An siRNA-based microbicide protects mice fromlethal herpes simplex virus 2 infection - siRNA

An siRNA-based microbicide protects mice fromlethal herpes simplex virus 2 infection
January 2006
Deborah Palliser Dipanjan Chowdhury Qing-Yin Wang, Sandra J. Lee, Roderick T. Bronson,David M. Knipe& Judy Lieberman
Nature
Herpes simplex virus 2 (HSV-2) infection causes significantmorbidity1and is an important cofactor for the transmission ofHIV infection2. A microbicide to prevent sexual transmission ofHSV-2 would contribute substantially to controlling the spread ofHIV and other infections3,4. Because RNA interference (RNAi)provides effective antiviral defence in plants and other organisms,several studies have focused on harnessing RNAi to inhibit viralinfection5. Here we show that vaginal instillation of small inter-fering RNAs (siRNAs) targeting HSV-2 protects mice from lethalinfection. siRNAs mixed with lipid are efficiently taken up byepithelial and lamina propria cells and silence gene expression inthe mouse vagina and ectocervix for at least nine days. Intra-vaginal application of siRNAs targeting the HSV-2 UL27 and UL29genes (which encode an envelope glycoprotein and a DNA bindingprotein6, respectively) was well tolerated, did not induce inter-feron-responsive genes or cause inflammation, and protected micewhen administered before and/or after lethal HSV-2 challenge.These results suggest that siRNAs are attractive candidates for theactive component of a microbicide designed to prevent viralinfection or transmission.
Most mammalian cells do not take up siRNAs without a transfec-tion reagent. We instilled fluorescein isothiocyanate (FITC)-labelledsiRNAs complexed with a transfection lipid into the mouse vagina.The vaginal and ectocervical epithelium, underlying lamina propriaand stroma efficiently took up the fluorescent siRNAs (Fig. 1a).When siRNAs targeting enhanced green fluorescent protein (EGFP)were administered intravaginally with lipid to transgenic GFP micethat express EGFP in every cell from theb-actin promoter7, GFPexpression three days later was down-modulated throughout thevagina and cervixof GFPsiRNA-treated mice, but not in control mice(Fig. 1b). Intravaginal siRNAs did not cause systemic silencing indistant organs such as the liver. Silencing persisted without dimin-ution for at least nine days (the total length of the experiments) underconditions in which epithelial turnover was reduced by treatmentwith medroxyprogesterone acetate (Fig. 1c). Further studies arerequired to determine how long silencing persists and to assess theeffect of menstrual variation on durability. Nonetheless, the extentand persistence of silencing suggests that siRNAs are attractivecandidates for the active component of a microbicide. Moreover,their durability suggests that an RNAi-based microbicide might notneed to be administered just before sexual intercourse, mitigatingone of the main problems with microbicides: compliance.
To determine whether topical siRNA application could protectagainst sexually transmitted infection, seven siRNAs targeting threeessential HSV-2 genes?UL5 (a component of the helicase?primasecomplex), UL27 (envelope glycoprotein B) and UL29 (a DNA-binding protein)6?were designed using the Dharmacon designprogram8. After overnight incubation, siRNA-treated NIH3T3(Fig. 2a) and Vero (Fig. 2b) cells were infected with HSV-2 strain186 at a multiplicity of infection (MOI) of 1, and viral replicationwasassessed by plaque assay 24 h later. UL5.2, UL27.2 and UL29.2siRNAs significantly reduced viral titre, but GFP siRNA and invertedUL29.2 siRNA did not (Fig. 2b, c).UL29.2 was the most effective siRNA, suppressing viral replicationby 62-fold in NIH3T3 cells and 25-fold in Vero cells. Viral replicationby UL29.2 was inhibited at siRNA concentrations of 25 nM, andreached a plateau at 100 nM siRNA (Fig. 2c and data not shown).Gene silencing was specific for the targeted gene. When UL27 andUL29 messenger RNAs were quantified by real-time polymerasechain reaction with reverse transcription (RT?PCR) in Vero cellstransfected one day earlier with UL27.2, UL29.2 or GFP siRNA andinfected with HSV-2, peak UL27 expression (6 h after infection) wassignificantly downregulated in response to UL27.2, but not to UL29.2or GFP siRNA (P , 0.004). Conversely, UL29, which is expressedearlier than UL27, was significantly downregulated both at 4 h and6 h, and only in response to UL29.2 siRNA (P , 0.0001 comparedwith GFP siRNA) (Fig. 2d). One day later, when infection hadamplified by cell-to-cell spread, the expression of all four viralgenes examined (siRNA-targeted UL5, UL27 and UL29 as well asthe viral thymidine kinase TK) was reduced by siRNAs targeting anyof the viral genes (Fig. 2e). These differences were all highlystatistically significant. Even the least effective siRNA (UL29.1)reduced viral replication (that is, TK expression; P , 0.002 com-pared with GFP siRNA). Control GFP siRNA did not affect viral genetranscription. Viral gene silencing roughly paralleled the inhibitionof viral replication, with UL29.2 siRNA proving the most effective,suppressing relative viral gene expression by 4?5-fold (P , 0.001compared with GFP siRNA). UL5.2 and UL27.2 siRNAs eachinhibited viral gene expression by ,3-fold (P , 0.002 for UL5.2,P , 0.001 for UL27.2 compared with GFP).To investigate whether siRNAs could protect mice from HSV-2infection, groups of 5?10 medroxyprogesterone-pretreated micewere given lipid-complexed UL29.2 intravaginally 2 h before and4 h after vaginal challenge with 2 LD50(2 ? 104plaque-formingunits, p.f.u.) of HSV-2 wild-type virus.
Mice treated with,250 pmol UL29.2 siRNA were not protected, mice treated with250 pmol siRNA were partially protected, and 500 pmol siRNA gaveoptimal protection (data not shown). We therefore administered500 pmol siRNA in subsequent experiments.UL29.2 siRNA provided highly significant protection, as assesseddaily by a clinical disease scoring system or by survival (Fig. 3a, b).Although 75% of infected mice treated with GFP siRNA (15/20) orno siRNA (13/17) died, only 25% of mice treated with UL29.2 (5/20)died (time to death comparison by log-rank test: P , 0.001 versusno treatment, P , 0.003 versus GFP siRNA). Although 55% ofUL29.2-treated mice developed some signs of infection, surviving mice were free of clinical disease by day 11. A longitudinal regressionanalysis of disease severity over time and between groups showedrobust protection in UL29.2-treated mice (P , 0.001 versus notreatment, P , 0.006 versus GFP siRNA when analysed with respectto time course; P , 0.001 versus either control when analysedbetween groups).Mice treated with UL27.2, which was less effective in vitro, wereless effectively protected. Sixty per cent (6/10) of mice survived thelethal vaginal challenge (P , 0.009 compared with untreated,P ? 0.10 compared with GFP siRNA). UL27.2 protection fromdisease severity was significant by longitudinal regression analysis(P , 0.001 compared with untreated, P , 0.005 compared withGFP siRNA with respect to time; P , 0.01 and P ? 0.05 whenanalysed between the respective groups). The clinical advantagewas also evident by quantifying shed virus six days after infection(Fig. 3c).
Although all infected mice not given siRNAs shed viruson day six, no virus was detected in 70% of UL29.2- and 50% ofUL27.2-treated mice. No virus was isolated from three out of nineGFP siRNA-treated mice, but this was not significantly different frommice not treated with siRNAs. Comparison of virus recovered fromUL29.2 siRNA-treated mice with GFP siRNA-treated mice also wasnot significant (P ? 0.09 by Wilcoxon rank sum test). However, thegeometric mean viral titre was reduced from 1,226 p.f.u. ml21inuntreated mice to 7.9 p.f.u. ml21in mice that received UL29.2(P , 0.01). Viral shedding at day six predicted survival, as 18 outof 19 mice from which virus was cultured died, whereas none out of15 mice with undetectable virus died.
One concern about using RNAi against viruses is escape fromRNAi by mutation of the targeted sequence. Escape mutation hasbeen shown for polio, HIV and hepatitis C9?11. We cloned and sequenced HSV-2 DNA from the day 6 vaginal swab from oneUL29.2-treated mouse that died and from one control mouse. Nomutations were found in 150-nucleotide stretches of UL29, whichincluded the targeted sequence, in 24 sequences analysed from eachmouse. Escape mutation is not anticipated to be as problematic forDNA viruses (such as HSV-2) as for RNA viruses.
NOTE: Please visit the website to read the article in its entirety in .pdf form.