The positions of G8 and A38 nucleobases in the active site resemble the orientation of the two histidine side chains in the active site of ribonuclease A, leading to the model that G8 and A38 mediate catalysis through a similar mechanism. Ribonuclease A provides a textbook example of concerted general acid base catalysis in which one histidine acts as a general base to activate nucleophilic attack by removing a proton from the 2' hydroxyl while a second histidine protonates the 5' oxygen leaving group to facilitate breaking the 5' oxygen-phosphorus bond. With an acid ionization constant near 6.5, significant fractions of histidine residues are in both protonated and deprotonated states so histidines are particularly adept at accepting and donating protons at neutral pH. Like ribonuclease A, hairpin ribozymes function well at neutral pH, exhibiting an apparent pKa between 6 and 7 (Fig. 2a). However, adenosine and guanosine undergo ionization only at pH extremes, at least as free nucleosides in solution, which seems to make them poorly suited for mediating proton transfer reactions at neutral pH.

We used 8-azaguanosine, a fluorescent nucleotide analog, to learn whether some feature of the ribozyme active site alters the ionization equilibria of adenosine or guanosine relative to their ionization behavior in solution and enhances their ability to serve as general acid base catalysts. 8-azaguanosine is analog of guanine displays a high fluorescent quantum yield when the N1 position is deprotonated and a low fluorescent quantum yield when N1 is protonated (Fig. 1b). The protonation-dependence of fluorescence emission intensity allows us to calculate ionization equilibria from changes in emission intensity with pH.

A hairpin ribozyme with 8-azaguanine in place of G8 displays full catalytic activity and cleaves with an apparent pKa value in the neutral range, similar to an unmodified ribozyme (Fig. 2 a,b). To distinguish among models that require protonated or deprotonated forms of G8, microscopic pKa values were determined for ionization of 8-azaguanosine substituted for G8 in the active site of fully functional ribozymes (Fig. 2c). Microscopic pKa values for deprotonation of 8 azaguanine in the hairpin ribozyme active site ranged from 9.2 to 10.4 under different salt conditions. These microscopic pKa values are about 3 units higher than the apparent pKa values determined from the pH dependence of self-cleavage kinetics. Thus, the increase in self-cleavage activity with increasing pH does not reflect deprotonation of G8 and the predominant form of G8 in the ribozyme active site is protonated at neutral pH. These results do not completely exclude a role in proton transfer, but a simple interpretation is that G8 functions in the protonated form, perhaps by donating hydrogen bonds that stabilize the transition state.

Our goal is to understand how RNAs fold into the correct three-dimensional structures inside cells. mRNAs have long been recognized as key intermediates in the transmission of information from the DNA sequence of genes to the amino acid sequence of proteins and noncoding RNAs are now known to perform several essential biological functions previously attributed to proteins. Although RNAs must adopt precise three-dimensional structures in order to perform these functions, RNAs tend to assemble into a mixture of properly folded and misfolded structures during assembly reactions carried out in vitro. We hope to learn how RNAs avoid misfolding and assemble into functional structures during biogenesis in vivo.

We have developed a way to use self-cleaving ribozymes embedded in chimeric mRNAs and noncoding RNAs to probe RNA assembly in living cells. Ribozyme self-cleavage provides a sensitive, quantitative signal that functional ribozyme structures have formed. We began by engineering ribozyme variants with flanking inserts that have the potential to form complementary base pairs with parts of the ribozyme sequence. The inserts are located either upstream or downstream of the ribozyme sequence so they will be transcribed either before or after the ribozyme. This allows us to examine how the sequential nature of RNA synthesis affects folding outcomes (Figure). Assembly of the alternative helix, AltH1, is incompatible with assembly of the ribozyme helix, H1, and prevents self-cleavage. If RNA helices assemble sequentially during transcription an upstream insert would inhibit ribozyme assembly more than a downstream insert. AltH1 and H1 also were designed with different numbers of stabilizing base pairs to learn how thermodynamic stability contributes to folding outcomes.

In the first set of ribozyme variants we examined, upstream or downstream AltH1 sequences with 10 base pairs were able to compete with assembly of H1 sequences with 8 base pairs. This result suggested that thermodynamic stability, and not the sequential nature of RNA synthesis, was the major determinant of folding outcomes in vivo. In vitro, an H1 helix with 8 base pairs dissociates very slowly, with a half time on the order of days, so this evidence that a downstream AltH1 can block assembly of an upstream H1 implied that some feature of the intracellular environment accelerates exchange between upstream and downstream structures.

In a second set of variants, H1 helices with 14 base pairs were combined with AltH1 helices with 12 base pairs. These changes slow helix dissociation rates even further and reverse the relative thermodynamic stability of H1 and AltH1 so that H1 helices are now more stable. Strikingly, assembly of these variants does not always produce the most thermodynamically favored outcome. As expected, significant cleavage does occur in the presence of a downstream insert, evidence that stabilizing H1 by adding base pairs can prevent competition from a shorter downstream 3'AltH1 that has lower kinetic and thermodynamic stability. However, the upstream 5'AltH1 also inhibits self-cleavage, even though H1 is longer and more stable, in contrast to previous evidence that the most thermodynamically stable structure predominates in vivo. Thus, ribozymes with 8 base pairs in H1 and 10 base pairs in AltH1 seem to reach thermodynamic equilibrium during intracellular assembly while ribozymes with 14 base pairs in H1 and 12 base pairs in 5'AltH1 seem to become trapped in an upstream 5'AltH1 with lower thermodynamic stability. Evidently, slow helix dissociation can trap RNAs in thermodynamically less favored structures in vivo but kinetic trapping requires much longer helices with much slower dissociation kinetics than expected from the behavior of similar RNA structures in vitro. Current efforts focus on understanding the molecular basis of accelerated exchange between adjacent helices during RNA biogenesis in living cells.