|Home | About | Journals | Submit | Contact Us | Français|
Brae Burn, Glastonbury CT 06033
These authors contributed equally.
One mechanism by which ribozymes can accelerate biological reactions is by adopting folds that favorably perturb nucleobase ionization. Herein we used Raman crystallography to directly measure pKa values for the Ade38 N1-imino group of a hairpin ribozyme in distinct conformational states. A transition-state analogue gave a pKa value of 6.27 ± 0.05, which agrees strikingly well with values measured by pH-rate analyses. To identify the chemical attributes that contribute to the shifted pKa we determined crystal structures of hairpin ribozyme variants containing single-atom substitutions at the active site and measured their respective Ade38 N1 pKa values. This approach led to the identification of a single interaction in the transition-state conformation that elevates the base pKa >0.8 log units relative to the precatalytic state. The agreement of the microscopic and macroscopic pKa values and the accompanying structural analysis support a mechanism in which Ade38 N1(H)+ functions as a general acid in phosphodiester bond cleavage. Overall the results quantify the contribution of a single electrostatic interaction to base ionization, which has broad relevance for understanding how RNA structure can control chemical reactivity.
RNA plays a central role in several key biocatalytic reactions including non-coding RNA processing, protein translation, and RNA splicing1–6. Distinct strategies identified in the protein world have been posited to account for this catalysis, including electrostatic complementation and general acid-base catalysis7–10. These strategies require base ionization, which necessitates the generation of an imine with a pKa near neutrality10. However, unlike proteins, there are no chemical groups in RNA that ionize near physiological pH, requiring the evolution of microenvironments that favorably perturb pKa values11. Although methods have been developed for measuring nucleotide base pKa values in a ribozyme, no study has quantified how specific stereochemical interactions modulate pKa shifting12. These knowledge gaps hinder efforts to model RNA structure-function relationships, including rigorous computational treatment of folding and catalysis13. Here, we used Raman crystallography, X-ray diffraction, and mutagenesis to pinpoint how specific functional groups influence the pKa of an essential imino group in the active site of an RNA enzyme known as the hairpin ribozyme (HPRZ), which is a model system for studying the architecture and function of RNA-mediated phosphoryl transfer.
The HPRZ is a member of the autolytic, small-ribozyme class8. It is derived from the negative-polarity strand of the satellite tobacco ringspot virus where it processes concatenated transcripts to unit length14. Like other small ribozymes, its chemical mechanism involves nucleophilic attack of the O2′ group from position −1 upon the scissile phosphorus at position +1 (Figure 1A). Unlike larger selfsplicing ribozymes, the HPRZ does not require divalent ions for activity8, making it ideal for investigating nucleobase-assisted catalysis. Indeed, the HPRZ pH-rate profile shows an apparent pKa of 6.2 ± 0.2, consistent with a single ionizable group8,15.
A synthetic variant of the HPRZ was developed in our lab for single-atom probing and structure-function investigations (Figures 1B and S1, Supporting Information). Crystallographic analyses of this minimal construct in the context of precatalytic and transition-state analogues revealed Gua8 and Ade38 adjacent to the scissile bond (Figures 1B and 1C)16. A functional analysis in which Gua8 was made abasic caused an 850-fold reduction in cleavage without altering the pH-rate profile17. By contrast, abasic replacement of Ade38 caused a 14,000-fold activity loss and a three log unit shift of the apparent pKa toward the basic15. The subtle exchange of Ade38 for N1-deaza-Ade also resulted in complete loss of detectable activity18, whereas similar active-site modifications at nearby Ade9 and Ade10 had modest effects18. These data indicate that the N1 imino of Ade38 has a critical role in catalysis, and an elevated N1 pKa was inferred.
Because the Ade38 protonation state figures prominently in any proposed HPRZ mechanism, we previously used Raman crystallography to measure directly the N1 pKa of Ade38 in the context of an inert precatalytic analog (PCA) (Figure 1C). The results revealed a pKa of 5.46 ± 0.0419. Although this value is elevated relative to AMP control measurements of 3.68 ± 0.06, it did not agree with the apparent reaction pKa of 6.2 ± 0.2, which reports on transition-state proton transfer9. We hypothesized that an active-site conformation closer to the transition state would fine tune the Ade38 N1 pKa and bring the microscopic pKa into better agreement with the apparent pKa of the reaction. To test this possibility, we utilized a transition-state analogue (TSA) to restrain the non-bridging oxygens of the scissile bond into a geometry consistent with the proposed transition state20,21 (Figure 1D). Using Raman crystallography we measured a pKa of 6.27 ± 0.05 for N1 of Ade38 in the context of the TSA, which represents a 0.81 log unit increase over the PCA value (Figure 1E and Table 1). This microscopic pKa is identical to the apparent pKa of the reaction15, thereby providing direct evidence that the N1 imino of Ade38 is the source of rate-limiting proton transfer.
To understand the molecular basis for the ~1 log unit difference between the PCA and TSA ionization constants, we compared their corresponding structures, which were derived previously from crystals of the same space group as those used for Raman measurements in Figure 1E. Two main differences were apparent in the active site. Whereas the TSA exhibits hydrogen bonds between the pro-Rp oxygen equivalent and the N6 amine of Ade38, as well as the O5′ leaving group and the Ade38 N1 imine, these interactions are missing in the PCA (Figure 1D versus 1C). By contrast, both active sites showed interactions between one non-bridging oxygen of the scissile bond and the N2 amino group of Gua8. To determine the importance of hydrogen bonding between the pro-RP oxygen of the scissile bond and the N6 group of Ade38 for pKa shifting, we substituted a purine nucleotide at position 38 that is devoid of the exocylic amine, and the N1 pKa was measured. The results revealed a PCA with a pKa value of 5.00 ± 0.06 (Table 1 and Figure S2A). By contrast, the control pKa for 2′-deoxynebularine in which N6 is a hydrogen, was 2.92 ± 0.04 (Figure S3), indicating a 2.08 log unit shift toward the basic in the context of the HPRZ active site. Application of the latter shift to AMP gave a pKa of 5.76, which is comparable to the wildtype PCA pKa (Table 1). To assess how the Ade38Pur substitution influences the active site geometry, we determined its crystal structure (Table S1), which shows an active site conformation similar to wildtype (Figure 2A). The phosphate group at the scissile bond adopted two conformations; one is similar to wildtype, but the other is rotated such that the pro-RP oxygen cannot hydrogen bond with N2 of Gua8. Neither of these conformations supports a hydrogen bond between the O3′ and N1 of position 38, as seen for the wildtype.
We then asked how loss of the N6 amine at position 38 would affect its N1 pKa in the context of the TSA. The Ade38Pur variant produced a pKa of 4.88 ± 0.06 (Table 1 and Figure S2B), representing a 2.0 log unit shift toward the basic relative to wildtype. When this shift is applied relative to the controls, the Ade38 N1 pKa is 5.66, which is significantly less than the pKa of 6.27 ± 0.05 measured for the wildtype TSA (Table 1). To understand how the Ade38Pur variant influences the HPRZ active site in the context of the TSA, we determined its crystal structure (Table S1). The results showed that the pro-RP oxygen of the scissile bond alters its conformation to be within hydrogen bonding distance of the N1 and O6 groups of Gua8 (Figure 2B). This represents a significant loss of scissile bond localization, and documents the extent to which the interaction between the pro-Rp oxygen and the N6 amine of Ade38 contributes to imine pKa shifting. Our pKa measurements and structural results are consistent with reports of Ade38Pur activity loss ranging from 102 to 103-fold15,22.
We next asked whether the N2 group of Gua8 influences Ade38 ionization. Structures of the wildtype HPRZ indicated that hydrogen bonding occurs between the exocyclic amine of Gua8 and the scissile bond in both PCA and TSA conformations (Figures 1C and 1D). To assess the importance of these interactions we replaced Gua8 with inosine, which lacks an exocylic amine. In the context of the PCA, the N1 pKa for Ade38 was 5.36 ± 0.05 (Figure S2C), which is nearly identical to wildtype (Table 1). A prior PCA structural comparison of the Gua8Ino variant to wildtype revealed only minor structural changes localized to the scissile bond16 (Figure 2C) even though the hydrogen bond is absent between the pro-Rp oxygen and the N2 amine of Gua8.
By contrast, the Gua8Ino substitution in the context of the TSA had a significant impact on Ade38 ionization via an indirect effect that repositioned the scissile phosphate. Our results revealed a pKa of 5.45 ± 0.09, which represents a 0.82 log unit decrease compared to wildtype (Figure S2D and Table 1). We determined the crystal structure of this variant in the context of the TSA (Table S1) and compared it to wildtype. The superposition indicates very subtle changes in the overall structure but a significant change in the hydrogen-bonding pattern to the scissile bond. In particular, the pro-Sp oxygen – unable to form a hydrogen bond to N2 of Ino8 – forms a new hydrogen bond to N1 (Figure 2D). This change shifts the pro-RP oxygen location so that it is unable to hydrogen bond with N6 of Ade38. Consequently, the O5′ leaving group is misoriented so that it does not interact readily with the Ade38 imine, consistent with the >10-fold loss in kcat for the Gua8Ino variant23.
Here we have shown that the microscopic pKa of Ade38 N1, in the context of a transition state analogue, is identical to the apparent pKa measured for the cleavage reaction for the most active, four-way-helical junction form of the HPRZ15. This agreement provides strong evidence that the N1 imine of Ade38 is involved in rate-limiting proton transfer. Independent evidence that the protonated form of Ade38 is operative in catalysis is supported by substitution of the base with 2-fluoro adenosine (pKa < 1), which significantly reduces activity24. Simulated pHrate profiles that assumed a pKa-shifted Ade38 and an unperturbed Gua8 pKa appeared similar to experimental pH-rate profiles, supporting the plausibility of general acid/base catalysis10. Herein, we also visualized HPRZ active sites that retained functional characteristics of the overall reaction, thereby allowing us to draw stereochemical inferences about the mechanism. Significantly, TSA crystal structures containing either vanadate20,21 or a 2′,5′ linkage at the scissile bond20 – identical to that employed here – do not show water localized at the N1 imine of Ade38, as proposed in a prior N1-imino pKa analysis25. As such, our observations strongly support a general acid role for Ade38. Similar roles have been ascribed previously to base Cyt75 of the hepatitis delta virus ribozyme26 and Ade756 of the Varkud satellite ribozyme27. We anticipate that when TSA structures become available for other small ribozymes, pKa-shifting interactions will be exhibited that are similar to those observed here. Our experimental analysis also provides a quantitative result that should be useful for computational biologists seeking ways to benchmark pKa prediction algorithms.
We thank R. Spitale, D. Perrin, N. Walter, P. Bevilacqua and D. Herschlag for helpful discussions. We thank the staff of SSRL for assistance with data collection. Portions of this research were carried out at the Stanford Synchrotron Radiation Laboratory (SSRL), a facility operated by Stanford University on behalf of the U.S. DOE. The SSRL Structural Molecular Biology Program is supported by the DOE, Office of Biological and Environmental Research, and by the NIH/NCRR, Biomedical Technology Program, and NIGMS.
Financial support was provided in part by grants from the NIH to PRC (GM84120) and to JEW (GM63162).
Supporting Information. Methods, the diffraction table, and figures. This material is available free of charge via the Internet at http://pubs.acs.org.
The authors declare no competing financial interest.
This work is dedicated to Professor David B. McKay in honor of his 67th birthday.