Pseudotyping of Glycoprotein D-Deficient Herpes Simplex Virus Type 1 with Vesicular Stomatitis Virus Glycoprotein G Enables Mutant Virus Attachment and Entry

Abstract

The use of herpes simplex virus (HSV) vectors for in vivo gene therapy will require the targeting of vector infection to specific cell types in certain in vivo applications. Because HSV glycoprotein D (gD) imparts a broad host range for viral infection through recognition of ubiquitous host cell receptors, vector targeting will require the manipulation of gD to provide new cell recognition specificities in a manner designed to preserve gD's essential role in virus entry. In this study, we have determined whether an entry-incompetent HSV mutant with deletions of all Us glycoproteins, including gD, can be complemented by a foreign attachment/entry protein with a different receptor-binding specificity, the vesicular stomatitis virus glycoprotein G (VSV-G). The results showed that transiently expressed VSV-G was incorporated into gD-deficient HSV envelopes and that the resulting pseudotyped virus formed plaques on gD-expressing VD60 cells, albeit at a 50-fold-reduced level compared to that of wild-type gD. This reduction may be related to differences in the entry pathways used by VSV and HSV or to the observed lower rate of incorporation of VSV-G into virus envelopes than that of gD. The rate of VSV-G incorporation was greatly improved by using recombinant molecules in which the transmembrane domain of HSV glycoprotein B or D was substituted for that of VSV-G, but these recombinant molecules failed to promote virus entry. These results show that foreign glycoproteins can be incorporated into the HSV envelope during replication and that gD can be dispensed with on the condition that a suitable attachment/entry function is provided.

Viruses have been extensively used for transferring genes into cells both in vitro and in vivo and remain the most efficient tools available for gene therapy applications. The ability of viruses to infect either specific cell types through specific receptors or a broad range of cell types through common cell surface determinants represents both advantages and limitations for the use of virus gene transfer systems. An important step forward in the development of virus-mediated gene transfer will be the development of methods to target viral infection to specific cell types, particularly in vivo, in order to achieve gene delivery to the tissue where transgene expression will be most effective in treating disease. The achievement of this goal will require modification of the virus host range by the manipulation of viral surface structures involved in cell recognition without compromising virus entry and transgene expression.

Herpes simplex virus type 1 (HSV-1) has received considerable attention as a gene transfer vector because of its large carrying capacity for foreign DNA sequences (2, 35, 53) and its ability to package vector DNA as concatemerized plasmids, referred to as amplicons (18, 19, 60). More recently, replication-incompetent genomic HSV vectors have been developed in which immediate-early (IE) functions are deleted, which eliminates both early and late gene expression and vector toxicity (36, 44, 56, 57, 69). Long-term transgene expression has been achieved in neurons (24, 31, 39), the natural host for latent viral infections, as well as in a variety of other cell types in which defective HSV vectors can persist with continued transgene expression (26; D. Wolfe, W. F. Goins, T. J. Kaplan, S. Capuano, M. Murphey-Corb, and J. C. Glorioso, submitted for publication; X. Zhang, J. P. Goff, D. S. Shields, J. Wechuck, R. J. R. Rouse, D. Wolfe, W. F. Goins, J. S. Greenberger, and J. C. Glorioso, submitted for publication). In addition, attenuated replication-competent HSV vectors have been tested in clinical applications. In these vectors, either the nonessential γ34.5 gene product, which disables the double-stranded-RNA-dependent protein kinase pathway involved in the inhibition of protein synthesis in virus-infected cells (7, 10), or the γ34.5 gene in combination with the UL39 viral ribonucleotide reductase large subunit (32, 49) was deleted to allow virus replication in dividing cells (e.g., glioblastoma cells) but not in certain postmitotic cells (e.g., brain neurons). However, the use of these attenuated viruses remains subject to safety concerns since they can infect nontumor cells, suggesting that their use may be limited to specialized applications, such as with brain tumors. Because HSV has a very broad host range, the utility of HSV vectors may be greatly increased by restricting viral infection to the cell types of interest, particularly in applications in which virus replication is required for effective gene delivery. Modification of the host range of HSV may prove to be a formidable task, considering the complexity of the virus envelope and the staged process of virus infection, requiring multiple essential viral envelope components (6, 11, 33, 42). Nevertheless, recent advances in understanding the viral envelope glycoproteins that mediate virus attachment and penetration suggest that it may be possible to target HSV infections.

HSV-1 attachment is known to be mediated by multiple glycoproteins. Binding to cell surface glycosaminoglycans, primarily heparan sulfate (27, 28, 61, 70) but also dermatan sulfate (4, 67) and chondroitin sulfate (3), is mediated by exposed domains of glycoproteins C (gC) (28, 40, 63, 65) and B (gB) (29, 40). This binding represents about 85% of the total binding to Vero cells, with gC contributing the major share (28, 40). This initial stage of virus-cell contact is not sufficient to trigger virus entry but accomplishes the task of positioning the virus for interaction with receptors recognized by glycoprotein D (gD). Recent work by several labs has identified gD cognate receptors utilized for both virus attachment and penetration. The first identified herpesvirus entry mediator, HVEM or HveA, is a member of the TNF-α/NGF receptor family (48, 50, 55, 64) and has a restricted cell type distribution. A second receptor for gD, HveC, is a member of the immunoglobulin superfamily and is widely expressed (20, 37, 38), providing a basis for the broad host range of HSV-1.

The consequence of the sequential attachment steps involving gB, gC, and gD is the fusion of the virus envelope with the cell surface membrane and subsequent virus entry. The events in penetration are not well understood, but multiple glycoproteins are required. A role for gD in virus penetration is supported by evidence that attached virus can be neutralized by anti-gD antibody and that virus mutants with a deletion of gD attach to cells but do not penetrate (17, 30). Mutants with a deletion of gH/gL or gB are also blocked in virus penetration but are not defective in attachment. Both gB and gD have been shown to be capable of inducing syncytia if expressed on the cell surface at low pH, supporting a possible role for both molecules in fusion (5). According to one model, gD, gB, and gH/gL act in succession to mediate fusion, entry, and virus release into the cytoplasm (16). The ability of gD to bind to receptors on a broad range of cell types suggests that the partial or complete replacement of gD with other sequences capable of mediating viral entry may provide a means for HSV vector targeting.

The present study was designed to evaluate the G spike glycoprotein of vesicular stomatitis virus (VSV-G) as a potential alternate to HSV gD in binding and entry. VSV is a member of the rhabdovirus family of RNA viruses. The VSV-G spike glycoprotein is involved in both virus attachment and entry into receptor-bearing host cells (15, 41, 58). VSV-G enables receptor-mediated endocytosis of VSV (47) and induces fusion of the viral envelope and the endosomal membrane (13, 14, 54), which results in the entry of VSV virions into the host cell cytoplasm. Although endocytosis involving VSV-G exposes the virus to a mildly acidic environment, several investigators have described VSV-G-mediated infection by other viruses pseudotyped with VSV-G, including retroviruses (1, 43, 52), measles virus (62), and adenoviruses (59).

Experiments were undertaken to determine (i) whether wild-type VSV-G or chimeric VSV-G/HSV glycoprotein derivatives could be incorporated into the envelope of gD-deficient HSV during virus replication and (ii) whether VSV-G or the chimeric molecules could provide an alternative means of HSV attachment and entry in lieu of gD. The results showed that VSV-G and derivatives containing the transmembrane domain (TM) of HSV gB or gD were incorporated into virus envelopes at high levels, although only native VSV-G had the ability to mediate infection by a gD-deficient virus. These studies demonstrate that VSV-G can substitute for gD in promoting virus attachment and entry.

Construction of plasmids and viruses.

A 10.1-kb KpnI fragment (HSV-1 positions 134789 to 144894) from the Us region of the EcoRI H plasmid (25) was cloned into the unique KpnI site of a pSP72 derivative lacking BamHI sites to produce pUs. A 4.2-kb BamHI fragment, containing a human cytomegalovirus IE promoter-lacZ expression cassette from pIEP-lacZ (2), was substituted for the 6.5-kb BamHI J fragment of pUs to create pUsΔ3-8Z, which contains sufficient flanking sequences for recombination into wild-type virus. A gD-null virus, KΔUs3-8Z (Fig. 1A), was generated by homologous recombination of wild-type KOS virus with pUsΔ3-8Z. lacZ-expressing recombinants were identified by staining with X-Gal [0.1% 5-bromo-4-chloro-3-indoyl-β-d-galactopyranoside (Roche, Indianapolis, Ind.) in a solution containing 100 mM Tris-HCl (pH 8.0), 13.5% dimethyl formamide (Sigma, St. Louis, Mo.), 14 mM K₄Fe(CN)₆, and 14 mM K₃Fe(CN)₆], and a selected blue plaque was purified in three rounds of limiting dilution. The resulting gD-null virus could be propagated on gD-complementing VD60 cells (42) but was unable to form plaques on Vero cells. The genotype of this virus was null for Us3 (protein kinase), Us4 (gG), Us5 (gJ), Us6 (gD), Us7 (gI), and Us8 (gE), as confirmed by the absence of the 6.5-kb BamHI J fragment in a Southern blot analysis (Fig. 1A). The extensive deletion of Us sequences precludes recombination between the complementing gD gene of VD60 cells and the KΔUs3-8Z viral genome. Southern blot analysis using a lacZ-specific probe also verified the presence of the human cytomegalovirus IEp-lacZ expression cassette in KΔUs3-8Z (Fig. 1A).

FIG. 1.

Construction of the KΔUs3-8Z recombinant virus and structure of the VSV-G/HSV chimeric proteins. (A) Schematic of the HSV-1 genome depicting the location of the essential and nonessential glycoprotein genes and replacement of the Us3-Us8 BamHI J fragment of strain KOS by a lacZ expression cassette to generate the gD-null virus KΔUs3-8Z. The accompanying Southern blots demonstrate the presence in KΔUs3-8Z of the 4.2-kb lacZ cassette (LacZ probe) and the absence of the 6.5-kb BamHI J fragment (BamHI J probe). (B) Wild-type and chimeric proteins (kindly provided by Hara P. Ghosh, McMaster University) represented as boxes corresponding to the EC (left), TM (center), and CT (right), with the relevant amino acid numbers provided in each case. VSV-G sequences are shown as open boxes, HSV-1 gB sequences are shown as black boxes, and HSV-1 gD sequences are shown as grey boxes. A deletion in the gB TM creating the gB₃ derivative is indicated.

A series of wild-type and chimeric VSV-G plasmids (Fig. 1B), constructed as previously reported (21, 51), was kindly provided by Hara P. Ghosh (McMaster University, Hamilton, Ontario, Canada). The VSV-G and chimeric genes in these plasmids were under the control of the adenovirus major late promoter. The nomenclature of these plasmids indicates the origin of the ectodomain (EC), TM, and cytoplasmic domain (CT) sequences. For example, the construct G/gD/G contained the EC of VSV-G, the TM of gD, and the CT of VSV-G.

Incorporation into budding viruses.

Previous studies examining chimeric glycoproteins (22) identified sequences within the TM of gB that are involved in localization of gB to the nuclear envelope, a potential point of membrane acquisition for budding capsids (21). Accordingly, we tested whether these sequences or potentially equivalent sequences in gD (51) could anchor the EC of VSV-G in the envelope of KΔUs3-8Z. 293T cells, in which the adenovirus major late promoter is active, were plated on 60-mm dishes and transfected with the wild-type or chimeric VSV-G, gD, or gB plasmids using Lipofectamine (Life Technologies, Inc., Gaithersburg, Md.). Twenty-four hours later, the cells were infected with KΔUs3-8Z at a multiplicity of infection of 3 for 1.5 h. Two hours postinfection (p.i.), the cells were treated for 1 min with 0.1 M glycine, pH 3.0, to inactivate nonpenetrating virus, and were incubated for 48 h at 37°C in Dulbecco modified Eagle medium–10% fetal bovine serum (Life Technologies, Inc.) containing [³⁵S]methionine-cysteine (NEN-DuPont, Boston, Mass.). Cell lysates (Fig. 2A) were prepared in lysis buffer containing 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% Triton X-100 (Roche), and 1 mM Nα-p-tosyl-l-lysine chloromethyl ketone (TLCK) (Roche) and were immunoprecipitated overnight at 4°C with antibodies against HSV-1 gB (lanes B), gC (lanes C), gD (lanes D), or VSV (lanes G) (30, 45, 46) (VSV polyclonal antibody kindly provided by Patricia Whitaker-Dowling, University of Pittsburgh). Samples were incubated for 1 h with protein A-Sepharose (Sigma) and centrifuged at 500 × g, and the immune complexes were washed with 600 μl of lysis buffer and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). In addition, media from the transfected, infected cultures were collected and cleared of cell debris and cell-associated virus by centrifugation. Virus from the cell supernatant was added to lysis buffer, immunoprecipitated, and subjected to SDS-PAGE analysis.

FIG. 2.

Requirements for incorporation into mature HSV-1 particles. 293T cells were transfected with the plasmids indicated above each panel and were infected with the gD-null virus KΔUs3-8Z. Metabolically labeled proteins in cell lysates and extracellular virus immunoprecipitated with antibodies (Ab) against gB (lanes B), gC (lanes C), gD (lanes D), or VSV (lanes G) were separated by SDS-PAGE and visualized by autoradiography. ∗ and #, wild-type VSV-G and VSV-G/HSV chimeric proteins, respectively.

As shown in Fig. 2A (panels 2 and 5 to 9), the wild-type and chimeric VSV-G glycoproteins were expressed at high levels by using immunoprecipitation with anti-VSV-G antibody (lanes G). However, only VSV-G, G/gD/G, and G/gB₃/G were incorporated into mature HSV-1 particles in significant amounts (Fig. 2B, lanes G, panels 2, 6, and 9). To a lesser extent, G/G/gB and G/gB/G were detected in released virions (panels 5 and 8). Potential background from the shedding of free radiolabeled protein into the media was a concern, but the observation of high levels of G/gD/gD in lysates, but not in released virions (panel 7), argued that the glycoproteins detected in the virus preparations were not simply the result of free molecules fortuitously associating with the virus in the media. These data are in agreement with the earlier report that the TM of gB, specifically the membrane-anchoring domain (gB₃), is sufficient for nuclear envelope localization (22). Although similar nuclear membrane localization studies have not been reported for the VSV-G/gD fusion proteins, it has been demonstrated that the TM of gD does not interfere with the fusion function of VSV-G (51). Our data now show that the gD TM also enabled efficient incorporation into virions, provided that it was flanked by the EC and CT of VSV-G (compare Fig. 2, panels 6 and 7). At the same time, HSV TM sequences were not essential since unmodified VSV-G was also incorporated (panel 2), although at a lower level. Because VSV-G is not known to be trafficked to the nuclear membrane, where HSV budding occurs, its incorporation into virus was surprising. Indirect immunofluorescence of unfixed cells transfected with the VSV-G construct or with VSV-G/HSV chimeras such as G/gD/G or G/gB₃/G demonstrated the presence of these glycoproteins at the cell surface (data not shown), suggesting that the virus may acquire VSV-G in a membrane exchange event, most likely in the Golgi apparatus. The significant levels of incorporation of wild-type VSV-G suggest that other foreign viral glycoproteins may also be correctly sorted and incorporated into mature HSV virions, which would create additional opportunities for pseudotyping HSV-1 particles with non-HSV proteins.

Complementation of gD function in HSV-1 entry.

Using pseudotyped viruses produced by the transient-expression protocol, we determined whether VSV-G or any of the chimeric proteins could complement the functions of gD in virus entry. 293T cells were transfected with the panel of glycoprotein constructs, infected with KΔUs3-8Z (multiplicity of infection = 3.0), treated with glycine, and incubated in complete media for 48 h. Media were collected and serial dilutions were incubated with 5 × 10⁵ VD60 cells in suspension for 1.5 h before the cells were plated in 35-mm dishes. Twelve hours p.i., the VD60 cells were overlaid with complete media containing 0.5% methyl cellulose (Aldrich, Milwaukee, Wis.), and at 72 h p.i., plaques were stained with X-Gal and counted. Virions capable of mediating virus entry in the absence of gD would produce plaques on VD60 cells, which constitutively express gD (42) and thereby provide the functions of gD required for the progression from the initial entry to plaque formation. Three experiments were performed in duplicate, and the compiled results are shown in Fig. 3A. Medium from mock-transfected cells yielded a background of less than 10 PFU, presumably representing gD-complemented KΔUs3-8Z virus remaining from the initial 293T-cell infection despite the subsequent acidic glycine wash. The wild-type gD plasmid produced a robust level of complementation (5 × 10³ PFU), but of the other constructs tested, only the VSV-G plasmid displayed complementing activity above background (∼100 PFU). While the complementation level of VSV-G was 50-fold lower than that of gD, the ability of this foreign glycoprotein to reproducibly substitute for gD in entry represents the first demonstration of an infectious, pseudotyped gD-null HSV-1 particle.

FIG. 3.

Complementation of gD functions in HSV-1 entry and neutralization of complementing virus. (A) Complementation measured as PFU on VD60 cells from the media of transfected, infected 293T cells. (B) Neutralization assay results for VD60 cells from the media of transfected, infected 293T cells. Results are expressed as percent reduction in the number of plaques (e.g., “100%” indicates complete neutralization).

Neutralization of complementing virus.

To validate these results, VD60 infectivity assays were repeated after incubation of the 293T media with complement-dependent neutralizing antibodies against HSV-1 glycoproteins (gB, gC, or gD) or VSV-G. Serial dilutions of virus-containing media were incubated with 5 μl of human complement (Life Technologies, Inc.) in the presence or absence of each antibody (2 μl) for 2 h at 37°C. VD60 cells (5 × 10⁵) were added to each sample and incubated in suspension for 1.5 h at 37°C with shaking. Infected cells were plated in 35-mm dishes, treated at 2 h p.i. with 0.1 M glycine as before, and overlaid with complete media containing methyl cellulose. Seventy-two hours p.i., plaques were stained with X-Gal and counted. The results are tabulated in Fig. 3B.

As shown in Fig. 3B, monoclonal antibodies against gB were able to neutralize all samples; likewise, neutralization of all samples was observed with gC-specific antibodies (data not shown), as was expected since KΔUs3-8Z is wild type for both gB and gC. Antibodies against gD completely neutralized wild-type-gD-complemented as well as mock-complemented virus, supporting the earlier suggestion that the background in this assay was due to gD-complemented KΔUs3-8Z remaining from the initial 293T-cell infection. The gD antibodies were also able to neutralize 18% of the activity in the VSV-G-complemented preparation, which is somewhat more than the expected neutralization of the background (4%) and may be due to nonspecific interactions. Likewise, anti-VSV-G polyclonal antibodies were able to neutralize 91% of the complementation by VSV-G but also displayed some neutralization of gD-complemented material (13%). These results validated the conclusion that VSV-G has a measurable, albeit limited, ability to complement the entry functions of HSV gD in a transient assay.

HSV shows great promise as a generally useful vector for the in vivo delivery of genes to a variety of cells, including bone marrow stem cells and cells of the nervous system, muscle, connective tissue, liver, and endothelium (2, 23, 26, 66, 68, 71; Wolfe et al., submitted). While the virus naturally persists in a latent state in sensory neurons, highly defective mutant viruses can also persist in other cell types (26; Wolfe et al., submitted; Zhang et al., submitted). HSV vectors can be targeted to specific sites in the body by direct injection, but there are circumstances in which safe and effective gene therapy will require cell-type-specific virus infection. Because natural HSV receptors that mediate virus entry are found on most cell types (e.g., HveC), it will be important to mutate the virus to prevent its attachment to these natural receptors and to equip the particle instead with ligands for cell-type-specific receptors.

The present study was initiated to determine whether gD is unique in its ability to mediate HSV entry or whether functionally related proteins from other viruses can substitute for gD, provided they are incorporated into the HSV envelope. We chose to test VSV-G because this protein has been used with considerable success for pseudotyping other viruses (1, 43, 52, 59, 62) and mediates both attachment and penetration of VSV (15, 41, 58). Nevertheless, VSV-G was not ideal, since the sites of HSV and VSV budding differ, raising the concern that VSV-G might not be incorporated into HSV envelopes. Moreover, since VSV entry involves fusion with endosomal vesicle membranes, which exposes the virus to an acidic environment, while HSV fuses with the cell surface membrane, VSV-G-directed entry could result in the destruction of the HSV particles in the endosomes. Remarkably, we found that VSV-G was included in the HSV envelope, although less abundantly than wild-type gD. Anticipating inefficient incorporation, we simultaneously tested chimeric constructs utilizing gB or gD sequences to direct the VSV-G EC to the HSV envelope. The results showed that the incorporation of VSV-G into virus was greatly improved when the TM of gB or gD was substituted for that of VSV-G, but these chimeric molecules were unable to direct HSV entry. The incorporation of unmodified VSV-G into virus may be accomplished by a membrane exchange in the Golgi apparatus or with the plasma membrane during virus egress (8, 9, 34), raising the possibility that expression during HSV infection may be sufficient for the inclusion of foreign glycoproteins in mature virus envelopes. Accordingly, other viral proteins may also be correctly sorted and incorporated into mature HSV virions, which would offer a range of opportunities for pseudotyping HSV-1 particles with non-HSV viral proteins. This suggestion is supported by the previous demonstration that cellular CD4 can be incorporated into HSV virions, albeit at low levels (12).

Although the amount of VSV-G in pseudotyped HSV envelopes was relatively low, VSV-G was able to mediate sufficient virus entry to allow plaque formation on VD60 cells. VD60 cells provide gD intracellularly for incorporation into genotypically gD-deficient virus, in this case replacing VSV-G, which was not produced by either the virus or the host cells. However, these events require prior entry of the virus into single cells, and the VD60 plaque assay therefore measured the success of this initial infection with pseudotyped gD-null virus, using intracellular gD to convert single infected cells into plaques. The assay showed a reproducible increase in plaque formation by the VSV-G-pseudotyped virus over the background level observed with mock-complemented virus and showed a return to background levels when the pseudotyped virus was preincubated with VSV-G- but not gD-neutralizing antibodies. The gC-specific antibodies neutralized all plaque-forming activity, including the background, while gD antibodies removed only a fraction, consistent with the interpretation that the background represented residual gD-complemented virus from the 293T-cell infection preceding the assembly of pseudotyped virus. These results, representing the first demonstration of an infectious gD-null HSV-1 particle, have now been confirmed using a VSV-G/KΔUs3-8Z recombinant virus (W. F. Goins et al., unpublished data). Preliminary experiments indicate that this recombinant virus enters by fusion with endosomal vesicles, like VSV but unlike wild-type HSV. Given this altered entry pathway, contributions from any HSV glycoprotein normally functioning in HSV infection may be dispensable in the presence of VSV-G, suggesting possibilities to examine regarding the roles of these glycoproteins, including gD, in postentry events such as budding and axonal transport.

The suggestion from our work that the entry pathway of HSV can be fundamentally altered by replacing gD with a foreign viral glycoprotein indicates that there is considerable flexibility in the mechanism of HSV entry, which in turn suggests that other foreign glycoproteins could be equally or more effective substitutes for gD. Among these, some may restrict the viral host range or represent attractive targets for the engineering of new host cell specificities. Our work suggests the intriguing possibility that gD-null HSV can be pseudotyped with separate molecules providing the cell recognition and fusion functions, reminiscent of the natural situation among paramyxoviruses, which could greatly facilitate the creation of targeting vectors.