key: cord-295257-iguhy1z8
authors: Calcagnile, Matteo; Forgez, Patricia; Iannelli, Antonio; Bucci, Cecilia; Alifano, Marco; Alifano, Pietro
title: ACE2 polymorphisms and individual susceptibility to SARS-CoV-2 infection: insights from an in silico study
date: 2020-04-24
journal: bioRxiv
DOI: 10.1101/2020.04.23.057042
sha: 
doc_id: 295257
cord_uid: iguhy1z8

The current SARS covid-19 epidemic spread appears to be influenced by ethnical, geographical and sex-related factors that may involve genetic susceptibility to diseases. Similar to SARS-CoV, SARS-CoV-2 exploits angiotensin-converting enzyme 2 (ACE2) as a receptor to invade cells, notably type II alveolar epithelial cells. Importantly, ACE2 gene is highly polymorphic. Here we have used in silico tools to analyze the possible impact of ACE2 single-nucleotide polymorphisms (SNPs) on the interaction with SARS-CoV-2 spike glycoprotein. We found that S19P (common in African people) and K26R (common in European people) were, among the most diffused SNPs worldwide, the only two SNPs that were able to potentially affect the interaction of ACE2 with SARS-CoV-2 spike. FireDock simulations demonstrated that while S19P may decrease, K26R might increase the ACE2 affinity for SARS-CoV-2 Spike. This finding suggests that the S19P may genetically protect, and K26R may predispose to more severe SARS-CoV-2 disease.

In principle, any new infectious agent that challenges a totally susceptible population with little or no immunity against it is able to totally infect the population causing pandemics. Pandemics rapidly spread affecting a large part of people causing plenty of deaths with significant social disruption and economic loss. However, if we look at the even worst pandemics in the human history we can realize that ethnic and geographical differences in the susceptibility to disease actually exist, in spite of the infectious sources and transmission routes that are the same for all individuals 1 . The current SARS covid-19 (a shortened form of "coronavirus disease of 2019") epidemic spread appears to be similarly influenced by ethnical and geographical factors. After its initial spread in China, the pandemic is now progressing at an accelerating rate in Western Europe and the United States of America 2 . In these regions, the causative agent, the severe acute respiratory syndrome corona virus -2 (SARS-CoV-2) is spreading incredibly quickly between people, due to its newness -no one on earth has immunity to SARS Covid-19 -and transmission route. Yet, in the other regions of the world, the kinetics of diffusion and mortality seem less impressive, although the world has become highly interconnected as a result of a huge growth in trades and travels 2 .

A multitude of factors may concur to explain the ethnic and geographical differences in pandemic progression and severity, including cultural, social and economic inequality, as well as health care system organization, and climate also. Mostly, considerable individual differences in genetic susceptibility to diseases may be involved 3 . Genomic predisposition is a major concept in modern medicine, and understanding of molecular bases of genetic predisposition can help to find prevention and treatment strategies for the corresponding diseases 3 . In the SARS Covid-19, even subtle inter-individual genetic differences may affect both the SARS-CoV-2 viral life cycle and the host innate and acquired immune response.

SARS-CoV-2 is an enveloped positive-stranded RNA virus that replicates in the cytoplasm, and uses envelope spike projections as a key to enter human airway cells 4 . In coronaviruses spike glycoproteins, which forms homotrimers protruding from the viral surface, are a primary determinant of cell tropism, pathogenesis, and host interspecies transmission. Spike glycoproteins comprise two major functional domains: an N-terminal domain (S1) for binding to the host cell receptor, and a C-terminal domain (S2) that is responsible for fusion of the viral and cellular membranes 5 .

Following the interaction with the host receptor, internalization of viral particles into the host cells is accomplished by complex mechanisms that culminate with the activation of fusogenic activity of spike, as a consequence of major conformational changes that, in general, may be triggered by receptor binding, low pH exposure and proteolytic activation 5 .

In some coronaviruses spike glycoproteins are cleaved by furin, a Golgi-resident protease, at the boundary between S1 and S2 domains, and the resulting S1 and S2 subunits remain non-covalently bound in the prefusion conformation with important consequences on fusogenicity 5 . Notably, at variance with SARS-CoV and other SARS-like CoV Spike glycoproteins, SARS-CoV-2 Spike glycoprotein contain a furin cleavage site at the S1/S2 boundary, which is cleaved during viral biogenesis 6 , and may affect the major entry route of viruses into the host cell 5 .

Productive entry of coronaviruses that harbor non-cleaved Spike glycoproteins (such as SARS-CoV) rely on endosomal proteases suggesting that this entry is accomplished by hijacking the host endocytic machinery 5 . Indeed, it has been reported that SARS-CoV infection is inhibited by lysomotropic agents because of the inhibition of the low-pH-activated protease cathepsin L 7 .

However, SARS-CoV is also able to fuse directly to the cell membrane in the presence of relevant exogenous proteases, and this entry route is believed to be much more efficient compared to the endocytic route 8 . In fact, proteases from the respiratory tract such as those belonging to the transmembrane protease/serine subfamily (TMPRSS), TMPRSS2 or HAT (TMPRSS11d) are able to induce SARS-CoV spike glycoprotein fusogenic activity 9,10,11,12 . The first cleavage at the S1-S2 boundary (R667) facilitates the second cleavage at position R797 releasing the fusogenic S2' subdomain 5 . On the other hand, there is also evidence that cleavage of the ACE2 C-terminal segment by TMPRSS2 can enhance spike glycoprotein-driven viral entry 13 . Notably, it has been very recently demonstrated that also SARS-CoV-2 cell entry depends on TMPRSS2, and is blocked by protease inhibitors 14 .

SARS-CoV-2 and respiratory syndrome corona virus (SARS-CoV) Spike proteins share very high phylogenetic similarities (99%), and, indeed, both viruses exploit the same human cell receptor namely angiotensin-converting enzyme 2 (ACE2), a transmembrane enzyme whose expression dominates on lung alveolar epithelial cells 6, 15, 16 . This receptor is an 805-amino acid long captoprilinsensitive carboxypeptidase with a 17-amino acids N-terminal signal peptide and a C-terminal membrane anchor. It catalyzes the cleavage of angiotensin I into angiotensin 1-9, and of angiotensin II into the vasodilator angiotensin 1-7, thus playing a key role in systemic blood pressure homeostasis, counterbalancing the vasoconstrictive action of angiotensin II, which is generated by cleavage of angiotensin I catalyzed by ACE 17 Although ACE2 mRNA is expressed ubiquitously, ACE2 protein expression dominates on lung alveolar epithelial cells, enterocytes, arterial and venous endothelial cells, and arterial smooth muscle cells 18 .

There is evidence that ACE2 may serve as a chaperone for membrane trafficking of an amino acid transporter B0AT1 (also known as SLC6A19), which mediates the uptake of neutral amino acids into intestinal cells in a sodium dependent manner 19 . Recently, 2.9 Å resolution cryo-EM structure of full-length human ACE2 in complex with B0AT1 was presented, and structural modelling suggests that the ACE2-B0AT1 can bind two spike glycoproteins simultaneously 20, 21 . It has been hypothesized that the presence of B0AT1 may block the access of TMPRSS2 to the cutting site on ACE2 20, 21 . B0AT1 (also known as SLC6A19) is expressed with high variability in normal human lung tissues, as shown by analysis of data available in Oncomine from the work by Weiss et al 22 .

Notably, a wide range of genetic polymorphic variation characterizes the ACE2 gene, which maps on the X chromosome, and some polymorphisms have been significantly associated with the occurrence of arterial hypertension, diabetes mellitus, cerebral stroke, coronary artery disease, heart septal wall thickness and ventricular hypertrophy 23, 24, 25 . The association between ACE2

polymorphisms and blood pressure responses to the cold pressor test led to the hypothesis that the different polymorphism distribution worldwide may be the consequence of genetic adaptation to different climatic conditions 25, 26 . In this study we have used a combination of in silico tools to analyze the possible impact of ACE2 single-nucleotide polymorphisms (SNPs) on the interaction with SARS-CoV-2 Spike glycoprotein. Results seem to suggest that ACE2 polymorphism can contribute to ethnic and geographical differences in SARS COVID-19 spreading across the world. 27 , 6LZG (SARS-CoV-2 Spike RBD /ACE2 complex) 28 , 6M0J (SARS-CoV-2 Spike RBD /ACE2 complex) 29 , 6VW1 (chimeric SARS-CoV/SARS-CoV-2 Spike RBD /ACE2 complex) 30 , and 6M17 (SARS-CoV-2 Spike RBD /ACE2/B0AT1 complex) 20,21 (Fig. 1a) . ClustalO alignments of human ACE2 amino acid sequences, In all models, similar to SARS-CoV RBM, SARS-CoV-2 RBM forms a concave surface that houses a convexity formed by two helices on the exposed surface of ACE2. Strong network of Hbond and salt bridge interactions mediate the receptor-ligand binding. Global energy and several distinctive features of the 3D models with and without glycosylation are reported in Supplementary   Table 1 . Contact residues are classified as: "permanent" (predicted as binding residues in all 10 models), "stable" (predicted as binding residues in 6 or 7 out 10 models), "unstable" (predicted as binding residues in 6 or 7 out 10 models), "hyper-unstable" (1 or 2 models out of 10). Twenty-seven substitutions were predicted to influence the ACE2/Spike interface in at least one of the different 3D PDB models. Fifteen and seventeen were predicted to affect the ACE2/B0AT1 and ACE2/ACE2 interfaces, respectively. Some residues, which are described in UNIPROT database (https://www.uniprot.org/uniprot/Q9BYF1) as important for the interaction between spike and ACE2, are permanent contact residues (predicted as binding) but all of these are non-polymorphic ( Fig. 1b) . In contrast, polymorphic residues are stable, unstable or hyper-unstable. A list of 18 SNPs from dbSNP (S19P , I21T, I21V, E23K, A25T, K26E, K26R, T27A, E35D, E35K, E37K, S43R, E75G, M82I, G326E, E329G, G352V; D355N), which were predicted to affect the ACE2/Spike interface, was used for further analysis.

SNPs possibly affecting ACE2 glysosylation. Supplementary Table 1 illustrates amino acid glycosylation sites, and structure of the glycosidic chains as inferred from different ACE2/Spike complex PDB models. Putative polymorphic sites (Q60R, N103H, N546D, N546S) from dbSNP database that may affect ACE2 glycosylation are also reported. One of these amino acid variations, N546D, is rather common in South Asia (Supplementary Table 4 ).

FireDock 34 was used to estimate the effects of removal of glycosidic residues or chains on ACE2 interaction with SARS-CoV-2 Spike RBD by calculating ΔG values. The data indicated that removal of glycosidic chains results in either an increased or a decreased ΔG values, depending on the PDB model ( Fig. 1c ). In particular, removal of glycosidic moieties apparently strengthened the ACE2/Spike interaction in SARS-CoV Spike/ACE2 in the 2AJF model, while it appeared to weaken the interaction between SARS-CoV-2 Spike and ACE2 in the 6VW1 model. In both cases, the effect was mostly due to removal of the terminal beta-mannose (BMA) (Fig. 1c) , which was predicted to decorate a glycosidic chain attached to aspartic amino acid residue at position 90 that maps in a helix that is involved in the interaction with Spike, as shown in Fig. 1d . Noteworthy, in the 6VW1 model, BMA is involved in two H-bonds and one pseudo-bond (Fig. 1e ), and these bonds are lost in non-glycosylated models. In contrast, in the 2AJF model, the BMA forms only one H-bond, and after removal of terminal BMA, the Thr-41 acquires more grads for binding thereby strengthening the interaction with ACE2. These results seem to suggest that ACE2 glycosylation may play a different role in modulating the interaction with SARS-CoV Spike and SARS-CoV-2

Spike. were, among the most diffused SNPs worldwide, the only two SNPs that were able to potentially affect the interaction of ACE2 with SARS-CoV Spike and SARS-CoV-2 Spike (Supplementary Table 3 ). In particular, the S19P SNP is rather common in African people with a frequency about 0.3%, while K26R SNP is frequent in European people with a frequency about 0.5% (Supplementary Table 4 ).

FireDock 34 results indicated that the S19P substitution decreased the affinity of ACE2 with Spike in 2AJF and 6VW1 models (Fig. 2c ) and similar results were obtained with all other models.

Moreover, this amino acid substitution seems also to affect the ACE2 N-terminal cleavage site (Fig.   2d ), and when FireDock 34 simulations were carried out on ACE2 with the alternative cleavage site, the effects of S19P SNP was much more impressive (Fig. 2 c) .

In contrast, the K26R and the less common K26E substitutions appeared to increase the affinity of ACE2 with SARS-CoV-2 Spike (2AJF model), and slightly decrease the affinity of ACE2 with SARS-CoV Spike (6VW1 and 6M17) models (Fig. 3a) . As 6VW1 was generated with a chimeric SARS-CoV/SARS-CoV-2 Spike, to support our results we performed an additional simulation by challenging the ACE2 structure from 6VW1 with the Spike structures that were generated by the different models (Fig. 3b) , and the results confirmed those shown in Fig. 3a . Noteworthy, the receptor-ligand interactions was much weaker in 6M17 (SARS-CoV-2 Spike RBD /ACE2/B0AT1 complex) with respect to the other models, confirming an inhibitory function of B0AT1. However, in this model, at lower energy values, the effects of K26R/E substitutions were much more evident. shows that ACE2 (computed by using 3D PDB model 6M17 as input file) is characterized by a high deformation tract that is located immediately upstream of the transmembrane domain (Fig. 4b) ,

whereas the C-terminal tail is characterized by high fluctuation (Fig. 4c ). suggest that the hydrophobic domain alone is highly unstable in the membrane confirming that a chaperone is required for correct topology maintenance. This function was assigned to the moonlighting amino acid transporter B0AT1 19 .

To investigate dynamic properties of ACE2 globular head, the trans-membrane helix and conserved domains were firstly mapped on a 3D structure. Then, Dynamut 36 simulation was carried out on ACE2 by using 6WV1 PDB model (without the transmembrane domain) (Fig. 5ab) . Results indicate that some residues of the ACE2 interface, which are involved in the interaction with SARS-CoV-2 Spike glycoprotein can actually fluctuate (Fig. 5cd ).

Dynamic properties of SARS-CoV-2 and SARS-CoV spike proteins were also investigated. between the two helices and beta-sheet, and between the residues of the beta-sheet are illustrated in Supplementary Fig. 4d (Fig. 6a, Supplementary Fig.   5 and Supplementary Table 7 ).

All 197 amino acid residues that were reported as polymorphic in dbSNP were analyzed. In that women are probably more prone to infection but often present a less severe disease. Although higher incidence of cardiac, respiratory and metabolic co-morbidities are probably responsible for more severe form of infection in men, estrogen-induced upregulation of ACE2 expression would explain increased susceptibility of women to a less severe and often asymptomatic form of disease.

Furthermore, the ACE2 gene is located on Xp22, in an area where genes are reported to escape from X-inactivation, further explaining higher expression in females 45, 46 .

On the other hand, it has been hypothesized that, regardless of sex, pharmacological (antihypertensive drugs, such as ACE inhibitors and sartans) or environmental factors (NO2 pollution), capable of inducing an overexpression of ACE2 could be responsible of increased susceptibility to infection and/or greater severity 47 . ACE2 plays an essential role in the renin-angiotensinaldosterone system, and its loss of function due to the massive binding of viral particles and internalization could constitute an essential element of the pathophysiology of pulmonary and cardiac damage during COVID-19 infection 47, 48 . In this context it should be underlined that ACE2

probably plays a dual role in the dynamic of infection and disease course. While at beginning ACE2 overexpression may increase the entry of the virus into the cell and its replication, its consequent viral-induced loss of function results in an unopposed accumulation of angiotensin II that further aggravates the acute lung injury response to viral infection. Indeed, in the rodent blockade of the renin-angiotensin-aldosterone system limits the acute lung injury induced by the SARS-CoV-1 spike protein 49 , suggesting that if ACE2 function is preserved (because of increased baseline expression, as especially seen in pre-menopausal women), clinical course of infection might be less severe.

It has been suggested that polymorphisms in the ACE2 gene could reduce the spike affinity, with subsequent lower susceptibility to infection: in this hypothesis, their geographical / ethnical distribution could explain the strong discrepancies in infection rate and/or lethality observed worldwide 47 . Effectively, we showed by Network plot and Non-Metric Multidimensional Scaling that most of the SNPs diffused worldwide did not affect significantly the interaction of ACE2 with SARS-CoV-2 Spike. S19P was one of the rare polymorphisms able to potentially affect this interaction, by lowering the affinity. This polymorphism is more frequent in African populations, but its diffusion (0.3%) remains too low to explain, except in minimal part, the reduced death toll . This allowed the authors to predict that hamsters could be infected, which was experimentally confirmed -underlining the reliability of in silico modeling-and could be subsequently at the origin of inter-animal transmission. However, hamsters, although developing clinical signs of the infection and relative histopathological changes, did not die 50 : we speculate that lethality may be related to Spike/ACE2 affinity. On the other hand, the lower affinity in bat could explain -besides a better immune control-why these animals are carriers without dying.

In the same study, Chan and colleagues 50 So, if modestly SNP-determined lower affinity between spike and ACE2 does not seem to explain the differences in the distribution and lethality of the disease in humans, we hypothesize that the question can be addressed in a specular way: perhaps polymorphisms responsible for higher affinity can be responsible of higher severity of disease, especially when very high affinity receptors are overexpressed because of the above mentioned environmental and pharmacological factors. Obviously underlying diseases would contribute to an even more severe course of the disease, with an intense viral replication capable of infecting in turn a large number of persons, including some individuals with similar ACE2 polymorphisms, and so on. Our in silico models allowed us to identify K26R and K26E as SNPs with a possible increase in Spike/ACE2 affinity.

K26R SNP is relatively frequent in European people with a frequency about 0.5%, which would correspond to a potential target population of 2,230,000 people at the European Union level.

In addition to FireDock 34, 35 simulations that led to predict the possible effects of S19P, K26R

and K26E ACE2 SNPs, Dynamut 36 and ENCoM 42 tools were used to compare dynamic features of ACE2 and its polymorphic variants in order to analyze the possible indirect effects on binding interfaces of SNPs that are located outside these interfaces. SNPs I468V, V488A and A501T were identified as the most common SNPs that may produce these indirect effects in dynamic models.

Although the precise effects of these SNPs on the interaction between ACE2 and SARS-CoV-2 or SARS-CoV Spike proteins have to be determined in more detail, nevertheless, it is desirable to use dynamic modeling to unmask indirect effects of SNPs.

It seems necessary to confirm in vivo that, among patients with serious disease and/or fatal outcome, polymorphisms responsible for a very high Spike/ACE2 affinity are more frequent than among patients with less severe/asymptomatic disease or even than in general population.

Obviously, the impact of these polymorphisms on severity of outcome should be weighted by appropriate demographic and clinical factors. If these differences were confirmed, this would pave the way for the identification, on a population scale, of healthy individuals whose molecular phenotypes would be responsible for more serious disease. Apart from the usual social distancing measures, which could be reinforced for these cases, targeted drug prevention strategies could be evaluated. It could be logical to assess pharmacological prophylactic interventions, as proposed in categories of healthy people at particular risk of exposure such as care-givers. In particular, chloroquine, interfering with N-terminal glycosylation of ACE2, could lower its affinity for spike, thus representing an interesting candidate. In our in silico model, we found that removal of glycosidic moieties weakened the interaction between SARS-CoV-2 spike and ACE2. The serine protease inhibitor camostat mesylate, approved in Japan to treat unrelated diseases, has been shown to block TMPRSS2 activity 54, 55 and is thus another interesting candidate.

On the other hand, the identification of broader categories of people with lower risk of developing severe disease, could allow a safer exit from the lock-down phase, while facilitating the establishment of a faster herd immunity, and waiting reliable serological tests and, above all, effective vaccines.

On the basis of our in silico study we speculate that infection and mortality are 

Databases. 3D structures of proteins were downloaded from PDB (RCSB Protein Data Bank 56 ).

We focused our analysis on 2AJF for SARS-CoV 27 57, 58 was used to identity the ACE2 receptor SNPs, and to select the most diffused ones.

Functional information was acquired by UNIPROT database 59 . Chimera 60 was used as a tool for Image generation, 3D mapping, PDB managing and to analyze the results.

Binding interface characterization. The selected PDB models were analyzed by a structural point of view using Chimera software in order to identify the glycosylation sites and the secondary structures of proteins involved in the binding between ACE2 and Spike protein receptor binding domain (RBD). To estimate the effect of glycosylations we implemented a static model. Chimera was used to remove glycosydic residues, while FireDock 34,35 was used to compute the global energy scores between the native structures and the de-glycosylated models. On the other side, starting from the entire list of SNPs, SSIPe (EVOEF) 32 was used to identify the residues involved in the binding interfaces. A second step, which was carried out with SSIPe (SSIPe) 33 , was aimed at estimating the effects of single SNPs, and to generate mutant models. Different SNPs lists were obtained, which were compared, and used to identify the most stable binding amino acid residues.

SSIPe analysis performed with the PDB model 6M17 was used to map: the ACE2/Spike protein interaction interface, ACE2 /ACE2 dimerization interface, and B0AT1/ACE2 interaction interface.

The model contains ACE2 in the dimeric form (with the hydrophobic domains) and B0AT1, while in all the others models the transmembrane domains, ACE2 /ACE2 interface and B0AT1/ACE2

interface are absent. Spike models reported in this study as ligands. In a similar manner, other static models from SSIPe were used as models to estimate the variation in terms of free energy related to polymorphisms that map on the ACE2 dimerization interface and ACE2-B0AT binding interface. Legends to Figures   Fig. 1. 3D Oscillation plot of the distance between ACE2 amino acid residues mapping in the two helices that are involved in binding with Spike proteins. b 3D images of regions containing the amino acid residues shown in panel a. c Variance of the distance between ACE2 amino acid residues mapping in the two helices that are involved in binding with Spike proteins. d-e Oscillation plot of the distance between the two helices and beta-sheet (d), and between the residues of the beta-sheet (e).

f-g Variance of the distance of amino acid residues between the two helices and beta-sheet (f), and between the residues of the beta-sheet (g). showing the results of SNPs that mapping outside the ACE2 interfaces.

Pandemics: waves of disease, waves of hate from the Plague of Athens to A.I.D.S

Are patients with hypertension and diabetes mellitus at increased risk for COVID-19 infection?

Genomic modulators of the immune response

Genomic characterization of the 2019 novel human-pathogenic coronavirus isolated from a patient with atypical pneumonia after visiting Wuhan

Mechanisms of coronavirus cell entry mediated by the viral spike protein

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Inhibitors of cathepsin L prevent severe acute respiratory syndrome coronavirus entry

Protease-mediated enhancement of severe acute respiratory syndrome coronavirus infection

Cleavage and activation of the severe acute respiratory syndrome coronavirus spike protein by human airway trypsin-like protease

Evidence that TMPRSS2 activates the severe acute respiratory syndrome coronavirus spike protein for membrane fusion and reduces viral control by the humoral immune response

Cleavage of the SARS coronavirus spike glycoprotein by airway proteases enhances virus entry into human bronchial epithelial cells in vitro

A transmembrane serine protease is linked to the severe acute respiratory syndrome coronavirus receptor and activates virus entry

TMPRSS2 and ADAM17 cleave ACE2 differentially and only proteolysis by TMPRSS2 augments entry driven by the severe acute respiratory syndrome coronavirus spike protein

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

Functional assessment of cell entry and receptor usage for SARS-CoV-2 and other lineage B betacoronaviruses

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

COVID-19 and the cardiovascular system

Tissue distribution of ACE2 protein, the functional receptor for SARS coronavirus. A first step in understanding SARS pathogenesis

A protein complex in the brush-border membrane explains a Hartnup disorder allele

Structure of dimeric full-length human ACE2 in complex with B0AT1

Structural basis for the recognition of SARS-CoV-2 by full-length human ACE2

Frequent and focal FGFR1 amplification associates with therapeutically tractable FGFR1 dependency in squamous cell lung cancer

ACE2 gene polymorphism and essential hypertension: an updated metaanalysis involving 11,051 subjects

The combination of ACE I/D and ACE2 G8790A polymorphisms revels susceptibility to hypertension: A genetic association study in Brazilian patients

Association of Angiotensin-Converting Enzyme 2 gene polymorphism and enzymatic activity with essential hypertension in different gender: A case-control study

Polymorphisms of ACE2 are associated with blood pressure response to cold pressor test: the GenSalt study

Structure of SARS coronavirus spike receptorbinding domain complexed with receptor

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor. Nature 1-9

Structural basis of receptor recognition by SARS-CoV-2. Nature 1-8

EvoDesign: Designing protein-protein binding interactions using evolutionary interface profiles in conjunction with an optimized physical energy function

EvoEF2: accurate and fast energy function for computational protein design

SSIPe: accurately estimating proteinprotein binding affinity change upon mutations using evolutionary profiles in combination with an optimized physical energy function

FireDock: fast interaction refinement in molecular docking

FireDock: a web server for fast interaction refinement in molecular docking

DynaMut: predicting the impact of mutations on protein conformation, flexibility and stability

CHARMM-GUI: a web-based graphical user interface for CHARMM

Membrane Builder for mixed bilayers and its application to yeast membranes

VMD: visual molecular dynamics

Scalable molecular dynamics with NAMD

CABS-flex 2.0: a web server for fast simulations of flexibility of protein structures

ENCoM server: exploring protein conformational space and the effect of mutations on protein function and stability

Task force COVID-19 del Dipartimento Malattie Infettive e Servizio di Informatica, Istituto Superiore di Sanità -Epidemia COVID-19. Aggiornamento nazionale

Transmission potential and severity of COVID-19 in South Korea

X-inactivation profile reveals extensive variability in X-linked gene expression in females

X chromosome gene expression in human tissues: male and female comparisons

Renin-angiotensin system at the heart of COVID-19 pandemic

Angiotensin converting enzyme 2: SARS-CoV-2 receptor and regulator of the renin-angiotensin system

A crucial role of angiotensin converting enzyme 2 (ACE2) in SARS coronavirus-induced lung injury

Simulation of the Clinical and Pathological Manifestations of Coronavirus Disease

Lethal infection in K18-hACE2 mice infected with SARS-CoV

SARS coronavirus infection of mice transgenic for the human angiotensinconverting enzyme 2 (hACE2) virus receptor

Mice transgenic for human angiotensin-converting enzyme 2 provide a model for SARS coronavirus infection

Simultaneous treatment of human bronchial epithelial cells with serine and cysteine protease inhibitors prevents severe acute respiratory syndrome coronavirus entry

Protease inhibitors targeting coronavirus and filovirus entry

The RCSB protein data bank: integrative view of protein, gene and 3D structural information

dbSNP: the NCBI database of genetic variation

Searching NCBI's dbSNP database

UniProt: the universal protein knowledgebase

UCSF Chimera-a visualization system for exploratory research and analysis

PAST: Paleontological statistics software package for education and data analysis

PatchDock and SymmDock: servers for rigid and symmetric docking

GRAMM-X public web server for protein-protein docking

The SWISS-MODEL Repository and associated resources

The I-TASSER Suite: protein structure and function prediction

SignalP 5.0 improves signal peptide predictions using deep neural networks

We wish to thank prof. Diane Damotte (University of Paris) for advice and critical reading of the manuscript.

M.A., P.A.: conception, coordination, designing, writing.M.C.: experimental set-up, pipeline development, in silico analysis; P.F., A.I., designing, data providing;

The authors declare no competing interests.

The authors declare no competing interests.

 File) Supplementary