Two distinct regions in Staphylococcus aureus GatCAB guarantee accurate tRNA recognition

Abstract

In many prokaryotes the biosynthesis of the amide aminoacyl-tRNAs, Gln-tRNA^Gln and Asn-tRNA^Asn, proceeds by an indirect route in which mischarged Glu-tRNA^Gln or Asp-tRNA^Asn is amidated to the correct aminoacyl-tRNA catalyzed by a tRNA-dependent amidotransferase (AdT). Two types of AdTs exist: bacteria, archaea and organelles possess heterotrimeric GatCAB, while heterodimeric GatDE occurs exclusively in archaea. Bacterial GatCAB and GatDE recognize the first base pair of the acceptor stem and the D-loop of their tRNA substrates, while archaeal GatCAB recognizes the tertiary core of the tRNA, but not the first base pair. Here, we present the crystal structure of the full-length Staphylococcus aureus GatCAB. Its GatB tail domain possesses a conserved Lys rich motif that is situated close to the variable loop in a GatCAB:tRNA^Gln docking model. This motif is also conserved in the tail domain of archaeal GatCAB, suggesting this basic region may recognize the tRNA variable loop to discriminate Asp-tRNA^Asn from Asp-tRNA^Asp in archaea. Furthermore, we identified a 3₁₀ turn in GatB that permits the bacterial GatCAB to distinguish a U1–A72 base pair from a G1–C72 pair; the absence of this element in archaeal GatCAB enables the latter enzyme to recognize aminoacyl-tRNAs with G1–C72 base pairs.

INTRODUCTION

Correct pairing of an amino acid with its cognate tRNA is an essential step to maintain the accuracy of translation. This is usually accomplished by aminoacyl-tRNA synthetases (aaRSs) that catalyze the direct attachment of an amino acid to its cognate tRNA (1). However, glutaminyl-tRNA synthetase (GlnRS) is absent in the majority of bacteria, in all known archaea, and in many organelles (2). These organisms utilize an indirect pathway for Gln-tRNA^Gln formation where a non-discriminating glutamyl-tRNA synthetase (ND-GluRS) synthesizes mischarged Glu-tRNA^Gln (3) that is then amidated to the cognate Gln-tRNA^Gln by glutamyl-tRNA^Gln amidotransferase (Glu-AdT) (2,4). Similarly, many prokaryotes lacking an asparaginyl-tRNA synthetase (AsnRS) generate Asn-tRNA^Asn by the combined actions of a non-discriminating aspartyl-tRNA synthetase (ND-AspRS) and an aspartyl-tRNA^Asn amidotransferase (Asp-AdT) (5,6).

Two types of AdTs exist: the heterotrimeric GatCAB (7) present in bacteria, archaea and organelles (2), and the heterodimeric GatDE found exclusively in archaea (8). GatDE specifically converts Glu-tRNA^Gln (8), while bacterial GatCAB acts as both a Glu-AdT and an Asp-AdT in vitro (2). The in vivo role of bacterial GatCAB is defined by the nature of ND-aaRS (ND-GluRS and/or ND-AspRS) present in the cell (2). In archaea that lack AsnRS, GatCAB is encoded (9). This enzyme (e.g. from Methanothermobacter thermautotrophicus) in vitro strongly prefers Asp-tRNA^Asn over the homologous Glu-tRNA^Gln (10); thus archaeal GatCAB may act in vivo as an Asp-AdT.

AdTs accurately distinguish their mischarged aa-tRNA substrates (Glu-tRNA^Gln and/or Asp-tRNA^Asn) from the cognate Glu-tRNA^Glu and Asp-tRNA^Asp species. Bacterial GatCAB and GatDE achieve this by recognizing the first base pair of the acceptor stem and the D-loop of their tRNA substrates (11–13). In contrast, archaeal GatCAB does not recognize the first base pair of Asp-tRNA^Asn (13,14). Instead the M. thermautotrophicus GatCAB makes use of the D-loop, the nucleotide in position 49, and to a lesser extent of the length of the variable loop to distinguish Asp-tRNA^Asn from Asp-tRNA^Asp (14), while the Methanosarcina barkeri GatCAB appears to use primarily the length of the variable loop for the same task (13).

Previous work implicated the tail domain of the GatB and GatE in D-loop recognition of their respective aa-tRNA substrates (11,12). These tail domains share homology with the standalone YqeY proteins of unknown function (PFAM id: PF09424) that are present in a diverse array of organisms (11,12,15,16). Previous crystallographic studies did not resolve the YqeY-like domain of AdTs (11,12,17,18) or of Deinococcus radiodurans GlnRS (15). Here, we present a full-length Staphylococcus aureus GatCAB structure (containing the YqeY domain) resolved at 1.9 Å. The structure reveals a Lys rich motif in the tail domain of GatB which is absent in GatE. We also identified by biochemical studies a bacterial GatCAB-specific element that enables this enzyme to distinguish a U1–A72 base pair from a G1–C72 pair.

MATERIALS AND METHODS

Preparation of S. aureus GatCAB

The enzyme was over-produced in an Escherichia coli B834 strain and purified over a HisTrap HP column (GE Healthcare) as described (11). The sample was then diluted 5-fold with Buffer A [50 mM Tris–HCl pH 7.5, 10 mM MgCl₂, 1 mM DTT and 10% (v/v) glycerol], and applied to a HiTrap Heparin HP column (GE Healthcare) equilibrated with buffer A. The column was washed with buffer A containing 50 mM NaCl, and proteins were eluted with a linear gradient of 50–500 mM NaCl. The enzyme eluted at ∼250 mM NaCl. The enzyme fractions were then loaded onto a HiLoad 26/60 Superdex 200 pg column (GE Healthcare) equilibrated with buffer B [20 mM Tris–HCl pH 7.5, 10 mM MgCl₂, 1 mM DTT and 10% (v/v) glycerol]. Pooled fractions were concentrated by ultrafiltration using Vivaspin devices (VIVASCIENCE) to a final concentration of ∼12 mg/ml. S. aureus GatCAB mutants were generated using the QuikChange^TM site-directed mutagenesis kit according to the manufacture’s protocol (Stratagene), and purified to homogeneity as described above.

Preparation of M. thermautotrophicus GatCAB and ND-AspRS

The over-production and purifications were as described (10). Methanothermobacter thermautotrophicus GatCAB mutants were generated as described above.

Preparation of S. aureus ND-GluRS

The gene encoding the ND-GluRS from S. aureus Mu50 was amplified by polymerase chain reaction (PCR). The S. aureus ND-GluRS is toxic for an E. coli, which does not possess GatCAB, therefore the gene was cloned into NcoI/XhoI site of a pTip vector, which is used to protein expression in Rhodococcus erythropolis, and then the S. aureus ND-GluRS protein was expressed in R. erythropolis as described (19). Cells were harvested (4000×g, 15 min at 4°C) and disrupted using sonication in buffer C [50 mM Tris–HCl pH 7.5, 300 mM NaCl, 5 mM MgCl₂, 10% (v/v) glycerol, 0.5 mg/ml lysozyme and 0.1 mg/ml DNase I]. All the following purification processes were carried out at 4°C. Cell debris was removed by centrifugation (40 000×g, 1 h), and clarified supernatant was applied to a HisTrap HP column as described (11). Pooled fractions were loaded onto a HiLoad 26/60 Superdex 200 pg column equilibrated with buffer D [20 mM HEPES-K pH 7.6, 5 mM MgCl₂, 100 mM KCl, 1 mM DTT and 10% (v/v) glycerol]. The purified S. aureus ND-GluRS was concentrated by ultrafiltration to a final concentration of 18 mg/ml, and then diluted 2-fold with 100% (v/v) glycerol, and stored at −30°C.

Preparation of S. aureus tRNA^Gln, and M. thermautotrophicus and C. trachomatis tRNA^Asn

The tRNA isoacceptors were in vitro transcribed and purified as described (10,11).

Preparation of aminoacyl-[³²P] labeled tRNA

The tRNA isoacceptors were [³²P]-labeled and aminoacylated as described (10,20) with minor modification. For glutamylation of S. aureus tRNA^Gln, 5 µM S. aureus ND-GluRS and tRNA^Gln were added in the aminoacylation reaction.

Crystallization and structure determination

The high quality single crystals of S. aureus GatCAB were obtained by using the micro-seeding technique (21) from hanging drops set up in a 1 : 1 : 0.1 ratio from protein, reservoir solution [25% (w/v) PEG 600, 5 mM MgCl₂, 50 mM HEPES–NaOH pH 7.2 and 3% (v/v) 2-methyl-2,4-pentanediol (MPD)] and a micro-seeds stock solution (21). The crystal of S. aureus GatCAB was rapidly soaked through the reservoir containing 50 mM MES-Na, pH 6.4, 25% (w/v) PEG 600, 5 mM MgCl₂, 3% (v/v) MPD and 10% (v/v) glycerol as a cryoprotectant, and then a data set was collected to 1.9 Å resolution at SPring-8 beamline 41XU (Hyogo, Japan) under cryogenic condition (−173°C). The data set was processed and scaled using the HKL2000 package (22). The structure of S. aureus GatCAB was solved by molecular replacement using AMoRe in the CCP4 suite (23), with the refined model of the previous (11) S. aureus GatCAB structure (PDB ID: 2G5H) as a search model. The model of S. aureus GatCAB was rebuilt by automatic refinement program LAFIRE (24,25) running with CNS (26), and modified manually by using Coot (27) followed yielded the final model with the crystallographic R/R_free factor of 19.5/21.4%. The summary of data statistics is presented in Table 1. All figures were generated by PyMol (28).

Table 1.

Data collection and refinement statistics^a

Data collection statistics
PDB ID	3IP4
Beamline	SPring-8 BL41XU
Wavelength	1.00 Å
Space group	P212121
Cell dimensions a, b, c (Å)	71.1, 92.7, 180.4
Resolution (Å)	50.00–1.90 (1.97–1.90)
Rmergeb	0.070 (0.457)
I/σ(I)	23.2 (3.5)
Completeness (%)	99.9 (100)
Redundancy	7.0
Refinement statistics
Resolution (Å)	20.0–1.90
No. reflections	94598
Rworkc/Rfreed	0.195/0.214
Number of atoms	8299/801/1
(protein/water/others)
Average B factors (Å2)	33.6/35.3/26.0
(protein/water/others)
RMSD bond length (Å)/angles (°)	0.006/1.3
Ramachandran plot (%)
Favored	90.2
Allowed	9.6
Generous	0.2

^aValues in parentheses are for the outermost resolution shell.

^bR_merge = Σ_hΣ_j|<I>_h − I_h,j |/Σ_hΣ_jI_h,j, where <I>_h is the mean intensity of symmetry-equivalent reflections.

^cR_work = Σ|F_obs − F_cal|/ΣF_obs, where F_obs and F_cal are observed and calculated structure factor amplitudes.

^dR_free value was calculated for R factor, using only an unrefined subset of reflections data.

Small-angle X-ray scattering

For preparation of the tRNA^Gln-bound S. aureus GatCAB, the S. aureus GatCAB was mixed with tRNA^Gln in a molar ratio of 1 : 4, and then purified by a gel filtration using a Superdex 200 10/300 GL column (GE Healthcare) in buffer B.

Small-angle X-ray scattering (SAXS) measurements were carried out at SPring-8 beamline 40B2 and 45XU (Hyogo, Japan) (29). The following SAXS analysis was carried out by using the data measured at BL45XU. A wavelength of 1.0 Å was used, and the specimen-to-detector distance was 246.5 cm. The scattering vector range was 0.008 Å⁻¹ < s < 0.334 Å⁻¹, where s = 4πsinθ/λ, with 2θ being the scattering angle. Data were measured at 5.0 mg/ml of samples at 25°C. Both buffers and samples were exposed 100 s. The SAXS data were normalized to the intensity of the incident beam and processed to background subtraction using the standard procedures with the program package PRIMUS (30). Maximum particle dimensions D_max were computed using the indirect transform package GNOM (31), which also gives the distance distribution functions p(r). The discrepancy between the calculated and the experimental scattering curves was minimized using the program CRYSOL (32) as described previously (33,34). The 16 ab initio models of tRNA-free and -bound S. aureus GatCAB were generated and averaged with the program DAMMIN (35), DAMAVER (36) and SUPCOMB (37) as described previously (34).

[³²P]tRNA/nuclease P1 amidotransferase assay

AdT activities of wild-type and mutant GatCAB enzymes were monitored using a [³²P]tRNA/nuclease P1 amidotransferase assay as previously described (38). Determination of kinetic parameters with the S. aureus GatCAB enzymes was carried out as described for the Helicobacter pylori GatCAB (20). Determination of the kinetic parameters with the M. thermautotrophicus GatCAB enzymes was carried out as described (10).

RESULTS

Overall structure of a full-length GatCAB

We crystallized S. aureus GatCAB by adding 3–10% (v/v) MPD to the previous crystallization condition (11), which dramatically improved the crystal quality (maximum resolution from 2.3 to 1.9 Å). Surprisingly, the high-resolution structure clearly shows the GatB C-terminal region encompassing amino acids 412–475 and a histidine-tag (Figure 1A and Supplementary Figure S1), whose electron density map was not visible in the previous crystal structures. Although the unit cell dimension and crystal packing of the full-length and the previous apo-form S. aureus GatCAB are very similar, in the current structure the C-terminal GatB region is sandwiched between two GatA molecules related by 2-fold screw axes. Successful resolution of the tail-domain may be due to MPD addition in the crystallization mixture making the protein domain less flexible.

Figure 1.

(A) Overall structure of the full-length GatCAB. GatA and GatC are colored blue and magenta, respectively. GatB contains three domains: the cradle, the helical and the tail domain (light green, dark green and yellow green, respectively). Mg atom (black) in the cradle domain and C-terminal His-tag (grey) are also shown. (B) The structure of the GatB tail domain. The residues involved in the hydrophobic core are shown as a grey stick model, along with hydrogen bonds between the inter-domain loop and the tail domain (indicated as dashed lines). L472 is indicated as a grey sphere model. (C) Superposition of the S. aureus tail domain (yellow green) and YqeY-C (cyan), where the Lys rich motif is indicated as a stick model. (D) Structural based sequence alignments of the tail domain of YqeY-like proteins. Secondary structure of S. aureus tail domain is indicated at top, and that of YqeY-C is indicated at bottom (α, α-helix; β, β-strand; η, 3₁₀- turn). The number at top is corresponding to S. aureus GatB sequence. The residues constructing the hydrophobic core of S. aureus tail domain are marked with filled triangles. The Lys rich motif and the GXXAXGX motif are shown as cyan box and magenta box, respectively. The species aligned are as follows: STAAM S. aureus, METTH, M. thermautotrophicus; DEIRA, D. radiodurans; PYRAB, P. abyssi; BACSU, B. subtilis.

The GatA apo-form is essentially the same as the previously deposited S. aureus GatA structures (11). Interestingly, the high-resolution structure shows an alternative conformation for R358, which recognizes the carboxyl group of the substrate Gln (Supplementary Figure S2). Furthermore, Y310, which made a hydrogen bond with D425 in the Gln-bound S. aureus GatCAB (PDB ID: 2F2A), is flipped out from the active site of GatA, indicating that Gln binding induces only minor conformational changes in the GatA active site. GatC is also nearly identical to the deposited S. aureus GatC structures except the six C-terminal residues (95–100) were disordered and could not be modeled.

GatB is comprised of three consecutive domains: a cradle domain (1–294), a helical domain (295–407) and a tail domain (412–475). The cradle and helical domains are identical to the S. aureus GatB apo-form (PDB ID: 2G5I), and the permanent Mg²⁺ in the catalytic pocket was visible (11,18). The tail domain forms an anti-parallel helix bundle with three amphiphilic helices (α13, α15 and α16) that construct a hydrophobic core with α14 (Figure 1B and Supplementary Figure S1). L472 participates in this hydrophobic core and is not exposed to the molecular surface, indicating L472 is important for maintaining the structure of the tail domain rather than directly recognizing the aa-tRNA substrate as previously suggested (11). The loop between α15 and α16 interacts with I412 and S413 by two main chain and one side chain hydrogen bonds. Therefore, the S. aureus GatB tail and helical domains are linked by a ∼13 Å long inter-domain loop (408–411).

Comparison of the GatB tail domain with YqeY

The C-terminal end of GatB, comprised of the helical and tail domains (295–475), belongs to the same protein family as standalone YqeY polypeptides of unknown function present in many bacteria and in yeast (PFAM ID: PF09424) (15,16). The YqeY-like tail domain appended to the D. radiodurans GlnRS enables the enzyme to productively bind to tRNA^Gln (15). A similar role is proposed for this structure in GatB and GatE via recognition of the D-loop (11,12). The only YqeY structure so far resolved is from Bacillus subtilis (PDB ID: 1NG6); it can now be compared with the YqeY region in the present full-length of S. aureus GatCAB structure. The anti-parallel helix bundle of the tail domain superimposes well with the C-terminal domain of B. subtilis YqeY protein (YqeY-C: 92-146) with an r.m.s. deviation of 1.97 Å for the 51 Cα pairs compared (Figure 1C). Interestingly, the GatB tail domain has an additional α-helix (α14) and a short loop between α14 and α15 (439–442) not found in the YqeY-C. Based on the multiple sequence alignment with the tail domains and YqeY-C, this extended region contains a Lys rich motif (KXGKXK) that is highly conserved in bacterial and archaeal GatB enzymes (Figure 1D and Supplementary Figure S3). The C-terminal extension of D. radiodurans GlnRS also contains this extended region with a similar motif (RGGKTA). In contrast to GatB and D. radiodurans GlnRS, the tail domain of GatE lacks α14 and the Lys rich motif. Interestingly, the Lys rich motif of GatB is instead replaced in GatE with a GXXAXGX motif that has been implicated in GatDE distinguishing Glu-tRNA^Gln from Asp-tRNA^Asn (16).

Docking tRNA^Gln into the S. aureus GatCAB structure

The GatB C-terminal tail is essential for GatCAB binding tRNA^Gln (11). In the co-crystal structure of the GatDE enzyme (PDB ID: 2D6F), the GatE tail domain was in the vicinity of the tRNA D-loop; however, its detailed structure could not be solved and the D-loop was fitted with the help of the B. subtilis YqeY-C structure (12). We decided to create a docking model of full-length S. aureus GatCAB with E. coli tRNA^Gln to understand how the enzyme uses the tail domain to bind the tRNA. The use of E. coli tRNA^Gln is reasonable as the sequences of the variable and D-loops of the S. aureus and E. coli tRNA^Gln are identical (Figure 2A), and E. coli Glu-tRNA^Gln serves as substrate for GatCAB in vivo (39).

Figure 2.

Comparison with tRNA^Gln and tRNA^Glu. (A) Secondary structures of tRNA^Gln tRNA^Glu from S. aureus, E. coli and T. thermophilus. The G18:C56 base pair are colored green. The two guanines conserved in the tRNA D-loop are enclosed with black boxes. U8, A14–G19, A21, C48 and U54–G57, which employed on superposing of tRNA^Gln and tRNA^Glu, are shown as black large characters. U47 of tRNA^Gln and U20_B of tRNA^Glu are colored red. (B) Superposition of the E. coli tRNA^Gln and T. thermophilus tRNA^Glu. Orange and cyan ribbon diagram indicate E. coli tRNA^Gln and T. thermophilus tRNA^Glu, respectively. Supplemental nucleotides of the variable loop of E. coli tRNA^Gln and the D-loop of T. thermophilus tRNA^Glu are shown as thick stick models. The U8–A14–A21 base-triple, the conserved ‘Levitt’ pair and tRNA D-loop and TΨC -loop associating nucleotides are indicated as thin stick models.

To construct the docking model we first superposed E. coli tRNA^Gln (PDB ID: 1QTQ) into the GatDE:tRNA^Gln co-crystal structure by aligning all atoms of nucleotides U8, A14–G19, A21, C48 and U54–G57 of the tRNAs. The U8–A14–A21 base triple in the augmented D-stem is commonly found in tRNA^Gln while nucleotide 15 in the D-loop and nucleotide 48 in the variable loop form the conserved Levitt pair (40,41). The tertiary interaction between the D- and TΨC- loops of tRNA is also well conserved.

Next, we docked S. aureus GatCAB to tRNA^Gln by taking advantage of the homology between GatB and GatE (8,16), and the co-crystal structure of GatDE:tRNA^Gln (12), superposing GatB with M. thermautotrophicus GatE. The cradle domain of S. aureus GatB and M. thermautotrophicus GatE superpose well with an r.m.s. deviation of 1.7 Å for 244 pairs of Cα atoms compared. However, initially there were severe clashes between the helical domain and tRNA^Gln. Therefore, we separated the helical and tail domains into three parts (293–363, 364–381 and 382–475) and then superposed into GatE independently. Such movements are predicted based on previous AdT structures (11,12,17). Finally, the cradle and helical domains of S. aureus GatB could be superposed into that of M. thermautotrophicus GatE with an r.m.s. deviation of 1.9 Å for 321 pairs of Cα atoms compared (Supplementary Figure S4A).

The GatB tail domain possesses a highly conserved hydrophobic pocket comprised by residues (V449, M452, G457, A459 and P461) from the α15, α16 and the loop between them (Figure 3A), surrounded by positively charged residues (Figure 3B). In contrast, the opposite surface of the tail domain is composed of non-conserved, mostly negative residues (Figure 3A and B). These observations suggest that this conserved hydrophobic pocket may recognize the shape of the tRNA^Gln D-loop with the surrounding positive residues interacting with the tRNA phosphate backbone (Figure 3C).

Figure 3.

(A) Surface representation with a ribbon diagram of S. aureus tail domain is shown, and is colored by the sequence conservation among bacterial GatB using the program ConSurf 3.0: from low to high (cyan to white to magenta). The conserved hydrophobic pocket and the Lys rich motif are shown as yellow and purple dashed circles, respectively. (B) Solvent-accessible surface with a ribbon diagram of S. aureus tail domain is shown in the same orientation as (A) and is colored according to the electrostatic potential calculated by the program APBS running on Pymol (blue for positively charged and red for negatively charged). (C) The S. aureus tail domain docking model. S. aureus tail domain is shown as in (B). Orange and cyan ribbon diagram indicate E. coli tRNA^Gln and T. thermophilus tRNA^Glu, respectively. Supplemental nucleotides of the variable loop of E. coli tRNA^Gln and the D-loop of T. thermophilus tRNA^Glu are shown as a stick model. The helical domain from the S. aureus GatB:tRNA^Gln docking model (described below) shown together. The inter-domain loop between the helical and the tail domain is shown as dashed lines.

Interaction of the tail domain with tRNA is expected, as deletion of the C-terminal portion of GatB gives rise to a GatCAB mutant enzyme unable to bind tRNA (11) and as mentioned the electron density map of the GatDE:tRNA^Gln structure places the YeqY-like tail domain in proximity of the tRNA D-loop (12). However, the initial S. aureus GatCAB:tRNA^Gln docking model placed the conserved hydrophobic pocket of the tail domain ∼12 Å away from the D-loop of tRNA^Gln (Supplementary Figure S4B). It is likely that the tail domain can move to interact with tRNA^Gln due to the flexible loop connecting the tail and helical domains. This domain flexibility may explain why it has been difficult to resolve the YqeY-like tail of previous AdT structures (11,12,17,18). Given the above, we manually fitted the pocket of the tail domain into the D-loop of the E. coli tRNA^Gln (Figure 3C). The distance between the N-terminal end of the tail domain (Q411) and the C-terminal end of the helical domain (N407) is ∼10 Å, a distance the inter-domain loop connecting the two domains can bridge (Figure 3C). Recognition of the D-loop is consistent with the fact bacterial GatCAB uses that tRNA element to distinguish transamidation substrates (Glu-tRNA^Gln and Asp-tRNA^Asn) from Glu-tRNA^Glu and Asp-tRNA^Asp (11,13).

In order to verify our final S. aureus GatB:tRNA^Gln docking model, we recorded small-angle X-ray diffusion spectra on the tRNA-free and -bound S. aureus GatCAB purified by a size exclusion chromatography (Supplementary Figure S5). The predicted curve calculated from the crystal structure of the full-length S. aureus GatCAB with the program CRYSOL (32) fits closely to the experimental curve of tRNA-free S. aureus GatCAB, as characterized by a discrepancy value χ of 0.050 (Figure 4A). Furthermore, the crystal structure fits well to the ab initio envelope calculated with the program DAMAVER (36), of 16 dummy atom models calculated with DAMMIN (35), as reflected by a normalized spatial discrepancy (NSD) value of 1.21 (4A). The predicted scattering curve of the tRNA^Gln docking model fits well the experimental curve of tRNA-bound S. aureus GatCAB (χ of 0.098) and the ab initio envelope (NSD of 0.97) (Figure 4A). However, the tRNA^Gln docking model before fitting the tail domain shows a significantly higher discrepancy value with the experimental curve of tRNA-bound S. aureus GatCAB and the ab initio envelope (χ of 0.186 and NSD of 1.08, respectively) than using the tail domain fitting model. These results suggest that our docking model with the tail domain interacting with the tRNA is consistent with the solution structure of the S. aureus GatCAB:tRNA^Gln complex.

Figure 4.

(A) Experimental and computed SAXS scattering data. (Upper) the tRNA-bound GatCAB, (lower) tRNA-free GatCAB. The logarithm of the scattering intensity is plotted against the momentum transfer s=4πsinθ/λ, where 2θ is the scattering angle and λ=1.0 Å is the X-ray wavelength. The plots are displaced along the ordinate for better visualization. Red, cyan and pink curves indicated the computed scattering from tRNA-bound GatCAB model after the tail domain fitting, before the tail domain fitting, and tRNA-free GatCAB, respectively. (B) Comparison of the model of the S. aureus GatB:tRNA^Gln complex (left) and the crystal structure of the M. thermautotrophicus GatE:tRNA^Gln complex (right). S. aureus GatB, M. thermautotrophicus GatE and M. thermautotrophicus tRNA^Gln are shown as green, magenta and grey ribbon diagram, respectively. The A1–U72 base pair of M. thermautotrophicus tRNA^Gln is shown as a stick model. The 3₁₀ turn and the hydrophobic core of S. aureus GatB and the short loop of M. thermautotrophicus GatE are also shown. (C) Structural based sequence alignments of the cradle domain of bacterial GatB, archaeal GatB and GatE. The number at top is corresponding to S. aureus GatB sequence. Secondary structure of S. aureus GatB is indicated at top, and that of M. thermautotrophicus GatE is indicated at bottom. The 3₁₀ turn of bacterial GatB and the corresponding region of archaeal GatB and GatE are shown as cyan box. The species aligned are as follows: CHLTR, Chlamydia trachomatis; THETH, Thermus thermophilus HB8; METJA, Methanococcus jannaschii; AERPE, Aeropyrum pernix; SULSO, Sulfolobus solfataricus; ARCFU, Archaeoglobus fulgidus.

In our docking model, the hydrophobic pocket of the GatB tail domain nicely accommodates the curve of the tRNA^Gln D-loop (Figure 3C). The S. aureus enzyme distinguishes tRNA^Gln from tRNA^Glu based on the presence of an extra base (U20_B) in the D-loop of tRNA^Glu (11). To gain a better understanding of how this is accomplished we superposed the structures of E. coli tRNA^Gln (PDB ID: 1QTQ) and Thermus thermophilus tRNA^Glu (PDB ID: 1G59) in a similar manner as described above. The D-loop of T. thermophilus tRNA^Glu like that of S. aureus tRNA^Glu possesses a U20_B supernumerary base (Figure 2A). That extra base flips out from the tRNA D-loop and TΨC-loop associating region (Figure 2B). In the docking model of the tail domain with tRNA, the flipped out U20_B in the D-loop of tRNA^Glu could not be accommodated; the extra base sterically clashes with the surface of the protein in the model (Figure 3C), suggesting that is the mechanism by which S. aureus GatCAB rejects tRNA^Glu.

In addition to recognition of the D-loop, the GatB tail domain may also bind to the variable loop of the tRNA. In our docking model, the GatB-specific Lys rich motif of the tail domain is situated in proximity to the variable loop of tRNA^Gln, in particular U46 which is pushed out from the tertiary core of the tRNA (Figures 2B and 3C). Consistent with this prediction, replacement of this Lys rich motif with the GXXAXGX motif from GatE results in a mutant S. aureus GatCAB enzyme with reduced affinity for tRNA (Supplementary Figure S6).

The GatB cradle domain recognizes the first base pair of the acceptor stem

Bacterial GatCAB enzymes recognize the U1–A72 base pair of the acceptor stem of Glu-tRNA^Gln and Asp-tRNA^Asn to discriminate them from Glu-tRNA^Glu and Asp-tRNA^Asp (11,13). In contrast, archaeal GatCAB does not use the first base pair of the tRNA acceptor stem to distinguish Asp-tRNA^Asn from Asp-tRNA^Asp, recognizing aa-tRNA species with either a U1–A72 or a G1–C72 base pair (10,13,14). However, the process of distinguishing a U1–A72 base pair from a G1–C72 base pair is not known.

In our GatCAB:tRNA^Gln docking model the tRNA^Gln acceptor stem U1–A72 base pair is located in the center of the cradle domain in a space constructed by α3, two internal loops between β12–α3 and β13–β14, and a 3₁₀ turn between α2–β11. The 3₁₀ turn is adjacent to the U1–A72 base pair (Figure 4B). This 3₁₀ turn is constructed by a hydrogen bond between the main-chain carbonyl oxygen of K183 and amide of E186. Furthermore, M184 in the 3₁₀ turn forms a hydrophobic core with L167, L189, C191 and F217, suggesting the 3₁₀ turn is fixed with α3 and β11 by a hydrophobic interaction. In the GatDE enzyme, GatE has a short loop in place of a 3₁₀ turn at the corresponding region (Figure 4B and C).

In an alignment of GatB and GatE sequences (Figure 4C), the region including the 3₁₀ turn (183–186 in S. aureus GatB) is conserved in bacterial GatB sequences, and the corresponding region in archaeal GatB is 1–3 residues shorter, suggesting archaeal GatB has a loop in place of the 3₁₀ turn like GatE. The loop region of GatE is well conserved and is rich in positive residues, however the corresponding region in archaeal GatB is not well conserved. The docking model and alignment suggest these structural differences between bacterial GatB, archaeal GatB and GatE may enable the differences in first base pair recognition between the enzymes. In particular, they imply that the 3₁₀ turn enables bacterial GatB to distinguish a U1–A72 from a G1–C72 base pair.

The 3₁₀ turn of bacterial GatB is crucial for tRNA^Gln acceptor helix U1–A72 base pair recognition

To evaluate the relation between the presence of the 3₁₀ turn in GatB and the specificity of first base pair recognition, we constructed a mutant S. aureus GatCAB enzyme in which the 3₁₀ turn (K183, M184 and E185) is replaced by the short putative loop from M. thermautotrophicus GatB. Based on the multiple sequence alignment (Figure 4C) we also constructed a mutant M. thermautotrophicus GatCAB where the 3₁₀ turn (residues Lys, Met and Glu) is inserted into GatB between residues Gly169 and Glu170. We then tested the transamidase activities (Tables 2 and 3) by the [³²P]tRNA/nuclease P1 assay (38) of these mutant GatCAB enzymes with a variety of tRNA substrates.

Table 2.

Transamidase activity of the S. aureus (Sa) GatCAB and M. thermautotrophicus (Mt) GatCAB with different mischarged tRNA substrates

	Transamidase activity (s−1)
aa-tRNAa	Sa GatCABb		Mt GatCABb
	wt	310Δ	wt	310 ins
Sa Glu-tRNAUAGln	0.71±0.03	0.05±<0.01	0.0035±0.0003	0.0047±0.0002
Sa Glu-tRNAGCGln	0.05±<0.01	0.03±<0.01	0.0014±0.0002	0.0004±0.0002
Mt Asp-tRNAGCAsn	0.05±0.03	0.04±0.01	0.05±0.01	0.01±<0.01
Ct Asp-tRNAUAAsn	0.24±<0.01	0.02±<0.01	0.04±<0.01	0.11±<0.01

Measurements were from three separate experiments. Standard deviations are reported. Reactions were carried out at 37°C in the presence of ATP (4 mM), amide donor (4 mM) and aa-tRNA (1.25 µM) indicated.

^aThe aa-tRNA substrates tested were the S. aureus (Sa) wild-type Glu-tRNA_UA^Gln and mutant Glu-tRNA_GC^Gln, the wild type of the M. thermautotrophicus (Mt) Asp-tRNA_GC^Asn and the C. trachomatis (Ct) Asp-tRNA_UA^Asn.

^bIn the reactions, concentrations of the GatCAB indicated [Sa wild-type (wt) GatCAB, Sa 3₁₀ turn deleted mutant GatCAB (3₁₀Δ), Mt wt GatCAB, or Mt 3₁₀ turn insertion mutant GatCAB (3_{10 ins})] ranged from 20 nM to 1 µM.

Table 3.

Kinetic data for the transamidase activity of the S. aureus and M. thermautotrophicus GatCAB mutants with different mischarged tRNA substrates

Enzyme	Sa Glu-tRNAUAGln			Mt Asp-tRNAGCAsn			L
	KM (mM)	kcat (s−1)	kcat/KM (s−1/mM)	KM (mM)	kcat (s−1)	kcat/KM (s−1/mM)
Sa wt	1.80±0.22	1.7±0.1	930±120	2.05±0.48	0.1±<0.1	60±15	15.5
Sa 310Δ	1.85±0.59	0.1±<0.1	54±18	0.61±0.40	0.04±<0.01	67±44	0.8
Mt wt	0.70±0.26	0.006±0.001	8±3	0.78±0.30	0.10±0.01	125±49	15.6
Mt 310 ins	0.97±0.22	0.009±0.001	9±2	2.51±0.51	0.03±<0.01	12±3	1.3

Measurements were from three separate experiments. Standard deviations are reported. Reactions were carried out at 37°C in the presence of excess ATP (4 mM) and Gln (4 mM). L is the relative catalytic efficiency [(k_cat/K_M of homologous substrate)/(k_cat/K_M of non-homologous substrate)]. GatCAB enzymes labeled like in Table 2.

As expected (10,13,14), wild-type S. aureus GatCAB prefers tRNAs with a U1–A72 base pair over those with a G1–C72 base pair, while wild-type M. thermautotrophicus GatCAB has no strong preference (Table 2). S. aureus Glu-tRNA^Gln was a poor substrate for wild-type M. thermautotrophicus GatCAB (Tables 2 and 3), like B. subtilis Glu-tRNA^Gln for unknown reasons (10). However, the transamidase activity of wild-type M. thermautotrophicus GatCAB with wild-type Glu-tRNA^Gln containing a U1–A72 base pair was approximately the same as with the G1–C72 mutant Glu-tRNA^Gln (0.0035 and 0.0014 s⁻¹, respectively; Table 2); this is in line with previous results archaeal GatCAB (10,13,14).

Deletion of the 3₁₀ turn from S. aureus GatCAB resulted in a mutant enzyme that was 17-fold less efficient than wild-type mostly due to a difference in k_cat (Table 3), possibly due to the enzyme no longer recognizing the first base pair of the substrate aa-tRNA. However, consistent with our model, the removal of the 3₁₀ turn from the S. aureus GatB resulted in a mutant S. aureus GatCAB that no longer strongly preferred tRNA substrates with a U1–A72 base pair to those with a G1–C72 base pair (Tables 2 and 3). For example the mutant S. aureus GatCAB could use the mutant S. aureus Glu-tRNA_GC^Gln as a substrate about as well as wild-type S. aureus Glu-tRNA_UA^Gln (0.03 and 0.05 s⁻¹, respectively, Table 2). In addition, the mutant S. aureus GatCAB could use M. thermautotrophicus Asp-tRNA^Asn, with its G1–C72 base pair, about as well as Chlamydia trachomatis Asp-tRNA^Asn, which has a U1–A72 base pair (Table 2) and was about as efficient in amidating M. thermautotrophicus Asp-tRNA_GC^Asn as S. aureus Glu-tRNA_UA^Gln (Table 3).

Consistent with our model, the insertion of the 3₁₀ turn into M. thermautotrophicus GatCAB gave a mutant enzyme with preference for tRNA substrates with a U1–A72 base pair (Tables 2 and 3). The mutant M. thermautotrophicus GatCAB was about as efficient as wild-type M. thermautotrophicus GatCAB using S. aureus Glu-tRNA_UA^Gln as a substrate (Table 3). However, compared to wild-type M. thermautotrophicus GatCAB, the mutant enzyme was 10-fold less efficient with the M. thermautotrophicus Asp-tRNA_GC^Asn substrate (Table 3). In addition, the mutant M. thermautotrophicus GatCAB preferred by 10-fold a substrate wild-type S. aureus Glu-tRNA_UA^Gln substrate over the G1–C72 mutant Glu-tRNA^Gln (Table 2). Also, while wild-type M. thermautotrophicus GatCAB was nearly as active with C. trachomatis Asp-tRNA_UA^Asn as with M. thermautotrophicus Asp-tRNA_GC^Asn, the insertion GatCAB mutant preferred C. trachomatis Asp-tRNA^Asn, which has a U1–A72 base pair (Table 2). These results indicate that insertion of the 3₁₀ turn into GatB enables M. thermautotrophicus GatCAB to distinguish aa-tRNA with a U1–A72 base pair from ones with a G1–C72 base pair. Taken with the mutant S. aureus GatCAB data, these results strongly suggest that the 3₁₀ turn in S. aureus GatCAB is crucial for distinguishing aa-tRNA with a U1–A72 from those with a G1–C72 base pair.

DISCUSSION

The bacterial GatCAB enzyme uses the first base pair of the acceptor stem and the tRNA D-loop for precise tRNA discrimination (11,13). Our results suggest that the U1–A72 base pair of tRNA^Asn and tRNA^Gln is recognized by a 3₁₀ turn in the cradle domain of GatB, while the tail domain of GatB binds the D-loop.

The transamidosome is a ternary complex of T. thermophilus GatCAB, ND-AspRS and tRNA^Asn stable during the overall catalytic process (42). Such a structure would enable Asn-tRNA^Asn formation without the risk of free mischarged Asp-tRNA^Asn being used in protein synthesis. The complex would also protect the Asn-tRNA^Asn product from deacylation until it will be bound by EF-Tu and transported to the ribosome (42,43). Complexes between ND-GluRS, tRNA^Gln, and either Glu-AdT (GatCAB or GatDE) are also predicted to exist (12,42). Formation of these complexes between ND-aaRS, AdT and tRNA may explain why the AdTs recognize the specific identity elements in their tRNA substrates.

The structural models of the transamidosomes predict the ND-aaRS enzyme binds to the acceptor stem and anticodon loop of the tRNA (tRNA^Gln or tRNA^Asn) (12,42) (Supplementary Figure S7) in the same fashion as they do in the absence of AdT (44,45). This places the tRNA 3′-end into the synthetase active site to be aminoacylated; initially the first base pair of the tRNA’s acceptor helix is not accessible to the 3₁₀ turn of bacterial GatB in its role to discriminate tRNA isoacceptors. However, the tRNA’s tertiary core (including the D-loop) is recognized (Supplementary Figure S7). Recognition of the D-loop by the AdT tail domain (11,12) may permit the amidotransferases to distinguish the ND-aaRS complexed with their tRNA transamidation substrates (tRNA^Gln or tRNA^Asn) from other tRNA isoacceptors (tRNA^Glu or tRNA^Asp) (12–14). For example, GatCAB recognition of the D-loop would enable the AdT to discriminate ND-AspRS bound to tRNA^Asn from the aaRS bound to tRNA^Asp despite the U1–A72 base pair being initially inaccessible to GatCAB. The variable loop of the tRNA in the transamidsome models is also accessible to the AdT, in particular the Lys rich motif in the tail domain of GatB, which may explain why archaeal GatCAB uses this element to discriminate tRNA^Asn from tRNA^Asp (13,14).

The transamidosome models (12,42) predict that after aminoacylation the tRNA’s 3′ CCA terminus flips from active site in the ND-aaRS to the kinase active site of the AdT (42), similar to the tRNA movements seen in certain aaRSs with editing domains (46,47). In the case of the transamidosome, this movement enables the aminoacyl-moiety of the mischarged tRNA to be amidated by the AdT in the complex. Once the 3′ end of the acceptor stem flips from the aaRS active site into the transamidase active site of the AdT, the U1–A72 base pair may become accessible to the 3₁₀ turn in bacterial GatB.

Recognition of the first base pair by bacterial GatCAB and the archaeal GatDE may be a proofreading step to ensure amidation of the mischarged tRNA substrate (Glu-tRNA^Gln and/or Asp-tRNA^Asn) and not the properly aminoacylated product of the ND-aaRS (Glu-tRNA^Glu or Asp-tRNA^Asp). Why this proofreading step is not required by the archaeal GatCAB is unclear, but presumably recognition of the tRNA tertiary core may be enough to ensure that GatCAB amidates Asp-tRNA^Asn and not Asp-tRNA^Asp (13,14).

It was speculated that the ancestor of GatB and GatE recognized the first base pair of its tRNA substrates (16). However, as the 3₁₀ turn in the cradle domain of the S. aureus GatB for recognition of the U1–A72 base pair in tRNA^Gln is conserved only in bacterial GatB and not in archaeal GatB and GatE, this may not be the case. Instead, the common ancestor of GatB and GatE may not have recognized the first base pair of its tRNA substrate, with bacterial GatCAB and GatDE independently evolving to recognize the first base pair of tRNA. Thus, bacterial GatCAB and GatDE both recognizing the first base pair may have been a case of convergent and not divergent evolution.

ACCESSION NUMBERS

Coordinate and structure factor have been deposited in PDB with accession number 3IP4.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

A.N. held a Research Fellowship of the Japan Society for the Promotion of Science for Young Scientists. Grants-in-Aid for Scientific Research (19370037 and 21370041 to M.Y.); Targeted Proteins Research Program (to I.T.) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; Institute of General Medical Sciences and the US Department of Energy (to D.S.).

Conflict of interest statement. None declared.