Bioinformatic analysis of riboswitch structures uncovers variant classes with altered ligand specificity

Abstract

Riboswitches are RNAs that form complex, folded structures that selectively bind small molecules or ions. As with certain groups of protein enzymes and receptors, some riboswitch classes have evolved to change their ligand specificity. We developed a procedure to systematically analyze known riboswitch classes to find additional variants that have altered their ligand specificity. This approach uses multiple-sequence alignments, atomic-resolution structural information, and riboswitch gene associations. Among the discoveries are unique variants of the guanine riboswitch class that most tightly bind the nucleoside 2′-deoxyguanosine. In addition, we identified variants of the glycine riboswitch class that no longer recognize this amino acid, additional members of a rare flavin mononucleotide (FMN) variant class, and also variants of c-di-GMP-I and -II riboswitches that might recognize different bacterial signaling molecules. These findings further reveal the diverse molecular sensing capabilities of RNA, which highlights the potential for discovering a large number of additional natural riboswitch classes.iboswitches are structured noncoding RNA domains that regulate gene expression in response to the selective binding of small-molecule or ion ligands. The discovery of numerous classes of riboswitches has helped reveal how RNAs can form exquisitely precise ligand-binding pockets using only the four common RNA nucleotides (1-4). Furthermore, each discovery links the riboswitch ligand to the protein products of the genes under regulation. Recent riboswitch findings have exposed unique facets of biology, such as the widespread molecular mechanisms that confer fluoride (5) or guanidine (6) resistance, that maintain metal ion homeostasis (7-9), and that control important bacterial processes such as sporulation, biofilm formation, and chemotaxis (10-14). Thus, the identification of additional riboswitch classes promises to offer insights into otherwise hidden biological processes and their regulation.Riboswitch variants have been reported previously, wherein the ligand-binding "aptamer" domain has mutated to accommodate a different metabolite or signaling compound. The identification of such RNAs provides rare opportunities to study how small changes in RNA sequence can lead to major changes in smallmolecule ligand affinity. There have been seven examples, either experimentally validated or suspected, of ligand specificity changes reported to date. These include guanine aptamer variations present in riboswitches for adenine (15) and 2′-deoxyguanosine (2′-dG) (16), c-di-GMP-I aptamer variations that result in riboswitches that bind the recently discovered bacterial signaling molecule c-AMP-GMP (13, 14), and coenzyme B 12 aptamer changes (17, 18) that yield riboswitches selective for aquocobalamin (19). Three additional ligand specificity changes are suspected. Namely, some molybdenum cofactor riboswitches appear to exploit an altered aptamer structure to selectively recognize tungsten cofactor (20), certain flavin mononu...

Riboswitch diversity and distribution

“…These variants appear to have adapted to sense thiamin (vitamin B 1 ) with an affinity similar to that of its natural phosphorylated derivatives (TMP and TPP). These findings, and similar observations with other riboswitch classes Kim et al 2007;Johnson et al 2012;Kellenberger et al 2015;Nelson et al 2015;Weinberg et al 2017), reveal that evolutionary forces can blur the lines between riboswitch classes through the acquisition of mutations within a ligand-binding pocket that can alter binding specificity.…”

“…This bioinformatics strategy has been used previously to reveal the existence of structural variants of preQ 1 -I, preQ 1 -II, and preQ 1 -III riboswitches (McCown et al 2014) and to uncover variants of guanine riboswitches that exhibit altered ligand specificities Mandal et al 2003;Kim et al 2007). Most recently, we have used a bioinformatics search pipeline, guided by the known atomic-resolution structures of riboswitches, to identify additional rare variants whose ligand-binding specificities have been altered (Weinberg et al 2017). …”

“…This problem was encountered for riboswitches such as adenine ) and 2 ′ -dG-I ) that are very similar in sequence and structure to guanine riboswitches. Similarly, riboswitches such as the c-AMP-GMP (Kellenberger et al 2015;Nelson et al 2015) and 2 ′ -dG-II (Weinberg et al 2017) classes remained hidden for many years after their parent classes had been discovered. Therefore, it is important to note that some bioinformatics hits assigned to a particular riboswitch class might actually represent unrecognized variants that have altered ligand specificity.…”

“…Only four examples of the 2 ′ -dG-I riboswitch class have been found, all of which occur in a single organism, Mesoplasma florum . Similarly, only 12 members of the distinct 2 ′ -dG-II class have been identified to date (Weinberg et al 2017), and all of these are present only in metagenomic sequence data.…”

“…This has occurred because there is a higher probability of encountering a member of a common riboswitch class versus a member of a rare class, regardless of whether the search strategy involved a genetics-based or a bioinformatics-based search. Some very rare riboswitch classes were discovered only because their representatives are very close derivatives of a more common riboswitch class, such as adenine or the 2 ′ -dG-I and -II riboswitches Weinberg et al 2017), all of which are variants of the vastly more common guanine riboswitch class initially discovered. Unfortunately, these close variants will remain hidden among the bioinformatics hits for a more prominent riboswitch class unless there are key distinctions, such as mutations to the ligand-binding core of the aptamer or gene associations that indicate a change in ligand specificity (Kellenberger et al 2015;Nelson et al 2015;Weinberg et al 2017).…”