|Home | About | Journals | Submit | Contact Us | Français|
Enveloped viruses enter cells by fusing the viral and cellular membranes, and most use a single viral envelope protein that combines receptor-binding and fusogenic functions. In herpesviruses, these functions are distributed among multiple proteins: the conserved fusion protein gB, various non-conserved receptor-binding proteins, and the conserved gH/gL heterodimer that curiously lacks an apparent counterpart in other enveloped viruses. Recent structural studies of gH/gL from HSV-2 and EBV revealed a unique complex with no structural or functional similarity to other viral proteins. Here we analyzed gH/gL structures and highlighted important functional regions. We propose that gH/gL functions as an adaptor that transmits the triggering signals from various non-conserved inputs to the highly conserved fusion protein gB.
Enveloped viruses enter host cells by fusing their envelopes with the cellular plasma membrane or the membrane of an endocytic vesicle. This process is initiated by binding of a virus to its cellular receptor and is catalyzed by a viral fusion protein. In most enveloped viruses, a single protein is responsible for the receptor binding and the fusogenic functions. Conformational changes in the receptor-binding module upon receptor interaction are thought to trigger fusogenic conformational changes in the fusion module. In some viruses, such as paramyxoviruses, these two functions are split between two viral proteins.
Herpesviruses are a large family of dsDNA, enveloped viruses. Eight human herpesviruses cause life-long infections and a variety of diseases, including skin lesions, encephalitis, cancers, and disseminated disease in the immunocompromised. Like all enveloped viruses, herpesviruses enter cells by fusing their envelopes with the host membrane. But, unlike most enveloped viruses, all herpesviruses utilize complex multiprotein entry machinery that consists of three conserved proteins: gB, gH, and gL, plus additional non-conserved proteins . This level of complexity is remarkable and raises the question virologists have asked for several decades: why do herpesviruses need several proteins to achieve entry?
In herpesviruses, the receptor-binding and the fusogenic functions are distributed among multiple proteins. The mechanistic details are best understood for Herpes Simplex viruses Type 1 and 2 (HSV-1 and HSV-2) and Epstein-Barr virus (EBV). The receptor-binding function is typically associated with a non-conserved protein, such as HSV gD  and EBV gp42 . Binding of gD or gp42 to their cellular receptors is thought to trigger the conserved membrane fusion machinery of herpesviruses composed of glycoprotein gB and the gH/gL heterodimer. gB is class III viral fusion protein , sharing a structural similarity with vesicular stomatitis virus glycoprotein G  and baculovirus gp64. The precise function of the gH/gL heterodimer is unclear, although it interacts with gB  and is required for gB function. Some herpesviruses such as Kaposi’s sarcoma herpesvirus (KSHV) use gH/gL to bind cellular receptors , but the requirement of gH/gL for entry in all herpesviruses indicates a more conserved function. Initially postulated to have fusogenic properties, gH/gL does not resemble any known viral fusion proteins nor does it have a known functional counterpart in other enveloped viruses.
This review focuses on how structural biology has shed a new light on this enigmatic component of the conserved fusion machinery of herpesviruses. Because structures are available for HSV-2  and EBV gH/gL complexes , the review primarily discusses these proteins. A partial structure of PRV gH  is available but is not discussed here in detail. We examine what is known about the gH/gL complex through the prism of the recent structural findings and map functional information onto structures to gain clues regarding the mechanism of gH/gL function. In essence, gH/gL receives a triggering signal from a receptor or receptor-binding protein and transmits it to gB, working as a middleman.
gH has a large ectodomain and a single C-terminal transmembrane anchor while gL lacks a transmembrane region. The two proteins form a stable 1:1 complex. The crystal structure of gH/gL ectodomain from HSV-2  revealed an unusual boot-shaped complex (Fig. 1a), reminiscent of the 15-20-nm curved glycoprotein spikes observed by cryo-electron tomography . Given the conservation of gH/gL in sequence and function, it was a surprise when the subsequently determined structure of EBV gH/gL  showed a cylindrical molecule (Fig. 1a). Yet despite different spatial arrangement, individual domains within HSV-2 gH align well with their EBV counterparts (Fig. 1b). To highlight fundamental similarities between HSV and EBV gH molecules and to reconcile somewhat different domain assignments proposed for the two structures, we divided gH into H1A, H1B, H2A, H2B, and H3 domains, from N to C terminus (Fig. 1a,b). The distinct shapes of two gH/gL complexes, boot vs cylinder, arise from differences in relative domain orientations, especially at the H1B/gL and H2A/H2B interfaces, demonstrating a certain structural plasticity.
gL is “sandwiched” between domains H1A and H1B and interacts extensively with both, illustrating why gH and gL are always found as a complex. H1A forms a continuous 5-stranded β-sheet with gL and is probably unstructured in the absence of gL. H1A is the target of several anti-HSV neutralizing (or inhibiting spread) antibodies [12, 13] and some mutations that disrupt HSV fusion  or affect VZV tropism . No functional role has yet been assigned to the much smaller H1A of EBV gH/gL. The relatively unstructured N terminus of domain H1B has many contacts with gL as well as a long lasso-like loop that interacts with H2A and H2B. The C terminus of domain H1B forms a 5-stranded β-sheet, a “picket-fence”  that wraps around domain H2A. Both EBV and HSV gH/gL contain non-conserved integrin-binding motifs within domain H1B.
Sequence conservation in gH increases from N to C terminus, so domains H2A, H2B, and H3 share a higher degree of structural similarity. The presence of a three-helix bundle in H2A evokes a structural similarity with other helical bundles, e.g., syntaxin, while the helical arrangement in the crescent-shaped H2B is reminiscent of HEAT repeats .
The C-terminal domain H3 has a β-sandwich fold. It is the most highly conserved domain and likely critical for fusion. H3 in HSV, EBV, and VZV gH is very sensitive to point mutations and insertions as most abolish fusion [15-19] or alter tropism in EBV [18-20]. A long extended polypeptide, termed a “flap” , wraps around one side of the β-sandwich (Fig. 2a,b). The hydrophobic surface patch underneath the flap was postulated to interact with membranes, were the flap to relocate . In support of this idea, replacement of some residues with more hydrophobic ones enhances fusion or rescues fusion reduced by other nearby mutations [15, 19] (Fig. 2a,b). The relocation of the flap and membrane interaction remains to be demonstrated.
gL is required for correct folding, surface expression, and function of gH in HSV [13, 21, 22] and EBV . gL is thus a scaffolding protein for gH , perhaps, with additional functions. Despite the lack of obvious sequence similarity, the HSV-2 and EBV gL structures can be aligned (Fig. 1c). Both structures feature a chemokine fold - a three-stranded β-sheet and a helix (Fig. 1c) - recognized earlier through bioinformatics analysis . A CC chemokine motif consisting of two adjacent conserved disulfide bonds is found in EBV gL, while in HSV gL, a single disulfide bond stabilizes the chemokine fold (Fig. 1c). These gL structures align well to both CC and CXC chemokines. Many chemokines dimerize by forming an intermolecular β-sheet. Interestingly, a structurally analogous β-sheet is formed by residues from domain H1A and gL (Fig. 1d). These similarities raise the interesting possibility that the gL gene may have originated from a chemokine. Surprisingly, the HCMV gL sequence cannot be aligned with either HSV or EBV gL and probably does not share a chemokine fold. Instead, CMV UL128, a component of CMV gH/gL/UL128/UL130/UL131 complex, shares sequence similarity with CC chemokines . Moreover, its rat homolog functions as a CC chemokine . HSV gL is secreted from mammalian cells in the absence of gH [27, 28], and soluble gL could, in principle, be produced during infection. It is unknown, however, whether gL can function as a chemokine, and this possibility awaits future investigation.
The ability of herpesviruses to enter multiple cell types is reflected in the variety of viral glycoproteins used to engage cellular receptors . HSV-1 and HSV-2 use gD to bind nectin-1 on epithelial cells and neurons and herpesvirus entry mediator (HVEM) on lymphocytes [2, 29, 30]. In EBV, gp42 binds HLA Class II molecules during B cell entry [3, 31] while gH/gL binds αvβ5, αvβ6, and αvβ8 integrins on epithelial cells [32, 33]. Entry of CMV into fibroblasts requires a gH/gL heterodimer  but entry into epithelial and endothelial cells requires a pentameric complex gH/gL/UL128/UL130/UL131 [35, 36].
Binding of these viral proteins to their cellular receptors triggers membrane fusion, likely by first activating gH/gL. In HSV, receptor-bound gD presumably interacts with gH/gL [6, 37], but the gD-gH/gL complex has not yet been captured directly and the gD-binding site on gH/gL has not been identified. More is known about the site in EBV gH/gL that receives a triggering signal from the receptor. During entry into epithelial cells, gH/gL receives a triggering signal directly from integrin receptors [32, 33]. During B cell entry, gH/gL receives a triggering signal from HLA class II through gp42, which it binds . The binding sites for integrins and gp42 in gH/gL likely overlap because mutation of the integrin motif KGD reduces not only binding to epithelial cells and epithelial cell fusion but also binding to gp42 and fusion with B cells . Thus, gp42 acts as a tropism switch by binding the integrin-binding site and inhibiting the gH/integrin interaction while simultaneously promoting entry into B cells. Importantly, regardless of where the triggering signal originates, EBV gH/gL receives it in the same location (Fig. 2b). The integrin-binding motif RGD in HSV gH  is neither conserved nor required for gH/gL function .
Activated gH/gL is thought to help gB achieve its fusogenic potential [8, 41]. In HSV, gB and gH/gL interact, presumably through their ectodomains, only after gH/gL has been exposed to receptor-bound gD [6, 42]. This suggests that gH/gL activates gB in response to a gD/receptor interaction rather than represses gB in its absence. The gB-gH/gL interaction precedes membrane fusion, is inhibited by neutralizing antibodies, and is required for fusion [8, 41, 42]. Although the gB-binding site on gH/gL has not yet been identified, anti-HSV-1 gH/gL neutralizing antibody LP11 blocks interaction. The LP11 epitope is defined by monoclonal antibody neutralization resistance (mar) mutations  and several insertion mutants that block LP11 binding . Because LP11 competes with gB for binding to gH/gL, the gB binding site on gH/gL likely overlaps the putative LP11 epitope and has been proposed to be located in a nearby groove  (Fig. 2a).
Little is known about how gB-gH/gL binding triggers membrane fusion. Interestingly, anti-HSV-1 gH/gL neutralizing antibody 52S blocks fusion without blocking the gB-gH/gL interaction . The 52S epitope in HSV-1 gH is defined by mar mutations  and maps to the opposite side of gH/gL from the LP11 epitope (Fig. 2c). This explains why LP11-resistant viruses are sensitive to 52S and vice versa . Less clear is how 52S blocks fusion. While it allows gB-gH/gL interaction, it may block the next step: gB activation. We propose that 52S prevents a conformational change that gH/gL must undergo to activate bound gB. In our model, gB binds to its putative site near the LP11 epitope (Fig. 2a,c). This binding impinges upon the well-aligned series of alpha helices in domains H2A and H2B, displacing them along with the helix containing the 52S epitope (Fig. 2c). We propose that gH/gL requires this or a similar kind of a conformational change to activate gB, and that 52S binding blocks this conformational change and thus gB activation. Such long-range conformational change would require a certain structural flexibility within gH/gL, which is consistent with the fact that both H2A and H2B tolerate insertions, a few in buried regions . A double point mutation in VZV gH, designed to destabilize helix-helix interactions, is fusion-null  and pinpoints a region that could be critical in transmitting the gB-activating intramolecular signal within gH/gL (Fig. 2c).
No known functional sites have yet been reported within a structurally analogous region of EBV gH/gL. Interestingly, in EBV gH/gL, a nearby site may be involved in binding gB or controlling its activation through a long-range effect (Fig. 2b). gL from a closely-related rhesus lymphocryptovirus (Rh-LCV) binds EBV gH, but does not substitute for EBV gL in virus-free cell fusion assays. Yet, replacing residues 54 and 94 in Rh-LCV gL with their EBV gL counterparts restores full function , suggesting that these residues somehow specify the selection of gB. Further experiments are needed to clarify where gH/gL binds gB and how this interaction activates gB for fusion. Considering conservation of gH/gL and gB, we expect that the gB-binding site is likely conserved in all herpesviruses. But, given the structural plasticity of gH/gL domain arrangement, the gB-activating conformational changes within gH/gL could vary.
Why do all herpesviruses require gH/gL, an additional protein shuttling signals from the receptor-binding protein to the fusion protein? Why not cut out this middleman? It is notable that fusion protein gB is highly conserved across all herpesviruses while cellular entry receptors and viral receptor-binding proteins vary greatly (Fig. 3). The evolutionary pressure on gB to maintain function as a fusogen may be too high to allow it to evolve to accommodate many different inputs. The high sequence and structure conservation of gB supports this idea. Perhaps, the solution was to use an “adaptor” that could relate different triggering inputs to a conserved fusion protein. We propose that the main function of gH/gL is to adapt and transmit activating signals to gB, thus allowing gB to be triggered by varied receptors in different herpesviruses and even within the same virus. While conserved, gH/gL sequences and structures are more divergent than those of gB. Structural plasticity in gH/gL may be uniquely suited for intermolecular signal transmission from the receptor or a receptor-binding protein to fusion protein gB. This plasticity may have permitted the divergent evolution of gH/gL and customization for different triggering partners. The versatility provided by gH/gL allows activation signals from a multitude of sources to be transmitted to gB to accommodate entry of herpesviruses into different cell types. Herpesviruses are among the most ubiquitous in nature, targeting nearly all mammals and even non-mammals. The presence of gH/gL as an evolutionarily malleable adaptor could have allowed herpesviruses to acquire new cell targets without compromising the function in fusion protein gB.
Herpesviruses require multiple viral proteins to achieve cell entry.
Receptor-binding and fusogenic functions are distributed among several proteins.
Conserved gH/gL lacks apparent functional analogs in other viruses.
HSV-2 and EBV gH/gL structures reveal structural plasticity and important functional regions.
gH/gL is an adaptor transmitting different signals to conserved fusion protein gB.
The authors thank Claude Krummenacher for critical reading of the manuscript. E.E.H. acknowledges funding support of the NIH grant 1DP20D001996 and the Pew Scholar Program in Biomedical Sciences.
Papers of particular interest, published within the period of review, have been highlighted as:
* of special interest