Pan-Cancer Analysis Reveals that the SARS-CoV-2 Receptor ACE2 is a Protective Factor for Cancer Progression

Background: The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has infected more than 13 million people and has caused more than 570,000 deaths worldwide as of July 13, 2020. The SARS-CoV-2 human cell receptor ACE2 has recently received extensive attention for its role in SARS-CoV-2 infection. Many studies have also explored the association between ACE2 and cancer. However, a systemic investigation into associations between ACE2 and oncogenic pathways, tumor progression, and clinical outcomes in pan-cancer remains lacking. Methods: Using cancer genomics datasets from the Cancer Genome Atlas (TCGA) program, we performed computational analyses of associations between ACE2 expression and antitumor immunity, immunotherapy response, oncogenic pathways, tumor progression phenotypes, and clinical outcomes in 12 cancer cohorts. We also identied co-expression networks of ACE2 in cancer. Results: ACE2 upregulation was associated with increased antitumor immune signatures and PD-L1 expression, and favorable anti-PD-1/PD-L1/CTLA-4 immunotherapy response. ACE2 expression levels inversely correlated with the activity of cell cycle, mismatch repair, TGF-β, Wnt, VEGF, and Notch signaling pathways. Moreover, ACE2 expression levels had signicant inverse correlations with tumor proliferation, stemness, and epithelial-mesenchymal transition (EMT). ACE2 upregulation was associated with favorable survival in pan-cancer and in multiple individual cancer types. Conclusions: ACE2 upregulation was associated with increased antitumor immunity and immunotherapy response, reduced tumor malignancy, and favorable survival in cancer, suggesting that ACE2 is a protective factor for cancer progression. Our data may provide potential clinical implications for treating cancer patients infected with SARS-CoV-2.


Background
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has infected more than 13 million people and has caused more than 570,000 deaths worldwide as of July 13, 2020 (https://coronavirus.jhu.edu/map.html). SARS-CoV-2 uses the angiotensin-converting enzyme 2 (ACE2) as a host cell receptor to infect humans [1,2,3,4]. ACE2 plays an important role in regulating cardiovascular and renal function [5]. This protein has recently received extensive attention for its role in SARS-CoV-2 infection [1,2,4]. Our recent study revealed that ACE2 is expressed in various human tissues [6], suggesting that SARS-CoV-2 may invade various human organs besides the lungs. Also, many studies have investigated the association between ACE2 and cancer [7][8][9][10][11][12][13]. For example, Yu-Jun et al. analyzed ACE2 expression in various cancers and revealed a positive association between ACE2 expression and survival prognosis in liver cancer [7]. Cai et al. described the genetic alteration, mRNA expression, and DNA methylation of ACE2 in over 30 cancer types and revealed genetic and epigenetic variations of ACE2 in various cancers [8]. Several studies demonstrated that ACE2 had antitumor effects by inhibiting tumor angiogenesis [10,11,13]. A recent study [14] showed that ACE2 expression was associated with increased tumor immune in ltration and was a positive prognostic factor in uterine corpus endometrial and renal papillary cell cancers. Nevertheless, a systemic investigation into the association between ACE2 expression and antitumor immunity, oncogenic pathways, tumor progression phenotypes, and clinical outcomes in pan-cancer remains lacking.
In this study, we investigated associations between ACE2 expression and antitumor immune signatures in 12 human cancer cohorts from the Cancer Genome Atlas (TCGA) program (https://cancergenome.nih.gov/). We also explored associations between ACE2 expression and multiple tumor phenotypes, including cell proliferation, stemness, epithelial-mesenchymal transition (EMT), oncogenic signaling, and clinical outcomes in these cancer cohorts. We also investigated the association between ACE2 expression and immunotherapy response in four cancer cohorts receiving the immune checkpoint blockade therapy. This study aimed to provide new insights into the association between ACE2 and cancer and the potential association between cancer and SARS-CoV-2 infection.
Evaluating the enrichment levels of immune signatures, pathways, and tumor phenotypes We evaluated the enrichment level of a pathway or tumor phenotype in a tumor sample by the singlesample gene-set enrichment analysis (ssGSEA) score [19]. The gene set included all marker genes of a pathway or tumor phenotype. A total of six cancer-associated pathways (cell cycle, mismatch repair, TGFβ, Wnt, VEGF, and Notch signaling) and three tumor phenotypes (cell proliferation, stemness, and EMT) were analyzed. We presented the marker genes of these pathways and tumor phenotypes in Supplementary

Gene-set enrichment analysis
We de ned high-ACE2-expression-level (upper third) and low-ACE2-expression-level (bottom third) tumors in each cancer type based on ACE2 expression pro les. We identi ed the KEGG [20] pathways highly enriched in both groups of tumors using GSEA [21] with a threshold of adjusted p-value < 0.05. Moreover, we used WGCNA [22] to detect the gene modules (gene ontology) differentially enriched between the highand low-ACE2-expression-level tumors in pan-cancer. We identi ed the hub genes as the genes connected to at least 5 other genes with a connectedness weight greater than 0.25 in a gene module and built their co-expression network.

Statistical analysis
We used Spearman's correlation test to evaluate the correlation (ρ) of ACE2 expression levels with the enrichment levels of pathways or tumor phenotypes, which were not normally distributed. We used Pearson's correlation test to evaluate the correlation (r) of ACE2 expression levels with the ratios of immune signatures, which was the log2-transformed values of the ratios between the mean expression levels of all marker genes in immune signatures and was normally distributed. We used the Benjamini and Hochberg method [23] to calculate the FDR for adjusting for multiple tests. We compared overall survival (OS), disease-speci c survival (DSS), progression-free interval (PFI), and disease-free interval (DFI) between the high-and low-ACE2-expression-level tumors. We utilized Kaplan-Meier curves to display survival time differences and the log-rank test to evaluate the signi cance of survival time differences.
The R package "survival" was used to perform the survival analyses.

Results
Association of ACE2 expression with immune signatures and immunotherapy response in cancer GSEA [21] identi ed many immune-related pathways highly enriched in the high-ACE2-expression-level tumors at least 5 cancer types. These pathways included cytokine-cytokine receptor interaction, hematopoietic cell lineage, viral myocarditis, natural killer cell-mediated cytotoxicity, chemokine signaling, Jak-STAT signaling, primary immunode ciency, antigen processing and presentation, autoimmune thyroid disease, T cell receptor signaling, intestinal immune network for IgA production, B cell receptor signaling, systemic lupus erythematosus, Leishmania infection, NOD-like receptor signaling, and epithelial cell signaling in Helicobacter pylori infection (Fig. 1A). Moreover, we found that ACE2 expression levels positively correlated with the pro-/anti-in ammatory ratios in pan-cancer (Pearson's correlation test, r = 0.26, p = 3.31 × 10 − 74 ) and in 11 individual cancer types (adjusted p-value (FDR) < 0.05) (Fig. 1B). This suggests that ACE2 expression has a stronger positive association with the proin ammatory signature than the anti-in ammatory signature in these cancer types. Altogether, these results suggest a prominent positive association between ACE2 expression and antitumor immune signatures in cancer. We found that ACE2 had a positive expression correlation with PD-L1 in pan-cancer and in 6 individual cancer types (FDR < 0.05) (Fig. 1C). We expected that the ACE2 expression would have a positive association with the response to anti-PD-1/PD-L1/CTLA-4 immunotherapy. We con rmed the anticipation in four cancer cohorts receiving immune checkpoint blockade therapy. In these cohorts, the high-ACE2-expression-level (> median) tumors displayed a higher rate of immunotherapy response than the low-ACE2-expression-level (< median) tumors (67% versus 17%, 80% versus 40%, 40% versus 20%, and 46% versus 25% for Nathanson (melanoma), Topalian (melanoma), Ascierto (renal cell carcinoma), and Snyder (bladder cancer) cohorts, respectively) (Fig. 1D). As a result, the former had better overall survival (OS) than the latter in the Nathanson cohort, which had related data available (log-rank test, p = 0.036) (Fig. 1E). These results suggest that the ACE2 expression is likely to be a positive predictor for anti-PD-1/PD-L1 immunotherapy.

Association of ACE2 expression with oncogenic pathways and tumor phenotypes in cancer
We quanti ed the activity of a pathway using the single-sample gene-set enrichment analysis (ssGSEA) [19] score of the set of genes included in the pathway. We found that ACE2 expression levels inversely correlated with the activity of cell cycle, mismatch repair, TGF-β, Wnt, VEGF, and Notch signaling pathways in 9, 6, 9, 7, 6, and 7 individual cancer types, respectively (Spearman's correlation test, FDR < 0.05) ( Fig. 2A). Moreover, we found that ACE2 expression levels had a signi cant inverse correlation with the expression levels of MKI67, which is a tumor proliferation index marker, in pan-cancer and 8 individual cancer types (Pearson's correlation test, FDR < 0.05) (Fig. 2B). Tumor stemness represents a stem cell-like tumor phenotype associated with tumor progression, metastasis, immune evasion, and drug resistance.

Identifying interaction networks of ACE2 in cancer
We identi ed 200 and 24 genes having marked positive and negative expression correlations with ACE2 in pan-cancer, respectively (|r| > 0.5) (Fig. 3A). WGCNA [22] identi ed four gene modules (indicated in yellow, red, pink, and turquoise color, respectively) highly enriched in the high-ACE2-expression-level tumors and three gene modules (indicated in black, blue, and green color, respectively) highly enriched in the low-ACE2-expression-level tumors (Fig. 3B). The GO terms highly enriched in the high-ACE2expression-level tumors mainly included immune response, induction of bacterial agglutination, regulation of microvillus length, and epidermal cell differentiation. In contrast, the GO terms highly enriched in the low-ACE2-expression-level tumors mainly included cell cycle, nervous system process, and microtubule-based process (Fig. 3B). Again, these results indicate that ACE2 expression has a signi cant positive association with antitumor immune response and a signi cant negative association with the cell cycle in cancer, suggesting the protective role of ACE2 from cancer progression.

Discussion
We investigated the association of ACE2 expression with immune signatures, oncogenic pathways, and tumor phenotypes in diverse cancer cohorts. Our results indicate that ACE2 is a protective factor for cancer progression. In particular, the ACE2 downregulation correlates with worse survival and tumor advancement in KIRC, also known as clear cell renal cell carcinoma (ccRCC). Previous studies demonstrated that ACE2 exerts antitumor effects by inhibiting tumor angiogenesis [10] and promoting tumor immune in ltration [14]. Our results are consistent with these previous ndings. Besides, we found that ACE2 upregulation was associated with reduced cell proliferation, stemness, and EMT, as well as the downregulation of oncogenic pathways, such as cell cycle, mismatch repair, TGF-β, Wnt, and Notch signaling. Moreover, we found that ACE2 had a negative expression correlation with PD-L1, an immunosuppressive molecule, and a predictive marker for an active response to immune checkpoint inhibitors. As a result, ACE2 upregulation correlates with a favorable response to anti-PD-1/PD-L1/CTLA-4 immunotherapy.
ACE2 also plays a protective role in hypertension and heart disease [27]. Moreover, ACE2 de ciency may exacerbate outcomes in patients with SARS-CoV-2 infection [27]. Indeed, a recent study showed that ACE2 was downregulated in virus-infected lung tissue [14], indicating a potential protective role of ACE2 in patients with SARS-CoV-2 infection. Thus, using ACE2 inhibitors for preventing and treating SARS-CoV-2 infections may not be an advisable strategy for individuals with hypertension, heart disease, or cancers.

Conclusions
ACE2 upregulation was associated with increased antitumor immunity and immunotherapy response, reduced tumor malignancy, and favorable survival in cancer, suggesting that ACE2 is a protective factor for cancer progression. Our data may provide potential clinical implications for treating cancer patients infected with SARS-CoV-2.
ZZ performed data analyses and helped in manuscript preparation. LL performed data analyses and helped in manuscript preparation. ML performed data analyses and helped in manuscript preparation. XW conceived the study, designed analysis strategies, and wrote the manuscript. All the authors read and approved the nal version of the manuscript.   Interaction networks of ACE2 in cancer. A. 200 and 24 genes having marked positive and negative expression correlations with ACE2 in pan-cancer, respectively (|r| > 0.5). B. Gene modules (gene ontology) enriched in high-ACE2-expression-level and low-ACE2-expression-level pan-cancer. C. Co-expression