MACC1 Correlates with Tumor Progression and Immune Cell Infiltration of Colon Adenocarcinoma and is Regulated by the lncRNA ZFAS1/miR-642a-5p Axis

Colon adenocarcinoma (COAD) is the most common pathologic type of colon cancer. Metastasis is responsible for the high mortality rate of patients with COAD. The gene, metastasis-associated in colon cancer 1 (MACC1), is a biomarker predictive of both metastatic and metastasis-free survival in patients with colon cancer and other solid tumors. However, the underlying mechanism by which MACC1 affect COAD progression and metastasis remains unknown. In this study, we analyzed the expression level and prognostic value of MACC1, as well as their correlation, in patients with various types of cancer included in The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) databases. High MACC1 expression was found to be significantly associated with poor prognosis in patients with COAD. Analysis of the potential upstream miRNA of MACC1 showed that miR-642a-5p was downregulated in COAD and was negatively correlated with MACC1 expression. Analysis of the upstream regulators of miR-642a-5p showed that the long non-coding RNA (lncRNA) ZFAS1was the most likely upstream regulator of miR-642a-5p. In addition, the expression of MACC1 correlated positively with tumor immune cell infiltration, as well as with the levels of biomarkers of five kinds of immune cells. In summary, these findings suggest that MACC1 contributes to COAD progression and immune cell infiltration via the ZFAS1/miR-642a-5p/MACC1 axis.


Introduction
Colon cancer is the third most common type of cancer [1] and the second most frequent cause of cancer-related deaths worldwide. It is estimated that more than one million people develop colorectal cancer every year [2]. Colon adenocarcinoma (COAD) is the most common type of colon cancer, accounting for approximately 90% of patients with these tumors. Despite advances in diagnosis and treatment, however, prognosis remains poor in patients with colon cancer, especially those with advanced disease, who are prone to recurrence and metastasis. Te 5-year relative survival rate of patients with colorectal cancer has been estimated at approximately 65%, ranging from 90% for patients with localized disease to low as 14% for patients with metastatic disease [3]. Te identifcation of therapeutic targets and the development of efective therapeutic agents to prevent the progression of early disease and to improve survival in metastatic colon cancer are therefore urgently needed.
Te metastasis-associated in colon cancer 1 (MACC1) gene, which is located on chromosome 7p21.1, was frst identifed in 2009. Genome-wide analysis of genes diferentially expressed in primary colon cancer, metastatic tumors and normal tissues has shown that MACC1 is an independent prognostic indicator of metastasis-free survival [4,5]. MACC1 plays a key role in the transition from adenoma to carcinoma in mice and humans [6,7] and has also been identifed as a metastatic and prognostic biomarker for other solid tumors, such as hepatocellular carcinoma, lung cancer, and breast cancer [8][9][10][11][12][13][14]. In vivo, MACC1 increases tumor cell proliferation and motility, supports cell survival, and regulates metabolism, thereby promoting metastasis. MACC1 has also been shown to be involved in tumor immunity. For example, MACC1 was found to regulate the expression of PDL1 and tumor immunity through the c-Met/AKT/mTOR pathway in gastric cancer [15]. However, the prognostic value of MACC1 in COAD and its correlation with immune cell infltration into tumors remain incompletely understood.
Te present study analyzed the expression levels and evaluated the prognostic value of MACC1 in multiple types of human cancers fnding that MACC1 was overexpressed in COAD, with this increased expression predictive poor clinical features and poor prognosis. Te potential carcinogenic mechanisms of MACC1 were assessed in COAD, as well as the correlation between MACC1 expression level and the degree of immune cell infltration. Te present study also analyzed the relationship between MACC1 expression and biomarkers of immune cells in COAD. Tese fndings showed that the ZFAS1/miR-642a-5p axis is a potential upstream pathway of MACC1 in COAD. Taken together, these results indicated that MACC1 contributes to COAD progression and immune cell infltration via the ZFAS1/miR-642a-5p/ MACC1 axis, suggesting that the ZFAS1/miR-642a-5p/ MACC1 axis may provide a novel therapeutic target and a valuable prognostic indicator for COAD.

Gene Expression Analysis.
Te Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/) is a cancer genomics program designed to identify and classify major oncogenic genomic changes to enhance our understanding of cancer [16]. Te Genotype-Tissue Expression Project (GTEx) is a data resource and tissue bank established by the National Institutes of Health Common Fund that can enable determination of the relationship between genetic variation and gene expression in a variety of human tissues [17]. RNA-seq data for MACC1 mRNA and relevant clinical data across 33 cancer types and normal tissues in the TCGA and GTEx database were downloaded from the UCSC XENA website (https:// xenabrowser.net/datapages/). Data for various tumor cell lines were downloaded from the Cancer Cell Line Encyclopedia (CCLE) database (https://portals. broadinstitute.org/ccle/). RNA-seq data on gene expression in the FPKM format were converted into TPM format and log2 transformed for further analysis. Te levels of expression of MACC1 in tumors and corresponding normal tissues in the data downloaded from the TCGA and GTEx databases were analyzed, as were the associations between MACC1 expression and clinical characteristics in patients with COAD. Diferences in expression were analyzed using the Mann-Whitney U test, with p < 0.05 defned as statistically signifcant. Results were visualized using the R package ggplot2.

GEPIA2 Database Analysis. Gene Expression Profling
Interactive Analysis 2 (GEPIA2; https://gepia2.cancer-pku. cn/#index) is a web-based tool for interactive analysis based on normal and cancer samples from the TCGA and GTEx databases. GEPIA2 ofers a series of functions such as differential expression analysis, profle plotting, correlation analysis, survival analysis, detection of similar genes, and dimensionality reduction analysis, all of which may be customized [18]. Te expression levels of MACC1 and lncRNAs in COAD tissues from the TCGA normal and GTEx databases as controls were analyzed using the "Expression DIY-Box Plot" module of GEPIA2 using the settings, p-value cutof � 0.001, |log2FC (Fold change)| cutof � 1, and "Match TCGA normal and GTEx data." Correlations between MACC1 expression and immune cell biomarkers in COAD were analyzed in the "Correlation Analysis" mode of GEPIA2, with correlation coefcients calculated using Spearman analysis. Te selection criteria were set as |R| > 0.1 and p < 0.05 for identifying statistically signifcant.

Analysis of the CPTAC and UALCAN Databases. Te
Clinical Proteomic Tumor Analysis Consortium (CPTAC, https://proteomics.cancer.gov/programs/cptac) is a centralized repository of publicly available proteomic sequence datasets. CPATC identifes the protein composition and characteristic proteome of each tumor sample by mass spectrometry [19]. UALCAN (https://ualcan.path.uab.edu) is a user-friendly interactive web-portal designed to facilitate analysis of the correlation between gene expression in tumor subgroups and patient survival [20]. UALCAN was used to analyze the expression of MACC1 total protein from CPTAC database, by entering "MACC1" and selecting the dataset of colon cancer and MACC1 proteomic expression profle based on sample types.

Analysis of Diagnostic and Prognostic
Value. Te diagnostic value of MACC1 in COAD was determined using receiver operating characteristic (ROC) curves. Te areas under the ROC curves (AUCs) were between 0.5 and 1, with AUCs >0.9 defned as being highly accurate. Te relationships between MACC1 expression and indicators of patient prognosis, including overall survival (OS), disease-specifc survival (DSS), and progression-free interval (PFI), were evaluated in 33 types of cancers, with the results presented as Forest plots and Kaplan-Meier curves. A uniformly standardized pan-cancer dataset from the UCSC XENA database (https://xenabrowser.net/datapages/), with samples from patients followed-up for less than 30 days and those with fewer than 10 samples in a single cancer species excluded. Hazard ratios (HRs) with 95% confdence intervals (CIs) were calculated by univariate regression analyses. Survival rates in patients with diferent levels of MACC1 expression were determined by the Kaplan-Meier method and compared by log-rank tests. Te optimal cut-of value for MACC1 expression that diferentiated patients into subgroups with high and low expression was calculated by the minimum p-value approach using the R package maxstat (Version 0.7-25). Statistical analyses were performed using the survival package in R software (version 3.6.3), with the forestplot and survminer packages used for visualization. A p value <0.05 was considered statistically signifcant.
2.5. starBasev2.0 Database Analysis. Te starBasev2.0 database (https://starbase.sysu.edu.cn/) allows systematic identifcation of RNA-RNA and protein-RNA interaction networks [21], which consist of several online prediction programs, such as PITA, RNA22, miRmap, microT, mi-Randa, PicTar, and TargetScan. Te miRNAs acting upstream of MACC1 were predicted by the starBasev2.0 database in the miRNA-mRNA panels. Predicted miRNAs appearing simultaneously in more than two programs were considered candidate miRNAs for MACC1 and included in subsequent analyses. Correlation analyses for miRNA and MACC1, lncRNA and hsa-miR-642a-5p, and lncRNA and MACC1 in COAD samples in the starBasev2.0 database were performed in the miRNA-Target CoExpression or RNA-RNA CoExpression panels. Te levels of expression of hsa-miR-642a-5p in COAD and normal tissue were also analyzed in the miRNA Diferential Expression panels, as well as to predict candidate lncRNAs that could potentially bind to hsa-miR-642a-5p. Te miRNA-mRNA and lncRNA-miRNA interactions networks were visualized by Cytoscape (version 3.8.2).
2.6. TIMER2.0 Database Analysis. Te TIMER2.0 web server (https://timer.cistrome.org/) allows a comprehensive analysis of gene expression and tumor infltrating immune cells in multiple cancers [22]. Te "Immune-Gene" module of this website was used to analyze the correlations between MACC1 expression levels and tumor infltration by immune cells, including B cells, CD8 + T cells, CD4 + T cells, Tregs, NK cells, macrophages, neutrophil cells, and dendritic cells, in COAD were selected. Te p-values and Rho values were obtained via the purity adjustment Spearman's rank test. Te correlations were shown as scatter plots, with p < 0.05 considered statistically signifcant.

Statistical Analysis.
Statistical analyses were performed using R software (V3.6.3), although some of the statistical analyses were performed automatically using the online databases described above. In all statistical analyses, p < 0.05 was considered statistically signifcant for all statistical analyses.

Expression of MACC1 mRNA in Human Cancers.
To investigate the potential role of MACC1 in carcinogenesis, its level of expression was determined in normal tissues, tumor cell lines and various primary human cancers. Assessment of MACC1 expression levels in 31 normal tissues from the GTEx database showed that MACC1 was expressed at low levels in most normal tissues, although its expression was higher in bone marrow ( Figure 1). Analysis of MACC1 expression in 21 tumor cells lines in the CCLE database showed that MACC1 was highly expressed in most cell lines (Figure 1(b)). Analysis of the diferential expression of MACC1 in primary tumors and adjacent normal tissues in data directly obtained from the TCGA database showed that the expression of MACC1 was signifcantly higher in eight types of human cancer, namely bladder urothelial carcinoma (BLCA), cholangiocarcinoma (CHOL), COAD, lung adenocarcinoma (LUAD), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), thyroid cancer (THCA), and uterine corpus endometrial carcinoma (UCEC), while signifcantly lower in fve types of cancer, namely head and neck squamous cell carcinoma (HNSC), kidney chromophobe tumors (KICH), liver hepatocellular carcinoma(-LIHC), lung squamous cell carcinoma(LUSC), and prostate adenocarcinoma(PRAD) (Figure 1(c)). Because the TCGA database included few normal tissue samples, data on tumor tissues in the TCGA database were combined with data on normal tissues in the GTEx database to analyze the diferential expression of MACC1 in 33 tumors. Compared with the corresponding normal samples, MACC1 expression was signifcantly higher in 20 types of cancer, namely BLCA, breast cancer (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), CHOL, COAD, difuse large B-cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma multiforme(GBM), kidney renal papillary cell carcinoma (KIRP), lower grade glioma (LGG), LUAD, ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), READ, STAD, testicular germ cell tumors (TGCT), THCA, thymoma (THYM), UCEC, and uterine carcinosarcoma (UCS). In contrast, MACC1 expression was signifcantly lower in eight other types of cancer, namely adrenocortical carcinoma (ACC), HNSC, acute myeloid leukemia (LAML), LIHC, LUSC, pheochromocytoma and paraganglioma (PCPG), PRAD, and skin cutaneous melanoma (SKCM). Te levels of expression of MACC1, however, did not difer signifcantly in kidney chromophobe (KICH) and kidney renal clear cell carcinoma (KIRC) samples (Figure 1(d)). Taken together, these fndings suggest that MACC1 may play diferent roles in the development of diferent tumors.

Te Prognostic Value of MACC1 in Human Cancers.
Te prognostic value of MACC1 expression in patients with 33 types of human cancers from the TCGA database was analyzed. Te correlations between MACC1 expression with overall survival (OS), disease-specifc survival (DSS) and progression-free interval (PFI) were determined using the R Package survival (version 3.2-7), with forest plots were used to evaluate the relationship between MACC1 expression and patient prognosis. Higher expression of MACC1 mRNA was indicative of shorter OS in patients with LGG (HR  ). Te levels of MACC1 expression, however, did not correlate with survival parameters in patients with other types of cancer. Taken together, these results suggested that the potential prognostic value of MACC1 mRNAs difers among diferent types of cancer. Moreover, Kaplan-Meier analyses showed that MACC1 was signifcantly correlated with OS, DSS, and PFI in patients with COAD, LGG, PAAD, UVM, BLCA, KIRC, KIRP, SKCM, KICH, and CHOL, suggesting that MACC1 expression may be a biomarker in these tumors ( Figure 3).

Expression of MACC1 in Patients with COAD.
Pan-cancer analysis showed that MACC1 mRNA was upregulated in COAD. To further understand the importance of MACC1mRNA and protein levels in COAD, MACC1 expression data from the TCGA and CPTAC databases were analyzed. Unpaired analysis of samples from the GEPIA2 database showed that MACC1 mRNA levels were signifcantly higher in 275 COAD samples than in 349 normal colon tissue samples ( Figure 4). Analysis of data from CPTAC in the UALCAN database showed that MACC1protein expression was signifcantly higher in COAD than in normal colon tissue samples (Figure 4(b)).

Predicted Upstream Potential miRNAs of MACC1.
Non-coding RNAs (ncRNAs) are an important class of gene regulators responsible for regulating the expression of many key genes at the transcriptional and post-transcriptional levels. miRNAs are mainly involved in the negative regulation of gene expression. To assess whether MACC1 is modulated by miRNAs, starBase2.0 was used to predict miRNAs acting upstream of MACC1. Tis analysis identifed 33 upstream miRNAs that could potentially bind to MACC1 (Supplementary Table S1). To improve visualization, Cytoscape software was used to establish a miRNA-MACC1 regulatory network ( Figure 6). Because miRNAs usually act by negatively regulating target gene expression, miRNAs were expected to negatively regulate MACC1. Evaluation of the correlation between miRNA and MACC1 expression showed that MACC1 mRNA negatively correlated with the levels of hsa-miR-141-3p, hsa-miR-142-5p, hsa-miR-126-5p, hsa-miR-186-5p, hsa-miR-155-5p, and hsa-miR-642a-5p in COAD ( Figure 6(b), Supplementary Table S1). Further determination of the levels of expression of these six miRNAs in COAD showed that hsa-miR-642a-5p was markedly downregulated in COAD ( Figure 6(c)) and that lower expression of hsa-miR-642a-5p was associated with poorer OS in patients with COAD ( Figure 6(d)). Tese fndings suggest that hsa-miR-642a-5p may regulate MACC1 mRNA in COAD.

Predicted
Upstream Potential lncRNAs of hsa-miR-642a-5p. Te starBase database predicted that 113 possible lncRNAs acted upstream of hsa-miR-642a-5p. Cytoscape software was used to improve visualization by constructing a lncRNA-hsa-miR-642a-5p regulatory network ( Figure 7). According to the hypothesis of competing endogenous RNA (ceRNA), these lncRNAs should be oncogenes in COAD, with the competitive binding of these lncRNAs to shared miRNAs enhancing the expression of MACC1 mRNA. Tus, the levels of these lncRNAs should correlate negatively with the levels of miRNA, but positively with the levels of MACC1 mRNA. Determination by GEPIA2 of the levels of expression of the predicted lncRNAs in COAD and corresponding normal tissues showed that, of the 113 possible lncRNAs, only four, MIR4435-2HG, LINC00511, MAFG-AS1, and ZFAS1, were expressed at signifcantly higher levels in COAD than in corresponding normal tissue samples (Figures 7(b)-7(e)). Te correlations of these four lncRNAs (MIR4435-2HG, LINC00511, MAFG-AS1, and ZFAS1) with the levels of hsa-miR-642a-5p and MACC1 mRNA in COAD were confrmed using the starBase database. ZFAS1 was negatively correlated with hsa-miR-642a-5p, but positively correlated with MACC1 mRNA (

Correlation between MACC1 Expression with Immune Cell Infltration and Biomarkers of Immune Cells in COAD.
Since MACC1 has been reported to be involved in tumor immunity, the correlation between MACC1 mRNA expression and immune cell infltration level was evaluated.    Journal of Oncology   (Table 3). Tese fndings suggest that MACC1 is involved in the immune regulation of COAD by regulating immune cell infltration.

Discussion
Although MACC1 was shown to be a critical regulator and biomarker for progression and metastasis in over 20 types of cancer, including colon cancer, HCC, bladder cancer, and esophageal cancers [23][24][25][26], the mechanisms by which MACC1 is involved in the development and progression of COAD remain unclear. Te present study analyzed the expression of MACC1 in 33 types of cancer using data from the TCGA and GTEx databases and found that MACC1 mRNA was more highly expressed in COAD than in corresponding normal colon tissues. Te expression of MACC1 mRNA in COAD was subsequently validated by analyses of the GEPIA2 and CPTAC databases. Survival analysis indicated that high MACC1 mRNA expression in patients with COAD was associated with poor prognosis, which was consistent with previous fndings [27]. MACC1 knockdown was also found to markedly inhibit cell proliferation, migration, invasion, colony formation, and tumorigenesis, both in vitro and in vivo, and to induce apoptosis in colorectal cancer (CRC) cells [28]. Tese fndings suggested that MACC1 has an important role in CRC carcinogenesis and progression through the β-catenin signaling pathway and mesenchymal-epithelial transition [28]. Combined with the results of the present study, these fndings demonstrate that MACC1 plays a carcinogenic role in COAD. Non-coding RNAs (ncRNAs), including small micro-RNAs (miRNAs), lncRNAs, and circular RNAs (circRNAs), were shown to be involved in multiple biological processes in various types of cancer by directly or indirectly interfering with gene expression by inter-communication through the ceRNA mechanism [29][30][31][32][33]. Tus, upstream regulatory miRNAs of MACC1 could be predicted using the program starBasev2.0, which includes PITA, RNA22, miRmap, microT, miRanda, PicTar, and TargetScan [34]. Of the 33 miRNAs identifed, most were found to be tumor suppressive in various types of cancers. For example, hsa-miR-18a-5p signifcantly reduced the hazard of dying in patients with CRC, regardless of tumor site [35], and low expression of miR-642a-5p was associated with a poor prognosis in patients with CRC. Moreover, miR-642a-5p was found to inhibit colon cancer cell migration, invasion, and EMT by targeting COL1A1 [36]. Overexpression of Neural Cell Adhesion Molecule 1 signifcantly inhibited the migration of ameloblastoma cells and was regulated by miR-141-3p [37]. In CRC, miR-186-5p acts as a cell cycle suppressor to inhibit tumor progression. Over-expression of miR-186-5p was found to signifcantly down-regulate the expression of SMAD6/7, resulting in decreased expression of CyclinD1 and c-Myc. Furthermore, over-expression of miR-186-5p promoted CRC cell apoptosis and inhibited their viability, proliferation, and migration [38]. Expression and correlation analyses showed that hsa-miR-642a-5p was the most likely tumor-suppressive miRNA targeting MACC1. No study to date, however, has assessed the role of hsa-miR-642a-5p targeting of MACC1 in CRC.
Based on the ceRNA hypothesis [39], the levels of expression of lncRNA and mRNA should correlate with each  other. Tus, the potential lncRNAs associated with the hsa-miR-642a-5p/MACC1 axis in COAD should be highly expressed. Analysis using starBase v2.0 software predicted 113 lncRNAs that could be associated with the hsa-miR-642a-5p/MACC1 axis. Expression and correlation analyses identifed ZFAS1 as the most likely upstream lncRNA. ZFAS1 lncRNA is a major protein regulator involved in various human cancers, including COAD. ZFAS1 has been shown to promote the proliferation, migration, and invasion of esophageal squamous cell carcinoma (ESCC) cells, and to inhibit their apoptosis by upregulating STAT3 and downregulating miR-124 [40]. ZFAS1 was also more highly expressed in glioma tissues and cells than in normal brain tissues and normal astrocyte HA cells. Moreover, high ZFAS1 expression was associated with poor prognosis among patients with glioma. Downregulation of ZFAS1 or upregulation of miR-1271-5p was found to inhibit the progression of  glioma by enhancing the apoptosis and inhibiting the repressing proliferation, migration, and invasion of glioma cells [41]. Te oncogenic activity of ZFAS1 and its signifcant upregulation, together with the elevated expression of DDX21 and POLR1B in CRC cells and tissues, further leads to poor clinical outcomes. Knockdown of ZFAS1 substantially suppressed CRC cell proliferation, invasion, and migration, and increased cell apoptosis [42]. ZFAS1 has also been shown to increase NOP58 and SNORD12C/78 expression in CRC cells and tissues [43].  Furthermore, ZFAS1 was found to promote the progression of CRC by competitively binding miR-150-5p, which plays a tumor suppressor role in CRC by targeting VEGFA [44]. Taken together, these fndings suggest that ZFAS1 plays a carcinogenic role in various human cancers and that the ZFAS1/hsa-miR-642a-5p/MACC1 axis constitutes the potential regulatory pathway that promotes the progression of COAD.
Immune cell infltration into tumors has been shown to afect the efectiveness of chemotherapy, radiotherapy, and immunotherapy, as well as the prognosis of cancer patients [45][46][47]. Te present study showed that MACC1 upregulation increases the infltration of various immune cells into COADs, including NK cells, macrophages, and neutrophils. Moreover, MACC1 expression showed signifcant positive correlations with biomarkers of NK cells, macrophages,  neutrophils, and dendritic cells. Tese results suggest that MACC1 is associated with immune cell invasion into COAD. However, the present study has several limitations. Te results of this study were based on bioinformatics analysis from various databases but were not validated experimentally. Confrmation of these fndings requires in vivo and in vitro experiments.
In conclusion, this study we have demonstrated that MACC1 is overexpressed in several types of human cancers and that MACC1 overexpression correlates with poor prognosis and increased tumor immune cell infltration in patients with COAD. Te present study also identifed the ZFAS1/hsa-miR-642a-5p axis (Figure 9) as the most likely upstream pathway of MACC1 in COAD. Tese results suggest that MACC1 participates in tumor immunity and progression of COAD through the ZFAS1/hsa-miR-642a-5p/MACC1 pathway.

Conflicts of Interest
Te authors declare that there are no conficts of interest.