GNG4, as a potential predictor of prognosis, is correlated with immune infiltrates in colon adenocarcinoma

Abstract The tumour microenvironment (TME) and immunosuppression play an important role in colon cancer (CC) metastasis, which seriously affects the prognosis of CC. G protein subunit gamma 4 (GNG4) has been shown to participate in tumour progression and the tumour mutation burden (TMB) in colorectal cancer. However, the effect of GNG4 on the CC TME and immunology remains elusive. Weighted gene coexpression network analysis (WGCNA) was employed for screening aberrantly expressed genes associated with the immune score, and GNG4 was then selected through prognostic and immune correlation analysis. Based on RNA sequencing data obtained from the TCGA and GEO databases, the expression pattern and immune characteristics of GNG4 were comprehensively examined using a pan‐cancer analysis. Upregulation of GNG4 was linked to an adverse prognosis and immune inhibitory phenotype in CC. Pan‐cancer analysis demonstrated higher GNG4 expression in tumours than in paired normal tissue in human cancers. GNG4 expression was closely related to prognosis, TMB, immune checkpoints (ICPs), microsatellite instability (MSI) and neoantigens. GNG4 promoted CC cell proliferation, migration and invasion and participated in immune regulation in the TME. Significantly, GNG4 expression was found to negatively correlate with tumour‐infiltrating immune cells, ICP, TMB and MSI in CC. GNG4 expression predicted the immunotherapy response in the IMvigor210 cohort, suggesting that GNG4 could be used as a potential biomarker in CC for prognostication and immunology. Moreover, the expression of GNG4 predicted the immunotherapy response of ICB in CC.

GNG4 on the CC TME and immunology remains elusive. Weighted gene coexpression network analysis (WGCNA) was employed for screening aberrantly expressed genes associated with the immune score, and GNG4 was then selected through prognostic and immune correlation analysis. Based on RNA sequencing data obtained from the TCGA and GEO databases, the expression pattern and immune characteristics of GNG4 were comprehensively examined using a pan-cancer analysis. Upregulation of GNG4 was linked to an adverse prognosis and immune inhibitory phenotype in CC. Pan-cancer analysis demonstrated higher GNG4 expression in tumours than in paired normal tissue in human cancers. GNG4 expression was closely related to prognosis, TMB, immune checkpoints (ICPs), microsatellite instability (MSI) and neoantigens. GNG4 promoted CC cell proliferation, migration and invasion and participated in immune regulation in the TME. Significantly, GNG4 expression was found to negatively correlate with tumour-infiltrating immune cells, ICP, TMB and MSI in CC. GNG4 expression predicted the immunotherapy response in the IMvigor210 cohort, suggesting that GNG4 could be used as a potential biomarker in CC for prognostication and immunology. Moreover, the expression of GNG4 predicted the immunotherapy response of ICB in CC.

K E Y W O R D S
colon cancer, GNG4, immune infiltration, immunotherapy, tumour microenvironment

| INTRODUC TI ON
Colorectal cancer (CRC), with the third highest morbidity and mortality rates, is one of the most common malignant tumours in the world. 1 Twenty percent of patients have metastatic CRC when first diagnosed with CRC, and another 25% of patients with local tumours will eventually develop metastasis. Fewer than 20% of patients with metastatic CRC survive for more than 5 years. 2 Cancer immunotherapy has shown promise in the treatment of recurrent or metastatic cancer. 3 ICB has demonstrated durable responses and long-lasting clinical benefits for a wide range of solid tumour types. Pembrolizumab and nivolumab are efficient for treatment of metastatic CRC with deficient mismatch repair (MMR) or high microsatellite instability (MSI-H), for which accelerated FDA approval have been granted for two promising programmed cell death 1 (PD1)-blocking antibodies 4 in CRC. However, the limited response of treatment in patients suggests the importance of effective markers for immunotherapy. 5 Currently, multiple potential factors have been identified, including the tumour mutation burden (TMB), 6 MSI status, 7 neoepitope load, 8 PD-L1 level, 9 CD8 + T-cell density, 10 interferonγ gene signature, 11 and MHC and T-cell receptor repertoire. 12 However, those biomarkers lack extensive validation and adequate data support.
Given the severe side effects and substantial economic burden of cancer treatments, novel and versatile biomarkers that can predict the ICB response are urgent. 13 Herein, a comprehensive analysis using public databases of colon cancer (CC) was conducted, in which GNG4 was identified as an immunotherapy marker associated with prognosis. GNG4, one of the 14 γ subunits of human G proteins, is crucial in guanosine triphosphatase (GTPase) activity, G protein-effector interactions and guanosine diphosphate (GDP) synthesis. Notably, hypermethylated GNG4 was found in bladder cancer and glioblastoma. 14,15 Furthermore, high expression of GNG4 has also been associated with the progression of CRC 16 and a variety of malignant phenotypes of lung adenocarcinoma. 17 GNG4 was demonstrated to be the key element of the CRC TMB, which is essential for immune checkpoint inhibitor (ICI) therapy of CRC. 18

| Immune score evaluation, WGCNA construction and gene set enrichment analysis (GSEA)
The tumour purity and immune score for 566 CC samples in the GSE39582 data set were calculated using the 'estimate' package in R based on their gene expression matrix. Then, the DEG matrix was further analysed by WGCNA. A correlation between the gene modules and an immune score greater than 0.3 was used to identify immune-related gene sets. Finally, GNG4 as a potential predictor of prognosis correlating with immune infiltrates was determined through the GEPIA and Timer databases.
To investigate the potential mechanism, the RNA expression profiles were divided into two groups according to the GNG4 median value. GSEA was applied using the R package 'clusterProfiler' based on the DEGs between the two GNG4 expression groups. The gene sets of c5.go.bp.v7.4.symbols.gmt were obtained from GSEA (http:// www.gsea-msigdb.org/gsea/index.jsp).

| GNG4 methylation analysis
The methylation data for the GNG4 promoter were downloaded from the TCGA database. CpG island methylation data were visualized by the MEXPRESS database (https://mexpr ess.be/). 21 2.5 | GNG4 expression and copy number variation (CNV) and drug sensitivity exploration CNV data of CC patients were also obtained from the SangerBox website. The CellMiner (https://disco ver.nci.nih.gov/cellm iner/) web tool was used to assess the relationship between GNG4 expression and pharmacological data in the NCI-60 cell line set. 22

| Cell culture
Human CC cells SW480, HCT116, DLD1 and RKO and the human colon epithelial cells NCM460 and HCoEpiC were obtained from the Shanghai Cell Bank of the Chinese Academy of Science and cultured in DMEM medium with 10% foetal bovine serum (FBS). All of the cell types were incubated in a humidified atmosphere with 5% CO 2 at 37°C. The cell medium was changed every 2 days based on this culture environment.

| Quantitative real-time PCR (RT-qPCR)
Total RNA from cells was extracted using an EZ-10 DNAaway RNA Mini-prep kit (Sangon Biotech Co., Ltd.). After the concentration and quality of RNA at 260/280 nm absorbance was determined, reverse transcription was performed by using PrimerScript RT Master mix (Takara Biotechnology Co., Ltd.). All of the PCR primers were obtained from Sangon Biotechnology. An ABI 7300 PCR system (Applied Biosystem) was used to perform the quantitative PCR reaction with SYBR Green Master Mix (Thermo Fisher Scientific). The primer sequences were as follows: GNG4 (forward primer, 5′-GCATC TCC CAA GCC AGG AAAGC-3′ and reverse primer, 5′-GCAGG CAC TGG AAT GAT GAGAGG-3′). Relative expression was normalized to GAPDH as an internal control and calculated using 2 −△△CT .

| Transwell assay
For the Transwell assay, 5 × 10 4 cells/well were resuspended in 200 μL of serum-free medium in the upper chamber and 600 μL of medium supplemented with 10% FBS were filled in the lower chamber (8μm pore size, Coster, Corning, USA). After incubation for 48 h (migration assay) or 72 h (invasion assay) in a humidified atmosphere (5% CO 2 at 37°C), the cells in the parietal chamber were removed.
Then the cells on the submucosal surface were immobilized in 4% paraformaldehyde for 30 min and stained with a crystal violet solution. The number of migrated cells in the four random regions of each membrane layer was counted under the microscope.

| GNG4 expression and immune subtypes, molecular subtypes and ICB
The TISIDB database was used to explore the correlations between

| Lentivirus packaging and infection
The GNG4 shRNA sequence was cloned into the pLKO-puro vector to generate the lentiviral shRNA constructs against human GNG4.

| Statistical analysis
All of the statistical analyses were performed using SPSS software (version 25.0) or R software (version 4.0.3). Spearman's coefficients was used to evaluate the correlation between GNG4 expression and variables. Statistical significance was determined as: NS, not significant; *p < 0.05; **p ≤ 0.01; ***p ≤ 0.001.

| GNG4 is an immunogenic-related gene affecting the prognosis of CC patients as screened by WGCNA
As shown in the volcano plots in Figure 1A, dysregulated genes between normal and tumour groups were identified based on the criteria of |logFC| < 1 and an adjusted p value <0.01.
We hypothesized that gene modules with high expression and a negative correlation with the immune score or gene modules with low expression and a positive correlation with the immune score in tumours could serve as potential therapeutic targets.
Ultimately, we identified 87 genes negatively associated with the immune score and 163 genes positively associated with the immune score ( Figure 1C). We identified three oncogenic genes that were positively correlated with prognosis, highly expressed and negatively correlated with immune invasion ( Figure 1D) from the GEPIA2 database. Similarly, as shown in Figure 1E, eight suppressor oncogenes were screened, which were negatively correlated with prognosis, with low expression in tumours and positively correlated with immune invasion. GNG4 was further identified by immune cell correlation analysis in the Timer database ( Figure 1F).
GNG4 was highly expressed in CC. High GNG4 expression was detrimental to patients, including a poorer prognosis, lower immune score and fewer immune cells ( Figure 1F,G), especially CD8 + T cells.

| GSEA of GNG4
To explore the possible mechanism of GNG4, RNA-seq data were grouped according to the GNG4 median value in the TCGA database, and the DEGs further underwent GSEA ( Figure 2A). The results of KEGG enrichment analysis revealed immune-related signalling pathways, including MYC targets ( Figure 2B,C), E2F targets ( Figure 2D), inflammatory response ( Figure 2E) and DNA repair ( Figure 2F).

| Pan-cancer immunological correlation of GNG4
Pan-cancer analyses aim to depict the immunological role of GNG4 comprehensively and thus to determine cancer types that may benefit from anti-GNG4 immunotherapy. Just as GNG4 was expressed inconsistently in tumours, the relationship between GNG4 and immunomodulators ( Figure 4A) or immunoregulatory cells was diverse ( Figure 4B).
Of note, GNG4 expression was negatively correlated with a majority of immunomodulators, including MHC molecules, receptors, chemokines and immunostimulators in CC. Similarly, there was a significant negative correlation between GNG4 expression and TIICs in the tumour microenvironment (TME) that were observed by using the 'ssGSEA' algorithm ( Figure 4B). Among them, the immune cells negatively correlated with GNG4 included activated CD4 + T cells, CD8 + T cells, MDSC, effector memory CD8 + T cells, central memory CD4 + T cells and type 2 T helper cells ( Figure 4C). Furthermore, we evaluated the correlation between GNG4 expression and immune checkpoints in multiple cancers and demonstrated that GNG4 was negatively associated with immune checkpoints, including CD274 (PDL1), LGA3, CTLA-4 and PDCD1 (PD1) in the COAD data set ( Figure 5A,B). The TMB and MSI status were found to be potential determinants of response and resistance to ICBs. Thus, we analysed the relevance between GNG4 expression and TMB, MSI and MASH. We noted that there was a strong negative correlation with TMB and MSI in CC and a significant positive correlation with MASH. This result implied the potential mechanisms of how GNG4 functions in tumour progression and ICB therapy ( Figure 5C, Figure S2).
In summary, pan-cancer analysis suggested that GNG4 is an on- expression in the TME were diverse, indicating that the potential regulation mechanisms by GNG4 differ. In CC, GNG4 had an obvious immunosuppressive effect, suggesting the potency of CC for anti-GNG4 immunotherapy.

| Expression, DNA methylation and mutational analyses of GNG4 in CC
The extent of malignancy associated with GNG4 in CC was further assessed in the TCGA, GSE39582 and GSE21510 data sets. GNG4 was stably and highly expressed in cancer tissues compared with normal adjacent tissues in the TCGA and GSE39582 data sets, as discussed previously ( Figure 6A,B). A similar result was obtained from GSE21510, which had 123 tumour and 25 normal tissues ( Figure 6C).
As is well known, DNA methylation is crucial in tumorigenesis. In addition, we found that GNG4 expression was lower in poorly differentiated colon cancer tissues compared with moderately/well differentiated colon cancer tissues ( Figure S1A).
The GNG4 methylation level was significantly lower in cancer tissues compared with normal tissues ( Figure S1B). As demonstrated, the GNG4 methylation level was significantly negatively correlated to expression levels in the TCGA database (COR = −0.708, p < 0.001, Figure 6D). Further analysis of the MEXPRESS database revealed a strong negative correlation between GNG4 expression and CpG island methylation ( Figure 6E).
In addition, the abundance of CD8 + T cells in the TME was negatively correlated with GNG4 expression (Figure 6F), and positively correlated with GNG4 methylation (Figure 6G), suggesting that the immunosuppressive role of GNG4 overexpression may be regulated by methylation. The molecular subtype can also be used for selecting neoadjuvant chemotherapy, radiotherapy and several targeted therapies by predicting the clinical response. The position of the GNG4 CNV on the chromosomes was plotted by the TCGA database ( Figure 6H). As shown in Figure 6F, the 15 most commonly mutated genes in CC had higher mutation rates in the low-expression GNG4 group ( Figure 6I). In addition, the level of CD8 + T cells varied with changes of the GNG4 CNV ( Figure 6J).
In the TISIDB database, the GNG4 CNV also showed a negative correlation with CD8 + T cells ( Figure 6K).
GNG4 expression also varied with different immune subtypes and molecular subtypes in CC. Immune subtype C6 (TGFβ dominant) patients had the lowest GNG4 expression, while immune subtype C1 (wound healing) patients had the highest GNG4 expression ( Figure 6L). Patients with molecular subtypes CIN were more likely to have the lowest GNG4 expression, while immune subtype C1 (wound healing) patients had the highest GNG4 expression. Patients with molecular subtypes CIN were more likely to have relatively high levels of GNG4, while molecular subtypes HM-indel patients had the lowest GNG4 expression levels ( Figure 6M).

| GNG4 promotes the proliferation, migration and invasion of CC cells
The relative expression of GNG4 on the mRNA level ( Figure 7A) in human CC cells SW480, HCT116, DLD1 and RKO, and human colon epithelial cells NCM460 and HCoEpiC was evaluated.
The level of GNG4 was high in SW480 and HCT116 cells, moderate in DLD1 and RKO cells and lower in NCM460 and HCoEpiC cells.
Next, we established GNG4 knockdown SW480 cells ( Figure 7B). There was no significant difference in average body weight between the control and shGNG4 groups ( Figure 7J). Collectively, these findings suggested that GNG4 was an agonist of CC cell proliferation, migration and invasion.

| GNG4 expression and the immune score and immune cells in CC
We explored the association of GNG4 with the immune score of CCs in the TCGA database and GSE38582 and GSE21510 data sets.
The results indicated that there were negative correlations between GNG4 expression and the immune score, but not the stromal score or estimate score in the TCGA ( Figure 8A), GSE39582 ( Figure 8B) or GSE21510 ( Figure 8C) data sets, further illustrating the specificity of GNG4. Next, we explored whether GNG4 also affected the expression levels of other cells in the TME. The activity of immune cells in the high GNG4 expression group decreased, including activated CD4 + T cells and CD8 + T cells, macrophages, Th1 cells, NK cells, DCsl and Th17 cells, leading to reduced TME infiltration, which promoted the growth of the tumour ( Figure 8D-F). Furthermore, GNG4 expression was negatively correlated with a variety of immune cells in the TCGA database ( Figure 8G, Figure S3), GSE39582 ( Figure 8H) and GSE21510 ( Figure 8I).

| GNG4 expression and immune checkpointing in CC
Given that chemokines and chemokine receptors can recruit immune cells, including CD8 + T cells into the TME, we evaluated the association of GNG4 with these factors in the TCGA database and validated      .c e ll A ct iv a te d .C showed that GNG4 expression was highly negatively correlated with chemokines such as GZMA, CCL5, IFNG and CD8A ( Figure 9A-C) and the chemokine receptors CD274, CD200, CD80 and TIGIT ( Figure 9D-F). These results revealed that GNG4 could be a potential immunomarker and therapeutic target by regulating T cells in the TME of CC.

| GNG4 predicts clinical benefit of ICB
The correlation between GNG4 expression and drug responses were revealed by the CellMiner (https://disco ver.nci.nih.gov/cellm iner/) web tool. Patients with high expression of GNG4 benefited from targeted drugs such as pentostatin, streptozocin and dacarbazine, while they had poor responses to a geldanamycin analog and alvespimycin ( Figure 10A and Figure S4). We then investigated whether GNG4 expression could predict an immunotherapy benefit using the GSE142693 data set and validating with the IMvigor210 data set.
There was no statistically significant difference in GNG4 expression to immunotherapy in GSE142693, possibly due to the small sample size ( Figure 10B). However, in the IMvigor210 data set, the results showed that patients with low GNG4 expression subtypes responded better to PD1 anti-PD-L1 therapy, whereas patients with high GNG4 subtypes had CR or PR ( Figure 10C). These results suggested that GNG4 can predict the effects of ICB treatment.

| DISCUSS ION
Immune infiltration is closely associated with the pathogenesis and progression of CC. TIICs, affecting angiogenesis and metastasis of colon cancer (mCC), are appealing therapeutic targets. 24 Recently, the ICBs for CC, including ICIs and CAR-T cells, have been less effective for patients, which was caused by the low immune cell infiltration level into the TME and by effector T cell exhaustion. 25  For mechanism studies of CC progression and the response to immunotherapy interventions, aberrantly expressed genes associated with an immune score were screened by WGCNA, and GNG4 was subsequently selected through prognostic and immune correlation analysis. Our study revealed the potential value of GNG4 as predictor of prognosis and clinical response to ICB.
Findings from preclinical research 26 showed that GNG4 could be used as a potential biomarker to predict the response of immunotherapy in bladder cancer, implying that GNG4 may be a broad-spectrum therapeutic target. Hence, we conducted a com-  CD274  PDCD1  CD276  ADORA2A  BTLA  CD200  CD200R1  CD80  CD86  CEACAM1  CTLA4  HAVCR2  IDO1  KIR3DL1  LAG3  LAIR1  LGALS3  PVR  TIGIT  VTCN1   PVR   CEACAM1 LGALS3 predicted shorter survival in COAD and LUAD, which was consistent with previous studies. 16,17,27 A pan-cancer immunological correlation of GNG4 aimed to determine cancer types that may benefit from anti-GNG4 immunotherapy. The results showed that the relationship between GNG4 and immunomodulators or immunoregulatory cells was diverse (Figure 2A,B), indicating the heterogeneity of GNG4 regulation. Of note, GNG4 expression was negatively correlated with a majority of immunomodulators, including MHC molecules, receptors, chemokines and immunostimulators in CC. Finally, CC was shown to be an ideal cancer for anti-GNG4 immunotherapy.
Despite the clinical success of antibodies against immunomodulators such as PD-L1/PD-1 and CTLA4, only a small fraction of individuals have a lasting benefit, suggesting the urgency of an efficient cancer-immunity cycle to iteratively proceed and expand, and thus generate anticancer immune responses. 28,29 Our research revealed that GNG4 may be involved in the initiation of several key steps in the cancer immune cycle. Tumour-infiltrating lymphocytes in the TME have been shown to be efficient in predicting prognosis and immunotherapeutic efficacy for cancer. 30 In the TCGA-COAD cohorts, the infiltration levels of several effector TIICs, such as activated CD8 + T cells, activated CD4 + T cells, activated B cells, Type 1 T helper cells, macrophages and natural killer cells, were significantly downregulated in CC patients with high GNG4 expression, which was verified in the validation group. Moreover, high GNG4 expression was also promoted in the C1 (wound healing) immune subtypes and CIN molecular subtypes, which indicated that GNG4 may be involved in TME remodelling.
We found that the GNG4 methylation status was closely related to its mRNA expression and positively correlated with CD8 + T cells.   GNG4 was shown to promote tumour progression in CRC 16 ; however, its potential mechanisms remain unclear. As a result, increased GNG4 induced the proliferation, migration and invasion of CC cells.
The GSEA analysis revealed a marked enrichment of MYC targets, E2F targets, the inflammatory response and DNA repair in the high GNG4 expression group. The MYC oncogene is a grand orchestrator of cancer growth and immune evasion, and it can regulate the TME by affecting both innate and adaptive immune effector cells and immune regulatory cytokines. 32,33 Relating to immune infiltration, E2Fs are potential biomarkers for prognosis of many cancers, including human gastric carcinoma, pancreatic adenocarcinoma and clear cell renal cell carcinoma. [24][25][26][27][28][29][30][31][32][33][34][35][36] In addition, low expression of GNG4 was associated with enrichment in the TNFγ and INFα signalling pathways. INFα belongs to Type I IFNs and has been reported to be associated with immune-mediated and inflammatory disorders. 37 IFNγ is mainly produced by natural killer (NK) and T helper cells and involved in inflammation and autoimmunity. 38 In summary, the invasion and metastasis of CC was promoted by GNG4 via multiple mechanisms, which may also explain the tumour immune escape. Furthermore, we found that patients with a low GNG4 expression subtype responded better to PD1 anti-PD-L1 therapy, suggesting that GNG4 expression can predict ICB treatment efficacy, based on the IMvigor210 data set. Although more studies are still needed for further confirmation, the immune microenvironment and the immunotherapy response of CC may be related to GNG4.
To conclude, we comprehensively analysed correlations between immune infiltration and oncogenes in the TME of CC. GNG4 was screened and found to affect the invasion and migration of CC cells and regulate the immune microenvironment remodelling in CC.
GNG4 has an obvious immunosuppressive effect that indicated the crucial roles of CC for anti-GNG4 immunotherapy.  We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.

CO N FLI C T O F I NTER E S T S TATEM ENT
The author declared no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study are available from the corresponding author upon request.