The pH Dependence of Hairpin Ribozyme Catalysis Reflects Ionization of an Active Site Adenine*

Understanding how self-cleaving ribozymes mediate catalysis is crucial in light of compelling evidence that human and bacterial gene expression can be regulated through RNA self-cleavage. The hairpin ribozyme catalyzes reversible phosphodiester bond cleavage through a mechanism that does not require divalent metal cations. Previous structural and biochemical evidence implicated the amidine group of an active site adenosine, A38, in a pH-dependent step in catalysis. We developed a way to determine microscopic pKa values in active ribozymes based on the pH-dependent fluorescence of 8-azaadenosine (8azaA). We compared the microscopic pKa for ionization of 8azaA at position 38 with the apparent pKa for the self-cleavage reaction in a fully functional hairpin ribozyme with a unique 8azaA at position 38. Microscopic and apparent pKa values were virtually the same, evidence that A38 protonation accounts for the decrease in catalytic activity with decreasing pH. These results implicate the neutral unprotonated form of A38 in a transition state that involves formation of the 5′-oxygen–phosphorus bond.

Understanding how self-cleaving ribozymes mediate catalysis is crucial in light of compelling evidence that human and bacterial gene expression can be regulated through RNA self-cleavage. The hairpin ribozyme catalyzes reversible phosphodiester bond cleavage through a mechanism that does not require divalent metal cations. Previous structural and biochemical evidence implicated the amidine group of an active site adenosine, A38, in a pH-dependent step in catalysis. We developed a way to determine microscopic pK a values in active ribozymes based on the pH-dependent fluorescence of 8-azaadenosine (8azaA). We compared the microscopic pK a for ionization of 8azaA at position 38 with the apparent pK a for the self-cleavage reaction in a fully functional hairpin ribozyme with a unique 8azaA at position 38. Microscopic and apparent pK a values were virtually the same, evidence that A38 protonation accounts for the decrease in catalytic activity with decreasing pH. These results implicate the neutral unprotonated form of A38 in a transition state that involves formation of the 5-oxygen-phosphorus bond.
Hairpin ribozymes (Hp Rz) 2 belong to one of several families of small self-cleaving RNAs that serve as useful models of RNA catalysis because they are relatively simple and amenable to chemogenetic analyses (1). The Hp Rz remains functional in the absence of divalent metals, relying exclusively on nucleotide functional groups for catalytic chemistry (2)(3)(4)(5). High resolution structures of the Hp Rz bound to transition state mimics show two active site purines, G8 and A38, positioned in a manner similar to the two histidines in RNase A that mediate general acid-base catalysis of the same reaction ( Fig. 1A) (6 -8), leading to the proposal that the Hp Rz uses the same concerted general acid-base mechanism (Fig. 1B) (9). A38(N1) is near the 5Ј-oxygen of the reactive phosphodiester, and A38(N6) lies within hydrogen bonding distance of the pro-R P non-bridging oxygen. Exogenous nucleobase rescue experiments confirmed that the amidine group of A38 interacts with the transition state, but its precise role remained unclear (10). In a general acid-base model, the cationic protonated form of A38 would act as a general acid during cleavage to donate a proton to the 5Ј-oxygen leaving group as the 5Ј-oxygen-phosphorus bond breaks. During the reverse ligation reaction, unprotonated A38 would accept a proton to activate the nucleophilic 5Ј-oxygen for attack on phosphorus as the 5Ј-oxygen-phosphorus bond forms. Cleavage and ligation rate constants both increase with increasing pH and exhibit the same apparent pK a values near 6.5 (3), consistent with the law of microscopic reversibility (11). The pH dependence of reaction kinetics can provide information about the identity of general acid-base catalyst when an apparent pK a value clearly corresponds to the microscopic pK a value for a particular active site functional group. However, no RNA functional groups exhibit protonation equilibria near pH 6.5, at least as free nucleotides in solution (12), although nucleotide functional groups are well known to shift in the context of specific RNA structures (13)(14)(15). In the case of the Hp Rz, however, no candidate for a general acid-base catalyst could be identified from comparison of apparent pK a values obtained from pHrate profiles and microscopic pK a values for protonation of particular RNA functional groups.
Considerable evidence implicates A38, but not G8, in the pH-dependent step of the reaction pathway. Substitution of A38 with an abasic linkage or with N 1 -deazaadenosine reduces activity by 10 4 -and 10 7 -fold, respectively (10,16). In Hp Rz with an abasic linkage in place of A38, pK a(app) values shift from 6 to 9 (10). Substitutions of A38 with nucleotide analogs that differ in intrinsic acidity have corresponding effects on the pH dependence of Hp Rz cleavage and ligation activity (10,17). Activity is restored to abasic A38 ribozymes with certain exogenous nucleobases that share the amidine group of adenine, and the pH dependence of chemical rescue reactions correlates with the intrinsic basicity of the rescuing nucleobase. In contrast, Hp Rz with an abasic linkage in place of G8 exhibit an 850-fold loss of self-cleavage activity, but the apparent pK a for the reduced activity remains near 6 (18). Our recent evidence that the microscopic pK a values are above 9 for 8-azaguanosine (8azaG) ionization in 8azaGHp Rz and do not correlate with the pK a(app) values below 7 obtained from 8azaGHp Rz pH-rate profiles also argues that the pH-dependent step in the Hp Rz reaction reflects the protonation state of A38 rather than G8 (19).
The general problem of kinetic equivalence limits the ability to distinguish among alternative catalytic mechanisms using kinetic data alone (12). In the case of the Hp Rz, the same increase in activity with increasing pH could be observed if donation of a proton to the 5Ј-oxygen by the protonated form of adenosine and withdrawal of a proton from the 2Ј-oxygen by hydroxide ion occurred in the transition state or if the reaction involved unprotonated adenosine alone (10,20).
Molecular dynamics simulations support a model in which the protonated form of A38 participates in catalysis as a general acid (21) and a model in which A38 both accepts a proton from the nucleophilic 2Ј-oxygen and donates a proton to the departing 5Ј-oxygen leaving group during cleavage (22). A38 might also facilitate catalysis through mechanisms other than general acid-base catalysis (1). Hydrogen bonding interactions between A38(N1) and the 5Ј-oxygen and between A38(N6) and the pro-R P oxygen might facilitate catalysis by positioning and orientating reactive groups in the trigonal bipyramidal geometry needed for an S N 2(P)-type in-line attack mechanism. Molecular dynamics simulations also support a model in which active site interactions provide electrostatic stabilization of the electronegative transition state without direct participation of RNA functional groups in proton transfer (23). Finally, the decrease in Hp Rz activity at low pH might just reflect the destabilization that is characteristic of all RNA structures at pH extremes that disrupt hydrogen bonds and might not be related to the catalytic mechanism at all. Information about the protonation state of A38(N1) in a functional ribozyme would help distinguish among these alternative possibilities.
We set out to gain a better understanding of how RNA functional groups can participate in catalysis and to distinguish among alternative roles for A38 in the Hp Rz mechanism by determining the protonation state of A38(N1) directly in a functional ribozyme. 8-Azaadenosine (8azaA) is an adenosine analog that exhibits high fluorescence emission intensity when N1 is unprotonated, whereas emission intensity is low when N1 is protonated ( Fig. 2A) (24). 8-Azaadenine differs from adenine only in having nitrogen in place of carbon at position 8 and maintains the Watson-Crick hydrogen bonding face of adenine (Fig. 2B). Moreover, Hp Rz with 8azaA in place of A38 retain full activity (25). We developed a novel method to place a single unique 8azaA at position 38 by using 8-azaadenosine 5Ј-mono-phosphate (8azaAMP) to prime in vitro transcription of a ribozyme fragment with A38 at the 5Ј-end (Fig. 3). Ligation of the 5Ј-8azaA38 RNA to an unmodified transcript containing the remaining ribozyme sequence produced a complete functional ribozyme.
Microscopic pK a values determined from pH-fluorescence profiles were virtually the same as pK a(app) values calculated from pH-rate profiles for the 8azaAHp Rz. Our results link A38(N1) protonation directly to the decrease in Hp Rz activity with decreasing pH. These results contradict the idea that the cationic protonated form of A38 is essential for catalysis. Instead, they point to a view of the Hp Rz transition state in which the neutral unprotonated form of A38 facilitates formation of the 5Ј-oxygen-phosphorus bond.

Preparation of 8azaAMP-For
Step 1 (Fig. 3, a- RNA Preparation-RNAs were prepared by bacteriophage T7 RNA polymerase transcription of linearized plasmid templates or of partially duplex synthetic oligonucleotide templates or by chemical synthesis (Dharmacon) as described previously (26,27). Plasmid templates encoding Hp Rz RNAs were prepared using conventional molecular cloning and PCR mutagenesis methods as described (26,27).
The self-cleaving Hp Rz variants used for these experiments assemble in the context of a four-way helical junction and have either 3 or 8 bp in the intermolecular H1 helix that forms between the 5Ј-and 3Ј-cleavage product RNAs (5ЈR and 3ЈP, respectively) (Fig. 4). Self-cleavage of Hp Rz with four-way helical junctions and 8 bp in the H1 helix forms a stable 8azaA5ЈR⅐3ЈP complex with an internal equilibrium that favors the ligated ribozyme relative to 5ЈR and 3ЈP under the conditions used for fluorescence experiments (26). In Hp Rz with just 3 bp in the H1 helix, self-cleavage products dissociate rapidly, preventing religation of bound products that could complicate interpretation of cleavage rate measurements (26,28). 8azaAR (Fig. 4, blue) was transcribed in reactions with 4 mM 8azaAMP and 4 mM each NTP. 8azaAR was ligated to the ⌬A⌬BHp48 RNA (Fig. 4, green) to generate the complete 8azaAHp Rz RNA (29). Before ligation, 40 M each ⌬A⌬BHp48 RNA and 8azaAR RNA were heated to 85°C and cooled to 15°C in the presence of 10 mM MgCl 2 . Ligation was carried out at 15°C for 18 h in reactions with 1 mM ATP and 2 units/l T4 RNA ligase 1. Unmodified ribozymes prepared through ligation exhibited virtually the same self-cleavage kinetics as ribozymes prepared as single transcripts. 8azaAHpm Rz RNAs were prepared in the same way as 8azaAHp Rz RNAs but with an inactivating Gϩ1A mutation.
Self-cleavage Kinetics-Self-cleavage rates were determined using ligation-chase assays with [ 32 P]pCp-labeled 3Ј-product RNAs and 5Ј-ribozyme RNAs as described (26). Assays were performed at 25°C in 50 mM buffers at various pH values   . Hp Rz RNAs. 8azaAHp is a Hp Rz with 8azaA substituted for adenosine at position 38 that was prepared by ligating 8azaAR (blue), an in vitro transcript with a 5Ј-terminal 8azaA, to ⌬A⌬BHp48 (green), an in vitro transcript prepared with unmodified NTPs. 8azaAHp Rz self-cleaves at the site indicated by the arrow to form a stable complex with a 5Ј-cleavage product RNA (8azaA5ЈR) and a 3Ј-cleavage product (3ЈP8). P8 and P3A are 3Ј-cleavage product analogs prepared through chemical synthesis that anneal with 8azaA5ЈR to form H1 helices with 8 or 3 bp, respectively. An 8azaAHp variant with 3 bp in the H1 helix forms through self-ligation with P3A and then cleaves to form an unstable complex that dissociates rapidly during ligationchase assays.
(PIPPS at pH 3-4.6, MES at pH 4.7-7, and HEPES at pH 7-8) with 10 mM MgCl 2 and 0.1 mM EDTA. Reported values are the mean of two or more independent experiments. Reported errors are standard deviations. Apparent pK a values were determined by fitting self-cleavage rates to Equation 1, and errors were obtained from least squares analyses of the fits.
Fluorescence Measurements-Fluorescence was measured using a SpectraMax M2e plate reader (Molecular Devices). 10 M 8azaAMP or 15 M 8azaAHp Rz RNA was prepared in 50 mM buffer (glycine at pH 1-2.4, PIPPS at pH 2.4 -4.6, MES at pH 4.7-7, HEPES at pH 7-8, or TAPS at pH 8 -9), 10 mM MgCl 2 , and 0.1 mM EDTA. RNAs were annealed by heating to 85°C for 1 min and slowly cooled to 25°C with the addition of MgCl 2 at 60°C. A 20-l aliquot was added to 384-well black microplates (Greiner). During excitation at 290 nm, emission spectra were recorded between 310 and 450 nm. Emission intensities for RNAs with adenosine rather than 8azaA at position 38 were subtracted from emission spectra for 8azaA RNAs to correct for nonspecific fluorescence and UV absorption.
pK a values were determined by fitting integrated fluorescence intensities to Equation 2, where F B and F BH are the normalized fluorescence emission intensities of unprotonated and protonated 8azaA, respectively, and K a is the acid dissociation constant for protonation of 8azaA in 8azaAMP and Hp Rz RNAs. Reported errors were obtained from least squares analyses of the fits.

RESULTS
Synthesis of 8azaAMP-8azaAMP was prepared using an in vitro enzymatic synthesis method beginning with the C5 phosphorylation of ribose by ribokinase (rbsK) to give ribose 5-phosphate, which was further phosphorylated at C1 by 5-phospho-D-ribosyl-␣-1-pyrophosphate synthetase (prsA), forming 5-phospho-D-ribosyl-␣-1-pyrophosphate (Fig. 3). Adenine phosphoribosyltransferase (aprT) efficiently coupled 8-azaadenine to 5-phospho-D-ribosyl-␣-1-pyrophosphate, forming 8azaAMP. The monophosphate was subsequently converted to 8azaATP by sequential phosphorylation of adenylate kinase (plsA) and creatine phosphokinase (ckmT) in good isolated yield. Due to the lack of a catalytically active 8azaATP pyrophosphatase, the second step of the synthesis involved essentially the reverse reaction to discharge the 8azaATP to 8azaAMP. Key features of this synthesis include the use of dATP as the phosphate donor and dATP being easily separated from the ribonucleotide analog by boronate affinity chromatography.
Hp Rz with 8azaA in Place of an Essential Active Site Adenosine-8azaAHp Rz is a hairpin ribozyme sequence variant that was designed to allow substitution of a single 8azaA for the critical adenosine at position 38 in the active site (Fig. 4). In our previous studies of 8azaG-substituted ribozymes, 8azaGHp Rz RNAs were prepared by transcription of a template with a unique cytosine in reactions with 8azaGTP in place of GTP and then ligating the 8azaG transcript to an unmodified transcript to give the complete ribozyme. We initially attempted to prepare a ribozyme with a unique 8azaA38 in the same way. However, ligation of three RNA fragments was required to make a complete Hp Rz RNA with a single 8azaA38 due to the presence of other important adenosines close to A38. Three-part ligations did not produce 8azaAHp Rz RNAs in sufficient yields to compensate for the lower quantum yield of 8azaA relative to 8azaG (Fig. 2). Therefore, we developed a new approach using 8azaAMP to place a unique 8azaA as the first nucleotide of an in vitro transcript, 8azaAR (Fig. 4, blue). Bacteriophage T7 RNA polymerase greatly prefers guanosine as the 5Ј-terminal nucleotide (30). To improve transcription efficiency, the uridine at position 39 in the wild-type ribozyme sequence was changed to guanosine. This change was not expected to affect self-cleavage activity (31) and increased transcription yields by Ͼ20-fold.
Optimal transcription conditions were identified in trials with different NTP, AMP, and 8azaAMP concentrations. The efficiency of 8azaAMP and AMP incorporation was monitored through reduction in [␥-32 P]ATP labeling. Overall transcription yields were monitored in parallel reactions with [␣-32 P]ATP. Optimized reactions with 4 mM each 8azaAMP, ATP, GTP, CTP, and UTP produced 32 transcripts per template. The transcription efficiency of reactions with 8azaAMP was somewhat lower than that of reactions with AMP, which produced 44 transcripts per template under the same conditions. These yields are comparable with yields obtained in conventional transcription reactions that included only NTPs (30).
5Ј-8azaAMP-terminated 8azaAR was ligated to a second unmodified transcript (Fig. 4, green) to form the complete ribozyme, giving 15 and 60% yields of 8azaAHp Rz and Hp Rz RNAs, respectively. The 5Ј-ends of transcripts that initiate with ATP would not undergo ligation. The 4-fold lower yields of 8azaAHp Rz RNAs relative to Hp Rz RNAs prepared through ligation of transcripts primed with AMP might reflect a larger fraction of transcripts with 5Ј-terminal ATP that do not undergo ligation, which might be higher in transcription reactions with 8azaAMP instead of AMP. Ligation efficiency might also be lower if 8azaA transcripts are poor substrates for T4 RNA ligase 1. As expected from early mutagenesis studies (31), the U39G Hp Rz exhibited virtually the same cleavage rate constants and pH dependence as Hp43, the parental Hp Rz characterized previously (26).
pH Dependence of 8azaAHp Rz Self-cleavage Activity-8azaAHp Rz exhibited ϳ10-fold lower activity than the same Hp Rz sequence with unmodified adenosine at position 38 (Fig.  5A). Both 8azaA-substituted and unmodified ribozymes exhibited pH-dependent self-cleavage kinetics that fit well to an equation that includes a variable for a single ionizable group. An apparent pK a value of 6 was determined from the pH dependence of self-cleavage kinetics of this Hp Rz sequence variant with unmodified adenosine at position 38. This value is similar to the pK a(app) values previously determined from the pH dependence of self-cleavage and ligation kinetics for other Hp Rz variants (10,18). An apparent pK a value of 4 was determined from the pH dependence of 8azaAHp Rz self-cleavage kinetics (Fig. 5A, black), a shift of 2 units in the acidic direction relative to the pK a(app) of unmodified Hp Rz (Fig. 5A, green). This shift is similar to the acidic shift in microscopic pK a values from 3.52 to 2.2 for N1 protonation in adenosine and 8azaA, respectively (24,32).
Microscopic pK a for N1 Protonation of 8azaA38 in the Hp Rz Active Site-pH-fluorescence profiles were measured for 8azaAMP and 8azaAHp Rz and 8azaAHpm Rz RNAs under the same conditions used for measuring self-cleavage kinetics. 8azaAHp Rz and 8azaAHpm Rz RNAs exhibited a 40-fold lower quantum yield than 8azaAMP. RNAs with 8azaG substitutions exhibited a similar level of fluorescence quenching relative to 8azaGTP (19,33). Quenching of 8-azapurine fluorescence in RNAs is likely attributable to orbital overlap with flanking nucleobases as characterized previously for 2-aminopurine in B-form helices (34 -36).
Microscopic pK a values for N1 protonation of 8azaA in 8azaAMP, 8azaAHp Rz, and 8azaAHpm Rz were calculated from the pH dependence of fluorescence emission intensities (Fig. 5B). The pK a value of 2.5 that was determined for 8azaAMP ionization is slightly more basic than the pK a value of 2.2 reported for 8azaA (24). A similar small basic shift is observed for other nucleotides relative to nucleosides, a shift that probably reflects the influence of the negatively charged phosphate (37).
RNA was not sufficiently soluble below pH 4.2 to permit measurements of fluorescence emission spectra. To learn how uncertainty in the determination of the fluorescence emission intensity of protonated 8azaA in RNA at low pH affected pK a calculations, we compared microscopic pK a values that were calculated by fixing or not fixing the minimum fluorescence emission intensity of fully protonated 8azaA at the value of 0.23 that we obtained from the pH dependence of 8azaAMP fluorescence across the full pH range. pK a values obtained from both calculation methods agreed within 0.2 pH units, with values calculated without fixing the minimum fluorescence emission intensity being slightly more basic. The reported fits and pK a values in Fig. 5B were obtained with the fixed value.
The microscopic pK a of 4.6 ( Fig. 5B, green) for ionization of 8azaA38 in 8azaAHp Rz RNA was shifted by 2.1 units in the basic direction relative to the pK a value of 2.5 obtained for 8azaAMP (Fig. 5B, black). Thus, some feature of the active site environment facilitates N1 protonation relative to the behavior of 8azaA in solution. 8azaA38 exhibited a similar but slightly more basic pK a shift of 2.5 units in 8azaAHpm Rz RNA (Fig. 5B,  red), evidence that the mutationally inactivated ribozyme retains the active site features that are responsible for the basic shift. These microscopic pK a values for 8azaA protonation in ribozyme RNAs are experimentally indistinguishable from the apparent pK a value of 4.0 measured for the pH-dependent step in the 8azaAHp Rz self-cleavage reaction.

DISCUSSION
The importance of nucleobase protonation states in RNA structure and function has been recognized for some time (13, 14, 38 -51). Nucleobase ionization properties are particularly important in catalytic RNAs, where nucleobases are thought to participate directly in proton transfer steps of catalytic chemistry (1). High resolution structures of Hp Rz complexes with transition state mimics place G8 and A38 near the reactive phosphate, positioned like the two histidines in RNase A that mediate general acid-base catalysis of the same reaction ( Fig.  1A) (6 -9), making the protonation state of these nucleobases the focus of considerable interest.
NMR with isotopically labeled RNAs and, more recently, Raman crystallography have been used to measure microscopic pK a values for protonation of N3 of pyrimidines and N1 of purines in catalytic RNAs (42,47,48,51). NMR analyses of A38 protonation in an isolated loop B domain of the Hp Rz gave a pK a value of 4.89 (40). Notably, the acid dissociation constant determined for N1 of A38 in loop B was intermediate between the values measured for other adenosines in the same RNA, evidence that adenosines in a variety of structural contexts exhibit different pK a values compared with adenosine in solution. Analysis of A38 protonation in a ribozyme inactivated by methylation of the nucleophilic 2Ј-oxygen using Raman crystallography revealed a pK a value of 5.46 (48). These values are somewhat lower than the pK a(app) values near 6 determined from the pH dependence of self-cleavage activity (10,18,19). However, the use of ribozyme fragments or ribozymes with inactivating modifications in these studies complicates the interpretation of these results in terms of nucleobase ionization equilibria in functional ribozymes. 8-Azapurines retain the Watson-Crick hydrogen bonding face of their natural counterparts and differ only slightly in intrinsic acidity at N1 (24), so ribozymes with 8azaG or 8azaA substituted for their natural analogs in Hp Rz remain functional (19,25,52).
We have reported the first application of 8azaA fluorescence to analyze a catalytically essential adenosine in a functional ribozyme. Application of this method required us to prepare 8azaAHp Rz RNA with 8azaA uniquely substituted for a single adenosine at position 38 in sufficient yields to give quantifiable emission spectra despite the low quantum yield of 8azaA. This was accomplished using a novel method for site-specific insertion of a modified nucleotide that involved ligation of two transcripts, one of which was primed with the 5Ј-monophosphate form of the modified nucleoside. T7 RNA polymerase transcription reactions primed with 8azaAMP gave relatively poor yields. However, reduced yields of 5Ј-monophosphate-terminated transcripts were balanced by the specificity of 8azaA incorporation that resulted from discrimination against 5Јtriphosphate-terminated transcripts during the subsequence ligation step that was used to prepare the complete ribozyme.
The activity of the Hp Rz with an 8azaA38 substitution was reduced by ϳ10-fold relative to an unmodified ribozyme. The reason for the moderate reduction in activity is unclear because the single N8-for-C8 change in 8azaA does not affect any interactions that are evident in high resolution structures (6 -8). However, deletion of A38 by substitution with an abasic linkage reduces Hp Rz self-cleavage activity by 14,000-fold (10), so 8azaA provides almost the same contribution to catalysis as its natural counterpart. The 8azaAHp Rz exhibited the same increase in self-cleavage activity with increasing pH that is characteristic of unmodified ribozymes (10,18,19).
The apparent pK a value determined from the pH dependence of 8azaAHp Rz self-cleavage kinetics was shifted by ϳ2 units in the acidic direction with respect to the apparent pK a(app) value of 6 determined for the unmodified Hp Rz. These shifts correspond to the differences in Brønsted acid strength between 8azaA, with a pK a value of 2.2, and adenosine, with a pK a of 3.52 for N1 protonation (24,32). Hp Rz with isoguanine in place of adenosine at position 38 showed a similar correspondence between the Brønsted acid strength of isoguanosine (53) and the pH dependence of reaction kinetics (10). Given the difference of 1.32 units in the Brønsted acid strength of 8azaA and adenosine (24,32), the pK a value of 4.6 that we determined for protonation of 8azaA38 in the Hp Rz active site corresponds to a pK a of 5.92 for protonation of A38 in an unmodified ribozyme. This microscopic pK a value is virtually identical to the apparent pK a value of 6 determined from the pH dependence of self-cleavage activity of the unmodified ribozyme. These strong correlations support the conclusion that the pHdependent step in the Hp Rz reaction pathway directly involves A38.
The molecular basis of the shift of nucleobase ionization constants in the active site of the ribozyme relative to their values in solution is not yet clear. High resolution structures of the Hp Rz active site reveal stacking interactions above and below A38 and hydrogen bonding interactions with virtually every functional group of A38 (6 -8) that might influence ionization equilibria. Most of these interactions are located within loop B. This might account for the similarity in microscopic pK a values between the functional Hp Rz and Hpm Rz, which is inactivated by a Gϩ1A mutation. A tertiary interaction between Gϩ1 in loop A and C25 in loop B is critical for interdomain docking (54,55). Although no bound metal cations were located near A38 in x-ray structures of transition state mimics, cations are required to stabilize the functional ribozyme structure (2)(3)(4)(5)56), and nonspecific electrostatic interactions might also influence A38 ionization.
The close correspondence between apparent pK a values for fluorescent and unmodified ribozymes and microscopic pK a values for ionization of N1 in 8azaA and by extension adenosine (24) shows that the pH-dependent step in Hp Rz depends directly on the protonation state of A38(N1). The neutral unprotonated form of A38 predominates at pH values above pH 7 that support optimal activity, and A38 protonation is responsible for the decrease in self-cleavage activity with decreasing pH. These results are not consistent with the conventional view of the Hp Rz mechanism that emphasizes the role of the cationic protonated form of A38 in donating a proton to the departing 5Ј-oxygen during 5Ј-oxygen-phosphorus bond breaking (Fig. 1B). Instead, these results support a view of the transition state that focuses on the 5Ј-oxygen-phosphorus bond formation step in which the neutral unprotonated form of A38 mediates general or specific base catalysis to activate the 5Ј-oxygen nucleophile for attack on the phosphorus (Fig. 6). Intriguingly, these results contrast with recent evidence that the cationic protonated form of an active site cytosine predominates at neu-FIGURE 6. A, model of the Hp Rz transition state in which A38 acts as a general base catalyst to activate the 5Ј-oxygen nucleophile and G8 acts as a general acid catalyst to protonate the departing 2Ј-oxygen during ligation. B, specific acid-base model of Hp Rz catalysis in which A38 and G8 facilitate proton transfer through water.
tral pH, where it is thought to serve as a general acid catalysis during self-cleavage of the hepatitis delta virus ribozyme (47). RNA functional groups clearly have the potential to participate in catalysis through a variety of mechanisms.
This study presents a novel method for determining microscopic pK a values for an essential adenosine in the active site of a functional Hp Rz. We applied this method to gain fundamental insight into how RNA functional groups can participate in catalysis. Our results likely have general implications for the mechanisms of other RNA enzymes that function without a requirement for divalent cation cofactors, such as the peptidyl transferase center of the ribosome. The ability to use the pHdependent fluorescence of a non-perturbing adenosine analog to monitor ionization states might also prove useful for dissecting structure-function relationships in other RNAs.