Cooperation of Sox4 with β-catenin/p300 complex in transcriptional regulation of the Slug gene during divergent sarcomatous differentiation in uterine carcinosarcoma

The pGL3B-Slug luc constructs, including −2125/−235 bp, −1859/−235 bp, −1587/−235 bp, and −813/−235 bp fragments, pcDNA3.1-HA-β-cateninΔS45, pcDNA3.1-Sox4, pcDNA3.1-Sox7, pcDNA3.1-Sox9, pcDNA3.1-HA-Slug, PCI-Flag-p300, pcDNA3.1-TCF4ΔN30 (dominant-negative form of TCF4), pG5 luc, and pM-β-cateninΔS45 were used as described previously [18‐21]. pM-Sox4 was constructed by inserting the Sox4 cDNA fragment into the pM DNA-BD vector (BD Biosciences Clontech, Worcester, MA, USA). Site-directed mutagenesis of putative Sox4 binding sites in the Slug promoter was performed using the PrimeSTAR Mutagenesis Basal kit (Takara Bio, Shiga, Japan).

The Em Ca cell lines, Ishikawa, Hec251, and Hec6 cells, were maintained in Eagle’s MEM with 10 % bovine calf serum. To establish cells stably overexpressing HA-Slug, the expression plasmids or empty vectors were transfected into Hec6 cells, and stable clones were established as described previously [20].

Anti-β-catenin and anti-p27^kip1 antibodies were purchased from BD Biosciences (San Jose, CA, USA). Anti-Sox4, anti-Sox6, anti-Sox7, anti-Sox9, anti-Sox11, and β-actin antibodies were obtained from Sigma-Aldrich Chemicals (St. Louis, MO, USA). Anti-Snail and anti-Slug antibodies were from Cell Signaling (Danvers, MA, USA). Anti-p21^waf1, anti-cyclin D1, and anti-CD44s antibodies were purchased from Dako (Copenhagen, Denmark). Anti-Sox2 and anti-cyclin A antibodies were from Abcam (Cambridge, MA, USA) and Novocastra (Newcastle, UK), respectively. Anti-HA and anti-E-cadherin antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and Takara (Shiga, Japan) respectively. Anti-CD133 antibody was from Miltenyi Biotechnology (Bergisch Gladbach, Germany).

STK2, which is a serum-free culture medium for mesenchymal stem cells, [22] was obtained from DS Pharma Biomedical (Osaka, Japan).

Transfection was carried out using LipofectAMINE PLUS (Invitrogen, Carlsbad, CA, USA) in duplicate or triplicate as described previously [18‐21]. Luciferase activity was assayed as described previously [18‐21].

cDNA was synthesized from 2 μg of total RNA. For quantitative analysis, real-time RT-PCR was carried out using a Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) with specific primers (Table 1). Fluorescent signals were detected using the ABI 7500 real-time PCR System, and data were analyzed using the associated ABI 7500 System SDS Software (Applied Biosystems). Primers for the GAPDH gene were also applied, as described previously [18‐21].

Total cellular proteins were isolated using RIPA buffer [20 mM Tris–HCl (pH7.2), 1 % Nonidet P-40, 0.5 % sodium deoxycholate, 0.1 % sodium dodecyl sulfate]. Aliquots of the proteins were resolved by SDS-PAGE, transferred to membranes, and probed with primary antibodies, coupled with the ECL detection system (Amersham Pharmacia Biotechnology, Tokyo, Japan).

For immunoprecipitation, cells cultured in STK2 were lysed with TNE buffer [10 mM Tris–HCl (pH7.6), 150 mM NaCl, 1 % NP-40, 1 mM EDTA]. Cell lysates were cleared and incubated with anti-Sox4 antibody, followed by incubation with Protein G-Sepharose (Amersham Pharmacia Biotechnology). Western blot assay was subsequently performed with anti-β-catenin and anti-Sox4 antibodies.

Cells were fixed using 70 % alcohol and stained with propidium iodide (Sigma) for cell cycle analysis. ALDH 1 enzyme activity in viable cells was determined using a fluorogenic dye based Aldefluor assay (Stem Cell Technologies, Grenoble, France) according to the manufacturer’s instructions. The prepared cells were analyzed by flow cytometry using BD FACS Calibur (BD Biosciences) and CellQuest Pro software version 3.3 (BD Biosciences).

Cells (x10³) were plated in low cell binding plates (Thermo Fisher Scientific, Yokohama, Japan) in STK2 or Eagle’s MEM with 10 % bovine calf serum. Uniform spheroids of at least 50 μm in size were counted approximately 2 weeks after plating.

ChIP analysis was performed using an EpiXplore ChIP assay kit (Clontech Laboratory, Mountain View, CA, USA). Briefly, after culture in STK2 for 1 week, cells were cross-linked with formaldehyde. Cell lysates were sonicated to shear DNA to lengths between 200 and 1000 bp, and then precipitated overnight using anti-Sox4 antibody or rabbit IgG as negative control, along with magnetic beads. After proteinase K digestion, DNA was extracted and analyzed by PCR. ChIP analysis was conducted with a reduction in the number of cycles from 30 to 25, using four specific primer sets.

We reviewed cases of comprehensively staged high-grade endometrial adenocarcinomas from the patient records of Kitasato University Hospital in the period from 1997 to 2012. According to the criteria of the 2014 World Health Organization classification, [23] tumors were designated as UCS if they had evidence of both malignant epithelial (endometrioid, serous, or clear cell) components and mesenchymal (homologous or heterologous) elements. Endometrioid adenocarcinomas with spindle elements and hyalinized stroma were specifically excluded. Finally, a total of 32 UCSs were investigated. Of these, 9 cases had endometrioid and 23 cases had non-endometrioid epithelial components, while 25 and 8 cases showed homologous and heterologous mesenchymal elements, respectively. All tissues were routinely fixed in 10 % formalin and processed for embedding in paraffin wax. Approval for this study was given by the Ethics Committee of the Kitasato University School of Medicine (B14-35). Signed informed consent forms were not required from the participants due to the retrospective approach of the study, which did not impact on their treatment.

IHC was performed using a combination of the microwave-oven heating and polymer immunocomplex (Envision, Dako) methods, as described previously [18‐21]. The immunoreactions were visualized with DAB (3,3′ diaminobenzidine), and the nuclei were counterstained with methyl green.

For evaluation of IHC findings, scoring of nuclear immunoreactivity was performed, on the basis of the percentage of immunopositive cells and the immunointensity, with multiplication of values of the two parameters, as described previously [18‐21].

Comparative data were analyzed using the Mann–Whitney U-test, and the Spearman’s correlation coefficient. The cutoff for statistical significance was set as p < 0.05.

To induce EMT in Em Ca cells, the three cell lines, including Ishikawa, Hec251, and Hec6 cells, were cultured in STK2, a serum-free medium for mesenchymal stem cells [22]. As shown in Fig. 1a, cells cultured in STK2 demonstrated a dramatically altered morphology toward a fibroblast-like appearance after 73 h, along with decreased E-cadherin and increased Slug expression at both mRNA and protein levels (Fig. 1b and c). In contrast, changes in Snail expression at the protein level (Fig. 1b), as well as the mRNA level, were relatively weak (data not shown).

Next, we examined whether Sox factors are directly involved in regulation of the EMT process observed in cells cultured in STK2, since some molecules are involved in the promotion of EMT [15]. The Ishikawa, Hec251, and Hec6 cells cultured in STK2 showed increased expression of Sox4, Sox7, and Sox9, but not Sox6 and Sox11, at both protein and mRNA levels (Fig. 1d and e). In addition, the Sox4 promoter activity was increased by 20–50 folds following transfection of Sox7, while changes in the promoter activity of both Sox7 and Sox9 in response to other Sox factors were relatively minor (Fig. 1f and Additional file 1: Figure S1A). These findings suggested that culturing Em Ca cells in STK2 was sufficient to induce EMT, along with downregulation of E-cadherin and upregulation of Slug. In addition, upregulation of some Sox factors through formation of complex transcriptional regulatory loops occurs during the EMT process in Em Ca cells.

To examine whether EMT is linked to CSC properties, Hec6 cells were selected since they showed higher Slug expression as compared to those in Ishikawa and Hecc251 cells after culturing in STK2 (Fig. 1b). Cultured Hec6 cells had a low cell proliferation rate, particularly in the exponential growth phase, which correlated with increased p21^waf1 but not p27^Kip1 expression, and a decreased S-fraction during cell cycle progression (Fig. 2a). The inhibitory effects were also observed in Ishikawa and Hec251 cells (data not shown). Aldefluor assay revealed an increase in the ALDH1^high activity population (Fig. 2b), in line with the significantly increased number of well-defined, round spheroids that were over 50 um in size (Fig. 2c). In addition, expression of other CSC markers, including CD44s and Sox2, but not CD133, was also increased in Hec6 cells as well as Ishikawa and Hec251 cells cultured in STK2 (Additional file 1: Figure S1B). These findings indicated that the Em Ca cells cultured in STK2 exhibited EMT/CSC properties.

To examine whether Slug is directly linked to induction of EMT/CSC properties, two independent Hec6 cell lines stably overexpressing Slug (H6SL#8 and #21) were established. The stable cells underwent a dramatic change in morphology to fibroblast-like mesenchymal features with decreased E-cadherin expression, independent of Snail status (Fig. 2d). In H6SL#8 cells, the proliferative activity was extremely low, along with an inhibition of the S- to G2/M-phase during cell cycle progression (Fig. 2e). In addition, the ALDH1^high activity population was increased in the stable cells (Fig. 2f). However, changes in expression of Sox4, Sox7, and Sox9 were relatively minor in H6SL#8 cells as compared to the mock (Additional file 1: Figure S1C). These findings suggested that exogenous overexpression of Slug is sufficient in induction of EMT/CSC properties in Em Ca cells, independent of Sox factors.

To examine whether Sox factors are directly involved in transcription of the Slug gene, the three Sox factors, whose expression was differentially regulated by culturing in STK2, were transfected into three Em Ca cell lines. Transient transfection of Sox4 resulted in increased activity of the Slug promoter, along with increased mRNA levels, which were further enhanced by cotransfection of Sox7 or Sox9. In contrast, such effects were relatively minor when only Sox7 and/or Sox9 were transfected (Fig. 3a and Additional file 2: Figure S2A and B).

Analysis of an approximately 2000 bp fragment upstream of the translation start site in the Slug gene (AF084243) revealed four potential Sox4-binding elements (BE) (A/T A/T CAA A/T G) (Fig. 3b). Using a series of 5′-truncated promoter constructs, we found that deletion from −2125 to −1587 bp had little effect on induction of the promoter activity by Sox4, as well as Sox7 and Sox9, whereas the -813/-235 bp deletion appeared to have prevented binding of the Sox factors and reduced the promoter activity to a very low level (Fig. 3c and Additional file 2: Figure S2C). This indicated that the region involved in the response to Sox4 is present between −2125 to −813 bp. Additional promoter constructs carrying four nucleotide alterations in all of the putative Sox4-binding sites, including Sox4-1, Sox4-2, Sox4-3, and Sox4-4 (Fig. 3b), resulted in considerable reduction of response to Sox4 (Fig. 3d). ChIP assays revealed that increased amount of Sox4 by STK2 culture caused its recruitment to these four Sox4-BEs within the promoter (Fig. 3e). These findings suggest that Slug is a target gene of Sox4.

Since it has been reported that β-catenin is a positive regulator of Slug expression, [18] we examined for an association between Sox4 and β-catenin in regulation of the Slug promoter. The promoter activity was increased by a combination of β-catenin and Sox4, and the effect was further enhanced by cotransfection of the multifunctional coactivator p300, but not by dominant-negative TCF4 (Fig. 4a and Additional file 3: Figure S3A). Cotransfection of GFP-Sox4, HA-β-cateninΔS45, and Flag-p300 resulted in formation of several enlarged dots in the nuclei (Fig. 4b), whereas such nuclear aggregates were not observed after cotransfection of GFP-Sox4 and HA-β-cateninΔS45 alone (Additional file 3: Figure S3B). One-hybrid assays revealed that the pG5 luc reporter activity was raised by 3–12 folds following cotransfection of either DNA-BD-fused full-length β-catenin or Sox4 fragment (pM-β-cat or pM-Sox4) and p300, but such effects were relatively minor with cotransfection of only pM-β-catenin and Sox4, Sox7 or Sox9 (Fig. 4c and Additional file 3: Figure S3C and D). Finally, coimmunoprecipitation using lysates of cells cultured in STK2 revealed a weak interaction between β-catenin and Sox4 (Fig. 4d). These findings suggested that Sox4 cooperates with β-catenin/p300 complexes in the transcriptional regulation of the Slug gene.

Representative images of IHC findings for Slug, β-catenin, and Sox factors are illustrated in Fig. 5a. Distinct nuclear immunostaining for Slug, Sox4, Sox7, and Sox9, and nuclear and cytoplasmic/membranous immunoreaction for β-catenin were observed in both the carcinomatous and sarcomatous components of UCSs. Average nuclear β-catenin and Slug scores were significantly higher in the sarcomatous elements than those in the carcinomatous components, in contrast to no significant differences in the scores of Sox4, Sox7, and Sox9 (Fig. 5b). The average Slug score was positively correlated with nuclear β-catenin score and combinations of the three Sox factors, including Sox4/Sox9, Sox7/Sox9, and Sox4/Sox7/Sox9. The nuclear β-catenin score also showed a positive correlation with combinations of the three Sox factors (Table 2).