The stabilization of yes‐associated protein by TGFβ‐activated kinase 1 regulates the self‐renewal and oncogenesis of gastric cancer stem cells

Abstract Gastric cancer (GC) is the most frequent digestive system malignant tumour and the second most common cause of cancer death globally. Cancer stem cell (CSC) is a small percentage of cancer cells in solid tumours that have differentiation, self‐renewal and tumorigenic capabilities. They have an active participation in the initiation, development, metastasis, recurrence and resistance of tumours to chemotherapy and radiotherapy. Gastric cancer stem cells (GCSCs) have been shown to be correlated with GC initiation and metastasis. In this study, we found that TAK1 expression level in GC tissues was significantly increased compared to the adjacent non‐cancerous tissues by RT‐qPCR, Western blot and immunohistochemistry. TAK1 has been identified as a critical molecule that promoted a variety of malignant GC phenotypes both in vivo and in vitro and promoted the self‐renewal of GCSCs. Mechanistically, TAK1 was up‐regulated by IL‐6 and prevented the degradation of yes‐associated protein (YAP) in the cytoplasm by binding to YAP. Thus, TAK1 promoted the SOX2 and SOX9 transcription and the self‐renewal and oncogenesis of GCSCs. Our findings provide insights into the mechanism of self‐renewal and tumorigenesis of TAK1 in GCSCs and have broad implications for clinical therapies.


| INTRODUC TI ON
Gastric cancer (GC) is the most frequent malignant neoplasm of the human digestive system. In 2018, it had the sixth highest incidence rate among all types of cancer worldwide and the second highest mortality rate. 1 Currently, surgical resection is the only possible way to cure GC.
However, most patients are already in late stage at the time of first diagnosed or during treatment. Besides, unsatisfactory surgical results can cause post-operative recurrence and metastasis. 2 In addition, GC patients are prone to show chemotherapy resistance after surgery. 2 Although cancer stem cells (CSCs) represent only a short tumour cell fraction, they are closely associated with tumour occurrence, recurrence, metastasis and chemotherapy resistance. 3 CSC occurrence have been confirmed in several tumour types, including gastric cancer stem cells (GCSCs). [4][5][6][7] GCSCs can be characterized by several biomarkers, such as CD44, 7 Lgr5, 8 CD133 9 and CD90. 10 Furthermore, GCSCs have been considered a relevant subset of targets for an efficient GC treatment. However, the GCSCs mechanism of action in GC has not yet been clarified.
TGFβ-activated kinase 1 (TAK1) is a mitogen-activated protein kinase kinase that acts on nuclear factor κB (NF-κB) and activator protein-1 (AP-1) activation. TAK1 plays essential roles in several biological responses, such as cell survival, development, metabolism, carcinogenesis, immune responses and chemoresistance. 11,12 Several studies have shown that TAK1 exerts a key role in tumour initiation, progression and metastasis and can behave as a tumour promoter [13][14][15][16] or suppressor. 17,18 Current evidence indicates that TAK1 performs a fundamental function in the stem cell regulation. 19,20 Furthermore, TAK1 has been described to be highly expressed in GC tissues and that may be associated to the GC occurrence and development. 21 Nevertheless, the detailed molecular mechanism by which TAK1 acts in GC remains elusive.
Recent advances highlight the Hippo pathway role in the tissue regeneration, organ development, stem cell self-renewal and tumorigenesis. 22,23 The main Hippo kinase cascade function is to supress the oncogenic transcriptional complex formed by the yes-associated protein (YAP), the transcriptional co-activator with PDZ-binding motif (TAZ) and TEA domain family members (TEAD). YAP/TAZ act as transcriptional activators that translocate between the nucleus and the cytoplasm and lead to expression of target genes by interaction with transcription factors of the TEAD family. Different studies have indicated that YAP and TAZ perform crucial roles in maintaining CSCs and cancer progression. [24][25][26][27] In this study, TAK1 has been identified as a tumour promoter positively correlated with poor prognosis and recurrence in GC. We demonstrated that TAK1 is up-regulated by IL-6 in the cell microenvironment and binds to YAP, thereby promoting self-renewal by regulating SOX2 and SOX9 in GCSCs.

| Clinical tissue samples
Gastric cancer (GC) and relative adjacent non-cancerous samples and detailed clinical and pathological data (including gender, age, drink, tumour size, location and differentiation, and T, N and TNM stages) were acquired from the clinical records of 200 patients diag-  100 U/mL penicillin-streptomycin (Gibco; Thermo Fisher Scientific Inc) in an atmosphere containing 5% CO 2 at 37°C. They were authenticated by STR profiling and tested to be pathogen and mycoplasma negative before the experiments (BioWing Biotechnology).

| Cell culture and cell lines construction
Sh1-TAK1, sh2-TAK1, shYAP, sh-NC, 3 × Flag-Vector, 3 × Flag-TAK1 and 3 × Flag-YAP were synthesized by GenePharma Co. Ltd. The corresponding nucleotide sequences are listed in Table S1. Knockdown and overexpression stable cell lines in the above-mentioned were established through lentiviral transduction. The established stable cell lines were attested by quantitative reverse transcription-PCR (RT-qPCR) and Western blot analyses and used for subsequent experiments.

| RNA Extraction and Quantitative real-time PCR (qPCR)
TRIzol reagent was used to remove total RNA from cells according to the manufacturer's instructions (Invitrogen; Thermo Fisher Scientific Inc). The first-strand cDNA was synthesized using a cDNA Reverse Transcription Kit (Applied Biosystems; Thermo Fisher Scientific Inc).
QPCR was conducted employing the SYBR Green PCR Master Mix (Applied Biosystems; Thermo Fisher Scientific Inc.) to detect the transcription levels. Expression levels were calculated in relation to the GAPDH control. The primer pairs used in the SYBR Green reactions are listed in Table S2. The expression relative levels were determined using the 2 −ΔΔCq method.

| Western blot (WB) analysis
For the protein expression level detection, WB was conducted.
In brief, the cells were lysed using RIPA cell lysis buffer (cat. no. P0013B; Beyotime) containing 1 mmol/L phenylmethylsulphonyl fluoride at 4°C for 30 minutes, with vortexing each 10 minutes, followed by centrifugation at 13 800 g for 10 minutes at 4°C. The protein concentration was quantified using a BCA kit (cat. no. P0009; Beyotime). Twenty micrograms of denatured protein for every sample were submitted to a 10% SDS-PAGE gel electrophoresis and transferred to a PVDF membrane. Non-fat milk (5%) in TBS buffer containing 0.05% Tween 20 (TBST) was used for membrane blocking for 2 hours at room temperature. Subsequently, the membranes were incubated with the following primary antibodies:

| Immunohistochemistry staining
Immunohistochemistry (IHC) staining was conducted using a standard methodology based on streptavidin-biotin-peroxidase complex according to the manufacturer's instructions (SA2010; Boster Biological Technology). A moist chamber was used to incubate the tissue sections overnight at 4°C in the presence of anti-TAK1 antibody (cat. no. ab109526, 1:200, Abcam). TAK1 expression levels were determined by establishing the percentage of positive tumour cell percentage and the positive staining intensity. The intensity of the staining was classified as follows: negative, 0; weak, 1; moderate, 2; and strong, 3. In addition, staining was evaluated according to the stained tumour cell percentage in the field of view as follows: negative, 0; 0%-25%, 1; 26%-50%, 2; 51%-75%, 3; and 76%-100%, 4.
The product of the intensity staining score and stained cell percentage represented the overall IHC score (0-12). For statistical analysis, scores of 0 to 7 were considered low expression and scores of 8 to 12 were considered high expression. Two independent pathologists accompanied and assessed the staining procedure and results.

| Cell proliferation, migration and invasion assays
Cell proliferation capacity was assessed by the clone formation assay. Approximately 2.0 × 10 2 cells per well were grown in 6-well plates containing DMEM with 10% FBS at 37°C. The cells were treated with a 10% formaldehyde solution and stained with 0.1% crystal violet (Sigma) after two weeks of culture. The calculation of the colonies formed was performed using a camera (Olympus). The cell migration ability was assessed by scratch wound assay. Approximately 4.0 × 10 5 cells were seeded in 6-well plates and cultured overnight to reach >80% confluence in an atmosphere containing 5% CO 2 at 37°C. Subsequently, a 200-μL pipette tip was used to make a longi-  Then, the spheres were photographed using a microscope and counted with ImageJ cell counter.

| Binding mode of TAK1 and YAP proteins
The crystal structure of protein TAK1 was obtained from Protein Data Bank (PDB ID: 2EVA). As the YAP crystal structure has not yet been solved, the three-dimensional structure of the YAP Pro101-Gln200 region was predicted ab initio using the online QUARK algorithm. 28,29 The protein-protein binding mode of TAK1 and YAP was predicted using the HDOCK server. 30 The residues involved in the interactions between TAK1 and YAP were showed by LigPlot 31 and PyMOL software (The PyMOL Molecular Graphics System, version 2.0 Schrödinger, LLC.).

| Co-culture of GC cell lines and cancerassociated fibroblasts
Cancer-associated fibroblasts (CAFs) were obtained from human GC tissue and normal fibroblasts (NFs) from the non-cancerous region at least 5 cm from the outer margin of the tumour in the same patient.

| Haematoxylin and eosin (HE) staining
The tissue specimens were sliced into 4 μm sections and mounted on silanized glass slides. After deparaffinization and hydration, the sections were stained by incubation with haematoxylin solution at 35°C for 3 minutes. Subsequently, the sections were immersed 5 times in 0.5% acid ethanol (1% HCl in 70% ethanol) and rinsed in distilled water. Then, the sections were stained by incubation with eosin solution at 35°C for 1 minutes, dehydrated with graded alcohol and cleared with xylene. Finally, the sections were examined under a light microscope (IX81, Olympus) at 20× magnification. The analysis was conducted on a Q Exactive mass spectrometer coupled to Easy nLC (Thermo Fisher Scientific) using a routine method.

| Immunofluorescence
To detect the TAK1 and YAP cell location, the GCSCs were incubated in 4% paraformaldehyde solution for 15 minutes and permeabilized in 0.5% Triton at room temperature for 5 minutes. A 5% BSA solution was used to block the samples, which were subsequently incubated with anti-TAK1 (cat. no. sc-7967, 1:500, Santa Cruz) and anti-YAP

| Luciferase reporter assay
Luciferase reporter assays were conducted through the Dualluciferase Reporter Assay System (Promega). The wild-type or F I G U R E 1 Up-regulation of TAK1 in human gastric tissues. A, The mRNA expression levels of TAK1 in GC and adjacent tissues (**P < .01); the protein expression levels of TAK1 in 6 paired GC(T) and adjacent tissues(N) were evaluated by Western blot. B, Representative IHC staining images of TAK1 in normal, primary cancer and recurrent cancer tissues of GC. C, The OS (left panel) and RFS (right panel) for the high and low TAK1 expression groups according to IHC staining intensity were analysed by Kaplan-Meier analysis. D, A multivariate analysis by the Cox multivariate proportional hazard regression model indicated that up-regulation of TAK1 may be an independent prognostic factor for the OS and RFS rates in patients with GC. The HRs are presented as the means (95% CI) mutant SOX2 that had the predicted binding site was settled and incorporated into a pGL3 dual-luciferase vector to constitute the pGL3-SOX2-wild type (SOX2-wt) or pGL3-SOX2-mutant (SOX2mut) reporter vector. The SOX2-wt or SOX2-mut co-transfection was performed with shYAP or negative control into HEK293T cells using Lipofectamine 2000. Luciferase activity was measured according to the manufacturer's guidelines after 48 hours of transfection.
The experiment was carried out in triplicate and expressed as the mean ± the standard deviation (SD).

| TAK1 is up-regulated in gastric cancer
To study the TAK1 potential role in GC, the TAK1 expression level was measured in GC and the related adjacent non-cancerous samples. As shown in Figure 1A, the TAK1 protein and mRNA levels in  vs. 40 months, respectively; log-rank = 9.023, P = .003; Figure 1C).

Multivariate analysis indicated that the expression level of TAK1
may be an independent risk factor for OS and RFS in GC patients ( Figure 1D). The high TAK1 expression group exhibited smaller OS and RFS rates (OS: hazard ratio (HR) = 2.011, 95% confidence interval (CI), 1.198-3.376, P = .008; RFS: HR = 1.786, 95% CI, 1.083-2.946, P = .023). These data suggest that the TAK1 expression level can be useful as an independent factor for GC prognosis prediction.

| TAK1 is expressed at high levels and promote self-renewal in gastric cancer stem cells
The CSC presence figures prominently in tumour proliferation, invasion and recurrence, resulting in poor outcomes and limited therapeutic options. Given the TAK1 role in promoting various GC malignant phenotypes, the TAK1 expression level has been estimated in GCSCs.
As self-renewal is a CSCs distinct feature, GCSCs were enriched by inducing the formation of GC spheroids. As shown in Figure 4A, an enhanced GC spheroid formation was observed in TAK1-overexpressing greater than that of CD44 + and/or CD44-GC cells alone ( Figure 4D).
These results indicated that TAK1 is overexpressed in GCSCs.

| TAK1 stabilizes YAP independently from its kinase activity
To further dissect the mechanism by which TAK1 contributes to self-renewal of GCSCs, we performed the liquid chromatography tandem mass spectrometry (LC-MS/MS) to qualitatively analyse components of TAK1 binding protein in GCSCs. LC-MS/MS analysis identified that TAK1 interacted with YAP, which inhibited by Hippo pathway ( Figure S2A). As there are no previous reports that TAK1 can interact with YAP and regulate biological processes in CSCs, immunoprecipitation was performed to attest the TAK1 binding to YAP ( Figure 5A). Further experiments showed that the TAK1 loss was associated with declines in YAP levels and negatively related to p-YAP. Other molecules of the HIPPO pathway were not shown to be related to TAK1 ( Figure 5B). Moreover, immunofluorescence (IF) analysis proved that there is a direct interaction between YAP and TAK1 in GCSCs ( Figure 5C).
The Hippo signalling pathway kinase cascade partially inhibits YAP by phosphorylation of its Ser127 residue, resulting in binding of YAP 14-3-3 and cytoplasmic retention. 32 To understand how TAK1 and YAP interact, the way of binding between TAK1 and YAP was predicted by protein-protein docking simulations. As shown in Figure 5D Figure 5G). In addition, spheroid-formation assays revealed that the TAK1 and YAP combination promotes GCSCs self-renewal ( Figure 5H). These results indicate that YAP can avoid being degraded in the cytoplasm by covering its phosphorylation sites with TAK1 and induces an increase in non-phosphorylation YAP, favours YAP nuclear localization and promotes YAP target genes in GCSCs.

| TAK1/YAP axis activates SOX2 and SOX9 transcription in the gastric cancer stem cells selfrenewal
Yes-associated protein exerts its role as a transcriptional coactivator mainly through interaction with the transcription factor TEAD1. 34 Then, the UCSC 35 and JASPAR databases 36 were used to examine the promoter region of several stemness-associated genes for TEAD1 binding sites ( Figure S2B). The competence for TEAD1 binding was assessed by a luciferase reporter system as SOX2 and SOX9 have been described as a relevant regulator of the population of GCSCs. 37,38 The dual-luciferase results show that shYAP reduced the SOX2 and SOX9 promoter activity ( Figure S2C). In addition, decreased SOX2 and SOX9 expression has been shown to be related to TAK1 interfered expression ( Figure 6A).
To study the role of TAK1/YAP axis in GC progression, it was investigated whether the TAK1/YAP axis regulates the GCSCs selfrenewal. The TAK1 knockdown decreased the GCSCs sphere formation ability. However, the sh-TAK1 inhibitory effect on GCSCs self-renewal was recovered by YAP overexpression. Furthermore, the GCSC spheroid-forming ability was not increased after YAP was interfered in the overexpressing TAK1 cell line ( Figure 6B). Flow cytometry analysis showed similar results ( Figure S3A). These results confirm that TAK1/YAP axis participates in the GCSCs self-renewal, thereby promoting the tumour growth in GC.

F I G U R E 4
Expression of TAK1 in gastric CSCs. A, Representative images of sphere formation in 3 × Flag-TAK1, 3 × Flag-Vector, sh-NC, sh-TAK1 in MKN45 and MGC803 cell lines (left and right panel). The numbers of sphere formation were measured and are shown in the bar graph. All data were mean ± SD and from three independent experiments (*P < .05; **P < .01), scale bars = 200 μm. B, Flow cytometry analysis of CD44 + or CD133 + populations in spheres derived from 3 × Flag-TAK1, 3 × Flag-Vector, sh-NC, sh-TAK1 in MKN45 cells.

| IL-6 promotes increased TAK1 expression in gastric cancer
Cancer stem cells are regulated by the tumour microenvironment, and some cytokines in the malignant tumour are closely related to the self-renewal, maintenance and growth of these cells. 39 IL-6 has been suggested to enhance the TAK1 expression level and induce the association between TAK1 and GNAI3. 40 Thus, TAK1 expression level was detected in IL-6-treated MKN45 and MGC803 cells to verify whether the TAK1 expression in GC is induced by IL-6. The results showed that the TAK1 expression levels increased with increasing IL-6 incubation time ( Figure 6C). In addition, the TAK1 expression levels in cancer cells that were co-cultured with CAFs were higher than those co-cultured with NFs ( Figure 6D). Similarly, we observed that IL-6 can enhance the expression of TAK1 at different expression levels of TAK1 ( Figure 6E).
GCSCs have been enriched by inducing spheroid formation through cocultured cells. As shown in Figure 6F, the tumour spheroid-formation frequencies are enhanced in the culture environment in which IL-6 was added. Flow cytometry analysis showed that the CD44 + and CD133 + GC cell proportion was increased in GC spheroids when co-cultured with IL-6 ( Figure S3B). These results reveal that increased TAK1 expression in GC cells can be regulated by the increase in the IL-6 levels.

| TAK1 inhibits the chemosensitivity of gastric cancer cells
As TAK1 was identified as a GCSCs self-renewal promoter, it was subsequently investigated whether TAK1 can affect the GC cell resistance to chemotherapeutic drugs 5-fluorouracil (5-FU) and cisplatin.
The cell viability was significantly increased in TAK1-overexpressing

| D ISCUSS I ON
TAK1 is a member of the serine/threonine family of protein kinases.   K63W , TAK1 E29A and YAP S127A , compared with wild type of TAK1 and YAP. F, TAK1 E29A does not interact with YAP, and YAP S127A also does not interact with TAK1 through Co-IP assay in MKN45-derived gastric cancer spheroids. G, After using the TAK1 kinase inhibitor 5Z-7-xox, TAK1 does not interact with YAP and TAK1 K63W (a mutation in the TAK1 kinase site) also does not interact with YAP through Co-IP assay in MKN45-derived gastric cancer spheroids. H, Representative images of sphere formation in TAK1, YAP S127A and TAK1 E29A in MKN45 cells. The numbers of sphere formation were measured and are shown in the bar graph. All data were mean ± SD and from three independent experiments (**P < .01; ***P < .001), scale bars = 200 μm GCSCs were sorted from the GC cells and identified as containing CD133 and CD44. Our data also showed that the CD44 + and CD133 + cell proportion in GCSCs is significantly increased and that the TAK1 mRNA expression level is positively correlated with this.
Transcriptional regulators YAP and TAZ have been reported as central malignancy determinants, due to their essential functions in tumour initiation, development, metastasis and chemoresistance. 23 In addition, they play a crucial role in tumorigenesis through F I G U R E 6 IL-6-dependent TAK1/YAP promotes the sphere formation in GC cells. A, Representative images showing the protein and mRNA expression levels of SOX2 and SOX9 after TAK1 knockdown in MKN45-derived gastric cancer spheroids. The quantification of bands represented in a bar graph (**P < .01, ***P < .001). B, Representative images of sphere formation in sh-NC, sh-TAK1, sh-NC + 3Flag-Vector, sh-TAK1 + 3Flag-YAP and 3Flag-TAK1 + shYAP in MKN45 cells. The numbers of sphere formation were measured and are shown in the bar graph. All data were mean ± SD and from three independent experiments (**P < .01), scale bars = 200 μm. C, Representative images showing the expression levels of TAK1 increased with the incubation time of IL-6 in MKN45 and MGC803 cell lines (upper and lower panel). D, Representative images showing the expression levels of TAK1 increased when co-cultured with CAFs, compared to co-cultured with NFs in MKN45 and MGC803 cell lines. E, The representative expression levels of TAK1 with the incubation of IL-6 in sh-TAK1 and 3×Flag-TAK1 MKN45 cell lines. F, Representative images of sphere formation in MKN45 and MGC803 cell lines with or without IL-6 incubation. The numbers of sphere formation were measured and are shown in the bar graph. All data were mean ± SD and from three independent experiments (**P < .01), scale bars = 200 μm. (g). IL-6 is secreted by CAFs in the cell microenvironment and promotes the expression of TAK1 in cells. TAK1 prevents the degradation of YAP in the cytoplasm by binding to YAP and promotes the transcription of SOX2 and SOX9 in the nucleus. Finally, TAK1 promoted the self-renewal ability of gastric cancer stem cells or its downstream signalling is capable of expanding CSCs in cancers of the breast, colon, ovary, lung, brain, and head and neck. [51][52][53] Our experiments showed that IL-6-induced TAK1 activation is a relevant factor pointing to TAK1 as a central molecule for the GCSCs self-renewal.
CSCs are highly resistant to chemotherapy and radiotherapy and, consequently, tend to co-operate with cancer recurrence. 54 There is evidence indicating that the TAK1 expression down-regulation can reduce the protein levels of YAP/TAZ in vitro and in vivo and modulate the pancreatic cancer intrinsic chemoresistance. 55 Then, we assessed the TAK1 potential therapeutic value in GC treatment.
Both in vivo and in vitro experiments suggested that regulating TAK1 expression in GC has a remarkable therapeutic potential for the GC treatment, particularly with regard to tumour cell chemosensitivity to chemotherapy.

| CON CLUS ION
In summary, the results presented here show that TAK1 promotes tumour propagation, can be activated by IL-6 and binds to YAP in GCSCs. Our results reveal a new critical paradigm in determining the GCSCs fate, which is a therapeutic promise for the GC treatment ( Figure 6G).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

CO N S E NT FO R PU B LI C ATI O N
Not applicable.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and/or analysed during the present study are available from the corresponding author on reasonable request.