Evolving Wonder-RNAs in a Test Tube

Abstract

RNA is a captivating polymer: it is chemically simple yet functionally diverse. RNA has been sought out by Mother Nature to take on many important biological functions, from genetic information encoding and transmission, to enzymatic catalysis and molecular recognition, to regulation of gene expression. Along the way, Nature has produced so many interesting RNA species, and, perhaps out of necessity, given them different sizes. At one end of the spectrum are microRNAs made of merely *22 nucleotides, and at the other end is the human 28S ribosomal RNA composed of 5,070 nucleotides! Yet, most functional RNA molecules, including many ribozymes and riboswitches, are middle-ofthe-roaders having 50–200 nucleotides. As far as the complexity of the structure and function is concerned, the size of the RNA does matter. The most fitting example is the ribosome, which is designed to execute an essential and demanding chemical transformation in a highly organized manner. Nature has adequately dealt with this challenge by creating one of most spectacular macromolecular assemblies in cells. Among several dozen macromolecules used to construct the ribosome are two super-sized RNAs that contain a few thousand nucleotides. One of Nature’s most intriguing but unsolved mysteries is the evolution process behind the selection of such long functional RNA molecules. It has been postulated that the modern size of ribosomal RNAs is a consequence of amalgamation of smaller fragments (Clark 1987; Gray and Schnare 1996) and sequence comparison studies have suggested that some ribosomal RNA domains are indeed older than others (Gray and Schnare 1996; Bokov and Steinberg 2009). Evidently, Nature has found a way to patiently build a complex function through progressive, domain-by-domain construction of ‘‘wonder-RNAs’’. A distinct trait of Homo sapiens is to learn and apply. Based on the principle of Darwinian evolution, in vitro selection technique was devised in 1990 (Ellington and Szostak 1990; Tuerk and Gold 1990), and has since been used to create numerous functional RNA (or single-stranded DNA) molecules. Key to the success of each in vitro selection experiment is the simultaneous exploration of a vast sequence space covered by a random library containing as many as 10 RNA siblings. Although many fascinating functional nucleic acids have been derived by this approach, large and complex wonder-RNAs are few and far between (Bartel and Szostak 1993). This may simply be an inevitable outcome of two inherent drawbacks associated with the random library approach: the inefficient sequence space coverage and the tyranny of simple structural motifs. A standard DNA synthesis produces *0.1 mg of DNA (*10 molecules), just enough to cover the entire sequence space of a 25-nucleotide sequence (4 & 10). When the random domain increases to 50 nucleotides (still smaller than a tRNA), the sequence space rises to *10 (4 & 10), translating to *35,000 t of DNA! Thus, the sequence space coverage for a long RNA is categorically infinitesimal. Even if we assume there are complex RNAs in such a library, the chance of finding them is further diminished due to the overabundance of simple structural motifs in the library (Schlosser and Li 2005). Computational analysis has shown that random libraries are primarily populated with simple structural motifs like hairpin loops and three-way junctions and have extremely few more intricate structural motifs often found in large biological RNAs (Gevertz et al. 2005). Y. Li (&) Department of Biochemistry and Biomedical Sciences, McMaster University, 1280 Main Street West, Hamilton, ON L8S 4K1, Canada e-mail: liying@mcmaster.ca