Insights into Herpes Simplex Virus Fusion Protein

Abstract

Herpes Simplex Viruses is a model Herpesvirus, of which eight afflict humans. Herpes Simplex Virus enters the host cell by the sequential conformation changes and activations in gD, gH and gL, culminating with gB, the fusion protein. While there are several structures for HSV gB in post fusion conformation up until recently there has been no high-resolution structures of any Herpes Virus gB in their pre-fusion state. HCMV’s gB in pre-fusion state was recently published and was achieved by the use of tomography on a Titan Krios equipped with a Volta Phase Plate. At UoL we have also used this technique to get a higher resolution structure than the previous along with developing an automated pipeline we can then apply to other samples.

Introduction

Herpesviruses infect many vertebrate and at least one invertebrate hosts. They include over 100 viruses, of which eight cause human infections, Human Herpes Viruses 1-8 (HHV1-8). These include Herpes Simplex Viruses (HSV) 1 and 2, Varicella-zoster virus (VZV), Cytomegalovirus (CMV), Epstein-Barr virus, HHV6,7 and 8. HHV1-8 are further classified into either alpha, beta or gamma herpes viruses. Alphaherpes viruses are neurotropic and include model Herpesvirus family viruses which this paper focuses on, HSV1 and HSV2.

Save your time!
We can take care of your essay

• Proper editing and formatting
• Free revision, title page, and bibliography
• Flexible prices and money-back guarantee

Place an order

HSV is a model system for the Herpesvirus family and has two serotypes, HSV-1 and HSV-2, that globally infect approximately 90% of the population. HSV inflicts lifelong infections by establishing latency in the host and undergoes periodic reactivations that can spread the virus. These infections normally manifest as mucocutaneous infections including keratitis, gingivostomatitis and genital warts. While both serotypes can cause genital lesions, HSV-2 usually results in more recurrences and viral shedding than HSV-1 [1] Furthermore, infection with HSV-2 increases the risk of acquiring and transmitting HIV [2]. While most HSV infections are unpleasant, more complicated infections can occur in neonates and the immunocompromised, a growing cohort [2-4]. Infection with HSV may also extend further than current knowledge with viruses being linked to many diseases that they were not orginally thought to be associated with, such as cancers and neurological disorders, such as Alzheimer’s [5].Specific antivirals limit the impact of HSV but none cure infection. Coupled with the lack of a preventative vaccine, this virus will continue to afflict the population, making it a global health burden of high priority.

HSV has a linear DNA genome of approximately 152Kb packaged tightly in an icosahedral capsid, which is 15nm thick and 125nm in diameter and exhibits icosahedral symmetry. The complete capsid was solved in 2018 to 3.1 Å [6] using cryo-EM. The capsid itself is encased in a matrix of 20 proteins (the tegument), that lies beneath a host derived lipid envelope, decorated with 10-12 glycoproteins [7]. Therefore, the size and complexity of HSV make structural studies extremely challenging.

As with all enveloped viruses, HSV infection begins with entry, a process that requires fusion of cellular and viral membranes. While the molecular details are still not known, all events are thought to follow the fusion-through-hemifusion pathway [8].

HSV membrane fusion is mediated by four glycoproteins: the primary receptor binding protein gD, a covalently linked heterodimer gH/gL, and the fusion protein, gB. HSV fusion begins with the interaction of gD with a cellular receptor. This interaction induces a conformational change in gD, prompting gH/gL to activate gB. Successive rearrangements of gB, from its initial metastable pre-fusion conformation to the more energetically favoured post-fusion conformation, lead to membrane curvature and disruption of cellular membranes, resulting in viral capsid release into the host cell [10].

Several structures of gD exist , including unliganded gD and in complex with its receptors (reviewed in [11]) and for a partially activated form of gH/gL [12]. However, only the post-fusion structure of gB has been solved [13]. This is because all purified forms of gB adopt the post-fusion conformation, and attempts to change this have been unfruitful [14]. This leaves an important gap in the knowledge of the HSV lifecycle.

HSV-1 gB is comprised of 904 residues and is a trimer in the post-fusion conformation. Side views depict it as a three-lobed structure. The truncated post-fusion structure identifies five domains that place the two fusion loops in domain I. Both domains I and V are at the “base” of the protein, in close proximity to the viral membrane. The central domain II is postulated to mediate interactions with gH/gL and is connected to the trimeric coiled-coil, domain III( ??). Domain IV, the “crown”, resides at the top, tethered by domain III [15]. The N-terminus (residues 31-102, putatively domain VI), is not resolved in the crystal structure due to its flexibility. Amino acids 730-904, which are missing in the purified proteins used for crystallographic studies, include the cytoplasmic tail, the transmembrane domain and the membrane-proximal region, all of which are involved in virus fusion and infectivity. However, in 2018 Cooper et al. [16] were able to resolve this missing information with the aid of specialised cryo-EM grids.

Viral fusion proteins are categorized into three distinct groups, I, II and III. As a class III fusion protein, gB is composed of -helices and -sheets, and contains two fusion loops per protomer. Class III fusion proteins are found in Herpesviruses, Vesicular stomatitis virus (VSV), Human Cytomegalovirus (HCMV) and Baculovirus. The VSV fusion protein, G, is the best characterized class III fusion protein and its post-fusion form shares features similar to gB [17, 18] (Figure x). Based on the structures of pre- and post-fusion G, Gallagher et al. created an in silico model for pre-fusion gB [19, 20]. To generate it, they proposed that gB’s pre-fusion domain arrangements are similar to G in its pre-fusion conformation, and accordingly gB’s fusion loops would point toward the viral membrane. Therefore, by analogy to G, during the transition from its pre- to post-fusion conformation, the fusion loops would first relocate to the top of gB, to interact with the target membrane. Further conformational changes would position the fusion loops of gB close to the transmembrane domains, leading to the merging of the cell and virus membranes. This model is supported by an in-depth structural study using fluorescent proteins (FP) to map gB’s domains, which suggested that regions allowing insertion of the FPs are exposed [20].

A second model of pre-fusion gB was thereafter proposed by Zeev-Ben-Mordehai et al. [21]. This was generated using cryo-electron microscopy (cryo-EM) to image microvesicles expressing full-length gB. Cryo-EM allows imaging of specimens at atomic or molecular resolution in close-to-native conditions. gB expressed in microvesicles adopted two different conformations: an elongated post-fusion form, and a compact form, putatively pre-fusion gB. They then calculated a 3D average of the compact form, fitting two post-fusion domains of gB (domains I & II) into the average. Based on VSV G, and like Gallagher et al., they assumed that the domains of gB are similar in the pre- and post-fusion conformations. The resulting model suggests that gB’s fusion loops (within domain I) point away from the viral membrane. Therefore, to produce fusion, gB would extend so the fusion loops could reach the target membrane, and then conformational changes, similar to the ones proposed by Gallagher et al., would merge the cell and virus membranes.

Recently, we augmented the microvesicle strategy [21] to produce gB in its pre-fusion form [22]. Using cryo-EM, we imaged vesicles expressing full-length gB bound to monovalent antibody fragments that do not possess an Fc region (Fabs) and to whole antibodies, along with gB containing genetically encoded FP insertions. Since the Fabs, antibodies and FPs were visible by cryo-EM, we used them as landmarks to map the position of gB domains in its pre-fusion conformation. According to our experimental data, we proposed that, initially, gB [22] has the fusion loops pointing toward the viral membrane, thereby agreeing with the model proposed by Gallagher et al. Additionally, some samples trapped intermediate conformations of gB, providing insights about how the pre- to post-fusion transitions could take place. Based on these intermediate conformations, we suggested that the fusion loops of gB that initially point toward the viral membrane are relocated to the top of the molecule as a second step in the fusion process, while gB maintains a compact conformation. This intermediate conformation would therefore be similar to the one proposed by Zeev-Ben-Mordehai et al., reconciling the two models for two conformations of gB. More data will be needed to unequivocally unravel the pre-fusion structure of gB and its transition to the post fusion form, thereby elucidating the mechanism of fusion.

More recently, Zhu Si et al. [23] solved the structure of HCMV’s fusion protein, gB utilising a Volta Phase Plate (VPP) to a resolution of ~21 Å. There are several options on the market to purchase for phase plates but currently and arguably the Volta Phase Plate (VPP) solves many issues its predecessors does not. Using these new technologies and a newly developed Subtomogram averaging pipeline at UoL we have been able to attain similar resolutions for HSV’s fusion protein, gB. This structure will help with rational drug design and vaccine development to tackle HSV infection.

Material & Methods

Microvesicle Production

293T Cells were cultured in DMEM containing 10% FBS and 100 µg/ml Penicillin-streptomycin. Transfection was done in 6 well plates seeded with x X 10X/per well. All media was changed to DMEM containing 10% exosome-depleted FBS and contained no penicillin-streptomycin once seeded. Exosome-depleted FBS was created through centrifugation for an extended period of time then the use of a filter to discard exosomes. Transfection reagent was then added to the gB plasmid then onto the cells. Media on the cells was collected for processing to extract the vesicles using cushions and centrifugation. Vesicles studded with gB was then resuspended overnight.

Cryo-Electron Grid Making

Grids were made using the Lecia Plunge Freezer. Sample was mixed 1:1 with 10nm gold beads then 3ul of sample mixture was added to the grid and various blot times were using between 2-5seconds. Humidity was always set at the maximum on the machine. Samples were then plunge frozen in liquid ethane cooled by liquid nitrogen. Grids were stored in liquid nitrogen.

Cryo-Electron Tomography

DIAMOND Data 48hr session 2016 using an automated tomography session via EPU (FEI). Performed by Corey Henderson.

UoL VPP

UoL Titan Krios equipped with a Volta Phase Plate was used for an automated tomography session via EPU (FEI). 12 Tomograms were originally acquired automatically using EPU (FEI) then another x amount was collected the same way.

Tomogram Reconstruction & Subtomogram Averaging

All tomograms were reconstructed via the IMOD package in Batch mode after motion correction using Motioncorr within RELION. A bin factor of 2 was univerwsally applied to all tomograms. All volumes in the tomograms were then individually trimmed for further processing. For reference generation a small amount of pre-fusion spikes alongside post-fusion spikes were manually picked and run through PEET (IMOD).

Trimmed volumes were then separated into two discrete groups, spherical and irregular. All spherical volumes were taken forward where seedSpikes (IMOD) was performed, then sequentially spikeInit (IMOD). From here iterations in PEET (IMOD) were preformed and reference refinement was able to proceed.

Results

Manual Tomogram Reconstruction in Comparison to Automatic Tomogram Reconstruction

Low-resolution results, from a 120Kv T12 (FEI) microscope, confirmed vesicles with two types of spikes were present (FIGURE X). Landmark mapping using Fabs and FPs to increase the attainable resolution the sample and pinpoint where key domains were was also preformed (Fontana et al 2017). This provided the evidence necessary to progress to a 300Kv Titan Krios Microscope. The first tilt series acquisitions consisted of 31 tomograms. These were reconstructed first using IMOD in manual mode, then in batch mode which is the automatic version. The results were then compared and showed no significant differences and so batch tomography was confirmed in the workflow pipeline for future datasets (FIGURE X).

Volta Phase Plate Data Compared to Standard Tomography

The next dataset was taken to improve the inherently low contrast in tomography. As shown in FIGURE X, particle picking was strained due to the low resolution achieved. A small preliminary dataset was taken using the UoL Volta Phase Plate with a defocus of -0.5um. This dataset of 12 tomograms was then automatically reconstructed using Batch tomo. FIGURE X, shows tomograms automatically reconstructed with and without a VPP, (FIGUREX). Manual picking of the spikes on the tomograms gave two distinct spikes of compatible resolution to previous datasets. Using x amount of spikes two references were generated (FIGUREX) for pre and post fusion gB.

From this we were also able to fit both proposed gB models into the densities for an initial comparison.

Volta Phase Plate Collection and Automatic Picking Pipeline

With the success of the preliminary data acquisition the next stage was to compare an automatic and manual picking pipeline. Automatic picking consisted of first defining the center of the vesicles so all circular vesicles were taken forward for this approach. Their center points were defined manually in 3Dmod then spikeInit (IMOD) and sequentially seedSpikes (IMOD) was used to automatically pick the entire surface. This was then put into PEET (IMOD) for a preliminary iteration using a manually picked reference. Successive iterations and searches getting gradually more comprehensive in Theta, Psi then Phi was utlisied. From this new references were generated and used in a final search in all Euler angles. ScoreHistogram was then able to produce a graph where the minimum cross correlation (MCC) was plotted (FIGURE X). MCC Groups particles in order of how similar they are to the provided reference in PEET. Several MCC values were applied using the createAligned model function (IMOD). From this a MCC of 0.15 was found to be the most effective in first classifying out the post-fusion spikes and other automatically picked points that don’t align with the previously generated reference.

The final Outcome was a density comparable to that of the manual pipeline (FIGURE X). The two proposed models were then again fit into this density.

Hight throughput Volta Phase Plate Pipeline

The final step was to use this automatic picking pipeline on a larger dataset. 67 Tilt series were acquired and automatically reconstructed into tomograms. Of these, most had several vesicles in a single tomogram so all spherical vesicles volumes were trimmed and taken forward. Identical rounds of PEET(IMOD) iterations were preformed and generated references symmetrized (C3) to generate the final volume (FIGURE X)

Discussion

Despite using similar techniques on vesicle production, it has been clear that larger vesicles are spiked with longer, post fusion spikes in comparison to the smaller vesicles. This was noted and for data collections smaller vesicles were normally selected for tilt series acquisition. However, this is in direct contradiction to Zeev-Ben-Mordehai et al. 2016, who reported that larger vesicles typically have the shorter, pre-fusion spikes. They theorised that the longer spikes, the presumed post-fusion preferred the higher curvature of the smaller vesicles. However, speculating it could be because the vesicles produced in this study, when larger varied at a higher frequency so the vast majority were not spherical and therefore the spikes differed to that of the previous study in 2016. However, their modelling suggests that the pre-fusion form has it’s fusion loops pointing outward, away from the viral membrane which could in theory explain why it is so difficult to capture gB in its pre-fusion state and be amenable to structural analysis. It is reasonable to assume that due to the native biological membranes derived into vesicles

The recent publication of HCMVs structure has shed light on the confusion of the position of fusion loops and the overall orientation of Herpes Virus fusion proteins. When the volume of our gB structure is compared to this, it more closely resembles this Herpes Virus fusion protein (FIGURE X). This, along with the model was originally generated from in sillico modelling the post fusion structure of HSV gB into the pre-fusion volume of VSV’s G protein, would suggest that, despite lacking a high-resolution structure of HSV’s gB in the pre fusion form more closely resembles the 2014 model and the 2016 model would therefore be an intermediate. This could also be because of discrepancies in vesicle preparations making their spikes exhibit intermediate forms as we find the shorter forms are more abundant on the smaller vesicles. The 2016 paper interpretation was also referred to as controversial in a recent paper ( 2019). Class I & III fusion proteins also exhibit their fusion loops buried in all the known pre-fusion structures

This improved resolution now matches the 2019 HCMV paper and overcame our earlier issues with large defocus values to increase contrast in the low SNR of tomography leading to lower resolution as discussed in 2019 HCMV paper.

While the use of Subtomogram averaging has provided vital information for proteins that exist in more than one state such as ( S. Hover et al. 2018).

Conclusions

The pipeline of after vesicle production has been improved from low resolution tomography on a 120Kv microscope to a 300Kv microscope equipped with an energy filter and Volta Phase Plate to optimize tilt series acquisition. From this we have been able to implement a higher throughput tomogram reconstruction and sequentially a Subtomogram averaging pipeline that rivals a manual pipeline. This has enabled higher resolution of the elusive HSV gB protein in its pre-fusion form which corresponds with a recently published paper for the equivalent protein in HCMV. Recently another dataset of gB100 (fontana et al) has been acquired using the Titan Krios equipped again with a Volta Phase Plate, energy filter and camera, reconstructed automatically using batch Tomo (IMOD). This will now be put through the automatic pipeline for Subtomogram averaging to map the antibody in greater detail, building on the 2017(?) Fontana et al. paper. Along with this all non-spherical vesicle tomograms already reconstructed will be taken forward with the use of MeshInit (IMOD) to increase particle numbers.

Background Information Not for Main Text

Cryo-Electron Microscopy

Cryo-electron microscopy (cryo-EM) is a powerful structural technique that has recently undergone a ‘resolution revolution’. Whole cells to macromolecules are amenable to cryo-EM. Cryo-EM employs the vitrifying of specimens via rapid freezing. This allows the imaging of specimens at high resolution in close-to-native conditions, thus rivalling x-ray crystallography. Near-atomic resolution de novo model-building is also possible and currently the highest resolution structure is at 1.8A [24].

FEI’s Titan Krios is the world leader in electron microscopes, featuring an ultra-stable Schottky Field emission gun (FEG), flexible high-tension (80kV-300kV), robotic loading (maximum 12 grids), automatic column cooling, automatic and fast data collection among many other cutting edge features [reviewed in [25]. Major limitations to cryo-electron microscopy, which hinder the ability to reach resolutions as high as x-ray crystallography include the signal-to-noise ratio (SNR), charging and sample movement due to exposure to the electron-beam. However, this is largely compensated for with the introduction of Direct Electron Detectors (DEDs).

Cryo-Electron Tomography

Cryo-Electron Tomography (ET) allows the generation of 3D images, such as single particle analysis (SPA). Tomography involves acquiring a tilt series of projection images which is subsequently aligned, computationally, to create a 3D image, a tomogram. Tilting of the specimen is achieved by rotating the stage the specimen is mounted on inside the microscope. Tilting is limited to a range of approximately ±60°, which, results in a “missing wedge” of information in Fourier space. Acquisition of tilt schemes involves a multitude of steps, including, but not limited to, focusing, tracking, alignments and image capture. This process is automated via the use of software packages such as EPU (E Pluribus Unum) & SerialEM. Typically, cryo-ET is sufficient to produce molecular resolution. However, the low SNR of cryo-ET may also be lessened with Subtomogram averaging whereby homogeneous features of a heterogeneous sample may be extracted and averaged.

1. Corey L., W.A.H.K., Sparling PF, Stamm WE, et al., Genital herpes, Sexually transmitted diseases. 2008: New York, McGraw Hill. p. 399-438.
2. WHO. Herpes Simplex Virus. 2017 [cited 2017 17/01]; Available from: http://www.who.int/mediacentre/factsheets/fs400/en/.
3. Dupuis, S., et al., Impaired response to interferon-alpha/beta and lethal viral disease in human STAT1 deficiency. Nat Genet, 2003. 33(3): p. 388-91.
4. Kimberlin, D.W., Herpes simplex virus infections in neonates and early childhood. Semin Pediatr Infect Dis, 2005. 16(4): p. 271-81.
5. Wozniak, M.A. and R.F. Itzhaki, Antiviral agents in Alzheimer's disease: hope for the future? Ther Adv Neurol Disord, 2010. 3(3): p. 141-52.
6. Yuan, S., et al., Cryo-EM structure of a herpesvirus capsid at 3.1 Å. Science, 2018. 360(6384): p. eaao7283.
7. Grunewald, K., et al., Three-dimensional structure of herpes simplex virus from cryo-electron tomography. Science, 2003. 302(5649): p. 1396-8.
8. Chernomordik, L.V. and M.M. Kozlov, Mechanics of membrane fusion. Nat Struct Mol Biol, 2008. 15(7): p. 675-83.
9. Harrison, S.C., Viral membrane fusion. Nat Struct Mol Biol, 2008. 15(7): p. 690-8.
10. Eisenberg, R.J., et al., Herpes virus fusion and entry: a story with many characters. Viruses, 2012. 4(5): p. 800-32.
11. Krummenacher, C., et al., Entry of herpesviruses into cells: the enigma variations. Adv Exp Med Biol, 2013. 790: p. 178-95.
12. Chowdary, T.K., et al., Crystal structure of the conserved herpesvirus fusion regulator complex gH-gL. Nat Struct Mol Biol, 2010. 17(7): p. 882-8.
13. Stampfer, S.D., et al., Structural basis of local, pH-dependent conformational changes in glycoprotein B from herpes simplex virus type 1. J Virol, 2010. 84(24): p. 12924-33.
14. Vitu, E., et al., Extensive mutagenesis of the HSV-1 gB ectodomain reveals remarkable stability of its postfusion form. J Mol Biol, 2013. 425(11): p. 2056-71.
15. Cooper, R.S. and E.E. Heldwein, Herpesvirus gB: A Finely Tuned Fusion Machine. Viruses, 2015. 7(12): p. 6552-69.
16. Cooper, R.S., et al., Structural basis for membrane anchoring and fusion regulation of the herpes simplex virus fusogen gB. Nature structural & molecular biology, 2018. 25(5): p. 416-424.
17. Roche, S., et al., Structures of vesicular stomatitis virus glycoprotein: membrane fusion revisited. Cell Mol Life Sci, 2008. 65(11): p. 1716-28.
18. Roche, S., et al., Structure of the prefusion form of the vesicular stomatitis virus glycoprotein G. Science, 2007. 315(5813): p. 843-8.
19. Atanasiu, D., et al., Dual split protein-based fusion assay reveals that mutations to herpes simplex virus (HSV) glycoprotein gB alter the kinetics of cell-cell fusion induced by HSV entry glycoproteins. J Virol, 2013. 87(21): p. 11332-45.
20. Gallagher, J.R., et al., Functional fluorescent protein insertions in herpes simplex virus gB report on gB conformation before and after execution of membrane fusion. PLoS Pathog, 2014. 10(9): p. e1004373.
21. Zeev-Ben-Mordehai, T., et al., Two distinct trimeric conformations of natively membrane-anchored full-length herpes simplex virus 1 glycoprotein B. Proc Natl Acad Sci U S A, 2016. 113(15): p. 4176-81.
22. Fontana, J., et al., The Fusion Loops of the Initial Prefusion Conformation of Herpes Simplex Virus 1 Fusion Protein Point Toward the Membrane. MBio, 2017. 8(4).
23. Si, Z., et al., Different functional states of fusion protein gB revealed on human cytomegalovirus by cryo electron tomography with Volta phase plate. PLOS Pathogens, 2018. 14(12): p. e1007452.
24. Merk, A., et al., Breaking Cryo-EM Resolution Barriers to Facilitate Drug Discovery. Cell, 2016. 165(7): p. 1698-1707.
25. Elmlund, D., S.N. Le, and H. Elmlund, High-resolution cryo-EM: the nuts and bolts. Curr Opin Struct Biol, 2017. 46: p. 1-6.