key: cord-0768157-l3hjzf8u
authors: Benková, Zuzana; Cordeiro, M. Natália D.S.
title: Structural Behavior of Monomer of SARS-CoV-2 Spike Protein during Initial Stage of Adsorption on Graphene
date: 2021-08-31
journal: Mater Today Chem
DOI: 10.1016/j.mtchem.2021.100572
sha: c9538410d66888b665be0b0ff91964523720136b
doc_id: 768157
cord_uid: l3hjzf8u

Spike glycoprotein of the SARS-CoV-2 virus and its structure play a crucial role in the infections of cells containing angiotensin-converting enzyme 2 (ACE2) as well as in the interactions of this virus with surfaces. Protection against viruses and often even their deactivation is one of a great variety of graphene applications. The structural changes of the non-glycosylated monomer of the spike glycoprotein trimer (denoted as S-protein in this work) triggered by its adsorption onto graphene at the initial stage are investigated by means of atomistic molecular dynamics simulations. The adsorption of the S-protein happens readily during the first 10 ns. The shape of the S-protein becomes more prolate during the adsorption but this trend, albeit less pronounced, is observed also for the freely relaxing S-protein in water. The receptor-binding domain (RBD) of the free and adsorbed S-protein manifests itself as the most rigid fragment of the whole S-protein. The adsorption even enhances the rigidity of the whole S-protein as well as its subunits. Only one residue of RBD involved in the specific interactions with ACE2 during the cell infection is involved in the direct contact of the adsorbed S-protein with the graphene. The new intramolecular hydrogen bonds formed during the S-protein adsorption replace the S-protein-water hydrogen bonds; this trend though less apparent, is observed also during the relaxation of the free S-protein in water. In the initial phase, the secondary structure of the RBD fragment specifically interacting with ACE2 receptor is not affected during the S-protein adsorption onto the graphene.

its strong interaction with incident visible light [16] . This is a pertinent feature for sterilization of material by heat generation. However, the adsorption of a virus onto graphene itself might cause significant disruption (denaturation) of the membrane structure, which in turn would lead to the loss of viral infectivity since the spike glycoprotein binds to ACE2 in its native and open conformation [7] . Molecular dynamics (MD) simulations of peptides and proteins interacting with graphene have revealed their strong adsorption on graphene and remarkable changes in their secondary structures induced by the adsorption [17] [18] [19] [20] . The fragment of viral protein R involved in the regulation of HIV genes through channel formation, Vpr13-33, underwent secondary structure transition from α-helix to β-sheet when the conformation of α-helix virtually disappeared already after 10 ns of the MD simulation [20] . The strength of adsorption of protein G-related albumin on graphene was compared with graphene oxide and it turned out that the disruption of the native conformation of this protein was more substantial when it interacted with graphene [19] .

Economic and reusable graphene masks have been already developed and are accessible

[21] and a graphene-based air purification scheme is under elaboration [22] . The mono-or multilayered graphene in the form of a mist spray or contained in surface cleaner wipes might be considered as an agent for disinfection of the surfaces infected with SARS-CoV-2 [23] . The capability of this equipment to exterminate the adsorbed SARS-CoV-2 viruses depends on the extent of the denaturation of the spike glycoprotein upon its adsorption.

In this work, the specific interactions between the non-glycosylated monomer of the spike glycoprotein trimer (further referred to as S-protein throughout this paper) and graphene during the initial phase of adsorption is studied using atomistic molecular dynamics (MD) simulations. These simulations should shed light on the kinetic of adsorption of the S-protein and the initial structural changes stemming from its adsorption onto the graphene. Special attention is devoted to the structural behavior of RBD as this fragment is decisive for the cell J o u r n a l P r e -p r o o f 5 infection. The extent of the secondary structure modification of the S-protein monomer renders a notion of the expected deformation of the secondary structure of the spike glycoprotein which in turn affects SARS-CoV-2 infectivity. The aim of this study is also to compare the structural changes of the S-protein interacting with the graphene in water with the structural changes induced by the relaxation of the free S-protein in water. This comparison allows to separate the effects brought about by the adsorption onto the graphene from the effects caused by the aqueous environment.

All-atom molecular dynamics simulations were used to investigate the interaction between S-protein and graphene sheet in water. The input structure of the non-glycosylated Sprotein monomer, composed of 1273 amino acid residues, was taken from the model of spike glycoprotein homotrimer obtained from Zhang group (QHD43416.pdb) [24] . This model is based on the incomplete crystal structure of the spike glycoprotein trimer determined using cryo-electron microscopy (PDB ID 6VXX, average resolution of 0.28 nm) [7] . This conformation corresponds to the RBD-down state, i.e., closed state. The missing residues, namely residues 1-26, 70-79, 144-164, 173-185, 246-262, 445, 446, 455-488, 502 , 621-640, 677-688, 828-853, 1148-1273, were modeled by the I-TASSER protocol. The omission of glycan units, which are important for the immunogenicity of the protein, may also influence the structure of the S-protein to certain extent. However, it was shown that the presence of glycan moieties did not influence the conformation of the spike glycoprotein at different temperatures [25] . If the trimer was considered the size of the simulation system would increase at least 4 times. Thus, the model used in this simulation study is capable of describing the secondary and tertiary structural changes induced during the initial adsorption stage of the S-protein but not the changes in the quaternary structure of the RBD of the spike protein trimer. In order to get J o u r n a l P r e -p r o o f an idea of the modification of the structure of RBD in the presence of glycans, the glycosylated RBD of the spike glycoprotein trimer with up-conformation of one RBD monomer and downconformation of two RBD monomers was also simulated during its initial adsorption onto the graphene. This initial conformation corresponded to the open state (prefusion conformation).

More details on the input conformation are provided in the Supplementary Information.

In order to adequately capture the hydrophobic interactions between the S-protein and water, which usually represent the most important contribution to the interactions, instead of creating only a thin hydration layer surrounding the S-protein, whole space between the graphene sheets was hydrated. The total number of atoms and water molecules along with the dimensions of the simulation box, and the separation between the graphene sheets in all simulated systems are summarized in Table 1 . Arginine and lysine were considered in their protonated forms, glutamate and aspartate were considered as deprotonated and histidine was neutral, which yielded a total charge of −7.

The C and N termini were deprotonated and protonated, respectively. In order to compensate J o u r n a l P r e -p r o o f for the overall negative charge of the system, seven Na + ions were added to the system. The atoms of the graphene were kept frozen and treated as neutral, i.e., only van der Waals interactions of the graphene with the S-protein and water molecules were taken into account.

In the initial conformation, the minimal distance between the atoms of S-protein and graphene was 0.8 nm. This value was more than twice the distance corresponding to the minimum of the van der Waals potential energy and it was still smaller than the interaction cutoff distance. The S-protein pointed with its S1 subunit containing the RBD (residues 319 to 541) towards the graphene. The graphene surface was aligned with the xy plane and placed at the bottom of the simulation box. To prevent interactions with periodic images along the z-direction, the second graphene plane was placed at a distance of 21 nm from the bottom graphene plane along the z axis and a vacuum slab of 42 nm was inserted above the upper plane. Periodic boundary conditions were applied in all three dimensions, and graphene being modeled as infinitely large.

The particle-mesh Ewald (PME) sum method [26, 27] was applied to handle the long-range electrostatic interactions. However, in the Ewald summation, the reciprocal sum was performed in three-dimensional space, but forces and potential were only applied in the z dimension to produce a pseudo-two-dimensional summation. The box height (63 nm) which is 3 times larger than the separation between the graphene sheets should be large enough to avoid electrostatic interactions between periodic images.

To obtain a reference system for the S-protein, MD simulations for a free S-protein in water were also carried out with the same initial conformation of S-protein as the one used for the MD simulations with the graphene. The MD simulations of pure water interacting with graphene provided a reference system for the water behavior at the interface with graphene.

J o u r n a l P r e -p r o o f All MD simulations were performed using the GROMACS software package [28] . The intra-and intermolecular interactions of S-protein and graphene were described by the CHARMM force field [29] . Water molecules were represented by the TIP3P model [30, 31] Using the non-polarizable force fields for the representation of graphene carbon atoms is quite common in the graphene-protein-water systems [32, 33] . Inclusion of polarization effects may influence the adsorption energies and relative strength of adsorption of the individual amino acids residues but it adds more interaction sites which makes simulations more time demanding.

The structure of an interacting protein might be modified directly through the interaction with the graphene as well as indirectly through the modification of the layer of interfacial water molecules due to the polarization effects.

Following an energy minimization by the steepest descent algorithm, the systems were allowed to equilibrate for 1 ns with a time step of 1 fs in the canonical ensemble (NVT). In this stage of the MD simulations, the positions of S-protein atoms were kept fixed and only the water molecules were relaxed. Finally, all atoms were relaxed during the next simulation runs in the (NVT) ensemble using a time step of 2 fs, but with all bond lengths constrained at their equilibrium lengths by the LINCS algorithm [34] . The temperature T was always kept fixed at 310 K by the velocity rescaling method with a relaxation time of 0.1 ps, in which a stochastic term was added to ensure sampling a proper canonical ensemble [35] . The overall simulation time was 40 ns out of which the first 20 ns was assumed as the equilibration period and the last 20 ns as the production phase. The time interval between sampling conformations was 10 ps.

The relaxation times, τ, extracted from the simple exponential autocorrelation functions fitting the initial parts of the plots, e −t/τ , of the mean values of the end-to-end distance/radius of gyration were 280 ps/400 ps and 1230 ps/1640 ps for the S-protein interacting with graphene and for the free S-protein, respectively. The relaxation time of the S-protein during its adsorption is considered. The plots and corresponding fits are shown in Figure S1 of the Supplementary Information. The faster relaxation of the S-protein interacting with the graphene when compared with the free protein may be attributed to its reduced degrees of freedom. Similar findings were reported for a geometrically constrained DNA in nanochannels where the relaxation times decreased with the decreasing diameter of the nanochannel, i.e., with the increasing confinement strength [36] . As referred to above, the long-range electrostatic interactions were dealt with the PME method with a real space cutoff of 1.3 nm. As to the Lennard-Jones (LJ) interactions, these were truncated at interatomic distances larger than 1.3 nm. To eliminate the discontinuity in the potential energy due to such cutoff, the LJ forces J o u r n a l P r e -p r o o f were smoothly switched to zero for interatomic distances between 1.0 nm and 1.3 nm. The neighbor list was maintained and updated using the Verlet cutoff scheme based on an energy drift with a buffer tolerance of 0.005 kJ•mol −1 •ps −1 [37] .

The kinetics of adsorption of the S-protein on graphene was monitored by the time dependence of the S-protein center of mass (COM) distance from the graphene plane, Rcom,z, as displayed in Figure 2 . For comparison, the number of atoms in contact with the graphene plane, Ncont, is also plotted in the same graph. Two distance criteria, 0.5 nm and 0.6 nm were adopted It should be mentioned at this point that the formation of a monolayer of adsorbed protein presumes disruption of the protein secondary structure. However, it has been found that there may be some sequences of amino acid residues which stabilize the secondary structure such as α-helix [38] . If these stabilizing interactions prevail over the interactions of the amino acid residues with the graphene the local secondary structure may persist. 

In contrast with small rigid molecules, the adsorption of sizeable proteins onto the solid surfaces is more complex and linked with their structural rearrangement leading to changes in surface affinity and many other phenomena. The adsorption of the S-protein on graphene is accompanied by changes of its shape as it is evident in Figure 3 showing the initial and final However, the time evolution of these components for both the interacting and free S-protein exhibits the same trend. The largest principal component slightly increases and is only modestly influenced by the adsorption, while the initially declining trend of the two remaining principal components is more supported by the adsorption. As is evident from the trends of the radius of gyration of the adsorbed S-protein and of the free S-protein, the S-protein becomes more compact in the initial stage of adsorption. The averaged values of the radius of gyration and its principal components along with the end-to-end distance of the adsorbed and free S-protein are collected in Table 2 . As can be seen the adsorption only modestly modifies the radius of gyration and its components while it is capable of shortening of the end-to-end distance Re.

J o u r n a l P r e -p r o o f 3.0 

The adsorption induced changes of the S-protein shape can also be quantified using the asphericity Δ and prolateness S parameters defined as follows [39] :

The asphericity Δ varies between 0 and 1 when going from a spherical to a rod-like shape. The prolateness parameter spans the range [−0.25, 2] with the lower and upper limit corresponding to an ideal disc and a rod shape, respectively. Negative values of the prolateness mean an oblate J o u r n a l P r e -p r o o f shape while positive values mean a prolate shape. Figure 5 shows the time evolution of both these parameters for the S-protein interacting with graphene and the free S-protein.

The shape of the starting conformation of the S-protein appears to be rather isotropic as can be deduced from the low values of respective Δ and S parameters and from the ratio λ1 : λ2 : λ3 = 2.30 : 1.55 : 1. For comparison, note that the asymptotic ratio of a random-walk chain is 3.48 : 1.65 : 1 [40] . Both processes, i.e., the relaxation of the free S-protein in water as well as its adsorption on the graphene surface in water contribute to the enhancement of the asphericity and prolateness parameters, more notably for the latter process. Thus, the asymmetry of the S-protein during its adsorption increases. However, flattening of the S-protein during its adsorption would lead to a decreasing trend of the prolateness parameter up to negative values when virtually a monolayer of atoms adsorbed on a surface might be formed.

Instead, during the simulations, the shape of the S-protein becomes more prolate and much more reduced Rgz components appear from the reorientation of the S-protein, which in turn enhances its prolate shape. This is also illustrated in Figure 3 . The ratio λ1 : λ2 : λ3 is changed to 2.78 : 1.39 : 1 and 2.15 : 1.16 : 1, respectively, for the relaxed free S-protein and the S-protein adsorbed on the graphene. Striking deformation of the spike glycoprotein adsorbed on the graphite was also observed in other molecular dynamics study [33] . 

Influence of the presence of graphene on the conformational dynamics of the S-protein can be inferred from the time dependence of the root-mean-square deviation RMSD of atom positions with respect to the initial conformation, which can be evaluated as follows:

where N it the number of atoms ri(t) is the position of atom i at time t of the structure superimposed with the reference structure at time t = 0. In addition to RMSD for the whole Sprotein, RMSD of its subunits S1 (residues 14-685) and S2 (residues 686-1273) as well as of RBD (residues 319-541) [41] were calculated for both systems ( Figure 6 ).

As for the structural quantities discussed above, RMSD values of the free and adsorbed S-protein are well equilibrated after 20 ns. During adsorption of the S-protein, the most abrupt changes in the conformation happen within the first 5 ns when RMSD attains the value of 1.14 nm. The time evolution of RMSD of S1 subunit, which is found to be the most flexible moiety, copies the same trend and its averaged value of 1. 

Another aspect deserving special attention is the identification of residues from the S1 subunit that are in direct contact with graphene and their contact frequencies, i.e., the number of stored frames for which that occurs. Here, the contact event is considered when the distance between any nonhydrogen atom and the graphene falls below or equals to 0.3 nm. The residues satisfying this criterion and having contact frequency larger than 100 events during the whole MD run are collected in Table 3 along with their respective contact frequencies.

In several previous simulation studies, the strongest adsorption on graphene has been found for Trp and Tyr with aromatic substituents and for polar basic Arg with guanidine substituent, when the π-π stacking (Trp and Tyr) and guanidine-π stacking (Arg) interactions J o u r n a l P r e -p r o o f are prominent [42] [43] [44] [45] [46] . The amide-π stacking interactions are possible for polar neutral Gln and Asn, which are adsorbed fairly well. On the other side, residues with aliphatic side chains do not display significant binding affinities for graphene in water. As can be seen in Table 3 , there is an abundance of Asn (7 residues) in direct contact with graphene while only 2 Gln residues and 1 Tyr residue are found in close vicinity to graphene. The orientations of these residues really correspond to the amide-π stacking (Asn, Gln) and π-π stacking (Tyr). The absence of contacting Trp follows from the rare occurrence of this residue within the S1 subunit. The surrounding residues prevent these Trp residues from creating contacts with the graphene.

There are more residues interacting with graphene through their aromatic side chains or through the amidic or guanidine groups introduced in their side chains with the distance of the contact atoms from the graphene in the interval of 0.3 nm -0.6 nm. These residues include Arg357 and Arg466 with guanidine-π stacking interactions, Gln23, Asn331 and Asn448 with amide-π stacking interactions, Trp258, Tyr144, Tyr449, Phe4, Phe79, Phe329, Phe464, and Phe562 with π-π stacking interactions. Worth noticing is also the highest contact frequency found for Pro330 during the last 20 ns of simulation, even though Pro is classified as a weak adsorbent when interacting with the graphene [42] [43] [44] [45] [46] . Its close proximity to the graphene, however, results from the neighboring residues, Phe329 and Asn331, which directly interact with graphene.

Similarly, seven Thr residues occur in close vicinity to graphene, though polar neutral Thr is regarded as a weak adsorbent [42] [43] [44] [45] [46] . In this case, the interactions of Arg21 with graphene brings Thr19, Thr20 and Thr22 to the contact with the graphene while Thr73, Thr333 and [12, 47, 48] . It should be emphasized at this point, that Asn331 is the only amino acid residue in contact with graphene which is glycosylated in the spike glycoprotein trimer. In the glycosylated RBD, this amino acid residue is slightly removed from the graphene.

However, it has been shown that the glycan moieties are only marginally involved into the contact with interacting surface and they occupy the lateral regions of the adsorbed spike glycoprotein [33] . The interactions of individual residues with the graphene do not seem to be mediated by water molecules; the water molecules in the first hydration layer, however, affect the orientation of those residues which do not possess some stacking interactions with graphene.

For instance, the hydroxyl groups of Ser and Thr residues form hydrogen bonds with the water molecules contained in the first hydration layer while the amide groups of these residues occupy the space between the first and the second hydration layer. As shown in Figure S2 of the Supplementary Information, all residues of S1 subunit of the S-protein found as being in contact with graphene are located at the periphery of the spike glycoprotein trimer (6VXX structure [49] ) and thus, these residues are expected to be available to interactions with graphene also in the spike protein trimer.

J o u r n a l P r e -p r o o f Table 3 . Residues of S1 subunit being in contact with the graphene during the whole MD simulation and their contact frequencies (CF), the bold-marked residues belong to RBD subunit. a A cutoff of 0.3 nm is considered as the criterion for a contact and it applies to the nonhydrogen atoms of the S1 subunit.

The structuring of interfacial water is not disturbed by the presence of the S-protein as can be seen in Figure 8 , where the density profile of interfacial water is compared with the density profile of interfacial water above the graphene in the absence of the S-protein. 

The secondary structure of the free and adsorbed S-protein has been analyzed using the STRIDE algorithm [50] , which combines hydrogen bond energy data with statistically derived backbone torsion angle data. A weighted product of hydrogen bond energy and torsion angle probabilities for α-helix and β-sheet determines the initiation and termination of secondary J o u r n a l P r e -p r o o f structure units based on empirically optimized recognition thresholds. The variation of the secondary structure with time for the free and adsorbed S-protein during the first and the last 10 ns of MD simulations is displayed in Figure 12 . As can be seen, the secondary structure of the S-protein is not subject to substantial changes during its relaxation in water but also during its adsorption on graphene. Similar behavior was reported in the study of the spike glycoprotein adsorption on the graphite and cellulose [33] . The S1 subunit is dominated by β-sheets while the S2 subunit is rich in α-helices. The content of helices and main β-sheet substructures and their composition in RBD of the S-protein in its initial conformation is compared with the corresponding final content within RBD of both systems in Table 4 and shown in Figure 13 .

The initial secondary structure of RBD is composed of 4 α-helices, one 5-stranded β-sheet and one 2-stranded β-sheet. During the adsorption, a new α-helix (Phe338-Gly339-Glu340-Val341-Phe342-Asn343) is created, and one α-helix (Glu406-Val407-Arg408-Gln409) transforms into a turn. The situation is different with RBD of the free S-protein. In agreement with the adsorbed S-protein, a new α-helix (Phe338-Gly339-Glu340-Val341-Phe342-Asn343) arises, however two α-helices (Pro384-Thr385-Lys386 and Ile418-Ala419-Asp420-Tyr421-Asn422) present in the initial RBD structure disappear. Thus, from comparison of the adsorbed S-protein with its free analogue, one can conclude that the adsorption of the S-protein is responsible for the stabilization of two α-helices (Pro384-Thr385-Lys386-Leu387-Asn388-Asp389 and Ile418-Ala419-Asp420-Tyr421-Asn422) and for diminishing of one α-helix (Glu406-Val407-Arg408-

originates from the relaxation in water rather than from the adsorption onto graphene. The main 5-stranded β-sheet is stabilized upon adsorption since it is preserved during the adsorption of the S-protein. The strands of this β-sheet are aligned parallel to the graphene with the strand Asn354-Arg355-Lys356-Arg357-Ile358 being closest to the graphene through the guanidine-π stacking interactions between Arg357 and graphene. This strand is shortened during the J o u r n a l P r e -p r o o f relaxation of the free S-protein and some new short β-sheets arise. The two strands of the short 2-stranded β-sheet slightly elongates during the adsorption of the S-protein as well as during its free relaxation in water. Some MD studies have pointed out the preference of β-sheets to αhelices near the graphene surface [20, 51] but also the opposite trend has been observed [52] . In the present MD simulation study, the diminished α-helix was not in direct contact with graphene during the adsorption. It should be mentioned that 310-helices appeared as transient substructures during the MD simulations, being converted from turns, coils, or terminal fragments of α-helices. The secondary structure of the fragments containing residues, which are directly involved in the interactions with the ACE2 receptor [12] , however, remained unaltered during the adsorption.

The glycosylated RBD subunit adsorbs onto the graphene through two monomers as can be seen in Figure S3 of the Supplementary Information. One of these two monomers is closer to and in direct contact with the graphene. It can be seen that the glycan units are not situated between the graphene and amino acid residues of this closest monomer and the secondary structure is similar to the secondary structure of the RBD of S-protein ( Figure 13b ). While, similarly to the case of S-protein, the 5-stranded β-sheet is also stabilized in this monomer of the glycosylated RBD during the adsorption, the helical elements are not stable and periodically appear and disappear during the adsorption. Only the 310-helix Ser366-Val367-Leu368 remains stable. This helix preserves also in the adsorbed and freely relaxed S-protein. The contacts between the closest monomer and graphene are also similar to the contacts between RBD of the S-protein and graphene this applies also to Tyr449 amino acid residue.

One should keep in mind that the separated monomer of the spike protein trimer is more flexible as it has been found in previous MD simulations [53] . The enhanced flexibility has been observed in the hinge region Gly700-…-Asn710 and in the region Gln784--Ser810 of the S2 subunit. In the trimeric spike protein, these two regions form β-sheets due to their mutual J o u r n a l P r e -p r o o f stabilization, i.e., the β-sheet of the S2 subunit of one monomer is stabilized by the β-sheet of the hinge region of other monomer. In separated monomer, these β-sheets are lost and transform into turns or coils [53] . In the present MD simulations, the same tendency has been observed.

J o u r n a l P r e -p r o o f 

Glu406-Gln409

Phe338-Asn343

Phe338-Asn343

Asn354-Ile358

Ser366-Asn370

J o u r n a l P r e -p r o o f Table 4 . Secondary structure of RBD of the S-protein in its initial conformation, of the adsorbed S-protein, and of the freely relaxed S-protein.

S-protein initial structure adsorbed S-protein free S-protein α-helix a Phe338-Gly339-

Phe338-Gly339-

Tyr365-Ser366-

Tyr365-Ser366-

Glu406-Val407-

Gly404-Asp405-

Ile418-Ala419-Asp420-Tyr421-

Ile418-Ala419-Asp420-Tyr421- 

a During the MD simulations, some short α-helices transiently transformed into 310-helices.

The present atomistic MD simulations have studied the initial phase of the adsorption of the S-protein (non-glycosylated monomer of the spike glycoprotein trimer) onto the graphene surface in order to describe the kinetics and dynamics of this adsorption and to find out how this adsorption affects the structure of the S-protein.

The adsorption of the S-protein onto the graphene is fast during the first 10 ns and after 20 ns of the simulation run the number of contacts as well as the distance of the S-protein center of mass from the graphene are levelled off. The time dependences of the root mean square deviation of the S-protein and of its S1

and S2 subunits and RBD again follow the same trend for the adsorbed and freely relaxed Sprotein. The S1 subunit is found to be much more flexible than the S2 subunit, which in turn exhibits slightly larger flexibility than RBD. The adsorption of the S-protein stabilizes the structure of S1 subunit maintaining the rigidity of RBD as well in the equilibrated phase. The The specific interactions between the S-protein and graphene, with the orientation of the S1 subunit pointing to the graphene, are dominated by amide-π stacking of Asn and Gln residues. Orientations of the side chains of two Arg residues and one Trp, two Tyr, and five Phe residues, respectively, correspond to guanidine-π stacking and π-π stacking. Only the Tyr449 residue, which is directly involved in the specific interactions between RBD and ACE2 during the cell infection, gets into direct contact with graphene during the adsorption of S-protein.

There is a negligible destruction of the interfacial water structure in the close vicinity of graphene due to the presence of the S-protein. The relaxation of the free S-protein in water and even more its adsorption on the graphene promotes formation of intramolecular hydrogen bonds. The number of hydrogen bonds between the S-protein and water drops more strikingly in the case of the adsorbed S-protein.

The secondary structures of the adsorbed and the freely relaxed S-protein experience only minor changes during the simulations. As a direct consequence of the adsorption, one αhelix of RBD transforms into a turn. The adsorption stabilizes two α-helices in RBD, which diminish during the free relaxation of the S-protein in water. One new α-helix created during the adsorption is not related directly to the adsorption process since this α-helix is created also during the free relaxation of the S-protein in water. The strands of the 5-stranded β-sheet adopt parallel alignment with the graphene and with the closest strand being stabilized by the presence of graphene. The secondary structure of the fragments containing the residues directly involved in the RBD-ACE2 interactions [12] is not affected by the adsorption.

This study shows that RBD is quite rigid substructure of the S-protein in water. This implies that its secondary structure might remain unchanged during the adsorption on some other surfaces as well as RBD might remain quite resistant to the external conditions. In this study, the motifs containing residues, which are directly involved in the interactions with the ACE2 receptor, are neither subject to the changes of their secondary structure nor, with the exception of Tyr449, in the direct contact with the graphene.

These MD simulations suggest the affinity between the spike monomer of SARS-CoV- The cigarette smoke is composed of aggregated aromatic particles, which interact with a protein in water through van der Waals interactions, π-π stacking and analogous interactions, and the hydrophobic interactions. The protein-graphene interactions in aqueous medium might reliably represent the latter two types of interactions while the van der Waals attractions are supposed to be underestimated. Since in these systems, the hydrophobic interactions usually dominate, the conclusions drawn for the interactions of the spike protein with the graphene in water might be valid for the interactions of the spike protein in water droplet with the aggregates present in the cigarette smoke. It would be of interest to scrutinize the behavior of SARS-CoV-2 virus in water droplet exposed to the cigarette smoke and to answer the question whether this virus can be spread through the cigarette smoke. As this study is based on some approximations 

Data are available at https://www.openicpsr.org/openicpsr/project/134381/version/V1/view.

The description of files is provided as the Supplementary Information.

Initial (a) and final (b) conformation of the S-protein interacting with the graphene surface.

J o u r n a l P r e -p r o o f

A Novel Coronavirus from Patients with Pneumonia in China

A new coronavirus associated with human respiratory disease in China

From SARS and MERS to COVID-19: A brief summary and comparison of severe acute respiratory infections caused by three highly pathogenic human coronaviruses

Airborne Transmission of SARS-CoV-2

Theoretical Considerations and Available Evidence Opinion

Exaggerated risk of transmission of COVID-19 by fomites

The Proteins of Severe Acute Respiratory Syndrome Coronavirus -2 (SARS CoV -2 or n -COV19), the Cause of COVID -19

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

Structure, Function, and Evolution of Coronavirus Spike Proteins

Receptor Recognition by the Novel Coronavirus from Wuhan : an Analysis Based on Decade-Long Structural Studies of SARS Coronavirus

Structural basis of receptor recognition by SARS-CoV-2

Repurposing approved drugs as inhibitors of SARS-CoV-2 S-protein from molecular modeling and virtual screening Repurposing approved drugs as inhibitors of SARS-CoV-2 S-protein from molecular modeling and virtual screening

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor

The impact of angiotensin-converting enzyme 2 (Ace2) expression on the incidence and severity of covid-19 infection

Graphene: The New Two-Dimensional Nanomaterial, Angew. Chemie

Engineering Optical Absorption in Graphene and Other 2D Materials: Advances and Applications

Adsorption of Villin Headpiece onto Graphene , Carbon Nanotube, and C60: Effect of Contacting Surface Curvatures on Binding Affinity

Atomic-Level Study of Adsorption, Conformational Change, and Dimerization of an r -Helical Peptide at Graphene Surface

Adsorption of GA module onto graphene and graphene oxide: A molecular dynamics simulation study

Molecular dynamics simulations of conformation changes of HIV-1 regulatory protein on graphene

Are graphene and graphene-derived products capable of preventing COVID-19 infection?

Investigation of the Effect of Temperature on the Structure of SARS-CoV-2 Spike Protein by Molecular Dynamics Simulations

Particle mesh Ewald: An N.log(N) method for Ewald sums in large systems

A smooth particle mesh Ewald method

5: a highthroughput and highly parallel open source molecular simulation toolkit

Extending the Treatment of Backbone Energetics in Protein Force Fields: Limitations of Gas-Phase Quantum Mechanics in Reproducing Protein Conformational Distributions in Molecular Dynamics Simulations

Comparison of simple potential functions for simulating liquid water

Structure and Dynamics of Liquid Water with Different Long-Range Interaction Truncation and Temperature Control Methods in Molecular Dynamics Simulations

Influence of Single-Stranded DNA Coatings on the Interaction between

Computer simulations of the interaction between SARS-CoV-2 spike glycoprotein and different surfaces

A Parallel Linear Constraint Solver for Molecular Simulation

Canonical sampling through velocity rescaling

Modeling the relaxation time of DNA confined in a nanochannel

A flexible algorithm for calculating pair interactions on SIMD architectures

What makes a good graphene-binding peptide? Adsorption of amino acids and peptides at aqueous graphene interfaces

Shape distribution and correlation between size and shape of tetrahedral lattice chains in athermal and theta systems

Shape asymmetry of star-branched random walks and nonreversal random walks

Structural and functional properties of SARS-CoV-2 spike protein: potential antivirus drug development for COVID-19

Stability of peptide (P1 and P2) binding to a graphene sheet via an all-atom to all-residue coarse-grained approach

Simulations of Peptide-Graphene Interactions in Explicit Water

Favorable adsorption of capped amino acids on graphene substrate driven by desolvation effect

Computation of the binding free energy of peptides to graphene in explicit water

Adsorption of amino acids on graphene: assessment of current force fields

SARS-CoV-2, an evolutionary perspective of interaction with human ACE2 reveals undiscovered amino acids necessary for complex stability

The proximal origin of SARS-CoV-2

Knowledge-Based Protein Secondary Structure Assignment

The adsorption mechanism and induced conformational changes of three typical proteins with different secondary structural features on graphene

Graphene Interactions Enhance the Mechanical Properties of Silk Fibroin

Distinct Structural Flexibility within SARS-CoV-2 Spike Protein Reveals Potential Therapeutic Targets

This work was supported by grant VEGA 2/0122/20 and by Project UID/QUI/50006/2020 (LAQV@REQUIMTE) with funding from FCT/MCTES through national funds.