Surfactants are commonly used as disinfection agents in personal care products against bacteria and viruses, including SARS-CoV-2. However, there is a lack of understanding of the molecular mechanisms of the inactivation of viruses by surfactants. Here, we employ coarse grain (CG) and all-atom (AA) molecular dynamics simulations to investigate the interaction between general families of surfactants and the SARS-CoV-2 virus. To this end, we considered a CG model of a full virion. Overall, we found that surfactants have only a small impact on the virus envelope, being inserted into the envelope without dissolving it or generating pores, at the conditions considered here. However, we found that surfactants may induce a deep impact on the spike protein of the virus (responsible for its infectivity), easily covering it and inducing its collapse over the envelope surface of the virus. AA simulations confirmed that both negatively and positively charged surfactants are able to extensively adsorb over the spike protein and get inserted into the virus envelope. Our results suggest that the best strategy for the design of surfactants as virucidal agents will be to focus on those strongly interacting with the spike protein.
In order to investigate the possible interactions between surfactants and the SARS-CoV-2 virus from a fundamental physico-chemical point of view, we perform here molecular dynamics simulations of this system. First, we use a coarse-grain (CG) model to perform molecular dynamics simulations of a SARS-CoV-2 virion in the presence of different surfactants. After that, we consider simulations with full atomic resolution. In this case, we consider a patch of a virus envelope membrane made of lipids and a full spike protein in contact with different surfactants. Overall, our simulations indicate that the most relevant mechanism in the interaction between surfactants and the virus is the interaction with the spike protein. Also, our atomistic simulations suggest a higher interaction of the spike with anionic surfactants as compared with cationic surfactants.
It is, of course, of great interest to understand how surfactants and their effects may contribute to inactivating the SARS-CoV-2 virus. Given the interactions that surfactants may have with proteins and surfactants, we can propose two possible mechanisms for the inactivation of the SARS-CoV-2 virus with surfactants ( Fig. 1 ). One possible mechanism is structural damage to the virus envelope as discussed before. 26 Another possible mechanism could be the denaturation of the highly exposed spike protein since surfactants are able to denaturate proteins.
Little is known about the mechanisms and interactions that occur when inactivating viruses by means of surfactants, 21 precluding its rational design for disinfecting applications. It is known that surfactants are able to denaturize or solubilize proteins, and also, they are able to destabilize lipid membranes due to their hydrophobic/hydrophilic nature 22–25 but it is not known how these effects may contribute to inactivating an enveloped virus. A recent study considering gastroenteritis viruses offers interesting clues 26 on this question. It is shown that viral particles with larger sizes and lower packing fractions kept their morphology intact after the application of mechanical force with AFM, whereas smaller viruses with higher packing fractions showed structural damage and content release. The virions can be made mechanically unstable in the presence of detergent and alcohol, and degradation of the virus structure can be obtained with AFM.
One of the most typical disinfecting agents is surfactants, which are employed in the formulation of soaps and other household cleaning products. Many common surfactants (such as cationic quaternary ammonium surfactants or anionic sulfate surfactants) are effective against different pathogens including viruses. 15–20 In the case of SARS-CoV-2, experimental evidence showed inactivation by sodium laureth sulfate, typically employed in soap. 18 Interestingly, experiments done using commercial hand soap indicate that SARS-CoV-2 is able to remain infectious after 5 min exposure to hand soap but not after 15 min exposure. 6
In the case of SARS-CoV-2, the studies developed during the present pandemic show that it is able to remain infectious over many different surfaces and remains stable under a wide range of environmental conditions. 6,7 For example, it remains infectious for 72 h over plastic surfaces, 5 7 days under a surgical mask, 6 and between 9 and 22 h over human skin depending on the virus variant. 8,9 Our molecular dynamics (MD) simulations indicate that the virus spikes can be attached to sebaceous human skin without any deformation or alteration and retain their hydration. 10 Also, our simulations indicate that they can remain adsorbed with a high affinity over common plastics without any substantial deformation. 11 Only some materials, such as metals 12 or carbon based materials, 13 seem to be able to have a substantial impact on the adsorbed virus spikes. These evidence justify the recommendations by health agencies about cleaning and disinfection of surfaces and hands. 14
The transmission of respiratory viruses like SARS-CoV-2 involves the expiration to the medium by an infected individual of droplets or aerosols that contain virus particles. The infection of another individual may be divided into two different mechanisms, direct or indirect transmission. The direct mechanism involves the inhalation of aerosols or the deposition of emitted droplets on mucosal surfaces. The indirect mechanism means that an expiratory droplet containing the virus may land on an environmental surface. Coronaviruses can remain viable over surfaces for extended periods of time (hours or days depending on the material), and eventually, another individual may touch the contaminated surface and consequently get infected when touching his/her mucose. 3–5
Coronaviruses are enveloped viruses, which means that the genetic material of the virus is protected by an envelope. In the case of SARS-CoV-2, the envelope is formed by phospholipids, membrane (M) proteins, and envelope (E) proteins. The envelope also has glycosylated spike (S) proteins that protrude from the envelope and are responsible for the infectivity of the virus.
The SARS-CoV-2 coronavirus emerged in December 2019 as a human pathogen that causes the COVID-19 respiratory disease 1,2 and at the time of writing this manuscript, the virus is still with us. Furthermore, COVID-19 is not the first pandemic that has been caused by a human coronavirus with a zoonotic origin, being the previous cases of SARS-CoV (2003) and MERS (2012), and future spillovers seem likely. 3 Therefore, investigations on basic scientific questions related to coronavirus inactivation and disinfection are of great importance. Here, we will consider a computational study with the aim to provide rational tools to select surfactants for disinfection processes suitable for SARS-CoV-2.
The surfactant–lipid cross interaction ( Fig. 3 ) was described also by the soft potential given by Eq. (4) . As in the case of the lipid–lipid and surfactant–surfactant interactions, the interactions of head beads with any other surfactant or lipid bead is purely repulsive. In order to model different types of surfactants (different strengths for the lipid–surfactant interaction), we have considered several different U 0 strengths for the attractive depth [Eq. (4) ] in the interactions between surfactant beads of tail type and lipid beads of interfacial type. We have also adjusted the value of U c for the different values of U 0 in Eq. (4) to avoid overlap between beads for the cases with the strongest attraction. The surfactant–protein interaction was modeled analogously to the lipid–protein interaction using Eq. (7) for all protein–surfactant beads, exploring different values of ɛ to mimic different degrees of interaction between surfactants and proteins.
Each CG surfactant has two beads, one with a hydrophilic character (head bead) and another with a hydrophobic character (tail bead). The size of the surfactant tail beads is taken equal to the size of the lipid interfacial beads, R I = 1.8 nm. The size of the surfactant head beads is taken as R SH = 1.5R I because, for this combination of bead sizes, the CG surfactants form micelles. 29 The interaction between surfactants was described, by Eq. (4) , using the same parameters as in the lipid case [Eqs. (5) and (6) ].
In order to model the surfactants that will interact with the virus in a way consistent with the virus CG model, we also used the CG model of Ref. 29 for describing the surfactants (see Fig. 3 ). This model was originally proposed as a generic and flexible CG model for describing self-assembly of amphiphilic molecules in vesicles, micelles, and other structures and includes both lipids and surfactants. In this model, surfactants always have zero charges since electrostatic interactions were not considered explicitly in the model. 29
In Eq. (7) , r 0 is the sum of the radii of the beads that interact, and r c is the cutoff, which is set to r c = 2r 0 . Unless stated otherwise, we will consider ɛ = 4 kcal/mol = 6.7k B T as in Ref. 30 . In order to explore the effect of a different strength of the lipid–protein interaction, we will also consider additional simulations with a weaker interaction with ɛ = 0.63 kcal/mol = 1.06k B T as considered in the original paper introducing this interaction. 32 We checked that for both values of ɛ, the structure of the virus model shown in Fig. 2 remains stable (proteins remain inserted in the bilayer envelope).
Cross interactions between the lipids and the structural proteins M, E, and S are responsible for the stability of these proteins in the envelope. These interactions were modeled using an attractive potential between lipid beads and the transmembrane domains of the M, E, and S proteins, given by
The cross interaction between head and interfacial beads was also considered to be purely repulsive and given by Eqs. (4) and (6) with r 0 = R I = 1.8 nm. Using these values, the model lipids spontaneously self-assemble into a vesicle. 29
In the virus model employed in the simulations ( Fig. 2 ), we have 19 290 interfacial beads and 38 580 head beads, modeling 38 580 phospholipids. The size of the beads was selected to be R I = 1.8 nm for the interfacial beads and R H = 0.75R I for the head beads. 30 The characteristic energies U c and U 0 are the same as in Ref. 30 and depend on the type of bead. For the interaction between interfacial beads, we have
In Eq. (4) , r 0 is the size of the bead, U c is a large repulsive energy preventing bead overlap, and −U 0 is the depth of the energy minima at r = r 0 . We note here that since Eq. (4) is not implemented in the standard release of LAMMPS, we have employed the custom code available in Ref. 31 (see details in the supplementary material ).
The lipids of the virus envelope were modeled using the bilayer CG model previously developed in Ref. 29 as done by Ref. 30 . This bilayer model consists of three beads per bilayer segment (i.e., “1.5” beads/lipid), with two hydrophilic headgroup beads (heads) connected by harmonic bonds to a hydrophobic bead designed as “interfacial bead” in Ref. 29 (see Fig. 3 ). The interaction potential between lipid beads has a soft core repulsion and a short range attraction,
where r c is equal to the bead size. We take the same values of A s and r c as in Ref. 30 . In the case of the S protein, the S1 and S2 domains were mapped to 60 and 50 CG beads, respectively, and the 22 glycans of S were each mapped to a single bead. Some beads of S are charged. Interactions between beads of the same S protein (intraprotein interactions) were modeled using a heteroelastic network model. Interprotein interactions were composed of excluded volume, attractive, and screened electrostatic terms. Excluded volume interactions were modeled using the soft core interaction given by Eq. (1) for M and E proteins. Attractive, nonbonded interactions between interprotein contacts were modeled as the sum of two Gaussian potentials,
The virion proteins M and E were coarse-grained in a way that each CG bead corresponds approximately to five aminoacids. All CG beads of M and E proteins have zero charge. The structure of these proteins is maintained by rigid bonds. The interactions between beads of different proteins were modeled using the following soft core cosine potential that avoids overlap between beads:
The CG model of the SARS-CoV-2 virus considers the virus envelope ( Fig. 2 ), and it neglects the internal structure of the virus (the viral RNA and the N proteins that encapsulate the genetic material inside the virion are not included in the model). In this model, the virus envelope is made of 1000 dimeric membrane (M) proteins and 20 envelopes (E) pentameric transmembrane proteins in a lipid bilayer of 38 580 lipids. It has also 30 Spike trimeric proteins (S) inserted at the envelope by a small transmembrane fragment (see Fig. 2 ).
Scheme of the non-bonded interactions between the components of the virus and surfactants used for the simulations. (*) Interaction not considered for simulations 1–3 in Table I . Surfactant in orange color.
Scheme of the non-bonded interactions between the components of the virus and surfactants used for the simulations. (*) Interaction not considered for simulations 1–3 in Table I . Surfactant in orange color.
Coarse grain model of the SARS-CoV-2 envelope developed by Ref. 30 . Spike protein in green, M protein in dark blue, E protein in red, and viral lipids in purple, cyan, and pink. (a) View of the full virus particle. (b) Cut off the view in (a) so that the internal structure of the envelope can be seen.
Coarse grain model of the SARS-CoV-2 envelope developed by Ref. 30 . Spike protein in green, M protein in dark blue, E protein in red, and viral lipids in purple, cyan, and pink. (a) View of the full virus particle. (b) Cut off the view in (a) so that the internal structure of the envelope can be seen.
We have performed Coarse-Grained (CG) MD simulations of a full SARS-CoV-2 virus in the presence of surfactants using a Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS), 19 September 2019 version. 27,28 Our simulations employ a CG model of surfactants developed by Ref. 29 and a compatible CG model of a full virion particle developed by Ref. 30 (see Fig. 2 ). All the elements of the employed CG models of the virus and the surfactants and their interactions are summarized schematically in Fig. 3 and described in detail below.
In order to test the influence of the parameters employed for simulating the supramolecular structure of the SARS-CoV-2 virus (in particular, the strength of the lipid–lipid interaction), we also ran a series of three additional simulations in which we weakened the lipid–protein interaction. In these simulations, the surfactant–protein interaction was considered equal to the lipid–protein interaction. These simulations are indicated as 7′, 8′, and 9′ in Table I .
We have performed simulations exploring different values of the strength of the interactions between surfactants and the virus components (proteins and lipids), as indicated in Table I (Simulations 1–9). As seen in the table, the surfactant–lipid interactions ranged from an attraction equal to that of the lipid–lipid interaction to a much larger attraction, and the surfactant–protein interactions ranged from no interaction at all to an attraction equal to that of the lipid–protein interaction.
The simulated system consists of a SARS-CoV-2 virus particle and 4000 surfactant molecules in a simulation box of dimensions 240 × 240 × 240 nm 3 , which corresponds to a surfactant concentration of 0.5 mM. In all the simulations, we have integrated the equations of motion using a timestep of 100 fs as in Ref. 30 . The temperature of the system was set at 300 K. We employed a Langevin thermostat as implemented in LAMMPS 27,33 with a decay time of t damp = 100 ps as in Ref. 29 .
The result from the RE simulation was used as starting point for an unbiased, standard MD simulation at 300 K. From the 24 final configurations obtained in the RE simulation, we took the one corresponding to the replica at 300 K as the initial configuration for the new simulation. This additional MD simulation at 300 K after the RE lasted for 1.5 · 10 7 and the interaction parameters used are also the ones from simulation 7 in Table I .
Convergence to metastable states is always an issue in MD simulations (particularly in CG models). In order to test the possible lack of convergence to the equilibrium of our simulations, we did a test using the Replica Exchange (RE) method. Starting with the same initial condition and parameters as in Simulation 7 in Table I , we have performed a Replica Exchange (RE) MD simulation in which we considered 24 different replicas separated by 20 K each one (T of the thermostats ranging from 300 to 760 K), with a frequency trial of a swap between replicas of each 100 time steps. This simulation was run for 1 · 10 6 time steps.
During the simulations, we monitored the number of surfactants adsorbed over the S protein and also the number of surfactants inserted into the membrane (see the supplementary material for details). The simulations were performed until the number of surfactants adsorbed over the spike protein and the number of surfactants inserted inside the membrane remain stable over time. This corresponds to 188 ns for the case of SDS and 127 ns for DTAB, as indicated in Table II . In order to study the stability of the results, we have performed additional simulations in which adsorbed surfactants were perturbed. We considered the last configuration of the simulation of the spike with SDS, and we apply an external force over selected surfactants that are adsorbed on the spike protein (20 of the adsorbed surfactants) so that they get fully detached from the S protein and returned to bulk. After this perturbation, we let the system equilibrate again until the number of surfactants adsorbed over the spike, and the number of surfactants inserted into the membrane is stable over time. Full detail of the protocol is given in the supplementary material .
The initial coordinates for the simulations were obtained by adding surfactants to the spike model developed in Ref. 35 (see Fig. 4 ), as described in the supplementary material . Once the system was equilibrated (see the supplementary material ), we performed production NVT simulations (see details in Table II ). In these conditions, the initial concentration of free DTAB and SDS surfactants in solution for each system is 11.4 and 14.5 mM, respectively. In order to avoid the crossing of surfactants from the external side of the membrane to the internal side across the periodic boundary conditions, we added an empty space at both sides, generating water/vacuum interfaces that preclude the crossing of the surfactants. The details of the simulations are summarized in Table II .
The force-field employed in the simulations was CHARMM36 37,38 as in Ref. 35 and as in our previous simulations of the SARS-CoV-2 spike glycoprotein. 10,11,13 This force field includes an appropriate parametrization for the surfactants. 39,40
The MD simulations were performed using NAMD 2.14 software 36 as described in the supplementary material . The temperature was set at 300 K in all simulations, employing the Langevin thermostat with a damping coefficient of 1 ps −1 .
Structures used in AA simulations. (a) Full spike glycoprotein inserted into the membrane. Model from Ref. 35 . Protein in NewCartoon representation (monomers in red, orange, and gray colors), glycans in line representation, and membrane represented by a surface; (b) dodecyltrimethylammonium bromide surfactant (DTAB); and (c) sodium dodecyl sulfate surfactant (SDS). Atom colors: Carbon in cyan, hydrogen in white, nitrogen in blue, oxygen in red, and sulfur in yellow.
Structures used in AA simulations. (a) Full spike glycoprotein inserted into the membrane. Model from Ref. 35 . Protein in NewCartoon representation (monomers in red, orange, and gray colors), glycans in line representation, and membrane represented by a surface; (b) dodecyltrimethylammonium bromide surfactant (DTAB); and (c) sodium dodecyl sulfate surfactant (SDS). Atom colors: Carbon in cyan, hydrogen in white, nitrogen in blue, oxygen in red, and sulfur in yellow.
The structure of the fully glycosylated and palmitoylated S protein (“up” conformation) inserted into a membrane was taken from a previously developed model, 35 and it is shown in Fig. 4(a) . The lipid membrane where the S protein is inserted is constituted by a mix of 1-palmitoyl-2-oleoyl-phosphatidylcholine (POPC), 1-palmitoyl-2-oleoyl-phosphatidylethanolamine (POPE), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (POPS), palmitoyloleoylphosphatidylinositol (POPI), and cholesterol lipids mimicking the composition of the endoplasmic reticulum–Golgi intermediate compartment. 35 The chemical structures of the surfactants considered in our simulations were shown in Figs. 4(b) and 4(c) .
We have performed all atomic MD simulations of the SARS-CoV-2 spike glycoprotein inserted into a bilayer membrane (corresponding to a small patch of the virus envelope) in contact with two different surfactants, the cationic Dodecyltrimethylammonium (DTAB) and the anionic Sodium Dodecyl Sulfate (SDS).
To illustrate the results of our CG simulations, we first describe in detail a specific particular case (simulation 4 in Table I), and after that, we will discuss how the results change for the different combinations of force field parameters considered in Table I.
In simulation 4, we performed two different runs corresponding to two different initial conditions, one with the surfactants randomly distributed [Fig. 5(a)] and another with the surfactants pre-assembled in micelles [Fig. 5(b)]. We recall that the surfactant concentration was the same in both cases. The results obtained from both initial conditions are also shown in Fig. 5.
FIG. 5.
View largeDownload slideSnapshots of the initial (0 steps), intermediate (after 106 MD steps), and final (1.5 × 107 steps) configurations of Simulation 4 in Table I, corresponding to the same simulation with two different initial conditions: (a) Initially, the surfactants are randomly dispersed in the medium and (b) surfactants are initially pre-assembled into equilibrium micelles in the initial configuration.
FIG. 5.
View largeDownload slideSnapshots of the initial (0 steps), intermediate (after 106 MD steps), and final (1.5 × 107 steps) configurations of Simulation 4 in Table I, corresponding to the same simulation with two different initial conditions: (a) Initially, the surfactants are randomly dispersed in the medium and (b) surfactants are initially pre-assembled into equilibrium micelles in the initial configuration.
Close modal
In the case of initial random dispersion of surfactants [Fig. 5(a)], the surfactants rapidly form micelles and interact with the virion. At the end of the simulation, there are no free surfactants remaining in the solution, and the integrity of the virus particle was preserved during the simulation [Fig. 5(a)]. The same results were obtained for the case of an initial dispersion of surfactants pre-assembled in equilibrium micelles [Fig. 5(b)].
Some interesting features of the simulation can be observed from the snapshots. As seen in Fig. 5, after equilibration, the spike proteins are covered by surfactants, and also, some surfactants are adsorbed onto the envelope membrane. In Fig. 6, we show, in detail, the evolution of the inactivation/coverage of a spike protein by surfactants (see also the movie provided in the supplementary material). In Fig. 6(b), we observe how a first micelle gets adsorbed on the spike protein. After some steps in Fig. 6(c), the spike is covered by other micelles, and finally, in Fig. 6(d), the spike protein tilts toward the viral envelope probably aided by surfactants that have been previously inserted in the envelope. The process of insertion of surfactants at the envelope is shown in detail in Fig. 7 (see also the movie provided in the supplementary material). We observe that, after approaching the envelope, a surfactant micelle gets easily adsorbed with the surfactants being incorporated into the membrane, creating a surfactant patch at the membrane exposed to the environment.
FIG. 6.
View largeDownload slideDetail of a snapshot of simulation 4 (see Table I) highlighting the process of coverage by surfactants of a particular virus spike. (a) Initial configuration, (b) adsorption of a micelle over the spike, (c) the spike becomes covered by surfactants, and (d) the spike modifies its conformation, collapsing over the envelope.
FIG. 6.
View largeDownload slideDetail of a snapshot of simulation 4 (see Table I) highlighting the process of coverage by surfactants of a particular virus spike. (a) Initial configuration, (b) adsorption of a micelle over the spike, (c) the spike becomes covered by surfactants, and (d) the spike modifies its conformation, collapsing over the envelope.
Close modal
A quantitative analysis of the number of surfactants in contact with the S protein and the number of surfactants inserted in the envelope is given in Fig. 8. In order to identify the contacts between surfactants and lipids and between surfactants and S proteins, we have first computed their corresponding radial distribution functions. As seen in Figs. 8(a) and 8(c), the RDFs show well defined peaks that allow a clear identification of the first coordination shells. We will thus consider that a surfactant is in contact with a lipid or a spike protein when it is in its first coordination shell. Using these RDFs, we have calculated the number of surfactants in contact with lipids [Fig. 8(b)] and spike proteins [Fig. 8(d)]. As seen in Figs. 8(b) and 8(d), the adsorption of surfactants can be considered equilibrated after 107 time steps. We observe that at equilibrium, the number of surfactants at the envelope (in contact with the lipids) is less than half the number of surfactants in contact with S protein (≈708 and ≈1708, respectively). Therefore, despite the weakness of the surfactant-S protein interaction, the surfactants are largely adsorbed at the highly exposed S proteins. Also, the simulation reveals that these surfactants not only cover the protein but also affect its orientation at the membrane and its exposure to the environment (see Fig. 6). This result indicates the interaction with S as a major mechanism for the inaction of the virion particle. This is remarkable, given that in this simulation, the strength of the surfactant–protein attraction is rather weak (of the order of the thermal energy), see Table I. In order to study in detail the impact of the strength of the surfactant–protein and surfactant–lipid interactions, we have considered a full exploration of different values for the parameters in the model (Simulations 1–9 in Table I). Since the initial condition had no impact on the final results, we have performed all other simulations in Table I, only with the initial condition of randomly distributed surfactants [Fig. 5(a)]. The results are summarized in Fig. 9 and Table III. In all cases, we obtain adsorption or incorporation of surfactants at the viral particle, which is deformed from its original spherical shape (see snapshots in Fig. 9) but maintains its integrity. Overall, the results show a dramatic impact of the interaction between surfactants and proteins, being this interaction more relevant than that between surfactants and lipids. This can be clearly seen in the data presented in Table III. In the case of simulations 1–3 (no interaction between surfactants and proteins), most of the surfactants remain free in the solution even for the case with the strongest surfactant–lipid interaction (recall Table I for the strength of the interactions). As long as some interaction between the surfactants and proteins is included, a massive condensation of surfactants over the virus is observed (even for the cases with a weak surfactant–protein interaction of the order of the thermal energy).
FIG. 8.
View largeDownload slideAnalysis of results for simulation 4 in Table I. (a) Radial distribution function (RDF) between surfactant tails and the hydrophobic beads of the lipids (b) Number of surfactants in contact with lipids (i.e., in the first coordination shell) as a function of time, as identified from the RDF in (a). (c) Radial distribution function (RDF) between surfactant beads and spike protein beads. (d) Number of surfactants in contact with spike protein (i.e., in the first coordination shell) as a function of time, as identified from the RDF.
FIG. 8.
View largeDownload slideAnalysis of results for simulation 4 in Table I. (a) Radial distribution function (RDF) between surfactant tails and the hydrophobic beads of the lipids (b) Number of surfactants in contact with lipids (i.e., in the first coordination shell) as a function of time, as identified from the RDF in (a). (c) Radial distribution function (RDF) between surfactant beads and spike protein beads. (d) Number of surfactants in contact with spike protein (i.e., in the first coordination shell) as a function of time, as identified from the RDF.
Close modal
FIG. 9.
View largeDownload slide(a) Final snapshots of simulations 1–9 corresponding to Table I. View of the internal part of the virus in each case. Lipid–surfactant interaction increasing from left to right, and protein–surfactant interaction increasing from top to bottom; (b) Number of surfactants in contact with the spike protein vs surfactants in contact with viral lipids for each of the nine simulations.
FIG. 9.
View largeDownload slide(a) Final snapshots of simulations 1–9 corresponding to Table I. View of the internal part of the virus in each case. Lipid–surfactant interaction increasing from left to right, and protein–surfactant interaction increasing from top to bottom; (b) Number of surfactants in contact with the spike protein vs surfactants in contact with viral lipids for each of the nine simulations.
Close modal
TABLE III.
Sim..
Incorporated w/lipids.
In contact w/S protein.
Interacting surfactants.
Free surfactants.
1 206 0 207 3793 2 292 0 309 3691 3 305 0 318 3682 4 708 1708 4000 0 5 1565 1498 4000 0 6 1830 1405 4000 0 7 809 2502 4000 0 8 1897 2340 3972 28 9 2543 1793 3910 90 Sim..
Incorporated w/lipids.
In contact w/S protein.
Interacting surfactants.
Free surfactants.
1 206 0 207 3793 2 292 0 309 3691 3 305 0 318 3682 4 708 1708 4000 0 5 1565 1498 4000 0 6 1830 1405 4000 0 7 809 2502 4000 0 8 1897 2340 3972 28 9 2543 1793 3910 90 View LargeIt is interesting to look, in more detail, at the number of surfactants that are in contact with lipids and in contact with the spike protein for each simulation [Table III and Fig. 9(b)].
For simulations 1–3, the contacts between the spike protein and surfactants are nonexistent as there is no attractive interaction between them. The number of surfactants at the envelope increases as the surfactant–lipid interaction increases from simulations 1–3, but as we said, these values are much lower than in any other simulations including surfactant–Spike interactions.
The results change dramatically when including a weak surfactant–protein interaction (simulations 4–6, Table I). We have not only a large adsorption of surfactants over the spike protein but also a substantial increase in the contacts between lipids and surfactants as compared with that obtained in simulations 1–3 [see Table III and Fig. 9(b)]. This result indicates that the addition of the surfactant–protein interaction induces a higher interaction between the surfactants and the viral lipids (we recall that the strength of the surfactant–lipid interaction is kept the same in simulations 1 and 4, 2 and 5, and 3 and 6, see Table I). As seen in the snapshots of Fig. 9, the virus particles have their spikes completely covered by surfactants, as in the case discussed earlier in this section. The adsorption of surfactants over the S protein is similar for the three simulations but decreases from simulation 4–6, as the surfactant–lipid interaction increases. Also, the number of surfactants in contact with the lipids increases from simulation 4–6 [Fig. 9(b)].
As we increase the surfactant–protein interaction (simulations 7–9), both the surfactant–spike and surfactant–lipid contacts increase [Fig. 9(b)]. The contacts between the surfactants and the spike protein are the highest in simulation 7, which corresponds to the simulation with the highest surfactant–protein interaction combined with the lowest lipid–surfactant interaction.
In addition to simulations 1–9, we have also performed additional simulations (simulations 7′, 8′, and 9′ in Table I) in order to test the influence of the parameters employed for simulating the supramolecular structure of the SARS-CoV-2 virus in the results. In these additional simulations, we have a weakened lipid–protein interaction (as compared with the original model). In these simulations 7′–9′, we consider that the surfactant–protein interaction is equal to the lipid–protein interaction, as in our previous simulations 7–9 (see Sec. II for details).
A comparison of the results obtained in simulations 7′–9′ with those obtained in simulations 7–9 is shown in Fig. 10. The most striking difference obtained with the modified (“weakened”) model of the virus is that now it is possible to observe a loss of integrity in the virus particle. This is observed in the case of simulation 9′ [see Fig. 10(a)] where a spike protein has been detached by the action of surfactants. Compared with the previous simulations, in the case of simulation 9′, we observe a larger deformation of the envelope [Fig. 10(a)]. It should be noted that the number of surfactants in contact with the lipids and in contact with the S protein is smaller in simulation 9′ as compared with simulation 9 [Fig. 10(b)]. Therefore, a smaller number of adsorbed surfactants produce a larger structural effect in the virus. In the case of simulations 7′ and 8′, the virus envelope retains a nearly spherical shape in contrast with the results in simulations 7 and 8 in which we observe substantial deformation of the envelope [Fig. 10(a)]. The number of surfactants in contact with the spike decreases in simulations 7′ and 8′ as compared with 7 and 8, respectively [Fig. 10(b)]. In fact, the number of surfactants in contact with lipids and proteins obtained in simulations 7′–9′ is more similar to that obtained for simulations 4–6 than for those obtained in simulations 7–9 [see Figs. 9(b) and 10(b)]. This is again in line with the conclusion that the most relevant parameter to interpret our simulations is the value of the surfactant–protein interaction.
FIG. 10.
View largeDownload slide(a) Final snapshots of simulations 7′, 8′, 9′, 7, 8, and 9 corresponding to Table I. View of the internal part of the virus in each case. Lipid-surfactant interaction increasing from left to right; (b) Number of surfactants in contact with the spike protein vs surfactants in contact with viral lipids.
FIG. 10.
View largeDownload slide(a) Final snapshots of simulations 7′, 8′, 9′, 7, 8, and 9 corresponding to Table I. View of the internal part of the virus in each case. Lipid-surfactant interaction increasing from left to right; (b) Number of surfactants in contact with the spike protein vs surfactants in contact with viral lipids.
Close modal
All the simulations discussed so far have been carried out at a constant temperature of 300 K, and the results were assumed to correspond to final equilibrium configurations. However, it could be possible that these configurations may actually correspond to a metastable state in which the system has become trapped. In order to explore this possibility, we have repeated one of the simulations (simulation 7 in Table I) using the replica-exchange (RE) technique in order to improve the sampling of configurations. The results from RE simulations are then used as an initial condition for subsequent standard simulation at 300 K (see Sec. II for details).
The results of RE simulations are shown in Fig. 11. During the RE simulations, the 24 different replicas sampled configurations corresponding to temperatures between 300 and 760 K [Fig. 11(b)]. A final snapshot of the replica ending at 300 K is shown in Fig. 11(a). As in the previous case of non-biased MD simulations (simulation 7 in Fig. 9), we observe a substantial deformation of the shape of the virus with the envelope retaining its integrity without any hole in the structure. The main difference with the previous results is that we see that some viral lipids are removed from the virus envelope [Fig. 11(a)]. Concerning the surfactants, some of them are incorporated in the envelope structure, others are adsorbed over the virion, and others are still forming micelles in the solution (in some cases interacting or encapsulating virion lipids).
FIG. 11.
View largeDownload slide(a) Initial and final configurations of the RE simulation. The final configuration corresponds to the replica at 300 K at time step 1 × 106. (b) Time evolution of the replica exchange simulation. We show that thermostat corresponds to each system replica as a function of time. Each thermostat is indicated with a different color.
FIG. 11.
View largeDownload slide(a) Initial and final configurations of the RE simulation. The final configuration corresponds to the replica at 300 K at time step 1 × 106. (b) Time evolution of the replica exchange simulation. We show that thermostat corresponds to each system replica as a function of time. Each thermostat is indicated with a different color.
Close modal
Using the last configuration corresponding to the replica at 300 K, we run an additional MD simulation run of 1.5 · 107 time steps at 300 K. The results are shown in Fig. 12. We see that after the MD run, the lipids that were removed from the envelope returned from the solution to the virus. The incorporation of these lipids into the membrane is not perfect, and in the snapshot, we see some of them at the surface of the membrane or surrounded by surfactants. The shape of the envelope remains deformed far from the initial spherical shape. The surfactants that were in the solution get adsorbed on the virus. We calculated the number of surfactants in contact with the virion lipids and the spike protein over time as described in Sec. II. At equilibrium, the number of surfactants in contact with virion lipids and the spike protein was ∼1145 and ∼2230, respectively. Comparing this result with that obtained for simulation 7 (Table III), we see that the number of contacts with the lipids has slightly increased, and the contacts with the spike have decreased. This is due to the fact that the viral lipids got dissolved during the RE simulation, and consequently, surfactants could interact better with the lipids, increasing then the number of contacts with the lipids.
FIG. 12.
View largeDownload slideEffect of an unbiased MD simulation at 300 K over the configuration obtained from the RE simulation. (a) Full virus and (b) Cuts showing the internal structure of the virus envelope. The images on the left correspond to the last configuration at 300 K of the RE simulation. The right images correspond to the last configuration after the MD simulation at 300 K.
FIG. 12.
View largeDownload slideEffect of an unbiased MD simulation at 300 K over the configuration obtained from the RE simulation. (a) Full virus and (b) Cuts showing the internal structure of the virus envelope. The images on the left correspond to the last configuration at 300 K of the RE simulation. The right images correspond to the last configuration after the MD simulation at 300 K.
Close modal
With high quality products and considerate service, we will work together with you to enhance your business and improve the efficiency. Please don't hesitate to contact us to get more details of Nonionic Surfactants in COVID-19.