Expression pattern of ACE2, TMPRSS2 and FURIN across different colorectal tissues
We first assessed the mRNA expression pattern of ACE2 in colorectal cancer tumors using in-house RNA-Seq, and found that the mRNA expression level of ACE2 was significantly higher in tumor tissues than in adjacent normal tissues (Figure 1A). However, the distinct expression was not validated in the TCGA database (Figure S1A); and even could not be found in meta-analysis of ACE2 expression using the combined in-house RNA-Seq dataset, TCGA and GEO datasets (Figure S1B). Therefore, we used TMT-based global proteomics to evaluate the protein expression of ACE2 in paired colorectal clinical tissues. The protein expression of ACE2 was higher in colorectal tumor tissues than in paired normal tissues (Figure 1B), which was also supported by immunohistochemical images that high staining and strong intensity of ACE2 existed in colon and rectum tumor tissues (Figure 1C). Subsequently, we analyzed the mRNA expression of TMPRSS2 and FURIN between colorectal cancer tissues and normal tissues via a meta-analysis of the RNA-Seq results and TCGA and GEO datasets. As shown in Figure 1D-E, the expression of both TMPRSS2 and FURIN was significantly decreased in colorectal cancer tissues compared with normal tissues. The immunohistochemical images of TMPRSS2 and FURIN staining also revealed that TMPRSS2 and FURIN were downregulated in colon tumor tissues (Figure 1F-G). Additionally, the heat map illustrated that these two genes were co-lower expressed in colorectal cancer tissues compared to paired normal tissues (Figure 1H). Furthermore, the expression levels of TMPRSS2 and FURIN gradually decreased along with malignant progression in the course of normal, adenoma and tumor (Figure 1I-L).
Given that ACE2 cooperating with co-factors TMPRRS2 and FURIN, plays key roles in regulating the binding and entry affinity of SARS-CoV-2, we further conducted Spearman’s correction analysis between ACE2 and its co-factors. As shown in Figure S1C-D, the expression of ACE2 was positively correlated with TMPRRS2 and negatively corrected with FURIN. A positive correlation was also detected between the expression levels of the two co-factors (Figure S1E). Therefore, we hypothesized that three genes were co-expressed and interacted in colorectal tumor tissues.
We then examined the role of onset age in the expression of these three genes. Higher ACE2 expression and lower FURIN expression were significantly related to late-onset colorectal cancer, while no association was found between onset age and TMPRSS2 expression (Figure S2). We also found no significant differences in the tumor site or tumor stage related to the expression of these three genes (Figure S2).
RNA and protein expression of ACE2 and its co-factors in pan-cancer tissues
Based on the GEPIA2 and HPA databases, we compared the mRNA and protein expression of ACE2 and its co-factors in different tissues. Notably, FURIN mRNA showed high abundance in most types of tissues compared with that of ACE2 and TMPRSS2 (Figure S3). However, the protein immunoreactivity of TMPRSS2 in most types of tumor tissues was generally weaker or negative compared with that of ACE2 and FURIN (Figure S4). ACE2 protein expression was observed to show moderate to strong cytoplasmic and membranous immunoreactivity in colorectal and renal cancers.
Co-expression pattern of ACE2 and its co-factors in colorectal cancer cells determined by single-cell RNA-Seq analysis
We next carried out single-cell RNA-Seq to evaluate the expression pattern of ACE2 and its co-factors in colorectal cancer tissues. As shown in Figure 2A-D, among colorectal tumor cell clusters, the expression of ACE2 and TMPRSS2 was heterogeneous, with some cells co-expressing both ACE2 and TMPRSS2 (9.33%), while some tumor cell clusters expressed only ACE2 (3.2%) or only TMPRSS2 (31.2%). Similarly, 15.2% of colorectal tumor cells co-expressed both ACE2 and FURIN, while 9.87% and 2.67% of cell clusters expressed only ACE2 or only TMPRSS, respectively (Figure 2E-H). These results provided the co-expression pattern of ACE2 and the co-factors in the colorectal tumor cells.
The infection rate of SARS-CoV-2 in colon cells
To evaluate the infection rate of SARS-CoV-2 in colon cell lines, we first tested the ACE2 expression level of normal colon epithelial cells (FHC and HCoEpiC) and colon tumor cells (SW620 and LoVo). None of these cells showed ACE2 expression that was detectable by Western blotting. Therefore, cell lines with stable overexpression of ACE2 were cultured, and the transduction efficiency was confirmed by GFP fluorescence and Western blotting (Figure 3A-B). Subsequently, we transfected the SARS-CoV-2 pseudovirus into cells and performed luciferase reporter assays. We found that relative luciferase activity was significantly increased in colorectal tumor cells compared with normal colon epithelial cells (Figure 3C). Finally, higher positive staining of SARS-CoV-2 was detected in colorectal tumor cells by immunofluorescent staining, while lower positive staining was observed in normal colon epithelial cells (Figure 3D). For the first time, we identified a higher infection rate of SARS-CoV-2 in ACE2-positive colorectal tumor cells than in normal colon epithelial cells.
Association of ACE2 and co-factor expression with immune infiltration in colorectal cancer tissues
Given the association of crucial gene expression with immune infiltration levels in colorectal cancer, we investigated whether ACE2 and its co-factors were associated with immune infiltration in different types of immune cells by using the TIMER algorithm (Table S2). We observed a significant negative association of tumor purity with FURIN (r = -0.180, P = 2.58 × 10-4), but no significant association of tumor purity with ACE2 or TMPRSS2 was found (Figure 4). Interestingly, we found that ACE2 expression presented a significant negative association with the infiltrating levels of neutrophils (r = -0.115, P = 2.11 × 10-2) and dendritic cells (r = -0.097, P = 5.16 × 10-2) in colon cancer tissues (Figure 4). Moreover, the expression of TMPRRS2 and FURIN presented a significant positive association with the infiltrating levels of neutrophils (r = 0.124, P = 1.29 × 10-2 in TMPRSS2; r = 0.225, P = 5.12 × 10-6 in FURIN) and dendritic cells (r = 0.121, P = 1.53 × 10-2 in TMPRSS2; r = 0.288, P = 4.20 × 10-9 in FURIN) in colon cancer tissues (Figure 4). These findings revealed that ACE2 and its co-factors play specific roles in immune infiltration in colorectal cancers, especially in of neutrophil and dendritic cells.
Prognostic analysis of ACE2 and its co-factors in colorectal cancer tissues
To explore the prognostic value of ACE2 and its co-factors in colorectal cancer, we performed Kaplan-Meier analysis based on the TCGA database. We found that low expression of ACE2 was significantly associated with a poor overall survival time (P = 0.028) (Figure 5A), while no significant association was observed between the overall survival time and the expression levels of TMPRSS2 (P = 0.073) or FURIN (P = 0.220) (Figure 5B-C). In addition, the overall survival time of patients in the higher-risk-score group was obviously longer than that of patients in the low risk score group for the combination of ACE2 and FURIN (P = 0.046) (Figure 5D). However, we only detected a weak association between a longer overall survival time and the higher risk score group for the combination of ACE2 and TMPRSS2 (P = 0.069) (Figure 5E) or the combination of three genes (P = 0.058) (Figure 5F). We also visualized the association between survival status and risk score rank as well as overall survival (Figure 5G-I).
Association of genetic variants in ACE2 and its co-factors with colorectal cancer risk and survival
Given that genetic variants in cancer-related genes are associated with colorectal cancer, we sought to investigate the genetic effect of SNPs in ACE2 and its co-factors on colorectal cancer susceptibility and prognosis. A summary of the experimental design and workflow is shown in Figure S5. A total of 19 candidate SNPs in ACE2 and its co-factors were included in the genotyping analysis after quality control and function prediction. Among these SNPs, no significant association was detected between candidate SNPs and the susceptibility of colorectal cancer (Table S3).
We then used 540 patients with follow-up information to confirm the genetic effect of candidate SNPs in predicting the prognosis of colorectal cancer (Table S4). Only two SNPs in the ACE2 gene were significantly associated with colorectal cancer survival time (HR = 1.40, P = 0.017 for rs2106809 and HR = 1.34, P = 0.038 for rs2285666) (Figure 6A-B). We also combined these two SNPs according to the number of risk alleles. As shown in Figure 6C, colorectal cancer patients with 1-4 risk alleles presented significantly poorer overall survival than those with 0 risk alleles (P < 0.05). Due to these two SNPs located in the ACE2 gene, we performed the functional annotation of rs2106809 and rs2285666 and the SNPs showing high linkage disequilibrium (rs4646142 and rs184697926) based on the Encyclopedia of DNA Elements (ENCODE) and Roadmap Epigenome Projects. The region encompassing these four SNPs was predicted in silico to present possible enhancer activity (Figure 6D), which indicated that genetic variants in the ACE2 gene could be useful in prognosis prediction for colorectal cancer.
Somatic mutation patterns of ACE2 and its co-factors in colorectal cancer tissues
We extracted mutational signatures from the TCGA database to identify somatic mutation patterns in ACE2 and its co-factors in colorectal cancer patients. As demonstrated in Figure S6A and Table S5, missense mutations in ACE2 (73.33%), FURIN (75.00%) and TMPRSS2 (83.33%) are particularly common in colorectal cancer tissues. Interestingly, we observed a significant increase in the mutation frequency of the ACE2 gene in the early-onset colorectal cancer group compared with the late-onset colorectal cancer group (P = 0.023) (Table S6). We also detected a decreasing trend in ACE2 expression in patients with ACE2 mutations (P = 0.059) (Figure S6B).
TMB is an emerging biomarker for the immunotherapy response of cancers. In this study, a higher TMB was detected in individuals with ACE2 mutations compared with those without ACE2 mutations (P = 3.32 × 10-5) (Figure S6C). We also found a significant decrease in TMB associated with increased ACE2 expression (r = -0.149, P < 0.001) (Figure S6D). Collectively, these results highlight the important roles of somatic mutations in the ACE2 gene and TMB in colorectal cancer progression.