Direct lipid interactions control SARS-CoV-2 M protein conformational dynamics and virus assembly

Abstract

M is the most abundant structural membrane protein in coronaviruses and is essential for the formation of infectious virus particles. SARS-CoV-2 M adopts two conformations, M_short and M_long, and regulated transition between states is hypothesized to coordinate viral assembly and budding. However, the factors that regulate M conformation and roles for each state are unknown. Here, we discover a direct M-sphingolipid interaction that controls M conformational dynamics and virus assembly. We show M binds Golgi-enriched anionic lipids including ceramide-1-phosphate (C1P). Molecular dynamics simulations show C1P interaction promotes a long to short transition and energetically stabilizes M_short. Cryo-EM structures show C1P specifically binds M_short at a conserved site bridging transmembrane and cytoplasmic regions. Disrupting M_short-C1P interaction alters M subcellular localization, reduces interaction with Spike and E, and impairs subsequent virus-like particle cell entry. Together, these results show endogenous signaling lipids regulate M structure and support a model in which M_short is stabilized in the early endomembrane system to organize other structural proteins prior to viral budding.

Figure 1.. SARS-CoV-2 M protein interacts with C1P.

(a) Sphingolipid snooper assay for M-lipid interactions. Western blot of immobilized lipids probed with HA-tagged M, a 50:50 mixture of EGFP-tagged and untagged M, His- and Sumo-tagged M, or a control sample without M protein. All M constructs tested displayed strong binding to C1P and dihydroC1P as well as appreciable binding to anionic So1P and Sa1P. (b) Bead pull down assay of M-lipid interactions. Western blot of input EGFP-M (left) or EGFP (right) expressing cell lysate or eluant from control uncoated, C1P-coated, and PA-coated beads. EGFP-M protein was robustly pulled down by C1P-coated beads. (c) Colocalization of SARS-CoV-2 assembly sites with CERK. HEK293 cells expressing CERK and SARS-CoV-2 structural proteins were imaged. CERK displayed colocalization with fluorescently labeled E protein. (d) Assay for C1P production. Thin layer chromatography of samples from mock-, CERK-, CERK and M-, or CERK, M, and N-transfected HEK293 cells treated with 2.5 μM NBD-ceramide and either DMSO or 100 nM CERK inhibitor NVP-231. CERK increased C1P levels and was inhibited by NVP-231. (e) Assay for M oligomerization in cells. Western blot for M (upper) or GAPDH (lower) in size exclusion chromatography fractionated CERK, M, N, E, and S-expressing HEK293 cell lysate treated with DMSO or 100 nM NVP-231. Later eluting fractions correspond to smaller species. Lower C1P levels altered M protein oligomerization state with an increased proportion of smaller species detected.

Figure 2.. C1P energetically stabilizes M_short.

(a) Structures of M protein viewed from the membrane plane. Collective variables that report on intersubunit distance and angle used to monitor conformational state during simulations are illustrated on the structures. (b) M protein conformational trajectory during molecular dynamics simulations. Intersubunit distance versus intersubunit angle of the structure at each time sample during 4 us simulations is plotted. Simulations were performed without (i,ii) or with C1P (iii, iv) and initiated from M_short (i,iii) or M_long (i, iv). Conformational switching was only observed from M_long to M_short in the presence (of C1P. (c) Free-energy landscape calculated from simulations without (above) or with C1P (below). Position of M_short, M_long, and a low energy state from each condition is indicated. (d) M-lipid contact frequency from simulations without (above) and with C1P (below). (e) M-C1P percentage occupancy by residue in subunit A (above) and subunit B (below).

Figure 3.. Cryo-EM structure of M_short bound to C1P

(a) 3.0 Å resolution cryo-EM map of SARS-CoV- 2 M in the short conformation viewed from the membrane. M subunits are purple, short conformation-selective Fabs (scFab) are gray, and C1P is yellow. Unsharpened map is shown transparent at low contour. (b) Zoomed in view of C1P binding site in one subunit. (c) Electrostatic surface view of M with C1P bound. (d) Single M_short-C1P subunit with C1P binding residues yellow, K⁺ binding residues pink, and residues displaying associated conformational changes blue. (e) Zoomed in view of K⁺ binding site in M_short-C1P (left) or apo M_short (right). Cryo-EM density is shown as a transparent surface. (f) Sequence alignment of M across coronaviruses with residues colored as in (d).

Figure 4.. Functional assessment of M protein mutations.

(a) Bead pull down assay of M^ΔC1P mutant-lipid interactions. Western blot of input EGFP-M^ΔC1P expressing cell lysate and eluant from uncoated, C1P-coated, or PA-coated beads. (b) M^ΔC1P disrupts colocalization with S observed with wild-type M. (c,d) M^ΔC1P disrupts Golgi localization of wild-type M. (e,f) Virus-like particle (VLP) entry assay. M^ΔC1P, M^ΔC1P/ΔTGN, and M^ΔN (which reduces M:N-interaction) derived VLPs show reduced cell entry compared to VLPs formed with wild-type M or M^ΔC1P. (g,h) Cell syncytia formation assay. Cells expressing S alone, but not S and M, form robust syncytia. In contrast, co-expression with M^ΔC1P significantly increased syncytia formation. Data are mean ± sem with differences assessed with a one-way Annova with Dunnet correction for multiple comparisons (****p<0.0001, ***p<0.001, **p<0.01).

Figure 5.. Model for lipid control of M conformational dynamics and coronavirus. assembly.

M protein adopts M_short and M_long conformations. Ceramide kinase localization to ERGIC membranes results in high C1P concentrations. C1P binds and stabilizes M_short in the ERGIC to facilitate recruitment of E and S. Decreasing C1P along the secretory pathway and other potential factors trigger a conformational switch to M_long that promotes higher membrane curvature and oligomerization necessary for virus assembly and budding.