Pathways and gene ontology associated with ACE2 expression
We identified highly enriched KEGG pathways in pan-tissue having high ACE2 expression levels (upper third) versus low ACE2 expression levels (bottom third) (Student’s t-test, adjusted P-value FDR < 0.05, fold change > 2) by GSEA (16) with a threshold of FDR < 0.05. These pathways were mainly involved in immune, stromal signature, metabolism, cell growth and proliferation, cancer, and other diseases (Fig. 1A). The immune-related pathways included cytokine-cytokine receptor interaction, leukocyte transendothelial migration, complement and coagulation cascades, TGF-β signaling, hematopoietic cell lineage, Jak-STAT signaling, adipocytokine signaling, chemokine signaling, viral myocarditis, intestinal immune network for IgA production, systemic lupus erythematosus, pathogenic Escherichia coli infection, epithelial cell signaling in Helicobacter pylori infection, and NOD-like receptor signaling. The stromal signature-related pathways included focal adhesion, ECM-receptor interaction, cell adhesion molecules, tight junction, adherens junction, regulation of actin cytoskeleton, axon guidance, and gap junction. The metabolism-related pathways included PPAR signaling, insulin signaling, arachidonic acid metabolism, drug metabolism-cytochrome P450, glutathione metabolism, retinol metabolism, fatty acid metabolism, tyrosine metabolism, metabolism of xenobiotics by cytochrome P450, glycolysis/gluconeogenesis, glycerophospholipid metabolism, phenylalanine metabolism, butanoate metabolism, and glycerolipid metabolism. The cell growth and proliferation-related pathways included MAPK, ErbB, p53, Wnt, VEGF, Notch, and mTOR signaling. The cancer-related pathways included pathways in cancer, small cell lung cancer, bladder cancer, melanoma, prostate cancer, glioma, acute and chronic myeloid leukemia, basal cell carcinoma, thyroid cancer, pancreatic cancer, renal cell carcinoma, and endometrial cancer, and the other diseases-related pathways included dilated cardiomyopathy, hypertrophic cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy, and prion diseases.
WGCNA generated 11 gene modules (indicated in green-yellow, cyan, yellow, light green, magenta, red, salmon, green, grey, black, and turquoise color, respectively) that were more highly enriched in the high-ACE2-expression-level than in the low-ACE2-expression-level pan-tissue (Fig. 1B). In contrast, two gene modules (indicated in blue and light-yellow color, respectively) were more highly enriched in the low-ACE2-expression-level pan-tissue (Fig. 1B). The gene ontology (GO) terms in the highly enriched gene modules in the high-ACE2-expression-level pan-tissue were mainly associated with immune response, cell cycle, reproductive process, brush border assembly, skin development, thyroid hormone metabolic process, muscle system process, cell projection organization, muscle contraction, blood vessel development, and cell growth and proliferation. The GO terms in the highly enriched gene modules in the low-ACE2-expression-level pan-tissue were mainly associated with nervous system development and muscle system process. The positive associations of ACE2 expression with immune response, cell cycle, and cell growth and proliferation in pan-tissue were consistent with the pathway analysis results.
Gene co-expression networks of ACE2
We found 2,983 and 74 genes having a significant positive and a significant negative expression correlation with ACE2 in pan-tissue, respectively (Pearson correlation coefficient |r| > 0.3) (Supplementary Table S1). Interestingly, when we analyzed female and male pan-tissue individually, we found that 3, 940 (or 587) and 87 (or 93) genes having a significant positive and a significant negative expression correlation with ACE2 in female (or male) pan-tissue, respectively (|r| > 0.3) (Supplementary Tables S2&S3). It indicates that more genes have a significant positive expression correlation with ACE2 in females than in male pan-tissue. Strikingly, we found 77 genes showing a significant positive expression correlation with ACE2 in females (r > 0.5) but a negative expression correlation with ACE2 in males (r < 0) (Supplementary Table S4). Fig. 2 shows 25 genes having the strongest positive and negative expression correlation with ACE2 in pan-tissue, female pan-tissue, and male pan-tissue (17). Notably, the gene encoding SLC6A19 (solute carrier family 6 member 19), which interacts with ACE2 (5), had a significant positive expression correlation with ACE2 in pan-tissue, female pan-tissue, and male pan-tissue (|r| > 0.3).
Associations of ACE2 expression and immune signatures with the expression of sex hormone receptor genes
Females have a lower disease severity and mortality risk than males infected with SARS-CoV-2 (18, 19). A potential explanation is a different host immune response to SARS-CoV-2 infection between females and males (8, 14, 20, 21). We analyzed the correlations between the expression levels of estrogen and androgen receptor genes (ESR1, ESR2, and AR) and ACE2 expression levels in pan-tissue and found that the correlations were consistently positive (Pearson’s correlation test, FDR < 1.0 × 10-60) (Fig. 3A). Interestingly, we found that the expression levels of ESR1 and ESR2 were positively associated with the enrichment levels of immune cells (B cells, CD8+ T cells, and NK cells) (Pearson’s correlation test, FDR < 1.0 × 10-10) (Fig. 3B). In contrast, the expression levels of AR inversely correlated with the enrichment levels of these immune cells (FDR < 1.0 × 10-40) (Fig. 3B). We obtained similar results in many individual tissues, including the blood vessel, breast, cervix uteri, colon, esophagus, pituitary, skin, small intestine, stomach, testis, and thyroid (Fig. 3C). These results suggest that females are likely to have a more robust immune defense system against SARS-COV-2 than males.