ACE2 gene variants may underlie interindividual variability and susceptibility to COVID-19 in the Italian population

COVID-19, caused by the novel SARS-CoV-2 coronavirus, rapidly became a global pandemic with wide interindividual variability in clinical severity, from asymptomatic infection to severe bilateral interstitial pneumonia and acute respiratory distress. While age and comorbidities influence risk, notable variability among younger individuals suggests host genetic factors may modulate susceptibility and disease severity. Prior work shows SARS-CoV-2 uses ACE2 as the cellular receptor and requires host proteases (e.g., TMPRSS2) for spike protein activation and viral entry. Specific ACE2 residues are implicated in spike receptor-binding domain interaction, and proteolytic cleavage of ACE2 can enhance viral entry. Given ACE2’s location on the X chromosome and variable X-inactivation across tissues, genetic variation within ACE2 may contribute to differences in infection susceptibility or severity. This study investigates ACE2 variation in the Italian population using large-scale whole-exome sequencing data, and evaluates whether specific variants may influence protein stability, spike binding, and clinical outcomes in COVID-19.

The study builds on evidence that SARS-CoV-2 shares high sequence homology with SARS-CoV and uses ACE2 as its entry receptor. Structural and modeling studies have identified key ACE2 residues involved in binding of the SARS-CoV-2 spike protein receptor-binding domain, and host proteases such as TMPRSS2 facilitate entry by cleaving ACE2 and activating spike. Variability in these interaction interfaces or cleavage-adjacent regions may therefore alter viral binding or internalization. Differences in ACE2 allele frequencies have been noted across populations, raising the possibility of population-specific genetic effects on COVID-19 susceptibility and severity.

• Cohorts and ethics: The Network of Italian Genomes (NIG) provided whole-exome sequencing (WES) data from 6930 unrelated, healthy Italian individuals (4171 males, 2759 females) recruited across five centers, with informed consent and ancestry verified through parental/grandparental origin. DNA was stored in the Telethon Network of Genetic Biobanks. A separate COVID-19 case cohort included 131 Italian patients (34 females, 97 males; median age 64 years, range 31–98) from the GEN-COVID Multicenter Study, with ethical approval and consent. Patients were categorized into four clinical severity groups by respiratory support needs (invasive ventilation, noninvasive ventilation/high-flow oxygen, conventional oxygen, no oxygen). A control cohort of 258 Italian individuals (129 males, 129 females) was used for case–control comparisons.
• Sequencing and variant calling: Targeted exome capture used Agilent SureSelect Human All Exon (V4/V5/V6/V7), Agilent Clinical Research Exome (V1/V2), Illumina Nextera Rapid Capture v1.2, TruSeq Exome Targeted Regions, or TruSight One Expanded V2. Sequencing platforms included Illumina Genome Analyzer, HiSeq2000, NextSeq500/550, and NovaSeq6000; some WES was outsourced (BGI; Mount Sinai; Broad Institute). Reads were aligned to hg19; variant calling and annotation followed in-house pipelines aligned with GATK Best Practices and used Annovar and VEP. gnomAD was used for population allele frequencies. Mean ACE2 exon coverage was 55x; variants with depth <20x were filtered per ASHG germline guidelines. Identified variants were submitted to LOVD (variant IDs provided in the paper).
• Computational analyses: The ACE2 structure (PDB 1R42) was used. DUET predicted effects of missense substitutions on protein stability. Molecular dynamics (MD) simulations (GROMACS 2019.3) were performed for selected variants and wild-type to assess structural stability via RMSD and motion correlations. Systems were solvated (TIP3P), neutralized with Na+/Cl−, energy-minimized (steepest descent, tolerance 1000 kJ mol−1 nm), heated to 310 K (1 ns, NVT), equilibrated (10 ns, NPT; V-Rescale thermostat; Berendsen barostat), then simulated for 100 ns for analysis. Graphs used XMGrace; structures visualized in PyMOL. High-performance computing resources included multi-node CPU/GPU infrastructure.
• Statistical comparison: ACE2 WES data from 131 patients and 258 controls were compared to evaluate allelic variability differences; significance threshold reported with P < 0.029.

• From 6930 Italian WES (9689 ACE2 alleles), three more common missense variants were identified: p.(Asn720Asp) [c.2158A>G] at frequency 0.011 (103/9689) in Italians (gnomAD overall 0.016; European non-Finnish 0.025; not reported in East Asians assessed), p.(Lys26Arg) [c.77A>G] at 0.0011 (vs European non-Finnish 0.0058; East Asian 6×10−5), and p.(Gly211Arg) [c.631G>A] at 0.0012 (vs European non-Finnish 0.0019; not reported in 14,822 East Asian exomes). These were predicted to influence ACE2 structure/stability; Asn720Asp lies near the TMPRSS2 cleavage region, potentially impacting internalization.
• Twenty-eight rare missense variants were observed, ten not previously reported in gnomAD, along with nine truncating variants also absent from gnomAD. Many truncating variants localized within the extracellular protease domain (which includes the receptor binding site), two in the neck (dimerization) domain, and one in the intracellular domain; all were very rare with no homozygous females detected, consistent with low tolerance for ACE2 loss-of-function.
• Specific missense changes predicted to be destabilizing included p.(Val506Ala), p.(Val209Gly), and p.(Gly377Glu). Two rare variants, p.(Leu351Val) and p.(Pro389His), were predicted to alter the hydrophobic core and induce conformational changes that could influence spike RBD interaction. p.(Pro389His) (European non-Finnish allele frequency 2.45×10−5) has not been reported in Asian populations and is predicted damaging by PolyPhen and deleterious by SIFT.
• MD simulations on selected variants (p.Val506Ala, p.Lys26Arg, p.Gly211Arg, p.Leu351Val, p.Pro389His) showed altered RMSD/flexibility patterns relative to wild-type, supporting predicted effects on ACE2 stability or spike-binding relevant regions.
• Case–control comparison revealed significantly higher ACE2 allelic variability in controls (n=258) than in COVID-19 patients (n=131), with P < 0.029, suggesting certain ACE2 variants may reduce susceptibility or severity.

The findings support the hypothesis that host genetic variation in ACE2 contributes to interindividual differences in susceptibility to SARS-CoV-2 infection and COVID-19 severity. Common variants enriched in Europeans and Italians compared to East Asians, including p.Asn720Asp near the TMPRSS2 cleavage region, may modulate ACE2 processing and viral entry. Rare variants predicted to alter ACE2 structural stability or the spike RBD interaction surface (e.g., p.Leu351Val and p.Pro389His) could impact the efficiency of viral binding and internalization. The observation of greater ACE2 allelic variability among controls than among patients suggests a potential protective effect of certain ACE2 variants, aligning with the mechanistic predictions from structural modeling and MD simulations. These genetic insights may help explain part of the heterogeneous clinical outcomes seen across individuals and populations and point toward the utility of genetic risk assessment in guiding personalized preventive or therapeutic strategies.

This study systematically characterized ACE2 coding variation in a large Italian cohort and identified both common and rare variants with predicted effects on protein stability and spike interaction. A significant excess of ACE2 allelic variability in controls compared with COVID-19 patients supports a role for ACE2 variation in modulating disease susceptibility/severity. The work contributes to understanding host genetic influences on COVID-19 and suggests that ACE2 genotyping could inform risk stratification and individualized approaches. Future research should include functional validation of prioritized variants, larger multi-ethnic case–control and cohort studies, integration with other host genes involved in viral entry (e.g., TMPRSS2), and exploration of regulatory variants and X-inactivation effects across relevant tissues.

• Functional effects were inferred primarily from in silico prediction (DUET, PolyPhen/SIFT) and MD simulations of selected variants; experimental validation in cellular or animal models was not presented.
• The case–control comparison involved a modest patient sample size (n=131) and a limited control set (n=258) relative to the discovery cohort, which may affect power and generalizability.
• Analyses were based on exonic variation; regulatory, non-coding, and structural variants were not assessed. Use of hg19 and heterogeneous capture kits/platforms may introduce technical variability, though filtered by coverage and standard pipelines.
• ACE2 is on the X chromosome and subject to variable X-inactivation across tissues, complicating genotype–phenotype interpretation, especially in females; tissue-specific expression/inactivation data were not integrated.
• Findings are specific to Italian/European populations; allele frequency and effect estimates may differ in other ancestries.