Multiple Immediate-Early Gene-Deficient Herpes Simplex Virus Vectors Allowing Efficient Gene Delivery to Neurons in Culture and Widespread Gene Delivery to the Central Nervous System In Vivo

Cell culture and virus propagation.

BHK and Vero cells were maintained by standard cell culture procedures. All viruses used in this study were propagated on BHK-derived complementing cell lines that have been described previously (B130/2 cells for virus strains 17+ 27− and 1764 27− [26] and 27/12/M:4 cells for virus strain 1764 27− 4− [62]). These cell lines only complement the genes that have been deleted, and there is no sequence overlap between the viral and cellular genomes. Complementing cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and 5% tryptose phosphate broth. Continual selection was achieved by maintaining the cells in 800 μg of G418 per ml (cell line B130/2) or 800 μg of G418 per ml and 750 μg of Zeocin per ml (cell line 27/12/M:4). HMBA (3 mM) was included in the growth media of viruses containing the in1814 mutation (1).

Promoter cassettes.

The pR20.5 and pR19 promoter cassettes have been described previously (45). The pR20.5 cassette (45) was inserted into a plasmid, allowing insertional inactivation of UL41, the gene encoding Vhs (insertion at the unique NruI site at nucleotide 91854 in UL41). The pR19lacZ and pR19GFP cassettes used here (Fig. 1) were recombined into the endogenous LAT regions between the two BstXI sites (nt 120220 and 120408). HSV-1 nucleotide numbers refer to GenBank file HE1CG. The structures of each of these shuttle plasmids are shown in Fig. 1.

Virus nomenclature.

All viruses were derived from HSV-1 strain 17syn+ (7). The term “1764” describes a virus with the in1814 mutation in the gene encoding VP16 (1) and with the genes encoding ICP34.5 and ORF P completely deleted (between nt 124945 and 125723) (12). The term 27− refers to the deletion of nucleotides 113273 to 116869, which contain the genes UL54, -55, and -56 (26). UL54 is the gene encoding the essential IE protein ICP27, and UL55 and -56 are both nonessential genes. Virus strains 1764 27− and 1764 27− 4− were also deleted for the endogenous LAT P2 regions (between nt 118768 and 120470, DdeI to HpaI) in order to prevent the recombinational instability that was otherwise found to occur after insertion of LAT P2-containing expression cassettes outside of the LAT region (45).

Viruses.

The shuttle plasmids described above were recombined into either virus strain 17+ 27− (26) or virus strain 1764 27− (12) by standard calcium phosphate techniques (60). Recombinant viruses were identified by reporter gene transfer and plaque purified. Genome structures were confirmed by Southern blotting.

Western blotting.

Samples for Western blot analysis were prepared by standard techniques. Extract from approximately 10⁵ cells (per lane) was loaded onto a sodium dodecyl sulfate (SDS)-polyacrylamide gel. Proteins were transferred to nitrocellulose and probed with appropriate antibodies by standard techniques. Antibody against HSV-1 ICP0 was purchased from ABI. Antibodies to HSV-1 ICP22, ICP6, and ICP47 were provided by Bernard Roizman (University of Chicago), Barklie Clements (MRC Institute of Virology, Glasgow, Scotland), and David Johnson (Oregon Health Sciences University, Portland), respectively. Detection was performed with the ECL enhanced chemiluminescence system (Amersham).

Primary neuronal cultures.

Organotypic hippocampal slice cultures were prepared as previously described (59) and maintained in a mixture of 50% MEM, 25 mM HEPES, 25% horse serum, and 25% Hanks' balanced salt solution supplemented with 2 mM l-glutamine and 6.4 mg of d-glucose per ml. Slices were maintained on porous rafts (Millicell; Millipore) over 1 ml of growth medium, which was changed every second day. Four organotypic slices were cultured on each raft. Slices were infected 1 week postplating by covering each raft in 2 ml of 10⁶ PFU of virus stock per ml diluted in growth medium supplemented with 2% TCM Supplement (ICN Biomedicals) for 1 h. Dorsal root ganglion (DRG) neurons from an adult rat were prepared as described previously (13). Neurons were plated on laminin-coated coverslips and maintained in DMEM supplemented with 10% horse serum, half of which was changed every third day. Neurons were infected by incubation for 1 h with an MOI of 10 of the appropriate virus diluted in serum-free DMEM.

In vivo gene delivery.

Adult (220-g) male Lewis rats were inoculated by stereotaxic injection of vector suspensions into the striatum, superior colliculus, or spinal cord (at the C6 level). At various times postinoculation, animals were perfusion fixed in ice-cold 4% paraformaldehyde and sectioned as described in the figure legends. Sections including the injection site and connected sites within the nervous system were stained with 1× phosphate-buffered saline (PBS) containing 5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆O · 6H₂O, 1 mM MgCl₂, and 150 μg of 4-chloro-5-bromo-3-indolyl-β-galactosidase (X-Gal) per ml in order to visualize lacZ expression. Some sections were counterstained in 0.02% neutral red.

Virus strain 1764 27− 4− does not express significant amounts of any IE gene.

In work aimed at generating a nontoxic HSV backbone, vectors were constructed by the sequential deletion or inactivation of ICP27, VP16, ICP34.5, ICP4, and Vhs. Here we aimed to combine IE gene deletions (ICP27 and/or ICP4) with VP16 inactivation such that the remaining IE genes, which had not been specifically deleted, would not be expressed in target cells. To test the levels of residual IE gene expression, noncomplementing BHK cells were infected at MOIs of 10, 5, and 1 with 17+ 27−, 1764 27−, and 1764 27− 4− viruses. Complementing 27/12/M:4 cells (62) were infected with the same viruses at an MOI of 1 as a positive control. Cells were harvested 48 h postinfection and analyzed for the expression of ICP0, ICP22, and ICP47 by Western blotting (Fig. 2).

It can be seen from Fig. 2 that the 17+ 27− virus expressed significant amounts of all three of the IE genes that have not been deleted (ICP0, ICP22, and ICP47). This is consistent with the phenotype of ICP27 deletion mutants (23, 40, 51). The subsequent inactivation of VP16 and deletion of ICP34.5 (the 1764 27− virus) leads to reduced but still significant levels of ICP0, ICP22, and ICP47 expression, although ICP22 and ICP47 are here expressed at higher levels at a low rather than high MOI, which is discussed later. Thus, the inclusion of the in1814 mutation to VP16 in the context of an ICP27-deleted virus is not sufficient to completely prevent the transactivation of the remaining IE gene promoters and leads to an unusual expression pattern for ICP22 and ICP47. However, prevention of transactivation of the other IE gene promoters can be achieved by the additional deletion of ICP4 (the 1764 27− 4− virus). The Western blots for ICP0, ICP22, and ICP47 reveal that there is no detectable expression of any of these IE genes from this virus, even at a high MOI.

The 1764 27− 4− virus does not therefore express significant amounts of any of the IE genes. The essential IE genes ICP4 and ICP27 have been completely deleted from the viral backbone, and the remaining IE genes ICP0, ICP22, and ICP47 are not expressed. This lack of IE gene expression requires the combination of VP16 inactivation and ICP4/ICP27 deletion and provides a level of disablement to the virus similar to that resulting from the individual deletion of each of the IE genes (52).

Virus strain 1764 27− 4− pR20.5 simultaneously expresses high levels of two exogenous genes.

The large genome size of HSV is often cited as one of the potential benefits of HSV as a gene delivery vector. However, to date, HSV has only been used to deliver multiple genes in a highly transient fashion (31). The pR20.5 cassette has been described previously (45) and consists of a central LAT P2 element flanked by two heterologous promoters (cytomegalovirus [CMV] and Rous sarcoma virus [RSV]) arranged in a back-to-back orientation (Fig. 1), designed to allow the simultaneous expression of two exogenous genes. This is demonstrated in Fig. 3a, where the same plaque can be seen to be simultaneously expressing two transgenes (lacZ and the gene coding for green fluorescent protein [GFP]) in complementing cells.

It has been previously reported that in the absence of IE gene products, the expression of inserted genes is undetectable in the majority of transduced cells even at short times postinfection (48, 52). It has been proposed that this absence of transgene expression is attributable to an active repression of both HSV and non-HSV promoters by a cellular factor, but that this effect is masked in less-disabled viruses by the presence of the potent viral transactivators VP16, ICP0, and ICP4 (48). In Fig. 3b, noncomplementing BHK cells were infected at an MOI of 10 and visualized by fluorescence microscopy or processed to detect lacZ expression 48 h postinfection. This shows abundant transgene expression from the 1764 27− 4− pR20.5 virus, even though IE gene expression levels are minimal (Fig. 2). This could be due to LAT P2 elements in the promoter preventing the transcriptional shutoff, which might otherwise occur (45), and/or might alternatively be due to low-level ICP0 expression, which, while undetectable in Western blots, might still aid expression from the CMV and RSV promoters used.

Exogenous gene expression from cassettes inserted into virus strain 1764 27− 4− is dose dependent in noncomplementing cells.

It can be seen from Fig. 2 that the expression level of ICP22 and ICP47 from the 1764 27− virus in noncomplementing cells is not dose dependent, the peak of expression occurring at an MOI of 1. This suggests that with this intermediately disabled virus, interactions between the virus genome and remaining viral factors are affecting virus gene expression. In order to examine this further, noncomplementing BHK cells were infected with a selection of disabled viruses over a wider range of MOIs, and the samples were harvested 48 h postinfection. Western blots were then probed for the expression of ICP22 and ICP47 as before. This experiment showed that in the case of the 17+ 27− virus, ICP22/47 expression is dose dependent, but with the 1764 27− virus, the peak of ICP22/47 expression is at an MOI of 0.5 (data not shown). As in Fig. 2, no expression of ICP22 was detected at any MOI from the 1764 27− 4− virus. This phenomenon was investigated further and found to also apply to exogenous genes inserted into the 1764 27− virus, but not the less-disabled 17+ 27− virus or the more-disabled 1764 27− 4− virus. As an example of an exogenous gene, a Western blot for CMV-driven GFP expression from all three viruses is shown (Fig. 4). Virus strains 17+ 27− pR19GFP and 1764 27− 4− pR20.5 demonstrate a dose-dependent pattern of GFP expression. However, with virus strain 1764 27− pR20.5, the pattern of GFP expression follows that of ICP22/47, with expression levels peaking at an MOI of 0.5. The reason for this expression pattern is unclear, but a similar dose effect has been reported previously when ICP0 and ICP4 were used in combination to activate IE gene target promoters in transient transfection assays (20, 21). Whatever the mechanism, it might be anticipated that a linearly dose-dependent pattern of transgene expression such as that directed by the most-disabled virus would be more desirable in gene delivery applications than the pattern shown by the less-disabled virus.

Virus strain 1764 27− 4− expresses ICP6.

Because ICP6 (the large subunit of ribonucleotide reductase, required for virus replication in neurons) is inefficiently expressed in the absence of prior viral protein synthesis, it was originally classified as an E gene (25). However, the identification of IE-type cis-responsive elements in the promoter for ICP6 expression has led it to now be considered a hybrid IE/E gene (61). Unlike classical E genes, high levels of ICP6 are expressed in the absence of ICP4 (14). We therefore examined whether virus strain 1764 27− 4− still allowed ICP6 to be expressed.

To test levels of ICP6 expression, noncomplementing BHK cells were infected at MOIs of 10, 5, and 1 with the viruses 17+ 27− pR19GFP, 1764 27− pR20.5, and 1764 27− 4− pR20.5. Complementing 27/12/M:4 cells (62) were infected with the same viruses at an MOI of 1 as a positive control. Cells were harvested 48 h postinfection and analyzed for the expression of ICP6 by Western blotting (Fig. 5).

Figure 5 shows that ICP6 was expressed from all three viruses and that the deletion of ICP4 made little difference to the levels observed, as might be expected from previous work (14, 54). However, because previous work has also shown ICP6 to be nontoxic (30), it was not anticipated that ICP6 expression would impair gene delivery when the vectors were used in the CNS.

Virus strain 1764 27− 4− directs high-efficiency gene delivery to cultured neurons.

In order to test the ability of 1764 27− 4− viruses to deliver exogenous genes to cultured neurons, primary cultures of DRG neurons were infected at an MOI of 10 with 1764 27− 4− pR20.5 (Fig. 6). Mock-infected neurons or neurons infected with virus strain 17+ 27− pR19GFP were used as controls. One week postinfection, neurons were photographed under phase-contrast and fluorescence microscopy. Both viruses were able to efficiently deliver exogenous genes to neurons in culture. However, by 7 days postinfection, the 17+ 27− virus caused considerable toxic effects, with loss of supporting cells, clumping of neuronal cell bodies, and retraction of neuronal processes. In contrast, cells infected with the 1764 27− 4− virus are indistinguishable in morphology from mock-infected cells under phase-contrast microscopy. The fluorescence photomicrograph of neurons infected with the 1764 27− 4− virus shows that both neuronal cell bodies and processes show abundant levels of GFP, demonstrating efficient gene delivery.

Virus strain 1764 27− 4− pR20.5 directs the expression of delivered genes for extended periods in neuronal and nonneuronal cells.

Previous work found that the genomes of viruses individually mutated for all of the IE genes are capable of persisting in Vero cells for at least 28 days postinfection (52), but that gene expression from the CMV promoter was completely repressed immediately after infection in the vast majority of cells. Gene expression could, however, be stimulated by superinfection with a less-disabled virus expressing ICP0 (52).

In order to test if the 1764 27− 4− virus could also establish a persistent infection, a similar experiment was performed. Vero cells at 50% confluence were infected at an MOI of 10 with virus strain 1764 27− 4− pR20.5. Following infection, cells were maintained in a low serum concentration at 34°C. At the last time point (23 days), a duplicate well was superinfected at an MOI of 5 with virus strain 17+ 27− (expressing no reporter gene). The results of this experiment are shown in Fig. 7. Here it can be seen that GFP expression was evident at all time points during the experiment and that, unlike in the previously published work (52), superinfection was not necessary for transgene expression to occur. GFP expression was present in decreasing numbers of cells over time, as in previous work (52), which is probably due to cell division in the culture and the finite life span of both transduced and untransduced cells. Superinfection of the cultures at 23 days did increase GFP expression levels slightly, but the effect was not marked. This demonstrates that the genome of virus strain 1764 27− 4− pR20.5, like that of the multiple IE gene-deleted viruses reported previously (52), is capable of persisting in the nucleus of infected cells, indicating the low toxicity of the vector. At all times up to the end of the experiment, the cells infected with virus strain 1764 27− 4− pR20.5 were indistinguishable from mock-infected cells (not shown). However, in contrast to previous work (52), virus strain 1764 27− 4− pR20.5 allowed continuous GFP expression without the need for superinfection. This previously unreported continued gene expression might be speculated to be due to the LAT P2 element in the promoter used (which has previously been shown to direct gene expression from heterologous promoters during herpesvirus latency [45]), residual ICP0 expression, or a combination of the two.

In order to extend these findings, the expression characteristics of virus strain 1764 27− 4− pR20.5 in rat organotypic hippocampal slice cultures was also investigated. At various times after infection of the cultures with virus strain 1764 27− 4− pR20.5, a single slice was photographed under a fluorescence microscope (Fig. 8).

Figure 8 shows that the high level of transgene expression observed at 2 days postinfection was still clearly detectable in the cultures 3 weeks after infection, albeit in a reduced number of cells. The processes of transduced neurons, even at the 3-week time point, were not swollen or beaded, suggesting that degeneration of transduced neurons had not occurred. Infected cultures maintained normal morphology and did not differ from mock-infected control cultures at any time. These results again suggest that the toxic effects of 1764 27− 4− pR20.5 on transduced cells are minimal, at least by morphological criteria, even over extended periods in culture.

Virus strain 1764 27− 4− gives widespread gene expression in the CNS following retrograde transport from the site of injection.

In order to examine transgene expression in the CNS in vivo with the viruses developed in this study, injections into the rat striatum, spinal cord, or superior colliculus were carried out with the 1764 27− 4− pR19lacZ virus. Transgene expression (lacZ) at both the injection site and connected sites within the nervous system was analyzed. The results of these experiments are shown in Fig. 9. In each case, the virus gave high-level gene expression at the site of injection and was also efficiently transported from the site of injection to the cell bodies at connected sites. This was clearly demonstrated by transport of the disabled virus from the striatum to the substantia nigra, from the superior colliculus via the optic nerve to the retinal ganglion cells, and from the C6 level of the spinal cord to DRG, the brainstem, and areas of the hind- and midbrain.

Both the promoter cassette and the vector backbone affect the levels of transgene expression in the CNS at inoculated and distant sites.

In order to test whether a promoter cassette containing the LAT P2 region was sufficient to drive the expression of a transgene and allow gene delivery to connected sites in the CNS regardless of the virus backbone, the following experiment was performed. Virus strain 17+ 27−, 1764 27−, or 1764 27− 4−, each containing the pR19lacZ cassette (Fig. 1), was stereotaxically injected (2 × 10⁵ PFU) into the striatum of adult rats. The results of injection of this set of viruses are shown in Fig. 10. All three viruses gave lacZ expression at the early time points, but expression levels from the 17+ 27− and 1764 27− viruses were dramatically reduced between 3 days and 1 week postinjection, and expression was only detectable in a few cells by 1 month. However, with the 1764 27− 4− virus, expression was strong at 1 week and was still clearly evident at 1 month postinjection, in both the striatum and the substantia nigra. This suggests that the lack of expression with the 17+ 27− and 1764 27− viruses at later time points was due to the residual toxicity of these less-disabled viruses, rather than transcriptional shutoff of the LAT P2/CMV promoter, because this was able to remain active in the 1764 27− 4− virus.

It is also noteworthy that at early times postinjection, the expression pattern differed according to the level of disablement of the virus. In the case of the 17+ 27− virus, the striatum, substantia nigra, and cortex were all stained at 3 days postinjection. However, by 1 week postinjection, the striatum was not labeled, and gene expression was only apparent at connected sites—the majority of gene expression was in the cortex, although some expression in the substantia nigra could still be seen. By 1 month postinjection, labeling of only a few cells was seen with the 17+ 27− virus. However, with the intermediately disabled 1764 27− virus, a different expression pattern was observed. Here, intense X-Gal staining was seen in the striatum, with some staining of the substantia nigra at 3 days postinjection. This pattern was maintained, but significantly reduced, by 1 week postinjection, and again only low-level gene expression was seen at 1 month postinjection. Unlike with the 17+ 27− virus, no gene delivery to cortical neurons other than those adjacent to the needle track occurred with this virus at any time point. With the 1764 27− 4− virus, lacZ expression was of considerably greater intensity in both the striatum and the substantia nigra at all time points during the experiment, but no delivery to cortical neurons was seen.