key: cord-0987140-t9rzicqw
authors: Kuzmin, Alexander; Orekhov, Philipp; Astashkin, Roman; Gordeliy, Valentin; Gushchin, Ivan
title: Structure and dynamics of the SARS-CoV-2 envelope protein monomer
date: 2021-03-19
journal: bioRxiv
DOI: 10.1101/2021.03.10.434722
sha: 328e60b3e87b6a31ac436dfbb07b799d068ad20a
doc_id: 987140
cord_uid: t9rzicqw

Coronaviruses, especially SARS-CoV-2, present an ongoing threat for human wellbeing. Consequently, elucidation of molecular determinants of their function and interaction with host is an important task. Whereas some of the coronaviral proteins are extensively characterized, others remain understudied. Here, we use molecular dynamics simulations to analyze the structure and dynamics of the SARS-CoV-2 envelope (E) protein (a viroporin) in the monomeric form. The protein consists of the hydrophobic α-helical transmembrane domain (TMD) and amphiphilic α-helices H2 and H3, connected by flexible linkers. We show that TMD has a preferable orientation in the membrane, while H2 and H3 reside at the membrane surface. Orientation of H2 is strongly influenced by palmitoylation of cysteines Cys40, Cys43 and Cys44. Glycosylation of Asn66 affects the orientation of H3. We also observe that the E protein both generates and senses the membrane curvature, preferably localizing with the C-terminus at the convex regions of the membrane. This may be favorable for assembly of the E protein oligomers, whereas induction of curvature may facilitate budding of the viral particles. The presented results may be helpful for better understanding of the function of coronaviral E protein and viroporins in general, and for overcoming the ongoing SARS-CoV-2 pandemic.

Coronaviruses (CoVs) (order Nidovirales, family Coronaviridae, subfamily Coronavirinae) 35 are enveloped viruses with a positive sense, single-stranded RNA genome of ~30 kb, one of the largest 36 among RNA viruses [1] . CoVs infect birds and mammals, causing a variety of fatal diseases. They can 37 also infect humans and cause diseases ranging from the common cold to acute respiratory distress have seen rapid progress since the beginning of the pandemic [5] . Whereas most of the key information 48 was obtained using experimental techniques, such as cryoelectron microscopy, X-ray crystallography 49 or NMR, computational approaches were key for some of the findings [6, 7] . Among the most notable 50 examples are detailed simulations of dynamics of the most important viral proteins [6, 8, 9] or even the 51 whole virion [10] , early generation of atomic models for all SARS-CoV-2 proteins [11] , and high-52 throughput virtual ligand screening of viral protease inhibitors [12, 13] . 53

The genomes of all coronaviruses encode four major structural proteins: the spike (S) protein, 54 nucleocapsid (N) protein, membrane (M) protein, and the envelope (E) protein [14] . The S protein is 55 involved in the host recognition, attachment and cell fusion. The N protein is involved in packaging of 56 the RNA genome and formation of the nucleocapsid. The M protein directs the assembly process of 57 virions through interactions with the other structural proteins and defines the shape of the viral envelope. 58

The E protein is possibly the most mysterious of them since it is associated with the assembly of virions, 59 effective virion transfer along the secretory pathway as well as a reduced stress response by the host 60 cell. Generally, it promotes virus fitness and pathogenesis [15] . Yet, experimental structure of the full-length wild type protein is not available at the moment, and the 111 influence of PTMs on it hasn't been studied. Moreover, there is little data on behavior of monomeric E 112 protein prior to its assembly into pentameric channels. In the present study, we applied molecular 113 dynamics to study the behavior of monomeric E protein from SARS-CoV-2 and identified the effects 114 of PTMs on the protein behavior. We have also observed that the protein induces curvature in the 115 membranes, and is attracted to the curved regions. These findings may be helpful in development of 116 anti-SARS-CoV-2 medications, and for understanding the function of viroporins in general. 117

Structure of the monomeric E protein 119 E protein from SARS-CoV-2 is a 75 amino acid-long protein that may be palmitoylated and 120 glycosylated in vivo. To assess the overall conformational space available to the protein, we conducted 121 first an extensive coarse-grained (CG) simulation of unmodified E protein, followed by atomistic 122 simulations of unmodified protein and CG simulations of the protein with modifications (Table S1) . 123 CG simulations are known to faithfully reproduce the major physicochemical properties of the studied 124 macromolecules while providing a considerable speedup compared to atomistic simulations [57, 58] . 125

In accordance with expectations, the simulations revealed that the protein is very flexible with 126 no particular tertiary structure ( Figure 1B ). Principal component analysis (PCA) shows that the first two 127 components describe most of the structural variation (~64%, Figure 2 ) and correspond to motions of 128 the helices H2 and H3 relative to each other and TMD near the membrane surface. 129

Atomistic simulations are considerably more computationally demanding, and thus the 130 exhaustive sampling of the conformational space can take a prohibitively long time. Consequently, we 131 simulated a number of atomistic trajectories starting from representative conformations from the CG 132 simulation. We divided the CG trajectory snapshots into four clusters, and used the centroids of the 133 clusters as the starting structures for atomistic simulations. For each starting structure, we obtained six 134 trajectories of the E protein: three with the protein embedded in the model membrane containing POPC, 135 and three with the membrane mimicking the natural ERGIC membrane (50% POPC, 25% POPE, 10% 136 POPI, 5% POPS, 10% cholesterol). No qualitative differences were observed between the simulations 137 conducted in these membranes. Overall, PCA shows that the atomistic simulations correspond to the 138 CG simulation and display roughly the same conformational space available to the E protein 139 ( Figure 2B ). Conformations observed in atomistic simulations are shown in the Supporting Figure S1 . 140

Atomistic simulations also show that while the secondary structure of the E protein is largely 141 conserved, the amphipathic α-helices H2 and H3 may partially unfold, with H2 being more disordered 142 (Figures 3 and S2 ). We observed both unfolding and refolding events. Overall, this observation is in 143 agreement with NMR experiments [55, 56] . 144 145 Position of the E protein elements relative to the membrane 146 Figure 4 shows the average positions of the secondary structure elements of the E protein 147 relative to the membrane surface. In all of the simulations, the TMD remained embedded in the 148 membrane. H2 is deeply buried in the lipid headgroup region, whereas H3 is slightly removed from the 149 membrane border, while still remaining in contact with it. In some atomistic trajectories, partial 150 unbinding of H3 from the membrane is observed ( Figure S3 ). 151

Interestingly, TMD, despite being a single transmembrane α-helix, has a preferable orientation (Table 1) . 156 H2, as an amphipathic helix, also has a preferred orientation ( Figure 6 ). Palmitoylation of the 157 three cysteines, Cys40, Cys43 and Cys44, in different patterns changes the physicochemical properties 158 of the helix and leads to its rotation around its axis ( Figure 6 , Table 1 ). The strongest effect on H2 159 orientation is observed when Cys40 and Cys44 are palmitoylated simultaneously: the helix is rotated (Table 1) . 163

Position of H3 was not significantly affected by palmitoylation. Yet, glycosylation of Asn66 164 had a pronounced effect on its dynamics (Figure 7 ). In the unmodified protein, we observed two major 165 orientations of H3: the first one with the hydrophobic residues Val62 and Leu65 facing the membrane, 166 and the less frequent orientation almost completely opposite to it, with H3 stacking with H2 while being 167 slightly above it (Figure 7 ). Glycosylation resulted in abolishment of the second orientation, presumably 168 due to the potential steric conflict between the sugar moiety and H2 ( Figure 7A,D) ; the helix was also 169 slightly rotated in the most frequent orientation ( Figure 7A,E) . 170

In all of the conducted simulations, we observed induction of curvature by the E protein: the 172 membrane bends towards the side where the C-terminus is located ( Figure 8A ). The effect is also 173 observed in larger systems containing four E protein monomers in opposite orientations ( Figure S4 ) and 174 in atomistic simulations ( Figure S5 ). Presumably, the curvature is induced by the amphipathic helices 175 that embed into the adjacent leaflet and expand it. 176

To check whether the curvature is indeed induced by H2 and H3, we conducted additional 177 simulations of artificial proteins consisting of only TMD or only H2 and H3 ( Figure 8B ,C). Isolated 178 TMD was tilted in a way similar to that observed in the simulations of the full-length protein. The 179 membrane was perturbed and thinned near the α-helix ( Figure 8B ), presumably because of the polar 180 residues on the respective sides (Glu8, Thr9, Thr11, Asn15, Ser16 at the N-terminal side, Thr30, Thr35 181 at the C-terminal side). Isolated H2 and H3 curved the membrane in the same way as the full-length 182 protein ( Figure 8C ). Thus, we conclude that the structural elements responsible for curvature induction 183

by the E protein are the amphipathic helices of the C-terminal domain. 184

Having observed the induction of curvature by the E protein, we were also interested to check 186 whether it has a preferable position in membranes that are already curved, such as the native ER, Golgi 187 and ERGIC membranes, especially during the budding of VLPs. As a test system, we used artificially 188 buckled membranes [59-62]. Irrespective of the starting positions, E protein monomers redistribute in 189 the membranes so that the C-termini localize to the convex regions ( Figure 9 ). The effect was observed 190 both in the membranes buckled in a single direction (non-zero mean curvature, zero Gaussian curvature, 191 Figure 9A ) and in the membranes buckled in both directions (positive Gaussian curvature, Figure 9B ). 192

Thus, we conclude that the monomeric E protein is curvature-sensitive. Our results show that monomeric SARS-CoV-2 envelope protein has rich conformational 205 dynamics strongly affected by post-translational modifications. The protein is organized as an α-helical 206 TMD and two amphipathic α-helices H2 and H3, flanked by short disordered N-and C-termini. 207

Whereas TMD is rigid and remains α-helical throughout the trajectories, helices H2 and H3 may 208 partially unfold. TMD, H2 and H3 mostly move freely relative to each other, so the monomeric E protein 209 can be considered an intrinsically disordered protein. Yet, all of its α-helices have preferred orientations 210 relative to the membrane. 211

The TM α-helix of the E protein is relatively long (28 amino acids, ~43 Å), and there is a 212 mismatch between the length of its hydrophobic segment and the thickness of the hydrophobic region 213 of the relevant membranes. Tilting of TM helices is a common mechanism for accommodating such 214 mismatch [64-68]. Accordingly, we find that E protein TMD is tilted at 25-40°, similarly to TMD in 215 pentameric E protein [56] and in other viroporins (Table S2) . It also has a preferred azimuthal rotation 216 angle, similarly to WALP peptides [69,70], for which the results obtained with MD simulations were 217 found to correspond well to those obtained in experiments [71] . However, the rotation angle of TMD 218 in a free E protein monomer is opposite to that in the E protein pentamer. This is likely a consequence 219 of the polar residues such as Glu8, Thr9, Thr11, Asn15, Ser16, Thr30 and Thr35 preferably facing the 220 solvent in the monomer and interior of the channel in the pentamer. 221

One of the most potentially important findings is a strong dependence of the E protein structure 222 on post-translational modifications. Previously, it was shown that palmitoylation is important for E 223 protein stability and overall assembly of VLPs, whereas the role of glycosylation is more elusive [17] . shorter (residues 8-38) and did not include the residues that could be modified (Mandala et al, 2020) . 233

Previous computational studies of the E protein also did not focus on the effects of PTM [11, 73, 74] . In 234 this work, we found that the average orientation of H2 is strongly dependent on palmitoylation pattern, 235

as the acyl chains act as anchors on the respective H2 cysteines and bring them closer to the membrane 236 core. On the other hand, positioning of H3 is affected by glycosylation as the glycan acts as a buoy on 237 H3 and prevents its interaction with H2. However, most or all of the E proteins in vivo are probably not 238 glycosylated [27], so the possible role of this modification remains unclear. While we did not study the 239 E protein in pentameric form, we believe that PTMs are likely to elicit effects in the assembled 240 oligomers similar to those that we observe in monomers. 241

In the last part of our work, we focused on interactions of the E protein with curved membranes. Here, we have found that the SARS-CoV-2 E protein can generate membrane curvature, and 254 this function can be ascribed to the amphiphilic C-terminal domain (α-helices H2 and H2). Such 255 viroporin-generated curvature may stabilize the budding viral particle and promote its formation [75] . 256

We have also found that the monomeric E protein can act as a curvature sensor and localize with the C-257 terminus at the convex regions of the membrane. Given that the C-terminus of E is oriented towards the 258 cytoplasm [27], the protein is likely to localize at the VLP budding sites and promote VLP budding. 259

Concentration in these curved areas may promote formation of pentameric channels. The assembled 260 channels are also likely to be curvature-sensitive due to their umbrella-like shape [11, 56] . On the other 261

hand, E protein is expected to be depleted at the concave inner surface of the VLP, in agreement with 

The authors declare no competing interests. 290 291

Supplementary information is available. 293

As a starting structure for simulations of the monomeric SARS-CoV-2 E protein, we used the 296 In all simulations, the N-and C-termini, residues Lys, Arg, Asp and Glu, and lipids POPS and 305 POPI were charged. The membranes were solvated with water; counter ions were added to neutralize 306 the systems. The simulations were performed using periodic boundary conditions. All systems were 307 energy minimized using the steepest descent method, equilibrated and simulated using GROMACS the TMD helix axis and the normal to the membrane (axis Z). The rotational angle was defined as the 335 angle between the Cα radial vector of a reference residue (Phe23, Cys43, Asn66) and the X-Z plane; the 336 helix was aligned so that its axis was in the X-Z plane (Figures 5-7) . We used the Ward's method from 337

MDTraj [101] to group the dataset into 4 clusters based on pairwise RMSD of coordinates of backbone 338 particles. Density distributions of TMD, H2 and H3 atoms were calculated using the density tool from 339 GROMACS. The secondary structure in the E protein was monitored using the Timeline plugin (version Average probability of observing the α-helical structure for each residue is shown. TMD remains fully 725 α-helical, whereas H2 and H3 can be sometimes disordered (H2 more often compared to H3). 726 

Coronaviruses: An Overview of Their Replication and Pathogenesis

Coronaviridae Study Group of the 351 International Committee on Taxonomy of Viruses, The species Severe acute respiratory 352 syndrome-related coronavirus : classifying 2019-nCoV and naming it SARS-CoV-2

Origin and evolution of pathogenic coronaviruses

Characteristics of SARS-CoV-2 and COVID-19

Structural biology in the fight 361 against COVID-19

Fighting COVID-19 Using Molecular Dynamics 364

COVID19 -Computational Chemists Meet the Moment

Simulations Go Exascale to Capture Spike Opening and Reveal Cryptic Pockets Across the 371

Beyond Shielding: 374 The Roles of Glycans in the SARS-CoV-2 Spike Protein

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

Modeling of Severe Acute Respiratory Syndrome Coronavirus

CoV-2) Proteins by Machine Learning and Physics-Based Refinement

Crowdsourcing drug discovery for pandemics, 383

COVID 402 Moonshot: Open Science Discovery of SARS-CoV-2 Main Protease Inhibitors by Combining

Computational Simulations, and Machine

The Molecular Biology of Coronaviruses

Severe Acute Respiratory Syndrome Coronavirus Envelope Protein Ion Channel 410

Activity Promotes Virus Fitness and Pathogenesis

The Coronavirus E Protein: Assembly and Beyond, Viruses. 4 413

Coronavirus envelope protein: current knowledge

Palmitoylation of the cysteine-rich endodomain of the SARS-coronavirus spike glycoprotein is 418 important for spike-mediated cell fusion

Structural basis for the membrane association of ankyrinG via 422 palmitoylation

Protein Palmitoylation 424 and Its Role in Bacterial and Viral Infections

Envelope Protein Palmitoylations Are 427 Crucial for Murine Coronavirus Assembly

Biochemical and functional characterization 430 of the membrane association and membrane permeabilizing activity of the severe acute 431 respiratory syndrome coronavirus envelope protein

Importance of Conserved Cysteine 434 Residues in the Coronavirus Envelope Protein

Analysis of the SARS-CoV-2 spike protein 437 glycan shield reveals implications for immune recognition

Mammalian glycosylation in immunity

Subcellular location and topology of severe acute 443 respiratory syndrome coronavirus envelope protein

446 SARS-CoV-2 envelope protein topology in eukaryotic membranes

The SARS-CoV-2 envelope and membrane proteins modulate maturation and retention of the 450 spike protein, allowing assembly of virus-like particles

Identification of a Golgi Complex-Targeting Signal in 453 the Cytoplasmic Tail of the Severe Acute Respiratory Syndrome Coronavirus Envelope

TGEV corona virus ORF4 encodes a 456 membrane protein that is incorporated into virions

Coronavirus 459 envelope (E) protein remains at the site of assembly

The ER-Golgi intermediate compartment (ERGIC): in 462 search of its identity and function

Absence of E protein arrests 465 transmissible gastroenteritis coronavirus maturation in the secretory pathway

Analysis of Constructed E Gene 468 Mutants of Mouse Hepatitis Virus Confirms a Pivotal Role for E Protein in Coronavirus 469

The cytoplasmic tails of infectious bronchitis virus E and M 472 proteins mediate their interaction

SARS-CoV envelope protein 475 palmitoylation or nucleocapid association is not required for promoting virus-like particle 476 production

The envelope protein of severe acute respiratory syndrome coronavirus interacts 480 with the non-structural protein 3 and is ubiquitinated

Construction of SARS-CoV-2 Virus-Like Particles by

Bcl-xL inhibits T-cell apoptosis induced by expression of SARS 487 coronavirus E protein in the absence of growth factors

The SARS coronavirus E protein interacts with PALS1 and alters tight 491 junction formation and epithelial morphogenesis

Comparing 494 the binding properties of peptides mimicking the Envelope protein of SARS-CoV

CoV-2 to the PDZ domain of the tight junction-associated PALS1 protein

CoV-2 protein interaction map reveals targets for drug repurposing

Coronavirus E protein forms ion channels with functionally and structurally-518 involved membrane lipids

Structural 521 Flexibility of the Pentameric SARS Coronavirus Envelope Protein Ion Channel

MERS coronavirus envelope 524 protein has a single transmembrane domain that forms pentameric ion channels, Virus 525 Research

SARS-CoV-2 E protein is a potential ion channel that can be 527 inhibited by Gliclazide and Memantine, Biochemical and Biophysical Research 528 Communications

Hexamethylene amiloride blocks E protein ion channels and 530 inhibits coronavirus replication

Structure and Inhibition of the SARS Coronavirus Envelope 534 Protein Ion Channel

Conductance and 537 amantadine binding of a pore formed by a lysine-flanked transmembrane domain of SARS 538 coronavirus envelope protein

Structure and drug binding of the SARS-CoV-2 envelope protein transmembrane domain in 542 lipid bilayers

Chapter Eight -Targeting the Channel Activity of Viroporins

Advances in Protein Chemistry and Structural Biology

Viroporins: structure and biological functions

Amantadine and rimantadine for 550 preventing and treating influenza A in adults

Severe acute 554 respiratory syndrome coronavirus E protein transports calcium ions and activates the NLRP3 555 inflammasome

Structure of a Conserved Golgi Complex-targeting 557 Signal in Coronavirus Envelope Proteins

Structural model of the SARS coronavirus E channel in LMPG 560 micelles

Lipids on the move: Simulations of membrane 563 pores, domains, stalks and curves

Effects of Coarse Graining and Saturation of Hydrocarbon Chains on 566 Structure and Dynamics of Simulated Lipid Molecules

Buckling Under Pressure: Curvature-Based Lipid 569 Segregation and Stability Modulation in Cardiolipin-Containing Bilayers

Computing Curvature Sensitivity of 572 Biomolecules in Membranes by Simulated Buckling

Curvature-Induced Sorting of 575 Lipids in Plasma Membrane Tethers, Advanced Theory and Simulations

Entropic forces drive clustering and 578 spatial localization of influenza A M2 during viral budding

Understanding COVID-19 via comparative analysis of dark proteomes of SARS-582

CoV-2, human SARS and bat SARS-like coronaviruses

Tilt Angle of a Trans-membrane Helix is Determined by Hydrophobic 585 Mismatch

Peptides in Lipid Bilayers: A Systematic Investigation of Hydrophobic Mismatch

On the Orientation of a Designed 591 Transmembrane Peptide: Toward the Right Tilt Angle?

The conformation of the pore region 594 of the M2 proton channel depends on lipid bilayer environment

Lipid Bilayer Topology of the 597

Transmembrane α-Helix of M13 Major Coat Protein and Bilayer Polarity Profile by Site-598

Orientation and dynamics of transmembrane peptides: the power of 601 simple models

Revisiting Hydrophobic Mismatch with Free Energy Simulation Studies of 604

Transmembrane Helix Tilt and Rotation

Hydrophobic mismatch of mobile 607 transmembrane helices: Merging theory and experiments

Stochastic palmitoylation of accessible cysteines in membrane proteins 612 revealed by native mass spectrometry

Structural insight into the role of novel SARS-CoV-2 E protein: A 615 potential target for vaccine development and other therapeutic strategies

618 Computational Study of the Ion and Water Permeation and Transport Mechanisms of the 619 SARS-CoV-2 Pentameric E Protein Channel

Membrane curvature at a glance

Interplay of proteins and lipids in generating membrane 624 curvature

Thermodynamics and Mechanics of Membrane 627 Curvature Generation and Sensing by Proteins and Lipids

Membrane curvature in cell biology: An integration of 630 molecular mechanisms

The Hydrophobic Insertion Mechanism of 633

Membrane Curvature Generation by Proteins

A general 636 amphipathic α-helical motif for sensing membrane curvature

Mechanism of Membrane Curvature Sensing by Amphipathic 639 Helix Containing Proteins

Antimicrobial peptides and induced membrane curvature: 642 Geometry, coordination chemistry, and molecular engineering, Current Opinion in Solid State 643 and Materials Science

Influenza Virus M2 Protein Mediates ESCRT-645 Independent Membrane Scission

Protein-ligand binding with the coarse-grained Martini model, 649 Nature Communications

A 651 series of PDB-related databanks for everyday needs

Dictionary of protein secondary structure: Pattern recognition of 654 hydrogen-bonded and geometrical features

Computational 657 Lipidomics with insane: A Versatile Tool for Generating Custom Membranes for Molecular 658 Simulations

The PyMOL molecular graphics system

Going 662 Backward: A Flexible Geometric Approach to Reverse Transformation from Coarse Grained to 663 Atomistic Models

CHARMM-GUI: A web-based graphical user interface for 666 CHARMM

GROMACS: 668 High performance molecular simulations through multi-level parallelism from laptops to 669 supercomputers

Canonical sampling through velocity rescaling

A unified formulation of the constant temperature molecular dynamics methods

Polymorphic transitions in single crystals: A new molecular 675 dynamics method

Parameterization of Palmitoylated Cysteine

Geranylgeranylated Cysteine, and Myristoylated Glycine for the Martini Force Field, 678

Extending the Martini Coarse-Grained Force Field to N-Glycans

CHARMM36m: an improved force field for folded and intrinsically disordered 684 proteins

LINCS: A linear constraint solver for molecular 686 simulations

JURECA: Modular supercomputer at Jülich Supercomputing Centre, 689 Journal of Large-Scale Research Facilities JLSRF

VMD Visual molecular dynamics

MDTraj: A Modern Open Library for the 695 Analysis of Molecular Dynamics Trajectories

Computational analysis of local membrane properties

) protein monomer with possible post-706 translational modifications. (A) Schematic model showing a fully palmitoylated and glycosylated E 707 protein. Transmembrane domain (TMD) is shown in blue and amphipathic helices H2 and H3 are shown 708 in green and red

Positions of the transmembrane helix were aligned for clarity; helices H2 and H3 are 710 mobile

Comparison of the E protein conformations observed in atomistic and coarse-grained 713 simulations using principal component analysis. (A) The scree plot for the top ten PCA eigenvalues

First two components describe ~64% of structural variations. (B) Comparison of the conformational 715 ensembles observed in AA (colored) and CG (gray) simulations projected onto PC1 and PC2

Starting conformations for AA simulations are shown 717 are labelled with stars. Trajectories from the first, second, third and fourth sets of AA simulations are 718 shown in red, blue, green and orange, respectively

Conformational changes associated with PC2. The structures are colored from blue to red according 720 to the PC projection value. Approximate membrane position is shown with lines

Average positions of TMD, H2 and H3 relative to the membrane in all atom 729 simulations. Average positions of lipid phosphate groups are shown using brown lines

TMD, H2 and H3 backbone atoms' positions are shown in blue, green and red, respectively

Figure 5. Orientation of TMD in POPC bilayer in coarse-grained (CG) and all atom (AA)

A) Distributions of the tilt angles. B) Distributions of the axial rotation angles. Vertical 736 lines indicate average values. C) Definitions of the tilt (α) and axial rotation (β) angles

Figure 6. Effects of palmitoylation on orientation of the helix H2 relative to the membrane

Rotation of Cys43 relative to the membrane plane (Y) viewed from the N-terminus is analyzed

B) Distributions for the Cys43 rotation angles relative to the membrane plane for different PTMs

Vertical lines indicate average values. (C) Schematics showing the H2 orientation with helical wheel 744 projections for selected variants. Palmitoylation affects the orientation of H2 because the respective 745 side chain becomes more hydrophobic

Effect of Asn66 glycosylation on orientation of the helix H3 relative to the membrane 750 in coarse-grained simulations

A) Distributions for the Asn66 rotation angle relative to the membrane plane 752 for unmodified and glycosylated variants. Vertical lines indicate average values

Representative conformations for ψ ≈ 0° and ψ ≈ 180°. (D) and (E) Schematics showing the H3 754 orientation with helical wheel projections for unmodified and glycosylated variants

Figure 8. Induction of curvature by the E protein monomer in coarse grained simulations

Upward displacement of each membrane boundary is shown in red, and downward displacement is 761 shown in blue. (A) Induction of curvature by the full-length E protein. (B) Membrane deformation by 762 an isolated TMD. The membrane is thinned around the TMD, but no buckling is observed

Membrane deformation by isolated H2 and H3 helices in coarse grained simulation. The membrane is 764 bent towards the α-helices H2 and H3. Each panel shows an exemplary protein position