To systematically investigate the candidate functional coding variants in ACE2 and the allele frequency (AF) differences between populations, we analyzed all the 1700 variants (Supplementary Table S1) in ACE2 gene region from the ChinaMAP (China Metabolic Analytics Project, under reviewing) and 1KGP (1000 Genomes Project)9 databases. The AFs of 62 variants located in the coding regions of ACE2 in ChinaMAP, 1KGP, and other large-scale genome databases were summarized (Supplementary Table S2). All of the 32 variants potentially affecting the amino acid sequence of ACE2 in databases were shown (Fig. 1a). Previous study showed that the residues near lysine 31, and tyrosine 41, 82–84, and 353–357 in human ACE2 were important for the binding of S-protein in coronavirus5. The mutations in these residues were not found in different populations in our study. Only a singleton truncating variant of ACE2 (Gln300X) was identified in the ChinaMAP (Fig. 1a). These data suggested that there was a lack of natural resistant mutations for coronavirus S-protein binding in populations. The effects of low-frequency missense variants in populations for S-protein binding could be further investigated. The distributions of seven hotspot variants (Lys26Arg, Ile468Val, Ala627Val, Asn638Ser, Ser692Pro, Asn720Asp, and Leu731Ile/Leu731Phe) in different populations were shown (Fig. 1b). Six low-frequency loci (rs200180615, rs140473595, rs199951323, rs147311723, rs149039346, and rs73635825) were found to be specific in 1KGP database, the AFs of which were also low in the gnomAD and TopMed10 database. Only two of these six variants (rs200180615 and rs140473595) could be found in CHB (Han Chinese in Beijing) population with the AF < 0.01. Interestingly, the SNP rs2285666 with the highest AF in the 62 variants exhibited much higher AF in the ChinaMAP (0.556) and CHS (Han Chinese South, 0.557) populations compared to others (AMR, Ad Mixed American, 0.336; AFR, African, 0.2114; EUR, European, 0.235). In addition, the homozygous mutation rate in males (0.550) was much higher than females (0.310) in the Chinese population (Supplementary Table S2). Taken together, the differences in AFs of ACE2 coding variants among different populations suggested that the diverse genetic basis might affect ACE2 functions among populations.
Fig. 1 The coding-region variants and eQTL variants for ACE2 in East Asian and other populations.
a Schematics of 32 coding variants in ACE2 identified in the ChinaMAP and 1KGP databases. Yellow stars indicate the nonsense variants; dots indicate the missense variants. The number of samples with hotspot variants was marked. b The distribution of hotspot missense mutations of ACE2 in different populations. The colors indicate different populations. c The distribution and the allele frequencies of representative eQTL variants for ACE2 in different populations. Pie charts depict the allele frequencies of an intron variant of ACE2 (rs4646127) in the world. Orange color denotes the frequency of alteration allele, and blue color denotes the reference allele. The allele frequencies of 15 eQTLs for ACE2 gene are shown in tables. The color gradient from blue to red indicates the increasing of allele frequencies. The allele frequencies of INDEL variant rs200781818 were annotated by the gnomAD database. EAS, East Asian; EUR, European; AFR, African; SAS, South Asian; AMR, Ad Mixed American.