805. Fatality in Mice Due to Oversaturation of Cellular Micro/Short Hairpin RNA Pathways

Abstract

Induction of RNA interference (RNAi) by expression of short hairpin RNA (shRNA) holds significant promise for functional genomics and human gene therapy. Therefore, numerous non-viral and viral shRNA expression vectors are currently being developed and evaluated in pre-clinical models of human disease. Here we report efficient expression of shRNAs in entire murine livers from intravenously delivered Adeno-associated virus (AAV) serotype 8, carrying a novel stabilized double-stranded (sds) vector genome with the shRNA cassette flanked by an intact AAV-2 and a truncated AAV-4 terminal repeat (ITR). As opposed to conventional dsAAV with two AAV-2 ITRs, the use of heterologous repeats prevented recombination of the genome ends and thus markedly enhanced vector stability and yield. We engineered a panel of 49 distinct sdsAAV-8 vectors to express shRNAs unique in stem length (19 to 29 nucleotides), sequence, loop structure, and target (human alpha-1-antitrypsin or luciferase gene, or human Hepatitis B or C virus genome). Strikingly, their in vivo evaluation in various wildtype mice, or in animals transgenic for the targets, showed that 36 shRNAs resulted in dose-dependent and often massive liver injury. Typical observations were rises in liver enzymes, emergence of icteric sera, drops in liver-derived albumin, and as a result of the latter, frequent occurrence of ascites. Moreover, histological analyses revealed the presence of megakaryocytes, indicative of liver damage and hepatocyte repopulation. In fact, shRNA-induced liver injury was often associated with a transient RNAi effect, peaking within weeks after vector delivery and completely lost thereafter. Immunohistochemical and Southern blot analyses confirmed that shRNA expression caused individual cell death and liver regeneration with hepatocytes lacking the AAV/shRNA, explaining the transient phenotype. Most strikingly, 23 of the tested shRNA vectors ultimately caused death of the treated animals within a month after the injection. Northern blot analysis of total liver RNA showed that morbidity correlated with high shRNA expression levels, which were in turn frequently associated with down-regulation of liver-derived micro-RNAs. The latter remained stable in alternative models of liver injury, implying a global competition mechanism specifically from shRNA overexpression. In fact, subsequent in vitro and in vivo studies hinted at the karyopherin Exportin-5 as a limiting factor in the common pathway. At this point, we hypothesize that this carrier is used by shRNAs and miRNAs for nuclear export, and that its saturation with over-abundant shRNAs is a main cause for the observed liver toxicity and animal morbidity. Our findings thus strongly imply that control over intracellular shRNA expression levels will be mandatory for the clinical RNAi use, and generally underscore the inherent risks of this technology. These can likely be minimized by optimizing shRNA dose and sequence, as shown by our findings of persistent and therapeutic RNAi against human Hepatitis B virus in vivo.