key: cord-0695217-535lw99y
authors: Su, Zhaoqian; Wu, Yinghao
title: A Multiscale and Comparative Model for Receptor Binding of 2019 Novel Coronavirus and the Implication of its Life Cycle in Host Cells
date: 2020-02-21
journal: bioRxiv
DOI: 10.1101/2020.02.20.958272
sha: 78e49fdb6f0aa9924a5b510341d52b618fff0ca6
doc_id: 695217
cord_uid: 535lw99y

The respiratory syndrome caused by a new type of coronavirus has been emerging from China and caused more than 1000 death globally since December 2019. This new virus, called 2019 novel coronavirus (2019-nCoV) uses the same receptor called Angiotensinconverting enzyme 2 (ACE2) to attack humans as the coronavirus that caused the severe acute respiratory syndrome (SARS) seventeen years ago. Both viruses recognize ACE2 through the spike proteins (S-protein) on their surfaces. It was found that the S-protein from the SARS coronavirus (SARS-CoV) bind stronger to ACE2 than 2019-nCoV. However, function of a bio-system is often under kinetic, rather than thermodynamic, control. To address this issue, we constructed a structural model for complex formed between ACE2 and the S-protein from 2019-nCoV, so that the rate of their association can be estimated and compared with the binding of S-protein from SARS-CoV by a multiscale simulation method. Our simulation results suggest that the association of new virus to the receptor is slower than SARS, which is consistent with the experimental data obtained very recently. We further integrated this difference of association rate between virus and receptor into a mathematical model which describes the life cycle of virus in host cells and its interplay with the innate immune system. Interestingly, we found that the slower association between virus and receptor can result in longer incubation period, while still maintaining a relatively higher level of viral concentration in human body. Our computational study therefore provides, from the molecular level, one possible explanation that the new disease by far spread much faster than SARS.

The coronavirus disease 2019 has emerged at the end of year 2019 2 from Wuhan, a city in China, as a new infectious disease [1, 2] . It has been found that the replicated RNA genome and synthesized viral proteins are finally assembled together into 10 new viruses, before they escape and attack other cells [7, 8] . As a result, the infections of 11 2019-nCoV normally come with the similar symptoms as SARS, including fever, 12 respiratory difficulty and pneumonia [9, 10] . Different from SARS, however, the new 13 COVID-19 seems to have longer incubation period and thus is more contagious [11] . The 14 disease has caused more than 70,000 confirmed cases with at least 1000 death globally, 15 according to the data from the World Health Organization (WHO) on February 16 th 2020. 16 Therefore, the development of vaccine or therapeutic treatment for this ongoing public 17 health crisis is highly demanding [12, 13] . 18 Almost all the coronaviruses recognize their host cells through spike (S) proteins 19 [14, 15] . S-protein is a glycoprotein expressed on the surface of viral envelop as homo-20 trimers [16] . Each S-protein further consists of two subunits. The S1 subunit includes a 21 region called receptor-binding domain (RBD) which is used to target receptors in host 22 cells, while the S2 subunit regulates the membrane fusion between virus and host cells 23 [17] . These roles of S protein suggest that it could be a key target for vaccine and 24 therapeutics developed to neutralize virus infection by blocking their invasion [18] . 25 Moreover, it has been confirmed in a recent report that the new virus 2019-nCoV uses the 26 same cell entry receptor ACE2 as SARS coronavirus [19] . The atomic structures of 27 complex between human ACE2 and the RBD regions from S-protein of SARS-CoV have 28 been obtained by x-ray crystallography [17] . It was also shown that the sequence of S-29 protein from SARS-CoV shares more than 70% identity with the S-protein from 2019-30 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi. org/10.1101 org/10. /2020 nCoV [6] . Therefore, it is reasonable to hypothesize that the new coronavirus uses the 1 similar binding interface with ACE2 as SARS to enter host cells of human. The obvious [20-26]. Therefore, here we developed a multiscale computational strategy to compare 10 the process of recognition between the SARS-CoV and host cells with the interactions 11 between the new coronavirus and host cells. A mesoscale model is used to simulate the 12 process in which the coronaviruses are captured by ACE2 receptors on cell surface. We 13 further constructed a structural model for complex formed between ACE2 and RBD of 14 2019-nCoV S-protein, so that the rate of their association can be estimated by a coarse-15 grained Monte-Carlo simulation and further compared with the binding of S-protein from 16 SARS-CoV. Our simulation indicates that association of the new virus to the receptor is 17 slower than SARS, which is consistent with the experimental data obtained very recently. 18 We integrated this difference of association rate between virus and receptor into a simple 19 mathematical model which describes the life cycle of virus in host cells and its interplay 20 with the innate immune system. Interestingly, we found that the slower association 21 between virus and receptor can result in longer incubation period, while still maintaining 22 a relatively higher level of viral concentration in human body. Our computational study 23 therefore explains, from the molecular level, why the new COVID-19 disease is by far 24 more contagious than SARS. In summary, this multiscale model serves as a useful 25 addition to current understanding for the spread of coronaviruses and related infectious 26 agents.

A rigid-body (RB) based model is first constructed to simulate the kinetic process 29 about how viruses are captured by the cell surface receptors on plasma membrane. In 30 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https: //doi.org/10.1101 //doi.org/10. /2020 brief, within a three-dimensional simulation box, the plasma membrane is represented by 1 a flat surface below the extracellular region. The area of the square is 1 µm 2 , while the 2 height of the simulation box is 500 nm. A number of ACE2 receptors (200) are initially 3 placed on the membrane surface (pink in Figure 1a ). They are represented by rigid 4 bodies of cylinders and their binding sites are located on top of the cylinders (red dots in 5 Figure 1a ). The height of each cylinder is 10nm and its radius is 5nm. On the other hand, 6 space above the plasma membrane represents the extracellular region. A number of 7 coronaviruses are located in this area (golden in Figure 1a) . Each virus is simplified as a 8 spherical rigid body with a given radius (40nm). Trimeric S-proteins are uniformly 9 distributed on the spherical surface of each virus (green dots in Figure 1a ). Each S-10 protein can interact with an ACE2 receptor on plasma membrane. After any S-protein on 11 one virus forms an encounter complex with a receptor, we assume that the host cell is 12 captured by the virus. The dissociation between the virus and the receptor is not 13 considered in the system, because we also assume that, after the association between S-14 protein and ACE2, the virus can enter the cell through membrane fusion. Following the 15 initial random configuration, the diffusion of receptors and viruses, as well as their 16 association, were simulated by a diffusion-reaction algorithm until the system reached 17 equilibrium. The detail process of the simulation is specified in the Methods. 18 Before the rigid-body simulation, in order to provide a more realistic estimation 19 on the binding between ACE2 and different coronaviruses, we specifically compared the 20 S-protein from 2019-nCoV with the S-protein from SARS. We applied our previously 21 developed residue-based kinetic Monte-Carlo (KMC) method to simulate the associate 22 processes of these two systems. In detail, the atomic coordinates of the complex between The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10. 1101 the distance between their binding interfaces is fallen within a given cutoff value d c of 20 1 Å. At the end of each trajectory, receptor and viral protein either form an encounter 2 complex through their binding interface observed in the complex structure, or diffuse 3 away from each other. Based on the simulation results collected from all the 500 4 trajectories, we counted how many times an encounter complex can be formed by the end 5 of the simulation time, which gives the probability of association. As a result, the 6 comparison of calculated probabilities of association for both systems is plotted in Figure   7 2a. 8 The figure shows that probability of association between ACE2 and the S-protein 9 from 2019-nCoV is remarkably lower than the probability of association between ACE2 10 and the S-protein from SARS. Specifically, among the 500 simulation trajectories of which is quantitatively consistent with our simulation results. 23 We then fed the information derived from the structure-based simulations into the 24 rigid-body model. Two specific simulation systems were compared. A relatively fast rate 25 of association between receptors and S-proteins on viral surfaces was adopted in the first 26 system to represent the binding process of SARS, while a relatively slow rate of 27 association between receptors and S-proteins on viral surfaces was adopted in the second 28 system to represent the binding process of 2019-nCoV. All the other parameters such as 29 diffusion constants and concentrations in both systems remain the same. As a result, the 30 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint total numbers of viruses that were captured by host cells are plotted in Figure 2b as a 1 function of simulation time. Without surprise, the figure shows that although almost all 2 viruses were attached to the cell surfaces by the end of both simulations, the kinetic 3 process in the SARS system is much faster than the 2019-nCoV system, which is resulted 4 from the difference in the association rate between receptors and their corresponding S-5 proteins. This leads into the fact that during the early stage of simulations, more SARS 6 viruses attach to host cells than 2019-nCoV. For instance, when the simulations in both 7 systems reached the first 10 5 nanoseconds, there have already been more than 40 SARS 8 viruses attached to the cell surfaces. In contrast, there were less than 20 viruses attached 9 to the cell surfaces within the same amount of time in the 2019-nCoV system.

Considering that the function of a bio-system is often under kinetic, rather than 11 thermodynamic, control [28, 29] , we suggest that this time-dependent behavior is 12 biologically more relevant. In reality, not all the viruses have the opportunity to find their 13 target receptors on host cells. Many of them will be recognized and removed by our 14 innate immune system. Therefore, the capability of how fast a specific type of 15 coronavirus can target its receptors is especially critical to the process of its invasion, as 16 well as the follow-up stages in its life cycle.

In order to further explore the impacts of our rigid-body simulation results on the cells. On the other side, in order to avoid viral spread, our innate immune system triggers 29 inflammatory signaling pathways in the infected cells [30] . For instance, the viral RNAs The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint receptors initiates the signaling cascade by interacting with the mitochondrial antiviral-1 signaling (MAVS) protein [32] . The aggregation of MAVS on the surface of 2 mitochondria will trigger the NF-κB signaling pathway that turns on gene expression of 3 specific cytokines [S] to stimulate the inflammatory responses [33, 34] . The inflammation 4 of host organism leads to the apoptosis of infected cells and the removal of virus by 5 recruited immune cells such as microphages. In summary, the change of concentration for 6 each variable in above system can be described by following set of ordinary differential 7 equations (ODE).

15

Equation (1) and k a represent rates of viral protein translation, assembly and releasing. Equation (5) 20 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint describes the stimulation of inflammatory signaling by viral RNA. The parameters s S , K m , 1 and d S in the equation represent the maximal activation rate, the saturation coefficient and 2 the rate of degradation of inflammatory signals. Equation (6) The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint are cleared up by the innate immune system. After the removal of all viruses, the healthy 1 cells in the system grow again, representing the recovery of the patient.

We further incorporated the results derived from the rigid-body simulations into 3 the mathematical model to compare the viral life cycle in SARS and COVID-19. The 4 rigid-body simulation suggests that SARS-CoV binds to receptor ACE2 faster than 19-5 nCoV, given the same amount of time. Therefore, we applied the mathematical model to 6 two comparative systems. A relatively fast viral binding rate k b was adopted in the first 7 system to represent the binding process of SARS, while a relatively slow rate was 8 adopted in the second to represent the binding process of 2019-nCoV. All the other 9 parameters such as diffusion constants and concentrations in both systems remain the The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi. org/10.1101 org/10. /2020 between human receptor and S-proteins from different coronavirus in, and provided the 1 possible mechanism from the molecular level to their impacts on, regulating the 2 dynamics of the entire viral life cycle.

The recent outbreak of COVID-19 has drawn substantial attention especially after 5 it spread to more than thirty countries and became a Public Health Emergency of 6 International Concern (PHEIC) [2, 36, 37 ]. The disease is caused by a new type of 

The atomic structures of complex between ACE receptor and different viral S-28 proteins are needed for the simulations of their association. The structural models of these 29 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi. org/10.1101 org/10. /2020 protein complexes were prepared as follows. The complex structure between human 1 ACE2 and the RBD domain from the S-protein of SARS was determined by the x-ray applied to study the association between ACE2 and both S-proteins from the two virus 29 systems. For each system, 500 trajectories are carried out. Each trajectory starts from a 30 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint relatively different initial conformation, but the initial distances between the binding 1 interfaces of S-proteins and receptors in all trajectories are below 20Å. The probabilities 2 of association were then derived and compared based on counting how many encounter 3 complexes formed among these trajectories in the two systems.

Model the cellular attachment of coronavirus by rigid-body diffusion-reaction algorithm 5 As described in the Results, a rigid-body (RB) based model is constructed to 6 simulate the binding between coronaviruses and cell surface receptors ACE2 on plasma 7 membrane. Given the model representation and a randomly-generated initial 8 configuration (Figure 2a) , the dynamics of the system is evolved by following a 9 diffusion-reaction algorithm [43] [44] [45] . Viruses or receptors are selected by random order Cartesian and compositional spaces, the system will finally reach equilibrium. (1) to equation (7) [46] . The

algorithm starts from the initiation of time and populations of each species in the 29 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https: //doi.org/10.1101 //doi.org/10. /2020 simulation system. Within each simulation step, the rates for all reactions are re-estimated 1 by the given parameters and updated population of corresponding species. One of these 2 reactions is then randomly selected based on the calculation of their relative weights. is specified on the right. The system can further be described by a set of ordinary 28 differential equations, as written from Equation (1) to Equation (7) in the main text. 29 author/funder. All rights reserved. No reuse allowed without permission.

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint Figure 4 : Given predefined weights for all the rate parameters and the initial values for 1 each variable in the model, the dynamics of the system is evolved as a function of time by 2 solving the mathematical model numerically with a stochastic simulation algorithm. The 3 figure suggests that the dynamics of the system can be divided into three stages (a). We 4 further applied the model to two comparative systems. A relatively fast viral binding rate 5 k b was adopted in the first system to represent the binding process of SARS, while a 6 relatively slow rate was adopted in the second to represent the binding process of 2019-7 nCoV. As shown in (b), we found that the first stage in the simulation of 2019-nCoV (red 8 curve) is longer than the simulation of SARS (black curve), while the level of free virus 9 at the end of the first period in 2019-nCoV is relatively higher than the corresponding 10 level of free virus in SARS. The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02.20.958272 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10. 1101 

Clinical features of patients infected with 2019 novel coronavirus in 2 Wuhan

A familial cluster of pneumonia associated with the 2019 novel 4 coronavirus indicating person-to-person transmission: a study of a family cluster

Novel Wuhan (2019-nCoV) Coronavirus. American journal of 7 respiratory and critical care medicine

Emerging understandings of 2019-nCoV

A new coronavirus associated with human respiratory disease in China

Genomic characterisation and epidemiology of 2019 novel coronavirus: 13 implications for virus origins and receptor binding

Characteristics of the Life Cycle of Porcine Deltacoronavirus

Replication Kinetics, Cellular Ultrastructure and Virion Morphology, and Evidence 17 of Inducing Autophagy. Viruses

The life cycle of SARS coronavirus in Vero E6 cells

The Novel Coronavirus: A Bird's Eye View. The 21 international journal of occupational and environmental medicine

The course of clinical diagnosis and treatment of a case infected with 23 coronavirus disease 2019

Special Expert Group for Control of the Epidemic of Novel Coronavirus Pneumonia of the 25

An update on the epidemiological characteristics of 26 novel coronavirus pneumonia(COVID-19). Zhonghua liu xing bing xue za zhi = 27 Zhonghua liuxingbingxue zazhi

Defining the Epidemiology of Covid

Potential Interventions for Novel Coronavirus in China: A Systematic 31

Coronavirus spike proteins in viral entry and 33 pathogenesis

Mechanisms of coronavirus cell entry mediated by the viral spike 35 protein

Structure, Function, and Evolution of Coronavirus Spike Proteins. Annual review of 37 virology

Structure of SARS coronavirus spike receptor-binding domain complexed with 39 receptor

The spike protein of SARS-CoV--a target for vaccine and therapeutic 41 development

A pneumonia outbreak associated with a new coronavirus of probable 43 bat origin

Stochastic reaction-diffusion simulation with MesoRD

All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the 21

SmartCell, a framework to simulate cellular processes that combines 1 stochastic approximation with diffusion and localisation: analysis of simple networks

Spatial stochastic modelling of the phosphoenolpyruvate-4 dependent phosphotransferase (PTS) pathway in Escherichia coli

Crowding and hydrodynamic interactions likely dominate in vivo 7 macromolecular motion

Diffusion, Crowding & Protein Stability in a Dynamic 10

Molecular Model of the Bacterial Cytoplasm

Statistical mechanical concepts in immunology

Manipulating the selection forces during affinity maturation to generate 14 cross-reactive HIV antibodies

Cryo-EM structure of the 2019-nCoV spike in the prefusion 16 conformation

Biomolecular diffusional association

Rate theories for biologists

Apoptosis: an innate immune response to virus infection. 21 Trends Microbiol

Immune signaling by RIG-I-like receptors

The adaptor MAVS promotes NLRP3 mitochondrial localization 25 and inflammasome activation

Mitochondrial-mediated antiviral immunity

A structural perspective of the MAVS-regulatory mechanism on the 29 mitochondrial outer membrane using bioluminescence resonance energy transfer

COVID-19: what is next for public health?

2020: p

Epidemiological and clinical characteristics of 99 cases of 2019 novel 34 coronavirus pneumonia in Wuhan, China: a descriptive study

Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-37 Infected Pneumonia

I-TASSER: a unified platform for automated protein 40 structure and function prediction

Receptor recognition by novel coronavirus from Wuhan: An analysis 42 based on decade-long structural studies of SARS

Mapping monomeric threading to protein-45 protein structure prediction

Protein-protein complex structure predictions by multimeric 47 threading and template recombination

No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the

Predicting Protein-protein Association Rates using Coarse-1 grained Simulation and Machine Learning

General principles of binding between cell surface 3 receptors and multi-specific ligands: A computational study

Elucidating the general principles of cell adhesion with a 6 coarse-grained simulation model

A coarse-grained model for the simulations of biomolecular 8 interactions in cellular environments

Stochastic simulation of chemical kinetics, in Annual Review of Physical 10 Chemistry