The intrinsically disordered nature of the CVB3 VP4 protein.

(A) Cross-sectional view of the CVB3 capsid (rendered from PDB 4GB3 [23]) showing the interior location of the VP4 protein (green). Other structural proteins VP1, VP2, and VP3 are colored gray, cyan, and pink, respectively. (B) Crystal structure of the basic capsid subunit formed by VP1–VP4 (PDB 4GB3 [23]). Residues 12–24 of VP4 are unresolved. A topology diagram highlights the myristoylation site and a hypothesized working model for myristoyl-enabled VP4 function. (C) Sequence alignment of VP4 from six viruses that require myristoylation for function. (D) Predicted structural states of the six VP4 proteins plotted on the Das-Pappu diagram (region 1: weak polyampholytes/polyelectrolytes). (E) Circular dichroism spectrum of CVB3 VP4 in phosphate buffer (pH 7.0). (F, I) Representative conformational snapshots of VP4 without (F) and with (I) the myristoyl modification. VP4 residues 2–68 are colored green; the myristoyl group is red. (G, J) Conformational ensembles projected onto a 2D plane defined by the end-to-end distance and radius of gyration (Rg). (H, K) Per-residue RMSF and the total non-helical secondary-structure propensity (Pn) derived from the simulations. Shaded areas represent the standard deviation from three independent 1 µs MD replicates.

MD simulations of monomric VP4 binding to membrane.

(A) Schematic of CGMD simulation setup. VP4 protein is colored green with myristoylation-modificated Glycine colored red. (B) Time-dependent z-coordinates of the COM for the myristoyl group and the VP4. Shaded area represents the standard deviation from 3×10 µs CGMD simulations for each condition. The dashed line indicates the average z-coordinate of the COM for the polar headgroups of POPC lipids in the upper leaflet. Snapshots taken at 2.5 µs intervals from a representative 10 µs trajectory are projected onto the XZ plane to illustrate the binding process. (C) Representative structures of VP4 in its membrane-bound state, showing minimal separation from the membrane. (D) Schematic illustrating the construction of two VP4 models with reduced conformational flexibility: VP4-partial and VP4-rigid. These were generated by introducing increasing amounts of intramolecular elastic restraints (blue lines). (E, F) RMSF and Rg of the VP4 models from simulations in solution, confirming their reduced conformational flexibility to the native VP4 IDP. (G) Time-dependent z-coordinates of the VP4-partial and VP4-rigid models during membrane interaction. Snapshots from the simulations are projected onto the XZ plane. (H) Time-dependent number of contacts between the membrane and the N-terminal (residues 1–20, deep green), middle (residues 21–40, green), and C-terminal (residues 41–68, light green) segments of VP4.

The myristoyl group promotes VP4 condensate formation on membranes.

(A) Confocal microscopy images of EGFP-tagged wild-type (WT) VP4 and the non-myristoylated G2A mutant expressed in HEK-293T cells. Scale bar: 20 µm. (B) Time-lapse imaging of VP4-MYR condensate dynamics on a supported lipid bilayer (SLB). Scale bar: 10 µm. (C) FRAP assay quantifying the internal mobility of VP4 condensates on SLBs. Scale bar: 10 µm and 5 µm. (D) Effect of 1,6-hexanediol (10% w/v) treatment on the stability of VP4 condensates. Scale bar: 10 µm. (E) Setup for CGMD simulations investigating VP4 condensation and membrane interaction. Six or twenty myristoylated VP4 proteins were initially placed with a minimum distance of 40 Å from the membrane; three independent 10 µs simulations were performed for each system. (F) Time-dependent analysis of the number of VP4 clusters and the size of the largest cluster during CGMD simulations. (G) Time-dependent distribution of myristoyl moieties along the membrane normal (z-axis). The axis was divided into 0.5 nm bins, and the count of myristoyl groups within each bin is plotted. (H) Representative simulation snapshot illustrating myristoyl-mediated membrane insertion and the nucleation of a VP4 condensate. (I) Control CGMD simulations of non-myristoylated VP4 proteins (three independent 10 µs replicates).

Dynamic VP4 condensates remodel the membrane to facilitate protein penetration.

(A) Autocorrelation function of the Rg for the 6-mer and 20-mer VP4 condensates. (B) Distribution of RMSF values for VP4 residues within the 6-mer and 20-mer condensates. A representative condensate structure is shown with residues color-mapped according to their RMSF. (C) Representative simulation snapshots of membrane configurations in the presence of a VP4 monomer, a 6-mer condensate, and a 20-mer condensate. (D) Quantitative analysis of membrane curvature. The local membrane topography is represented by the height function h(x, y) of the lipid midplane (contour map). Cross-sectional profiles (xz and yz planes) illustrate the membrane deformation induced by the condensates. (E) PMF profiles for the translocation of a VP4 protein from pre-formed condensates into the membrane. The thermodynamically stable end state (ES), characterized by a fully buried myristoyl moiety (red), is labeled. (F) Structural illustrations of the ES configurations for the three systems (monomer, 6-mer, and 20-mer).

Myristoylation stabilizes a multimeric pore formed by VP4 N-terminal helices

(a) Free energy profile of the coil-to-helix transition for the N-terminal 20 residues of VP4 in aqueous and lipid environments, with and without myristoylation (MYR), respectively. (b) Two-dimensional free energy landscape from umbrella sampling, showing the relationship between the helicity propensity and the membrane embedding depth of the VP4 N-terminus. Asterisks mark representative conformations at the 4 and 10 kcal/mol free energy contours. (c) Structural evolution of a model pore formed by six VP4 N-terminal helices. The initial configuration featured six vertically arranged helices, which were subsequently refined via three independent 500 ns MD simulations. Snapshots from one representative simulation are shown. (d) Range of residue Z-coordinates sampled during the MD simulations. Error bars represent the standard deviation across the three replicates. (e) Lipid density projected onto the XY plane from one 500 ns simulation. Data from the remaining two runs are shown in Fig. S8. (f) Time-dependent number of water molecules within a cylindrical volume encompassing the pore, projected along the Z-axis (n=3 independent runs). Data from the remaining two runs are shown in Fig. S9.

The structural basis for myristoylation’s multi-faceted role in enabling the disordered VP4 protein to breach the host cell membrane.

Molecular dynamics simulations performed in this study.

Sequence alignment of VP4 proteins that do not need myristoylation modification to function.

MD simulations of reduced flexibility VP4 binding to membrane.

Panel (a) shows the Martini coarse-grained molecular dynamics (CGMD) sampled bound states of native, disordered VP4 with the membrane. Panels (b) and (c) show the corresponding states for a VP4 protein with partially reduced flexibility and a rigid VP4 protein, respectively. The N-terminal myristoylation moiety is colored red. The bar plot at the bottom shows the time-dependent number of contacts between the membrane and the myristoylated GLY residue (orange) and between the membrane and the first 20 residues of VP4 (green).

The myristoylated GLY residues form hydrophobic cores within the condensate.

(a) A snapshot of the final 6-mer condensate formed on the membrane, with the myristoylation moiety colored red. The heatmap shows the average residue-residue contact number within the 6-mer condensate. The right-most panel shows the average contact number mapped onto a single VP4 protein structure. (b) The same analysis as in panel (a) for the 20-mer condensate. (c) The top five residue pairs with the highest contact frequencies. ”MG” refers to the myristoylated glycine residue.

Effects of annulling myristoylation within the formed VP4 condensate.

(a) Schematic of the simulation strategy: myristoylation sites in a pre-formed 6-mer VP4 condensate were mutated to glycine, followed by molecular dynamics (MD) investigation. (b) Time evolution of the number of VP4 clusters and the size of the largest cluster after the mutation. (c) Radius of gyration (Rg) of the condensate following the mutations. (d, e) Average residue-residue contact numbers between VP4 residues in the wild-type condensate (d) and the mutant condensate (e). These contact numbers are also mapped onto single VP4 protein structures. (f) Top five residue pairs with the highest contact numbers in the wild-type condensate and the mutant condensate, respectively.

Condensation behavior of VP4-MYR monomer in pure aqueous solution.

(a) Schematic of Martini CGMD simulations with 20 VP4-MYR monomers in pure aqueous solution (3 independent runs of 10 µs each). (b) Time evolution of the number of VP4 clusters and the size of the largest cluster in solution. (c) Radius of gyration (Rg) of condensates formed in solution compared to those formed on the membrane. (d) Distribution of RMSF values calculated from VP4 condensates formed in solution versus those formed on the membrane. A representative condensate structure is shown, with residues color-mapped according to their RMSF values. (e) Average residue-residue contact numbers calculated from VP4 condensates formed in solution and on the membrane. (f) Top five residue pairs with the highest contact frequencies in condensates formed in solution versus those formed on the membrane.

Time-dependent analysis of residue-residue contacts within the condensate.

Average residue-residue contact numbers calculated from the VP4-MYR condensates formed on the membrane at different time points (2, 4, 6, 8, and 10 µs). (a) Results for the 6-mer condensate. (b) Results for the 20-mer condensate.

Energetics of VP4 penetrating an artificially flattened membrane.

(a) Schematic illustrating the application of pressure in the xy-plane to flatten a membrane that had been remodeled by a 20-mer VP4-MYR condensate. (b) Potential of mean force (PMF) profile along the translocation reaction coordinate for a single VP4 molecule penetrating the artificially flattened membrane, compared to the unflattened membrane.

Lipid density profiles across the membrane were calculated from two 0.5 µs all-atom molecular dynamics (MD) simulations investigating the stability of the pore formed by six N-terminal VP4 helices.

Results are shown for the system (a) without and (b) with the MYR (myristoylation) modification.

The time-dependent count of water molecules within the pores was analyzed from two 0.5 µs all-atom MD simulations.

The distribution of water molecules along the membrane normal (Z-axis) is plotted for the system (a) without and (b) with the MYR (myristoylation) modification.