Emergence of a dual-catalytic RNA with metal-specific cleavage and ligase activities: The spandrels of RNA evolution

Abstract

In vitro selection, or directed molecular evolution, allows the isolation and amplification of rare sequences that satisfy a functional-selection criterion. This technique can be used to isolate novel ribozymes (RNA enzymes) from large pools of random sequences. We used in vitro evolution to select a ribozyme that catalyzes a novel template-directed RNA ligation that requires surprisingly few nucleotides for catalytic activity. With the exception of two nucleotides, most of the ribozyme contributes to a template, suggesting that it is a general prebiotic ligase. More surprisingly, the catalytic core built from randomized sequences actually contains a 7-nt manganese-dependent self-cleavage motif originally discovered in the Tetrahymena group I intron. Further experiments revealed that we have selected a dual-catalytic RNA from random sequences: the RNA promotes both cleavage at one site and ligation at another site, suggesting two conformations surrounding at least one divalent metal ion-binding site. Together, these results imply that similar catalytic RNA motifs can arise under fairly simple conditions and that multiple catalytic structures, including bifunctional ligases, can evolve from very small preexisting parts. By breaking apart and joining different RNA strands, such ribozymes could have led to the production of longer and more complex RNA polymers in prebiotic evolution.The in vitro selection of functional RNA molecules from random sequence pools has successfully isolated far more classes of ribozymes than have yet been found in nature. The classic experiment (1) revealed an abundance of RNA ligase ribozymes in a random sampling of 10 15 unique sequences. The first experiments to probe in vitro evolution of ribozymes searched for improved or altered versions of the naturally occurring Tetrahymena group I ribozyme. For example, Green et al. (2) used in vitro evolution to restore and to improve activity of a pool of molecules based on the Tetrahymena group I intron, whereas Joyce and colleagues subsequently isolated variants of the group I ribozyme that performed subtly different tasks, such as cleavage of a DNA substrate (3) or acquisition of a novel dependency on a specific divalent metal cation (4). The ability of directed evolution and ribozyme engineering to evoke secondary activities of group I variants such as cleavage of aminoacyl ester or amide bonds (5-7; reviewed in ref. 8) suggests that RNA has an intrinsic capacity to adapt to novel contexts or substrates, although in these cases the acquisition of new catalytic activity was aided by careful design of the substrate to mimic the natural phosphodiester bond.In this paper we report the in vitro selection and evolution of a very small ribozyme that catalyzes the ligation of two substrate RNAs but also possesses the unexpected ability to undergo a separate self-cleavage reaction. Substitution of a divalent metal ion in the absence of further sequence evolution triggers the switch from selected ligase to unselected selfcleaving RNA. The fa...

Metal Ion Requirements for Structure and Catalysis of an RNA Ligase Ribozyme

“…The differences between the rates in the presence of different metal ions correlates with the difference between their pK a 's, such that the rate with Mn 2+ (pK a ) 10.6) is 6-10-fold higher than with Mg 2+ (pK a ) 11.4), and the rate with Mg 2+ is 13-fold higher than with Ca 2+ (pK a ) 12.9). Likewise, a 2′,5′ ligase ribozyme, which was selected in vitro and performs a reaction similar to the class I ligase, demonstrates the same relationship between its activity and the pK a of the metal ion (51). Most in vitro-selected and natural ribozymes, however, tend to show more complex behavior.…”

“…The ligase not only performs the chemistry of RNA polymerization, a reaction crucial to the RNA world hypothesis, but it is also one of the fastest ribozymes. Under optimal conditions, selfligation proceeds at a rate of 800 min -1 , comparable to that of the RNA component of RNase P (360 min -1 ) (61), and both ribozymes have rates 10 times faster than the observed rates of most other ribozymes (17,31,51,(62)(63)(64). However, the optimal conditions for these ribozymes differ from each other and from physiological conditions.…”

The effect of cytidine on the structure and function of an RNA ligase ribozyme

“…There are several known ribozymes that catalyze the template-directed ligation of oligonucleotide substrates+ These include naturally occurring ribozymes, such as the hammerhead and hairpin (Hegg & Fedor, 1995;Hertel & Uhlenbeck, 1995), and ribozymes that were obtained through in vitro evolution (Bartel & Szostak, 1993;Hager & Szostak, 1997;Jaeger et al+, 1999;Landweber & Pokrovskaya, 1999;Robertson & Ellington, 2000)+ Most in vitro evolution experiments have focused on RNA molecules that catalyze attack of either the 29-or 39-hydroxyl of an oligonucleotide substrate on the 59-triphosphate of the ribozyme, forming a 29,59-or 39,59-phosphodiester linkage, respectively+ The most proficient and best studied of the in vitro evolved ligase ribozymes is the class I motif (Ekland et al+, 1995)+ It forms a 39,59-phosphodiester linkage with a catalytic rate of ;100 min Ϫ1 + The class I ligase has been adapted to catalyze the template-directed polymerization of nucleoside triphosphates (Ekland & Bartel, 1996) and has been made to undergo continuous in vitro evolution (Wright & Joyce, 1997)+ The product of the first continuous in vitro evolution experiment was the E100 ligase+ This molecule contains 29 mutations relative to the class I ligase and is able to ligate a chimeric DNA/RNA substrate that contains the sequence of the T7 RNA polymerase promoter element (Wright & Joyce, 1997)+ The E100 ligase was used as a starting point to develop a ligase ribozyme that completely lacks cytidine (Rogers & Joyce, 1999)+ Cytidine was chosen for elimination because of its tendency to undergo spontaneous deamination to uridine, which has led some to suggest that cytidine was not present in the first genetic material (Levy & Miller, 1998)+ In developing the cytidinefree ligase, the rate of cytidine deamination was greatly accelerated by treatment with sodium bisulfite and CTP was withheld from the in vitro transcription mixture+ After 24 successive rounds of in vitro selective amplification, a cytidine-free ligase was obtained that formed a 29,59-phosphodiester linkage with a catalytic rate of 0+01 min Ϫ1 + This is about 10 5 -fold faster than the uncatalyzed rate of reaction (Rohatgi et al+, 1996a), but about 10 3 -fold slower than the rate of the E100 ligase+ The cytidine-free ligase adopted a completely different secondary structure compared to the class I ligase and formed a 29,59 rather than a 39,59 linkage (Rogers & Joyce, 1999)+ The reduced catalytic rate and altered regiospecificity of the cytidine-free ligase might be a reflection of limited evolutionary opportunities that were available starting from the E100 ligase+ To evaluate this possibility, a second in vitro evolution experiment was carried out starting from a large pool of random-sequence, cytidine-free RNAs+ This resulted in the isolation...…”

Optimization and optimality of a short ribozyme ligase that joins non-Watson–Crick base pairings

“…Despite the fact that a canonical ligation junction was built into the constant regions of the original pool (Fig+ 1B), the L1 ligase, like many previously selected ligases (Chapman & Szostak, 1995;Cuenoud & Szostak, 1995;Ekland et al+, 1995;Hager & Szostak, 1997;Landweber & Pokrovskaya, 1999), generated a ligation template or guide sequence from its random sequence core (Fig+ 1A)+ Pairings between the template and the sequences to be ligated are primarily canonical+ However, at the ligation junction itself there are apparently two adjacent, non-Watson-Crick pairings, a U:G base pair stacked on a G:A base pair (Fig+ 1A)+ This was unusual, as a variety of previously selected ribozyme and deoxyribozyme ligases (Chapman & Szostak, 1995;Cuenoud & Szostak, 1995;Ekland et al+, 1995;Hager & Szostak, 1997) have all exhibited canonical WatsonCrick pairings at their junctions (however, it should be noted that no pairings, Watson-Crick or non-WatsonCrick, are proposed for the 59 side of the ligation junction of the Class II Bartel ligase)+ Only one other ligase, selected by Landweber and Pokrovskaya (1999) appears to have a non-Watson-Crick pairing (a G:G pairing on the 59 side of the ligation junction)+ To determine if these apparent non-Watson-Crick pairings were important for ligase activity, they were altered to canonical base pairings (Fig+ 1C)+ Greatly reduced ligation activities were observed with these canonical pairings+ The requirement for noncanonical pairings at the ligation junction differs from requirements at other positions, as it was found that several Watson-Crick pairings between the ribozyme and its substrate could be substituted by other canonical pairings without significant loss of activity + However, it is interesting to note that another non-Watson-Crick pairing, a G:U base pair directly below the G:A base pair at the junction, cannot be changed to a G:C pairing without a 20-fold loss in activity (Fig+ 1C)+ It may be that the nonWatson-Crick stack of U:G, G:A, and G:U pairings forms a secondary structure that is particularly conducive to ligation+…”