Activation of the glmS Ribozyme Nucleophile via Overdetermined

Aug 8, 2017 - Abstract Image. RNA enzymes, or ... Daniel D. Seith , Jamie L. Bingaman , Andrew J. Veenis , Aileen C. Button , and Philip C. Bevilacqua...
0 downloads 0 Views 1MB Size
Communication pubs.acs.org/biochemistry

Activation of the glmS Ribozyme Nucleophile via Overdetermined Hydrogen Bonding Jamie L. Bingaman,†,‡ Inanllely Y. Gonzalez,† Bo Wang,† and Philip C. Bevilacqua*,†,‡,§ †

Department of Chemistry, Pennsylvania State University, University Park, Pennsylvania 16802, United States Center for RNA Molecular Biology, Pennsylvania State University, University Park, Pennsylvania 16802, United States § Department of Biochemistry and Molecular Biology, Pennsylvania State University, University Park, Pennsylvania 16802, United States ‡

S Supporting Information *

remarkable ∼106-fold. Early studies supported a role for GlcN6P as the general acid in the ribozyme.6,7 We recently uncovered three additional roles for GlcN6P: alignment of the active site, stabilization of charge accumulation, and activation of the 2′-OH.4 However, the molecular mechanism by which the ribozyme and cofactor are responsible for activation of the 2′-OH remained unclear. Herein, we investigate the molecular mechanism of 2′-OH nucleophile activation in the glmS ribozyme through thio substitutions in the background of ribozyme variants and a GlcN6P analogue. Typically, thio substitutions at NBO atoms are used to test for divalent metal ion binding at these positions. However, we recently showed that the glmS holoribozyme does not bind divalent metal ions at the NBO atoms of the scissile phosphate and that normal thio effects observed in the holoribozyme are due to disruption of hydrogen bonding within the active site.4 Thus, thio effects in this study largely reflect changes in hydrogen bonding patterns within the active site. Crystal structures of the glmS holo- and aporibozymes show multiple potential hydrogen bonding interactions with each NBO atom of the scissile phosphate (Figure 1 and Table S1). We hypothesized that the 2′-OH nucleophile is activated through hydrogen bond competition for the pro-RP NBO, as this is the site where the cofactor binds. Thus, we expected that elimination of single hydrogen bond donors to the pro-R P NBO in the wild-type (WT) holoribozyme would induce an inhibitory 2′-OH to pro-RP NBO hydrogen bond. This would result in an inverse thio effect at the RP position due to breaking of the inhibitory hydrogen bond upon thio substitution. To test this model, we substituted adenine at G57, which contacts the pro-RP NBO (Figure 1). In-line probing (ILP), which provides nucleotide level information, shows little difference in hydrolysis in both the apo and holo backgrounds of G57A compared to that in the WT ribozyme, supporting the conclusion that a G57A change does not disrupt the overall glmS ribozyme structure (Figure S1). We also synthesized a deoxy analogue of GlcN6P that lacks the hydroxyl group at position 1 (Scheme S1 and Figure S2). The G57A holoribozyme and WT+1-deoxy-GlcN6P variants, which

ABSTRACT: RNA enzymes, or ribozymes, catalyze internal phosphodiester bond cleavage using diverse catalytic strategies. These include the four classic strategies: in-line nucleophilic attack, deprotonation of the 2′-OH nucleophile, protonation of the 5′-O leaving group, and stabilization of developing charge on the nonbridging oxygen atoms of the scissile phosphate. In addition, we recently identified two additional ribozyme strategies: acidification of the 2′-OH and release of the 2′OH from inhibitory interactions. Herein, we report inverse thio effects in the presence of glmS ribozyme variants and a 1-deoxyglucosamine 6-phosphate cofactor analogue and demonstrate that activation of the 2′-OH nucleophile is promoted by competitive hydrogen bonding among diverse ribozyme moieties for the pro-RP nonbridging oxygen. We conclude that the glmS ribozyme uses an overdetermined set of competing hydrogen bond donors in its active site to ensure potent activation and regulation by the cofactor. Nucleophile activation through competitive, overdetermined hydrogen bonding could be a general strategy for ribozyme activation and may be applicable for controlling the function of ribozymes and riboswitches in the laboratory.

T

he small, self-cleaving ribozymes catalyze scission of an internal phosphodiester bond through attack of a 2′-OH group on the adjacent phosphate, leaving 2′,3′-cyclic phosphate and 5′-OH termini.1,2 These ribozymes utilize multiple catalytic strategies to enhance their rates of cleavage. The Breaker lab articulated four catalytic strategies: in-line nucleophilic attack, deprotonation of the 2′-OH nucleophile, protonation of the 5′O atom, and stabilization of charge on the nonbridging oxygen (NBO) atoms.1,3 We recently identified two additional strategies: acidification of the 2′-OH and release of the 2′OH nucleophile from inhibitory interactions.4 Together, these six strategies allow ribozymes to self-cleave in an efficient and specific manner. The glmS ribozyme is the only known small, self-cleaving ribozyme that is also a riboswitch. In the precleaved state, the downstream gene is translated to produce a protein associated with cell wall biosynthesis in bacteria, during which glucosamine 6-phosphate (GlcN6P) is generated.5 GlcN6P in turn serves as the ligand for the ribozyme, regulating self-cleavage a © XXXX American Chemical Society

Received: July 13, 2017 Revised: August 1, 2017

A

DOI: 10.1021/acs.biochem.7b00662 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry

Figure 1. Scissile phosphate NBO atom contacts of glmS holo- and aporibozymes reveal similar architectures. Active-site architecture and NBO hydrogen bonding contacts of the glmS (A) holoribozyme [Protein Data Bank (PDB) entry 2NZ4]7 and (B) aporibozyme (PDB entry 3G8S).8 In both cases, the 2′-O-methyl substitution at the −1 base has been manually changed in PyMOL to a 2′-OH to reflect the WT ribozyme. The numbering of the GlcN6P cofactor positions is provided in panel A.

eliminate single hydrogen bond donors to the pro-RP NBO, were expected to induce the inhibitory 2′-OH to pro-RP NBO interaction and provide an inverse thio effect with the RP substrate. Previously, we reported large, normal thio effects (kO/kS) for RP, SP, and dithio substrates in the WT holoribozyme of 200 ± 50, 15 ± 5, and 9000 ± 3000, respectively (Table S2).4 A fraction cleaved versus time plot for the oxo, RP, SP, and dithio substrates in the G57A holoribozyme is provided in Figure 2A.

and G57A holoribozymes indicates that the G57A holoribozyme variant alleviates the WT holoribozyme normal thio effect stereospecifically at the RP position. The thio effect for the dithio substrate in the G57A holoribozyme has weakened compared to that in the WT holoribozyme, reflecting the product of the two single-thio substitutions. These observations suggest that in the background of G57A, the SP side of the active site is intact while the RP side is not as well aligned for reactivity as with the WT ribozyme. Overall, the thio effects at the RP and SP positions are additive and noncooperative in G57A. Slowing of the oxo substrate suggests that all hydrogen bond donors to the pro-RP oxygen are needed for potent activation of the nucleophile, but a single-nucleotide change does not lead to an inverse thio effect. The WT+1-deoxy-GlcN6P ribozyme data are provided in Figure 2B. In this background, the RP, SP, and dithio substrates still show normal thio effects of 7.0 ± 0.3, 26 ± 1, and 190 ± 10, respectively. The magnitudes of thio effects are diminished for the RP and dithio substrates, suggesting that the active site is not as well aligned for reactivity as with the unmodified cofactor. The dithio effect is again roughly the product of the two single-thio effects. While the decreased magnitudes of the RP and dithio thio effects support the importance of the 1-OH of the sugar in activation of the nucleophile, we again do not see an inverse thio effect for the RP substrate. Neither the G57A holoribozyme nor the WT+1-deoxyGlcN6P single variants, which both eliminate one hydrogen bond donor to the pro-RP NBO of the scissile phosphate, induced the hypothesized RP inverse thio effect. Apparently, the nucleophilic 2′-OH is not competing with only a single hydrogen bond donor. We thus revised our model for activation of the 2′-OH nucleophile to account for more than two potential hydrogen bond donors to the pro-RP NBO. In other words, elimination of one NBO hydrogen bond donor may not induce the inhibitory 2′-OH to NBO hydrogen bond if there is a third hydrogen bond donor close enough to interact with the NBO lone pair freed upon the single elimination. Analysis of the glmS holoribozyme crystal structure8 reveals three potential WT hydrogen bond donors to the pro-RP NBO: GlcN6P(O1), GlcN6P(N2), and G57(N2) (Table S1). We next tested the activity of the oxo and various thio substrates in the background of double-variant ribozymes involving these donors. To test the effect of double-hydrogen bond donor elimination at the pro-RP NBO, we first probed the thio effects

Figure 2. Single-variant ribozymes reveal non-thio or normal thio effects, while double-variant ribozymes reveal inverse thio effects. Oxo (black), RP (blue), SP (green), and dithio (purple) substrate reactivity for (A) G57A holo and (B) WT+1-d-GlcN6P, (C) G57A+1-dGlcN6P, and (D) G57A apo in 50 mM NaHEPES (pH 7.0) and 10 mM MgCl2 at 37 °C.

The thio effect for the RP substrate in the G57A holoribozyme is 1.08 ± 0.05 due to slowing of the oxo substrate. While the normal thio effect of 200 for the RP substrate in the WT holoribozyme is alleviated in the G57A holoribozyme background, the expected inverse thio effect is not observed. The SP and dithio substrates in the G57A holoribozyme background display normal thio effects of 12 ± 2 and 10.1 ± 0.4, respectively. The similar thio effect for the SP substrate in WT B

DOI: 10.1021/acs.biochem.7b00662 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry in a G57A+1-deoxy-GlcN6P double-variant background, where two WT hydrogen bond donors to the pro-RP NBO, GlcN6P(O1) and G57(N2), are eliminated (Figure 2C). In this double variant, we observe the anticipated stereospecific inverse thio effects for the RP and dithio substrates, which are sizable at 260 ± 20 and 180 ± 20, respectively, while the SP substrate has a non-thio effect of 1.1 ± 0.1 (Table S2). Again, the dithio effect is additive. This result suggests that elimination of two of the three possible hydrogen bond donors to the proRP NBO exposes one lone pair on the pro-RP oxygen, inducing an inhibitory 2′-OH to pro-RP NBO interaction. When a sulfur atom is substituted at the pro-RP position, the inhibitory 2′-OH interaction is disrupted, giving rise to inverse thio effects for the RP and dithio substrates. Next, we tested the effect of a G57A aporibozyme double variant wherein all three WT pro-RP hydrogen bond donors, GlcN6P(O1), GlcN6P(N2), and G57(N2), are eliminated (Figure 2D). In this double variant, we again observe a stereospecific inverse thio effect for the RP substrate, with a value of 12 ± 1, while the SP and dithio substrates have small normal thio effects of 1.5 ± 0.7 and 1.8 ± 0.2, respectively (Table S3). These observations suggest that when all three possible hydrogen bond donors to the pro-RP NBO are eliminated, at least one lone pair on the pro-RP NBO is exposed, allowing for an inhibitory interaction with the 2′-OH and leading to an inverse thio effect with the RP substrate. For the first time, the dithio substrate does not give rise to an additive effect. Thus, upon deletion of all three WT hydrogen bond donors, the effects of sulfur substitution at the pro-RP and pro-SP positions become cooperative.9 To test whether appearance of the inhibitory 2′-OH hydrogen bonding interaction and the associated inverse thio effect are specific to the pro-RP NBO, single and double variants involving the pro-SP NBO atom were investigated, including G32A holoribozyme, G32A+1-deoxy-GlcN6P, and G32A aporibozyme variants (Figure S3). The G32A holoribozyme single variant eliminates just one hydrogen bond donor to the pro-SP NBO; the G32A+1-deoxy-GlcN6P double variant eliminates one hydrogen bond donor to each NBO, and the G32A aporibozyme double variant eliminates two hydrogen bond donors to the pro-RP NBO and one to the pro-SP NBO. We note that the G32A change has only minor structural effects on the ribozyme in the holo and apo backgrounds according to ILP (Figure S1). Neither the G32A holoribozyme single variant nor the G32A+1-deoxy-GlcN6P double variant leads to inverse thio effects, suggesting that the inhibitory 2′-OH hydrogen bonding interaction is specific to double hydrogen bond donor elimination at the pro-RP NBO (Figure S3 and Table S2). This conclusion is further supported by findings for the G32A aporibozyme, which gives inverse thio effects for the RP and dithio substrates (Table S3). Thio effects for the various singleand double-variant ribozymes are summarized in Figure 3 and Figure S3D. The results herein lead to a new model for 2′-OH activation depicted in Figure 4. According to this model, the WT (oxo) holoribozyme is activated by competitive hydrogen bonding for the pro-RP NBO by three hydrogen bond donors: GlcN6P(O1), GlcN6P(N2), and G57(N2) (depicted with green dashed lines in Figure 4A). The existence of simultaneous hydrogen bonds between the pro-RP oxygen and any two of these three donors efficiently precludes hydrogen bond formation with the 2′-OH nucleophile, releasing it for proton abstraction by the general base. This overdetermined network

Figure 3. Inverse thio effects for G57A apo and G57A+1-deoxyGlcN6P variant ribozymes. RP (blue), SP (green), and dithio (purple) thio effects for WT holo,4 WT apo,4 G57A holo, WT+1-deoxyGlcN6P, G57A+1-deoxy-GlcN6P, and G57A apo in 50 mM NaHEPES (pH 7.0) and 10 mM MgCl2 at 37 °C. The color of the variant indicates the NBO atom with hydrogen bond donor elimination. The number in parentheses indicates the number of donors eliminated.

Figure 4. Model for nucleophile activation through overdetermined hydrogen bonding. (A) The WT holoribozyme exhibits an activated conformation due to three possible hydrogen bond donors to the proRP oxygen. (B) Aporibozyme, (C) G57A+1-deoxy-GlcN6P, and (D) G57A aporibozyme variants exhibit inactivated conformations because of fewer than two hydrogen bond donors to the pro-RP oxygen, allowing an inhibitory 2′-OH to pro-RP hydrogen bond. Panels B−D are associated with inverse thio effects.

of hydrogen bonds is essential for potent activation of the 2′OH nucleophile as any of the single variants leads to strong C

DOI: 10.1021/acs.biochem.7b00662 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry

revealed by the general additivity of the thio substitutions. This suggests the presence of an overdetermined active site in the glmS ribozyme with a noncooperative, additive network of hydrogen bonds that supports specificity,13 also consistent with the similarity of holo- and aporibozyme crystal structures. This extensive network of hydrogen bonds prevents interaction of the 2′-OH nucleophile with the scissile phosphate in the holo state and assures a maximal reaction rate. Such a rigid active site may be critical for regulating the ribozyme through hydrogen bond competition, thus avoiding leaky reaction in the apo state. The overdetermined set of hydrogen bond donors is akin to overengineering. It will be of interest to see the extent to which nature regulates the activity of other ribozymes and riboswitches through overdetermined networks of hydrogen bonds, and the extent to which overengineering of hydrogen bonding can be exploited in the design of RNA switches.

inhibition of the oxo substrate. In the WT aporibozyme, which has an inverse thio effect at the pro-RP position,4 two of three potential hydrogen bond donors at the pro-RP oxygen are eliminated. This leaves one pro-R P oxygen lone pair unoccupied, allowing it to hydrogen bond with the 2′-OH nucleophile (depicted with a red dashed line in Figure 4B). Similarly, in the G57A+1-deoxy-GlcN6P double variant, which also gives an inverse thio effect at the pro-RP position, two of three potential hydrogen bond donors at the pro-RP oxygen are abolished. This again leaves one pro-RP oxygen lone pair to hydrogen bond with the 2′-OH nucleophile (depicted with a red dashed line in Figure 4C). Notably, this is the first inverse thio effect for the glmS ribozyme reported in the presence of a cofactor, albeit a variant one. Deletion of all three potential hydrogen bond donors involving the pro-RP oxygen in the G57A aporibozyme also leads to an inverse thio effect at the pro-RP position due to the release of both pro-RP oxygen lone pairs. This again allows for the inhibitory 2′-OH nucleophile to pro-RP NBO interaction (depicted with a red dashed line in Figure 4D). Inverse thio effects in the glmS ribozyme occurred at the proRP NBO position, but it is conceivable that an inhibitory 2′-OH to NBO interaction could occur at a pro-SP position. Indeed, we reported SP-specific inverse thio effects in the HDV ribozyme in the presence of low-charge density ions.10 Thus, ribozymes can exhibit interaction of the 2′-OH with the pro-RP or pro-SP NBO atoms depending on ribozyme architecture. For both the glmS and HDV ribozymes, inverse thio effects occurred in less reactive variants or conditions suggesting that WT ribozymes may prevent spurious 2′-OH nucleophile hydrogen bonds to optimize nucleophilic attack. To gain more insight into the preference of the 2′-OH for the pro-RP and pro-SP NBO positions, we investigated the occurrence of the inhibitory 2′-OH to NBO interaction within a chimeric oligonucleotide system, which lacks the preorganized and crowded active site of a ribozyme and whose cleavage is known to be imidazole-catalyzed11 (Figure S4 and Table S4). We included NH4Cl in these reactions to provide a cationic species without promoting interaction of the metal ion with the NBO atoms, as might be true with divalent metal ions. We added PEG8000 to help mimic the crowded nature of a ribozyme active site and promote possible formation of an inhibitory 2′-OH to NBO atom hydrogen bond that could be alleviated by thio substitution. However, the RP, SP, and dithio model oligonucleotides did not give rise to inverse thio effects (Figure S4 and Table S5). The occurrence of inverse thio effects in natural ribozymes but not model RNA oligonucleotides suggests that an inhibitory 2′-OH to NBO interaction is promoted by the compact and preorganized nature of an active site. One role for the two hydrogen bonding states of the 2′-OH depicted in Figure 4 is pKa tuning. We previously reported that the pKa of the 2′-OH in the glmS ribozyme has a remarkable value of 20 when it is hydrogen bonded to a NBO atom, which is ∼5 pKa units higher than the solution pKa of ∼15.12 Thus, the prevention of this inhibitory 2′-OH to NBO interaction through competitive hydrogen bonding could acidify the pKa of the 2′-OH by at least 5 pKa units, helping to activate the nucleophile. These results suggest that the intricate hydrogen bonding networks that exist within ribozyme active sites may be crucial for selective activation and regulation of ribozymes. The active site of the glmS ribozyme is robust. Single-atom changes did not compromise the integrity of the active site, as



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biochem.7b00662. Experimental details; supporting text; mass spectrometry of oligonucleotides; data tables; in-line probing; 1H NMR, 13C NMR, and 31P NMR spectra of synthesis compounds; single- and double-variant ribozyme plots involving pro-SP NBO contacts; and model oligonucleotide plots (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Philip C. Bevilacqua: 0000-0001-8074-3434 Funding

This work was supported by National Science Foundation Grant CHE-1213667 (J.L.B. and P.C.B.). Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Erica Frankel and Kyle Messina for assistance with fast reaction time points and Ken Feldman for helpful discussions.



REFERENCES

(1) Breaker, R. R., Emilsson, G. M., Lazarev, D., Nakamura, S., Puskarz, I. J., Roth, A., and Sudarsan, N. (2003) RNA 9, 949−957. (2) Cochrane, J. C., and Strobel, S. A. (2008) Acc. Chem. Res. 41, 1027−1035. Ferré-D’Amaré, A. R., and Scott, W. G. (2010) CSH Perspectives in Biology, Vol. 2. (3) Emilsson, G. M., Nakamura, S., Roth, A., and Breaker, R. R. (2003) RNA 9, 907−918. (4) Bingaman, J. L., Zhang, S., Stevens, D. R., Yennawar, N. H., Hammes-Schiffer, S., and Bevilacqua, P. C. (2017) Nat. Chem. Biol. 13, 439−445. (5) Winkler, W. C., Nahvi, A., Roth, A., Collins, J. A., and Breaker, R. R. (2004) Nature 428, 281−286. Collins, J. A., Irnov, I., Baker, S., and Winkler, W. C. (2007) Genes Dev. 21, 3356−3368. (6) Klein, D. J., and Ferré-D’Amaré, A. R. (2006) Science 313, 1752− 1756. Viladoms, J., and Fedor, M. J. (2012) J. Am. Chem. Soc. 134, 19043−19049.

D

DOI: 10.1021/acs.biochem.7b00662 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry (7) Cochrane, J. C., Lipchock, S. V., and Strobel, S. A. (2007) Chem. Biol. 14, 97−105. (8) Cochrane, J. C., Lipchock, S. V., Smith, K. D., and Strobel, S. A. (2009) Biochemistry 48, 3239−3246. (9) Moody, E. M., and Bevilacqua, P. C. (2003) J. Am. Chem. Soc. 125, 2032−2033. Moody, E. M., and Bevilacqua, P. C. (2003) J. Am. Chem. Soc. 125, 16285−16293. (10) Thaplyal, P., Ganguly, A., Hammes-Schiffer, S., and Bevilacqua, P. C. (2015) Biochemistry 54, 2160−2175. (11) Anslyn, E., and Breslow, R. (1989) J. Am. Chem. Soc. 111, 4473− 4482. (12) Zhang, S., Ganguly, A., Goyal, P., Bingaman, J. L., Bevilacqua, P. C., and Hammes-Schiffer, S. (2015) J. Am. Chem. Soc. 137, 784−798. (13) Moody, E. M., Feerrar, J. C., and Bevilacqua, P. C. (2004) Biochemistry 43, 7992−7998.

E

DOI: 10.1021/acs.biochem.7b00662 Biochemistry XXXX, XXX, XXX−XXX