Cell. Mol. Life Sci. 65 (2008) 1716 – 1728
`1420-682X/08/111716-13
`DOI 10.1007/s00018-008-7534-3
` Birkhäuser Verlag, Basel, 2008
`
`Cellular and Molecular Life Sciences
`
`Review
`
`Structures of vesicular stomatitis virus glycoprotein:
`membrane fusion revisited
`
`S. Roche*,+, A. A. V. Albertini, J. Lepault, S. Bressanelli and Y. Gaudin*
`
`CNRS, UMR2472, INRA, UMR1157, IFR 115, Virologie MolØculaire et Structurale, 91198, Gif sur Yvette
`(France), e-mail: gaudin@vms.cnrs-gif.fr, Fax: + 33 1 69 82 43 08
`
`Received 21 November 2007; received after revision 29 January 2008; accepted 30 January 2008
`Online First 17 March 2008
`
`Abstract. Glycoprotein G of the vesicular stomatitis
`virus (VSV) is involved in receptor recognition at the
`host cell surface and then, after endocytosis of the
`virion, triggers membrane fusion via a low pH-
`induced structural rearrangement. G is an atypical
`fusion protein, as there is a pH-dependent equilibrium
`between its pre- and post-fusion conformations. The
`atomic structures of these two conformations reveal
`that it is homologous to glycoprotein gB of herpesvi-
`ruses and that it combines features of the previously
`characterized class I and class II fusion proteins.
`
`Comparison of the structures of G pre- and post-
`fusion states shows a dramatic reorganization of the
`molecule that is reminiscent of that of paramyxovirus
`fusion protein F. It also allows identification of
`conserved key residues that constitute pH-sensitive
`molecular switches. Besides the similarities with other
`viral fusion machineries, the fusion properties and
`structures of G also reveal some striking particular-
`ities that invite us to reconsider a few dogmas
`concerning fusion proteins.
`
`Keywords. Vesicular stomatitis virus, rhabdovirus, paramyxovirus, glycoprotein, membrane fusion, viral entry,
`conformational change.
`
`Introduction
`
`To initiate a productive infection, all viruses must
`translocate their genome across the cell membrane
`[1]. For enveloped viruses, this step is mediated by
`virally encoded glycoproteins that promote both
`receptor recognition and membrane fusion. Both
`tasks can be achieved by a single or by separate
`glycoproteins acting in concert. Activation of the
`fusion capacity involves large structural rearrange-
`ments of the fusogenic glycoproteins upon interaction
`with specific triggers (e.g.
`low pH and cellular
`
`* Corresponding author.
`+ Present adress: Max Planck Institute of Biochemistry, Am
`Klopferspitz 18, 82152 Martinsried (Germany)
`
`receptors). These conformational changes result in
`the exposure of a fusion peptide or fusion loops, which
`then interact with one or both of the participating
`membranes, resulting in their destabilization and
`merger [2]. Triggering of the conformational change
`in the absence of a target membrane leads to
`inactivation of the fusion properties of the fusogenic
`glycoprotein.
`Experimental data suggest that the membrane fusion
`pathway is very similar for all the enveloped viruses
`studied so far whatever the organization of their
`fusion machinery [3 – 5] (Fig. 1). It is generally accept-
`ed that fusion proceeds via the formation of inter-
`mediate stalks that are local
`lipidic connections
`between the outer leaflets of the fusing membranes
`[6]. Radial expansion of the stalk would induce the
`
`Page 1 of 13
`
`KELONIA EXHIBIT 1015
`
`

`

`Cell. Mol. Life Sci.
`
`Vol. 65, 2008
`
`Review Article
`
`1717
`
`Figure 1. Stages of membrane fusion according to the stalk-pore model [82].
`
`formation of a transient hemifusion diaphragm (i.e. a
`local bilayer made by the two initial inner leaflets).
`Depending on the experimental system, hemifusion
`may be restricted (i.e. without lipid exchange between
`the two membranes) or unrestricted (i.e. without any
`restriction of lipid diffusion). Restriction of lipid flux
`has been proposed to be due to a ringlike aggregate of
`fusogenic glycoproteins surrounding the hemifusion
`diaphragm [3, 4]. The next step would be the forma-
`tion of a pore in the fusion diaphragm. The initial pore
`is small and is often opening and closing repeatedly
`(the so-called flickering pore) before its enlargement
`that leads to complete fusion [7].
`
`Rhabdovirus glycoprotein G
`
`Rhabdoviruses are widespread among a great diver-
`sity of organisms (including plants, insects, fishes,
`mammals, reptiles and crustaceans) [8]. This family
`includes vesicular stomatitis virus (VSV) as well as
`significant human pathogens like rabies virus (RV) or
`Chandipura virus [9]. Rhabdoviruses are enveloped
`viruses and have in common a bulletlike shape. Their
`genome is a single RNA molecule of negative polarity.
`It associates with the nucleoprotein N, the viral
`polymerase L and the phosphoprotein P to form the
`nucleocapsid. The nucleocapsid is condensed by the
`matrix protein M into a tightly coiled helical structure,
`which is surrounded by a lipid bilayer containing the
`viral glycoprotein G.
`G forms the spikes that protrude from the viral
`surface. After cleavage of the aminoterminal signal
`peptide, the complete mature glycoprotein is about
`500 amino acids long (495 for VSV Indiana). The bulk
`of the mass of G is located outside the viral membrane
`and constitutes the amino-terminal ectodomain. As
`this ectodomain is the only outer component of the
`viruses, it is the target of neutralizing antibodies [10 –
`16].
`G plays a critical role during the initial steps of virus
`infection. First, it is responsible for virus attachment to
`specific receptors. The nature of the receptor remains
`a matter of debate for both VSVand RV. In the case of
`
`VSV, although phosphatidylserine has been consid-
`ered to be the viral receptor for a long time [17], recent
`results indicated that it is not [18]. In the case of RV,
`many molecules, including gangliosides [19], phos-
`pholipids [20], the nicotinic acetylcholine receptor
`[21, 22], neuronal cellular adhesion molecules [23] and
`the low-affinity nerve growth factor receptor [24],
`have been proposed to be viral receptors.
`After binding, the virions enter the cell by the
`endocytic pathway. Subsequently, the viral envelope
`fuses with a cellular membrane within the acidic
`environment of the endosome [25]. Fusion is triggered
`by the low pH of the endosomal compartment and is
`mediated by the viral glycoprotein. The pH depend-
`ence is very similar from one rhabdovirus to another
`and the fusion is optimal around pH 6 [26 – 28].
`Preincubation of the virus at low pH in the absence of
`a target membrane leads to inhibition of viral fusion.
`However, this inhibition is reversible, and readjusting
`the pH to above 7 leads to the complete recovery of
`the initial fusion activity. This is the main difference
`between rhabdoviruses and other viruses fusing at low
`pH, for which low pH-induced fusion inactivation is
`irreversible [29].
`G can assume at least three different conformational
`states having different biochemical and biophysical
`characteristics [26, 30]: the native, prefusion state
`detected at the viral surface above pH 7; the activated
`hydrophobic state, which interacts with the target
`membrane as a first step of the fusion [31]; and the
`post-fusion conformation, which is antigenically dis-
`tinct from the native and activated states [32]. There is
`a pH-dependent equilibrium between the different
`states of G that is shifted toward the post-fusion state
`at low pH [32]. This indicates that, differently from
`fusogenic glycoproteins from other viral families, the
`low-pH induced-conformational change is reversible
`and thus that the native conformation is not meta-
`stable. In fact, the reversibility of the fusogenic low-
`pH-induced conformational change is essential to
`allow G to be transported through the acidic compart-
`ments of the Golgi apparatus and to recover its native,
`prefusion state at the viral surface [33].
`
`Page 2 of 13
`
`

`

`1718
`
`S. Roche et al.
`
`VSV glycoprotein structure
`
`Figure 2. Overall
`structure of
`the pre- and post-fusion forms
`of tick-borne encephalitis virus
`(TBEV) glycoprotein E, a repre-
`sentative member of class II viral
`fusion proteins. (A) Ribbon dia-
`gram of the dimeric pre-fusion
`structure (PDB: 1SVB [41]). (B)
`Ribbon diagram of the trimeric
`post-fusion
`structure
`(PDB:
`1URZ [83]). Domains are col-
`oured as in [41]. The location of
`the fusion loop is indicated. The
`magenta dotted lines represent
`the missing part of the ectodo-
`main that is connected to the
`transmembrane domain. PDB:
`Protein Data Bank.
`
`Class I and class II fusion proteins
`
`Before the structure determination of VSV G, two
`classes of viral fusion proteins had been identified
`(Fig. 2, 4). The viral fusion proteins belonging to class
`I, of which the best-characterized members are the
`influenza virus hemagglutinin (HA) [34, 35] and the
`fusion protein (F) of the paramyxoviruses [36, 37] but
`which also include fusion proteins from retroviruses
`[38] and filoviruses [39], are organized in trimers.
`Each subunit (or protomer) constituting the trimer
`results from the proteolytic cleavage of a precursor
`into two fragments. The C-terminal fragment bears at
`or near its amino-terminal end (i.e. at or near the
`cleavage site) a hydrophobic fusion peptide, buried at
`a trimer interface in the prefusion state. In the post-
`fusion conformation, this region refolds as a trimeric
`coiled coil at
`the N-terminal end of which are
`displayed the three fusion peptides and against
`which are packed, in an antiparallel manner, the
`segments
`abutting
`the
`transmembrane
`region
`(Fig. 4B). The protomer shape is thus an elongated
`hairpin-like structure with the fusion peptide and the
`transmembrane domain located at the same end, as
`expected at the end of the fusion process [40].
`The class II fusion proteins contain E protein of
`flaviviruses and E1 of alphaviruses [41 – 43]. They
`display a molecular architecture completely different
`from that of class I proteins (Fig. 2). Their fusion
`peptide is internal, located in a loop between two b-
`strands. They are synthesized and folded as a complex
`with a second viral envelope protein that plays a
`chaperone role. Proteolytic cleavage of the chaperone
`primes the fusion protein to trigger membrane merger
`[44]. In their native conformation (Fig. 2A), they form
`homo- (flaviviruses) or hetero- (alphaviruses) dimers
`that are organized in an icosahedral assembly [42, 45].
`They lie flat or nearly flat at the viral surface and their
`
`fusion loops are buried at a dimer interface. Upon low-
`pH exposure, dimers dissociate and the protomers
`reassociate in a trimeric structure [46, 47]. Similar to
`the structure of post-fusion class I proteins [48], the
`fusion loops and the transmembrane domains are then
`located at the same end of an elongated molecule that
`is now perpendicular to the membrane [49, 50]
`(Fig. 2B). Thus, even though the structures of class I
`and class II fusion proteins are unrelated, the mech-
`anisms for refolding share key common features. First,
`the fusion peptides/loops are exposed and projected
`toward the top of the glycoprotein, allowing the initial
`interaction with the target membrane. Second, the
`folding back of the C-terminal region onto a trimeric
`N-terminal region leads to the formation of a post-
`fusion protein structure with the outer regions zipped
`up against the inner trimeric core [2].
`For both class I and class II fusion proteins that trigger
`membrane merger at low pH, the proteolytic cleavage
`priming the proteins to undergo their low-pH-induced
`conformational change occurs in the trans-Golgi
`network or at the host cell surface [44, 51, 52]. This
`precludes premature activation of the fusion protein
`in the acidic compartments of the Golgi apparatus.
`Thus, reversibility of the low-pH-induced fusogenic
`transition is not necessary for these proteins.
`Biochemical, structural and functional properties of
`rhabdovirus G suggested that it was distinct from both
`class I and class II viral fusion proteins that had been
`already described [29, 53]. Indeed, the pH-dependent
`equilibrium between the different states of G, the
`absence of predicted a-helical coiled-coil motif char-
`acteristic of class I viral fusion proteins [40] and the
`absence of activating cleavage (neither in G nor in an
`accompanying protein) strongly suggested that G
`could define a new category of fusogenic glycopro-
`teins.
`
`Page 3 of 13
`
`

`

`Cell. Mol. Life Sci.
`
`Vol. 65, 2008
`
`Review Article
`
`1719
`
`Figure 3. (A) Overall structure
`of the pre- and post-fusion forms
`of VSV glycoprotein G. Ribbon
`diagrams of the pre-fusion struc-
`ture of G trimer (top left) (PDB:
`2J6J [55]); of
`the post-fusion
`structure of G trimer (top right)
`(PDB: 2CMZ [54]); of the pre-
`fusion structure of G protomer
`(residues 1 – 413) (bottom left);
`and of the post-fusion structure
`of G protomer (residues 1 – 410)
`(bottom right). G protein is col-
`oured by domains (domain I: red,
`domain II: blue, domain III:
`orange, domain IV: yellow) with
`the fusion loops in green and the
`C-terminus in magenta. The pro-
`tomers are superimposed on
`their fusion domains (DIV) and
`the trimers on the rigid blocks
`made of DI and the invariant part
`of DII. In the protomer diagrams,
`the dotted lines represent the
`missing C-terminal segment of
`the ectodomain that leads to the
`transmembrane segment.
`(B)
`Overall structure of HSV-1 gB.
`Ribbon diagrams of gB trimer
`(top) and gB protomer (bottom)
`(PDB: 2GUM [57]). gB protein
`is coloured by domains as their
`homologous
`counterparts
`of
`VSV G.
`
`VSV G structure
`
`We have recently determined the atomic structures of
`both the pre- and post-fusion forms of the VSV-G
`ectodomain, generated by limited proteolysis with
`thermolysin (Gth, aa residues 1 – 422)
`[54, 55]
`(Fig. 3A). The dimensions of the two conformations
`– consistent with the electron microscopy data ob-
`tained on RV G [30, 56] – together with the position of
`the antigenic sites allowed a clear-cut identification of
`the pre- and post-fusion structures. The structural
`organization of the two conformations of G is very
`different from that of other viral fusion proteins
`described so far. However, amino acid sequence
`alignment of different G proteins of rhabdoviruses
`belonging to different genera shows that all these
`glycoproteins have the same fold.
`
`Four distinct domains of Gth have been identified: a b-
`sheet-rich lateral domain (domain I), a central domain
`that is involved in the trimerization of the molecule
`(domain II), a pleckstrin homology domain (domain
`III) and the fusion domain (domain IV) inserted in a
`loop of domain III. Major antigenic sites are located in
`both domains I and III [55].
`After the end of the trimerization domain (after
`amino acid residue 405), there remain 40 amino acids
`for the polypeptide chain to reach the G transmem-
`brane domain (Fig. 3A, bottom), but their structural
`organization is unknown after amino acid residue 413
`for the pre-fusion conformation and after amino acid
`residue 410 for the post-fusion conformation.
`
`Page 4 of 13
`
`

`

`1720
`
`S. Roche et al.
`
`An unexpected homology
`
`The structure of herpes simplex virus 1 (HSV1, a
`double-stranded DNA virus) glycoprotein gB [57] was
`published at the same time as that of Gth post-fusion
`state. Comparison of the two structures revealed that
`their folds are the same and that they have a common
`evolutionary origin that could not be detected by
`looking at the amino acid sequences (Fig. 3B). This
`was completely unexpected and suggests that rhabdo-
`viruses, and most probably all viruses belonging to the
`Mononegavirales order, have the ability to steal genes
`from their host (or from another virus during co-
`infection of a host cell). This might occur when the
`viral polymerase jumps from the antigenomic tem-
`plate onto an RNA messenger (either of cellular or
`viral origin) during genomic RNA synthesis. Never-
`theless, the exact scenario of G gene acquisition by the
`rhabdovirus ancestor will still remain a matter of
`debate for a long time.
`The orientation of the central helix relative to the viral
`membrane in the determined structure of HSV1 gB
`suggests that gB is in its post-fusion conformation.
`Nevertheless, as the fusion machinery of herpesvirus-
`es is much more complex than that of rhabdoviruses
`(with four glycoproteins that are essential for virus
`entry) and as its mode of activation is completely
`different [58], the extent of gB conformational change
`cannot be inferred from its homology with VSV G.
`
`The conformational change of VSV G
`
`Comparison of the pre- and post-fusion structures of
`VSV G reveals a dramatic reorganization of the
`molecule (Fig. 3A). During the conformational
`change, domains I, III and IV retain their tertiary
`structure. Nevertheless, they undergo large rearrange-
`ments in their relative orientation due to secondary
`structure changes in the hinge regions between the
`fusion and pleckstrin homology domains and major
`refolding of
`the central
`trimerization domain
`(Fig. 4C). In fact, the pre- and post-fusion states are
`related by flipping both the fusion domain and the C-
`terminal segment relative to a rigid block constituted
`of the lateral domain and the part of the trimerization
`domain that retains its structure during the molecule
`refolding.
`to post-fusion
`Global refolding of G from pre-
`conformation exhibits striking similarities to that of
`class I proteins such as paramyxovirus fusion protein
`(F) (Fig. 4A) and influenza virus hemagglutinin sub-
`unit 2 (HA2) (Fig. 4C) [35, 37]. Particularly, the
`reversal of the molecule around the rigid block
`involves the lengthening of the central helices (that
`
`VSV glycoprotein structure
`
`form the trimeric central core of the post-fusion
`conformation, thus displaying the fusion domains –
`through the PH domains – at their N-termini) and the
`refolding of the three carboxy-terminal segments into
`helices that position themselves in the grooves of the
`central core in an antiparallel manner to form a six-
`helix bundle (Fig. 4C). This structural organization is
`obviously very similar to that of the post-fusion
`hairpin structure of class I proteins (Fig. 4B), even
`though, for VSV G, the central helices are not coiled
`and remain parallel.
`
`Interaction between fusion domains and membranes
`
`The structural organization of the G fusion domain
`resembles that of class II fusion proteins. The main
`difference is that the membrane interacting motif of
`the fusion domain is bipartite (as previously proposed
`for viral haemorrhagic septicaemia virus, another
`rhabdovirus [27]), made of two loops, and that the
`loop sequences are not conserved among rhabdovi-
`ruses. However, as in class II fusion proteins, these
`loops always contain aromatic residues and are
`located at the tip of an elongated three-stranded b-
`sheet. Note that in striking contrast to class I and class
`II viral fusion proteins, the fusion loops are not buried
`at an oligomeric interface in G pre-fusion conforma-
`tion (Fig. 3A). Indeed, these loops are much less
`hydrophobic than the amino-terminal fusion peptides
`of class I proteins (even when the three fusion
`domains of G are grouped together in the post-fusion
`conformation). That these loops are indeed an essen-
`tial part of the membrane interacting motif is con-
`sistent with previous mutagenesis work performed on
`rhabdoviruses [59, 60] (Table 1) and has since been
`confirmed for both VSV G [61] (Table 1) and Her-
`pesviruses gB [62, 63].
`Although hydrophobic photolabeling experiments
`have demonstrated the ability of G fusion domain to
`insert into the target membrane as a first step of the
`fusion process [31, 64], it is clear, from the presence of
`charged residues in the vicinity of the loops (as in the
`fusion domain of class II fusion proteins), that any
`deep penetration inside the membrane is precluded
`(Fig. 5). Rather, the tryptophans and tyrosines that are
`found in the fusion loops of all the rhabdoviral G
`proteins (Fig. 5) act as sticky fingers by positioning
`themselves at the interface between the fatty acid
`chains and head group layers of lipids. It is probable
`that this interfacial interaction involving only a few
`residues does not create a strong point of anchoring
`that can be used to pull the target membrane toward
`the viral one. Rather, we propose that by perturbating
`the outer leaflet of the target bilayer, it facilitates the
`
`Page 5 of 13
`
`

`

`Cell. Mol. Life Sci.
`
`Vol. 65, 2008
`
`Review Article
`
`1721
`
`Figure 4. (A) Ribbon diagrams of paramyxovirus F trimers in the pre-fusion (left) and post-fusion (right) conformations colored by
`domains (PDB: 2B9B, 1ZTM [36, 37]). Domains and segments are named according to [37]. Note that the two conformations are related by
`flipping the fusion peptide (FP) and the C-terminal transmembrane domain (TM) relative to a rigid block made of DI and DII. (B)
`Comparison of the central six-helix bundle of the post-fusion conformation of VSV G with the one of hPIV3 F. hPIV3 F segments are
`colored according to (A). VSV G helices of the trimerization domain are colored as their hPIV3 F counterpart. The position of the
`pleckstrin homology (PH), fusion and transmembrane (TM) domains of VSV G are indicated as well as the fusion peptides (FP) and the
`transmembrane (TM) domains of hPIV3 F. (C) Comparison between the conformational changes of influenza HA2 (left) (PDB: 1HGG,
`1HTM [34, 35]) and the trimerization domain of VSV G. For both polypeptides, unchanged secondary structures are colored in blue; the
`segment that refolds to form the central helix in the post-fusion conformation is in orange; and the C-terminal part, linked to the
`transmembrane (TM) domain, that positions itself in the groove of the central core is in red. The position of the pleckstrin homology (PH),
`fusion and transmembrane (TM) domains of VSV G are indicated as well as the fusion peptide (FP) and the transmembrane (TM) domains
`of HA2.
`
`formation of pointlike protrusions that have been
`proposed to be stalk precursors [65].
`In Gth pre-fusion conformation, the fusion loops point
`toward the viral membrane (Fig. 3A). The distance
`between these loops and the viral membrane essen-
`tially depends on the organization of the C-terminal
`segment of the ectodomain (amino acids residues
`414 – 446) that leads to the TM domain. The structure
`of this segment is not known, but in the case of RV G, it
`has been proposed to adopt an a-helical conformation
`[66] having a strong amphipathic signature. Depend-
`ing on the orientation of this putative helix (either
`
`lying flat at the membrane surface or as part of a
`central trimeric helical bundle perpendicular to the
`viral membrane), the fusion loops may also interact
`with the viral membrane when G is in its pre-fusion
`conformation.
`
`pH sensitive molecular switches
`
`Both structures revealed conserved clusters of resi-
`dues that play the role of pH-sensitive molecular
`switches. In the pre-fusion conformation [55] there is a
`
`Page 6 of 13
`
`

`

`1722
`
`S. Roche et al.
`
`VSV glycoprotein structure
`
`Table 1. Mutations in rhabdoviral G that affect either the fusion properties or the conformational change of VSV G.
`
`Mutation
`
`Corresponding
`residue in VSV G
`
`Phenotype
`
`Molecular explanation
`
`VSV F2L*1
`
`stabilization of the pre-fusion
`structure
`
`Residue F2 makes Van der Waals contact with residues I387 and
`M391 of the outer helix in the post-fusion hairpin.
`
`RV M44V*2
`RV M44I*2
`
`Q42
`Q42
`
`slower transition from the pre-
`to the post-fusion structure
`
`Residue located in the hinge region between the fusion and PH
`domain.
`
`VSV G108A3
`VSV G108E4
`VSV F109Y3
`VHSV F132K5
`VHSV P133K5
`VSV P111D4
`VSV P111G3
`VSV P111L3
`
`VSV W72V6
`VSV W72A6
`VSV Y73V6
`VSV Y73A6
`VHSV F100K5
`VHSV W139K5
`VSV Y116V6
`VSV Y116A6
`VSV A117R6
`VSV A117H6
`VSV A117K4,6
`
`F109
`P110
`
`Y73
`Y116
`
`RV E282K*2
`VSV Q285R*1
`
`L283
`
`RV V392G*2
`RV M396T*2
`
`A402
`E406
`
`fusion poorly efficient and/or
`only detected at lower pH
`
`These residues are located in the polyproline helix upstream the
`second fusion loop. This structure is involved in lateral interactions
`with the neighbouring protomer in the post-fusion conformation.
`
`no fusion activity
`
`Replacement of a hydrophobic residue by a charged one in the fusion
`loops or replacement of an aromatic residue by an aliphatic one that is
`less prone to destabilize the interface between the fatty acid chains
`and head group layers of lipids.
`
`stabilization of the pre-fusion
`structure
`
`Residues located in the central helix. Their mutations probably affect
`the trimeric interface in one or both conformations.
`
`stabilization of the pre-fusion
`structure
`
`Residues located in the hinge region between the trimerization
`domain and the C-terminal part of the protomer.
`
`For RV G and VHSV G mutants, the residue of VSV G that aligns with the mutated position is indicated. The asterisks indicate that the
`corresponding mutations have been found in a mutant virus. Other mutations were introduced in a plasmid allowing G expression (and the
`fusion was assayed by syncytia formation). 1 From [79]. 2 From [80]. 3 From [81]. 4 From [59]. 5 From [60]. 6 From [61].
`
`Figure 5. Sequence alignment of
`the two regions corresponding to
`the tip of the fusion domain of
`rhabdoviral G and close-up view
`of
`the tip of VSV G fusion
`domain showing the two fusion
`loops. Amino acids that consti-
`tute the fusion loops are in bold
`letters. Hydrophobic residues in
`the fusion loops are in black.
`Charged residues that impede
`deep penetration of the fusion
`domain in the membrane are in
`grey. BEFV: Bovine ephemeral
`fever virus. VHSV: Viral hemor-
`rhagic septicaemia virus.
`
`cluster of three histidines: H60 and H162, both located
`in the fusion domain, and H407, located in the C-
`terminal part of the protein. Low pH-induced proto-
`nation of these residues leads to a cluster of positive
`charges that might trigger the movement of the fusion
`domain toward the target membrane. This is consis-
`tent with the observed inhibition of VSV fusion by
`
`diethylpyrocarbonate, a reagent known to modify
`specifically histidine residues [67].
`Conversely,
`in the post-fusion state [54] a large
`number of acidic amino acids are brought close
`together in the central six-helix bundle. In this protein
`conformation, they are protonated and form hydrogen
`bonds. The deprotonation of these residues at higher
`
`Page 7 of 13
`
`

`

`Cell. Mol. Life Sci.
`
`Vol. 65, 2008
`
`Review Article
`
`1723
`
`Figure 6. Ribbon diagrams of the pre- and post-fusion conformations of VSV G trimers. The position of the mutations that affect either the
`conformational change or the fusion properties of G are colored as indicated in Table 1. Close-up views of the mutations located in the
`trimeric interface (blue) and of the environment of residue F2 (red) are also shown.
`
`pH will induce strong repulsive forces that destabilize
`the trimer and initiate the conformational change
`back toward the pre-fusion state (in which these
`residues are solvent-exposed). As the post-fusion
`form makes the majority of the population up to pH
`6.5 [32], the pKa of these residues in the post-fusion
`conformation is much higher than their pKa in the pre-
`fusion conformation in which they are solvent-ex-
`posed. Thus, the post-fusion conformation has a
`stronger affinity for the protons than the pre-fusion
`one. This explains the cooperativity of rhabdoviral G
`structural transition as a function of pH [32].
`
`Mutations affecting the structural transition and/or
`the fusion properties
`
`A large number of mutations have been described that
`affect either the structural
`transition,
`the fusion
`properties of rhabdoviral G or both. The phenotype
`of some of these mutations most probably results from
`folding defects often associated with a decrease in
`transport efficiency. Once this trivial explanation has
`been discarded, the remaining mutants can be classi-
`fied in different categories (Table 1). In general, the
`
`phenotype of the mutants that affect the stability of
`the pre-fusion structure can be explained at a molec-
`ular level simply by localizing the residue (e.g. those
`located in hinge regions of the molecule) or by
`comparing its environment in the pre and post-fusion
`conformations (Fig. 6). It is also easy to understand
`why mutations of residues located in the fusion loops
`abolish G fusion activity (Fig. 6, residues in green).
`Nevertheless, the fusion phenotype of G having
`mutations in the segment (Fig. 6, residues in yellow)
`that adopts a polyproline conformation is not that easy
`to explain. In the post-fusion conformation, interac-
`tions between this segment and b-strand d of the
`fusion domain of the neighbouring partner close the
`bottom of the trimers internal cavity and keep
`together the lower part of the fusion domains and
`thus the fusion loops of the three protomers. Although
`this underlines the importance of this segment in the
`regulation of the structural transition and suggests its
`possible involvement in the oligomerization of inter-
`mediates, it is not clear how these mutations affect the
`efficiency of the fusion process or decrease the
`optimal fusion pH.
`
`Page 8 of 13
`
`

`

`1724
`
`S. Roche et al.
`
`VSV glycoprotein structure
`
`Figure 7. Two possible pathways for the transition from pre- to post-fusion of VSV G at the surface of the viral membrane (depicted by a
`gray dotted line). Coloured as in Figure 1, with the magenta dotted lines representing the missing part of the ectodomain. 1, pre-fusion
`trimer, 6, post-fusion trimer, (2’) and (5), pre- and post-fusion protomers, respectively, in parentheses to indicate that these forms are likely
`transient. The remainder of the forms depicted here are possible intermediates including some but not all of the structural changes. (A)
`Transition assuming the refolding events occur in the order shown in [55]. 1 to 2 (rotation around the DIV-DIII hinge) can occur without
`trimer dissociation, but 1 to 4 (adding refolding of the DIII-DII connections) very likely entails monomerization, as the lengthening of the
`central helix F would lead to clashes at the pre-fusion trimer interface. Refolding of the C-terminus of DII then leads to the post-fusion
`protomer (5). This last transition likely occurs at the same time as or even after trimer formation, as both events bury acidic amino acids in
`interfaces (Fig. 3 of [54]). (B) Transition assuming refolding of the C-terminus of DII into helix H (shown in red in Fig. 4C) occurs first (1 to
`3’). This implies trimer dissociation [shown as (2)], as this segment makes almost all the contacts with the neighbouring protomer. Note
`that, as for 4, 3’ projects the fusion loops away from the viral membrane. 4’, putative intermediate adding the refolding of the DII-DIII and
`DIII-DIV connections. Finally, 4’ to (5) and 6 is achieved by the packing of helix H against DI (forming the hydrophobic cluster I387-M391-
`F2 shown in Fig. 6) and DII (bringing D393 and E276 into hydrogen-bonding distance, Fig. 3A of [54]).
`
`Intermediates on the fusion pathway
`
`One intermediate on the fusion pathway exposes the
`fusion domain at the top of the molecule. This allows
`the interaction of the fusion loops with the target
`membrane but is also responsible for the low pH-
`induced viral aggregation. For rabies virus,
`this
`intermediate activated state can be stabilized at low
`temperature and pH 6.4 [30, 31]. EM data [30] and
`antibody binding [32] suggest that the structure of this
`state is not that different from the native state, raising
`some questions about the structure of this intermedi-
`ate (Fig. 7).
`Although, for Gth in solution, it seems possible to go
`from the pre-fusion trimeric form to its post-fusion
`counterpart without dissociation and without break-
`ing the threefold symmetry, this seems difficult at the
`viral surface when the conformational change occurs
`in the absence of a target membrane (Fig. 7). As there
`are large differences between the trimeric interfaces
`of the pre and post-fusion conformations [55], a
`possible scheme for the pre- to post-fusion transition is
`that it goes through a transient monomer (as it is the
`
`case for class II fusion proteins). This view is also
`consistent with previous results showing that the pre-
`fusion trimer monomerizes after detergent solubiliza-
`tion [68] and that there is an equilibrium between
`monomers and oligomers in solution [69]. Two
`plausible pathways are depicted in Figure 7. The first
`one assumes that the refolding of the molecule implies
`the initial rotation of the fusion domain (rotation
`around the DIV-DIII hinge), whereas the second
`postulates that the refolding of the C-terminus of DII
`into helix H (shown in red in Fig. 4C) occurs first.
`
`Cooperativity between a large number of
`glycoproteins
`
`The activation energy of the fusion process has been
`estimated to be in the range of 40 kcal/mol [26, 70, 71],
`most of which is required by enlargement of the initial
`fusion pore. For class I and class II fusion proteins, it
`has been proposed that the energy released during the
`irreversible fusogenic transition is used to achieve the
`energetically expensive membrane-f