The positively charged active site of the bacterial toxin RelE causes a large shift in the general base pKa

Abstract

The bacterial toxin RelE cleaves mRNA in the ribosomal A-site. Although it shares a global fold with other microbial RNases, the active site contains several positively charged residues instead of histidines and glutamates typical of ribonucleases. The pH dependences of wild-type and mutant RelE indicate it uses general acid-base catalysis, but either the general acid (proposed to be R81) or the general base must have a substantially downshifted pK_a. However, which group is shifted cannot be determined using available structural and biochemical data. Here, we use a phosphorothiolate at the scissile phosphate to remove the need for a general acid. We show this modification rescues nearly all of the defect of the R81A mutation, supporting R81 as the general acid. We also find that the observed pK_a of the general base is dependent on the charge of the side chain at position 81. This indicates that positive charge in the active site contributes to a general-base pK_a downshifted by more than 5 units. Although this modestly reduces the effectiveness of general acid-base catalysis, it is strongly supplemented by the role of the positive charge in stabilizing the transition state for cleavage. Furthermore, we show that the ribosome is required for cleavage but not binding of mRNA by RelE. Ribosome functional groups do not directly contact the scissile phosphate, indicating that positioning and charge interactions dominate RelE catalysis. The unusual RelE active site catalyzes phosphoryl transfer at a comparable rate to similar enzymes, but in a ribosome-dependent fashion.

Graphical Abstract

Introduction

Phosphoryl transfer reactions, including RNA cleavage, are ubiquitous in biology¹. In bacteria, several toxin-antitoxin systems act by cleaving mRNA, therefore inhibiting translation either globally or for specific transcripts². These systems may be selfish gene elements, which cannot be removed without killing the organism due to the longer lifetime of the toxin molecules³. However, chromosomal versions respond to regulatory stimuli such as nutrient stress, suggesting they may improve the fitness of the overall population. Contrary to a proposed role in bacterial persistence by inducing a dormant state, deletion of toxin-antitoxin systems does not reduce persistence⁴. To repress translation, some toxins (MqsR, MazF) may bind to and cleave free mRNA, but others (RelE, YafQ, YoeB, and HigB) only cleave ribosome-bound mRNA. Toxins in both cases often have structural homology to microbial RNases^5,6.

RelE is the toxin component of the RelBE system, and is ubiquitous in bacteria and archaea^7–10. It has some specificity for the RNA target sequence, but an absolute requirement for ribosome-bound RNA, although no ribosome functional groups are in direct contact with the scissile phosphate¹¹. Under single-turnover conditions, RelE cleaves RNA at more than 100 s⁻¹, comparable to the fastest RNases^12–14. This represents a 10¹⁰-fold increase over the uncatalyzed reaction in similar conditions¹⁵. Although it shares the microbial RNase fold, the active site residues are unusual (Figure 1)¹¹. Prototypical RNases have residues with neutral pK_as that can act as general acid/base partners to promote the reaction¹⁶. Instead, RelE has arginines, lysines, and a tyrosine in the active site. These residues have substantial effects on catalysis. Mutation to alanine results in defects of 10²- to 10⁶-fold on the chemical step, but their unperturbed pK_as are well-removed from neutrality¹².

Figure 1.

Active site of RelE. A) Precleavage structure (pdb entry 4V7J) of R81A mutant with 2’-O-methyl-modified nucleophile (nuc). 5’-oxygen leaving group is labeled (lg). R81A and scissile bond are colored red. Phosphate backbone is not in inline conformation, and catalytic residues are in the vicinity but not positioned for catalysis. B) Postcleavage structure (pdb entry 4V7K) of wild type. K52, K54, and Y87 are within hydrogen-bonding distance of the nucleophile, but the leaving group is not present. R81 and the scissile phosphorus are colored red.

To probe the role of acid-base catalysis, the pH dependences of wild-type and mutant RelE were measured¹⁷. Wild-type RelE has been shown to have no pH-rate dependence over a wide pH range (pH 5.5 to 9.0). This is consistent with acid-base catalysis only if the pK_as of the general acid and general base are widely separated so that the traditional bell curve is broadened and only the wide, flat top of the curve is within the measurable range. Because the active site exclusively contains groups with high pK_as in isolation (lysines and arginines) this implies that one of the acid or base must have a large pK_a downshift. This could be driven by the large amount of positive charge in the active site, which will favor the deprotonated state. Similar pK_a shifts have been observed in other systems, driven either by a hydrophobic or a charged environment^18–21.

Several alanine mutants in the active site result in nucleases with positive pH-rate dependences. This suggests that in these contexts, there is no longer a group with a low pK_a contributing to catalysis. In particular replacement of R81, the proposed general acid, by alanine results in a log-linear pH-rate relationship with a positive slope of approximately one. This suggests that the general base pK_a is high in this mutant. This could be because R81 is the group with a low pK_a in wild-type RelE, combining with a general base with high pK_a to create the flat pH-rate dependence. Alternatively, the general base pK_a in wild type is low, but the loss of positive charge at position 81 causes it to increase substantially. These two reaction mechanisms cannot be resolved using the available data. To differentiate between them, we measured the cleavage rate of wild-type and mutant RelE with phosphorothiolate-substituted RNA. We combined this with pH-rate measurements for RelE mutants with different charges in the active site. Additionally, we measured the affinity of RelE for free mRNA to determine whether ribosomes contributed to the chemical step or only to binding. Our results demonstrate the atypical catalytic strategies that RelE employs for ribosome-dependent cleavage of RNA.

Materials & Methods

Enzyme and Substrate Preparation

Wild-type and mutant enzymes were purified as previously described¹².

Unmodified mRNA 5’-GGCAAGGAGGUAAAAAUGUAGAAAAACAAU-3’ and fragments were obtained from the Keck Oligonucleotide Synthesis facility at Yale University. Phosphorothiolate-modified oligos were prepared as previously described²², with the following modifications. Synthesized dinucleotide ApG (with phosphate linkage) and ApsG (with phosphorothiolate linkage) were phosphorylated using γ−³²P-ATP, resulting in a radioactive phosphate on the 5’-end. The 5’-fragment 5’-GGCAAGGAGGUAAAAAUGU-3’, 3’-fragment 5’-pAAAAACAAU-3’, and DNA splint 5’-CCTCATTTTTACATCTTTTTGTTA-3’ were added in a single ligation mixture. First, annealing reactions containing 24 μM labeled ApG or ApsG dinucleotide, 110 μM 5’-fragment, 100 μM 3’-fragment, 2.5 mM EDTA, and 50 mM KCl were heated to 90 °C for 3’ and cooled 0.1 °C/s to 16 °C in a PCR machine. RNA ligase mix contained 1 part 10x RNA ligase 2 reaction buffer, 4 parts PEG 8k, and 2 parts RNA ligase 2 (New England Biolabs). Equal volumes of ligase mix and annealing reaction were mixed and incubated at 16 °C for 16 hours. Full-length 30-nucleotide mRNA was purified using 20% acrylamide TBE gels. ApG dinucleotide was converted with quantitative efficiency to full length; ApsG dinucleotide was only 0–6%. The presence of phosphorothiolate substitution was confirmed with a silver nitrate test as previously described.

mRNA with a 3’-Alexa Fluor 488 modification for fluorescence intensity and anisotropy was prepared as previously described¹⁷.

Reaction complex containing 300 nM 70S ribosomes, 400 nM tRNA^fmet, and trace radiolabeled mRNA in formation buffer (50 mM Tris-HCl pH 7.5, 70 mM NH₄Cl, 7 mM MgCl₂, 1 mM dithiothreitol was incubated at 37 °C for 30–60 minutes. Complexes were spun at 120,000 rpm for 2 hours in a Beckman Optima Max TL centrifuge with TLA 120.1 rotor. Pellets were resuspended in MMB buffer (50 mM MES, 50 mM MOPS, 50 mM Bicine, 70 mM NH₄Cl, 30 mM KCl, 2 mM MgCl₂, 1 mM dithiothreitol) at a given pH.

To remove the 2’ protecting group, ribosome complexes were placed in the focus of a 100 W, 365 nm UV lamp for 5 minutes. mRNA with all phosphate linkages was cleaved at the same rate by wild-type and mutant RelE as previously reported, indicating this irradiation did not substantially affect ribosome activity (data not shown).

Kinetics

For reactions with a half-life greater than 10 seconds, enzyme and reaction complex were mixed by pipette. Trace radiolabeled reaction complex was mixed with at least 30 μM RelE at 25 °C. 2 uL aliquots were removed at defined times and added to 8 uL of quench buffer (80% formamide, 50 mM Tris-MES pH 6.5, 65 mM EDTA). Faster reactions were mixed using a KinTek RQF-3 Quench Flow instrument. Timepoints were run on 16% polyacrylamide gels and quantitated using a Storm Phosphorimager. Reactions were fit to a single exponential (equation 1). Reactions that displayed a substantial residual were fit to a double exponential (equation 2), in which case the fast rate is reported.

$$\mathit{\text{fraction}} \mathit{\text{reacted}}=A\left(1-e^{-kt}\right)+Y_o$$ (equation 1)

$$\mathit{\text{fraction}} \mathit{\text{reacted}}=A_{\mathit{\text{fast}}}\left(1-e^{-k_{\mathit{\text{fast}}}t}\right)+A_{\mathit{\text{slow}}}\left(1-e^{-k_{\mathit{\text{slow}}}t}\right)+Y_o$$ (equation 2)

Binding Measurements by Fluorescence

Fluorescence intensity and anisotropy were measured using a PTI QuantaMaster 40. For anisotropy, increasing amounts of RelE were added to 50 nM Alexa-mRNA. All slit widths were 4mm, excitation wavelength was 485 nm, and emission was measured from 515 to 535 nm. Average anisotropy was plotted against concentration, corrected for dilution, and fit to equation 3, where A₀ is the anisotropy in the absence of RelE and A_∞ is the anisotropy at saturating [RelE].

$$\mathit{\text{fraction}} \mathit{\text{bound}}=\left(A_{\infty }-A_o\right)\frac{\left[\mathit{\text{RelE}}\right]}{\left[\mathit{\text{RelE}}\right]+ K_D}+A_o$$ (equation 3)

For intensity measurements, increasing amounts of unmodified mRNA were added to 100 nM RelE. The excitation wavelength was 290 nm. The area under the curve for emission from 300 to 320 nm was corrected for dilution and fit to equation 3 with [RNA] in place of [RelE].

Simulated pH-rate profiles

The dependence of rate on pH for given combinations of pK_as was simulated using partition functions as previously described¹⁷. To test the contributions of linked pK_as to the shape of the curve, the spreadsheet in ref 23 was used.

Calculated Contributions to Catalysis from General Acid and General Base

The fold enhancement on k₁ (the intrinsic rate of the chemical step) comparing two functional groups was calculated using equation 4²⁴.

$$\mathit{\text{fold}} \mathit{\text{enhancement}} \left(k_1\right)= \frac{k_1^{\mathit{\text{ group}} 1}}{k_1^{\mathit{\text{ group}} 2}}={10}^{\left(\beta *\left(p{K}_a^{\mathit{\text{ group}} 2}-p{K}_a^{\mathit{\text{ group}} 1}\right)\right)}$$ (equation 4)

For RelE, the values $\beta =0.5, p{K}_a^{\mathit{\text{ group}} 1}=7.5, and p{K}_a^{\mathit{\text{ group}} 2}=10$ were used to give the following:

$$\mathit{\text{fold}} \mathit{\text{enhancement}} \left(k_1\right)= \frac{k_1^{\mathit{\text{ group}} 1}}{k_1^{\mathit{\text{ group}} 2}}={10}^{\left(0.5*\left(10-7.5\right)\right)}=18$$ (equation 4)

The total enhancement on k_obs, the observed rate of catalysis at a particular pH, may be attenuated by the fraction of the enzyme that is not in the correct protonation state, calculated using equation 5²⁴:

$$k_{obs}=k_1*\frac{1}{1+{10}^{p{K}_a^{\mathit{\text{ base}}}-pH}}*\frac{1}{1+{10}^{pH-p{K}_a^{\mathit{\text{ acid}}}}}$$ (equation 5)

For enzymes with pK_a^base = 5 and pK_a^acid = 10, k_obs = k₁. For pK_a^base = pK_a^acid = 7.5, k_obs = k₁ * ¼. Therefore the net rate enhancement on the observed rate for pK_as near neutrality is 4.5-fold.

Results

R81A defect is rescued by phosphorothiolate modification at the scissile phosphate

Crystal structures and rate measurements of wild-type and mutant RelE have established the importance of several residues located near the scissile phosphate for catalysis^11,12. In particular, R81 has been proposed to act as a general acid to protonate the leaving group oxygen. However, the pre-cleavage structure has an R81A mutation, the post-cleavage structure lacks the leaving group, and several residues have similar rate defects. Therefore, further study is required to determine whether R81 acts as a general acid.

Because the general acid is responsible for protonating the leaving group oxygen, substitution of this oxygen with a leaving group that does not need to be protonated in the rate-limiting step can rescue the rate defect of deleting the general acid. In particular, phosphorothiolate substitution, where the bridging oxygen is replaced by sulfur, has been used to identify the general acid in both protein and RNA enzymes^25–29.

We made the phosphorothiolate substitution at the scissile phosphate of the mRNA substrate for RelE, and determined the rate effects on wild-type and mutant enzymes. At pH 7.5 with unsubstituted mRNA, R81A has more than a 10⁵-fold defect compared to wild type¹². However, R81A is only 10-fold slower than wild type when reacted with a phosphorothiolate mRNA (Table 1). R81A is the only enzyme tested whose rate goes up substantially with the phosphorothiolate relative to the unsubstituted mRNA. A similar uncatalyzed reaction has a 100-fold increase in reaction rate when no general acid is present³⁰. In contrast, other RelE mutants (K54A and R61A, Table 1) catalyze cleavage of the phosphorothiolate and unsubstituted mRNAs at similar rates. The increase in reaction rate for R81A alone upon phosphorothiolate substitution, together with the available structural data, strongly supports the identification of R81 as the general acid.

Table 1.

Rates for cleavage of unmodified (PO) and phosphorothiolated (PS) mRNA by wild-type and mutant RelE at pH 7.5. Rates ± standard deviations are listed, with number of trials in parenthesis.

	PO rate (s−1)a	krel(PO)	PS rate (s−1)	krel(PS)	krel(PS)/ krel(PO)
Wild type	400 ± 100	≡ 1	0.8 ± 0.3 (4)	≡ 1	≡ 1
R81A	0.0053 ± 0.0006	1.3 × 10−5	0.09 ± 0.02 (4)	1 × 10−1	7700
K54A	0.17 ± 0.02	4.2 × 10−4	0.05 ± 0.02 (3)	6 × 10−2	140
R61A	0.00032 ± 0.00008	8.0 × 10−7	0.0004 ± 0.0002 (3)	5 × 10−4	630

^adata from ref 17.

Part of this rescue is due to a 500-fold decrease in the rate of wild-type cleavage of phosphorothiolate mRNA. Phosphorothiolate effects on wild-type systems have varied from a 12-fold enhancement for the Hairpin ribozyme²⁷ to a 6-fold reduction for the Varkud Satellite (VS) ribozyme²⁹. These effects reflect a combination of two factors. First, phosphorothiolate substitution can change the transition state of the cleavage reaction³⁰, which can reduce the effectiveness of active site residues³¹. Second, by alleviating the need for a general acid, only the general base needs to be in the correct protonation state to cleave phosphorothiolate-substituted substrate. For pH values above the general acid pK_a, a higher fraction of enzyme will be active for cleavage of the phosphorothiolate substrate relative to the unmodified substrate. For VS, the general acid pK_a is low and phosphorothiolate substitution increases the fraction correctly protonated ribozyme by 100-fold. Because the observed effect is a 6-fold decrease, the effect on the chemical step is estimated to be a 600-fold decrease. For RelE, if the general acid pK_a were low, then phosphorothiolate substitution would also increase the fraction in the correct protonation state by at least 100-fold. The true effect of the phosphorothiolate on the chemical step would then be at least 50,000-fold, far greater than the effects observed for other systems. Alternatively, the general acid has a high pK_a, and the chemical step is slowed by 500-fold by phosphorothiolate substitution, similar to the value for VS.

Charge at position 81 affects pK_a of general base

The flat pH-rate dependence for wild-type RelE over a wide pH range is consistent with the general acid and general base pK_as being widely separated, but does not distinguish between two kinetically equivalent mechanisms. In mechanism 1, the general acid is higher than the measurable range, and the general base pK_a is perturbed so it is lower than the measurable range. In mechanism 2, the pK_as are reversed so that the acid is low and the base is high. Mutations of charged groups in the RelE active site cause the rate to increase with increasing pH, but it is unclear whether this change in pH dependence is due to the loss of groups with positive charge or an inability to transfer protons.

As a strategy to separate the roles of charge and proton transfer at position 81, we determined the rate and pH dependence of the R81K mutant (Figure 2a). Replacement of arginine with lysine would be expected to result in a lower pK_a. The maximum reaction rate for R81K decreases 200-fold relative to wild type. This could be due to defects in proton transfer, changes in charge distribution at the active site, or structural effects. Importantly, the pH-rate dependence is different from both wild type and R81A, with a decrease in reaction rate with decreasing pH and an observed reaction pK_a of 7.5. This result is consistent with a low general-base pK_a in wild type (mechanism 1) that is dependent on positive charge at position 81 (Figure 2b). If K81 cannot act as a general acid, then R81A and R81K differ mainly by the charge at position 81. Introducing charge to the active site with lysine instead of alanine lowers the general-base pK_a down from above 9 to 7.5.

Figure 2.

pH dependence of R81 mutants. A) WT RelE (squares) has little pH dependence. R81A (triangles) has a log-linear pH dependence with an apparent general base pKa greater than 8. R81K (diamonds), which preserves the charge but not the geometry, has an apparent pKa of 7.5. Error bars correspond to one standard deviation and are not shown when smaller than the points. B-D) Simulated curves showing the shapes of the pH-rate dependence for different combinations of acid and base pKas. Only the relative rates for each pair due to change in protonation state are shown.

If instead the general acid pK_a were low (mechanism 2), replacement of arginine with lysine would drive it even lower. This would retain the flat pH-rate dependence observed in wild-type. If the general acid pK_a were low in wild type and lysine were completely unable to serve as a general acid, then the reaction would have a very high pK_a corresponding to the general base, as in the R81A mutant (Figure 2c). Finally, if the general acid pK_a were high (mechanism 1) but decreased upon substitution with lysine, then the reaction rate should decrease at high pH (Figure 2d). None of these are observed. Instead the observed reaction pK_a increases with decreasing charge at position 81.

RelE binds to mRNA in the absence of the ribosome

RelE only cleaves mRNA bound to the ribosome. This could be simply because the ribosome is required for RelE to associate with mRNA. Alternatively, the ribosome could increase the rate of the chemical step. Based on crystal structures the ribosome does not contribute functional groups to the active site, but it may participate through substrate positioning or other indirect effects¹¹. We sought to determine whether RelE could bind to RNA in the absence of the ribosome using fluorescence anisotropy. We observed a clear, concentration-dependent increase in anisotropy upon addition of RelE (Figure 3). The calculated dissociation constant for the RelE-mRNA interaction is 21 ± 5 μM, which is seven-fold weaker than that for RelE and the ribosome-mRNA complex determined by cleavage kinetics, 3 μM¹². Intrinsic fluorescence of RelE was previously used to monitor the RelE-RelB interaction³² and we used this physical property of RelE to monitor its binding to mRNA. We also saw an increase in RelE fluorescence upon addition of RNA with a similar K_d to that determined by anisotropy (data not shown). This indicates that RelE is capable of binding RNA in the absence of ribosomes, yet there is no indication that the RelE:mRNA binary complex is competent for catalysis.

Figure 3.

Binding of RelE to free RNA. Addition of RelE to Alexa-labeled RNA in the absence of ribosomes causes an increase in fluorescence anisotropy. Error bars correspond to one standard deviation and are not shown when smaller than the points.

Discussion

Shift in pK_a of general base

Although RelE has a similar global fold to other microbial RNases, the active site only contains residues whose unperturbed pK_as are well above neutrality. Previous results indicated that RelE used acid-base catalysis, but those studies could not distinguish between two distinct mechanisms¹⁷. One possibility is that the general base pK_a (proposed to be K54 based on position and pH dependence^12,17) is downshifted below 5.5 so that it is always deprotonated at physiological pH, and the general acid (R81) retains a pK_a above 9 and is always protonated at physiological pH (mechanism 1). Alternatively, the general acid is downshifted while the general base is unperturbed (mechanism 2). Both cases would give identical, flat pH-rate profiles, consistent with the previous observations. The datasets presented here using phosphorothiolate-substituted RNA and mutant enzymes allow us to distinguish between these two models, which has interesting implications for the mechanism of catalysis.

We compared two mutations at position 81 to wild type to determine the role of charge on the pK_a of the general base. When the charge is eliminated in R81A, a positive pH-rate dependence is observed with a reaction pK_a at or above the observable range. In R81K the charge is nearly preserved, and the general base pK_a rises only to 7.5. We conclude that R81 in wild type is essential for pushing the general base pK_a below the observable range (mechanism 1).

If instead the general acid (R81) pK_a were low (mechanism 2), it would have to increase upon substitution of arginine by lysine, opposite of the trend in their unperturbed pK_as³³. Furthermore, R81 would require a pK_a shift of 8 or more units, compared with only 5 units for lysine. Similar shifts for lysine residues have been observed in other systems^18,21. In one case, lysines substituted for residues in the hydrophobic core of staphylococcal nuclease had pK_as as low as 5.3¹⁹. In contrast, arginines were always charged, even when in the hydrophobic core³⁴. No shifts for arginine consistent with the observed pH dependence of RelE have been reported²⁰.

In addition to R81, other residues also yield a positive pH-rate dependence when mutated to alanine¹⁷. R61A, Y87A, and K52A have positive slopes with pK_as of 7, 8.5, and greater than 9, respectively. Higher pK_as indicate the loss of a more important interaction for the pK_a shift of the general base. In particular, K52A has the highest pK_a of any mutant tested, suggesting K52 plays a substantial role in decreasing the pK_a of the general base. If K54 is the general base, then the greater effect of K52 on the general base pK_a could be due to its close proximity. In acetoacetate carboxylase, an adjacent lysine drives down the pK_a of the Schiff-base-forming lysine³⁵. For this and other systems, pK_a perturbation can be accomplished via charge-charge interactions or by positioning effects³⁶. Importantly, arginines at positions 61 and 81 would be expected to retain a positive charge at all measurable pHs, consistent with their role in stabilizing the transition state charge and acting as the general acid, respectively.

Contributions to general acid and base catalysis of widely separated pKas

In textbook cases of acid/base catalysis the pK_as of both groups are close to the reaction pH, yielding the classic bell curve for the pH-rate dependence. This optimizes the contribution of each group, since most groups will be in the correct state, but will also have a low barrier to proton transfer. The contribution to catalysis is related to the pK_a of the candidate catalytic groups and the Brønsted coefficient²⁴. This Brønsted coefficient reflects the extent of proton transfer in the transition state. High coefficients indicate a high degree of proton transfer and will strongly favor catalytic groups capable of transferring a proton.

Even groups with pK_as two or more units removed from the reaction pH, as may be the case here, can provide substantial catalytic power²⁴. This is because in the absence of a general acid or base, proton transfer would have to occur through water molecules. Because the pK_a of water is so far from neutrality, even a titratable group with a pK_a of 5 or 10 positioned for proton transfer provides significant catalytic potential. For a Brønsted value of 0.5, a 2.5-unit difference in pK_a from the reaction pH produces only a 4.5-fold decrease in the observed rate (see Materials and Methods).

This mechanism involving acid-base catalysis by groups with pK_as far from neutrality is most effective when the Brønsted coefficients are small; i.e. when proton transfer is relatively weak. For uncatalyzed reactions of phosphodiesters, both bond breaking and bond formation occur in the rate-limiting transition state^1,37. If proton transfer from the general acid to the leaving group oxygen is weak, then this could lead to an associative transition state with high bond order to the leaving group and a buildup of negative charge on the phosphate. Mutation of positively charged residues or modifying the charge distribution of the phosphate both have significant impacts on RelE catalysis^12,17. In particular, mutation of R61, which contacts the scissile phosphate, has the largest detrimental effect when mutated to alanine. This could be from decreased ability to stabilize increasing negative charge in the transition state. Additionally, the geometry of the phosphate could change during progression to the transition state, again leading to improved stabilization by R61 and other positively charged residues. These charge interactions appear to be more important than acid-base catalysis in RelE.

Binding of RelE to free and ribosome-bound mRNA

RelE binds to free mRNA with only a 7-fold worse affinity than binding to the mRNA:ribosome complex. This indicates that the additional energy of binding to the ribosome is used to drive conformational changes. Indeed, binding of RelE causes a substantial movement of the mRNA in the ribosome, such that the scissile phosphate is in an inline conformation¹¹. It is not known whether mRNA adopts a similar conformation when bound to RelE in the absence of ribosomes. Since the ribosome does not appear to contribute functional groups to catalysis, it is likely that its role is to stabilize a catalytic conformation. In addition to electrostatic stabilization, conformational effects – in particular promoting an inline conformation – could dominate catalysis by RelE.

One possible reason that RelE’s active site differs from other RNases is the requirement for ribosome binding. However, other ribosome-dependent RNases such as YoeB and YafQ have conventional active sites including negatively charged residues and histidines (Figure 4)^38,39. Alternatively, RelE offers a different solution to catalyzing phosphoryl transfer, depending to a larger extent on charge and conformational stabilization. This bears a remarkable resemblance to ribozymes, which lack an equivalent to the histidine side chain and therefore often use functional groups with either perturbed pK_as or with pK_as far from the reaction pH. These strategies in both RNA and protein enzymes are capable of catalyzing phosphodiester transfer at the rates required for biological function.

Figure 4.

A) Proposed mechanism of catalysis by RelE. Limits of observed pK_as are shown, with unperturbed pK_as in italics. The general base has not been definitively established but is shown here as K54 (red), consistent with previous biochemical data. B) Mechanism of catalysis by YoeB, which is also ribosome-dependent. The unperturbed pK_as of the general acid and base are in italics.

ABBREVIATIONS

mRNA

messenger ribonucleic acid

VS

Varkud Satellite ribozyme

MES

2-(N-morpholino)ethanesulfonic acid

MOPS

(3-(N-morpholino)propanesulfonic acid)

EDTA

Ethylenediaminetetraacetic acid

TBE

Tris borate EDTA