CXXC finger protein 4 inhibits the CDK18‐ERK1/2 axis to suppress the immune escape of gastric cancer cells with involvement of ELK1/MIR100HG pathway

Abstract Gastric cancer, is the fourth most common tumour type yet, ranks second in terms of the prevalence of cancer‐related deaths worldwide. CXXC finger protein 4 (CXXC4) has been considered as a novel cancer suppressive factor, including gastric cancer. This study attempted to investigate the possible function of CXXC4 in gastric cancer and the underlying mechanism. The binding of the ETS domain‐containing protein‐1 (ELK1) to the long non‐coding RNA MIR100HG promoter region was identified. Then, their expression patterns in gastric cancer tissues and cells (SGC7901) were detected. A CCK‐8 assay was used to detect SGC7901 cell proliferation. Subsequently, SGC7901 cells were co‐cultured with CD3+ T cells, followed by measurement of CD3+ T cell proliferation, magnitude of IFN‐γ+ T cell population and IFN‐γ secretion. A nude mouse model was subsequently developed for in vivo validation of the in vitro results. Low CXXC4 expression was found in SGC7901 cells. Nuclear entry of ELK1 can be inhibited by suppression of the extent of ELK1 phosphorylation. Furthermore, ELK1 is able to bind the MIR100HG promoter. Overexpression of CXXC4 resulted in weakened binding of ELK1 to the MIR100HG promoter, leading to a reduced proliferative potential of SGC7901 cells, and an increase in IFN‐γ secretion from CD3+ T cells. Moreover, in vivo experiments revealed that CXXC4 inhibited immune escape of gastric cancer cells through the ERK1/2 axis. Inhibition of the CXXC4/ELK1/MIR100HG pathway suppressed the immune escape of gastric cancer cells, highlighting a possible therapeutic target for the treatment of gastric cancer.


| INTRODUC TI ON
Gastric cancer is currently the second most common cause of cancer-related deaths in China and worldwide, while being also the second most commonly diagnosed type of cancer. 1,2 Interestingly, gastric cancer can be treated with relative ease if caught at an early stage. However, patients with gastric cancer at an advanced stage show extensive invasion and metastasis beyond the stomach leading to many difficulties in treatment and a generally poor prognosis. 3 Lack of accurate molecular markers for early-stage gastric cancer can lead to progression of this disease in up to 70% of patients. 3 Generally, the treatment for gastric cancer contains standard gastrectomy as well as neoadjuvant chemotherapy, radiotherapy and molecular-targeted therapies. 3 Furthermore, according to the description of tumour status (T/N/M and stage), combined treatments which relieve suffering and acquire best results are performed. 3 Immunotherapy is a novel treatment option using immune tumour vaccines or antibodies against tumour antigens to stimulate one's immune system to fight the tumour. Immune checkpoint inhibitors such as programmed death 1/programmed cell death-ligand 1 (PD-1/PD-L1) are now widely used as targets for immunotherapy, as they activate effector T cells to kill stomach cancer cells. 4 CXXC proteins are a group of zinc finger proteins that preferentially bind to CpG islands on genomic DNA thus regulating transcription. 5 CXXC4 is a tumour suppressor expressed at low levels in tumours compared to normal tissues. Thus, very low expression of CXXC4 correlates with poor survival outcome. 6 An intriguing finding is that down-regulation of CXXC4 can activate the Wnt pathway, which leads to malignant renal cell carcinoma. 6 Thus, CXXC4 functions as a tumour suppressor in gastric cancer, and can be negatively controlled by enhancer of zeste homolog 2 and participates in mitogen-activated protein kinase (MAPK) axis to suppress tumour growth. 6,7 More importantly, CXXC4 has been found to inhibit the extent of ETS domain-containing protein-1 (ELK1) phosphorylation, 8 which is relevant as phosphorylation of ELK1 promotes the development of gastric cancer. 9 ELK1 is a transcription factor binding to purine-rich DNA sequences and has been shown to be up-regulated in various cancers. 10 ELK1 may contribute to CD8+ T cell differentiation by activating the extracellular regulated protein kinases (ERK); this may help eradicate cancer cells by activating T cells. 11 However, the relationship between CXXC4 and ELK1 in gastric cancer is not clear. In this study, we conducted a series of bioinformatic analyses to explore potential genes involved in gastric cancer. Based on the microarray dataset of gastric cancer GSE49051 in the GEO database, we successfully isolated CXXC4 as a significantly down-regulated gene in gastric cancer, while the StarBase database revealed a significant up-regulation of ELK1 in gastric cancer. Interestingly, we found CXXC4 and ELK1 were co-expressed in gastric cancer using the Chipbase website. Furthermore, Chipbase also revealed that ELK1 and MIR100HG are also co-expressed, and, at the same time, ELK1 binding sites in MIR100HG promoter region were predicted by JASPAR. MIR100HG is a long non-coding RNA (lncRNA) reported to be an oncogene in multiple cancers including colorectal, laryngeal squamous cell carcinoma and triple-negative breast cancer. [12][13][14] In gastric cancer, high expression of MIR100HG is positively correlated with clinical stage, tumour invasion and distal metastasis, thus potentially serving a potential prognostic biomarker for gastric cancer. 15 However, how these molecules can play a role in tumour immunotherapy remains unclear. Based on reported studies and bioinformatic analyses, we have been suggested the potential role of CXXC4/ELK1/MIR100HG in gastric cancer.
In our investigation, we unravelled the function of CXXC4 in T cell stimulation. CXXC4 inhibited the phosphorylation of ELK1 reducing its nuclear translocation, which restrained the expression of MIR100HG. Finally, the signalling of cyclin-dependent kinase 18-extracellular signal-regulated kinases1/2 (CDK18-ERK1/2) was prohibited and activated T cells reduced the proliferation of gastric cancer cells.

| Ethics statement
Our experiments were carried out with the approval of the Ethics   Under the manufacturer's instructions, pGCSIL-PUR lentivirus of encoding shRNA against CXXC4 (sh-CXXC4) and Lenti-OE lentivirus of overexpressing CXXC4, MIR100HG or CDK18 from Shanghai Genechem Co., Ltd., as well as ERK1/2 pathway inhibitor GDC-0994 from MedChemExpress (10 µmol/L, HY-15947), were used for cell transfection. After 6 hours, the culture medium was renewed and the cells were further cultured for 48 hours, followed by subsequent experiments.

| Cell proliferative potential detection
After overnight culture in a 96-well plate (10 4 cells per well), the cells were transfected and cultured for 0, 1, 2 and 3 days. Then, 10 µL of the cell counting kit 8 (CCK-8) solution was added for 2 hours of incubation. The absorbance value of each well at 450 nm was measured for plotting a growth curve. 16,17

| Isolation and identification of T cells
The heparin anticoagulant was put into a centrifuge, and the serum was discarded. Red blood cell lysis buffer was used for 10 minutes to lyse haemocytes at room temperature. After 5-minute centrifugation and supernatant removal, the cell pellets were washed and resuspended in 5 mL sterile phosphate buffer saline (PBS) solution, followed by cell counting. After that, the cells positive for PE-CD3+ (130-113-129; 1:50; Miltenyi biotec) were selected using flow cytometry.

| Proliferation detection of T cells co-cultured with gastric cancer cells
A 96-well plate was coated with anti-CD3/anti-CD28 tetramer antibodies (STEMCELL Technologies Inc). IL-2 (20 IU/mL) was added to the collected CD3+ T cells. The CD3+ T cells labelled by CFSE were co-cultured with gastric cancer cells in the RPMI 1640 medium at 37°C with 5% CO 2 . The content difference of CFSE was measured by flow cytometry, and cells with low CFSE signal were considered to be proliferating cells. 19

| Enzyme-Linked Immunosorbent Assay (ELISA)
The cytokine IFN-γ secreted by T cells was determined by ELISA.
In short, IFN-γ specific antibody and cells were incubated in a 96-well plate at 4°C for 24 hours. Then, the cells were washed with PBS and sealed with blocking buffer. A total of 100 mL detection antibody was added into the plate and then 100 μL of diluted streptavidin peroxidase. The reaction was stopped by adding 3,3′,5,5′-Tetramethylbenzidine substrate and sulphuric acid. Enzyme activity was measured as optical density value at the wavelength of 450 nm by using ELISA reader. 22

| Dual-luciferase reporter gene assay
The gastric cancer gene microarray GSE49051 was obtained from the Gene Expression Omnibus database (https://www.ncbi.nlm. The luciferase activity was expressed by the ratio of firefly RLU to renilla RLU.

| Chromatin immunoprecipitation (ChIP)
The cells were fixed and cross-linked with 1% formaldehyde for 10 minutes. Then, the chromatin was broken by ultrasonic wave and centrifuged at 12 000 g for 10 minutes at 4°C. The supernatant was collected and divided into two tubes then incubated with antibody to immunoglobulin G (IgG) (ab109489; 1:300; Abcam Inc) for NC and the specific antibody to phosphorylated ELK1 (p-ELK1) (ab28818; 1:100; Abcam) at 4°C overnight. Protein Agarose/Sepharose was used to precipitate DNA protein complex. After centrifugation for 5 minutes at 12 000 g, the supernatant was discarded and the nonspecific complex was washed. De-crosslinking was conducted at 65°C overnight, and the phenol/chloroform was used to extract, purify and collect DNA fragments. The binding of ELK1 protein to MIR100HG promoter was assessed by RT-qPCR.

| RNA isolation and quantitation
Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was carried out under the instructions of the TaqMan Gene Expression Assays protocol (Applied Biosystems, Thermo Fisher Scientific), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the internal reference ( Table 1). The relative expression of each target gene was calculated by 2 −ΔΔCt method. 23,24
The relative expression of protein was expressed as the ratio of gray value of protein band to be tested to that of internal reference band (β-actin).

| CXXC4 overexpression curbed the proliferative potential of gastric cancer cells and promoted the activation of T cells
In an attempt to investigate the effects of CXXC4 on the development of gastric cancer, the differential analysis on the gastric cancer microarray dataset GSE49051 was performed; 6924 DEGs were obtained, of which 3449 genes were highly expressed with the remaining 3475 genes being poorly expressed ( Figure 1A).
Subsequently, differential analysis of the microarray dataset GSE49051 showed that CXXC4 was expressed at a low level in gastric cancer ( Figure 1B). The StarBase database revealed that ELK1 was highly expressed in gastric cancer ( Figure 1C). The co-expression relationship between CXXC4 and ELK1 in gastric cancer was obtained from Chipbase ( Figure 1D). Then, RT-qPCR revealed that the expression of CXXC4 in gastric cancer tissues was notably decreased compared to that of adjacent normal tissues ( Figure 1E).
According to Western blot analysis, compared with adjacent normal tissues, the expression of CXXC4 in gastric cancer declined while the level of p-ELK1 increased ( Figure 1F). Then, the expres-  Figure 1G). Next, we measured the protein expression of ELK-1 and p-ELK1 in GES-1 and SGC7901 cells, and found that the p-ELK1 level was remarkably increased in SGC7901 cells compared to GES-1 cells ( Figure 1H).
In order to understand the effect of CXXC4 on the proliferative potential and immune escape capability of gastric cancer cells, we overexpressed CXXC4 in SGC7901 cells. As detected by Western blot analysis, the phosphorylation level of ELK1 decreased after overexpression of CXXC4 ( Figure 1I). Moreover, the proliferation of SGC7901 cells measured by CCK-8 assay revealed that the proliferative ability of SGC7901 cells was greatly decreased after overexpression of CXXC4 ( Figure 1J). Then, as assessed by flow cytometry

| CXXC4 suppressed the expression of MIR100HG in gastric cancer cells
To further study the mechanism of action for ELK1/MIR100HG in gastric cancer, the co-expression relationship between ELK1 and MIR100HG was retrieved and confirmed using the Chipbase website ( Figure 2A). The expression of MIR100HG in gastric cancer was measured by RT-qPCR, and the results showed that compared with normal tissues, the expression of MIR100HG was increased in gastric cancer tissues ( Figure 2B). Cell lines' data also showed a similar trend ( Figure 2C). The binding sites of ELK1 in the MIR100HG promoter region were predicted by JASPAR, and all three predicted binding sites were truncated ( Figure 2D). A dual-luciferase reporter gene assay found that the possible binding site of ELK1 in the MIR100HG promoter region was site 2 ( Figure 2E), which was mutated further. Then, ELK1 was found to improve the luciferase activity of wild-type MIR100HG, but had no significant effect on the mutant ( Figure 2F), indicating that ELK1 might bind to the site 2 of the MIR100HG promoter. ChIP assay revealed the combination of p-ELK1 and MIR100HG promoter ( Figure 2G). As a transcription factor, ELK1 regulates downstream genes by conducting nuclear import, an event triggered by ELK1 phosphorylation. 25 Therefore, we speculated that CXXC4 might inhibit the nuclear import of ELK1 by suppressing the phosphorylation of ELK1, and then inhibiting the expression of MIR100HG. Therefore, after overexpression of CXXC4 in SGC7901 cells, the distribution of p-ELK1 in the cells was detected by subcellular fraction. The results showed that phosphorylated ELK1 was translocated into the nucleus, while overexpression of CXXC4 inhibited the phosphorylation of ELK1 and thus suppressed the nuclear import of ELK1 ( Figure 2H). ChIP assay further suggested that the binding of the ELK1 to the promoter of

| CXXC4 overexpression inhibited the proliferation of gastric cancer cells and promoted the activation of T cells by downregulating the expression of MIR100HG
The aforementioned results have determined that highly expressed

| CXXC4 overexpression curbed the proliferation of gastric cancer cells and enhanced the activation of T cells by suppressing the expression of CDK18
With an attempt to further explore the correlation between CDK18 and gastric cancer, the expression of CDK18 in gastric cancer tissues and cells was measured, and the results showed that the expression of CDK18 was increased in gastric cancer tissues and cells ( Figure 4A,B). After overexpressing MIR100HG in SGC7901 cells, an elevated expression of CDK18 was observed ( Figure 4C). As re- ( Figure 4F). We then co-cultured the transfected gastric cancer cells F I G U R E 1 Overexpressed CXXC4 inhibits the proliferation of gastric cancer cells and promotes the activation of T cells by suppressing the phosphorylation of ELK1. A, Volcano map, the x-axis represents the different log10 P value, and the y-axis indicates logFoldChange. Each point on the map represents a gene; red colour represents up-regulated genes, and green refers to down-regulated genes in gastric cancer samples. B, The expression box plot of CXXC4 in the gastric cancer microarray dataset GSE49051. The x-axis represents the sample type, while the y-axis represents the expression value. C, The expression box plot of ELK1 in gastric cancer in the Starbase database. D, The co-expression box plot of CXXC4 and ELK1 according to Chipbase E, The expression of CXXC4 in gastric cancer and adjacent normal tissues measured by RT-qPCR, n = 37. *P < .05 tumour tissues vs adjacent normal tissues. F, The expression of CXXC4, ELK-1 and p-ELK1 in gastric cancer and adjacent normal tissues normalized to β-actin assessed by Western blot analysis, n = 37, *P promoter assessed by ChIP assay, n = 3, *P < .05 oe-CXXC4 vs oe-NC. J, The expression of MIR100HG measured by RT-qPCR, n = 3, * P < .05 oe-CXXC4 vs oe-NC. Measurement data were expressed as mean ± SD. Comparisons between tumour tissues and adjacent normal tissues in panel B were analysed using paired t test. The data conforming to normal distribution and homogeneous variance between two groups in panels C and E-J were analysed by unpaired t test. Comparisons among multiple groups in panel G were analysed using one-way ANOVA, followed by Tukey's post hoc test. The experiment was repeated three times independently Subsequently, we silenced CXXC4 and inhibited ERK1/2 axis in SGC7901 cells. After silencing CXXC4, the extent of ERK1/2 phosphorylation was elevated ( Figure 5D), and the proliferation ability of SGC7901 cells was also significantly increased ( Figure 5E). In contrast, the proliferation of CD3+ T cells and the proportion of IFN-γ+ T cells were notably reduced ( Figure 5F) and the cytokine IFN-γ secreted by the CD3+ T cells declined ( Figure 5G). However, the inhibition of ERK1/2 axis would counterbalance the action of silencing CXXC4 alone. These results suggest that highly expressed CXXC4 could inhibit the CDK18-ERK1/2 axis through MIR100HG, thus suppressing the proliferation of gastric cancer cells and promoting the activation of T cells.

| In vivo inhibitory role of overexpressed CXXC4 in immune escape of gastric cancer cells through the ERK1/2 axis
With an attempt to further validate our results in vivo, we es-  Figure 6E). Also, the results were reversed by further inhibition of the ERK1/2 axis.
To sum up, the CXXC4/CDK18-ERK1/2 axis was verified to be functional in vivo. F I G U R E 4 Enforced expression of CXXC4 leads to suppression of gastric cancer cells proliferation and the activation of T cells by inhibiting the expression of CDK18. A, The expression of CDK18 in gastric cancer and adjacent normal tissues measured by RT-qPCR, n = 37, *P < .05 tumour tissues vs adjacent normal tissues. B, The expression of CDK18 in GES-1 and the gastric cancer cell line SGC7901 assessed by RT-qPCR, n = 3, *P < .05 SGC7091 cell line vs GES-1 cell line. C, The expression of CDK18 in gastric cancer cell line SGC7901 normalized to β-actin measured by Western blot analysis, n = 3, *P < .05 oe-MIR100HG vs oe-NC. D, The expression of CDK18 in gastric cancer cell line SGC7901 normalized to β-actin assessed by Western blot analysis, n = 3. E, The expression of CXXC4, MIR100HG and CDK18 measured by RT-qPCR, n = 3. F, The proliferation of SGC7901 cells detected by CCK-8 assay, n = 3. G, The proliferation ability and the proportion of activated IFN-γ+ T cells of CD3+ detected by flow cytometry. H, The cytokine IFN-γ secreted by T cells measured by ELISA, n = 3, *P < .05 oe-CXXC4 + oe-NC vs oe-NC, # P < .05 oe-CXXC4 + oe-CDK18 vs oe-CXXC4 + oe-NC. Measurement data were expressed as mean ± SD. Comparisons between tumour and adjacent normal tissues in panel A were analysed using paired t test. The data conforming to normal distribution and homogeneous variance between two groups in panels B and C were analysed by unpaired t test. Comparisons among multiple groups in panels D, E, G and H were analysed using the one-way ANOVA, followed by Tukey's post hoc test. The data at different time points in panel F were analysed by the repeated measures ANOVA followed by Bonferroni's post hoc test. The experiment was repeated three times independently  F I G U R E 5 Re-expression of CXXC4 suppresses the proliferation of gastric cancer cells and promotes the activation of T cells by inhibiting CDK18-ERK1/2 axis through MIR100HG. A, The expression of CDK18, ERK1/2 and p-ERK1/2 normalized to β-actin measured by Western blot analysis in SGC7901 cells, n = 3, *P < .05 oe-CDK18 vs oe-NC. B, The expression of ERK1/2 and p-ERK1/2 normalized to β-actin assessed by Western blot analysis in SGC7901 cells, n = 3, *P < .05 oe-CXXC4 + oe-NC vs oe-NC, # P < .05 oe-CXXC4 + oe-MIR100HG vs oe-CXXC4 + oe-NC. (C), The expression of ERK1/2 and p-ERK1/2 normalized to β-actin measured by Western blot analysis, n = 3. *P < .05 oe-CXXC4 + oe-NC vs oe-NC, # P < .05 oe-CXXC4 + oe-CDK18 vs oe-CXXC4 + oe-NC. D, The expression of CXXC4, ERK1/2 and p-ERK1/2 normalized to β-actin assessed by Western blot analysis in SGC7901 cells, n = 3. E, The proliferation of SGC7901 gastric cancer cells detected by CCK-8, n = 3. F, The proliferation of CD3+ T cells and the proportion of IFN-γ+ T cells measured by flow cytometry, n = 3. G, The IFN-γ cytokine secreted by T cells assessed by ELISA, n = 3. *P < .05 sh-CXXC4 + DMSO vs sh-NC + DMSO, # P < .05 sh-CXXC4 + GDC-0994 vs sh-CXXC4 + DMSO. Statistical data were measurement data and presented as mean ± SD. Unpaired student's t test was used for comparison between the two groups in panel A, and comparisons among multiple groups were analysed by the one-way ANOVA in panels B-D, F and G. The data at different time points in panel E were analysed by the repeated measures ANOVA, followed by Bonferroni's post hoc test. The experiment was repeated three times independently In this research, the functions of CXXC4 in regulating the fate of tumour cells and differentiation at the T cell level were intriguing.

| D ISCUSS I ON
We elucidated one signalling axis from CXXC4, ELK1, MIR100HG to ERK1/2. However, previous researches have shown that the ELK1 protein is a downstream target of ERK1/2, activated ERK efficiently phosphorylating nuclear ELK1. 33,34 Another study shows reciprocal regulation of the transcription activation, and repressive activities are collaborated by MAPK-mediated phosphorylation of ELK1. 35 Given these results, we speculated that ELK1 and ERK1/2 might form a circular-feedback regulation mechanism. However, there are still lots of work to do to explore the involving participant molecules as many as possible to better understand the mechanism. Of note, only NCs were set in our investigation due to the limited conditions, and further studies should be performed by including internal positive controls for validation of our results. Additional attention should also be paid when relating the findings from animal models to the clinical setting because of physiological and pathological differences. Collectively, CXXC4 is a tumour suppressor that controls the growth of tumour cells and the statuses of T cells which will further target the tumour cells and kill them. MIR100HG and ERK axis can mediate the immune system activity and hence should be considered as chemical or antibody target for immunotherapy in the future.

| CON CLUS ION
CXXC4 overexpression has been shown to exert inhibitory effects on immune escape of gastric cancer cells, by suppressing SGC7901 cell proliferation and promoting release of IFN-γ by CD3+ T cells via the inactivated CDK18-ERK1/2 signalling pathway through down-regulation of ELK1 phosphorylation and MIR100HG expression.

ACK N OWLED G EM ENTS
We would like to thank our researchers for their hard work and reviewers for their valuable advice.

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

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data not shared.