Topical administration of chol-siRNAs targeting an HSV-2 viral gene and a host receptor induced sustained viral resistance in the genital tract of female mice. Mice were protected against challenge with 2LD50 of HSV-2 for at least a week independently of whether the challenge was performed just before siRNA administration or delayed until a week later. The administered siRNAs were stabilized by a single PS linkage to protect them from degradation by genital fluid RNases and were cholesterol-conjugated to facilitate intracellular entry.
These results improve upon our earlier study that used lipoplexed unmodified antiviral siRNAs (Palliser et al., 2006
). We found that we needed to modify the approach we had developed for a topical siRNA-based antiviral microbicide because the transfection lipid used to deliver the siRNAs into cells on its own facilitated viral infection and because protection was short-lived. Our results suggest that the lipid transfection reagent induced slight, but significant, inflammation as measured by an increase in infiltrating hematopoietic cells, and this may be sufficient to enhance viral uptake, resulting in a decrease in mouse survival. Alternatively, the transfection lipid might enhance viral entry by facilitating fusion of the viral envelope with the cell membrane. Therefore lipoplexed siRNAs should probably not be considered for topical delivery of siRNAs to any mucosal surface, since they might also increase transmission of other enveloped viruses. Although unmodified siRNAs can apparently be taken up and induce silencing in the pulmonary epithelium without the aid of a transfection reagent (Bitko et al., 2005
), we did not find that to be the case in the genital tract of mice. It is possible that application of significantly higher siRNA concentrations could knock-down gene expression in the tissue, but we did not evaluate this since chol-conjugated siRNAs provided an efficient and apparently nontoxic method for genital tract delivery. However, our evaluation of toxicity was limited to gross assessments of epithelial cell integrity and inflammatory infiltrates and induction of inflammatory cytokines. Although mice treated with a 5-fold higher dose of siRNA did not appear to become sick, this dose did not afford any viral protection. This may be because of dose-dependent off-target effects or toxicity, which would need to be carefully evaluated for further preclinical development.
Ivag administration of chol-siRNAs did not induce substantial interferon or inflammatory responses. In contrast to two recent studies that reported induction of high levels of inflammatory chemokines, IFNβ, IFNγ and IL12 by some siRNA preparations (Kleinman et al., 2008
, Robbins et al., 2008
), we found no indication that these cytokines were significantly induced following siRNA treatment at the low dose we used (1 mg/kg, twice). In addition, no survival advantage was observed in mice given siRNAs targeting GFP compared with mice receiving no siRNAs. Therefore, the protective effect is likely a direct consequence of specific siRNA-mediated gene silencing.
HSV-2 infects epithelial cells and dorsal root ganglion neurons. Our results support efficient chol-siRNA delivery to epithelial cells, but this acute HSV-2 infection model does not address whether neuronal processes that extend into the tissue are effectively transduced. Topically applied siRNAs might be useful to treat and prevent reactivation and sexual transmission of clinically latent HSV-2 infection, which is not truly latent since chronically infected women without clinical symptoms of active disease are constantly shedding infectious virus (Wald, 2004
). An ideal drug would also prevent viral replication in the dorsal root ganglion viral reservoir. Further studies are therefore needed to examine siRNA transduction of these cells. We would also like to apply the results of this study to developing a topical microbicide to prevent HIV-1 infection, an urgent global public health need given the failures of HIV vaccine development. Since HIV, like HSV-2 is an enveloped virus, transfection lipid-mediated delivery would likely be inadvisable. To prevent HIV transmission, we need to develop a way to transduce immune cells in the genital tract responsible for HIV transmission, especially Langerhans cells, macrophages and T cells. Because these cells are rare in the uninflamed genital tissue of mice maintained under pathogen-free conditions, we do not yet know whether chol-siRNAs will be suitable, but are evaluating this.
This study confirms that nectin-1, a cell adhesion molecule that localizes at the adherans junctions of epithelial cells, is an important HSV-2 receptor (Geraghty et al., 1998
, Cocchi et al., 1998
, Shukla et al., 2000
). In addition to nectin-1, herpesvirus entry mediator (HVEM) has been reported as an efficient entry receptor for HSV (Montgomery et al., 1996
). However, in mice genetically deleted for HVEM, little difference in HSV-2 ivag infectivity was observed. By contrast, mice lacking nectin-1 exhibited reduced infection levels in the vaginal epithelium and decreased viral spread to the nervous system. Death was delayed in nectin-1 knockout mice and approximately half the mice succumbed to disease (Taylor et al., 2007
). The superior protection from HSV-2 infection afforded by nectin-1 knock-down, where 80% of mice were protected, compared to knockout mice could be due to differences in mouse or virus strains or potentially to compensatory changes in expression of members of the nectin and nectin-like family in knockout animals. The coordinated change during diestrus in nectin-1 expression and subcellular localization to the luminal surface with susceptibility to HSV-2 infection in mice further supports the importance of nectin-1 as a viral receptor, as previously reported (Linehan et al., 2004
). Of note, nectin-1 is expressed at all stages of the menstrual cycle in humans (Linehan et al., 2004
Therapeutically knocking down a host gene can be deleterious if the gene is required for normal cellular function. Because nectin-1 localizes to adherens junctions, it might be required to maintain the integrity of the epithelial barrier. However, we saw no changes in histology of the epithelial layer by hematoxylin and eosin staining. Nectin-1 has been identified as a gene mutated in cleft lip or cleft palate syndrome. The incidence of this mutation in some populations may be a result of selection due to a survival advantage conferred by relative resistance to HSV-1 and HSV-2 (Suzuki et al., 2000
). However, whether this mutation confers a protective effect is, as yet, undetermined. Therefore, nectin-1 may be needed for development but may be a nonessential gene (and good microbicide target) after that. However, the use of nectin-1 siRNAs might not be without risk during pregnancy.
siRNAs targeting viral genes were effective over a limited time frame, but siRNAs targeting a putative host receptor provided more sustained protection against both in vitro and in vivo viral challenge. We previously found that siRNAs targeting CCR5, the host coreceptor for HIV protected macrophages from HIV-1 infection in vitro for weeks, whereas protection waned after a few days when uninfected cells were transduced with HIV-1 gag siRNAs (Song et al., 2003
). These data, together with the observation in C. elegans
that siRNAs are more stable when a target mRNA is expressed (Plasterk et al., 2002
), suggested that siRNAs targeting host genes might persist longer and provide more sustained protection than siRNAs targeting viral genes. Although the outcome was as expected, namely silencing nectin-1 provided more durable protection than UL29 siRNAs, the explanation was not what we expected. The half-lives of nectin-1 and UL29 siRNAs in uninfected NIH3T3 cells are in fact comparable. Similarly we did not find a difference in half-life of GFP siRNAs in GFP-expressing and nonexpressing HeLa cells (data not shown). Therefore the stability of an siRNA does not appear to be affected by the expression of its target gene. However, there was significantly more intracellular nectin-1 siRNA compared to UL29 siRNA at all time points. This may be because the nectin-1 siRNA was more efficiently taken up into RNA-induced silencing complexes (RISC) and available for gene silencing. In fact, the level of intracellular nectin-1 siRNA 7 days after transfection was similar to the level of UL29 siRNA on the first day after transfection. What this suggests is that the difference in durable protection may not be due to targeting a host receptor vs a viral gene, but rather to the use of a more effective siRNA. Therefore identifying more effective siRNAs that are either more efficiently incorporated into RISC or more active at directing target mRNA degradation might not only generate siRNAs active at a lower dose, but that last longer. For this study and our previous study, we only tested a few siRNAs for each target gene; a systematic approach to identify host or viral gene siRNAs active at pM concentrations in vitro would be advisable for preclinical development. These could be directed against either viral or host genes. Use of optimized siRNA sequences or further chemical modification of siRNAs to enhance stability in the genital tract might lower the siRNA dose or extend durability of effective protection against sexual transmission.