). In vivo STAT3 activation, induced by LIF alone, is restricted to day 4 of pregnancy, resulting in the localization of STAT3 specifically to the nuclei of cells in the luminal epithelium, which coincides with the onset of uterine receptivity (Cheng Jr et al., 2001). Targeted disruption of Stat3 revealed that this protein is essential for early embryonic development (Takeda et al., 1997). STAT3 signaling also plays a crucial role in the induction of antigen-specific T-cell tolerance (Cheng et al., 2003). Thus, in order to elucidate the role of STAT3 in the process of embryo implantation, we investigated which proteins this transcription factor interacts with at the time of embryo implantation.

In this study, we show that STAT3 physically and functionally interacts with its downstream target myeloid cell leukemia-1 (MCL-1) during the ‘window of implantation’. We demonstrate the interaction using co-immunoprecipitation, far-western analysis, colocalization of the two proteins in uterine sections, and with co-transfection studies in MCF7 cells, and show that the interaction is strongly mediated by the ovarian hormones estrogen and progesterone. The epithelial mesenchymal transition (EMT) is characterized by changes in protein expression, increased invasion and resistance to apoptosis. In this study, we show that STAT3 interacts with MCL-1, which leads to increased apoptosis, decreased invasion and upregulation of certain proteins, such as E-cadherin (also known as cadherin-1) and cytokeratins, which are normally downregulated during the EMT.

Like many proteins, STAT3 mediates its action by interacting with a number of other proteins. Some of the known interacting partners of STAT3 are coactivators, such as CBP–p300 and NCoA/SRC1a, which possess intrinsic histone acteyltransferase activity and enhance the transcriptional activity of STAT3 (Giraud et al., 2002). TIP60 (also known as KAT5) interacts with STAT3; it binds to the α-chain of the IL-9 receptor, suggesting that Tip60 might be involved in IL-9 signaling (Xiao et al., 2003). Although STAT3 is known to play a key role during embryo implantation, the protein networks that it interacts with in the uterus are not known. Thus, we initiated experiments to identify partners of STAT3 that enable it to perform its functions.

Immunoprecipitation of STAT3 using nuclear or cytosolic extracts, prepared from uteri, resulted in a limited number of protein bands that stained positively with Coomassie dye. Each band was trypsin digested and then subjected to matrix-assisted laser desorption/ionization (MALDI) analysis. The results identified possible interacting partner proteins of STAT3 (Fig. 1A,B). The band at the 37-kDa-position in the nuclear immunoprecipitation (GB8) was subjected to MALDI analysis, this analysis strongly pointed towards the possibility of the protein being MCL-1, a member of the BCL-2 family proteins, which regulate apoptosis (Fig. 1C). MCL-1 has sequence similarity to BCL-2. BCL-2 is involved in normal lymphoid development and a t(14; 18) chromosome translocation in the gene results in lymphoma (Kozopas et al., 1993). The presence of STAT3 in the nuclear immunoprecipitate was confirmed by peptide fingerprint, followed by Mascot analysis, of the band at 88 kDa (GB4) (Fig. 1C). Therefore, we focused our efforts towards establishing the relationship between the STAT3 and MCL-1 proteins during the time of embryo implantation.

To establish the authenticity of the interaction between STAT3 and MCL-1, we investigated whether MCL-1 co-precipitated with STAT3. Using antibodies against STAT3, STAT3 was immunoprecipitated from uterine nuclear and cytosolic extracts. An MCL-1-positive band was detected when the STAT3 precipitates were probed with antibody to MCL-1 (Fig. 1D,E). The presence of STAT3 in these immunoprecipitates was confirmed by western blotting with an antibody against STAT3. For nuclear and cytosolic extracts, protein input was shown by histone H3 and actin, respectively (lower panels). We further confirmed the interaction by immunoprecipitating MCL-1 from the nuclear (Fig. 2A) and cytosolic (Fig. 2B) extracts from different days of pregnancy. MCL-1 immunoprecipitates showed a STAT3-positive band (Fig. 2Ca) in the uterine nuclear extracts on all the tested days of pregnancy, suggesting the existence of an association between STAT3 and MCL-1 during the window of implantation. The western blot of MCL-1 confirms the presence of MCL-1 in the immunoprecipitate (Fig. 2Cb). In the case of immunoprecipitation using the uterine cytosolic extracts, MCL-1 did not pull down STAT3, and vice versa, in the cytosol from the sample taken on day 4 (D4) 10:00 (Fig. 2Cb). A possible explanation is that the low MCL-1 levels indicate a lack of cytosolic distribution at D4 10:00 and thus might be unavailable for interaction with STAT3 at this stage. It is evident from the results of the immunoprecipitations that the STAT3–MCL-1 interaction is prominent in the nucleus at the early peri- and peri-implantation period (D4 16:00 and D5 05:00), which is then reduced in the post-implantation period (D5 10:00). Although the STAT3–MCL-1 interaction is retained in the cytosol during the peri-implantation phase, there is a conspicuous lack of interaction between the two proteins in the cytosol at the pre-implantation period (D4 10:00). Mock-immunoprecipitation and secondary-antibody-alone controls show the specificity of the antibodies used (supplementary material Fig. S1).

To test the interaction between STAT3 and MCL-1, far-western analysis was performed. The day D4 16:00 cytosolic extract, where the interaction between STAT3 and MCL-1 was prominent, was used as the prey. Purified STAT3 protein was used as the bait and bovine serum albumin (BSA) was used as a control. The blot probed with an antibody against STAT3 revealed the classical 41 kDa band that indicates MCL-1_L, an MCL-1 variant that is encoded by the myeloid cell leukemia-1 gene and known to migrate at 40–42 kDa. The band indicating MCL-1_L is marked with an asterisk and arrow (Fig. 2Da). The position of the band was confirmed when the blot was developed using an antibody against MCL-1 (Fig. 2Db). The secondary-antibody alone shows two nonspecific albumin-immunopositive bands (NS1 and 2) in the cytosolic extract and BSA samples (Fig. 2Dc). Fig. 2Dd shows the protein load by Coomassie staining. The membrane that was probed with an antibody against STAT3 showed several bands because recombinant STAT3 would have also interacted with other proteins in the cytosolic extract.

To establish further an interaction between STAT3 and MCL-1, we performed co-immunoprecipitation with an MCL-1 antibody using total extracts from MCF7 cells transfected with STAT3 or MCL-1 alone, or with STAT3 and MCL-1 in combination. The MCL-1 immunoprecipitates were then western blotted with an antibody against STAT3, which showed that STAT3 interacts with MCL-1 in cells overexpressing STAT3 (Fig. 2E).

Having established that an interaction between MCL-1 and STAT3 occurs, we sought to identify the expression pattern of the two proteins during the crucial time of embryo adhesion in order to understand the functional relevance of the interaction. Consequently, immunocytochemical studies were performed to visualize the localization of STAT3 and MCL-1 in the uterine sections on different days of pregnancy. STAT3 was labeled with secondary antibody conjugated to fluorescein isothiocyanate (FITC) and MCL-1 was labeled with secondary antibody conjugated to tetramethylrhodamine (TRITC). The nuclei of the endometrial epithelium is seen to be devoid of MCL-1 and STAT3 as both proteins show localization to the membrane. By contrast, stromal cell nuclei showed weak STAT3 expression at D4 10:00 of pregnancy. An increased expression of STAT3 and MCL-1 is evident at D4 16:00 in comparison with D4 10:00 (Fig. 3A,B) with their localization prominently in the membrane of the endometrial epithelia and cytosol, as well as nuclei of stromal cells. This increased expression is maintained at peri-implantation stages but there is a spatio-temporal variation in the distribution pattern. The early peri-implantation uteri (D4 16:00) and peri-implantation uteri (D5 05:00) are evident by the presence of STAT3 and MCL-1 in the the nuclei of cells in the stroma and, the stromal expression of the proteins showing a dotted appearance, suggesting that the two proteins accumulate in nuclear speckles (Fig. 3B). The fluorescence, indicating STAT3 and MCL-1, was redistributed as the uterus prepared to receive an invasive-phase embryo – showing that STAT3 began to translocate from the nucleus to membrane and cytosolic compartments in stroma. STAT3 also showed increased expression in epithelial cells at the late peri-implantation time period, D5 10:00 (Fig. 3A). The remaining STAT3 and MCL-1 continued to colocalize in nuclear speckles (Fig. 3B). Reorganization of MCL-1 and STAT3 in the nucleus of cells towards the endometrial epithelium and stroma, where the embryo has to implant, could signal the death of cells in the luminal epithelium and stroma in order to accommodate the invading embryo. The overlapping expression pattern of STAT3 and MCL-1 further strengthens the possibility that the two proteins associate, and their presence in the implantation site indicates that STAT3 and MCL-1 play a key role in embryo implantation.

As implantation is regulated by the steroidal hormones estrogen and progesterone, we investigated the effect of these two hormones on the interaction between MCL-1 and STAT3 in a time-dependent manner. The experiments were performed using MCF7 cells, a cell line in which STAT3 and MCL-1 have previously been shown to be expressed (Ding et al., 2007; Berishaj et al., 2007). The results showed that, before treatment with hormones (0 hour), STAT3 and MCL-1 were located in the cytosol (Fig. 4A). In agreement with reported nuclear shuttling of STAT3 (Meyer and Vinkemeier, 2004), STAT3 was also observed in the nucleus. After 30 minutes of treatment with estrogen, STAT3 began to accumulate in foci in the nucleus, and reduced levels of STAT3 were observed in the cytosol; however, the STAT3 remaining in the cytosol was seen to colocalize with MCL-1. After 24 hours of treatment with estrogen, the majority of STAT3 became nuclear, whereas a substantial portion of MCL-1 remained cytosolic; however, in the nucleus, STAT3 and MCL-1 colocalized in a dotted pattern (Fig. 4B). Colocalization of STAT3 with MCL-1 was predominantly in the nucleus after 30 minutes and 24 hours of progesterone treatment (Fig. 4A,B). The pattern of expression was comparable to that after 24 hours of estrogen treatment. The same pattern of expression was seen when progesterone and estrogen were given in combination for 24 hours. Hence, both hormones regulate the expression and nuclear translocation of STAT3 and MCL-1. The expression of STAT3 and MCL-1 is more targeted to peri-nuclear bodies (indicated by the yellow dots in the merged image of STAT3 and MCL-1 fluorescence, and the white dots in the merged images of STAT3, MCL-1 and DAPI fluorescence; Fig. 4A,B). This pronounced colocalization of STAT3 with MCL-1, which is especially apparent in nuclear bodies, is still visible after 24 hours of treatment with either hormone alone, or in combination (Fig. 4A,B). Treatment of the cells with estrogen and progesterone together causes a large fraction of MCL1 and STAT3 to relocate to the cytosol and also results in reduced expression of the two proteins.

The MCF7 cells were co-transfected with STAT3–EGFP and MCL-1–DsRed using Lipofectamine 2000 (Invitrogen). The proteins were expressed for 48 hours after transfection, and the pattern of expression was observed by using a confocal microscope. As shown in Fig. 5A, colocalization of STAT3 and MCL-1 was restricted to the cytosol before treatment with hormones. Although some nuclear expression of STAT3 is seen, the nuclei are devoid of MCL-1 at t = 0 (0 hour). After 24 hours of treatment with 1 nM estrogen, both STAT3 and MCL-1 showed increased accumulation in the nucleus, a substantial fraction also surrounded the nucleus. The merged image clearly shows overlay of staining of the proteins, indicating that they colocalize in the cell. After 24 hours of treatment with 10 nM progesterone, a fraction of STAT3 but not MCL-1 entered the nucleus, even though the two proteins clearly colocalized in the cytosol. This mimics the scenario of D4 10:00 and D4 16:00 uteri, which are not subjected to 24 hours of estrogen surge in vivo. The lack of endogenous MCL-1 entry into the nucleus in comparison with the nuclear entry of overexpressed exogenous MCL-1 in MCF7 cells (Fig. 4 compared with Fig. 5A) could be due to the different expression levels of the proteins. STAT3 and MCL-1 evidently colocalized in the nucleus when estrogen and progesterone were given in combination (Fig. 5A), which is in accordance with the colocalization of endogenous STAT3 and MCL-1 in MCF7 cells. This is also in line with our observations in uterine sections of D5 05:00, that is, a stage equivalent to 24 hours of estrogen treatment in a progesterone-stimulated environment (Fig. 3). Thus, it appears that both hormones are required for efficient functioning of STAT3 and MCL-1.

Having demonstrated an interaction between MCL-1 and STAT3, we wanted to probe which domains of STAT3 mediate this interaction. MCF7 cells were transfected, as described above, with MCL-1–DsRed and constructs encoding domains of STAT3 tagged with enhanced green fluorescent protein (EGFP), namely the protein interaction domain (STAT3-PID), alpha domain, DNA-binding domain (DBD) and SH2 domain (STAT3-SH2) (Fig. 5B). Colocalization of MCL-1 with STAT3-PID and STAT3-SH2 in the nucleus was detected by 24 hours of estrogen treatment; however, in transfections using the constructs encoding the STAT3 alpha domain and DBD, MCL-1 did not enter the nucleus (Fig. 5C), suggesting that MCL-1 interacts with PID and SH2 domains to enter the nucleus.

JAK-mediated tyrosine phosphorylation regulates the dimerization of STATs and is a prerequisite for the establishment of a classical JAK–STAT3 signaling pathway. Tyr705 phosphorylation is attributed to STAT3 activation (Bowman et al., 2000; Decker and Kovarik, 2000; Levy and Lee, 2002); therefore, we examined the phosphorylation status of STAT3 upon its interaction with MCL-1. We immunoprecipitated MCL-1 from MCF7 cells that had been co-transfected with plasmids encoding STAT3 and MCL-1 and treated with or without estrogen. The precipitates were then probed for phosphorylated STAT3. We observed decreased levels of phosphorylated STAT3 in estrogen-treated cells that had been co-transfected with MCL-1 and STAT3 (Fig. 6A).

To analyze whether the activity of STAT3 at the promoters of response genes is affected by the interaction with MCL-1, luciferase assays were performed using the Cignal STAT3 Reporter Assay Kit. Plasmids encoding STAT3 and MCL-1 were transfected into MCF7 cells along with the STAT3-responsive luciferase construct. The constitutively expressing Renilla construct encodes the Renilla luciferase reporter gene and acts as an internal control for normalizing transfection efficiencies and monitoring of cell viability. Luciferase activities of the test samples were normalized with promoter-alone controls and then plotted. In cells transfected with STAT3 and MCL-1, the assay showed a 4-fold decrease in the activity of STAT3 on the promoter upon treatment with 1 nM estrogen in comparison with cells without estrogen (P<0.003) (Fig. 6B). The STAT3 promoter activity was high in the absence of estrogen in co-transfected cells. The controls, STAT3 alone and MCL-1 alone, showed high STAT3 promoter activity in the absence of estrogen. In the presence of estrogen, cells transfected with STAT3 alone showed a 21-fold downregulation (P<0.0196) of promoter activity. By contrast, cells transfected with MCL-1 alone showed a statistically insignificant downregulation of promoter activity in estrogen-treated cells compared with untreated cells that had been transfected with MCL-1 (P<0.05904) (Fig. 6B). As the presence of MCL-1 rescued the inhibition of the transcriptional potential of STAT3 upon treatment with estrogen from 21-fold to 4-fold, the results indicate that MCL-1 is a positive modulator of STAT3 function in the presence of estrogen and acts to relieve the repressive effects of estrogen on STAT3 promoter activity.

To analyze whether the STAT3–MCL-1 interaction varies with the cell cycle dynamics, we performed cell cycle analysis using fluorescence-activated cell sorting. The cells were synchronized using serum starvation and then treated with estrogen. We observed a substantial increase in the number of cells in the G2/M stage when cells were transfected with only MCL-1, or co-transfected with STAT3 and MCL-1 compared with controls (Fig. 6C). This correlates with earlier reports that transient transfection with MCL-1 increases the expression of phospho-Ser345 Chk1 in the absence of DNA damage and leading to accumulation of cells in G2 (Jamil et al., 2008), which suggests a new role for MCL-1 in generating an appropriate response to DNA damage and in the maintenance of chromosome integrity (Jamil et al., 2008).

The loss of uterine epithelial cells surrounding the embryo is crucial for bringing the blastocyst into close proximity with the uterine endometrium for successful embryo implantation. It has been shown previously that the epithelial cells are phagocytosed by trophoblast cells after apoptotic cell death (Parr et al.,1987); hence, apoptosis is vital during embryo implantation, and we, therefore, used the Dual Apoptosis Assay Kit to measure the degree of apoptosis stimulated by the association of STAT3 with MCL-1 in MCF7 cells. Cells transfected with MCL-1 alone showed a 2-fold increase in apoptosis in the presence of estrogen; by contrast, cells transfected with STAT3 alone showed a 2.7-fold increase in apoptosis in the presence of estrogen (Fig. 6D). This correlates with the downregulation of STAT3 promoter activity in the presence of estrogen. A previous report has shown that inhibition of STAT3 activity induces apoptosis and promotes expression of anti-apoptotic genes in pancreatic cancer cell lines (Glienke, 2011). In cells co-transfected with STAT3 and MCL-1, the degree of apoptosis was reduced to a level that is less than that of cells transfected with MCL-1 alone (∼73% reduction) in the presence of estrogen. Hence, MCL-1 protects cells from the effect of estrogen on STAT3 activity, proving the anti-apoptotic role of MCL-1. The action of progesterone on these transfected cells was similar to that of estrogen on cells transfected with STAT3 or MCL-1 individually. Cells transfected with MCL-1 alone showed a 1.8-fold increase in the levels of apoptosis in the presence of progesterone, and cells transfected with STAT3 alone showed a 2.6-fold increase in the presence of progesterone. However, when STAT3 and MCL-1 were coexpressed, the degree of apoptosis was increased to almost 5-fold in the presence of progesterone, suggesting that MCL-1 and STAT3 together cannot rescue the apoptotic effect of progesterone. In the presence of both estrogen and progesterone, cells transfected with MCL-1 alone showed a 5-fold increase in apoptosis, STAT3 alone showed an increase of 3.5-fold and coexpression of STAT3 and MCL-1 increased the apoptosis rate by 6-fold in comparison with cells that were left untreated. Thus, it appears that estrogen and progesterone have an additive effect on apoptosis and that expression of MCL-1 is unable to abrogate this response.

The EMT is a well-defined phenomenon that is characterized by the increased expression of certain proteins, such as N-cadherin, vimentin, MMP-2, fibronectin, SNAI1 and SNAI2, and the decreased expression of E-cadherin, cytokeratin, desmoplakin and occludin. In MCF7 cells co-transfected with STAT3 and MCL-1, we detected increased expression of E-cadherin and cytokeratin, which was most pronounced in STAT3 and MCL-1 co-transfected cells treated with estrogen and progesterone. Interestingly, a decreased expression of N-cadherin and vimentin was detected in the co-transfected cells (Fig. 7A). Progesterone, which is known to increase expression of vimentin (Lin et al., 2003; Uchida et al., 2012), has the same effect in co-transfected cells, but estrogen alone or in combination with progesterone reverses this effect. Thus, it appears that cells co-transfected with STAT3 and MCL-1 and treated with estrogen and progesterone in combination are equivalent to implantation-state cells because they exhibit a mesenchymal to epithelial transition (MET).

Functional markers of the EMT include increased invasion and resistance to apoptosis. Coexpression of STAT3 with MCL-1 in MCF7 cells resulted in increased apoptosis upon treatment with estrogen and progesterone, as shown in Fig. 6B. As a marker for cell invasion, we analyzed the expression of MMP-2 and its regulator TIMP-2. Although the level of MMP-2 was found to be increased in cells treated with estrogen and progesterone together, its levels were clearly downregulated in cells treated with progesterone alone, compared with that of control, and, concomitantly, TIMP-2 expression was high in cells co-transfected with STAT3 and MCL-1 in all hormone treatments (Fig. 7B). This can be explained by the fact that estrogen and progesterone modulate the window of implantation but embryo invasion occurs at the time when only progesterone is present and the effects of estrogen have diminished.

To confirm the effects of STAT3 and MCL-1 in promoting a MET, real-time analysis of the EMT markers was performed using cells transfected with vector (pcDNA), STAT3 or MCL-1 alone or STAT3 and MCL-1 together. The relative expression was calculated using the ΔΔCt method with 18S rRNA as the endogenous control. Estrogen-treated untransfected cells served as the calibrator. The primers used for the real-time experiments are listed in the supplementary material Table S2. The data obtained for each marker corroborated with the results obtained by western blotting. The mesenchymal marker vimentin (P<0.02) showed a significant decrease in expression compared with pcDNA-transfected control cells and N-cadherin (P<0.05) showed a significant decrease in comparison with STAT3-transfected cells (Fig. 7C). By contrast, the epithelial markers E-cadherin (P<0.03) and cytokeratin 7 (P<0.02) showed an increased expression in estrogen-treated cells that had been co-transfected with STAT3 and MCL-1, which suggests that induction of the MET occurs in these cells (Fig. 7C). The invasion markers MMP-2 (P<0.02) and MMP-9 (P<0.01) exhibited decreased expression in cells co-transfected with STAT3 and MCL-1 and treated with estrogen; by contrast, the MMP inhibitor TIMP-2 showed a corresponding increase in expression in co-transfected cells compared with the control (Fig. 7C), which points to a possible decrease in invasive capabilities of these cells.

As the real-time expression data indicated that co-transfection of MCF7 cells with STAT3 and MCL-1 might inhibit the invasive potential of these cells, we performed invasion assays using the BD BioCoat Tumor Invasion Fluoroblok cell culture insert. The data showed that, indeed, cells co-transfected with STAT3 and MCL-1 had reduced potential to invade the membrane in response to the chemoattractant FBS. Compared with cells transfected with STAT3 alone (P<0.001) and MCL-1 (P<0.0002) alone, cells transfected with both plasmids showed a decreased invasion capacity (Fig. 7D). Cells transfected with empty pEGFP-N1 and pDsRedExpress-C1 vectors served as controls.

The foremost finding of our work is the demonstration that, using peptide mass fingerprinting, MCL-1 is associated with STAT3 during embryo implantation. MCL-1, a gene first identified in a screen for differentiation-induced genes activated in the human monocyte leukemia cell line (Kozopas et al., 1993), was isolated during the early differentiation of a human embryonic carcinoma cell line, an event that serves as a model of early embryogenesis (Sano et al., 2000). Expression of MCL-1 is observed when a cell is committed to differentiate, suggesting that it might be a crucial factor at decision points determining cell fate. Increased expression of MCL-1 is often associated with cell survival, whereas decreased expression corresponds to cell death (Craig, 2002). Given the capacity of MCL-1 to promote cell survival and its presence at various junctures of proliferation and differentiation, it has been suggested that MCL-1 has a possible role in the immune response and reproduction. Previous work has shown the role of MCL-1 in oocyte survival versus atresia, fertility (affecting ovulation), embryogenesis and placental development (Sano et al., 2000). Thus, our finding that MCL-1, along with STAT3, is a key protein required for implantation points to another possible mechanism for regulation of STAT3 action.

Our results demonstrate that STAT3 can be co-precipitated with MCL-1, that the two proteins colocalize during the late pre- and peri-implantation period in uterine sections and that STAT3 can bind to MCL-1 using far-western blotting analyses. Co-transfection studies with various domains of the STAT3 proteins identified the PID and SH2 domain of STAT3 as being important for the interaction with MCL-1. It is known that STAT3 interacts with other proteins using its PID and SH2 domain (Zhang et al., 1999). STAT3-mediated expression of MCL-1 is essential for the survival of primary human macrophages that have been differentiated in vitro (Liu et al., 2003). Studies have revealed a STAT3 regulatory element in the MCL-1 promoter, and transfection with dominant-negative STAT3 diminished MCL-1 mRNA and protein levels (Isomoto et al., 2005). Chromatin immunoprecipitation analysis has demonstrated a direct binding of STAT3 to the putative STAT3-binding sequences in the MCL-1 promoter (Isomoto et al., 2005). A STAT3-binding serum-inducible element (SIE) at position −86 to −93 has been identified in the MCL-1 promoter region, and activated STAT3 can bind to the SIE-related element in the murine MCL-1 promoter, which indicates a possible role for STAT3 in regulating expression of MCL-1 (Epling-Burnette et al., 2001a). Although STAT3 has been previously shown to bind to the MCL-1 promoter, our work here is the first report of a direct interaction between the two proteins.

Steroid hormones orchestrate the process of embryo implantation and pregnancy. As shown by our data, steroid hormones (estrogen and progesterone) modulate the expression of both MCL-1 and STAT3, their co-translocation into the nucleus and the promoter activity of STAT3. Steroid hormones are known to modulate STAT3 nuclear translocation and affect its transcriptional activation potential (Wang et al., 2001). Our results support these established findings and, additionally, show that MCL-1 translocates with STAT3 under hormonal control. The distinct localization in nuclear bodies, after estrogen and progesterone treatment, has important functional implications. Nuclear bodies include the well-characterized Cajal bodies, the nucleolus, perinucleolar and perichromatin regions, and promyelocytic leukemia (PML) nuclear bodies (which have an important role in DNA replication, surveillance, and repair, as well as mRNA and rRNA synthesis and assembly). Nuclear bodies are very dynamic and mobile within the nuclear space and are regulated by cellular stress, such as heat shock, apoptosis, senescence, exposure to heavy metals, viral infection and DNA damage responses (Zimber et al., 2004). STAT3 has been reported to be present in these nuclear bodies (Herrmann et al., 2004); thus, the colocalization of STAT3 with MCL-1 in nuclear bodies will have implications in mediating STAT3-directed functions.

Our cell cycle analysis, which showed increased G2/M arrest in cells transfected with MCL-1 alone or co-transfected with STAT3 and MCL-1, is in line with the data of Jamil and colleagues, which shows that increased levels of MCL-1 induce G2/M arrest (Jamil et al., 2008). This suggests that the association of STAT3 with MCL-1 does not affect the cell cycle, in comparison with the transfection of MCL-1 alone.

Apoptosis plays an important regulatory role in mammalian embryogenesis and development. MCL-1 is an anti-apoptotic BCL-2-related gene, MCL-1 deficiency results in peri-implantation embryonic lethality due to an inability of the embryo to implant in utero, and the null blastocysts fail to develop or attach in vitro, indicating a defect in the trophectoderm; however, the inner cell mass could grow in culture. This indicates that MCL-1 is essential for preimplantation development and implantation, suggesting that it has a function beyond regulating apoptosis (Rinkenberger et al., 2000). Cooperative regulation of MCL-1 by Janus kinase, STAT and phosphatidylinositol 3-kinase contributes to delayed apoptosis in human neutrophils, which is stimulated by granulocyte-macrophage colony-stimulating factor (Epling-Burnette et al., 2001b). MCL-1 is an important anti-apoptotic factor for the survival of T cells at multiple stages of their life-cycle in vivo. MCL-1 is essential for the survival of double-negative and single-positive thymocytes, as well as naive and activated T cells. MCL-1 functions together with Bcl-xL to promote double-positive thymocyte survival (Dzhagalov et al., 2008). MCL-1 plays an important role in the development of various carcinomas (Shigemasa et al., 2002; Zhou et al., 2001). The invading peri-implantation embryo has been, rightly, compared to invasive tumor cells; yet, a notable difference is that the embryo is differentiated into the trophectoderm and the inner cell mass. The overall expression of MCL-1 can increase during the induction of cell death, as well as during the induction of differentiation (Zhan et al., 1997). The role of steroid hormones, together with the interaction between STAT3 and MCL-1, increase the understanding of the process of apoptosis of uterine cells during embryo implantation.

Our apoptosis assay using MCF7 cells demonstrates that steroid hormones, especially progesterone in combination with estrogen, can increase the rate of apoptosis in cells transfected with both STAT3 and MCL-1, and that the STAT3–MCL-1 interaction changes the function of MCL-1 from an anti-apoptotic to a pro-apoptotic protein, which correlates with the report that progesterone inhibits growth and induces apoptosis in breast cancer cells through downregulation of BCL-2 and the upregulation p53 (Formby and Wiley, 1999). Our luciferase assay data show that estrogen downregulates the activity of STAT3 on promoters, which was almost restored by MCL-1, lending further support to our hypothesis.

The passage of a cell through the EMT is marked by several discrete processes, such as activation of transcription factors, cytoskeletal remodeling and differential expression of surface proteins and ECM components. Previous reports suggest that spheroids formed from the choriocarcinoma cell line JAR, under the influence of estrogen and progesterone, induce the EMT in Ishikawa cells (Uchida et al., 2012). Our analysis of the EMT markers found in cells overexpressing STAT3 and MCL-1 demonstrate that the association of STAT3 with MCL-1 prevents the EMT. We found increased expression of E-cadherin and cytokeratin and decreased expression of N-cadherin, vimentin and MMP-2. The western blot data of the EMT markers was further confirmed with real-time gene expression analysis and invasion assays. Real-time analysis showed a substantial increase in epithelial markers and subsequent downregulation of mesenchymal markers. Invasion assays and real-time analysis demonstrated that cells co-transfected with STAT3 and MCL-1 have less invasive potential. In addition, our observation of a substantial reduction in phosphorylated STAT3 in estrogen-treated cells that had been co-transfected with STAT3 and MCL-1 enabled us to hypothesize that lower levels of phosphorylated STAT3 could have implications for the EMT-to-MET shift. Thus, the STAT3–MCL-1 interaction controls the switch from an EMT phenotype to the MET phenotype (Fig. 8). Our results are strongly supported by the recent observation that MET occurs during decidualization when fibroblast-like endometrial stromal cells are converted into polygonal epithelium (Zhang et al., 2013). The colocalization of STAT3 with MCL-1 is clearly evident in stromal cells at the post-implantation stage, which is in line with the onset of decidualization. Increased apoptosis and decreased invasion contribute to the disruption of EMT by the interaction of STAT3 with MCL-1. We hypothesize that overexpressed MCL-1 suppresses STAT3 activity by interacting with it, serving as a control mechanism for successful embryo implantation. This interaction serves as a control to prevent excessive embryo invasion, in contrast to tumors, which are characterized by rapid and extensive growth and invasion in the host.

In summary, our results lead us to postulate that STAT3 interacts with MCL-1 in the uterus during embryo implantation. MCL-1 is known to control two distinct processes, apoptosis and the cell cycle; our demonstration of an interaction with STAT3 now suggests that MCL-1 might have other, currently unknown, functions as a result of its association with STAT3. Our studies suggest that the association of MCL-1 with STAT3 causes a MET and is, thus, crucial for regulating epithelial plasticity during implantation. Further work will be essential to investigate the role of the interaction between MCL-1 and STAT3 during implantation failure, development and the progression of cancer.

Swiss strain albino mice, with a regimen of 14 hours of day and 10 hours of night, were fed with water and food ad libitum. Swiss strain male mice of proven fertility were used for mating the females. The day of the vaginal plug was designated as day 1 of pregnancy. Uterine tissue was extracted from the females, cleared of adhering fat and retrograde flushing with PBS was performed to remove the embryos. Embryos were observed by a microscope to determine the correct stage of pregnancy. All animal work was approved by the institute's Animal Ethical Committee.

Tissues, taken from different days of pregnancy, were thawed on ice and washed with PBS (10 mM). The tissues were minced and homogenized in buffer [10 mM PIPES, 100 mM KCl, 1 mM MgCl₂, 1.5 mM EGTA, 3 mM NaCl, 3.4 mg PMSF, phosphatase inhibitor cocktail, complete mini protease inhibitor cocktail (Roche)] using the Polytron homogenizer. The homogenate was then incubated on a Genei Rocker-100 on ice for 30 minutes and centrifuged at 200 g at 4°C for 5 minutes in an Eppendorf 5820R centrifuge to pellet unbroken cells. The supernatant was aspirated and centrifuged at 1000 g at 4°C for 10 minutes to separate the nuclei. The post-nuclear supernatants were further centrifuged at 12,000 g at 4°C to remove mitochondria and other large organelles. The obtained supernatant was centrifuged at 100,000 g at 4°C in a Beckman L8-M ultracentrifuge for 1 hour to separate membrane extract. The supernatant obtained (cytosolic extract) was stored at −80°C. The nuclear pellets obtained after centrifuging at 1000 g were washed with PBS (three times, 10 minutes each) and vigorously pipetted up and down. The nuclear protein fraction was extracted on a rocking bench for 30 minutes in hypertonic buffer [20 mM HEPES, 10 mM KCl, 1 mM MgCl₂, 400 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol (DTT), 1 mM sodium orthovanadate, 200 µM PMSF, 20% glycerol, 1% Triton X-100, phosphatase inhibitor cocktail, complete mini protease inhibitor cocktail (Roche)]. The homogenate was then subjected to centrifugation at 20,000 g at 4°C for 30 minutes. The supernatant collected (nuclear extract) was stored at −80°C. Protein concentrations were determined using the Bio-Rad DC Protein assay kit.

For the identification of the interacting partners, immunoprecipitation studies followed by MALDI TOF analysis were performed as described previously (Padmanabhan et al., 2011). Briefly, 200 µg of nuclear or cytosolic extracts were incubated with 1 µg of primary antibody. The extracts were then incubated with protein A agarose at room temperature for 60 minutes and then centrifuged in an Eppendorf 5820R centrifuge (Eppendorf, Hamburg, Germany) at 14,000 r.p.m. at 4°C for 30 minutes. The pellet was denatured in Laemmli buffer and the proteins were separated electrophoretically by SDS-PAGE (Laemmli, 1970). The gels were stained using Coomassie and the protein bands were subjected to trypsin digestion using a Trypsin Profile IGD Kit (Sigma) as per the manufacturer's protocol. The peptide fragments were then subjected to analysis and the results computed using Mascot (http://www.matrixscience.com/). Mock immunoprecipitation was also performed to prove the specificity of the antibodies (supplementary material Fig. S1).

Immunoprecipitated samples were separated by SDS-PAGE and transferred to a polyvinylidene fluoride (PVDF) membrane as described previously (Towbin et al., 1979). The blots were developed using the following primary antibodies: STAT3 (sc482), MCL-1 (sc819), vimentin (sc5565), E-Cadherin (sc7870), N-cadherin (sc31031), MMP-2 (sc13595), TIMP-2 (sc5539), cytokeratin 7 (sc17116), p-STAT3 (sc7993). All primary antibodies were purchased from Santa Cruz Biotechnology. Secondary labeling was performed using suitable secondary antibodies. To visualize the band, the membrane was either subjected to color development using 3,3′-diaminobenzidine tetrahydrochloride dihydrate (DAB) as an electron donor, H₂O₂ as the enzyme substrate and NiCl₂ as metal enhancer, or was visualized using an enhanced chemiluminescence detection system. Images were captured using a Bio-Rad FluorS Multi Imager. The specificity of the antibodies was monitored by performing secondary-antibody-only controls (supplementary material Fig. S1).

The far-western analysis technique was employed to identify proteins that interacted, as described previously (Kihm et al., 1998). Samples [cytosolic extract (D4 16:00, CE), BSA and purified STAT3] were resolved using SDS-PAGE on a 10% gel and were then electrophoretically transferred to a PVDF membrane. The PVDF membranes were incubated in a blocking solution (50 mM Tris-HCl, 50 mM NaCl, 1 mM EDTA, 1 mM DTT, 3% BSA, pH 7.9) for 1 hour at 4°C. Recombinant STAT3 protein (200 ng in 10 ml blocking solution) was added and the membrane was gently agitated for 2 hours at 4°C. The membrane was thoroughly washed with washing buffer (20 mM Tris-HCl, 50 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.2% Tween-20, pH 7.9) three times for 10 minutes each, and bound protein was detected by immunoblotting with STAT3. As a control, another panel of samples was probed with an antibody against MCL-1. A control using secondary antibody alone was also performed.

The endogenous peroxidase activity was quenched using 3% hydrogen peroxide in absolute methanol. Antigen unmasking was performed by autoclaving slides for 10 minutes after placing them in a coplin jar with 10 mM sodium citrate buffer (pH 6.0). After cooling, the slides were rinsed in PBS and then incubated with blocking solution (5% normal goat serum) for 1 hour. The washed slides were incubated with the primary antibody (1∶200 dilution) overnight at 4°C. After extensive washing with PBS, the sections were incubated with the fluorescein isothiocyanate (FITC)-conjugated secondary antibody in the dark for 60 minutes. The unbound secondary antibody was washed off by repeated buffer replacements. The slides were then washed and blocked and stained for the second protein, using a tetramethylrhodamine (TRITC)-conjugated secondary antibody, as described above. To detect the nuclei, the sections were further processed for staining by incubation in 1∶1000 dilution of 50 µg/ml solution of DAPI for 10 minutes. After proper washing, the sections were again dehydrated and then imaged using a Leica TCS SP2 confocal laser scanning microscope. All experiments were repeated five times. Secondary-antibody-alone controls were performed for both FITC- and TRITC-conjugated antibodies.

To identify the localization of the proteins under in vitro conditions, studies were performed using the MCF7 cell line. The cells were maintained in Dulbecco's modified Eagle's medium (DMEM) with nonessential amino acids, 1% antibiotic-antimycotic and 10% fetal bovine serum. The cells were grown to 80% confluence in the culture flask. 0.25% trypsin-EDTA was then used to detach and isolate the cells. Cells were seeded on to coverslips at a density of 10⁵ cells per well in a 24-well culture plate. The cells were supplemented with their respective media and allowed to grow. Water-soluble estrogen and progesterone were administered to cells at concentrations of 1 nM and 10 nM per well, respectively. The coverslips were taken out at 0, 30 minutes or 24 hours after the treatment with estrogen or progesterone, or a combination of the two, after 24 hours. The cells were then fixed using 4% paraformaldehyde, neutralized in 50 mM ammonium chloride and permeabilized with 0.25% Triton X-100. The cells were blocked with 5% goat serum, probed with the primary antibody (STAT3, catalogue number SC-482, Santa Cruz Biotechnology; MCL-1, catalogue number SC-819, Santa Cruz Biotechnology, 1∶200 dilution) and fluorochrome-conjugated secondary antibody (1∶200 dilution). DAPI was used to stain the nucleus. After thorough washings, the coverslips were mounted on the slides using 1∶1 glycerol∶PBS. Images were captured using a Nikon confocal microscope. Secondary-antibody-alone controls were performed for both Alexa-488- and Alexa-568-conjugated antibodies.

Total RNA was prepared from mouse uterus using Trizol (Sigma). The purity of the total RNA was established by determining the A_260/280 ratio and the concentration was then adjusted (100 ng–5 µg/µl). cDNA from total RNA was prepared using Ready-To-Go T-primed First-Strand Synthesis kit (GE Healthcare), which utilizes the moloney murine leukemia virus (M-MuLV) reverse transcriptase and an oligo(dT)18 primer. All oligonucleotides were purchased from Ocimum Biosolutions. Stat3 and Mcl-1 were amplified from mouse uterine cDNA, and the sequence showed complete similarity with the available Genbank sequences of Stat3 transcript variant 1 (accession no. NM_213659) and Mcl-1 (accession no. NM_008562.3), respectively. The product of Stat3 was cloned into pEGFP-N1 (Clontech) for co-transfection studies and pcDNA 3.1(-) (Invitrogen) for luciferase and apoptosis assays. The primers used to generate STAT3–EGFP were: STAT3_forward XhoI 5′-GGTGGTCTCGAGATGGCTCAGTGGAACCAGCTG-3′ and STAT3_reverse EcoRI 5′-GGTGGTGAATTCGCATGGGGGAGGTAGCACACT-3′, and for STAT3 pcDNA STAT3_pcDNA_forward 5′-CTCGAGGTTATGGCTCAGTGGAACCAGCTG-3′ and STAT3_pcDNA_reverse 5′-GAATTCTCACATGGGGGAGGTAGCACACT-3′ were used. Mcl-1 was cloned into pDsRedExpress-C1 (Clontech) using primers MCL-1_forward EcoRI 5′-GAATTCTATGTTTGGCCTGCGGAGAAACGCGGT-3′ and MCL-1_reverse SalI 5′-GTCGACCTATCTTATTAGATATGCCAGACCAGC-3′. MCL-1 pcDNA was generated using primers MCL-1_pcDNA_forward 5′-GAATTCACCATGGTTGGCCTGCGGAGAAAC-3′ and MCL-1_pcDNA_reverse 5′-GGATCCCTATCTTATTAGATATGCCAGACC-3′. EGFP-bound constructs of different domains of STAT3 [protein interaction domain (PID), alpha domain, DNA-binding domain (DBD) and SH2 domain] were also generated for co-transfection assays with MCL-1–DsRed (supplementary material Table S1).

The MCF7 cell line was cultured in DMEM media (Gibco-BRL) supplemented with 10% FBS (Sigma) and appropriate antibiotics. For transfection, the cells were incubated with Opti-MEM (Life Technologies) without antibiotics for 20 minutes. The cells were then transfected with plasmids encoding STAT3–EGFP and MCL-1–DsRed by using the Lipofectamine reagent (Invitrogen). The appropriate amount of plasmids and the Lipofectamine reagent were diluted separately in Opti-MEM and incubated at room temperature for 5 minutes (0.8 µg plasmid per 50 µl of Opti-MEM; 1 µl Lipofectamine per 50 µl Opti-MEM). The two mixtures were then combined and incubated for 20 minutes at room temperature.100 µl of the mix was added to each well and incubated in 5% CO₂ for 4–6 hours. Fresh medium, supplemented with FBS and antibiotics, was added after 6 hours. We also performed co-transfection assays of MCL-1–DsRed with EGFP constructs of different domains of STAT3: protein interaction domain (PID), alpha domain, DNA-binding domain (DBD) and SH2 domain. MCF7 cells transfected with empty EGFP and empty DsRed plasmids served as negative controls.

To analyze whether the activity of STAT3 on promoters is affected by a STAT3–MCL-1 interaction, luciferase assays were performed using a Cignal STAT3 reporter assay kit from SABiosciences. The cells (10⁶ per well) were seeded in a 12-well tissue culture plate the day before transfection. Cells were transfected with constructs encoding STAT3 and MCL-1, along with a STAT3-responsive luciferase construct, which contains the firefly luciferase reporter gene under the control of a minimal cytomegalovirus promoter and tandem repeats of the SIE (sis inducible element) transcriptional response element. Each sample was tested in triplicate. After transfection, the cells were incubated with 1 nM of estrogen and then lysed using the passive lysis buffer (Dual Luciferase Assay kit, Promega) 24 hours after transfection. Luciferase assays were performed with 20 µl of the cell extract and 100 µl of luciferase assay buffer. The enzyme activity was measured for 2 seconds using a TD 20/20 luminometer.

Apoptosis is a key event during the process of embryo implantation. We used a Dual Apoptosis Assay Kit with NucView™488 Caspase-3 Substrate and Sulforhodamine 101-Annexin V kit (Biotium) for measuring the degree of apoptosis created by the association of STAT3 with MCL-1 using MCF7 cells. The Dual Apoptosis Assay Kit allows simultaneous detection of caspase-3 activation and phosphatidylserine translocation in apoptotic cells. The assay was performed as per the manufacturer's instructions using the Tali™ Image-based cytometer (Invitrogen). Cells were transfected with STAT3 pcDNA, MCL-1 pcDNA or STAT3 and MCL-1 in combination. Hormone treatments [estrogen (1 nM) or progesterone (10 nM) individually, or estrogen and progesterone in combination] were performed to analyze the effect of hormones on apoptosis induced by the association of STAT3 with MCL-1.

Colocalization of STAT3 and MCL-1 in uterine sections was imaged using a Leica TCS SP2 Confocal Laser Scanning Microscope DMIRE2 using a ×63 NA 1.4 objective lens. Images were captured using a Leica camera (DC350F) and the Leica acquisition software. Co-transfection studies and endogenous localization of MCL-1 and STAT3 in MCF7 cells were observed using a Nikon Confocal Microscope with the objective lens Plan Apo VC 60× oil DIC N2, NA 1.4. Images were captured using the Nikon acquisition software.

MCF7 cells were cultured in 35-mm dishes and cells were transfected with vector alone, STAT3 or MCL-1 alone or STAT3 with MCL-1 in combination. The cells were treated with 1 nM of estrogen for 24 hours after transfection. These cells were used for flow-cytometric analysis of cell cycle progression and relative quantification of EMT markers.

For analyzing the cell cycle status, after transfection and treatment with estrogen, cells were fixed with 70% ethanol at 4°C overnight and subsequently stained with propidium iodide solution [0.1% (v/v) Triton X-100, 10 µg/ml propidium iodide (Sigma) and 100 µg/ml DNase-free RNase A in PBS]. The cells were filtered to remove clumps and the stained cells were analyzed using a FACS Aria II flow cytometer (FACSDiva v5 software, Becton Dickinson). Cells transfected with the empty pcDNA vector served as controls.

Total RNA from transfected cells were extracted using Trizol (Sigma). The RNA samples were quantified using the ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE). cDNA from 100 ng of total RNA was prepared using Superscript® VILO cDNA Synthesis Kit, and quantitative PCR amplification was performed with Power SYBR®Green PCR Master Mix (Life Technologies, CA). Oligonucleotides were designed using Primer Express version 3 and were purchased from Ocimum Biosolutions. The relative expression was calculated using the ΔΔCt method using 18S rRNA as the endogenous control. Estrogen-treated untransfected cells served as the calibrator. The run was performed on the 7900HT Fast Real-Time PCR system (Applied Biosystems, CA) under standard cycling conditions. Melting curve analysis was performed to ensure product specificity.

Invasion assays using cells that had been transfected with a vector encoding STAT3–EGFP or MCL-1–DsRed alone, co-transfected with STAT3–EGFP or MCL-1–DsRed, or the empty EGFP and DsRed vectors (control) were performed using the BD BioCoat Tumor Invasion Assay System containing the BD Falcon FluoroBlok 24-multiwell insert according to the manufacturer's instructions. The fluorescence was measured using the Infinite 200 Tecan microplate reader.

In all the statistical analyses, P values were obtained by applying a two-tailed, type 2 Student's t-test using Microsoft Excel. P<0.05 was considered significant.

We gratefully acknowledge the confocal imaging core at Rajiv Gandhi Centre for Biotechnology, the technical assistance of P. Manoj in the sequencing core and V. Jiji and K. G. Anurup in confocal imaging.