key: cord-1029958-ddqth18x
authors: Oliveira, A. Sofia F.; Shoemark, Deborah K.; Davidson, Andrew D.; Berger, Imre; Schaffitzel, Christiane; Mulholland, Adrian J.
title: SARS-CoV-2 spike variants differ in their allosteric response to linoleic acid
date: 2022-04-22
journal: bioRxiv
DOI: 10.1101/2022.04.21.489022
sha: a0a588a1917dc7d98801af083dbb6994cfaa1924
doc_id: 1029958
cord_uid: ddqth18x

The SARS-CoV-2 spike protein contains a fatty acid binding site, also found in some other coronaviruses (e.g. SARS-CoV), which binds linoleic acid and is functionally important. When occupied by linoleic acid, it reduces infectivity, by ‘locking’ the spike in a less infectious conformation. Here, we use dynamical-nonequilibrium molecular dynamics (D-NEMD) simulations to compare the response of spike variants to linoleic acid removal. These simulations show that the fatty acid site is coupled to functional regions of the protein, some of them far from the site (e.g. in the receptor-binding motif, N-terminal domain, the furin cleavage site located in position 679-685 and the fusion peptide-surrounding regions) and identify the allosteric networks involved in these connections. Comparison of the response of the original (‘Wuhan’) spike with four variants: Alpha, Delta, Delta plus and Omicron BA.1 show that the variants differ significantly in their response to linoleic acid removal. The allosteric connections to the fatty acid site on Alpha are generally similar to the original protein, except for the receptor-binding motif and S71-R78 region which show a weaker link to the FA site. In contrast, Omicron is the most affected variant exhibiting significant differences in the receptor-binding motif, N-terminal domain, V622-L629 and the furin cleavage site. These differences in allosteric modulation may be of functional relevance, e.g. in differences in transmissibility and virulence. Experimental comparison of the effects of linoleic acid on different variants is warranted.

The spike glycoprotein, which is located on the surface of the virion, mediates SARS-CoV-2 entry into host cells by binding primarily to the receptor angiotensin-converting enzyme 2 (ACE2) (1, 2) as well as to other receptors, such as neuropilin-1 (3, 4) and potentially nicotinic acetylcholine receptors (5, 6) . The spike ectodomain ( Figure 1A and Figure S2 ) contains the domain that directly binds to the human receptors (named receptor-binding domain, RBD) as well as all the machinery needed to fuse the host and viral membranes, including the fusion peptide (FP) (7) (8) (9) . The SARS-CoV-2 spike contains two proteolytic cleavage sites (7): a furin cleavage site located at the S1/S2 junction (residues N679-R685), which distinguishes SARS-CoV-2 from other betacoronavirus spike proteins and affects the overall stability of the protein and modulates infectivity (10) (11) (12) ; and, a cleavage site in the S2 subunit ( Figure S2 ) that releases the fusion peptide (10) . A free fatty acid (FA) binding site was also discovered in the SARS-CoV-2 spike by members of this team in late 2020 (13) . Subsequently, equivalent FA sites have been identified in other closely related spikes (e.g. (13) (14) (15) ), including in the pangolin coronavirus (14) and the original SARS-CoV (13) . This discovery opened the door for potential new spike-based therapies based on free fatty acids. The FA pocket is located at the interface between two neighbouring RBDs on adjacent monomers in the homotrimeric spike (13) ( Figure   1A ). This hydrophobic pocket is formed by two RBDs, with one RBD providing the aromatic and hydrophobic residues to accommodate the FA hydrocarbon tail and the RBD of the adjacent chain providing the polar (Q409) and positively charged (R408 and K417) residues that bind the FA carboxylate headgroup (13) (Figure 1B) . The presence of linoleic acid (LA) in the FA pocket stabilises the locked spike conformation (in which the receptor-binding motifs are buried inside the RBD trimer, rendering them inaccessible for binding to ACE2), reducing infectivity (13) . (13) . The spike protein is a homotrimer (7): each monomer is shown in a different colour: green, red and blue. LA molecules are highlighted with yellow spheres. Note that each FA binding site is located at the interface between two neighbouring monomers and is formed by residues from two adjacent receptor-binding domains. (B) Detailed view of the FA binding site: this pocket is lined by hydrophobic and aromatic residues, and the LA acidic headgroup is located near R408, Q409 and K417. (C) Models used as starting points for the equilibrium MD simulations of the Alpha, Delta, Delta Plus and Omicron variants (16) . The yellow spheres represent the LA molecules. The red spheres highlight the position of mutations, deletions and insertions in the four simulated variants. Biomolecular simulations have provided molecular level insight into SARS-CoV-2 spike structure and dynamics, uncovering the effect of mutations, predicting interactions and 4 revealing allosteric connections in the protein (e.g. (3, 13, (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) ), including providing insight into the complex role of the FA site (13, 17, 29, 30) . Equilibrium molecular dynamics (MD) simulations indicated persistent and stable interactions between LA and the spike trimer (13, 17) , and that binding of LA rigidifies the FA binding site (17) . Recently, in our previous work, using dynamical-nonequilibrium molecular dynamics (D-NEMD) simulations, we showed that the FA site is allosterically coupled to functional motifs for membrane fusion or with antigenic epitopes (29, 30) . These simulations revealed that the removal of LA from the FA sites induces long-range structural responses in the receptor-binding motif (RBM), N-terminal domain (NTD), furin cleavage site and FP-surrounding regions (29, 30) . D-NEMD simulations have also highlighted different allosteric and dynamical behaviours between the original (also known as 'Wuhan' or early 2020) spike and the D614G and BriSΔ (a variant containing an eight amino-acid deletion encompassing the furin recognition motif and S1/S2 cleavage site) variants (29, 30) .

Here, we use D-NEMD simulations (31) (32) (33) (34) to analyse the response of four spike variants to G339D, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R,   G496S, Q498R, N501Y, Y505H, T547K, D614G, H655Y, N679K, P681H, N764K, D796Y, N856K, Q954H, N969K and L981F) (41) . Omicron is more transmissible than Delta and any other preceding variants but apparently causes less severe disease than previous strains (e.g. (42) (43) (44) ). Figure S1 ). In this work (and similarly to our previous work (29, 30) ), the perturbation introduced to the protein was the instantaneous deletion of all LA molecules from the FA sites. Glycans are crucial to the biological functions of the spike, being important for shielding (18) and infection (e.g. by altering the dynamics of RBD opening (18, 27) ). To 6 investigate e.g. interactions with receptors, including the glycan shield is important (e.g. (6, 18) ). The aim here was to compare the response of the variant spike proteins. The allosteric communication networks and the response of the protein to an internal structural perturbation (here, LA removal) are unlikely to be qualitatively altered by the presence of glycans, as they cover the outside of the protein. The cryo-EM structure of the original spike in a locked conformation with LA bound (13) only contains glycans on the exterior.

In the D-NEMD simulations, the instantaneous deletion of the LA molecules from the FA sites prompts the structural response of the protein as it adapts to LA removal. Note that the D-NEMD simulations are not intended to model the physical process of LA binding or dissociation. The perturbation used here, which was the same as in our previous work (29, 30) , is designed to induce a rapid response and force signal transmission within the protein, hence allowing mapping of the mechanical and dynamical coupling between the structural elements involved in this process. The time evolution of the response is extracted using the Kubo-Onsager relation (31) (32) (33) (34) . Multiple D-NEMD simulations are performed and compared with the equilibrium trajectories to identify the protein's structural response. This response is averaged over multiple trajectories, removing noise (31) (32) (33) (34) . Given that the same perturbation (LA removal) is used for all systems, the structural response of the different variants can be directly and meaningfully compared. Note that the D-NEMD approach allows for the statistical significance of the structural rearrangements to be determined ( Figures S4-S8 ) (34) . We focus on the differences shown here to be significant. Figure 2 ). In the systems simulated, LA removal induces a conformational response in the FA pocket, which contracts due to movement of the hydrophobic and aromatic sidechains that line it ( Figure S10 ). Structural changes are then rapidly transmitted to the RBD and NTD, and then 8 to V622-L629, the furin cleavage site, and residues surrounding the FP ( Figure S9 ). As can be seen in Figures 2, 3 and S9, 0.1 ns after LA removal, significant structural rearrangements can already be observed in the RBD, mainly in the RBM, and in specific regions of the NTD, such as S71-R78, H146-E156 and L249-G257. The RBM contains the residues that directly interact with the host ACE2 receptor (8, 49) and is a known target for neutralising antibodies (50) (51) (52) .

The NTD is also a major target for neutralising antibodies (e.g. (53) (54) (55) (56) ). In particular, the H146-E156 and L249-G257 segments were shown to directly mediate the interaction between spike and specific antibodies (e.g. 4A8 monoclonal antibody) (54). The S71-R78 region is an antigenic epitope and has also been suggested to be involved in binding to other receptors besides ACE2 (57) . 

Significant variations in the response of the RBM (mainly of the A475-C488 segment) were observed between the five virus spike proteins (Figures 3 and S3 ). In the Alpha spike, the response of A475-C488 is notably weaker and more diffuse (the amplitude of the structural changes is lower) than the original protein (Figures 3 and S3) . In contrast, Delta, Delta plus and Omicron show a larger response for the A475-C488 segment than the original spike 

The response of the NTD differs significantly between variants, particularly in the S71-R78, H146-E156 and L249-G257 segments (Figure 4) . For example, Delta shows larger structural rearrangements in the region H146-E156 than the original protein, whereas Alpha and Omicron show reduced responses (Figures 4 and S3) . These variants contain deletions and mutations either in, or in direct contact with, the H146-E156 region: Alpha contains a one-residue deletion in position 144 (Y144Δ); Delta has a three-residue deletion from position 156 to 158 (E156Δ, 

The rearrangements induced by LA removal are not restricted to the regions near the FA site, and they propagate as far as the V622-L629 segment, the furin cleavage site and the regions surrounding the FP (Figure 5 and S9) . The furin cleavage site, which is located at the S1/S2 interface, shows significant differences in response to LA removal between variants ( Figures   5 and S3 ). In Alpha, the furin cleavage site is less impacted by LA removal than in the original protein. In contrast, in Delta, Delta plus and Omicron, the furin cleavage site is more affected ( Figure S14 ). Alpha, Delta, Delta plus, and Omicron contain residue substitutions close to the furin cleavage site. The proline residue in position 681 is mutated to histidine in Alpha and Omicron (P681H) and to an arginine in Delta and Delta Plus (P681R). In addition to P681H, in Omicron, asparagine 679 is also replaced by lysine (N679K). The extra positively charged residues near the cleavage site (P681R in Delta and Delta plus and N679K in Omicron) strengthen the allosteric connection to the FA pocket ( Figure S14 ). The addition of flanking positively charged residues to the P681-R685 stretch has been suggested to improve proteolytic processing (59) .

The allosteric coupling between the FA site and V622-L629 is substantially weaker in the variants containing the D614G mutation (Delta, Delta Plus and Omicron): the structural changes are smaller than the original spike ( Figures 5 and S14 ). D614G significantly reduces signal propagation and allosteric coupling between the FA site and V622-L629. The D614G mutation is located at the interface between two monomers, where it disrupts the trans-interface salt-bridge and hydrogen bond networks (e.g. (60, 61) ) and alters the dynamics of this region (e.g. (29) ). The D614G substitution has been shown to increase transmission, infectivity and viral fitness (e.g. (60, (62) (63) (64) (65) (66) (67) ). Our results here indicate that it may have a role in limiting the allosteric effects of the FA site.

The regions surrounding the fusion peptide, notably D808-S813 and the FPPR, are also affected by LA removal (Figures 5 and S3 ). The response of D808-S813 is similar between all simulated variants ( Figure S14 ). This segment is located upstream of the FP and immediately preceding the S2' protease cleavage site. The S2' site is essential for infection (e.g. (10)), and its cleavage 15 is mediated by the transmembrane protease serine 2 (TMPRSS2) after spike binding to ACE2 (e.g. (10, 68) ). Finally, the FPPR shows reduced response in all the variants compared to the original protein ( Figure S14 ). This diminished response in Alpha, Delta, Delta plus and Omicron indicates a weakened allosteric connection of the FPPR to the FA site. The results here show that the mutations occurring in or close to the furin cleavage site and V622-L629, such as D614G (in Delta, Delta plus and Omicron), H655Y and N679K (in Omicron), P681H

(in Alpha and Omicron) or P681R (in Delta and Delta plus), alter the allosteric networks connecting the FA site to the regions surrounding the FP, particularly the FPPR.

Our findings show that SARS-CoV-2 variants differ significantly in their allosteric response to fatty acid binding (Table 1 ). These differences are of potential functional importance in the regulation of viral infectivity by LA. It may also have implications for efforts to target the FA site with natural, repurposed or specifically designed ligands (17) .

The allosteric connections in Alpha are generally similar to the original ('Wuhan') spike protein, except for the RBM and S71-R78 (Table 1) . Delta and Delta plus exhibit significant differences in the responses of the NTD, furin cleavage site and V622-L629 but not in the RBM (Table 1) . Omicron, the most infectious variant simulated, displays significant changes in the response of the NTD, RBM, furin cleavage site and V622-L629 compared to the original spike protein (Table 1 ). In Omicron, S71-R78, H146-E156 and V622-L629 exhibit a weaker connection to the FA site, whereas the L249-G257 region, RBM and the furin cleavage site show stronger coupling to the FA site.

In Delta, Delta plus and Omicron, the allosteric connection between the FA site and the furin cleavage site is increased compared to the original spike, whereas the link to V622-L629 is diminished. Also, in all variants, FPPR displays weaker connection to the FA site (Table 1) .

This indicates that mutations affect how the structural rearrangements are propagated to the FPPR. insertions, and deletions affect the amplitude of the structural response of specific regions in the S1 and S2 subunits and alter the rates at which these rearrangements propagate to them.

While some mutations (such as L18F, T19R, G142D, E156Δ-F157Δ, T478K or P681H/R) strengthen the links to the FA pocket, others (such as H69Δ-V70Δ, Y144Δ and D614G) abate them.

The coupling of the FA site to the NTD, in particular to the S71-R78, H146-E156 and L249- Our results demonstrate that D-NEMD simulations are a valuable tool to study allostery, identify and explore the allosteric networks operating within the spike, and predict the impact of future mutations in these pathways. D-NEMD simulations provide an effective approach to characterise allosteric effects (e.g. (29, 30) ), effects of pH changes (26) and other functionally important properties and will be helpful to investigate further emerging variants. We note that the simulations here compare spike proteins without glycans; as discussed above and elsewhere (29) , the presence of glycans (covering the outside of the protein) is unlikely to qualitatively alter the internal mechanical response of the protein. The differences in allosteric response to LA between variants revealed by D-NEMD simulations here may have functional relevance and should be investigated by experiments.

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

Structural and functional basis of SARS-CoV-2 entry by using human ACE2

Neuropilin-1 is a host factor for SARS-CoV-2 infection

Neuropilin-1 facilitates SARS-CoV-2 cell entry and infectivity

A Functional Interaction Between the SARS-CoV-2 Spike Protein and the Human α7 Nicotinic Receptor

A potential interaction between the SARS-CoV-2 spike protein and nicotinic acetylcholine receptors

Structure, function, and antigenicity of the SARS-CoV-2 spike glycoprotein

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

Distinct conformational states of SARS-CoV-2 spike protein

Proteolytic activation of SARS-CoV-2 spike protein

Characterisation of the transcriptome and proteome of SARS-CoV-2 reveals a cell passage induced in-frame deletion of the furin-like cleavage site from the spike glycoprotein

SARS-CoV-2 and bat RaTG13 spike glycoprotein structures inform on virus evolution and furin-cleavage effects

Free fatty acid binding pocket in the locked structure of SARS-CoV-2 spike protein

Bat and pangolin coronavirus spike glycoprotein structures provide insights into SARS-CoV-2 evolution

Structural analysis of full-length SARS-CoV-2 spike protein from an advanced vaccine candidate

Molecular dynamics of spike variants in the closed conformation; their RBD interfaces, fatty acid binding and furin cleavage sites

Molecular simulations suggest vitamins, retinoids and steroids as ligands of the free fatty acid pocket of the SARS-CoV-2 spike protein

Beyond shielding: the roles of glycans in the SARS-CoV-2 spike protein

AI-driven multiscale simulations illuminate mechanisms of SARS-CoV-2 spike dynamics

A multiscale coarse-grained model of the SARS-CoV-2 virion

The flexibility of ACE2 in the context of SARS-CoV-2 infection

Biomolecular Simulations in the Time of COVID19, and After

Distant residues modulate conformational opening in SARS-CoV-2 spike protein

SARS-CoV-2 simulations go exascale to predict dramatic spike opening and cryptic pockets across the proteome

SARS-CoV-2 Spike Protein Mutations and Escape from Antibodies: A Computational Model of Epitope Loss in Variants of Concern

#COVIDisAirborne: AI-Enabled Multiscale Computational Microscopy of Delta SARS-CoV-2 in a Respiratory Aerosol

A glycan gate controls opening of the SARS-CoV-2 spike protein

Dynamic Network Modeling of Allosteric Interactions and Communication Pathways in the SARS-CoV-2 Spike Trimer Mutants: Differential Modulation of Conformational Landscapes and Signal Transmission via Cascades of Regulatory Switches

The fatty acid site is coupled to functional motifs in the SARS-CoV-2 spike protein and modulates spike allosteric behaviour

Structural insights in cell-type specific evolution of intra-host diversity by SARS-CoV-2

Thought-experiments by molecular dynamics

Computer simulation in material science

Non-equilibrium by molecular dynamics: a dynamical approach

Dynamical nonequilibrium molecular dynamics reveals the structural basis for allostery and signal propagation in biomolecular systems

Estimated transmissibility and impact of SARS-CoV-2 lineage B.1.1.7 in England

Potential transmission chains of variant B.1.1.7 and co-mutations of SARS-CoV-2

Emergence of a new SARS-CoV-2 variant in the UK

What We Know About the Science

Omicron SARS-CoV-2 variant: a new chapter in the COVID-19 pandemic

Omicron emerges

Mutational and phylogenetic analyses of the two lineages of the Omicron variant

SARS-CoV-2 transmission dynamics in South Africa and epidemiological characteristics of the Omicron variant

SARS-CoV-2 Omicron strain exhibits potent capabilities for immune evasion and viral entrance

Omicron severity: milder but not mild

Identification of the initial steps in signal transduction in the α4β2 nicotinic receptor: insights from equilibrium and nonequilibrium simulations

A general mechanism for signal propagation in the nicotinic acetylcholine receptor family

F508del disturbs the dynamics of the nucleotide binding domains of CFTR before and after ATP hydrolysis

Allosteric communication in class A β-lactamases occurs via cooperative coupling of loop dynamics

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

Convergent antibody responses to SARS-CoV-2 in convalescent individuals

Isolation of potent SARS-CoV-2 neutralising antibodies and protection from disease in a small animal model

Broad neutralisation of SARS-related viruses by human monoclonal antibodies

The N-terminal domain of spike glycoprotein mediates SARS-CoV-2 infection by associating with L-SIGN and DC-SIGN

A neutralising human antibody binds to the N-terminal domain of the Spike protein of SARS-CoV-2

SARS-CoV-2 proteome microarray for mapping COVID-19 antibody interactions at amino acid resolution

Potent SARS-CoV-2 neutralising antibodies directed against spike Nterminal domain target a single supersite

Role of the GTNGTKR motif in the N-terminal receptor-binding domain of the SARS-CoV-2 spike protein

Antibody escape and cryptic cross-domain stabilisation in the SARS-CoV-2 Omicron spike protein

SARS-CoV-2 spike and its adaptable furin cleavage site

Structural and functional analysis of the D614G SARS-CoV-2 spike protein variant

D614G mutation alters SARS-CoV-2 spike conformation and enhances protease cleavage at the S1/S2 junction

SARS-CoV-2 spike D614G change enhances replication and transmission

SARS-CoV-2 D614G spike mutation increases entry efficiency with enhanced ACE2-binding affinity

SARS-CoV-2 D614G variant exhibits efficient replication ex vivo and transmission in vivo

Spike mutation D614G alters SARS-CoV-2 fitness

Evaluating the effects of SARS-CoV-2 spike mutation D614G on transmissibility and pathogenicity

Tracking changes in SARS-CoV-2 spike: evidence that D614G increases infectivity of the COVID-19 virus

SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 and is blocked by a clinically proven protease inhibitor

All D-NEMD simulation data (including input and trajectories files) are openly available from the MolSSI/BioExcel COVID-19 public data repository for biomolecular simulations (https://covid.molssi.org/simulations/).