Small Molecule Rescue and Glycosidic Conformational Analysis of the Twister Ribozyme

Abstract

The number of self-cleaving ribozymes has increased sharply in recent years giving rise to elaborations of the four known ribozyme catalytic strategies α, β, γ, and δ. One such extension is utilized by the twister ribozyme, which is hypothesized to conduct δ, or general acid catalysis, via N3 of the syn adenine +1 nucleobase indirectly via buffer catalysis at biological pH and directly at lower pH. Herein, we test the δ catalysis role of A1 via chemical rescue, and the catalytic relevance of the syn orientation of the nucleobase by conformational analysis. Using inhibited twister ribozyme variants with A1(N3) deaza or A1 abasic modifications, we observe chemical rescue effects greater than one hundred-fold in the presence of protonatable biological small molecules such as imidazole and histidine, similar to observed rescue values previously reported for C75U/C76Δ in the HDV ribozyme. Brønsted plots on the twister variants support a model in which small molecules rescue catalytic activity via a proton transfer mechanism, suggesting that A1 in the wild-type is involved in proton transfer, most likely general acid catalysis. Additionally, through glyosidic conformational analysis in an appropriate background that accommodates the bromine atom, we observe that an 8BrA1-modified twister ribozyme is up to 10-fold faster than a non-modified A1 ribozyme supporting crystallographic data that A1 is syn when conducting proton transfer. Overall, this study provides functional evidence that the nucleotide immediately downstream of the cleavage site participates directly or indirectly in general acid-base catalysis in the twister ribozyme while occupying the syn conformation.

Graphical Abstract

INTRODUCTION

Small self-cleaving RNA enzymes (ribozymes) are a class of RNAs that catalyze the cleavage of a specific phosphodiester bond.^1,2 Ribozyme catalytic strategies can be categorized under four general types: in-line arrangement of the 2’O-P-5’O atoms (α catalysis), neutralization of the negative charge building of the scissile phosphate’s nonbridging oxygen atoms (β catalysis), activation of the 2’-OH nucleophile (γ catalysis), and stabilization of the 5’O leaving group (δ catalysis).^3–5 Most ribozymes use multiple of these catalytic strategies, with some using all four strategies at once.^3–6

The twister ribozyme is one of the most active self-cleaving ribozymes, with a half-life measured in the seconds under biological conditions.^6,7 Due to its high catalytic activity, the twister ribozyme was initially suspected to use multiple catalytic strategies to conduct its rapid cleavage activity.⁶ Numerous studies have investigated the catalytic mechanism of the twister ribozyme finding that it likely uses all four catalytic strategies; nonetheless, uncertainty remains toward key features about the mechanism.⁸ Several studies support that the twister ribozyme conducts α catalysis; β catalysis via N2 of G55 (env9 numbering); and γ catalysis via N1 of the same guanine.^9–12

The mechanism of δ, or general acid, catalysis for the twister ribozyme is in need of further investigation. Crystal structures consistently indicate that the N3 of A1 is appropriately positioned for general acid catalysis,^9–11,13 and the importance of this N3 is supported by several mechanistic and mutagenic studies that show a severe loss in activity with a 3-deazaadenine modification.^11,12 The importance of A1(N3) to catalysis was established by Wilson et al. (2016) who showed that an N3C modification was inactivate. Via a set of rigorous mechanistic experiments, they associated this atom with general acid catalysis for cleavage.¹² However, NMR experiments with the env22 twister ribozyme determined an experimental pK_a of just 5.1 for A1, nearly 2 pK_a units lower than the experimental pK_a of ~6.9 observed for self-cleavage activity from several pH-rate profile analyses.^7,9,11 Thus, A1(N3) may be an effective general acid for direct proton transfer only below physiological pH.⁷

Our lab recently provided evidence that the twister ribozyme engages in a three-channel mechanism, with different roles proposed for A1(N3) in various channels and fractional channel flux dependent on solution pH and Mg²⁺ concentration (Scheme 1).⁷ We observed that at biological Mg²⁺ concentrations (0.5 and 1.0 mM)¹⁴ and sub-biological pH (below 7.0), catalytic activity was buffer-independent.⁷ However, upon raising the pH to biological or greater values (7.5 to 9.0), catalytic activity became dependent upon buffer concentration and identity. Moreover, when the Mg²⁺ concentration was raised to 10 mM, twister was buffer dependent-throughout the entire pH range of 5.5 to 9.0. To account for this behavior, we proposed a three-channel mechanism comprised of one buffer-independent and two buffer-dependent channels (Scheme 1C). In the buffer-independent channel, the general acid appears to be the protonated N3 of A1, while in the buffer-dependent channel, the general acid appears to be a protonated small molecule that conducts indirect proton transfer possibly through a proton shuttle to the 5’-O leaving group involving the N3 of A1 (Scheme 1A and 1B). While this multi-channel mechanism is consistent with several studies,^7,11,12 the proton transfer and conformational role of A1 in the various channels is not fully established. Herein, we test the contribution of N3 of A1 in general acid catalysis via Brønsted analysis of small molecule rescue and glycosidic conformational analysis of its syn orientation.

Scheme 1.

Mechanistic Schemes for the Twister Ribozyme. (A) Indirect and (B) Direct Protonation Mechanisms and (C) Multi-Channel Mechanism.

Figure 1.

Kinetic traces of 3dzA1 experiments show that exogenous small molecules rescue activity. Experiments were performed at 23 °C with 10 mM MgCl₂ at pH 7.5 in the presence of either 30 mM Tris (circle), 180 mM histidine/7.5 mM HEPES (square) or 240 mM imidazole/10 mM HEPES (diamonds). The k_obs values were determined to be (2.0 ± 0.1 h⁻¹) × 10⁻⁴, (25 ± 2 h⁻¹) × 10⁻⁴ and (130 ± 10 h⁻¹) × 10⁻⁴ for the Tris, histidine and imidazole trials, respectively. Data were fit to either a straight line (Tris) or a single exponential (Equation 1, histidine and imidazole). Values are for individual trials and errors are from fitting. Average k_obs values for multiple trials are presented in Table 1.

Our study provides evidence that the A1 nucleotide of the twister ribozyme assists in the catalytic mechanism via general acid catalysis while in a syn conformation. The role of A1 in general acid-base catalysis is explored via a chemical rescue approach whereby two inhibited twister constructs, one lacking the N3 of A1 and the other lacking the entire adenine base, are incubated with protonatable small molecules to test for rescue effects. Additionally, the catalytic conformation of A1 is investigated via glycosidic conformational analysis whereby an 8Br-modified adenine is used to pre-structure the A1 nucleobase in the syn conformation.

MATERIALS AND METHODS

Reagents for Kinetics Experiments.

For chemical rescue experiments with small molecules, buffers were prepared at 2–5X concentration depending on small molecule solubility. Solutions contained either MES (pH 5.5) or HEPES (pH 6.5, 7.5 or 8.2) buffer and one small molecule including adenosine, 1,2-dimethylimidazole, 2-ethylimidazole, imidazole, 2-isopropylimidazole, L-glycine, L-histidine, L-lysine, 2-methylimidazole, or pyrazole. The pH of buffers was measured and adjusted using a room temperature-calibrated (~23 °C) pH probe, with buffer and small molecules dissolved in solution. After pH adjustment, all solutions were filter sterilized using a 0.2 μM filter. Buffers were used at final concentrations of 10–20 mM MES or 5–10 mM HEPES, while small molecules were used at final concentrations of 120–240 mM. An exception was the HEPES/adenosine buffer, which was prepared at a 2X concentration and used at a final concentration of 5 mM HEPES with 10 mM adenosine owing to low solubility of adenosine. Experiments containing only HEPES or only Tris were used at final concentrations of 5, 30 or 250 mM. For the experiments at 37 °C, the pH was adjusted using a room temperature-calibrated pH probe.

For the pH-rate profiles with 8BrA1 substrate and the wild-type enzyme strand, buffers were prepared at 0.5 M stocks and included MES (pH 5.5), HEPES (pH 6.5), and Tris (pH 7.5–8.5). The pH of the buffers was measured using a room temperature-calibrated pH probe. After pH adjustment, all buffer solutions were filter sterilized using a 0.2 μM filter.

Twister Enzyme Strand Preparation.

The two-piece construct used in kinetic assays, consisting of an enzyme and substrate, was based on a previously characterized twister construct (Figure S1).⁶ The wild-type and A38AA twister enzyme strands were prepared via in vitro transcription from a hemi-duplex DNA template (IDT) with a top-strand (forward) T7 promoter sequence and a bottom-strand template:

Top-Strand Primer: 5’-TAA TAC GAC TCA CTA TAG G −3’

Wild-Type Enzyme Bottom-Strand DNA Template: 5’-CGC GAC ATT ACT CTG CTA TTT TTG CGG GCT TGT AAC CGC TTT ATT GCC CCT ATA GTG AGT CGT ATT A-3’

A38AA Enzyme Bottom-Strand DNA Template: 5’-CGC GAC ATT ACT CTG CTA TTT TTG CGG GCT TTG TAA CCG CTT TAT TGC CCC TAT AGT GAG TCG TAT TA-3’

The one-nucleotide difference between the two templates is highlighted in bold and underlined font. The transcribed twister enzyme strands contain a 5’-GG-overhang as previously described.⁶ The enzyme transcript was purified after in vitro transcription, via a 8.3 M urea denaturing 10% PAGE, visualized via UV shadowing, excised, and eluted overnight at 4 °C into 10 mM Tris (pH 7.5) and 250 mM NaCl (TN₂₅₀). EDTA, which is often added to elution buffers, was omitted here to prevent unwanted sequestration of Mg²⁺ in kinetic experiments, which can be especially problematic with low Mg²⁺ concentrations. The RNA was ethanol precipitated, resuspended in sterile water, and quantified via Nanodrop spectroscopy.

Twister Wild-Type Substrate Strand Preparation.

The twister wild-type substrate strand (IDT) was 5’-end labeled with [γ−³²P] ATP by T4 polynucleotide kinase (New England Biolabs).

Twister Wild-Type Substrate RNA: 5’-CGC GGC AUA AUG CAG CUU UAU UGC C −3’

Samples were purified via an 8.3 M urea denaturing 10% PAGE, excised, and eluted overnight at 4 °C into TN₂₅₀. The RNA was ethanol precipitated, resuspended in sterile water, and quantified by scintillation counting.

Twister N3 Deaza A1 Substrate Strand Preparation.

The twister N3 deaza (3dzA1) substrate strand was synthesized from two separate RNA strands, (1MT) and (2) (see below), which were ligated together with the 5’-ligation reactant (1MT) containing the A1(N3) deaza nucleotide, denoted as “3dzA”, at its very 3’-end. This synthetic strategy was adopted because it was simpler to prepare 3dzA as solid support rather than as a phosphoramidite. The 5’-piece was prepared both with the mutation (1MT) and without the mutation (1WT) as a control for ligation preparation. The two strands were ligated via a DNA splint reaction using T4 DNA Ligase (New England Biolabs). Note that the splint is intentionally longer than the ligated product so that the two can be well separated on denaturing PAGE.

(1MT) 5’−3dzA1 RNA Substrate: 5’-CGC GGC AU3dzA-3’

(1WT) 5’-RNA Substrate: 5’-CGC GGC AUA-3’

(2) 3’−3dzA1 RNA Substrate: 5’-pAUG CAG CUU UAU UGC C-3’

(1MT+2) Ligated 3dzA1 RNA Substrate: 5’-CGC GGC AU3dzA AUG CAG CUU UAU UGC C-3’

(1WT+2) Ligated WT RNA Substrate: 5’-CGC GGC AUA AUG CAG CUU UAU UGC C-3’

Ligation Splint DNA: 5’-TCG AAC TTA GCA ATA AAG CTG CAT TAT GCC GCG A-3’

The DNA splint ligation reaction was incubated at 25 °C for 2.5 h. The ligated product was purified via an 8.3 M urea denaturing 16% PAGE, visualized via UV shadowing, excised, and eluted overnight at 4 °C into TN₂₅₀. The RNA was ethanol precipitated, resuspended in sterile water, and quantified via Nanodrop spectroscopy. As described above, the 3dzA1 substrate strand was radiolabeled with [γ−³²P] ATP by T4 polynucleotide kinase and purified via PAGE for kinetic experiments. As a control, a WT substrate was also produced identically through ligation. It was ethanol precipitated, radiolabeled, and its activity tested at pH 5.5, 10 mM MgCl₂.

Twister 8BrA1 Substrate Strand Preparation.

In contrast to the 3dzA1 substrate, the 8BrA1 twister substrate strand was synthesized as a single RNA oligonucleotide. It was HPLC-purified using reverse phase chromatography on a Waters ACQUITY Arc System with an Xbridge C18 5 μm, 4.6 × 150 mM column. The mobile phase consisted of 0.1 mM ammonium acetate with 2% acetonitrile (Buffer A) and 0.1 mM ammonium acetate solution with 50% acetonitrile (Buffer B). Running conditions started with 100% Buffer A for 10 min followed by a linear gradient to 10% Buffer B over 30 min. The sample eluted at ~29.5 min and was captured by an automated fraction collector. To remove ammonium ions, collected samples were dried down by SpeedVac after which they were dissolved in water and dried down again. After a total of 5 such cycles, the samples were ethanol precipitated to remove any residual salts. The 8Br substrate strand was radiolabeled with [γ−³²P] ATP by T4 polynucleotide kinase for kinetic experiments, as described above.

Chemical Synthesis of Oligoribonucleotides Containing 3-deazaadenosine and 8-bromoadenosine Residues.

Chemical synthesis of 3-deazaadenosine was performed according to the procedure of Erlacher et al.¹⁵ First, 3-deazaadenosine was converted into 5’-O-(4,4’-dimethoxytrityl)-2’(3’)-O-(tertbutyldimethylsilyl)-N6-benzoyl-3-deazaadenosine using standard procedures.¹⁵ This derivative was transformed with succinic anhydride into the respective ester and condensed with long chain alkylamine controlled pore glass (lcca-CPG) to get 3-deazaadenosine loaded support. This prepared support was used for chemical synthesis of 3’ 3-deazaadenosine-terminated oligonucleotides.

To synthesize oligonucleotides with 8-bromoadenosine, commercially available 8-bromoadenosine phosphoramidite (ChemGenes) were utilized. Oligonucleotides were synthesized on a BioAutomation MerMade12 DNA/RNA synthesizer using β-cyanoethyl phosphoramidite chemistry and commercially available phosphoramidites (ChemGenes, GenePharma). For deprotection, oligoribonucleotides were treated with mixture of 30% aqueous ammonia/ethanol (3/1 v/v) for 16h at 55°C. Silyl 2’-protecting groups were cleaved by treatment trimethylamine trihydrofluoride. Deprotected oligonucleotides were purified by silica gel thin layer chromatography (TLC) in 1-proponal/aqueous ammonia/water (55/35/10 v/v/v), as described previously.^16,17 For deprotection, oligonucleotides with 8-bromoadenosine were incubated in a mixture of 30% aqueous ammonia/ethanol (3/1 v/v) for 48h at room temperature.¹⁸ The remaining deprotection and purification steps were the same as described above.^16,17

Kinetic Assays and Data Analysis.

All kinetic assays were conducted under single turnover conditions with 0.25 nM of radiolabeled substrate, 100 nM twister enzyme (saturating), and 100 mM NaCl. Chemical rescue experiments of 3dzA1 and AbA1 were performed at 23 °C. Conformational analysis experiments with the wild-type enzyme were performed at 23 °C as well, while conformational analysis experiments with the A38AA enzyme were performed at 37 °C to enhance dynamics. Buffer concentrations and experimental conditions are provided in the Figures and Tables.

Reactions with half-lives slower than 10 seconds were conducted by conventional hand-mixing, whereas reactions faster than this were performed on a KinTek-RQF3 rapid quench instrument (see below). For reactions completed via hand-mixing, 100 μL reaction solutions containing enzyme, substrate, buffer, and NaCl, but not MgCl₂, were renatured at 90 °C for 2 min. Samples were removed and cooled to room temperature on the bench for at least 10 min. Samples were spun down for ~30 s to collect droplets, transferred to a 23 °C or 37 °C heat block as appropriate, and allowed to equilibrate for at least 3 min. A pre-initiation time point (5 μL) was removed, and the reaction was initiated by addition of the larger volume RNA/buffer/salt solution to the smaller volume 5X MgCl₂ solution to facilitate mixing. Aliquots (5 μL) were removed at appropriate times and quenched with an equal volume of 20 mM EDTA in 2X Formamide Loading Buffer (FLB). Quenched aliquots were immediately placed on powered dry ice and then stored in a −20 °C freezer until gel analysis.

For reactions performed via rapid quench, 300 μL 2X RNA and 2X MgCl₂ solutions were prepared as previously described.¹⁹ The 2X RNA solution was renatured at 90 °C for 2 min and cooled to room temperature for at least 10 min. Both solutions were spun down for 30 s to collect droplets, and a 20 μL aliquot was removed from the 2X RNA solution and mixed with 150 μL of the EDTA stop mix for a pre-reaction time point.¹⁹ Reaction samples were quenched with enough EDTA to produce a final EDTA concentration of 2X the respective MgCl₂ concentration. A 5 μL aliquot was removed from the quenched reaction solution and combined with an equal volume of 2X FLB (2X FLB sample). Both the full volume quenched reaction and small volume FLB sample were then placed on powdered dry ice and later stored in a −20 °C freezer until gel analysis.

Samples prepared by hand-mixing or rapid-quench kinetics were fractionated on an 8.3 M urea denaturing 18% PAGE. Gels were dried and bands visualized with a Typhoon Phosphorimager (Molecular Dynamics). Resulting bands were analyzed via ImageQuant software whereby the band intensity was initially background corrected, and the fraction cleaved of each lane was calculated as f_cleaved = P/(P+R), where f_cleaved is the fraction of RNA substrate cleaved, P is the intensity of the product band and R is the intensity of the reactant band. The data were plotted via KaleidaGraph (Synergy Software) and fit to the single exponential equation

$$f_{\text{cleaved }}=\text{A}+{\text{Be}}^{-k_{\text{obs}}\text{t}},$$ (1)

where f_cleaved represents the fraction of RNA cleaved after time t, A is the final fraction of RNA cleaved, –B is the amplitude of the phase, k_obs is the observed rate constant, and A + B is the fraction cleaved at time zero. In cases where the fraction cleaved was less than 0.3, data were graphed and fit to a straight line.

Additional Equations Used for Data Analysis.

Rescue effects were calculated as k₊/k₋, where k₊ is the rate constant with small molecules and k₋ is the rate constant without small molecules, typically a control containing 5 mM HEPES.

Prior to producing the Brønsted plots, all measured k_obs values were corrected by subtracting the average k_obs value of the HEPES control trial, k_HEPES, from the average k_obs value of a small molecule trial. HEPES-corrected k_obs values that resulted in a negative value were excluded from the final Brønsted plots. The HEPES-corrected k_obs values were then extrapolated to expected values for 100% protonated or 100% deprotonated, as is typical.^7,20 To do this, the fraction of protonated small molecule was first determined at pH 7.5 according to the equation

$$f_{\text{prot}}=\frac{1}{1+{10}^{\text{pH}-\text{p}K\text{a}}},$$ (2)

where f_prot represents the fraction of small molecule that is protonated, pH is the experimental value of 7.5, and pK_a is the pK_a of the small molecule of interest (11 buffers were tested in total). Once f_prot was determined, the k_obs value of a small molecule was divided by the value of f_prot to extrapolate the k_obs value to 100% small molecule protonation. To extrapolate the k_obs values to 100% small molecule deprotonation, the k_obs value of a small molecule was similarly divided by the value of f_deprot, calculated as 1−f_prot. Brønsted plots were constructed at pH 7.5 for various small molecules added to 3dzA1 or AbA1 twister ribozymes by plotting values of log (k_obs – k_HEPES) versus pK_a of the small molecule and fitting to a straight line.

Rate-pH profiles were fit to eq (3), which was derived from a single ionization model

$$k_{obs}=\frac{k_{\text{max}}}{1+{10}^{\text{p}K\text{a}-\text{pH}}}\text{ or log}{k}_{\text{obs}}=\text{log}\left(\frac{k_{\text{max}}}{1+{10}^{\text{p}K\text{a}-\text{pH}}}\right).$$ (3)

RESULTS

Small Molecules Rescue the Activity of A1 Variant Twister Ribozymes

We previously reported that biological small molecules efficiently catalyze the wild-type env9 twister ribozyme (WT) at biologically relevant pH, with stimulatory effects up to 5-fold.⁷ This appears to be accomplished via proton donation to the 5’-oxygen leaving group through a proton shuttle involving small molecules and the N3 of A1 (Scheme 1A). A study by the Lilley lab demonstrated that the N3 of A1 is critical for catalytic activity of the twister ribozyme as a general acid, with an A1(N3)deaza (3dzA1) modification resulting in reactions taking days as opposed to minutes.¹² Given these observations, we hypothesized that if A1 engages in general acid catalysis, then chemical rescue of a 3dzA1 env9 twister ribozyme construct via the introduction of small protonatable molecules may be possible. Small molecule rescue of general acid catalysis was originally reported for N3 of C75/76, with both the genomic and antigenomic HDV ribozymes.^21–23

We began our study by benchmarking the activity of 3dzA1, which lacks the N3 general acid atom, in the env9 twister background used in our prior studies,⁷ against that of the related 3dzA1 ES2 utilized by the Lilley lab.¹² Our 3dzA1 substrate was prepared by a ligation of two oligonucleotides (see Materials and Methods). As a control, we prepared the WT substrate via a similar ligation. We obtained results on this substrate that were the same as those with a standard one-oligonucleotide WT substrate, within error. With 30 mM Tris (pH 7.5) and 10 mM MgCl₂, k_obs of 3dzA1 env9 was very slow, at 0.0002 h⁻¹ (Figure 1, circles). Under equivalent conditions, WT env9 has a k_obs of 1600 h⁻¹, equating to approximately a 10⁷ faster k_obs, confirming the prior report that the N3 of A1 is critical for twister activity.¹² After 11 days, we observed that the 3dzA1 construct was only 4.8% cleaved. This severely reduced activity of 3dzA1 env9 is slightly slower than that previously reported for the ES2 construct with 3dzA1, which had a k_obs of 0.002 h⁻¹ with ~20% cleaved after 11 days, although that study was conducted at pH 8.5.¹² Indeed, our assays with our 3dzA1 env9 conducted at pH 8.2, rather than pH 7.5, showed increased activity with a k_obs of 0.00063 h⁻¹ and ~8.2% cleaved at 5 days, showing that higher pH conditions are more conducive to self-cleavage (Table 3). In summary, our 3dzA1 env9 construct has similar activity to that published for 3dzA1 ES2. Thus, conclusions drawn herein should be applicable to other twister constructs.

Table 3.

pH Dependence of Observed Rate Constants and Rescue Effects of 3dzA1 Twister Ribozyme with Imidazole and 2-Methylimidazole

	No Small Molecule	Imidazolea		2-Methyl Imidazolea
pH	kobs (h−1)b × 10−4	kobs (h−1)b × 10−4	Rescue Effectc	kobs (h−1)b × 10−4	Rescue Effectc
5.5	0.25 ± 0.10	5.9 ± 0.9	24 ± 10	0.57 ± 0.26	2.3 ± 1.4
6.5	0.46 ± 0.06	50 ± 24	110 ± 55	3.1 ± 0.1	6.8 ± 0.9
7.5	1.9 ± 0.1	130 ± 10	66 ± 3	19 ± 1	9.9 ± 0.5
8.2	6.3 ± 0.2	120 ± 20	18 ± 4	29 ± 1	4.5 ± 0.2

^aIn 20 mM MES (pH 5.5) or 10 mM HEPES (pH 6.5– 8.2), 240 mM small molecule.

^bValues are the average of three experiments and errors are the standard deviation of three experiments.

^cRescue effect is defined as the k_obs of trials with small molecules divided by the k_obs of trials with no small molecules. Values are the average of three experiments and errors are calculated as typical for error propagation.

Next, we tested if the 3dzA1 construct could be rescued by addition of small molecules. We first incubated 3dzA1 with imidazole or histidine, which were chosen as these had significant stimulatory effects on the WT in our prior experiments.⁷ In the WT, we previously used 120 mM of small molecule with 5 mM of HEPES as a background buffer,⁷ while here we doubled the concentrations to enhance effects, although significant stimulatory effects were also observed at 120 mM imidazole (see below). Imidazole was 240 mM in 10 mM HEPES, while histidine was 180 mM in 7.5 mM HEPES owing to solubility limits. Strikingly, both small molecules restored activity quite significantly, with fold-effects of 66 ± 3 and 10 ± 3 (Figure 1 and Table 1). These recovery effects strongly support the conclusion that the N3 of the twister ribozyme engages in general acid catalysis in the wild-type ribozyme.

Table 1.

Observed Rate Constants and Rescue Effects for 3dzA1 and AbA1

	3dzA1 Twister Ribozyme		AbA1 Twister Ribozyme
Reaction Conditionsa	kobs (h−1)b × 10−4	Rescue Effectc	kobs (h−1)b × 10−4	Rescue Effectc
5 mM HEPES	1.9 ± 0.1	1.0 ± 0.1	1.4 ± 0.1	1 ± 0.1
10 mM Adenosine	2.6 ± 0.1	1.4 ± 0.1	2.7 ± 0.3	1.9 ± 0.3
250 mM Tris	2.7 ± 0.3	1.4 ± 0.1	1.1 ± 0.2	0.37 ± 0.06
210 mM Lysine	5.2 ± 0.3	2.8 ± 0.2	22 ± 1	16 ± 1
240 mM Glycine	3.1 ± 0.1	1.6 ± 0.1	2.4 ± 0.4	1.7 ± 0.3
180 mM Histidine	19 ± 5	10 ± 3	13 ± 1	9.5 ± 0.8
240 mM Imidazole	130 ± 10	66 ± 3	240 ± 14	180 ± 17
240 mM 2-Methylimidazole	19 ± 1	9.9 ± 0.5	23 ± 1	17 ± 1
240 mM 2-Ethylimidazole	11 ± 1	6.0 ± 0.6	19 ± 2	14 ± 2
180 mM 2-Isopropylimidazole	7.4 ± 0.3	3.9 ± 0.2	11 ± 1	7.7 ± 0.8
240 mM 1,2-Dimethylimidazole	9.9 ± 1.5	5.2 ± 0.8	7.7 ± 0.4	5.6 ± 0.5
240 mM Pyrazole	1.8 ± 0.1	0.92 ± 0.07	1.8 ± 0.2	1.3 ± 0.2

^aAll trials, excluding 5 mM HEPES and 250 mM Tris, contain 5–10 mM HEPES (pH 7.5) as a background buffer.

^bValues are the average of three experiments and errors are the standard deviation of three experiments.

^cRescue effect is defined as the k_obs of trials with small molecules divided by the k_obs of the 5 mM HEPES control. Errors are calculated as typical for error propagation.

Seeing that two different small molecules can rescue 3dzA1, the scope of the study was expanded to include additional small molecules to construct Brønsted plots. We tested small molecules that were not very bulky to avoid steric effects and that were previously shown to have effects on the WT⁷ including Tris, glycine and lysine; several derivatives of imidazole including 2-methylimidazole, 2-ethylimidazole, 1,2-dimethylimidazole, 2-isopropylimidazole, and pyrazole; and a nucleoside (Figure 2A). Mirroring the WT, the strongest rescue effect of 3dzA1 was with imidazole (66-fold), with the imidazole derivatives producing 4- to 10-fold rescue (Figure 2B and Table 1).

Figure 2.

Exogenous small molecules rescue catalytic activity of 3dzA1 and AbA1 twister ribozymes. (A) Chemical structures, pK_as and names of small molecules used in 3dzA1 and AbA1 rescue experiments. Atom with provided pK_a is in bold. All structures are drawn in their deprotonated state. References for pK_as.^12,24,25 (B) Chemical rescue experiments for 3dzA1 at 23 °C in 10 mM MgCl₂ with HEPES background buffer (5–10 mM, pH 7.5). Small molecules were present at 240 mM, with the exceptions of Tris (250 mM), lysine (210 mM), histidine (180 mM), 2-isopropylimidazole (180 mM), and adenosine (10 mM) due to solubility limits. Rescue

Equivalent experiments were performed with an abasic A1-modified env9 twister ribozyme (AbA1), as we hypothesized that the presence of the A1 nucleobase in 3dzA1 could exclude some small molecules from the active site due to steric clash. Removal of the entire A1 nucleotide was slightly more detrimental to catalytic activity than removing the N3, with a k_obs of 0.00014 h⁻¹ as compared to 0.00019 h⁻¹ (Table 1). Like the 3dzA1, the AbA1 activity was recovered by small molecule addition. Overall, the k_obs values for AbA1 in the presence of the small molecules were nearly identical to those observed for 3dzA1 (Figure 2D and Table 1). Rescue of AbA1 by the same panel of small molecules is shown in Figure 2C and contrasted to 3dzA1 in Figure 2D. For example, a 180-fold rescue effect was observed when AbA1 was incubated with imidazole (Figure 2C and Table 1). The imidazole derivatives exhibited rescue effects ranging from 6- to 17-fold, similar to the range of rescue effects associated with the 3dzA1 construct (Figure 2D and Table 1). We did observe that lysine had an enhanced rescue effect with AbA1 compared to 3dzA1, 16- versus 2.8-fold, suggesting a steric effect on 3dzA1.

Experiments with imidazole and 2-methylimidazole revealed that rescue effects are concentration dependent (Table 2). For example, when the concentration of imidazole and 2-methylimidazole were decreased from 240 mM to 120 mM—the concentration used in our previous study⁷—rescue effects dropped from 66- to 33-fold and from 9.9- to 6.3-fold, respectively, but remained significant. This helps explain the low rescue effect with adenosine, where we were forced to use only 10 mM of adenosine owing to solubility. Extrapolating to 240 mM adenosine predicts a 34- to 46-fold rescue effect for 3dzA1 and AbA1 constructs, respectively. Overall, rescue effects result from several factors including the pK_a and concentration of the small molecule.

Table 2.

Rescue Effect Concentration Dependence with 3dzA1 Twister Ribozymes

Reaction Conditions	kobs (h−1)a × 10−4	Rescue Effectb
120 mM Imidazolec	63 ± 3	33 ± 2
240 mM Imidazoled	130 ± 10	66 ± 3
120 mM 2-Methylimidazolec	12 ± 0.2	6.3 ± 0.3
240 mM 2-Methylimidazoled	19 ± 1	9.9 ± 0.5

^aValues are the average of three experiments and errors are the standard deviation of three experiments.

^bRescue effect is defined as the k_obs of trials with small molecules divided by the k_obs of the 5 mM HEPES control. Errors are calculated as typical for error propagation.

^cIn 5 mM HEPES, pH 7.5

^dIn 10 mM HEPES, pH 7.5

To assess the extent to which small molecules catalyze self-cleavage by proton transfer, we generated several Brønsted plots. Figure 3A and 3B provide Brønsted plots of log(k_obs–k_HEPES) versus the pK_a for 11 different small molecules for 3dzA1, with k_obs extrapolated to either 100% protonated or 100% deprotonated as described in the Materials and Methods. For 3dzA1, these plots show a linear dependence of log (k_obs– k_HEPES) on small molecule pK_a, with slopes of −0.67 ± 0.10 (R² of 0.86) for 100% protonated and +0.33 ± 0.10 (R² of 0.60) for 100% deprotonation (Figure 3A and 3B). Similar trends are observed for the Brønsted plots for AbA1, with slopes of −0.60 ± 0.06 (R² of 0.91) for 100% protonated and +0.40 ± 0.06 (R² of 0.83) for 100% deprotonated (Figure 3C and 3D). Thus, the strength of rescue is highly correlated with the pK_a of the small molecule, suggesting that small molecule rescue effects are a result of proton transfer. The larger magnitude of the slope for protonated small molecules, as well as positioning of A1 for proton donation in all crystal structures^9–11,13, favors a critical role for A1(N3) in protonating the leaving group 5’O in the reaction.

Figure 3.

Brønsted plots of small molecule rescue support proton transfer. Data for the 3dzA1 (A, B) and AbA1 (C, D) were collected at pH 7.5 and 10 mM MgCl₂, with k_obs values extrapolated to either 100% protonated (A, C) or 100% deprotonated (B, D) using equation (2). Slopes were calculated to be −0.67 ± 0.10 (R² of 0.86), +0.33 ± 0.10 (R² of 0.60), −0.60 ± 0.06 (R² of 0.91) and +0.40 ± 0.06 (R² of 0.83) for A-D respectively. Data for pyrazole were excluded from plots A and B, and data for Tris were excluded from plots C and D, as the HEPES-corrected k_obs values resulted in a negative number. (E) Identities of the 11 small molecules and their respective pK_as in plots A-D. Small molecules concentrations are as described in Figure 2. References for pK_a’s are provided in the legend to Figure 2A.

Small Molecule Rescue Effects are pH-Dependent

Given the potential of small molecules for general acid-base catalysis described in the previous section, we reasoned that the magnitude of the rescue effects could be influenced by pH. We therefore measured the pH-dependence of rescue of 3dzA1 by imidazole and 2-methylimidazole (Figure 4 and Table 3), as these small molecules had significant rescue effects with 3dzA1 (Figure 2). The logarithm of 3dzA1 increases linearly with pH and reaches a maximum at pH 7.5 for imidazole and 8.2 for 2-methylimidazole (Figure 4A, circles and squares). When fitting these data to equation 3, we observed pK_as of 6.8 ± 0.2 and 7.3 ± 0.15 for imidazole and 2-methylimidazole, respectively, which are near their free small molecule pK_as of 7.0 and 7.9.^24,25 Notably, no rate saturation is observed in the absence of an imidazole (Figure 4A, diamonds), supporting the conclusion that the pK_as in the presence of the imidazoles are assignable to the imidazoles themselves. Leveling off of the rate-pH profile at lower pH in the absence of imidazoles may be due to an alternative reaction channel, and a slope near unity at higher pH suggests proton transfer but from a functional group with a pK_a above the upper measured pH of 8.2.²⁶

Figure 4.

Small molecule rescue effects are pH-dependent. (A) Logarithmic rate-pH profiles of 3dzA1 show an increase in k_obs with pH and reveal a pK_a near neutrality in the presence of imidazole (circle) and 2-methylimidazole (square) but not in the absence of small molecule (diamond). Data are fit to the logarithmic version of equation (3) for imidazole and 2-methylimidazole indicating pK_as of 6.8 ± 0.2 and 7.3 ± 0.15 and k_max values of 0.014 ± 0.002 and 0.0032 ± 0.001, respectively; no pK_a was found in the absence of small molecule so these data points are connected by line segments. (B+C) Rescue profiles of 3dzA1 with imidazole (B) and 2-methylimidazole (C) show maximal increased rescue effects near neutrality. Rescue effects were calculated by dividing the average of three trials with small molecules (+) by the average of three trials with no small molecules (−). Note that the scale is 10X larger in panel (B) than (C). Single-turnover experiments were performed with 3dzA1 at 23 °C in 10 mM MgCl₂ with 240 mM imidazole, 240 mM 2-methylimidazole, or no small molecule. Background buffers to maintain pH were 20 mM MES at pH 5.5 and 10 mM HEPES at pH 6.5–8.2. Data points are the average of three trials and error bars are the standard deviation of three trials for panel A or determined via typical error propagation for panels B and C. Data are tabulated in Table 3.

The range of rescue effects for imidazole is 18 to 110, while for 2-methylimidazole it is 2.3 to 9.9, with the actual value depending on pH (Figure 4B, 4C, and Table 3). Maxima occur near pH 6.5 for imidazole and pH 7.5 for 2-methylimidazole, which are near the pK_as of 7.0 and 7.9, respectively. These rate-pH profiles are also consistent with proton transfer by small molecule.

8BrA1 Modification Enhances the Rate of the Twister Ribozyme in an Appropriate Background

In all available twister ribozyme crystal structures^9–11,13, the A1 nucleobase is in the syn conformation, with the N3 of A1 oriented towards the 5’O leaving group of the scissile phosphate with distances ranging from 3.2 to 5.3 Å depending on pH.⁷ Prior studies from our lab reveal that the presence of the rare syn glycosidic conformation is often associated with a functional role in ribozyme active sites or riboswitch binding sites.²⁷ Crystal structures of ribozymes do not always reflect the catalytic form of the active site, however, and can never capture the transition state. Thus, testing such functional features experimentally is critical.^28–30 In particular, no study to date has determined if the syn conformation of A1 is catalytically relevant. One method of investigating glycosidic conformation is to introduce a bromine modification at the 8-position of a purine, which forces the syn conformation, as the increased steric bulk of the Br is disfavored above the ribose sugar.¹⁸ Indeed, studies from our lab with 8Br-substitutions in the leadzyme led to significant decreases in catalytic activity at anti sites due to steric clashes, and modest increases in catalytic activity at syn sites due to pre-structuring of the active site, which revealed the catalytic relevance of a computer model of the ribozyme over NMR and crystallographic ones.²⁸ Thus, we hypothesized that if A1 is indeed syn in the active form of the twister ribozyme, then pre-structuring of the active site via an 8BrA1 modification should increase catalytic activity modestly. Conversely, if A1 is anti in the transition state, catalytic activity should decrease significantly with 8Br substitution, with values up to 23-fold measured for the leadzyme.²⁸

We began with a rate-pH profile of 8BrA1 from pH 5.5 to 8.5 at 1 and 10 mM MgCl₂ (Figure 5 and Table 4). In the wild-type enzyme background, the activity of 8BrA1 was several-fold lower in all conditions compared to WT. For both MgCl₂ conditions, the smallest fold-decreases were observed at pH 5.5, with decreases in k_obs ranging from 5.8 to 16-fold, while the larger fold-decreases were observed from pH 6.5 to 8.5, with decreases in k_obs ranging from 16 to 29-fold. Additionally, we observed a downward shift of the pK_a by 0.3 and 0.5 units with the 8Br substitution at 1 and 10 mM Mg²⁺, respectively (Figure 5). This pK_a shift is similar in magnitude to those observed with brominated monocyclic and bicyclic organic molecules, including aniline, phenol and 2-naphthalamine, which have downward pK_a shifts ranging from 0.4 to 2.0.³¹ Overall, pK_a shifts are relatively minor and effects of 8Br are observed throughout the rate-pH profile, suggesting that observed inhibitory effects of 8Br on rate are likely due to other effects than pK_a differences. We reasoned that the rate loss of the 8BrA1 could be the results of a steric clash of the bromo substitution in the active site.

Figure 5.

8BrA1 has decreased k_obs compared to WT. WT (circle) and 8BrA1 (square) experiments were performed at 23 °C with 1 mM (A) and 10 mM (B) Mg²⁺ with 30 mM MES (pH 5.5), HEPES (pH 6.5), or Tris (pH 7.5–8.5). Data are fit to equation (3) providing (A) pK_as of 6.7 ± 0.3 and 6.4 ± 0.2 and k_max values of 5.1 ± 0.7 and 0.18 ± 0.01 for WT and 8BrA1, respectively and (B) pK_as of 6.6 ± 0.7 and 6.1 ± 0.7 and k_max values of 20 ± 6 and 0.70 ± 0.18 for WT and 8BrA1, respectively. Data are the average of three trials and error bars are the standard deviation of three trials. Data are tabulated in Table 4.

Table 4.

Observed Rate Constants and Fold-Effects of the WT and 8BrA1 Twister Ribozyme Substrates with the Wild-Type Twister Ribozyme

	1 mM Mg2+			10 mM Mg2+
pH	WT kobs (min−1)a	8BrA1 kobs (min−1)a	Fold Effectb	WT kobs (min−1)a	8BrA1 kobs (min−1)a	Fold Effectb
5.5	0.33 ± 0.02	0.021 ± 0.004	0.063 ± 0.012	1.7 ± 0.1	0.29 ± 0.01	0.17 ± 0.01
6.5	1.5 ± 0.3	0.10 ± 0.01	0.064 ± 0.015	6.9 ± 1.9	0.35 ± 0.13	0.052 ± 0.023
7.5	5.2 ± 0.6	0.19 ± 0.01	0.037 ± 0.005	26 ± 2	0.91 ± 0.07	0.035 ± 0.004
8.5	4.5 ± 1.3	0.17 ± 0.01	0.037 ± 0.011	14 ± 4	0.56 ± 0.02	0.040 ± 0.011

^aValues are the average of three experiments and errors are the standard deviation of three experiments.

^bFold effect is defined as the k_obs of the 8BrA1 twister ribozyme divided by the k_obs of the WT twister ribozyme. Values are the average of three experiments and errors are calculated as typical for error propagation.

We considered if the bulky nature of the 8Br modification could result in steric clashes. We initially modeled an 8BrA1 into the crystal structures of the env9 and env22 twister ribozymes, intentionally omitting energy minimization in order to visualize any clash (Figure 6A and 6B). Modeling of an 8BrA1 into all available twister ribozyme crystal structures revealed clashes with similar distances ranging from 1.7 to 2.5 Å (Table S1). We observed that the modeled bromine atom is just 2.5 Å (env9) or 1.7 Å (env22) from loop P4, specifically between PK1 and PK2 (Figure 6 and Table S1). Converting the env9 crystal structure to a space-filling model revealed that the phosphate backbone and nucleotide orbitals of A38 in loop 4 overlap with those of the 8BrA1 modification (Figure 6C and 6D). As proper formation of PK1 and PK2 is integral for formation of the active site, this would likely have a significant negative effects on catalytic activity potentially preventing A1 from adopting the proper syn conformation and explaining the fold-decrease effects observed herein.³²

Figure 6.

Modeling of 8BrA1 with the env9 (4QJH)¹³ and env22 (4RGE)¹⁰ crystal structures reveals a steric clash. Stick models of (A) env9 and (B) env22 twister ribozymes show that the bromine modification (orange) is within 1.7 to 2.5 Å of the phosphate backbone between PK2 and PK1 (green). Space-filling models of the env9 twister ribozyme without (C) and with (D) the bromine modification emphasize the unfavorable steric clash. An 8H is shown explicitly in panel 6C. Coloring is as follows: A1 nucleotide (blue), 8H (white, transparent in panel C), 8Br modification (orange, transparent in panel D), PK1 (purple), PK2 (yellow), and nucleotides between PK1 and PK2 (green). Measured distances are tabulated in Table S1.

In an attempt to provide space for the bulky 8Br, we designed a variant env9 enzyme strand with an additional adenine between the two pseudoknots, similar to that found for env22 (Figure 6B). The new enzyme strand, which we denoted “A38AA enzyme strand,” was tested with both the WT and 8Br substrates at 37 °C, at pH 7.5 with 1 and 10 mM MgCl_2, for 24 hours to determine the initial rates. Higher temperature was chosen to enhance dynamics of the variant ribozyme. Both the WT and 8Br substrates had significantly lower activity in the A38AA enzyme background, suggesting some structural distortion (Table 5). However, when comparing activity of the WT and 8Br substrates in the background of the A38AA enzyme strand, the 8Br substrate was now faster by 2.1-fold at 1 mM Mg²⁺ and a striking 9.6-fold at 10 mM Mg²⁺ (Figure 7 and Table 5). This stands in stark contrast to the aforementioned up to 29-fold decrease in k_obs for the 8BrA1 substrate with the WT enzyme strand. These data strongly support A1 as syn in the transition state, with tight packing of 8HA1 in the wild-type assisting in maintaining the syn conformation (Figure 6C).

Table 5.

Observed Rate Constants and Fold-Effects of the WT and 8BrA1 Twister Ribozyme Substrates With the A38AA Enzyme Strand

MgCl2 (mM)	WT Substrate kobs (min−1)a × 10−4	8BrA1 Substrate kobs (min−1)a × 10−4	Fold Effectsb
1	1.0 ± 0.3	2.2 ± 0.1	2.1 ± 0.6
10	0.87 ± 0.08	8.4 ± 0.4	9.6 ± 1.0

^aValues are the average of three experiments and errors are the standard deviation of three experiments.

^bFold effect is defined as the k_obs of the 8BrA1 twister ribozyme divided by the k_obs of the WT twister ribozyme. Values are the average of three experiments and errors are calculated as typical for error propagation.

Figure 7.

An 8BrA1 substrate strand stimulates catalytic activity over a WT substrate strand in the background of the A38AA enzyme strand. Fold-effects at 1 and 10 mM MgCl₂ at 37 °C and pH 7.5 are 2.1 ± 0.6 and 9.6 ± 1.0, respectively. Fold-effects were calculated by dividing the average of three trials of the 8BrA1(A38AA) twister ribozyme by the average of three trials of the WT(A38AA) twister ribozyme. Error bars were determined via typical error propagation methods. Data are tabulated in Table 5.

Given that 8Br facilitates twister ribozyme cleavage in the A38AA background, we postulated that perhaps 8BrA would drive cleavage in just an oligonucleotide, possibly providing a pared-down ribozyme. To test this notion, we prepared two model oligonucleotides with 8BrA, similar in sequence to those previously reported and under similar conditions of 1 mM spermidine and 0.1% PVP.³³ Initial cleavages observed at very long times (24 h) were eliminated when the samples were first boiled in 0.1% SDS, suggesting small amounts of ribonuclease that was inactivated (Figure S2 and S3). Because cleavages were removed or reduced upon boiling in SDS, we conclude that possibly aiding α catalysis alone by conformational biasing is not enough to attain rate acceleration or cleavage specificity.

Discussion

While mechanistic characterization of the twister ribozyme has firmly established roles for the catalytic guanine in β and γ catalysis, the A1 nucleotide has been suggested to play different roles in general acid catalysis depending on the reaction channel.⁷ Crystallographic data suggest that the N3 of a syn A1 interacts with the 5’O leaving group owing to their proximity, measured at 3.2 to 5.3 Å.^9–11,13 Biochemical experiments indicate that the N3 of A1 is critical for catalysis, as an A1(N3) deaza modification renders the ribozyme catalytically inactive.^11,12 However, confounding the general acid role at biological pH, NMR experiments indicate that the pK_a of A1(N3) is 5.1,¹¹ which is significantly lower than the apparent pK_a of the twister ribozyme at ~6.9 determined from the rate-pH profile.^7,9 The difference of nearly 2 pK_a units suggests a more complex, multichannel mechanism at biological pH (Scheme 1). Recent data from our lab suggest that buffer catalysis is important at biological pH,⁷ and that the role of A1(N3) under this reaction channel needs attention.

Herein, we investigated the contribution of the A1(N3) to the catalytic mechanism of the twister ribozyme in terms of both proton transfer and glycosidic conformation. To quantify the contribution of the N3 of A1 to proton transfer, we conducted small molecule rescue experiments with two inhibited twister constructs, one with a 3 deaza A1 modification and the other with an abasic A1 modification. We found that small molecules rescue the activity of both constructs in a similar fashion, with rescue effects up to 180-fold (Figure 2D and Table 1). Brønsted analysis shows strong linearity for the general acid plots, which supports the conclusion small molecules recover activity via a proton transfer mechanism, implicating A1 in a general acid-base role (Figure 3). Stronger slopes of the protonated plots (α = −0.60 to −0.67) than the deprotonated plots supports a general acid role for A1, as implicated by crystal structures. A general acid role is further supported by the rate-pH profiles of the 3dzA1 variant ribozyme in the presence of imidazole and 2-methylimidazole, in which the rate is log-linear at lower pH and levels off at higher pH, with a pK_a near that of the small molecule; importantly, this behavior is lost in the absence of added small molecule (Figure 4A). Additionally, we determined that the catalytically relevant conformation of A1 is syn, as found in crystal structures, by testing a twister ribozyme with an 8BrA1 upon compensatory mutation of A38AA to provide space for the Br atom (Figure 7 and Table 5). Indeed, the 8BrA1 substrate is nearly 10-fold faster than a non-modified A1 substrate in the A38AA ribozyme background at 10 mM Mg²⁺.

Comparison of Twister Ribozyme Rescue Effects to Prior Ribozyme Studies

Chemical rescue experiments have been used with other small self-cleaving ribozymes to identify functional residues involved in general acid-base catalysis. For example, extensive chemical rescue experiments were conducted with the genomic and antigenomic HDV ribozymes. Rate-pH studies with functional group variations implicated C76 of the antigenomic HDV ribozyme (equal to C75 of the genomic HDV ribozyme) in proton transfer.^21,34 The general acid-base role of C76 was identified via imidazole rescue of a C76U variant.²¹ The mutation decreased self-cleavage activity by 10⁶-fold and imidazole rescued activity at least 250-fold. For a C76Δ antigenomic HDV construct, a similar decrease in self-cleavage activity was observed, but larger rescue effects up to 1000-fold were observed for cytosine and imidazole analogues.²² General acid-base catalysis of the genomic HDV ribozyme was also tested via similar chemical rescue experiments with equivalent changes at the C75 position.²³ For both the C75Δ and C75U genomic HDV ribozymes, k_obs decreased by ~10⁶-fold and was rescued up to 1000-fold when incubated with cytosine or imidazole analogues. Rescue effects supported C75/C76 in general acid-base catalysis.^22,23 Additionally, Brønsted studies on the antigenomic ribozyme provided β values of ~0.5²² with an updated value of −0.8 ± 0.2 for the antigenomic ribozyme using a variety of C75 analogues differing only in the positioning of ring nitrogens or addition of fluorine.³⁵ Subsequent 5’-bridging sulfur substitution studies determined that C75/C76 residues play the role of a general acid in the catalytic mechanism of the HDV ribozyme.³⁶

In our current study, observed small molecule rescue of both 3dzA1 and AbA1 twister ribozymes with diverse exogenous small molecules strongly supports a model in which proton transfer plays a major role for the twister ribozyme. For the 3dzA1 and AbA1 twister ribozymes, we observed activity losses up to ~10⁷-fold and rescue effects up to 180-fold in the presence of protonatable small molecules (Figure 2B, 2C and Table 1). The present results on the twister ribozyme generally parallel those of the genomic and antigenomic HDV ribozymes, which had activity losses of approximately 10⁶-fold, with rescue effects of 5- to 1000-fold.^21–23 The similarity between both the activity losses and the rescue effects, supports the conclusion that A1(N3) participates in the catalytic mechanism of the twister ribozyme via general acid-base catalysis and serves a similar role as C75(N3). This is corroborated by the Brønsted plots for 3dzA1 and AbA1 variant twister ribozymes, which show a strong linear dependence of log (k_obs–k_HEPES) on the pK_a of the exogenous small molecule and significant α values for proton donation of −0.67 ± 0.10 (R² = 0.86) and –0.60 ± 0.06 (R² = 0.91) for 3dzA1 and AbA1, respectively, strongly supporting the conclusion that small molecules participate in proton transfer (Figure 3). Additionally, the dependence of rescue on the identity of small molecule supports that the active site is solvent accessible, as for the wild-type twister ribozyme.⁷ The slope for the abasic twister of −0.60 ± 0.06 is in the same range as the α values for the HDV ribozyme reported as −0.5 to −0.8 for the antigenomic ribozyme.^22,35 Additionally, the observed α values for 3dzA1 and AbA1 are larger than we previously observed with the wild type ribozyme at pH 6.5 and 7.5 of −0.23 and −0.46, respectively.⁷ This suggests that small molecules play a major role in the proton transfer mechanism for the twister ribozyme at higher biological pH, including where A1 is now unable to participate in proton transfer owing to changing it to 3dzA1 and AbA1. As both the 3dzA1 and AbA1 twister ribozymes lack the A1 nucleotide, both direct and indirect proton transfer via A1 are no longer possible. Thus, small molecule proton transfer presumably becomes the predominate method of general acid catalysis in the inhibited constructs, mimicking Channel 1 of the mechanism with direct proton transfer to the leaving group. Proton transfer thus appears to be especially important in this channel.

8BrA1 Activity Supports the Catalytic Conformation of the A1 Glycosidic Bond as Syn

Nucleotides can occupy different orientations above the ribose sugar, with limiting conformations denoted as anti and syn.²⁷ The anti conformation is characterized by the Watson-Crick face of the nucleobase being rotated away from the sugar, while the syn conformation involves the Watson-Crick face rotated above the sugar. For most nucleobases, the syn conformation is higher in energy than the anti conformation and thus less populated. The active sites of riboswitches and ribozymes are enriched in syn bases, however, with such bases often involved in functional roles.²⁷

For the twister ribozyme, the conformation of A1, the general acid, is syn in all crystal structures, placing the N3 of A1 in the vicinity of the 5’O leaving group.^9–11,13 However, transition states cannot be crystallized; thus, any catalytic relevance of a syn A1 conformation needs to be tested experimentally. Indeed, our prior conformation analysis of the leadzyme ribozyme using 8Br modifications revealed that a syn base in leadzyme crystal structures was not catalytic.²⁸ As such, we sought to test the glycosidic conformation of A1 in the twister ribozyme via an 8BrA1 modification.

In the wild-type background, the 8BrA1 modification proved detrimental to catalytic activity, with a 6- to 30-fold decrease in k_obs as compared to the wild-type enzyme, dependent on pH and MgCl₂ concentration (Figure 5 and Table 4). This inhibitory effect was suspected to result from a steric clash between the bulky 8Br modification and nearby nucleotides located between the conserved PK1 and PK2 (Figure 6). This observation suggested tight packing of the imidazole ring of A1 into the core of the ribozyme. To provide space for the bromine substitution, we introduced an additional nucleotide between PK1 and PK2, in A38AA, to increase the space around the bulky 8-bromine modification. This approach is analogous to a bump-hole experiment.³⁷ The additional A diminished the baseline self-cleavage activity of the twister ribozyme by 10⁴ to 10⁵-fold compared to the wild-type ribozyme, consistent with the importance of the pseudoknots.³² Strikingly, A38AA twister ribozymes were 10-fold faster with an 8Br substrate strand than a WT substrate strand (Figure 7 and Table 5). This enhancement supports the catalytic relevance of the syn conformation of A1; furthermore, it suggests that tight packing might be how the syn conformation of normal 8H is stabilized in the wild-type. To investigate this idea further, we modeled an anti conformation of A1 into all six available twister ribozyme crystal structures. Indeed, we found that our modeled anti A1 introduced a severe steric clash between its Watson-Crick face and conserved nucleotides located between PK1 and PK2 (see example of env9 in Figure S4B). We determined distances as short as 2.2 Å between the C2 of A1 and the phosphorus, and distances as short as 1.6 Å between the C2 of A1 and the bridging oxygens (Table S2). Thus, the syn conformation is populated by disfavoring the anti conformation by a strong steric clash.

While population of the syn conformation appears to be driven primarily by destabilization of the anti conformation via sterics, we also note the potential for favorable interactions in the syn conformation. In particular the N6 of A1 donates hydrogen bonds to nonbridging oxygens (NBOs) of two consecutive residues, which are the pro-S_P oxygen of C25 and the pro-R_P oxygen of C26 at the start of 5’ strand of PK2, with heteroatom distances of 3.0 and 2.7 Å, respectively as measured in 4OJI, and conserved in an equivalent position in 4RGE with distances of 4.0 and 2.7 Å (Table S1).^9,10 The syn orientation in general does not correspond to a single conformation, but rather sweeps out 45° in a full syn conformation and a full 180° in going to intermediate syn;²⁷ the N6-NBO interactions may thus serve to finely tune the position of A1 so that the N3 is optimally positioned for catalysis. Additionally, the N6 interactions with the anionic NBOs may serve to help shift the pK_a of A1(N3) upward and favor its protonation.

In conclusion, our study supports a mechanism in which A1 makes a major contribution to twister ribozyme catalysis via proton transfer as a general acid (α ~ −0.67), and that it does so by adopting the syn conformation in the transition state. This syn conformation appears to be populated primarily by tight packing within the active site that destabilizes the anti conformation. Fine-tuning of the syn conformation via N6 hydrogen bond donation to nearby phosphate oxygens appears to be important both for orientation of the general acid and shifting of its pK_a.

ABBREVIATIONS

3dzA1

env9 twister ribozyme with an N3 deaza A1 modification

8Br

8 Bromine modification

A38AA

ribozyme enzyme strand with an additional adenine nucleotide in loop 4, located between PK1 and PK2

AbA1

env9 twister ribozyme with an abasic A1 modification

env9

ninth example of a type P3 twister ribozyme from an environment sample

FLB

formamide loading buffer

HDV

Hepatitis Delta Virus

PAGE

polyacrylamide gel electrophoresis

TN250

10 mM Tris (pH 7.5), and 250 mM NaCl

WT

wild-type env9 twister ribozyme