The asparagine-transamidosome from Helicobacter pylori: a dual-kinetic mode in non-discriminating aspartyl-tRNA synthetase safeguards the genetic code

Abstract

Helicobacter pylori catalyzes Asn-tRNA^Asn formation by use of the indirect pathway that involves charging of Asp onto tRNA^Asn by a non-discriminating aspartyl-tRNA synthetase (ND-AspRS), followed by conversion of the mischarged Asp into Asn by the GatCAB amidotransferase. We show that the partners of asparaginylation assemble into a dynamic Asn-transamidosome, which uses a different strategy than the Gln-transamidosome to prevent the release of the mischarged aminoacyl-tRNA intermediate. The complex is described by gel-filtration, dynamic light scattering and kinetic measurements. Two strategies for asparaginylation are shown: (i) tRNA^Asn binds GatCAB first, allowing aminoacylation and immediate transamidation once ND-AspRS joins the complex; (ii) tRNA^Asn is bound by ND-AspRS which releases the Asp-tRNA^Asn product much slower than the cognate Asp-tRNA^Asp; this kinetic peculiarity allows GatCAB to bind and transamidate Asp-tRNA^Asn before its release by the ND-AspRS. These results are discussed in the context of the interrelation between the Asn and Gln-transamidosomes which use the same GatCAB in H. pylori, and shed light on a kinetic mechanism that ensures faithful codon reassignment for Asn.

INTRODUCTION

Accurate expression of genetic information is essential to all living organisms. Integrity of the message encoded within a gene is safeguarded by mechanisms that cells have evolved at each step of gene expression. In protein synthesis, tRNA aminoacylation is a crucial step where each tRNA species is bound to its cognate amino acid in reactions catalyzed by the aminoacyl-tRNA synthetases (aaRSs) (1,2). The amino acids are first activated as aminoacyl∼AMPs and then transferred onto their cognate tRNAs to synthesize the final aminoacyl-tRNA products. The latter are then carried to the ribosome by an elongation factor where they are used to decode the codons on the mRNA. Fidelity of translation depends on accurate tRNA aminoacylation, i.e. correct matching of an amino acid to its set of isoacceptor tRNAs. This coupling is determined by the existence of a particular aaRS for each amino acid/tRNA pair. The capacity to select homologous substrates is conferred by active site geometry and by correction processes that are used by aaRSs when the wrong aminoacyl-tRNAs are formed (3,4). Moreover, aaRSs are split into two classes, according to sequence/structure (5) and kinetic (6–8) properties. Members of class I bind tRNA on the minor grove side and acylate the 2′-OH of the terminal adenosine whereas the members of class II approach tRNA from the major grove side and transfer the amino acid moiety on the 3′-OH of the terminal adenosine. Further, it seems that most class I aaRSs display a rate-limiting step for the release of aminoacyl-tRNA, whereas the transfer of the activated amino acid onto tRNA is rate-limiting for class II aaRSs (8).

Global genome analyses and biochemical data indicate that about half of prokaryotes are deprived of asparaginyl-tRNA synthetase (AsnRS), and that the vast majority lack a glutaminyl-tRNA synthetase (GlnRS). This pattern negates the old ‘dogma’ that postulated the presence of a specific aaRS for each amino acid and tRNA pair (9,10). These organisms, including mitochondria (11), all synthesize amidated aminoacyl-tRNAs thanks to an enzymatic system that includes a non-discriminating (ND) aaRS, ND-aspartyl-tRNA synthetase (ND-AspRS) and/or ND-glutamyl-tRNA synthetase (ND-GluRS), providing a misacylated aminoacyl-tRNA species (Asp-tRNA^Asn or Glu-tRNA^Gln); these aminoacyl-tRNAs are then modified into Asn-tRNA^Asn or Gln-tRNA^Gln through a phosphorylation/transamidation mechanism catalyzed by a tRNA-dependent amidotransferase called GatCAB in bacteria (12–20). The synthesis of Asn-tRNA^Asn and Gln-tRNA^Gln is ensured by two separate bi-enzymatic systems, consisting in functionally related enzymes brought together by a scaffold/substrate tRNA. The resulting aaRS/tRNA/GatCAB complexes are named transamidosomes. These particles vary in cohesive strength, with the Asn-transamidosome of Thermus thermophilus (21) showing the highest stability while the Gln-transamidosomes of Thermotoga maritima and Helicobacter pylori are more dynamic (22–24). Within all of these complexes, the tRNA moiety can be used either as a scaffold for the nucleoprotein to form Asn or Gln or as a substrate for their synthesis. This makes the misacylation/transamidation system a matrix-assisted amino acid biosynthesis pathway of particular interest in H. pylori, a well-known and widespread human pathogen, which uses these indirect pathways to form Asn, Gln-tRNA^Gln and Asn-tRNA^Asn (24–28).

Although it was shown that mischarged Asp-tRNA^Asn and Glu-tRNA^Gln species are deprived of significant affinity for the elongation factor EF-Tu (29,30), several investigations have shown that they can lead in vivo to incorporation of Asp (31,32) or Glu (33–35) in place of Asn and Gln into proteins. Overexpression of either misacylating aaRS is toxic in Escherichia coli (31–35) while in yeast a disabled amidotransferase resulted in significant incorporation of Glu at Gln codons in mitochondrial proteins (36). Asn- and Gln-transamidosomes, which control the release of misacylated intermediates, may therefore be another layer of aminoacyl-tRNA quality control.

Here, we report how this quality control is achieved for Asn-tRNA^Asn synthesis in H. pylori through the formation of an Asn-transamidosome. Moreover, we show that ND-AspRS exhibits surprisingly sensitive kinetic properties for Asp-tRNA^Asp and Asp-tRNA^Asn formation: fast release of Asp-tRNA^Asp for translation and slow release of Asp-tRNA^Asn to allow transamidosome formation by GatCAB binding. This safeguard mechanism resembles an editing process and may shed light on evolutionary processes that led to Asn infiltration into the present-day genetic code. Our results suggest that, in H. pylori, ND-AspRS and GatCAB co-evolved to generate an appropriate kinetic system able to reduce translational error level and to ensure a faithful codon reassignment for Asn.

MATERIALS AND METHODS

Overproduction and purification of enzymes and tRNAs

The H. pylori (Hp) ND-AspRS and GatCAB were overproduced from the cloned aspS, gatC, gatA and gatB genes as described (28,32). For protein purification, the cells from an overnight culture were harvested by centrifugation at 4000g, washed in cold Tris–HCl buffer 100 mM, pH 8.0, sedimented and suspended in the cell disruption buffer (1/3, w/v) formed by supplementing the above buffer with 5 mM 2-mercaptoethanol, 0.1 mM Na₂EDTA, 1 mM benzamidine and 1 mM of protease inhibitors [4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride, AEBSF, Pefablock]. The cell extract, obtained by sonication of the cell suspension and centrifugation at 105 000g, was dialyzed against potassium phosphate buffer 20 mM, pH 7.2, containing 0.5 mM Na₂EDTA and 5 mM 2-mercaptoethanol, and adsorbed on a DEAE-cellulose column followed by elution with a linear gradient of potassium phosphate from 25 to 250 mM pH 7.2.

The AspRS-containing fractions, eluted between 130 and 150 mM salt, were then adsorbed on a Ni-NTA column in 50 mM Tris–HCl buffer pH 7.5 before elution of the proteins with a linear gradient from 25 to 450 mM of imidazole. Pure AspRS (10 mg at 95% purity) were obtained from 30 g cells. The GatCAB containing fractions eluted on DEAE-cellulose from 160 to 190 mM potassium phosphate were adsorbed on a Ni-NTA column in 50 mM Tris–HCl buffer pH 7.5. After washing with the buffer containing 25 mM imidazole, GatCAB was eluted with 450 mM imidazole. Pure GatCAB (50 mg at 95% purity) was obtained from 80 g of cells.

The reconstituted tRNA^Asn gene from T. thermophilus was cloned in the pKK223 vector and purified as described (21). The E. coli (Ec) tRNA^Asp gene was amplified from genomic DNA and cloned into the pUC18 vector for expression in the E. coli JM103 strain. Transfer RNA^Asp (800 mg from 120 g of cells; accepting capacity 30.6 nmol.mg⁻¹) was obtained by phenol extractions followed by chromatographies on DEAE-cellulose and nmol.mg Sepharose 4B as described (37). All tRNA^Asp and tRNA^Asn mutants used in this study were cloned into the pUC18 vector downstream a T7 promoter, transcribed and purified as described (38).

Gel-filtration

Experiments were conducted as described (21) using an ÄKTA Purifier and a 24-ml Superdex G200 column (GE Healthcare) equilibrated with 50 mM Na–HEPES buffer pH 7.2 containing 30 mM KCl, 6 mM MgCl₂, 0.1 mM Na₂EDTA and 5 mM 2-mercaptoethanol at 12°C. Samples of 0.2 ml containing the enzyme and tRNA partners at the indicated concentrations were diluted in the equilibration buffer and analyzed. K_D values [free enzyme] × [free tRNA^Asn]/[enzyme/tRNA^Asn complex] were determined by evaluation of the areas of enzyme-bound tRNA and free tRNA peaks, the sum of which equals the total tRNA present in the sample. The contaminating non-specific tRNA was subtracted from free tRNA. The quantity of free enzyme was determined by subtracting the quantity of enzyme-bound tRNA from total amount of enzyme in the sample.

Dynamic light scattering

Dynamic light scattering (DLS) experiments were conducted as described (21) at 20 ± 0.1°C in a Zetasizer NanoS instrument (Malvern, UK). The 20- to 40-µl samples contained 10 µM ND-AspRS (20 µM subunits) and/or 20 µM GatCAB with or without 20 µM tRNA^Asn in the gel-filtration buffer. When present, ATP, Asp and Gln were at a concentration of 2 mM. Hydrodynamic diameters were corrected for solvent refractive index (n = 1.3353) and absolute viscosity (η = 1.041 mPa.s).

Aminoacylation of tRNA

The reaction mixture contained 100 mM Na–HEPES buffer pH 7.2, 30 mM KCl, 12 mM MgCl₂, 2 mM ATP, 20–50 µM [¹⁴C] L-Asp (330 cpm.pmol⁻¹, GE Healthcare), 0.1 mg.ml⁻¹ bovine serum albumin, 10 µM T. thermophilus (Tt) tRNA^Asn or Ec tRNA^Asp, or 0.04–6 µM Tt tRNA^Asn for K_M measurements, and 0.01–0.5 µM Hp ND-AspRS. The reactions were conducted at 37°C unless otherwise indicated. The rate-limiting step was determined with 0.5, 1.0 and 1.5 µM Hp ND-AspRS and 10 µM tRNA^Asn or tRNA^Asp without or with GatCAB at the indicated concentrations in the presence of 15% glycerol at 4°C. For K_M measurements of Asp and ATP in the presence of tRNA^Asp or tRNA^Asn, the small substrate concentrations varied from 10 to 300 µM and tRNA was present at a 10 µM concentration. Inhibition experiments with aspartol-adenylate (Asp-ol-AMP), kindly provided by Prof. R. Chênevert (Université Laval, Québec), were performed with concentrations ranging from 10 to 200 µM. The inhibition constants were determined as described (39). The [¹⁴C] aminoacyl-tRNA formed in 10 or 20 µl aliquots withdrawn at various time intervals was determined as described (1).

ATP-PPi exchange

The reaction mixture contained 100 mM Na–HEPES buffer pH 7.2, 10 mM MgCl₂, 2 mM L-Asp, 2 mM ATP, 2 mM [³²P] PPi (2 cpm.pmol⁻¹, Perkin Elmer) and 0.5 µM ND-AspRS. The [³²P] ATP formed at 37°C was determined in 40 µl aliquots at various time intervals as described (1).

Transamidation of Asp-tRNA^Asn

Reactions were conducted at 37°C for 10 min in a mixture containing 100 mM Na–HEPES buffer pH 7.2, 30 mM KCl, 12 mM MgCl₂, 2 mM ATP, 10 µM [¹⁴C] L-Asp (330 cpm.pmol⁻¹), 1 mM L-Gln, 0.1 mg.ml⁻¹ bovine serum albumin and 10 µl of the transamidosome fraction eluted from size-exclusion chromatography. Transamidation was measured as described (21,24).

RESULTS

An Asn-transamidosome in H. pylori

Association of ND-AspRS, tRNA^Asn and GatCAB from T. thermophilus leads to a stable ternary complex named Asn-transamidosome (21). Gel-filtration and DLS experiments conducted with the H. pylori partners show that the ND-AspRS can bind both tRNA^Asp and tRNA^Asn but with different affinities (K_D values of 7.9 and 21.4 µM respectively, Figure 1A) and that GatCAB can form a much more stable binary complex with tRNA^Asn than ND-AspRS (K_D value of 2.1 µM, Figure 1B). These divergent K_D values differ from those obtained using the T. thermophilus partners, where tRNA^Asn bound to ND-AspRS with higher affinity than to GatCAB (21). The asymmetric K_D values determined with the H. pylori system were confirmed by DLS (Figure 2). Indeed, a higher-sized particle was clearly seen in the presence of ND-AspRS and tRNA^Asp (12.6 nm), compared to free ND-AspRS and tRNA (10.9 and 4.9 nm, respectively) but not when ND-AspRS and tRNA^Asn were mixed (Figure 2, lanes 1–4) (10.9 nm). Similarly, we detected an association between GatCAB and tRNA^Asn (10.9 nm) compared to their isolated counterparts (9.4 and 4.9 nm, respectively) (Figure 2, lanes 5 and 6). No association of the protein partners in the absence of tRNA was detected by either technique (data not shown). Finally, when the three partners were mixed, a new ribonucleoprotein (RNP) of significantly higher size (13.5 nm) appeared (Figure 2, lane 7). Isolation of this complex by gel-filtration (Figure 1C) and analysis of its components by SDS–PAGE (Figure 1D) revealed the presence of ND-AspRS, the three GatCAB subunits and tRNA^Asn. Functional analysis confirmed that it was fully able to synthesize Asn-tRNA^Asn in the presence of free Asp, ATP and Gln (Figure 1E). Because the association of the protein partners is tRNA^Asn-dependent, this complex constitutes a bona fide Asn-transamidosome according to previous arguments (21).

Figure 1.

Size-exclusion chromatography shows formation of a transamidosome complex by ND-AspRS, GatCAB and tRNA^Asn. (A) Comparison of isolated ND-AspRS and ND-AspRS/tRNA^Asp and ND-AspRS/tRNA^Asn complexes. The observed association enabled an estimation of the K_D value for binding of each tRNA to ND-AspRS (see ‘Materials and Methods’ section). K_D values for the binding to ND-AspRS are 7.9 and 21.4 µM for tRNA^Asp and tRNA^Asn, respectively. (B) Comparison between free GatCAB and GatCAB/tRNA^Asn complexes. The K_D value for this association is 2.1 µM. (C) Elution profile of a mixture containing GatCAB, ND-AspRS and tRNA^Asn. (D) SDS–PAGE profile of a SEC fraction (lane 3) compared to GatCAB alone (lane 1) and ND-AspRS alone (lane 2). (E) TLC plate demonstrating that the complex eluted in the first peak (termed ‘isolated’) is able to produce Asn-tRNA^Asn alone, or when supplemented with excess tRNA^Asn (termed ‘+2 µM tRNA’), compared to a negative control where GatCAB was omitted. All partners were mixed to a final concentration of 20 µM each. All K_D values determined varied within 10%.

Figure 2.

Characterization of the complexes involving Hp ND-AspRS, GatCAB, tRNA^Asn and tRNA^Asp by DLS and analysis of the stability of the transamidosome under catalytic and non-catalytic conditions. Each value is an average of five independent measurements with errors <5%. All the experiments were conducted under the same conditions as size chromatography experiments (see ‘Materials and Methods’ section). Ternary 1, 2 and 3 all refer to the transamidosome under non-catalytic (1), aspartylation (2) and aspartylation/transamidation (3) conditions.

Dynamics of the H. pylori Asn-transamidosome

In T. thermophilus, the Asn-transamidosome is characterized by its remarkable stability in gel filtration (21,22), whereas for H. pylori, it is less stable (Figure 1C). Since this observation could reflect a difference in dynamic properties, we tested the stability of the Hp Asn-transamidosome by measuring its hydrodynamic diameter in the presence of small substrates, e.g. under catalytic conditions. The particle size decreased from 13.6 nm under non-catalytic conditions to 10.9 nm under aspartylation conditions (i.e. presence of Asp and ATP), and from 13.6 to 12.9 nm under transamidation conditions (i.e. presence of Asp, ATP and Gln). Furthermore, gel-filtration with a mix of GatCAB, ND-AspRS and preformed Asp-tRNA^Asn revealed elution of the binary GatCAB/Asp-tRNA^Asn complex and a separate ND-AspRS peak (data not shown). These results suggest that charging of tRNA^Asn may trigger destabilization of the transamidosome (Figure 2, lanes 7–9), apparently through release of ND-AspRS. Hence the Hp Asn-transamidosome displays completely different dynamics than does its T. thermophilus counterpart (21).

Helicobacter pylori ND-AspRS, an enzyme with dual kinetic behavior

Table 1 summarizes the steady-state kinetic parameters of Hp ND-AspRS for aminoacylation of tRNA^Asp and tRNA^Asn. The enzyme aspartylates tRNA^Asp 2.5-fold faster than tRNA^Asn (0.33 and 0.14 s⁻¹), as reported previously (32). However, pre-steady-state kinetics conducted under conditions that allowed an examination of the first catalytic cycles of the enzyme (4°C and in the presence of 10 or 15% of glycerol) revealed intriguing differences in aminoacylation of both tRNAs since in contrast to tRNA^Asp the charging kinetics of tRNA^Asn were biphasic. Aminoacylation of tRNA^Asp remains linear after completion of the first catalytic cycle. Thus, this first cycle occurs with the same rate as subsequent ones (0.041 s⁻¹) (Figure 3). Since ATP-PPi exchange rate is significantly faster than tRNA charging (15.9 and 0.33 s⁻¹ at 37°C), the steady-state rate of tRNA^Asp aminoacylation is dictated by the rate of transfer of activated Asp onto tRNA^Asp. Surprisingly, when tRNA^Asn is used as a substrate, biphasic kinetics arise, which exhibit a burst of Asp-tRNA^Asn formation (∼0.04 s⁻¹) followed by a significantly slower linear phase (0.0033 s⁻¹) (Figure 3). Extrapolation of the linear phase at t₀ points to formation of one Asp-tRNA^Asn per ND-AspRS active site during the fast phase (Figure 3) and suggests that the first tRNA^Asn is aspartylated significantly faster than those following. Interestingly, the first tRNA^Asn is aminoacylated with a rate equivalent to that observed for tRNA^Asp (Figure 3). Burst and steady-state rate values are both dependent on enzyme concentration (Supplementary Figure S1A). In the slow phase, the steady-state rate of tRNA^Asn charging increases with the pH but not significantly with the ionic strength (Supplementary Figure S1B). This kinetic behavior is consistent with the release of Asp-tRNA^Asn being the rate-limiting step at the steady-state of the reaction (40). The slow dissociation of the Asp-tRNA^Asn product agrees with the absence of detectable hydrolysis of its ester bond in the presence of ND-AspRS (28).

Table 1.

Kinetic constants for aspartylation of tRNA^Asp and tRNA^Asn and a tRNA variant with H. pylori ND-AspRS in the absence and presence of GatCAB

	−GatCAB		Burst	+GatCAB
	kcat (s−1)	KM (µM)		kcat (s−1)	KM (µM)
tRNAAsp	0.33 ± 0.01	n.d.	−	0.31	n.d.
tRNAAsn	0.14 ± 0.02	0.08 ± 0.01	+	0.25 ± 0.03	0.3 ± 0.03
yeast tRNAAsp	0.29 ± 0.03	n.d.	−	n.d.	n.d.
tRNAAsn (G1–C72, U20A)	0.26 ± 0.02	n.d.	+	0.28 ± 0.02	n.d.

Burst indicates the presence of an initial fast phase of tRNA aminoacylation.

Figure 3.

Aminoacylation kinetics for tRNA^Asp and tRNA^Asn with Hp ND-AspRS: a dual response to substrates. Experiments were conducted in 15% glycerol at 4°C for tRNA^Asp (black diamonds) and tRNA^Asn (gray triangles). Results show that tRNA^Asp and tRNA^Asn are aminoacylated differently. In case of tRNA^Asp, the first cycle (1 Asp-tRNA formed) has the same rate as the subsequent ones (0.04 s⁻¹). However, in case of tRNA^Asn, the first cycle (0.04 s⁻¹), which is equivalent to that for tRNA^Asp, is significantly faster than the subsequent ones (0.0033 s⁻¹). Extrapolation of this slower phase at t₀ points to the formation of 1 Asp-tRNA^Asn per ND-AspRS active site. This pattern suggests a rate-limiting step which would be the release of Asp-tRNA^Asn. SE was <5% on all values.

GatCAB modifies the kinetic properties for tRNA^Asn

Table 1 shows the effect of GatCAB on kinetic constants of Hp ND-AspRS for tRNA^Asn and tRNA^Asp aspartylation. The addition of GatCAB increases the steady-state rate of Asp-tRNA^Asn formation according to a hyperbolic curve (Figure 4A), but has no effect on the steady-state rate of tRNA^Asp charging (Figure 4A, inset and Table 1), indicating that this phenomenon is specific to tRNA^Asn. When saturating concentrations of GatCAB are reached, the rate of tRNA^Asn aspartylation increases 1.8-fold (0.14–0.25 s⁻¹) and nearly fits that of tRNA^Asp (0.33 s⁻¹, Table 1), while the K_M value of ND-AspRS for tRNA^Asn increases 3.8-fold, leading to an increase in the overall charging efficiency (k_cat/K_M) of 1.8-fold (Table 1). Thus, in the presence of GatCAB, aminoacylation of tRNA^Asn occurs with a similar efficiency as that of tRNA^Asp (k_cat/K_M, respectively of 0.27 and 0.30 µM⁻¹ s⁻¹). Pre-steady-state kinetics reveal that this GatCAB-mediated effect originates in an increase in the rate of aminoacylation during the slow phase (0.0033–0.012 s⁻¹, Figure 5A and B). The rate of the fast phase does not seem to be affected (Figure 5B). Considering that tRNA^Asn binds GatCAB and ND-AspRS with K_D values of 2.1 and 21.4 µM, respectively (Figure 1), and that aspartylation of tRNA^Asn is more efficient when it is ‘labeled’ with or ‘presented’ by GatCAB (Table 1), the GatCAB/tRNA^Asn complex may constitute a better substrate for misacylation by the ND-AspRS. We named this complex the tRNA^Asn-presentation complex (tRNPC). Since this complex is non-productive in absence of ND-AspRS and can also be a substrate, tRNPC can be considered as a bona fide transfer ribonucleoprotein (tRNP).

Figure 4.

Effects of GatCAB on aminoacylation of tRNA^Asn using Hp ND-AspRS versus Ec D-AspRS. The graph shows the effect of increasing GatCAB concentrations on initial rates of both AspRSs in the presence of saturating tRNA^Asn concentration. Results (v) were normalized using the steady-state rate value of the corresponding AspRS in absence of GatCAB (v₀). Concentration of the GatCAB/tRNA^Asn complex increases together with GatCAB concentration. (A) GatCAB increases the steady-state rate of Hp ND-AspRS when tRNA^Asn is present (gray triangles), but not when tRNA^Asp is used (inset). Error bars represent the SD from three independent experiments. (B) GatCAB decreases the steady-state rate of tRNA^Asn misacylation with Ec D-AspRS.

Figure 5.

GatCAB influences aminoacylation kinetics for tRNA^Asn with Hp ND-AspRS. Pre-steady-state kinetics were conducted under the same conditions of pre-steady-state as previously (4°C, 15% glycerol). (A) In the presence of saturating GatCAB concentration (10 µM), the steady-state rate of tRNA^Asn misacylation increases and the burst value slightly decreases. These results suggest that GatCAB increases the dissociation of Asp-tRNA^Asn. (B) Zoom on the 4 first aminoacylation cycles (in the presence of GatCAB), corresponding to the gray zone in (A). SE was <5% on all values.

Interestingly, the tRNPC has opposing effects on the discriminating AspRS of E. coli (D-AspRS) (Figure 4B) in vitro. Ec D-AspRS is able to aspartylate tRNA^Asn, in addition to its cognate tRNA^Asp, albeit much less efficiently (14). As shown for other mischarging kinetics of non-cognate tRNAs, the transfer step is rate-limiting (41). GatCAB decreases the rate of aminoacylation of tRNA^Asn by the D-AspRS (Figure 5B). This result can be interpreted through the strong binding of tRNA^Asn on GatCAB and the low affinity of the D-AspRS for tRNA^Asn. In contrast to the H. pylori system, GatCAB decreases the rate of Asp-tRNA^Asn formation by Ec D-AspRS presumably through sequestration of tRNA^Asn limiting the free tRNA^Asn available for aminoacylation. Thus, the enhancing effect of GatCAB is restricted to tRNA^Asn aspartylation by ND-AspRS.

In vivo experiments confirmed that expression of Hp ND-AspRS in the transformed E. coli ER Asn-auxotrophic strain was toxic presumably because of the accumulation of Asp-tRNA^Asn (Supplementary Figure S2), as previously shown (32). An antibiogram-like test with an IPTG-soaked paper disk clearly triggers a growth defect phenotype compared to controls (Supplementary Figure S2A–C). Co-expression of Hp GatCAB reduces this toxic effect, since the enzyme is able to convert Asp-tRNA^Asn into Asn-tRNA^Asn allowing growth of the recombined strain (Supplementary Figure S2D), although with lesser efficiency.

GatCAB helps to release Asp-tRNA^Asn from the ND-AspRS active site

The acceleration of the steady-state rate of tRNA^Asn aspartylation by GatCAB (Figure 5) suggests that GatCAB controls the release of Asp-tRNA^Asn from the ND-AspRS active site. To test this hypothesis, we used a strategy to nullify this effect without altering GatCAB or ND-AspRS. In other words, we asked whether this effect was solely due to the ability of GatCAB to bind Asp-tRNA^Asn. To this end, we investigated aspartylation of tRNA^Asn variant mutated at positions known to determine recognition by GatCAB (38). The U₁–A₇₂ base-pair constitutes the major transamidation determinant, but recognition of tRNA^Asn is further favored by the absence of the U_20A position in the D-loop (38). Importantly, these positions are not critical for ND-AspRS recognition (38,42). Thus, we constructed a G₁–C₇₂:U_20A-containing tRNA^Asn variant and verified that it is a suitable substrate for ND-AspRS but is not recognized by GatCAB (Figure 6 and 38). The aminoacylation kinetics of this tRNA^Asn variant remained biphasic and saturating levels of GatCAB had no effect on the steady-state rate of aminoacylation (Figure 6A and B). Thus, removing the ability of GatCAB to bind this mutant Asp-tRNA^Asn eliminated the impact of GatCAB on ND-AspRS kinetics. These results indicate that GatCAB enhances ND-AspRS kinetics by recognizing the wild-type aspartylated extremity of tRNA^Asn according to the transamidation determinants, consequently freeing the ND-AspRS active site. As a consequence, the overall turnover of ND-AspRS increases in the presence of GatCAB.

Figure 6.

GatCAB pulls Asp-tRNA^Asn from the ND-AspRS active site. (A) Pre-steady-state kinetics performed with a tRNA^Asn mutated at the GatCAB identity elements: G₁–C₇₂:U_20A replace the natural U₁–A₇₂:Δ_20A positions. Addition of GatCAB (black diamonds, +GatCAB) has no effect on the steady-state rate, compared to a reaction without GatCAB (gray triangles, −GatCAB). (B) Pre-steady-state kinetics performed with a wild-type tRNA^Asn. Addition of GatCAB (black diamonds, +GatCAB) accelerates the steady-state rate, compared to a reaction without GatCAB (gray triangles, −GatCAB).

Origin of the rate-limiting step

What are the structural elements that trigger different aminoacylation mechanisms with respect to tRNA^Asp and tRNA^Asn? The major differences between bacterial tRNA^Asp and tRNA^Asn lie within the acceptor arm (displaying a G₁–C₇₂ or a U₁–A₇₂ base pair, respectively) and the anticodon loop (positions C₃₆ and U₃₆ and C₃₈ and A₃₈ in tRNA^Asp and tRNA^Asn, respectively). Yeast cytoplasmic tRNA^Asp provides a natural variant that displays a GUC ‘Asp’ anticodon and a bacterial ‘tRNA^Asn-like’ acceptor-end that almost avoids aminoacylation by Ec D-AspRS (43). Similar to bacterial tRNA^Asp, aminoacylation of yeast tRNA^Asp is monophasic (data not shown). Thus, the elements leading to the biphasic kinetics for aminoacylation of tRNA^Asn by the Hp ND-AspRS are absent in yeast tRNA^Asp. Further, substitution of C₃₆ by U₃₆ in bacterial tRNA^Asp leading to a GUU ‘Asn’ anticodon did not trigger a biphasic kinetic response (data not shown). Hence, the divergent kinetic responses of Hp ND-AspRS for tRNA^Asp versus tRNA^Asn aminoacylation are probably triggered by different overall conformations of the aminoacyl-tRNA end-products, rather than by distinct affinities for the tRNA substrates. This hypothesis is in line with the fact that the two k_cat's of aspartylation (Table 1) show opposing trends with respect to the affinities of the enzyme for the two tRNAs (Figure 1). Moreover, the steady-state rate of tRNA^Asn aspartylation is more affected by increase of the pH than by that of the ionic strength (Supplementary Figure S1), suggesting that the aminoacylated tRNA^Asn extremity, and mostly the Asp moiety, plays a prevalent role in the slow dissociation of Asp-tRNA^Asn, in accordance with other studies on the rate-limiting step of tRNA aminoacylation (40) and the inhibition properties of Asp-ol-AMP. Interestingly, this non-hydrolyzable Asp-AMP mimic, competitive with respect to Asp and ATP, shows different inhibition patterns for aminoacylation of tRNA^Asp and tRNA^Asn (39 and Supplementary Table S1). Indeed K_i values of 8 and 32 µM were determined in presence of tRNA^Asp and tRNA^Asn, respectively (Supplementary Table S1). This 4-fold increase in the K_i value when Asp-tRNA^Asn is formed suggests a strong binding of the tRNA^Asn-bound Asp moiety within ND-AspRS active site, compared to Asp-tRNA^Asp. This is further confirmed with differential inhibition patterns using Asp-ol-AMP (Supplementary Table S1) in presence of tRNAAsp or tRNAAsn, the latter tRNA protecting the active site from inhibition more efficiently as a consequence of its retention. This is in agreement with a previous work on Pseudomonas aeruginosa ND-AspRS (39) showing that the limiting step for Asp-tRNAAsn synthesis could be general for ND-AspRS of bacterial type.

To obtain further insights into the origin of the distinct dissociation rates of the two aminoacyl-tRNAs from ND-AspRS, we performed Arrhenius activation energy (E_A) measurements (Supplementary Figure S3). The E_A values are the same for aminoacylation of tRNA^Asp (68.6 ± 0.6 kJmol⁻¹K⁻¹) and tRNA^Asn (66.9 ± 1.9 kJmol⁻¹K⁻¹). Thus, both E_A are determined by the same step, presumably the transfer step, since the release of aminoacyl-tRNA does not contribute significantly to E_A. From these similar E_A values, one can suggest that the steps preceding the release of Asp-tRNA are equivalent and that the differences occur after the transfer step. It is therefore reasonable to assume that the formation of Asp-tRNA^Asp, or Asp-tRNA^Asn trigger different structural arrangements of the active site. This conformational difference may explain the difference in Asp-tRNA release and in inhibition patterns by Asp-ol AMP.

DISCUSSION

The Asn-transamidosome in H. pylori

Previous work has suggested that an Asn-transamidosome should exist in H. pylori (28). Indeed, this organism does not import Asn (44), and given the fact that no Asn synthetase or AsnRS could be detected in the genome using BLAST, the indirect misacylation/transamidation pathway seems to be the sole route for Asn and Asn-tRNA^Asn biosynthesis, as is also the case for Gln-tRNA^Gln (24). Hp ND-AspRS and GatCAB participate together with tRNA^Asn in the formation of a ribonucleoprotein homologous to that shown for T. thermophilus (Figures 1 and 2). Contrary to the T. thermophilus Asn-transamidosome (21), we noticed that, in the Hp system, uncharged tRNA^Asn has a 10-fold stronger affinity for GatCAB (the second enzyme of the pathway) than for ND-AspRS (the first enzyme in the pathway) (Figures 1 and 2). Thus, in H. pylori, ND-AspRS presumably docks onto the tRNPC, rather than to free tRNA^Asn (Scheme 1). In conditions where GatCAB is in excess, tRNA^Asn may thus be mostly presented by GatCAB, ensuring the proper channeling of Asp-tRNA^Asn toward GatCAB. Consistently, the tRNPC seems to be a better substrate for aminoacylation than free tRNA^Asn (Table 1 and Figure 4). Upon aspartylation, the Asp-tRNA^Asn acceptor-end shifts toward the GatCAB active site and triggers the release of ND-AspRS from the complex (Figure 2, lanes 7–9), explaining why the steady-state rate of ND-AspRS increases in the transamidosome. Indeed, when GatCAB is present, Asp-tRNA^Asn gets pulled out of the ND-AspRS active site, freeing ND-AspRS from its product and allowing a new catalytic cycle to occur as evidenced by the faster steady-state rate of aspartylation.

Scheme 1.

Formation of Asn-tRNA^Asn in H. pylori. (A) The major pathway used in H. pylori (the main ‘Asn-transamidosome pathway’) starts with (1) the formation of the tRNPC, linking GatCAB and tRNA^Asn, onto which (2) ND-AspRS can then dock to form the Asn-transamidosome. In this ribonucleoprotein, tRNA^Asn gets aspartylated more efficiently (3), and the aspartylated CCA-end of tRNA^Asn shifts from the ND-AspRS active site to the GatCAB phosphorylation/amidation active site. Asn-tRNA^Asn biosynthesis proceeds thanks to amide donor hydrolysis (here free Gln) to provide the ammonia moiety necessary for transamidation (4) (28). When Asn-tRNA^Asn is formed, it is released together with ND-AspRS, and used in protein synthesis (5). (B) When GatCAB is not in sufficient amounts to produce enough tRNPC, free tRNA^Asn binds to ND-AspRS (6) and gets aminoacylated less efficiently (7). The dissociation of Asp-tRNA^Asn is rate-limiting, but if Asp-tRNA^Asn is released (8), GatCAB can still recognize it (9) and transamidate it (10) (28) into Asn-tRNA^Asn, which can be released to fuel translation (11). (C) Since the release of Asp-tRNA^Asn from ND-AspRS is rate-limiting, Asp-tRNA^Asn remains within the ND-AspRS active site for longer times, enabling free GatCAB to bind to the ND-AspRS/Asp-tRNA^Asn complex (12) to form an Asn-transamidosome, in which Asn-tRNA^Asn can be synthesized through transamidation (4) rejoining the main Asn-transamidosome cycle. This latter mechanism can be qualified as a ‘rescue mechanism’. The table summarizes the different pathway characteristics. The aspartylation and transamidation steps are highlighted to stress the fact that the Asn-transamidosome pathway couples both reactions, while they are sequential when GatCAB is absent or when the tRNPC is not involved.

This scheme strongly differs from that determined for the T. thermophilus Asn-transamidosome, where the complex remains stable over the entire catalytic cycle, according to a peculiar flip-flop mechanism involving a non-substrate scaffold tRNA (_scaftRNA) (22). Indeed, Tt ND-AspRS, unlike Hp ND-AspRS, changes its release mode and retains Asp-tRNA^Asn, to preserve the tRNP stability when GatCAB is present, e.g. within the complex. These major differences in stability and in mechanism could be due to the differences between Tt and Hp ND-AspRSs. Indeed, while Tt ND-AspRS belongs to the archaeal structural-type, Hp ND-AspRS falls into the bacterial structural-type which contains an additional domain, called the GAD domain. This domain is appended to the active site where it could prevent the scaffold tRNA from playing its role, or it could destabilize the overall complex (22). Either of these effects would result in lesser stability and greater dynamics for the bacterial-type Asn-transamidosome. On the other hand, while Tt ND-AspRS does not retain its Asp-tRNA^Asn product in the absence of GatCAB (22) (i.e. outside the complex), Hp ND-AspRS might be able to trap this aminoacyl-tRNA thanks to the GAD domain, which lies next to the active site in a position where an editing domain is found in the class II homologous ThrRS (45)

ND-AspRS displays a dual mode of release for its two Asp-tRNA products

ND-AspRS recognizes and aspartylates both tRNA^Asp and tRNA^Asn equivalently, which makes it a ND enzyme for substrate recognition and aminoacylation. Nevertheless, kinetic studies clearly show a difference in product release. Asp-tRNA^Asn is retained upon aminoacylation, while Asp-tRNA^Asp is not (Figure 3). This difference seems to be triggered by the overall conformation of tRNA rather than by specific identity elements. Our results suggest that the shapes of tRNA^Asp and tRNA^Asn could induce different active site configurations that lead to these divergent modes of release. According to the predominant effect of pH on the steady-state rate (slower phase) (Supplementary Figure S1), the CCA-Asp moiety may play an important role in Asp-tRNA^Asn retention on ND-AspRS. This is further confirmed with differential inhibition patterns using Asp-ol-AMP (Supplementary Table S1) in presence of tRNA^Asp or tRNA^Asn, the latter tRNA protecting the active site from inhibition more efficiently as a consequence of its retention. This is in agreement with a previous work on P. aeruginosa ND-AspRS (39) showing that the limiting step for Asp-tRNA^Asn synthesis could be general for ND-AspRS of bacterial type.

This scheme strongly shares analogy with editing mechanisms, where a tRNA gets misacylated with a wrong or near-cognate amino acid and retained on the enzyme to be edited. However, here, a cognate amino acid is loaded onto a ‘non-cognate’ tRNA^Asn and the Asp-tRNA^Asn product gets retained, perhaps to statistically favor further processing of the misacylated intermediate by GatCAB prior to its release (Scheme 1). We named this a ‘rescue mechanism’; it may be essential under conditions where GatCAB is not found in sufficient amounts within the cell or when it is occupied in the glutaminylation pathway that also exists in H. pylori.

Taking into account the hypothesis forged from a kinetic study (8), the rate-limiting step of ND-AspRS, like all class II aaRSs, is expected to be the transfer of activated Asp onto tRNA. In contrast, in class I aaRSs, the rate-limiting step is typically product release. Clearly, Hp ND-AspRS shifts to a class I-like kinetic behavior with tRNA^Asn but retains its class II character with tRNA^Asp. Previously, it has been demonstrated that the class II phosphoseryl-tRNA synthetase (SepRS) used in the Cys-tRNA^Cys biosynthesis indirect pathway in Methanocaldococcus janaschii, also has a rate-limiting step for the release of the aminoacyl-tRNA (46). The authors hypothesized that this characteristic could have evolved from modifications of the motifs implicated in aminoacylation, accelerating the transfer step over release, leading to a new rate-limiting step compared to other class II enzymes that is more similar to class I. ND-AspRS is a new example of an aaRS that can shift from a class II to a class I-like kinetic behavior, according to its tRNA substrate. This phenomenon fully correlates with the fates of Asp-tRNA products, since Asp-tRNA^Asn should be retained on ND-AspRS until GatCAB transforms it into Asn-tRNA^Asn, while Asp-tRNA^Asp should be released in line with its ‘cognate’ status. Together, these results allow us to make an analogy between the misacylation/transamidation pathway and editing strategies encountered with several classical aaRSs.

Transamidosomes are editing-like systems that enlarged the genetic code

AaRSs are responsible for matching a tRNA and its cognate amino acid. Many of these enzymes acquired catalytic peculiarities that nowadays ensure homologous aminoacyl-tRNA production, with an error rate compatible with survival of the cell (4,47,48). In spite of their specificity, aaRSs are known to use non-cognate amino acid or metabolic precursors to form mischarged aminoacyl-tRNAs. ValRS and IleRS can respectively produce Thr-tRNA^Val and Val-tRNA^Ile (49,50), and LeuRS can misacylate isoleucine, methionine, homocysteine, α-amino butyrate and norvaline onto tRNA^Leu (51). The release of mischarged aminoacyl-tRNA is normally prevented by correction mechanisms [either hydrolysis of the non-cognate aminoacyl∼AMP intermediate (pre-transfer editing) or hydrolysis of the ester linkage of the mischarged aminoacyl-tRNA (post-transfer editing)] (4,49–53). Post-transfer correction is thought to rely on aminoacyl-tRNA dissociation as the rate-limiting step providing time for the misacylated aminoacyl-tRNA acceptor arm to shift from the active site to the editing site. Interestingly, this mechanism completely relies on the presence of the cognate tRNA (54). Post-transfer editing can also be achieved by autonomous trans-editing factors (4). Indeed, as an example, ProRS and tRNA^Pro form a stable complex together with YbaK to avoid accumulation of Cys-tRNA^Pro and subsequent mistranslation (55).

Asn and Gln-transamidosomes constitute an assembly of an ND-aaRS (ND-AspRS or ND-GluRS/GluRS2), a ‘non-cognate’ tRNA (tRNA^Asn or tRNA^Gln) and a homeotopic modification enzyme (56); together these components ensure accurate synthesis of the cognate aminoacyl-tRNA pair. Since they reduce toxicity of misacylated species by preventing their release into the cellular medium, Asn- and Gln-transamidosomes could also be considered as kinetic systems related to editing systems that increase the faithfulness of aminoacylation and protect the integrity of the genetic code (Supplementary Figure S2 and 21,24). In H. pylori, the Asn-transamidosome mimics a post-transfer editing system, in which GatCAB would play a role similar to a trans-editing factor, except that the misacylated aminoacyl-tRNA does not get hydrolyzed but is rather modified into a cognate aminoacyl-tRNA. In line with this scenario, it is noteworthy that indirect pathways are thought to be ancient routes that allowed ancient ‘Asp’ and ‘Glu’ codons to be shifted into ‘Asn’ and ‘Gln’ codons and could thus be regarded as codon reassignment systems (24). Indeed, before GatCAB appeared, tRNA^Asn and tRNA^Gln were considered as normal ND-AspRS and GluRS substrates. However, among past Asp-tRNAs, the present-day tRNA^Asn emerged with a different kinetic behavior that favored its modification by GatCAB. While editing systems restricted and still restrict the combinations of small molecules that can be linked to tRNAs, Asn and Gln-transamidosomes were ribonucleoprotein systems that enlarged the genetic code by shifting some Asp-tRNA and Glu-tRNA species into Asn-tRNA^Asn and Gln-tRNA^Gln.

GatCAB is a node for transamidation

Helicobacter pylori belongs to a subclass of organisms that lack both GlnRS and AsnRS, in which GatCAB plays a dual role in glutaminylation and asparaginylation. In contrast, archaea possess two tRNA-dependent amidotransferases each specific for one pathway (57). In H. pylori, GatCAB is able to bind very efficiently to uncharged tRNA^Asn (this study) but also to tRNA^Gln (24) with equivalent affinities (2.1 and 2.0 µM, respectively) (Supplementary Scheme S1). This capacity provides two types of tRNA-presentation complexes: tRNPC and the tRNA^Gln-presentation complex (tRQPC), which may represent a way to target GatCAB toward ND-AspRS and asparaginylation on one hand and GluRS2 (the enzyme specific for tRNA^Gln misacylation in Hp) (25,34) and glutaminylation on the other hand (Supplementary Scheme S1). This need to share GatCAB between two distinct aaRSs could explain the reason why the Asn- and Gln-transamidosomes are both more dynamic in H. pylori than in T. thermophilus (22) or T. maritima (23). Hence, transamidosomes may be adapted to the metabolic context and physiology of the organism from which they originate.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online: Supplementary Table 1, Supplementary Scheme 1 and Supplementary Figures 1–3.

FUNDING

Natural Sciences and Engineering Research Council of Canada (9597-2010 to J.L.); Fond de Recherche sur la Nature et les Technologies du Québec (PR-105093 to J.L., Robert Chênevert and Paul H. Roy, and graduate fellowship to J.L.H.); Regroupement Québecois de Recherche sur la Fonction, la Structure et l'Ingénierie des Proteines (PROTEO) (graduate fellowship to J.L.H.); ‘Ministère du Développement Économique, de l'Innovation et de l'Exportation (MDEIE) du Québec’ [grant PSR-SIIRI-095/ESFARN to J.L., Sheng-Xiang Lin and Robert Chênevert, in collaboration with Prof. You-Xin Jin at the Shanghai Institute of Biological Sciences (SIBS), Chinese Academy of Sciences]; Ministère de l'Education Nationale, de la Recherche et de la Technologie (graduate fellowship to F.F.); Université de Strasbourg, Centre National de la Recherche Scientifique and Association pour la Recherche sur le Cancer (to D.K.) and National Institutes of Health (R01GM071480 to T.L.H.). Funding for open access charge: Centre National de la Recherche Scientifique, Université de Strasbourg, Association de la Recherche sur le Cancer.

Conflict of interest statement. None declared.