Proteins with AT-rich interactive domains (ARIDs), which are helix–turn–helix motif-based DNA-binding domains spanning ~100 amino acid residues, constitute a highly conserved family. Interestingly, not all ARID-containing proteins conform to a specific pattern of binding sequences and AT-rich sequences (1). However, ARID family proteins still participate in a comprehensive range of DNA interactions (2). In humans, 15 members have been identified. They fall into seven classes based on the degree of sequence identity, and these classes are named ARID1 through ARID5, JARID1 and JARID2 in mammals (3, 4). Members of the ARID family are involved in the modification of chromatin structure and regulate the expression of specific genes that are directly associated with cellular growth, differentiation, and development (5, 6). The precise functions of all the ARID family proteins are still not entirely clear.

Digestive organs, including the liver, gallbladder, pancreas, oesophagus, stomach and intestine, provide suitable conditions for the transport and absorption of food and nutrient metabolism. These digestive organs are impacted by our diet to a great extent, and they are easily susceptible to various diseases with obvious symptoms. Patients with solid tumours that originate from these organs tend to have higher morbidity and mortality compared with other organ systems (7). Previous studies have shown that there are common intersections between tumorigenesis mechanisms and regulatory networks in different digestive cancers (8–10).

The ARID gene family has been confirmed to participate in tumour growth and invasion. Downregulation of ARID1A and ARID1B, the variant subunits of the BRG1/BRM-associated factor complex from the SWI/SNF chromatin remodelling family, in various cancer cells might lead to deficiency in DNA repair (11). Another chromatin remodelling factor, ARID2, plays a metastasis suppressor role in hepatocellular carcinoma and rectal cancer, and mutants of this gene exhibit tumorigenic functions (12, 13). ARID3A promotes cell proliferation, migration and invasion in colorectal cancer and is significantly associated with the prognosis of patients (14). The mechanisms by which ARID family members function in different types of cancer often vary and even have contrary outcomes. JARID2 has been reported to facilitate tumorigenesis and epithelial-mesenchymal transition in bladder cancer and acts as a tumour suppressor in myeloid neoplasms (15, 16). The close connections among family members and the certain similarities in the tumorigenesis and development of digestive cancer are the basis for the use of systematic analysis. Stem cell-like features effectively work in tumour initiation and are closely related to resistance to drug treatment. Tumour microenvironment refers to the initiation, development and metastasis that are closely related to the internal and external environment of tumour cells more than just the location. Stemness and tumour microenvironment can be used as auxiliary indicators reflected the status of cancer.

In the present study, we investigated the expression pattern of the ARID gene family members and their association with the stemness, the tumour microenvironment and patient outcomes in digestive cancer. Our study highlights the essential differences in the ARID genes that exist between digestive cancers and the need to study each ARID as a separate entity.

Previous studies have confirmed that the ARID gene family is highly associated with tumour growth in proliferation, apoptosis, migration, and invasion. As a conserved family, the expression and functions of these genes retain tightly interactive relationships. However, few studies have focused on the integral study of the 15 genes. Our study provides the first systemic analysis of ARID gene family members in the digestive system. We found noticeable heterogeneity in the expression levels of ARID genes among different tumours which suggested the works by these genes were not conservative. There was intersection among the functions of gene family in digestive cancers. However, the functions of different genes or in different cancers were still variant and the expression heterogenicity was an external manifestation. ARID3C, working as a transcriptional coactivator and a potential regulator of early processes in B cell antigen receptor signalling, was expressed preferentially in B lineage lymphocytes, and ARID3A had the highest expression levels in activated follicular B cells (24, 25). ARID3C had notable differential expression between tumour and normal tissues in CHOL, and our immune analysis showed that ARID3C expression was negatively associated with immune score in this type of cancer. Considering the role of ARID3C in lymphocytes, we inferred that it might be effectively activated through B cell antigen receptors to influence CHOL progression. It prompted the different regulator roles of ARIDs like activator or inhibitor in pathways were at work in the family heterogenicity among different types of digestive cancers.

ARID1A and ARID1B are two mutually exclusive regulatory subunits within the multiprotein chromatin remodelling complex SWI/SNF. Indeed, we found that they had similar expression patterns in most of the tested cancer types. Furthermore, they had similar expression patterns in immune infiltration subtypes, as they were highly expressed in immune subtypes enriched with IFN-γ and TGF-β infiltrates. Previous studies have confirmed that ARID1A deficiency leads to the loss of the tumour suppressive function of TGF-β (26). Mutations in ARID1B and TGF-β were simultaneously detected in liver metastases, but direct evidence of their association is still lacking in existing studies (27). ARID2 could function as a tumour suppressor gene in colorectal cancer and constitute an important downstream element in the regulatory pathway of TGF-β in LIHC (28, 29). IFN-γ is a cytotoxic molecule secreted from immune cells and promotes immunomodulation and anticancer activity. A pronounced difference in PD-L1 induction by IFN-γ treatment existed between ARID1A-deficient and control gastric cancer cells (30). Downregulation of ARID2 inhibited DNA methyltransferase, which increased methylation in the promoter region and increased the transcription of IFN-γ (31). Therefore, similar downstream mechanisms of ARIDs might share common characteristics in tumour regulation, and we could utilise these findings for the further investigation of other genes in digestive cancer.

The RNAss and DNAss that we referred to in this study were calculated based on machine learning which reflected the relationships between expression and stemness respectively from transcriptional and epigenetic regulation which might not be fully coincident. Strong correlation indicated family gene meddled the stemness of digestive cancer. Positive association theoretically contributed to the cancer stem cell and the negative promoted dormancy. The clinical value of ARID1A and ARID1B was profound in CHOL patients (32). Interestingly, although ARID1A and ARID1B had a positive association with DNAss estimated to have epigenetic modifications in CHOL, these two genes showed either a negative correlation or a correlation that was not significant with stemness when measured by RNAss. Our results indicated that ARID1A and ARID1B could alter the biological behaviour of CHOL cells through epigenetic regulation of cancer stemness. The expression of ARID1A impacted the biological functions, including migration, invasion and sphere formation activity, of CHOL and was confirmed to exhibit a tumour-suppressive effect through the regulation of ALDH1A1, a stemness marker, with histone acetylation (33). The role of ARID1B in stem cell-like features was unclear in previous CHOL studies. However, evidence supports that ARID1A and ARID1B work together in DNA modulation, which might provide a basis for subsequent investigations of stemness regulation (34). On these grounds, the associations with stemness might not be direct intervention.

The expression of 6 of the 15 gene family members was associated with the prognosis of LIHC patients, and the results were coincident with the hazard regression analysis. We chose the LIHC transcriptome for further analysis. Except for prognostic significance, most of the family members were associated with the immune subtype and cancer stage, indicating that these genes play roles in tumour progression and immune microenvironment abnormalities. Bioinformatics analysis of public databases and expression analysis of LIHC tissue samples showed that ARID4B was highly expressed in LIHC compared with adjacent normal tissues and strongly correlated with clinical features (35). The expression of ARID1A and ARID2 inhibited the proliferation and migration of LIHC cells and, thus, act as tumour suppressors (36, 37). Therefore, the predicted roles of the ARID gene family members as a tumour suppressor or promoter in LIHC were in accord with our results, but validation from further experiments is required for an in-depth exploration.

The expression signature presented in our study might still have some limitations that may need further improvement. First, we performed an analysis based on transcriptome data from public databases, and more clinical trials should be included for confirmation in larger populations. Second, biological experiments and in-depth investigations of the potential mechanisms for the ARID gene family members in our study may better support the signature. Third, we selected digestive cancer for research because almost every ARID gene is related to several tumours in these organs and the internal heterogeneity among specific systems is less than that among whole organs. However, studies of ARIDs in pan-cancer may have their own advantages in further expanding this research.

We provided a comprehensive and systematic study of the expression of all ARID gene family members in patients with digestive cancer. Our results indicated that the putative tumour promoter or suppressor role of ARIDs was not consistent among the gene family members within a specific cancer type and that even the specific ARID member behaved differently in single or various cancer types. Taken together, our work uncovered the roles of the ARID gene family members in tumorigenesis, including the immune response and tumour microenvironment of digestive cancer, and might help to develop personalised treatment.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

WF had designed the study. YL and XZ had collected data. ZL, WW, and XC had analysed and interpreted the data. All authors were involved in writing paper and approved of the submitted and published versions.

This work was supported by grants from National Natural Science Foundation of China (Nos. 91959110 and 81972702), Natural Science Foundation of Beijing (No. 7204324) and National Multidisciplinary Cooperative Diagnosis and Treatment Capacity building project for major diseases: comprehensive diagnosis and treatment of gastrointestinal tumours.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.