Engineering a thermostable RNA ligase for use in an unbiased microRNA sequencing protocol

Abstract

MicroRNAs are small, non-coding RNA sequences that act to regulate gene expression at a post translational level. In humans, irregular expression of microRNAs has been associated with the development of numerous diseases. The ability to determine alterations in microRNA expression may, therefore, provide a novel method for predicting the onset, severity, and outcomes of these diseases. Identifying changes in microRNA expression relies on accurately sequencing and profiling microRNAs. Current microRNA sequencing protocols, however, are plagued by bias. To successfully sequence microRNAs, adapters must be ligated to both ends of the molecule. Current protocols utilise two separate reactions to achieve this. In the first, a pre-adenylated, 3′ amino modified DNA adapter is ligated to the 3′ end of the microRNA. In the second, a ligase must adenylate the 5′ phosphate of the microRNA, utilising ATP, beforeligating it to the 3′ end of a 5′ dephosphorylated RNA. It is in the ligation of these adapters that the bias originates. Different microRNA secondary and tertiary structures have been shown to differentially alter the ability of ligase enzymes to interact with the microRNA. By running the adapter ligation steps of the sequencing protocol at temperatures high enough to melt microRNA structures, this bias should be eliminated. The RNA ligase from the hyperthermophilic archaeon Pyrococcus furiosus (Pfu) was identified as a target of interest for use in our sequencing protocol due to its remarkable thermostability. To optimise the activity of the ligase for ligation of a pre-adenylated 3′adapter to microRNA, its adenylation activity needed to be removed. This activity is associated with the generation of undesirable ligation products. Preliminary research, carried out by Dr Tifany Oulavallickal, identified residues K92 and K238 as targets of interest for mutagenesis. A K92A variant of the Pfu RNA ligase was generated as a proof of concept, displaying significantly reduced adenylation activity while retaining ligation activity at the desired temperatures. The research conducted for this thesis built upon this work, characterising all other possible K92 substitutions, and utilising that information to inform separate, and co-substitution of K238 (i.e., single, and double mutants). All the possible K92 variants were successfully generated. Ligation activities with both DNA and RNA sequences were then assessed by endpoint TBE-urea gel assays, with K92A, K92G, K92S, K92T, and K92Y being identified as substitutions of interest. These same amino acids (A, G, S, T, and Y) were then used to replace the second active site lysine, K238. Double mutants were also generated by substituting K238 of the K92A variant for A and Y. All K238 single mutants, and both double mutants were successfully characterised. All K92 variants displayed significant decreases in adenylation activity, with the amino acid substitutions A, G, S, T, and Y resulting in increased ligation activities. All K238 variants, bar K238Y, displayed significant ligation activities, but continued to display adenylation activity. As such, K92A was identified as the best candidate for the 3′ DNA adapter ligation. This variant displayed the most promising ligation activity of all K92 variants, while displaying low enough adenylation activity to minimise the production of undesirable ligation products. The adenylation activity observed in the K238 variants made them ideal for the 5′ RNA adapter ligation, as this requires the microRNA to be adenylated. As such, the most active K238 variant, K238G, was identified as the best candidate for this adapter ligation.