Vector Targeting

One method to achieve cell-specific expression of the therapeutic gene is targeting vector recognition and infection of cells to unique receptors present principally on target cells. Exploration of this approach has been limited thus far, but several recent reports have shown convincing targeting of adenovirus (77-79) or retrovirus (80-83). Targeted viral infection requires (1) the identification of cell-specific surface re-ceptor(s) to which viral binding/entry can be directed, and (2) the modification of viral glycoproteins to recognize novel receptors while eliminating the binding of these viral ligands to native receptors, a process that ideally should be accomplished without compromising infectivity. These steps can be combined if it is possible to replace the natural receptor-binding domain of viral glycoproteins responsible for infection with binding domains specific for alternate receptors.

The complexity of the glycoproteins in the viral membrane and the fact that multiple glycoproteins are required for the sequential steps of virus attachment and entry has made redirecting HSV infection difficult. Early electron microscopic studies showed that the trans-plasmalemmal entry of HSV into cells occurs in two morphologically definable stages: cell surface attachment and virion-cell fusion. It is convenient to consider these stages separately, although in reality the transition from one to the other is rapid and seamless.

Initial virus attachment is cooperatively mediated by numerous glycoproteins (5,6,8). Binding of viral particles to cell surface heparan sulfate (HS) and other glycosaminoglycans (GAGs) (10,84-90) is mediated by exposed domains of glyco-proteins C (91-94) and B (86,92), as evidenced by the diminished cell attachment of gC and gB mutants, and the interference with cell attachment by (1) pretreatment with anti-gB or anti-gC antibodies, (2) enzymatic removal of cell surface GAGs, and (3) competition studies using heparin. The role of gB is complex as this glycoprotein functions both in the initial binding of virus to the cell surface (nonessential for cell entry) and in later stages in the virion-cell fusion program (essential for cell entry). Molecular dissection of the domains of gB responsible for each distinct process has allowed the generation of a gB mutant that shows isolated loss of GAG-binding functions, dissociated from preserved cell entry functions. This mutant, generated by deletion of a positively charged lysine-rich region within gB (designated gB:pK~), is an important resource in the generation of viral vectors with targeted cell binding and preserved entry (see below). The initial gB/ gC-HS/GAG-mediated cell attachment greatly enhances but is not essential for subsequent events in the cell entry cascade; thus, cell attachment represents a reasonable target for strategies aimed at effecting restricted cell entry.

In initial studies to manipulate the cell attachment properties of HSV, we focused on the elimination of HS binding by removal of appropriate binding domains from the virus envelope, leaving intact the viral determinants mediating the downstream events in cell entry. We hypothesized that subsequent replacement of the HS-binding ligands with avid spe

Figure 5 Strategies for HSV-1 vector design. (A) The production of defective full-length genomic HSV vectors is carried out in cell lines engineered to provide the deleted essential genes in trans. These vectors can be produced in high titers, are capable of long-term persistence in neurons in vivo, can accommodate large or multiple transgenes, and are incapable of replicating in neurons or other cells because of the missing essential genes. (B) Helper-free amplicons can be readily propagated in bacteria using the bacterial origin of replication (E. coli ori), and then transfected into a noncomplementing cell line along with 5 cosmids that encompass the entire HSV genome. Unlike the standard amplicon system in which the final preparation consists either of a mixture of amplicon concatemers and defective HSV particles, only amplicon concatemers get packaged into new virus particles because the overlapping cosmids lack the HSV packaging sequence ("a" sequence). The helper-free amplicon preparations suffer from low titer yields, decreased stability of the amplicon DNA, and decreased transgene payload.

Figure 5 Strategies for HSV-1 vector design. (A) The production of defective full-length genomic HSV vectors is carried out in cell lines engineered to provide the deleted essential genes in trans. These vectors can be produced in high titers, are capable of long-term persistence in neurons in vivo, can accommodate large or multiple transgenes, and are incapable of replicating in neurons or other cells because of the missing essential genes. (B) Helper-free amplicons can be readily propagated in bacteria using the bacterial origin of replication (E. coli ori), and then transfected into a noncomplementing cell line along with 5 cosmids that encompass the entire HSV genome. Unlike the standard amplicon system in which the final preparation consists either of a mixture of amplicon concatemers and defective HSV particles, only amplicon concatemers get packaged into new virus particles because the overlapping cosmids lack the HSV packaging sequence ("a" sequence). The helper-free amplicon preparations suffer from low titer yields, decreased stability of the amplicon DNA, and decreased transgene payload.

cific receptor-binding sequences might achieve at least partial targeting. A double-mutant virus, KgBpK-gC-, was derived from wild-type KOS strain (Figs. 6A and B). In this mutant, the coding sequence for the nonessential gC gene was removed from the viral genome and the wild-type gB sequence replaced by the gB:pK~ mutation (92) (Fig. 6B). The resulting virus demonstrated an 80% reduction in binding to Vero cells compared with wild-type virus (Fig. 6D). By replacing the HS-binding domain of gC with the coding sequence of erythropoietin (EPO) in the background of the KgBpK~gC " mutant virus (Fig. 6C), we demonstrated that the gC:EPO fusion protein was incorporated into the budding virion and that recombinant virus was specifically retained on a soluble EPO-receptor column (95). The gC:EPO virus demonstrated a 2fold increase in infection of K562 cells (Fig. 6D) (Laquerre and Glorioso, unpublished data, 1998), which express the EPO receptor. The EPO-expressing particle also stimulated the proliferation of FD-EPO cells (95), an EPO-dependent cell line, indicating that virus binding to specific cellular receptor had occurred. However, virus attachment was followed by delivery of the viral particle to the endosome compartment resulting in virus degradation, thus preventing the normal pathway of entry resulting in productive infection. These studies stress the importance of appropriate receptor interactions to ensure virus entry by the normal route of virus penetration.

An alternative strategy for effecting targeted cell entry through engineering cell attachment was subsequently considered. In the gC-EPO experiments described above, entry of HSV into the endosome prevented cytoplasmic penetration, probably because of the low pH of the endosome compartment; trans-plasmalemmal cellular entry of HSV occurs at physiological pH, and it is unlikely that the native viral entry determinants would be functional in the acidic environment of the endosome. Several viruses, however, enter cells through a trans-endosomal route, exploiting the low pH of the endosome to effect fusion-triggering alterations in viral receptors. An example is provided by vesicular stomatitis virus (VSV). The VSV-G spike glycoprotein had previously been successfully used to redirect the tropism of lentiviruses, expanding their host cell range and reconfiguring the usual trans-plas-malemmal route of entry of lentiviruses into cells, to enable a trans-endosomal route to be taken. We thus explored pseudotyping HSV using VSV/HSV fusion proteins. A mutant virus was used in which the short unique segment of the HSV genome US3-8 (encoding gD in addition to nonessential glycoproteins gE, gG, gJ, and gl) was deleted. The gD-null phe-notype of this vector was transiently rescued by using a series of plasmid expression cassettes encoding chimeric VSV-G/ HSV glycoprotein fusion proteins (96). Chimeras containing the ectodomain of VSV-G linked to either the C-terminal or

Figure 6 Targeted binding of HSV-1 particles expressing gC: EPO fusion molecules to EPO receptor (EPO-R)-bearing cells. (A) Diagram of the KOS wild-type HSV-1 genome depicting the location of the two HSV glycoproteins (gB and gC) involved in heperan sulfate-binding. The heperan sulfate binding domain of gB consisting of a series of polylysine (pK) residues is shown in greater detail. (B) The KgBpK~gC~ recombinant virus deleted for binding to heperan sulfate was constructed by deleting the pK region from the essential gB gene and deletion of the nones-sential gC gene by insertion of an HCMV IEp-lacZ expression cassette into the gC locus. (C) The gC:EPO2 recombinant (KgBpK ~gCEPO2) was constructed by introducing EPO into the gC gene, replacing aa#1-162 in the KgBpK~gC~ recombinant virus, which could readily be purified by X-gal staining. (D) The percentage of radiolabeled wild-type HSV-1 (KOS), the recombinant deleted for heperan sulfate binding (KgBpK ~gC~), and the EPO-expressing (KgBpK ~gCEPO2) viruses that bound to K566 cells bearing the EPO-R was determined and is expressed compared with input virus. These data demonstrate that the KgBpK " gCEPO2 recombinant virus binds to the EPO-R and this binding conferred increased infectivity of the KgBpK ~gCEPO2 virus for cells bearing the EPO-R (K566).

transmembrane domain of either gD or gB were generated. A number of different constructs were tested to facilitate the identification of chimaeras that were effectively packaged into the virion, because it was previously suggested that the structural requirements for incorporation of glycoproteins into the envelope of the mature virus particle were stringent with respect to transmembrane or endodomains (97-100). It was demonstrated that VSV-G chimeras containing the transmembrane domain of gD, or a truncated gB transmembrane domain, were incorporated into the viral envelope efficiently, and that the wild-type VSV-G protein was incorporated rather less efficiently (96). The latter, however, was able to partially rescue the gD-deficient phenotype, whereas the chimeric proteins were nonfunctional. Neutralizing anti-VSV-G antibodies blocked the partial VSV-G mediated rescue (96). The poor efficiency of phenotypic rescue of the US3-8 null virus by VSV-G may be attributable to either (1) inefficient incorporation of the foreign viral glycoprotein, or (2) acid degradation of HSV in the endosome compartment. We have studied this issue further in order to elucidate and exploit the mechanisms contributing to partial rescue. A recombinant virus was generated in which the VSV-G expression cassette was incorporated into the genome of the US3-8 deleted virus (the resulting vector is null for gD, gE, gG, gJ, and gI, but expresses VSV-G). The recombinant particle enters cells possessing the VSV receptor. However, an abortive infection ensues, culminating in endosomal degradation of the virion at low pH, similar to that observed when the gC-EPO-expressing recombinant enters cells (Goins and Glorioso, unpublished data, 2003). The use of lysomotropic agents that raise endosomal pH, such as chloroquine (101), enables release of viral contents from the endosome into the cytoplasm, resulting in plaque formation. An acidic environment is required for the activation of VSV-G fusion functions (102,103). It follows that, in the presence of lysomotropic drugs, fusion of the VSV-G expressing, US3-8 null, recombinant with the endosomal membrane must be mediated by HSV fusion glycoproteins (gB or gH/gL). It thus appears likely that, in this situation, VSV-G functions as a ligand for an internalized receptor, rather than a pH-dependent mediator of viral envelope fusion with the endosome membrane. We are currently examining the mechanism in more detail because this strategy might be exploited to allow specific targeting of HSV to internalized cellular receptors.

Viral gC/gB-mediated binding to cell surface HS is followed by a second binding event between viral gD and specific cellular receptors. An essential role for gD in virus penetration is supported by the finding that recombinant virus deleted for gD is capable of binding to cells but is unable to penetrate (9,11). In addition, infection can be neutralized by anti-gD antibodies that do not block attachment of virus to the cell (9,11). Several distinct cellular gD receptors have been identified. The first HSV gD receptor herpesvirus entry mediator (HVEM or HveA), isolated by screening a cDNA expression library in HSV refractory Chinese hamster ovary (CHO) cells, was determined to be a member of the tumor necrosis factor (TNF)/nerve growth factor (NGF) receptor family (104). Subsequently, a number of other entry mediators have been identi fied, including HveB [poliovirus receptor-related protein 2 (Prr2)] (105), and HveC [nectin-1, poliovirus receptor-related protein 1 (Prr-1)] (106). The latter, an alternatively spliced member of the immunoglobulin superfamily that bears no structural relation to HVEM/HveA is the major gD receptor and is widely expressed. Finding multiple, unrelated receptors capable of mediating HSV infection via gD suggests that several distinct receptor-binding domains exist within gD. The recent publication of the crystal structure of the HveA-gD receptor-ligand complex is an essential first step in the recognition of gD domains that are essential for interaction with specific cellular ligands. This knowledge may assume great importance in the rational design of gD mutants that are restricted for binding/entry functions through prespecified cellular HSV receptors, and ultimately for engineering gD to target new ligands. Targeted infection might be accomplished by manipulation of the specificity of gD while leaving the HS-binding activity of gC and gB intact. The ability of gD to bind to a range of cellular receptors suggests that the partial or complete substitution of gD with other sequences capable of mediating viral entry may provide a means of HSV vector targeting. Mutation of, or antibody binding to, N-terminal amino acids of gD eliminates HveA binding while leaving HveC binding and the penetration function intact, but the reverse has yet to be accomplished (107). Initial mutagenesis studies suggest that the binding and entry functions of gD may be dissociable and dependent on overlapping but distinct subsets of amino acid residues within the HveA/HVEM- and HveC/nectin-1/Prr-1-binding domains of gD (Bai and Glorioso, unpublished data, 2002), suggesting that targeted entry may be at least theoretically possible. A subsequent study has shown limited evidence that cell entry might occur through engineered gD (108). A recombinant virus was generated, in which the binding sites for HS in gB and gC were deleted. The amino-terminal of gC was replaced by IL-13 and a second copy of IL-13 was inserted into gD, disrupting the binding site for HveA/HVEM but not the binding site for HveC/nectin/ Prr-1. The recombinant and wild-type viruses replicated in a variety of cell lines that expressed HveC/nectin/Prr-1. The recombinant failed to replicate in a cell line that does not express HveC/nectin/Prr-1, but did replicate in a derived clone that expressed the IL-13 receptor. Although mutants that are unable to enter cells through HveA/HVEM, but which have preserved HveC/nectin/Prr-1-mediated entry functions, have been previously described, this study provides preliminary evidence that gD engineering might allow viral entry through generation of a novel binding specificity. However, the crucial test of this strategy for viral targeting will be whether viral entry functions of gD remain preserved after disruption of the HveC/nectin/Prr-1-binding site, an observation that was notably absent from this report. Ongoing studies will determine whether engineering a novel binding ligand into the gD N-terminus will substitute for gD binding to HveA and HveC and preserve the role of gD in penetration.

An alternative strategy might use a bispecific soluble adapter molecule to bind to the HveC/nectin/Prr-1 site of gD, inducing the viral entry cascade. The molecule would also contain a binding site for a specific cellular ligand, thus providing the following 3 properties essential for targeted entry: (1) targeted cellular binding, (2) promotion of viral fusion, and (3) blocking the gD HveC/nectin/Prr-1-binding domain that would enable entry to nontargeted cells expressing this receptor. We have shown proof of principle that viral entry might be triggered by a soluble HveC/nectin/Prr-1 fragment containing the gD-binding domain (Bai and Glorioso, unpublished data, 2002) in the absence of a specific interaction between the soluble adapter and the cell surface.

Binding of gD to its cognate cellular receptor is followed by gB/gH/gL-mediated virion envelope-plasmalemmal fusion. As might be predicted, mutants deleted for gH/gL or gB are defective for virus penetration but are not defective in attachment (1,2,109). Furthermore, antibodies directed against gB, gH, or gL are able to prevent entry without interfering with attachment. The mechanisms of viral envelope-plas-malemmal fusion remain uncertain. In the absence of a specific role ascribed to a cellular receptor, it seems unlikely at present whether this part of the cell entry cascade will be amenable to manipulation resulting in targeted cell entry.

Was this article helpful?

0 0

Post a comment