697. New Insights Into the Mechanisms of Argonaute Protein Competition and Implications for RNAi Gene Therapies

Abstract

RNA interference (RNAi) has become an established tool for basic gene annotation and a promising option for therapeutic concepts that require specific inhibition of dysregulated or exogenous genes. However, we and others accumulated strong evidence that overexpression of short hairpin RNAs (shRNAs, an expressable and widely used form of RNAi trigger) can cause cytotoxicity leading up to organ damage and lethality in shRNA-treated animals. A meticulous dissection of the underlying mechanisms is critical as it will ultimately guide the design of safer RNAi therapies and thus foster their clinical translation. One informative finding was that co-expression of Ago2 (Slicer, as it cleaves targeted mRNA), a key component of the RNA-induced silencing complex RISC, increases shRNA efficiency in cells and animals, implying that Ago2 is a rate-limiting factor whose saturation may be involved in cytotoxicity. Moreover, over-expression of the three other mammalian Ago proteins – Ago1, Ago3 or Ago4 (all slicing-incompetent) – dampens RNAi potency, underscoring that a delicate balance of all four Ago variants is key to efficient and safe gene silencing. Here, we investigated three hypotheses that could explain how relative Ago1-4 levels may affect the functionality of shRNA-loaded RISC and hence the strength of target knockdown: over-abundance of non-Slicer Ago proteins could (i) dysregulate Ago2 expression (transcription and/or translation), (ii) re-localize intracellular Ago2 away from sites of RNAi activity, or (iii) quantitatively sequester shRNAs into slicing-incompetent RISC. Using codon-optimized Ago variants that are distinguishable by real-time PCR from the endogenous counterparts, as well as Western blotting, we could eliminate model (i). Likewise, wide-field and confocal microscopy analyses revealed no evidence for Ago2 re-localization upon Ago1/3/4 over-expression and vice versa. We therefore focused on hypothesis (iii) and fused the Ago1-4 cDNAs with GFP or RFP tags that enabled their specific immunoprecipitation, alone or in combination, using the TRAP system. Indeed, we found that Slicer and non-Slicers compete for shRNA loading and mRNA knockdown in a dose-dependent manner, readily explaining all prior observations of RNAi enhancement by Ago2 over-expression, or shRNA inhibition by Ago1/3/4, respectively. Importantly, the combination of GFP/RFP-Ago fusion constructs with the co-precipitation protocols newly established during this work should be widely useful to address other vital questions in the RNAi field, such as the selective processing of different cellular small RNAs by the four Ago proteins. Moreover, our notion that over-expression of non-Slicers has no adverse effect on endogenous Ago2 levels or localization paves the way for novel strategies to fine-tune therapeutic RNAi in a safe and potent manner. As a prototype, we present a new shRNA/Ago co-expression system whereby Ago1/2 levels can be regulated independently through exogenous macrolides, permitting a tight control of target knockdown efficiencies and providing an original avenue to reduce toxicity in future RNAi gene therapy applications.