The Process of Wrapping Virus Revealed by a Force Tracing Technique and Simulations

Viral entry into the host cell is the first step of virus infection; however, its dynamic process via endocytosis remains largely elusive. Here, the force tracing technique and single particle simulation are combined to investigate the invagination of single human enterovirus 71 (HEV71, a positive single‐stranded RNA virus that is associated with hand, foot, and mouth disease) via cell membranes during its host cell entry. The experimental results reveal that the HEV71 invaginates in membrane vesicles at a force of 58 ± 16 pN, a duration time of 278 ± 68 ms. The simulation further shows that the virus can reach a partially wrapped state very fast, then the upper surface of the virus is covered by the membrane traveling over a long period of time. Combining the experiment with the simulation, the mechanism of membrane wrapping of virus is uncovered, which provides new insights into how the cell is operated to initiate the endocytosis of virus.


Construction of the recombinant bacmid (BAC-P1-3CD).
Based on previous studies [1] , we construct the recombinant baculovirus (Ac-P1-3CD) carrying P1 protein and 3CD protein by Bac-to-Bac baculovirus expression system. The Polymerase Chain Reaction (PCR) product of P1 was inserted into downstream of the pH promoter of the transfer vector pFastBacDual (Invitrogen), then 3CD was inserted to the downstream of the p10 promoter to construct the donor plasmid pFastBac-P1-3CD. The recombinant bacmid (Bac-P1-3CD) ( Figure S1) was generated through translocating the donor plasmid to the AcMNPV bacmid (Ac-BAC) in the presence of transposases. The verified bacmid DNA was transfected into Sf9 cells to obtain the recombinant virus (Ac-P1-3CD).

The examination of VLPS.
HEV71 VLPs were produced by infecting Hi5 cells with the recombinant baculovirus Ac-P1-3CD. The infected cells were harvested at 72h post-infection, and the cell lysate was collected to purify the VLPs by sucrose density gradient centrifugation. The VLPs were examined by immunofluorescence assay (IFA) transmission and electron microscope (TEM). The green fluorescence particles were distributed in the cytosol and on the membrane surface of VLPincubated Vero cells ( Figure S2c), which was consistent with the result of the immunofluorescence assay of HEV71 virus ( Figure S2a). To further confirm the HEV71 VLPs infection in Vero cells, we used chlorpromazine (CPZ) as an inhibitor to inhibit the internalization process of the HEV71 viruses and VLPs. Figure S2b and S2d show that VLPs 4 endocytosis was inhibited largely. In the control experiment, no green fluorescence particle was observed in the cytosol of cells ( Figure S2e).

Engaging the tip
African green monkey kidney (Vero) cells, which can be infected by HEV71, were used to study the mechanism of virus entry. In this particular study, we prepared a dense monolayer of Vero cells on the glass surface functionalized with (3-Aminopropyl) triethoxysilane (APTES).
Usually, the cells were cultured for 2 or 3 days on the glass slide. Before performing Force Tracing, a CCD camera was used to locate the AFM tip on the top of the Vero cell monolayer ( Figure S3a). To determine whether the tip can be approached to the cell surface with high precision, we approached the AFM tip to contact point at a rate of about 50 nm/s repeatedly ( Figure S3d). The results show that relative position between the cell and the tip has remained within ± 7nm range ( Figure S3e), which illustrates that this method engage the AFM tip to the living cell surface with high precision. Then, we engaged the functionalized AFM tip, and we did force-distance curves on the living cell so that we can locate the contact point between the AFM tip and cell surface, as illustrated in Figure S3f. We took advantage of the feedback system to approach the AFM tip to the contact point between the AFM tip and cell surface at a rate of about 50 nm/s and then turned off the feedback system ( Figure S3g I). As a result, the AFM tip suspended on the cell surface, as shown in Figure S3c and S3g II. Upon viral internalization, the cantilever of the AFM tip bent downward, and the force signal was recorded via a Force Tracing curve, as shown in Figure S3d

Computational Model and Methodology
Simulations were based on dissipative particle dynamics (DPD), a mesoscopic coarse-grained simulation method for soft matter and biomembrane systems [2][3][4][5][6] . In DPD simulation, a small group of atoms is treated as a single bead located at the center of mass of the group. Beads i and j interact with each other via a pairwise force consisting of a conservative force representing excluded volume effect, a dissipative force representing viscous drag between moving beads, and a random force representing stochastic impulse. Both and act together as a 8 thermostat for beads. Similar to molecular dynamics simulation, time evolution is also governed by the Newton's equation of motion. The total force on bead i can be expressed as where is the maximum repulsive force, the distance, ̂ the unit vector and the relative velocity between beads i and j; denotes a random number with zero mean and unit variance, is a normalized distribution function, being the cutoff radius; and are parameters related to each other as , being the unit of energy. The standard values and are used in our study [3][4][5] . The mass, length, and time scales are all normalized in the DPD simulations, with the unit of length taken to be the cutoff radius , the unit of mass to be that of the solvent bead, and the unit of energy to be . All other quantities are expressed in terms of these basic units. The reduced DPD unit can be converted to SI unit by examining the membrane thickness and the lipid diffusion coefficient. The simulated value of the bilayer thickness is and the effective time scale of simulation can be determined from the simulated lateral diffusion constants of the lipid bilayer [4] . The DPC bilayer thickness is about 4 nm with a diffusion coefficient around , [7] by comparison with typical experimental values, it can be shown that one DPD length unit corresponds to approximately 0.8 nm in physical unit and the time unit to ps. The time step is taken as . All simulations are based on LAMMPS package [8] .
The dimension of the simulation box is . Periodic boundary conditions are adopted in three directions.The system contains a total of 5,329,858 particle, 5,113,858 water beads, 187,200 lipid beads and 28,800 nanoparticle beads, with a particle density of approximately 3 [4] . The solvent beads are not shown for clarity. The simulation system consists of the bilayer membrane with 14,400 lipid molecules (50% of them are assumed as receptors), a sphere viruswith stiff hydrophobic core coated by a single layer of hydrophilic ligands. The lipid molecule is represented by a coarse-grained model proposed by Groot and Rabone [2] , as shown in S5, with 3 lipid hydrophilic head beads and 10 lipid hydrophobic tail beads. The parameters 9 for repulsive interactions between different types of beads are detailed in Table 1. Within a lipid molecule, an elastic harmonic force, ̂ is used to connect two consecutive beads, where and are the spring constant and equilibrium bond length respectively. Here we use , for lipid molecule [9] .
Hydrophilic ligands are linked to each nanoparticle core bead with , . The bending resistance of the lipid chain is represented as an additional force due to a harmonic constraint on two consecutive bonds,

[ ]
where , and are the bending constant, inclination angle and equilibrium angle respectively. For three consecutive lipid tail beads or three consecutive lipid head beads in a lipid molecule, we take and ; for the last head-bead and the top tail-beads and ; for the bottom two consecutive head beads and the first bead in each tail and [3,4] .  For clarity, the water is not shown in all figures.