Substrate-Specific Kinetics of Dicer-Catalyzed RNA Processing

Abstract

The specialized ribonuclease Dicer plays a central role in eukaryotic gene expression by producing small regulatory RNAs—microRNAs (miRNAs) and short interfering RNAs (siRNAs)—from larger double-stranded RNA (dsRNA) substrates. Although Dicer will cleave both imperfectly base-paired hairpin structures (pre-miRNAs) and perfect duplexes (pre-siRNAs) in vitro, it has not been clear whether these are mechanistically equivalent substrates and how dsRNA binding proteins such as trans-activation response (TAR) RNA binding protein (TRBP) influence substrate selection and RNA processing efficiency. We show here that human Dicer is much faster at processing a pre-miRNA substrate compared to a pre-siRNA substrate under both single and multiple turnover conditions. Maximal cleavage rates (V_max) calculated by Michaelis–Menten analysis differed by more than 100-fold under multiple turnover conditions. TRBP was found to enhance dicing of both substrates to similar extents, and this stimulation required the two N-terminal dsRNA binding domains of TRBP. These results demonstrate that multiple factors influence dicing kinetics. While TRBP stimulates dicing by enhancing the stability of Dicer–substrate complexes, Dicer itself generates product RNAs at rates determined at least in part by the structural properties of the substrate.

Introduction

MicroRNAs (miRNAs) and short interfering RNAs (siRNAs) play central roles in controlling eukaryotic gene expression and have attracted increasing attention as potential therapeutic agents. In the cytoplasm, many of these RNAs are generated from cleavage of longer double-stranded RNA (dsRNA) substrates by the enzyme Dicer, a large multidomain, ribonuclease III family protein.1, 2 However, the precursors to these RNAs are structurally distinct: pre-miRNAs are predicted to be imperfectly base-paired hairpins with a variable-size loop, whereas pre-siRNAs are usually perfectly base-paired duplexes of variable length with no loop. Some species have functionally distinct Dicers that process different kinds of substrates. For example, in Drosophila melanogaster, Dcr-1 and Dcr-2 are responsible for the biogenesis of miRNAs and siRNAs, respectively.³ However, there is only one known Dicer in the human system that putatively processes both pre-miRNAs and pre-siRNAs. Understanding how processing rates for human Dicer are affected by substrate differences and by Dicer's protein-binding partners is paramount to identifying the molecular mechanisms that underlie regulatory RNA production.

The catalytic core of Dicer, represented by the structure of the enzyme from Giardia intestinalis, comprises a PAZ domain N-terminal to two ribonuclease III domains.⁴ This structure, likely to be common to all Dicer enzymes, led to a model for dsRNA recognition in which substrates are positioned for cleavage to produce dsRNA products of a specific 21- to 27 -bp length.¹ However, it remains unclear how Dicer accommodates substrate RNAs with the different helical structures and thermodynamic stabilities that characterize pre-miRNAs and pre-siRNAs.

In most eukaryotes, Dicer contains additional domains beyond those found in the Giardia enzyme, including an N-terminal DExH/D-box helicase, a domain of unknown function (DUF 283) and a C-terminal dsRNA binding domain (dsRBD) (Fig. 1a). These domains function at least in part to recruit dsRNA binding proteins that are central players in RNA-mediated gene silencing.6, 7, 8, 9

Studies of RNA samples isolated from mammalian cells having or lacking Dicer show that a majority of Dicer products are miRNAs.10, 11 Some examples of endogenous siRNAs have been identified, but their overall occurrence is relatively uncommon in these cell types.12, 13 Nevertheless, siRNA precursors such as short hairpin RNAs are thought to be processed by human Dicer, and some studies suggest that such processing has a vital role in enhancing siRNA efficiency by facilitating optimal RNA-induced silencing complex (RISC) loading.14, 15

In addition to Dicer's potential for differential substrate recognition, double-stranded RNA binding proteins that directly interact with Dicer could influence its activity.¹⁶ These proteins, typically comprising two to three consecutive dsRBDs connected by variable-length linkers, include RDE-4 in Caenorhabditis elegans,¹⁷ R2D2 and Loquacious/R3D1 in D. melanogaster18, 19 and trans-activation response (TAR) RNA binding protein (TRBP) and PACT in Homo sapiens.6, 7 In C. elegans, RDE-4 is thought to be exclusively involved in the siRNA pathway.²⁰ In D. melanogaster, the substrate preferences detected for R2D2 and Loquacious/R3D1 suggest roles in selecting pre-siRNA and pre-miRNA substrates, respectively, and in specifying guide strand selection during RISC assembly.19, 21, 22, 23 In humans, TRBP ensures efficient recruitment of miRNA onto Argonaute 2 (Ago-2) in RISC and enhances the stability and efficiency of both Dicer and RISC.6, 14, 24, 25, 26 Perhaps as a result, changes in steady-state levels of certain miRNA families correlate with TRBP truncation mutations that characterize some colorectal cancers.²⁷ PACT has also been shown to bind to human Dicer,7, 26 although its effect on dsRNA binding or processing remains unclear.

To investigate the effects of substrate variants and TRBP on dicing kinetics, we compared the processing rates of two distinct dsRNA substrates containing two-nucleotide 3′ overhangs: a 73-nucleotide hairpin with an imperfect stem (human pre-let-7a) and a 35-bp perfect duplex (Fig. 1a). Dicer showed a marked preference for processing pre-let-7a over the 35-bp duplex, an effect most pronounced under multiple turnover conditions in which substrate is in excess over enzyme. TRBP increased the rate of endonucleolytic cleavage of both substrates by ∼ 4- to 5-fold under multiple turnover conditions. This stimulatory effect required the two N-terminal dsRBDs of TRBP. These results reveal substantial variability in Dicer-catalyzed small RNA processing depending on substrate sequence and secondary structure. Although TRBP stimulates dicing activity, substrate-dependent differences in dicing rates are inherent to human Dicer itself. Our findings imply that levels of miRNAs and siRNAs are influenced at least in part by precursor structure and/or sequence. Furthermore, levels of regulatory RNA in the cell could change as a consequence of mutations that alter substrate RNA structure or stability.