Aberrant claudin-6–adhesion signal promotes endometrial cancer progression via estrogen receptor α

Cell adhesion proteins not only maintain tissue integrity but also possess signaling abilities to organize diverse cellular events in physiological and pathological processes; however, the underlying mechanism remains obscure. Among cell adhesion molecules, the claudin (CLDN) family often possesses aberrant expression in various cancers, but the biological relevance and molecular basis have not yet been established. Here, we show that high CLDN6 expression promotes endometrial cancer progression and represents the poor prognostic marker. The second extracellular domain and Y196/200 of CLDN6 were required to recruit and activate Src-family kinases (SFKs) and to stimulate malignant phenotypes. Importantly, we demonstrate that the CLDN6/SFK/PI3K-dependent AKT and SGK (serum- and glucocorticoid-regulated kinase) signalings target Ser518 in the human estrogen receptor α and ligand-independently activate target genes in endometrial cancer cells, resulting in cancer progression. The identification of this machinery highlights regulation of the transcription factors by cell adhesion to advance cancer progression.

Kaplan-Meier plots revealed significant differences in overall survival and recurrence-free survival between 23 the two groups ( Figure 1C and Supplementary Figure S2). The five-year survival rate in the high CLDN6 24 expression group remained at approximately 30%, whereas that in the low expression group was 90%. Among 25 the clinicopathological factors, the high CLDN6 expression was significantly associated with surgical stages 6 III/IV (p<0.001), histological type (p=0.030), histological grade 3 (p=0.004), lymphovascular space involvement 1 (LVSI; p=0.005), lymph node metastasis (p=0.001) and distant metastasis (p<0.001), but not with younger age 2 (Supplementary Table S1). In addition, using the Cox multivariable analysis, stages III/IV (hazard ratio [HR] 3 10.93, p=0.002), distant metastasis (HR 4.68,p=0.006) and high CLDN6 expression (HR 3.50,p=0.014) 4 possessed independent prognostic variables for overall survival of endometrial cancer patients (Supplementary 5 8 We subsequently generated, using the lentiviral vector system, the human endometrial carcinoma cell line 9 Ishikawa expressing CLDN6 (Ishikawa:CLDN6; Figure 2A). CLDN6 was detected along the cell borders in 10 Ishikawa:CLDN6 cells, indicating that CLDN6 acted as a cell adhesion molecule ( Figure 2B). BrdU assay 11 revealed that cellular proliferation was significantly increased in Ishikawa:CLDN6 cells compared with parental 12 Ishikawa cells (Figure 2C and D). In contrast, on the TUNEL assay, few apoptotic cells were observed in both 13 cell lines (Supplementary Figure S3). Moreover, wound healing assay demonstrated that cell migration in 14 Ishikawa:CLDN6 cells was significantly accelerated compared with that in Ishikawa cells ( Figure 2E and F). 15 We then validated whether the high CLDN6 expression also promoted malignant phenotypes of human 16 endometrial carcinoma cells in vivo. Four weeks after inoculation in SCID mice, the tumor weight of 17 Ishikawa:CLDN6 xenografts was significantly increased compared with that of Ishikawa ( Figure 2G and H). 18 Neither lymph node nor distant metastasis was grossly evident in these xenografts. Microscopically,19 Ishikawa:CLDN6 xenografts were equivalent to Grade 3 endometrial carcinomas that were rich in solid 20 components ( Figure 2I). Furthermore, intratumor heterogeneity of CLDN6 expression was observed in 21 Ishikawa:CLDN6 xenograft tissues as in the high CLDN6 expression cases of endometrial cancer subjects. It is 22 also noteworthy that invasion into the fibrous capsule around the tumor was prominent in Ishikawa:CLDN6 23 xenografts but hardly in Ishikawa ones. showed that pSFK appeared to be concentrated to cell boundaries together with CLDN6 in Ishikawa:CLDN6 5 cells ( Figure 3A). When Ishikawa:CLDN6 cells were exposed to C-terminal half of CPE (C-CPE), which binds 6 to the EC2 of CLDN6 and excludes CLDN6 from cell membranes without alteration in its total protein levels 7 (15,18), the pSFK immunoreactivity was markedly reduced. On Western blot, the levels of pSFK were elevated 8 in Ishikawa:CLDN6 cells compared with Ishikawa cells, and decreased in both Ishikawa: CLDN6Y196A and 9 Ishikawa:CLDN6Y200A cells ( Figure 3B). Immunoprecipitation assay revealed that CLDN6 was associated with 10 pSFK in Ishikawa:CLDN6 cells, and the CLDN6/pSFK complex was diminished in Ishikawa:CLDN6 cells on C-11 CPE treatment as well as in Ishikawa:CLDN6Y196A and Ishikawa:CLDN6Y200A cells ( Figure 3C and D). 12 We also demonstrated that CLDN6 was highly tyrosine-phosphorylated in Ishikawa:CLDN6 cells, and the 13 phospho-tyrosine levels were suppressed by C-CPE exposure and in both Ishikawa:CLDN6Y196A and 14 Ishikawa:CLDN6Y200A cells ( Figure 3E and F). In addition, the promoted cell proliferation and migration in 15 Ishikawa:CLDN6 cells were reversed by C-CPE treatment ( Figure 2C-F). Moreover, the CLDN6-enhanced cell 16 proliferation was prevented in Ishikawa:CLDN6Y196A or Ishikawa:CLDN6Y200A cells ( Figure 3G). Taken 17 collectively, these results indicated that the CLDN6 signaling activated SFKs and accelerated endometrial cancer 18 progression in the EC2-and Y196/200-dependent manners. 19 We subsequently validated the involvement of PI3K and the two major downstream cascades AKT and SGK 20 (serum-and glucocorticoid-regulated kinase), which shares the high degree of homology and the same consensus 21 phosphorylation motif (22), in the CLDN6/SFK signaling, using the respective protein kinase inhibitors 22 LY294001, AKT inhibitor VIII and SGK1 inhibitor. The enhanced cell proliferation in Ishikawa:CLDN6 cells 23 was significantly prevented by these inhibitors and the SFK inhibitor PP2 (Supplementary Figure S4A). In 24 addition, the CLDN6-facilitated cell migration in endometrial cancer cells was reversed by these four inhibitors, 25 8 though slight but significant difference in migration between Ishikawa and Ishikawa:CLDN6 cells remained 1 upon the AKT inhibitor VIII treatment (Supplementary Figure S4B).

3
The CLDN6/SFK/PI3K-dependent AKT and SGK signalings target ERα in endometrial carcinoma cells 4 To evaluate whether the CLDN6-adhesion signaling stimulates the malignant behavior of endometrial carcinoma 5 cells via ERα, we then generated both Ishikawa:ESR1 −− and Ishikawa:CLDN6:ESR1 −− cells, and compared their 6 phenotypes. Knockout of ESR1 genes in both cell lines was confirmed by DNA sequence, Western blot and 7 immunostaining ( Figure 4A-C). In the absence of ERα CLDN6 did not alter cell proliferation or migration 8 capacity in Ishikawa cells ( Figure 4D-G). 9 We also used HEC-1A cells, in which neither CLDN6 nor ERα were expressed, and established cell lines 10 expressing either CLDN6, ERα or both together ( Figure 5A-C). Cell growth was significantly elevated in HEC-11 1A:ESR1:CLDN6 cells but not in HEC-1A:CLDN6 or HEC-1A:ESR1 cells compared with parental HEC-1A 12 cells, ( Figure 5D). Cell migration was also significantly increased in HEC-1A:ESR1:CLDN6 cells compared 13 with HEC-1A and HEC-1A:ESR1 cells, and was raised in HEC-1A:CLDN6 cells less efficiently than in HEC-14 1A:ESR1:CLDN6 cells ( Figure 5E). Taken together, these results strongly suggested that the CLDN6-adhesion 15 signaling links to ERα in endometrial cancer cells. In addition, exposure of HEC-1A:ESR1:CLDN6 cells to C-16 CPE reversed the increase in cell proliferation and migration ( Figure 5D and E), again indicating the critical role 17 of the EC2 in the CLDN6 signaling. 18 Notably, AKT and SGK1 were associated with transiently introduced ERα but not with ERαC, in 19 Ishikawa:ESR1 −− cells ( Figure 6A and B), indicating that both kinases target either the LBD/AF2 or F region of 20 ERα We next determined whether the CLDN6 signaling directed to ERαS518 and ligand (estradiol)-21 independently stimulated the ERα activity in endometrial cancer cells, as in . To this end, we 22 generated Ishikawa:CLDN6:ESR1 −− :ESR1-wt (wild-type) and Ishikawa:CLDN6:ESR1 −− :ESR1S518A cells, in the 23 latter of which ERαS518 was substituted for an alanine residue, and cells were grown in phenol red-free medium 24 with charcoal-treated FBS to exclude fat-soluble ligands. As expected, the transcript levels of the four ER target 9 genes (BCL2, CCND1, MYC, and VEGFA; 23) were significantly higher in Ishikawa:CLDN6 cells than in 1 Ishikawa cells ( Figure 6C). More importantly, the expression levels of these target genes were significantly  Figure 6E). Hence, the CLDN6-adhesion signaling directed to ERαS518 for promoting the ERα activity and 8 malignant phenotypes in endometrial cancer cells.

10
The CLDN6 signaling ERα-dependently and independently modulates gene expression in endometrial 11 carcinoma cells 12 To identify downstream molecules that expression is altered by the CLDN6 signaling, we next compared, 13 using RNA sequencing, the transcriptome in Ishikawa:CLDN6 cells with that in Ishikawa cells ( Figure 7A). 14 Among the CLDN6-activated genes, the gene products associated with malignant phenotypes, including 15 ADAMTS18 (a disintegrin and metalloproteinase with thrombospondin motifs; 24) and the transmembrane 16 receptor-associated tyrosine kinase AXL (25), as well as the soluble factors CTGF (Connective tissue growth 17 factor; 26)，CXCL1 (C-X-C motif ligand 1; 27)，FGFBP1 (Fibroblast growth factor binding protein 1; 28)， 18 NRG1 (Neuregulin 1; 29)，NTN4 (Netrin 4; 30) and TGFB2 (Tumor growth factor beta 2; 31), were detected. 19 We then by semi-quantitative RT-PCR clarified the expression of these eight genes in Ishikawa, 20 Ishikawa:CLDN6, Ishikawa:ESR1 −− and Ishikawa:CLDN6:ESR1 −− cells ( Figure 7B). CLDN6 appeared to 21 induce the expression of ADAMTS18, AXL, CTGF, NRG1, NTN4 and TGFB2 transcripts via ERα. By contrast, 22 CLDN6 activated the mRNA expression of CXCL1 and FGFBP1 in an ERα-independent manner. Thus, the 23 CLDN6-activated genes can be classified into at least two groups with distinct ERα-dependence. In the present study, we demonstrated that high CLDN6 expression in endometrial cancer tissues, in which the 2 strong and moderate signal intensity on cell membranes was observed at greater than 30% and 50%, 3 respectively, was significantly related to several clinicopathological features such as surgical stages III/IV, 4 histological type, histological grade 3, LVSI, lymph node metastasis and distant metastasis. Importantly, the high 5 CLDN6 expression represented an independent prognostic factor (HR 3.50), and the five-year survival rate was 6 about 30%, which was one third of that in the low expression group. Thus, the aberrant CLDN6 expression 7 appeared to corelate with poor outcome in patients with endometrial cancer. Taken together with the finding that 8 CLDN6 is barely expressed in normal adult cells as described above, the established anti-human CLDN6 mAbs 9 would provide powerful tools that selectively recognize CLDN6 protein in a range of cancer tissues. Extremely 10 high 11 CLDNs comprise a gene family as described above, and some anti-CLDN Abs are known to react not only 12 with the corresponding CLDN but also with other CLDN subtypes (32). Therefore, it is of particular importance 13 to verify the specificity of the anti-CLDN Abs used. Along this line, we previously established the anti-CLDN 14 pAbs that selectively recognize CLDN1, CLDN5, CLDN6, CLDN7, CLDN12 or CLDN15 as far as we 15 determined (33-35). The anti-CLDN6 pAb is one of the most reliable anti-CLDN6 Abs, and is used for 16 immunohistochemical staining of formalin-fixed paraffin-embedded human tissues (17,35,36). However, we 17 noticed in the present work that it also reacted with highly expressed CLDN4 and CLDN5 less efficiently than 18 CLDN6, reinforcing the importance of validating the selectivity of each anti-CLDN Ab. 19 We also showed that CLDN6 accelerated endometrial cancer progression in vitro and in vivo. This was 20 obvious because introduction of the human CLDN6 gene was enough to promote cell proliferation and migration Another finding of the present study is that the EC2 and Y196/200 of CLDN6 are responsible for recruiting 1 and activating SFKs in endometrial cancer cells, as well as promoting the malignant properties. This conclusion 2 was drawn from the following results: 1) the pSFK levels were increased in Ishikawa:CLDN6 cells but not in 3 Ishikawa:CLDN6Y196A or Ishikawa:CLDN6Y200A cells; 2) colocalization of CLDN6 and pSFK along cell 4 boundaries was evident in Ishikawa:CLDN6 cells, and diminished by C-CPE treatment; 3) a CLDN6-pSFK 5 complex was formed in Ishikawa:CLDN6 cells, and their association was decreased upon C-CPE exposure and 6 in Ishikawa:CLDN6Y196A and Ishikawa:CLDN6Y200A cells; 4) the increased cell growth and migration in both 7 Ishikawa:CLDN6 and HEC-1A:ESR1:CLDN6 cells were abrogated upon C-CPE treatment; 5) the CLDN6-8 stimulated cell proliferation was not detected in Ishikawa:CLDN6Y196A or Ishikawa:CLDN6Y200A cells. We 9 also demonstrated that SFKs in turn phosphorylated CLDN6 at both Y196 and Y200, and tyrosine-10 phosphorylation of CLDN6 was governed by the EC2 domain. We previously reported that similar reciprocal 11 regulation between CLDN6 and SFKs is also observed in mouse F9 stem cells (18), further strengthening our 12 conclusion. Moreover, using the respective protein kinase inhibitors, we revealed that the PI3K-dependent AKT 13 and SGK cascades contributed to the CLDN6/SFK signaling in endometrial cancer progression. 14 The most important conclusion of the present work is that the CLDN6/SFK/PI3K-dependent AKT and SGK 15 signalings target ERα in endometrial cancer cells. This was apparent because CLDN6-accerelated cell growth 16 and migration were hindered in Ishikawa:CLDN6:ESR1 −− cells. Using HEC-1A expressing CLDN6 and/or ERα, 17 it was confirmed that the CLDN6 signaling in endometrial cancer advancement was mediated via ERα. 12 should be determined not only in endometrial cancer tissues, but also in other hormone-dependent tumors, such 1 as ovarian cancer and breast cancer, in future experiments. 2 Our RNAseq analysis revealed that a variety of gene expression, including the SGK1 gene, was altered 3 between Ishikawa and Ishikawa:CLDN6 cells. Among eight representative gene products associated with tumor 4 progression in various cancers, the ADAMTS18, AXL, CTGF, NRG1, NTN4 and TGFB2 genes were activated by 5 the ERα-dependent CLDN6 signaling. By contrast, the expression of CXCL1 and FGFBP1 transcripts was 6 induced by CLDN6 in an ERα-independent manner. Interestingly, the novel AKT/SGK-consensus 7 phosphorylation motif is conserved in 14 of 48 members of human nuclear receptors (18). Taken together, 8 CLDN6 may also target these nuclear receptors and possibly other transcription factors in order to regulate the 9 expression of certain genes. Importantly, we previously reported that the CLDN6 signal targets RAR in mouse 10 F9 stem cells to initiate epithelial differentiation (18). 11 Genomic and non-genomic heterogeneity among distinct cell populations within cancers is known to 12 influence tumour behaviors (41). Our immunohistochemical study revealed intratumor heterogeneity on the 13 CLDN6 expression within human endometrial cancer and Ishikawa:CLDN6 xenograft tissues. These tumors 14 were composed of CLDN6-positive and negative subpopulations, even in endometrial cancer tissues with high 15 CLDN6 expression. Hence, the expression of CLDN6 should be carefully evaluated when small biopsy 16 specimens and tissue arrays were subjected to immunohistochemistry. Of note, since the gene expression of 17 various diffusive factors was induced in Ishikawa:CLDN6 as described above, non-cell-autonomous paracrine 18 effects between CLDN6-positive and negative cancer cells may also contribute to the enhancement of tumor 19 progression. 20 In summary, we here established that high expression of CLDN6 protein in endometrial cancer leads to more 21 aggressive tumors and predicts poor prognosis. We also demonstrated that the CLDN6/SFK/PI3K-dependent 22 AKT and SGK cascades direct to S518 in human ERα and stimulated its activity, resulting in progression of 23 tumor behaviors in endometrial cancer. Therefore, in addition to the PI3K/AKT pathway, which is frequently 24 altered in endometrial cancers (42-45), the CLDN6/SFK, SGK and ERαS518 may be promising therapeutic 13 targets for endometrial cancer. It would also be interesting to determine whether a similar link between cell 1 adhesion and nuclear receptor signaling regulates tumor progression in various types of cancers. The antibodies used in this study are listed in Supplementary Table S3. A rabbit pAb against CLDN6 was 4 generated in cooperation with Immuno-Biological Laboratories as described previously (21). 5 Rat mAbs against CLDN6 were established using the iliac lymph node method (20). In brief, a polypeptide, 6 (C)SRGPSEYPTKNYV corresponding to the cytoplasmic domain of CLDN6, was coupled via the cysteine to 7 Imject TM Maleimide-Activated mcKLH (Thermo Fisher SCIENTIFIC). The conjugated peptide was 8 subcutaneously injected with Imject TM Freund's Complete Adjuvant (Themo Fisher SCIENTIFIC) into the 9 footpads of anesthetized eight-week-old female rats. The animals were sacrificed 14 days after immunization, 10 and the median iliac lymph nodes were collected, followed by extraction of lymphocytes by mincing. Extracted 11 lymphocytes were fused to a SP2 mouse myeloma cell line by polyethylene glycol. Hybridoma clones were 12 maintained in GIT medium (Wako) with supplementation of 10% BM-Condimed (Sigma-Aldrich). The For immunostaining, uterine endometrial cancer tissues were obtained, and the 10% formalin-fixed and 28 paraffin-embedded tissue blocks were sliced into 5-μm-thick sections, then deparaffinized with xylene and 29 rehydrated using a graduated series of ethanol. The sections were immersed in 0.3% hydrogen peroxide in 30 methanol for 20 min at room temperature to block endogenous peroxidase activity. Antigen retrieval was 31 performed by incubating the sections in boiling citric acid buffer (pH 6.0) for antigen retrieval in a microwave. 32 After blocking with 5% skimmed milk at room temperature for 30 min, the sections were incubated overnight at 1 Kit for rat (VECTOR LABORATORIES) was used for 3',3'-diaminobenzidine (DAB) staining. 2 Immunostaining results were interpreted by three independent pathologists and one gynecologist using a 3 semi-quantitative scoring system, immunoreactivity score (IRS; 46). The immunostaining reactions were 4 evaluated according to signal intensity (SI: 0, no stain; 1, weak; 2, moderate; 3, strong) and percentage of 5 positive cells (PP: 0, <1%; 1, 1 to 10%; 2, 11 to 30%; 3, 31 to 50%; and 4, >50%). The SI and PP were then 6 multiplied to generate the IRS for each case. We divided the samples into two groups based on the results of the 7 immunostaining in the tissues: low expression (IRS<8) and high expression (IRS≥8) (Supplementary Table S5). Immunoprecipitation and immunoblot 13 Immunoprecipitation was performed using an Immunoprecipitation kit (Protein G, Sigma), following the 14 manufacturer's protocol. Immunoblot analysis was performed as previously described (51) Cells were grown on coverslips coated by Cellmatrix Type I-A (Nitta gelatin). The samples were fixed in 1% 29 paraformaldehyde and 0.1% Triton-X for 10 min at room temperature. After washing with PBS, they were 30 preincubated in PBS containing 5% skimmed milk. They were subsequently incubated overnight at 4ºC with 31 primary antibodies in PBS, then rinsed again with PBS, followed by a reaction for 1 h at room temperature with 32 appropriate secondary antibodies. All samples were examined using a laser-scanning confocal microscope 1 Rasband National Institutes of Health).

3
Cell proliferation, migration, and apoptosis assays 4 Cell proliferation index was evaluated by incorporation of bromodeoxy uridine (5-Bromo-2-DeoxyUridine, 5 BrdU, sigma). Cells were exposed to BrdU for 5 min after passage. The specimens were fixed with 4% 6 paraformaldehyde and 0.1% Triton-X, followed by immunostaining with anti-BrdU antibody (BD) and its 7 standard protocol. 8 For evaluating cell migration, wound areas were generated by scratching with disposable 1,000 µl pippette 9 tips 24-48 h after passage. Culture media were changed daily. Photographs of the wound areas were taken at the 10 same locations, using a phase-contrast microscope. Wound healing was calculated as the percentage of the 11 remaining cell-free area compared with the initial wound area using ImageJ software.  21 We used the chi-squared test to evaluate the relationship between CLDN6 expression and various 22 clinicopathological parameters (age, stage, histological type, histological grade, LVSI, lymph node metastasis, 23 distant metastasis, 5-year OS, and 5-year RFS). Survival analysis was performed using the Kaplan-Meier 24 method, and differences between the groups were analyzed using the log-rank test. The Cox regression 25 multivariate model was used to detect the independent predictors of survival. Two-tailed P-values less than 0.05 26 were considered to indicate a statistically significant result. All statistical analyses were performed using SPSS 27 software version 23.0 (IBM). 28 The PCR values are presented as the mean ± SD from three samples. Original values were quantified by 29 ImageJ software (Wayne Rasband National Institutes of Health). The expression levels of the target genes in RT-30 PCR were divided by the corresponding GAPDH signal intensity. Their relative levels were analyzed by paired 31 sample two-tailed t-test to evaluate statistical significance. Relative pSFK levels Relative CLDN6 levels Relative pSFK levels Relative pSFK levels    Relative gene expression   Figure.

Proliferation index
CSRGPSEYPT