Viral Membrane Fusion: A Dance Between Proteins and Lipids

There are at least 21 families of enveloped viruses that infect mammals, and many contain members of high concern for global human health. All enveloped viruses have a dedicated fusion protein or fusion complex that enacts the critical genome-releasing membrane fusion event that is essential before viral replication within the host cell interior can begin. Because all enveloped viruses enter cells by fusion, it behooves us to know how viral fusion proteins function. Viral fusion proteins are also major targets of neutralizing antibodies, and hence they serve as key vaccine immunogens. Here we review current concepts about viral membrane fusion proteins focusing on how they are triggered, structural intermediates between pre- and postfusion forms, and their interplay with the lipid bilayers they engage. We also discuss cellular and therapeutic interventions that thwart virus-cell membrane fusion.


Supplemental
The fusion proteins schematically depicted are HIV Env (class I), Semliki forest virus (SFV) E1 (class II), and herpes simplex virus-1 (HSV-1) gB (class III).Subsequent to priming (for class I and II proteins; not depicted), and in response to specific triggers (T), the pre-fusion structures proceed through stages of extended pre-hairpin formation, fold-back (F), and zippering (Z) while the membranes progress from separated bilayers through hemifusion, fusion pore formation, and (not shown) fusion pore expansion.F, F1 and F2 denote progressive stages of fold-back; these designations are not meant to infer that there are only one or two stages of fold-back.Some pre-fusion states have been shown to be ensembles of structures, for examples, see References (1)(2)(3)(4)(5).For class I and III proteins, triggering involves 'cross-talk' between the receptor binding head and the stalk regions (6)(7)(8)(9)(10)(11)(12)(13) (2,(12)(13)(14).For class I proteins without a disulfide bond between the receptor binding and fusion subunits (e.g., HIV Env as depicted (gp120, gray) and SARS-CoV-2 S), the receptor binding subunit is shed (15,16); for others (e.g., influenza HA), the head domains move away from the trimer axis.It has been suggested that class I fusion proteins form membraneembedded trimeric pre-hairpins in a sequential asymmetric manner (17)(18)(19), perhaps with insertion of one fusion peptide as a first step.For an avian retroviral Env and HIV Env, glimpses of membrane-embedded pre-hairpins have been observed by cryo-EM (20,21).On mature virions at neutral pH, the SFV fusion protein (class II) is a heterodimer of E1 (colored subunit) and E2 (gray subunit), as depicted in the far-left image, that are further arrayed as trimeric spikes.In response to the fusion trigger, low pH (~pH 6), the B domain of E2 is dislodged revealing the fusion loops (asterisks) on E1 (second image), E2 dissociates, and E1 forms a membraneembedded pre-hairpin.Where studied, target membrane interaction is first mediated by an E1 monomer (not shown) prior to coalescing to the (membrane-embedded) trimeric pre-hairpin (22).See main text and (23)(24)(25)(26) for how class III fusion proteins are triggered and change conformation.Color-coding: Class I: red, fusion peptide; blue, heptad repeat 1 (HR1); green, HR2; orange, transmembrane domain (TMD); gray, HIV gp120.Class II: E1: green, domain I; red, domain II; purple, domain III; orange, TMD of E1, asterisk, fusion loop; (SFV) E2: gray, domains A, B, C and TMD.Class III: red, domain I; light brown, domain II; blue, domain III; green, domain IV; purple, domain V; pink, membrane proximal external region (MPER); orange, TMD; asterisks, fusion loops.The orientation is target membrane (light gray) on top and viral membrane (light gray) below.Thin black lines below TMDs denote cytoplasmic tails.Other notes: Flexibility in MPERs, located just prior to TMDs, can be required for foldback stages (27).The HIV peptide inhibitor T20 prevents fusion by blocking six-helix bundle formation (28).Fusion pore formation and/or fusion pore expansion (not depicted) may involve changes to matrix layers underlying the viral membrane (29)(30)(31).Eukaryotic and archaeal fusion proteins display structural similarity to class II viral fusion proteins (32,33).The overall  ; green denotes TMDs.HA1 is not seen in the postfusion structure.An evolutionary analysis of the fusion subunits of class I fusion proteins has recently been presented (43).As depicted in Figure 1a, receptor binding subunits (e.g., HIV gp120 and influenza HA1) are said to 'clamp' the fusion protein in an inactive state.Upon triggering, the clamp is released, allowing conformational changes to proceed.For HA, key changes include the B-loop to helix (magenta) transition, helix D (neon green) to loop transition, and packing of the Cterminal 'leash' (orange) in the grooves of the central N-terminal coiled-coil (44)(45)(46)(47).Structures of the TMD of an H1N1 influenza HA in detergent micelles were visualized by cryo-EM revealing both straight and tilted TMD helices in trimeric bundles (27).
(B) Class II: Dengue virus type 2 protein E is shown in pre-(left, PDB: 4UTB, side view) and postfusion (right, PDB: 1OK8) conformations.On virions, dengue E is present as a homodimer that is further arrayed along with its associated protein (prM; not shown) on the virion surface (48, 49) (41).In the left panel (pre-fusion), one E ectodomain monomer is shown in gray and the other is coded with Domains I, II, and III in yellow, blue and purple, respectively; the fusion loops at the tips of beta-strands comprising domain II are shown in red, and the TMDs are depicted in green.Some class II proteins (those of alpha-and flaviviruses) have one fusion loop; some in the Bunyavirales order have 2 or 3 fusion loops.For dengue virus E, packing of stem regions into the 'seam' of the trimer has been shown to be important (50,51).All class II fusion proteins, including those involved in eukaryotic cell-cell fusion and archaeal fusion (32,33), have the same basic 3-domain architecture, The companion (associated) proteins to the class II viral fusion proteins (prM for flaviviruses (not shown)) are also structurally conserved (52).Class II fusion proteins have been observed in arrays post fusion (53)(54)(55).
(C) Class III: HCMV gB is shown in (left 2 panels) in pre-(PDB: 7KDP) and (far right) post-fusion (PDB: 7KDD) conformations; an excised monomer is shown on the far left.Domains I, II, III, IV and V are shown in blue, magenta, teal, orange, and yellow, respectively.The MPER is shown in purple, the TMD in green and the fusion loops (FLs) in red.HMV gB also has a long cytoplasmic tail (~130 residues, not shown).Rhabdoviral Gs have similarly constituted regions corresponding to domains I-IV (56).All class III proteins display fusion loops at the tips of a beta-stranded domain (Domain I for gB; FD (fusion domain) for VSV G).A hinge (between DII and DIII for gB) allows the ~180 o flip (of DI and DII) that positions the fusion loops for insertion into the target membrane (56,57).Some herpesvirus gBs are cleaved by furin (e.g., EBV gB), but others are not (e.g., HSV-1 gB).Nonetheless, gBs are considered to be metastable as they rapidly transit to their post-fusion structures.In contrast, rhabdovirus Gs are not primed and are not metastable; they reversibly sample post-fusion structures upon pH modulation.For rhabdovirus Gs studied, evidence suggests that the extended (target membrane-embedded) intermediate is a monomer that reassembles into a trimer as fusion progresses (5,26,(58)(59)(60).There is no evidence that herpesvirus gBs (or viral membrane anchored class I proteins) are monomers as they insert into the target membrane (though, as stated above, fusion peptides/loops may insert sequentially).In contrast to rhabdoviral Gs, herpesvirus gBs work in concert with additional proteins to elicit fusion (see main text).Class III fusion proteins have been seen in (altered) arrays post fusion (59,61,62).

Figure 3. Energetics and membrane dynamics during fusion. (A)
An approximated schema of fusion energetics as exemplified by the HIV Env glycoprotein (protein symbols as in the key to Figure 4).The blue curve traces the approximate free energy from receptor binding (slightly negative due to the enthalpy of gp120/CD4 binding), to hemifusion (a higher energy intermediate and local minimum), to pore opening and finally through pore expansion to the ultimately lower energy post-fusion membrane.The energies depicted are approximate but the transition energies (the peaks) range from ~10-100k B T (47,63,64).(B) Lipids can have positive (lysophosphatidylcholine, (lysoPC)), negative (phosphatidylethanolamine (PE)) or no (phosphatidylcholine (PC)) intrinsic membrane curvature.Negative intrinsic curvature of the exterior leaflet stabilizes hemifusion intermediates and positive intrinsic curvature in the interior leaflets stabilizes fusion pores.

Figure 4. Effects of membrane lateral heterogeneity on viral membrane fusion as exemplified by HIV. (A)
In target cell membranes, lipid nanodomains can organize and concentrate receptors and/or fusion triggers.For HIV, the receptor, CD4, partitions to ordered nanodomains (rich in saturated (teal) phospholipids and cholesterol) while the co-receptor, CCR5, partitions to the domain boundary as described in (65).(B) In the viral membrane, lipid nanodomains influence the spacing of fusion proteins.On immature HIV particles, Env is relatively immobile.Upon maturation via proteolytic cleavage of the juxtamembrane Gag polyprotein, Env diffuses more rapidly to form clusters that facilitate fusion (66,67).Env has multiple sequences that promote association with ordered nanodomains (yellow phospholipids; cholesterol not depicted) including a cholesterol recognition amino acid consensus (CRAC) motif within the MPER (68), and palmitoylation sites and an additional cholesterol interacting domain within the C-terminal cytoplasmic tail of gp41 (69).While Env generally partitions to ordered domains, this may be cell type and HIV strain dependent (68).(C) In addition to concentrating receptors and fusion proteins, lateral heterogeneity affects the energetics of fusion.After one or more Env trimers bind receptor and co-receptor, the fusion subunit, gp41, forms a pre-hairpin with fusion peptides preferentially inserted at the discontinuity in bilayer thickness between ordered and disordered lipid nanodomains as described in (70).Fold-back to the trimer-of-hairpins promotes hemifusion, fusing the outer bilayer leaflets.Joining two ordered domains produces one larger domain upon full fusion with a lower ratio of perimeter/area than the starting smaller domains.This minimizes line tension at the domain boundary and contributes favorably to the energetics of fusion as described in (71).Figure 5.A complex between fusion peptides/loops and MPER-TMD regions completes the fusion protein refolding process that accompanies fusion pore opening.Shown on top is a cartoon of the fusion pore stage of ebolavirus GP mediated fusion; only GP2 is shown and it is color-coded: purple and brown, respectively, the N-and C-terminal heptad repeats; blue, fusion loop; red, MPER-TMD.Gray represents the recently merged membrane.Below is a blow up of the region boxed in the cartoon in which the NMR structures in membrane mimetics of the ebolavirus fusion loop (blue) (72,73) and MPER-TMD (red) (74) were docked based on experimental interaction constraints from fluorescence and NMR data.Green denotes interacting residues.Gray represents membrane.Figure adapted from (74).A recent study presents a cryo-EM

Figure 1 .
Figure 1.Fusion pathways of class I, II and III viral fusion proteins.The fusion proteins schematically depicted are HIV Env (class I), Semliki forest virus (SFV) E1 (class II), and herpes simplex virus-1 (HSV-1) gB (class III).Subsequent to priming (for class I and II proteins; not depicted), and in response to specific triggers (T), the pre-fusion structures proceed through stages of extended pre-hairpin formation, fold-back (F), and zippering (Z) while the membranes progress from separated bilayers through hemifusion, fusion pore formation, and (not shown) fusion pore expansion.F, F1 and F2 denote progressive stages of fold-back; these designations are not meant to infer that there are only one or two stages of fold-back.Some pre-fusion states have been shown to be ensembles of structures, for examples, see References(1)(2)(3)(4)(5). For class I and III proteins, triggering involves 'cross-talk' between the receptor binding head and the stalk regions (6-13)(2,(12)(13)(14).For class I proteins without a disulfide bond between the receptor binding and fusion subunits (e.g., HIV Env as depicted (gp120, gray) and SARS-CoV-2 S), the receptor binding subunit is shed(15,16); for others (e.g., influenza HA), the head domains move away from the trimer axis.It has been suggested that class I fusion proteins form membraneembedded trimeric pre-hairpins in a sequential asymmetric manner(17)(18)(19), perhaps with insertion of one fusion peptide as a first step.For an avian retroviral Env and HIV Env, glimpses of membrane-embedded pre-hairpins have been observed by cryo-EM(20,21).On mature virions at neutral pH, the SFV fusion protein (class II) is a heterodimer of E1 (colored subunit) and E2 (gray subunit), as depicted in the far-left image, that are further arrayed as trimeric spikes.In response to the fusion trigger, low pH (~pH 6), the B domain of E2 is dislodged revealing the fusion loops (asterisks) on E1 (second image), E2 dissociates, and E1 forms a membraneembedded pre-hairpin.Where studied, target membrane interaction is first mediated by an E1 monomer (not shown) prior to coalescing to the (membrane-embedded) trimeric pre-hairpin(22).See main text and(23)(24)(25)(26) for how class III fusion proteins are triggered and change conformation.Color-coding: Class I: red, fusion peptide; blue, heptad repeat 1 (HR1); green, HR2; orange, transmembrane domain (TMD); gray, HIV gp120.Class II: E1: green, domain I; red, domain II; purple, domain III; orange, TMD of E1, asterisk, fusion loop; (SFV) E2: gray, domains A, B, C and TMD.Class III: red, domain I; light brown, domain II; blue, domain III; green, domain IV; purple, domain V; pink, membrane proximal external region (MPER); orange, TMD; asterisks, fusion loops.The orientation is target membrane (light gray) on top and viral membrane (light gray) below.Thin black lines below TMDs denote cytoplasmic tails.Other notes: Flexibility in MPERs, located just prior to TMDs, can be required for foldback stages(27).The HIV peptide inhibitor T20 prevents fusion by blocking six-helix bundle formation(28).Fusion pore formation and/or fusion pore expansion (not depicted) may involve changes to matrix layers underlying the viral membrane(29)(30)(31). Eukaryotic and archaeal fusion proteins display structural similarity to class II viral fusion proteins(32,33).The overall Figure layout is similar to Fig.1in(34), with the class II pathway stylized after Fig.3in(22).
Figure 1.Fusion pathways of class I, II and III viral fusion proteins.The fusion proteins schematically depicted are HIV Env (class I), Semliki forest virus (SFV) E1 (class II), and herpes simplex virus-1 (HSV-1) gB (class III).Subsequent to priming (for class I and II proteins; not depicted), and in response to specific triggers (T), the pre-fusion structures proceed through stages of extended pre-hairpin formation, fold-back (F), and zippering (Z) while the membranes progress from separated bilayers through hemifusion, fusion pore formation, and (not shown) fusion pore expansion.F, F1 and F2 denote progressive stages of fold-back; these designations are not meant to infer that there are only one or two stages of fold-back.Some pre-fusion states have been shown to be ensembles of structures, for examples, see References(1)(2)(3)(4)(5). For class I and III proteins, triggering involves 'cross-talk' between the receptor binding head and the stalk regions (6-13)(2,(12)(13)(14).For class I proteins without a disulfide bond between the receptor binding and fusion subunits (e.g., HIV Env as depicted (gp120, gray) and SARS-CoV-2 S), the receptor binding subunit is shed(15,16); for others (e.g., influenza HA), the head domains move away from the trimer axis.It has been suggested that class I fusion proteins form membraneembedded trimeric pre-hairpins in a sequential asymmetric manner(17)(18)(19), perhaps with insertion of one fusion peptide as a first step.For an avian retroviral Env and HIV Env, glimpses of membrane-embedded pre-hairpins have been observed by cryo-EM(20,21).On mature virions at neutral pH, the SFV fusion protein (class II) is a heterodimer of E1 (colored subunit) and E2 (gray subunit), as depicted in the far-left image, that are further arrayed as trimeric spikes.In response to the fusion trigger, low pH (~pH 6), the B domain of E2 is dislodged revealing the fusion loops (asterisks) on E1 (second image), E2 dissociates, and E1 forms a membraneembedded pre-hairpin.Where studied, target membrane interaction is first mediated by an E1 monomer (not shown) prior to coalescing to the (membrane-embedded) trimeric pre-hairpin(22).See main text and(23)(24)(25)(26) for how class III fusion proteins are triggered and change conformation.Color-coding: Class I: red, fusion peptide; blue, heptad repeat 1 (HR1); green, HR2; orange, transmembrane domain (TMD); gray, HIV gp120.Class II: E1: green, domain I; red, domain II; purple, domain III; orange, TMD of E1, asterisk, fusion loop; (SFV) E2: gray, domains A, B, C and TMD.Class III: red, domain I; light brown, domain II; blue, domain III; green, domain IV; purple, domain V; pink, membrane proximal external region (MPER); orange, TMD; asterisks, fusion loops.The orientation is target membrane (light gray) on top and viral membrane (light gray) below.Thin black lines below TMDs denote cytoplasmic tails.Other notes: Flexibility in MPERs, located just prior to TMDs, can be required for foldback stages(27).The HIV peptide inhibitor T20 prevents fusion by blocking six-helix bundle formation(28).Fusion pore formation and/or fusion pore expansion (not depicted) may involve changes to matrix layers underlying the viral membrane(29)(30)(31). Eukaryotic and archaeal fusion proteins display structural similarity to class II viral fusion proteins(32,33).The overall Figure layout is similar to Fig.1in(34), with the class II pathway stylized after Fig.3in(22).

Figure 2 .
Figure 2. Pre-and post-fusion structures of class I, II, and III viral fusion proteins.For more details on the pre-and post-fusion structures of viral fusion proteins see (22-25, 35-40) (41).Pre-fusion, Class I and III proteins project perpendicularly, whereas Class IIs lie parallel, to the viral membrane.All viral fusion proteins are projecting trimers in their post-fusion states.Fusion peptide/loop sequences can be found at https://viralfp.bio.di.uminho.pt/.As discussed in the text, intermediates between the pre-and post-fusion forms have recently been detected by advanced biophysical techniques.