Complex Structure of Pseudomonas aeruginosa Arginine Rhamnosyltransferase EarP with Its Acceptor Elongation Factor P

Posttranslational rhamnosylation of EF-P plays a key role in Pseudomonas aeruginosa, establishing virulence and antibiotic resistance, as well as survival. The detailed structural and biochemical characterization of the EF-P-specific rhamnosyltransferase EarP from P. aeruginosa not only demonstrates that sugar donor TDP-Rha binding enhances acceptor EF-P binding to EarP but also should provide valuable information for the structure-guided development of its inhibitors against infections from P. aeruginosa and other EarP-containing pathogens.

ABSTRACT

A bacterial inverting glycosyltransferase EarP transfers rhamnose from dTDP-β-l-rhamnose (TDP-Rha) to Arg32 of translation elongation factor P (EF-P) to activate its function. We report here the structural and biochemical characterization of Pseudomonas aeruginosa EarP. In contrast to recently reported Neisseria meningitidis EarP, P. aeruginosa EarP exhibits differential conformational changes upon TDP-Rha and EF-P binding. Sugar donor binding enhances acceptor binding to EarP, as revealed by structural comparison between the apo-, TDP-Rha-, and TDP/EF-P-bound forms and isothermal titration calorimetry experiments. In vitro EF-P rhamnosylation combined with active-site geometry indicates that Asp16 corresponding to Asp20 of N. meningitidis EarP is the catalytic base, whereas Glu272 is another putative catalytic residue. Our study should provide the basis for EarP-targeted inhibitor design against infections from P. aeruginosa and other clinically relevant species.

IMPORTANCE Posttranslational rhamnosylation of EF-P plays a key role in Pseudomonas aeruginosa, establishing virulence and antibiotic resistance, as well as survival. The detailed structural and biochemical characterization of the EF-P-specific rhamnosyltransferase EarP from P. aeruginosa not only demonstrates that sugar donor TDP-Rha binding enhances acceptor EF-P binding to EarP but also should provide valuable information for the structure-guided development of its inhibitors against infections from P. aeruginosa and other EarP-containing pathogens.

INTRODUCTION

Glycosylation of proteins is the most ubiquitous posttranslational modification in nature and plays an important role in protein structure and function (1). For a long time, protein glycosylation was deemed to be restricted to eukaryotes. So far, it has been clearly established that bacteria, including pathogens possess both O- and N-linked glycosylation pathways that display many commonalities with their eukaryotic counterparts, as well as some unexpected variations (2, 3). N-linked protein glycosylation commonly occurs at the side chain amide of an Asn residue in the sequences Asn-X-Ser/Thr of the targeted proteins, where X can be any amino acid except proline. Recently, rare N-glycosylation of arginine residues was shown to be performed by the type III secretion effector NleB from the attaching/effacing pathogens. NleB modifies arginine residues of host proteins with N-acetylglucosamine (GlcNAc) to inhibit antibacterial and inflammatory host responses (4–6).

More recently, a new type of arginine glycosylation was reported as an l-rhamnosylation modification on a specific arginine residue within bacterial translation elongation factor P (EF-P) (7). EF-P is orthologous to eukaryotic/archaeal elongation factor 5A (e/aIF-5A), being required for preventing ribosome stalling at proline stretches (8–10). In contrast to the two-domain structure of e/aIF-5A, bacterial EF-P exhibits a three-domain structure. The N-terminal domains I and II of EF-P are equivalent to the N- and C-terminal domains present in e/aIF-5A (11), whereas the additional C-terminal domain III confers an overall L-shape similar to tRNA on EF-P (12, 13). The function of both EF-P and e/aIF-5A in assisting polyproline translation depends on posttranslational modification (8, 14). In this regard, EF-P can be β-lysylated (15–17), 5-aminopentolylated (18–20), and rhamnosylated (7, 21, 22). The eukaryotic/archaeal ortholog e/aIF5A, in contrast, strictly depends on (deoxy)hypusination (23). In all of these cases, the acceptor amino acid is either lysine or arginine and located at the tip of domain I. When bound to the ribosome, the modification protrudes toward peptidyl transferase center, where it stabilizes the CCA sequence end of the P-site tRNA to promote an optimal geometry for peptide bond formation (24–26). Whereas lysyl lysine, 5-aminopentolyl lysine, and (deoxy)hypusine resemble chemically related structures, arginine rhamnosylation is exceptionally distinct. To date, EF-P rhamnosylation has been discovered in Shewanella oneidensis (7) and the human pathogens Pseudomonas aeruginosa (21) and Neisseria meningitidis (22).

The rhamnosylation of Arg32* of EF-P was shown to be performed by EarP, a conserved glycosyltransferase encoded at a position adjacent to efp and employing dTDP-β-l-rhamnose (TDP-Rha) as the donor substrate (7, 21, 22). Li et al. and Wang et al. reported that rhamnose is α-linked to Arg32* in bacterial EF-P proteins, demonstrating that EarP inverts the sugar of its donor substrate (27, 28). During the preparation of our manuscript, two independent works on the structure of EarP appeared. R. Krafczyk et al. were the first to report the crystal structure of EarP from Pseudomonas putida bound to TDP-Rha (29). These researchers identified that EarP contains two opposing Rossmann fold domains, which classify EarP in glycosyltransferase superfamily B (GT-B). EarP was then built into the carbohydrate active enzyme (CAZy) database and now represents the new glycosyltransferase family GT104. Nevertheless, this structure missed structural elements in the N-terminal domain (NTD). T. Sengoku et al. determined the crystal structures of EarP from N. meningitidis in the apo- and TDP-Rha-bound forms, as well as in complex with domain I of EF-P (30). These researchers showed that EarP binds the entire β-sheet structure of EF-P domain I and recognizes its conserved residues through numerous side chain-specific interactions. These researchers also described a rotational reorientation of the NTD relative to the C-terminal domain (CTD) and a conformational change in a conserved TDP-Rha binding loop after EF-P binding.

GT-B enzymes often exhibit a global domain movement upon binding of donor and acceptor substrates, differing in the type and degree of motion in a catalyst-specific manner (31). This open-to-closed conformational transition typically brings acceptor and donor into close proximity and is also accompanied by several loop movements. For some GT-B glycosyltransferases, such as MurG (32, 33), MshA (34), and hOGT (35), these dynamic structural changes play critical roles in determining the enzymes’ sequential ordered mechanism, with donor binding first and acceptor binding second. However, the role of substrate-induced structural changes that occur during catalysis in EarP remains obscure.

Most of the inverting glycosyltransferases employ a direct-displacement S_N2-like reaction by a base catalyst that deprotonates the incoming nucleophile of the acceptor and by Lewis acid activation of the departing phosphate leaving group (36). The key question in examining the catalytic mechanism of inverting glycosyltransferases is therefore the identity of the acid/base catalyst. In P. putida EarP, the three negatively charged residues Asp13, Asp17, and Glu273 were identified as potential candidates to catalyze the glycosylation reaction, based on these residues being in the vicinity of the rhamnose moiety in the active pocket and alanine substitution of each of them eliminating in vitro EF-P rhamnosylation detected by Western blotting. In N. meningitidis EarP, Asp20 directly interacts with the acceptor Arg32* η-nitrogen atoms and replacement of Asp20 with alanine or asparagine abolished in vivo EF-P rhamnosylation. Thus, the conserved Asp20 other than Asp16 (corresponding to Asp17 and Asp13 of P. putida EarP, respectively) was identified as the general base of N. meningitidis EarP. In vivo EF-P rhamnosylation assay was carried out by coexpressing N. meningitidis EF-P with EarP in Escherichia coli cells and analyzing purified EF-P rhamnosylation by mass spectrometry. More-detailed structural information and robust enzymatic assay of EarP would help to clarify the catalytic mechanism of EarP.

EF-P proteins contribute to the infectivity of pathogenic bacteria by controlling the translation of proteins critical for establishing virulence (37), modulating antibiotic resistance (21), or maintaining cell viability (38). Deleting EF-P or the corresponding rhamnosyltransferase EarP increases the susceptibility of P. aeruginosa to a range of antibiotics (21). EF-P and Arg32* are essential for the cell viability of N. meningitidis (22). In addition to P. aeruginosa and N. meningitidis, the genomes of certain pathogenic betaproteobacteria and gammaproteobacteria encode conserved EF-P containing Arg32* and EarP, suggesting that rhamnosylated EF-P might be equally important in other clinically relevant species. However, most of the bacteria living in the human body do not rely on EarP for EF-P modification. In contrast to the three-dimensional structure of EF-P, which is mostly composed of convex surfaces, EarP has a substrate-binding pocket and may generally be suitable for the development of its inhibitors. Such inhibitors are expected to function as potential narrow-spectrum antibacterial reagents against these EarP-containing pathogens, with reduced side effects and a lower risk of the emergence of drug-resistant bacteria.

Here, we report crystal structures of P. aeruginosa EarP in the apo form, in binary complexes with TDP or TDP-Rha, and in a ternary complex with both TDP and full-length EF-P, which represent different enzymatic states during the catalytic reaction. In contrast to N. meningitidis EarP, P. aeruginosa EarP exhibits more major conformational change in substrate-binding loops upon TDP-Rha binding, whereas EF-P binding induces further localized structural change at the substrate-binding pocket rather than an interdomain movement. Comparative structural analyses combined with isothermal titration calorimetry experiments revealed that conformational change induced by TDP-Rha binding plays a role in the adjustments required for EF-P binding. Moreover, the conserved Asp16 serves as the general base, whereas the conserved Glu272 is another putative catalyst in the S_N2 reaction, based on in vitro EF-P rhamnosylation and EarP active-site geometry.

RESULTS AND DISCUSSION

EarP overall fold.

The crystal structure of P. aeruginosa EarP in the apo form was determined at 1.75-Å resolution (Fig. 1A) using single-wavelength anomalous diffraction of SeMet-derivative EarP (SeMet-EarP) crystals. The two binary-complex structures of EarP bound to TDP at 1.85-Å resolution (Fig. 1B) and bound to TDP-Rha at 2.15-Å resolution (Fig. 1C) were obtained by cocrystallizing native EarP with TDP-Rha and by soaking SeMet-EarP crystals with TDP-Rha, respectively. Although EarP was cocrystallized with TDP-Rha, only a TDP moiety could be located in the electron density map (see Fig. S1A in the supplemental material), suggesting the wild-type EarP hydrolyzes the rhamnose of TDP-Rha in the absence of the acceptor protein during the course of the cocrystallization trials. The TDP-Rha cocrystal structure shows clear electron density for the TDP-Rha ligand (Fig. S1B). Only one EarP molecule was found in the asymmetric unit (ASU) of the apo-, TDP-Rha-, and TDP-bound EarP crystals. The ternary complex structure of EarP with both TDP and full-length EF-P in a stoichiometry of 1:1:1 was solved at 2.30-Å resolution (Fig. 1D) by cocrystallization. The ASU of this structure consists of two of such complexes, which are almost identical to each other. All of the complex crystal structures of EarP were solved by molecular replacement using SeMet-EarP as the searching model.

FIG 1.

Overall structure of P. aeruginosa EarP in the apo form (A), in complex with TDP (B), in complex with TDP-Rha (C), and in complex with both TDP and full-length EF-P (D). The hinge region of EF-P is colored yellow.

P. aeruginosa EarP consists of two Rossmann-like domains separated by a deep cleft. The NTD (residues 1 to 138, as well as residues 364 to 376 from the C-terminal helix) is composed of a four-stranded parallel β-sheet in the center and a β-hairpin on each side, surrounded by four α-helices and a 3₁₀-helix, while the CTD (residues 155 to 363) contains a five-stranded parallel β-sheet and an antiparallel strand on one side, surrounded by ten α-helices and a 3₁₀-helix (Fig. 1A). The C-terminal α-helix L364-L376 extends from the CTD and forms part of the NTD, securing interdomain contacts. In addition, the two domains are connected by a long loop (residues 139 to 154). The structure of EarP is basically consistent with the structural fold possessed by GT-B enzymes but with some differences. In the classical GT-B fold, each domain adopts the Rossmann-type fold that contains a central six-stranded parallel β-sheet with a 321456 topology flanked by α-helices. By comparison, the third parallel strand is replaced with a β-hairpin in each domain of EarP, while the sixth parallel strand is replaced by a β-hairpin in the NTD and absent in the CTD of EarP. In a word, EarP displays an atypical GT-B fold. Sequence alignment indicated that P. aeruginosa EarP shows 41% identity, 52% similarity, and 8% gaps to N. meningitidis EarP and 60% identity and 68% similarity to P. putida EarP in amino acid sequence. The overall structures of P. aeruginosa and N. meningitidis EarP proteins are quite similar, giving an RMSD of about 1.1 Å over 229 Cα atoms after superposition. As the reported P. putida EarP structure missed structural elements in the NTD, the CTD structures of P. putida and P. aeruginosa EarP proteins were compared, giving a root mean square deviation (RMSD) of 0.618 Å over 169 Cα atoms after superposition.

EarP interacts with domain I of EF-P mainly through the NTD (Fig. 1D). EF-P is comprised of three β-barrel domains. Previously determined crystal structures of EF-P proteins from some bacterial species, including P. aeruginosa (PDB ID 3OYY) (13) have shown the structural flexibility of EF-P with the rotational motion of domain I with respect to domains II and III. In accordance, the domain motion analysis through the program Dyndom of the two EF-P molecules in the ASU of our ternary complex structure shows that domain I (residues 1 to 64) rotates about 13.2° with respect to domains II (residues 67 to 128) and III (residues 129 to 188), with residues 64 to 66 as a hinge region (Fig. 1D and see Fig. S2 in the supplemental material). The largest 67° rotational domain motion was observed after superposition of four P. aeruginosa EF-P structures from this study and those previously determined (Fig. S2). This EF-P domain motion should be attributed to crystal packing rather than EarP binding, since only domain I of EF-P contacts with EarP in the ternary complex structure (Fig. 1D).

Donor binding site of EarP.

The donor TDP-Rha is located in a cavity in the cleft between the NTD and the CTD (Fig. 1C and Fig. S3A) and is surrounded mainly by residues from the CTD (Fig. 2A). The TDP moiety superimposes well in the TDP- and TDP-Rha-bound structures with all TDP hydrogen bonding atoms, and hydrophobically interacting residues superimposed well in these two binary complex structures (Fig. S1C). In detail (Fig. 2A), the O2, N3, and O4 atoms of the thymine moiety form hydrogen bonds with the main-chain atoms of Met252. Moreover, the thymine base stacks against the aromatic side chains of Phe251 and Tyr257, and the thymine methyl group forms hydrophobic interactions with Phe190, P216, and Tyr257. The thymidine-binding pocket is too small to accommodate a purine by steric clashes with Pro216. The O3 of the deoxyribose moiety forms two hydrogen bonds with the carboxylate of Asp273 and the Nɛ2 of Gln254.

FIG 2.

Binding of TDP-Rha. (A) Binding of the thymine, deoxyribose, and pyrophosphate moieties. (B) Binding of the rhamnose moiety. (C) ITC titration and fitting curves of the D16N and wild-type EarP proteins with TDP-Rha and TDP, respectively.

The α- and β-phosphate groups are located at the N-terminal ends of helix α1 from the NTD and helix α9 from the CTD (Fig. 1 and Fig. 2A), where they have a stabilizing interaction with the positive charges of the two helix dipoles. The α-phosphate forms hydrogen bonds with the side chain hydroxyl group of Ser274, as well as with the main-chain nitrogen atoms of Glu272, Asp273, and Ser274. The β-phosphate forms hydrogen bonds with the guanidinium group of Arg270 and the aromatic hydroxyl of Tyr192, as well as with the main-chain nitrogen atoms of Tyr14 and Gly15 (Fig. 2A).

Most of the residues contributing to the interaction between EarP and TDP-Rha are conserved among P. putida, N. meningitidis, and P. aeruginosa EarP proteins (see Fig. S4 and S5 in the supplemental material). Since the electron density for the rhamnose moiety was unclear in the TDP-Rha complexed structure of P. putida EarP (Fig. S4A), the TDP-binding sites of P. putida and P. aeruginosa EarP proteins were compared, which were superimposed well in the two TDP-Rha complexed structures (Fig. S4B). In contrast, in the structure of N. meningitidis EarP bound with TDP-Rha, the β-phosphate lies farther away from helix α1 (Fig. S4C), forming water-mediated hydrogen bonds with helix α1 (30).

Similar to what was observed for N. meningitidis EarP, the rhamnose ring in our structure also assumes the ¹C₄ chair conformation but has a different orientation relative to the TDP moiety (Fig. S4C). The rhamnose-accommodating cavity in our structure provides more hydrogen bonds than that observed for N. meningitidis EarP (30). In detail (Fig. 2B), the O of the rhamnose forms hydrogen bonds with Nδ2 of Asn13 and the aromatic hydroxyl of Tyr192. The O2 of the rhamnose makes hydrogen bonds with the side chain carboxylate and the main-chain nitrogen atom of Asp16. The O3 of the rhamnose forms a hydrogen bond with the side chain carboxylate of Asp16, as well as water-mediated hydrogen bonds with the side chain carboxylate of Asp16 and the main-chain oxygen of Gly15.

Conformational change of EarP after donor binding.

Analysis with the program Dyndom revealed that P. aeruginosa EarP does not undergo a domain rotation upon TDP-Rha or TDP binding. Similarly, both N. meningitidis and P. putida EarP enzymes have no global interdomain movements upon TDP-Rha binding indicated by Dyndom analysis and small-angle X-ray scattering (29), respectively. Superposition of P. aeruginosa EarP active site in the apo- and TDP-Rha-bound forms indicated that donor binding fosters more drastic conformational changes in substrate-binding loops of P. aeruginosa EarP (see Fig. 4A) than that observed for N. meningitidis EarP (Fig. S6A). First, loop_β1-α1 (containing residues 11 to 14, corresponding to residues 15 to 18 of N. meningitidis EarP), which could not be traced in the apo structure (Fig. 1A), becomes visible (Fig. 1C and 4A; see also Fig. S7) and contacts with the β-phosphate and the rhamnose moieties in the TDP-Rha bound structure (Fig. 2A and B). Second, Arg219 in loop_β10-α7 swings its side chain toward the pyrophosphate moiety and positions its guanidinium group at hydrogen bonding distance to Oδ1 of Asn13 upon TDP-Rha binding (see Fig. 4A; see also Fig. S7). However, this hydrogen bonding interaction between the CTD and the NTD is not conserved in N. meningitidis EarP (see Fig. S4C and S5 in the supplemental material). Third, loop_β14-α10 containing residues 290 to 296 is invisible in the apo structure (Fig. 1A) but could be traced in the TDP-Rha bound structure with the density for the aromatic ring of Tyr290 being clearly visible (Fig. 1C and 4A; see also Fig. S7 in the supplemental material), possibly because this residue mediates van der Waals contact with the rhamnose moiety. Consistently, the alanine substitution of the corresponding residue (Tyr291) of P. putida EarP caused significantly reduced activity compared to the wild type. In addition, Phe190 adjusts the position of its aromatic ring to avoid a steric clash with the methyl of the thymidine. Phe251 moves its aromatic ring to stack against the thymidine heterocycle. Arg270 moves its side chain to approach the β-phosphate at a hydrogen bonding distance. Asp16 swings its side chain due to the electrostatic repulsion with the pyrophosphate moiety, forming a hydrogen bond with the side chain hydroxyl group of Tyr112 (see Fig. 4A).

FIG 4.

Substrate-induced conformational changes of EarP. (A) Structural comparison between the apo form and the TDP-Rha complex of P. aeruginosa EarP. (B) Structural comparison between the TDP-Rha complex and the EF-P complex of P. aeruginosa EarP. (C) ITC titration and fitting curves of the D16N mutant of EarP (40 µM) in the absence or presence of approximately saturated TDP-Rha (240 µM) with EF-P.

Acceptor binding site of EarP.

The acceptor EF-P is located at the entrance of the substrate pocket (Fig. S3B). The buried surface area between EarP and domain I of EF-P was 1,030 Ų provided by PISA. This is similar to the interface areas of 897.5 Ų between N. meningitidis EarP and EF-P (30) and 973 Ų between Escherichia coli EF-P and its modification enzyme GenX/EpmA (39).

Most of the residues contributing to the interaction between EarP and EF-P are conserved (Fig. S5). In detail (Fig. 3), the guanidinium group of Arg32* form hydrogen bonds with the side chain carboxyl groups of Asp16 and Asp12 and the side chain hydroxyl group of Tyr112, as well as the main-chain oxygen of Asp12. The hydrophobic moiety of the Arg32* side chain forms a CH-π interaction with the aromatic rings of Tyr290 and Tyr112. Tyr290 in turn makes a hydrogen bond with the side chain oxygen of Asp12. The tip loop of EF-P is stabilized by hydrogen bonds between the main-chain oxygens of Arg32*, Ser30*, and Lys29* and the side chains of Lys300 and Gln292, as well as intramolecular hydrogen bonds. The side chain amino group of Lys29* salt bridges with the carboxylate of Glu294, whereas the side chain hydroxyl group of Ser30* (corresponding to a glycine in N. meningitidis EF-P) hydrogen bonds with the carboxyl group of Glu111.

FIG 3.

EF-P recognition by EarP. (A) EF-P residues that directly interact with EarP are shown as green sticks. (B) Recognition of the tip loop containing Lys29* to Ala35*. (C) Recognition of EF-P other than the tip loop. (D) ITC titration and fitting curves of wild-type EarP (40 µM) in the presence of approximately saturated TDP (240 µM) with wild-type EF-P or its mutants.

Furthermore, the side chain amino group of Lys55* hydrogen bonds with the main-chain oxygen atoms of Ala87 and Cys88, and the side chain carboxylate of Glu84, as well as salt bridges with the side chain carboxylate of Glu89. The main-chain atoms of Val53* hydrogen bond with the main-chain oxygen of Pro126 and the side chain oxygen of Ser127. The side chain hydroxyl group of Thr52* hydrogen bonds with the main-chain oxygen of Pro126 and main-chain nitrogen of Leu128. In addition, Asn28*, Val36* (corresponding to a lysine in N. meningitidis EF-P), and Val53* form hydrophobic interactions with Trp118, Phe139, and Phe137. On the end opposite to Arg32*, Ile15* forms a hydrophobic interaction with Leu128. Arg124 of N. meningitides EarP corresponds to Ser121 of P. aeruginosa EarP, which does not provide hydrogen bonds to Glu26*.

Conformational change after EF-P binding.

Analysis with the program Dyndom revealed that P. aeruginosa EarP does not undergo a domain rotation upon TDP and EF-P binding, different from that observed for N. meningitidis EarP, in which the NTD rotated by about 8° relative to the CTD after TDP and EF-P binding. However, EF-P binding fosters further conformational changes inside the active site of P. aeruginosa EarP revealed by structural comparison between the TDP-Rha complex and the EF-P complex (Fig. 4B). First, Asp16 and Tyr290 that have been reoriented by TDP-Rha binding fine-tune their side chain conformations to interact directly with Arg32*. A new hydrogen bond is formed between Tyr290 and Asp12 accompanying Arg32* binding. Similar phenomenon has also been observed by structural comparison of N. meningitidis EarP at different enzymatic states (Fig. S6). Asp20 and Tyr288 of N. meningitidis EarP shifted their side chain conformations upon TDP-Rha binding to facilitate their EF-P binding (Fig. S6B). These observations suggested that structural changes induced by TDP-Rha binding contribute directly to EF-P binding. Thus, donor binding would assist acceptor binding to EarP.

In order to investigate the affinity of EarP for donor and acceptor substrates and to test whether the donor binding enhances the acceptor binding, we performed isothermal titration calorimetry (ITC) experiments. Since wild-type EarP could hydrolyze the rhamnose of TDP-Rha in the absence of the acceptor protein, a deactivated mutant D16N (see below) was used to examine the affinity of EarP for TDP-Rha. The D16N mutant bound TDP-Rha with a K_d value of 6.29 ± 0.28 μM, whereas wild-type EarP bound TDP with a K_d value of 31.4 ± 2.38 μM (Fig. 2C). In order to determine the binding affinity of EarP bound with the donor for EF-P, most of the EarP proteins in the sample cell should be bound with the donor instead of being in the free form. As shown in Fig. 2C, when the molar radio of the donor to EarP was about 6 the heat signal approximately approached zero, which indicated that EarP was mostly bound with the donor. The free D16N mutant bound EF-P with a K_d of 32.7 ± 12.3 μM, whereas the interaction between the D16N mutant of EarP and EF-P was enhanced about 6-fold with a K_d value of 4.97 ± 0.34 μM when TDP-Rha was added into the sample cell with a molar radio of about 6 to EarP-D16N (Fig. 4C). The calculated binding stoichiometry (N) values for all ITC experiments were approximately 1, which is consistent with the expected binding stoichiometry. These results indicated that TDP-Rha binding indeed enhances EF-P binding to EarP.

ITC experiments showed that wild-type EarP bound EF-P with a K_d of 0.24 ± 0.02 μM in the presence of TDP with a molar radio of about 6 to EarP (Fig. 3D). Comparatively, when the molar radio of TDP to EarP was also about 6 in the sample cell, the K29A* and T52V* mutants of EF-P showed only slightly decreased binding, with K_d values of 0.61 ± 0.04 μM and 0.35 ± 0.03 μM, respectively, whereas the K55A* mutant showed remarkably decreased binding with a K_d value of 52.6 ± 27.7 μM (Fig. 3D), suggesting that Lys55* contributes significantly to the binding of EarP. Moreover, wild-type EarP exhibited no binding to the peptide corresponding to the EF-P tip loop sequence containing residues Asn28* to Ala35* in the presence of TDP with a molar radio of about 6 to EarP (data not shown), suggesting that overall shape complementarity is required for the recognition of EF-P by EarP.

Catalytic mechanism of EarP.

The ternary complex structure of P. aeruginosa EarP with TDP and EF-P shows that Asp12, Asp16, and Tyr112 interact with the acceptor Arg32* η-nitrogen atoms directly. Asp16 of P. aeruginosa EarP corresponds to Asp20 of N. meningitidis EarP, which has been reported as the general base (30). Moreover, replacement of Glu273 of P. putida EarP, corresponding to Glu272 of P. aeruginosa EarP, with a glutamine resulted in undetectable EF-P rhamnosylation in vivo (29). To further analyze the roles of these conserved key residues, we performed site-directed mutagenesis combined with an in vitro assay for EF-P rhamnosylation, detected by mass spectrometry. The D16N and E272Q mutations completely abolished rhamnosylation activity. In contrast, the D12N and Y112F mutations retained rhamnosylation activity (Fig. 5). Asp16 is the only carboxylate residue to be in a position to deprotonate the η-amino group of Arg32* (Fig. 3B; see also Fig. S8B and C in the supplemental material). Thus, Asp16 of P. aeruginosa EarP is the best candidate for the general base that activates the acceptor Arg32* η-nitrogen for the nucleophilic attack at the C-1 atom of the donor TDP-Rha in the S_N2 reaction. Glu272 does not interact with TDP-Rha (Fig. 2B) or EF-P (Fig. 4B) in our structures, but it does play a critical role in catalysis. In the structure of the TDP complex, a water molecule was observed at hydrogen-bonding distance to the carboxyl groups of Asp16 and Glu272 (Fig. S8A). In the structure of the EF-P complex, clear electron density was observed close to the β-phosphate group of the bound TDP and was modeled as glycerol (Fig. S8B and C), which was probably derived from the crystal cryoprotectant. Each of the two glycerol models in the ASU, which may mimic part of the rhamnose moiety of the donor, hydrogen bonds to the carboxyl group of Glu272 and forms water-mediated hydrogen bonds with the Arg32* η-nitrogen and the side chain carboxyl group of Asp16. Glu272 may work as an acid catalyst. A Michaelis complex structure is needed to fully elucidate the role of Glu272.

FIG 5.

Mass spectrometric analyses to monitor EF-P rhamnosylation by EarP (A) and its mutants (B to E) in vitro. Each experiment was repeated three times. Representative results are shown. WT, wild type.

When we superimposed the structures of the TDP-Rha complex and the EF-P complex, we noted that the rhamnose moiety sterically clashes with a Tyr290 aromatic ring (Fig. 4B) other than Arg32* observed for N. meningitidis EarP (Fig. S6B). Thus, the rhamnose ring should be forced to change its conformation by the conformational change of loop_β14-α10 after EF-P binding. The hydrogen bond between the rhamnose moiety and the carboxyl group of Asp16 in the TDP-Rha complex (Fig. 2B) should be disrupted by the conformational changes of the rhamnose ring and loop_β1-α1 after EF-P binding. Asp16, Tyr112, Glu272, and Arg32* should undergo local conformational changes that facilitate the formation of the catalytically competent Michaelis complex. When Arg32* is monorhamnosylated, it would cause steric tension with the substrate pocket, triggering rhamnosylated EF-P leaving, followed by the release of the second product TDP.

Concluding remarks.

Crystal structures of P. aeruginosa EarP in the apo, TDP- and TDP-Rha-bound forms, as well as in complex with both TDP and full-length EF-P, show stepwise structural changes in EarP that occur upon binding of TDP-Rha and EF-P during catalysis. Based on comparative structural analyses and ITC assays, we proposed a putative catalytic cycle of EarP (Fig. 6). EarP works with the sugar donor binding first and induces conformational changes in substrate-binding loops, followed by the acceptor binding. In the apo-form state, the NTD and CTD are somewhat separated with an open interdomain crevice (Fig. 6, catalytic state a). Upon binding to the donor TDP-Rha, loop_β1-α1 and loop_β10-α7 flip toward the active site (state b). The donor-binding-induced fit facilitates the binding of the sugar acceptor EF-P, accompanied by the movement of loop_β14-α10 (state c) and the formation of glycosidic bond on the α-face of the sugar (state d). The reaction would trigger the substrate-binding loop_β1-α1 and loop_β14-α10 to kick outward to release rhamnosylated EF-P (state e). Finally, upon the release of the second product TDP, the enzyme is turned over to the apo form and ready for the next catalytic cycle (state a).

FIG 6.

Schematic diagram of the putative catalytic cycle. (a to e) Five consecutive catalytic states. The NTD and CTD are colored blue and pink, respectively.

In addition, based on in vitro rhamnosylation assay, Asp16 and Glu272 of P. aeruginosa EarP were identified as potential candidates to catalyze the glycosylation reaction. Asp16 acts as the general base that activates the acceptor Arg32* η-nitrogen for the nucleophilic attack at the C-1 atom of the donor TDP-Rha in the S_N2 reaction.

Finally, structural characterization of P. aeruginosa EarP may provide a platform for the development of new narrow-spectrum antibacterial agents to combat infections from P. aeruginosa and other EarP-containing pathogenic bacteria.

MATERIALS AND METHODS

Cloning, expression, and purification.

The EarP and EF-P genes were amplified by PCR from genomic DNA prepared from P. aeruginosa PAO1 and cloned into a modified pET28a vector, in which the thrombin protease sites were substituted for tobacco etch virus cleavage sites. All of the mutants were generated by using a MutanBEST kit (TaKaRa). All clones were verified by DNA sequencing.

All the proteins were expressed in Escherichia coli BL21-Gold(DE3) cells. Cells were grown in Luria-Bertani medium at 37°C until an optical density at 600 nm of about 0.8 was reached. The proteins were then induced with 0.1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) at 16°C for 24 h. The cell pellet was resuspended and sonicated in buffer A (20 mM Tris-HCl [pH 8.0], 500 mM NaCl). After centrifugation, the supernatant was loaded on a His-Trap nickel-Sepharose column (GE Healthcare), and the protein was eluted with buffer A containing an imidazole gradient. The protein was further purified on a Superdex 200 HiLoad 16/60 column (GE Healthcare) preequilibrated with buffer B (20 mM Tris-HCl [pH 7.5], 200 mM NaCl). The purified proteins were then concentrated for subsequent analysis.

Crystallization, data collection, and structure determination.

All crystallization trials were carried out via the sitting-drop vapor diffusion method with 1 μl of protein solution mixed with 1 μl of reservoir solution at 293K. EarP crystals in the apo form were obtained using both N and C terminally His₆-tagged SeMet-derivative EarP protein with reservoir buffer A [0.1 M 2-(N-morpholino)ethanesulfonic acid (MES; pH 6.5), 12% PEG 20K]. Crystals of SeMet-derivative EarP in complex with TDP-Rha were obtained by soaking SeMet-derivative EarP crystals in reservoir buffer A supplemented with 5 mM TDP-Rha for 5 min. TDP-Rha was purchased from Carbosynth. Crystals of EarP in complex with TDP were grown by cocrystallizing native EarP and TDP-Rha at a molar radio of 1:4 with reservoir buffer A. Although TDP-Rha was used in this crystallization trial, only TDP could be observed in the final structure due to the enzymatic activity of EarP. To crystallize EarP in complex with EF-P, we mixed the purified N-terminally His₆-tagged native EarP protein, SeMet-derivative EF-P protein, and TDP at a molar ratio of 1:1.5:4 and further purified the ternary complex on a Superdex 200 HiLoad 16/60 column. The complex crystals were obtained with reservoir buffer B (0.2 M ammonium sulfate, 0.1 M MES [pH 6.5], 20% PEG 8K). All the crystals were harvested using reservoir buffer containing 25% glycerol as cryoprotectant.

All the X-ray diffraction data sets were collected on beam line 19U1 at Shanghai Synchrotron Radiation Facility (SSRF) at a wavelength 0.978 Å. The data sets were indexed, integrated, and scaled by HKL2000 (40). The initial crystallographic phases were calculated using PHENIX.autoSolve by the single-wavelength anomalous dispersion of SeMet signals from a single crystal of SeMet-derivative EarP (41). An initial model of SeMet-derivative EarP was then automatically built by ARP/wARP and further built and refined using coot and Phenix.refine (41–43). The other complexes were solved by molecular replacement using the program Phaser employing the SeMet-derivative EarP structure as the searching model (44). The EF-P model in the ternary complex was automatically built by ARP/wARP (42). All the structures were further manually built and refined using coot and Phenix.refine (41, 43). All data collection and refinement statistics are summarized in Table 1. Figures were prepared using PyMOL (DeLano Scientific LLC).

TABLE 1.

Data collection and refinement statistics^a

Parameter	Se-EarP	EarP/TDP	Se-EarP/TDP-Rha	Earp/TDP:Se-EF-P
Data collection
Wavelength (Å)	0.978	0.978	0.978	0.978
Space group	P212121	P212121	P212121	C2
Cell parameters
a, b, c (Å)	45.7, 92.2, 119.9	45.3, 92.0, 119.1	45.5, 92.4, 119.5	223.5, 56.7, 132.4
α, β, γ (°)	90, 90, 90	90, 90, 90	90, 90, 90	90, 125.03, 90
Resolution (Å)	40.00–1.75 (1.78–1.75)	40.00–1.85 (1.92–1.85)	40.00–2.15 (2.19–2.15)	40.00–2.30 (2.38–2.30)
Rmerge (%)	8.3 (99.3)	7.5 (69.0)	9.0 (56.9)	11.3 (46.8)
I/σI	57.2 (2.1)	41.7 (2.8)	31.4 (2.2)	22.3 (1.9)
Completeness (%)	98.9 (99.8)	100.0 (100.0)	98.7 (98.9)	98.6 (98.0)
Avg redundancy	7.1 (7.0)	13.1 (13.5)	6.6 (6.9)	3.9 (3.8)
Refinement
No. of reflections (overall)	51,409	43,297	27,742	57,951
No. of reflections (test set)	2,608	2,163	1,352	2,809
Rwork/Rfree (%)	17.52/20.91	16.92/20.15	22.68/25.70	18.81/24.16
No. of atoms
Earp/Efp	3,022/–	3,095/–	3,098/–	6,000/2,797
TDP/TDP-Rha	–/–	25/–	–/35	50/–
Water	352	306	220	354
B factors (Å2)
Earp/Efp	34.22/–	36.28/–	35.66/–	41.7/61.21
TDP/TDP-Rha	–/–	35.27/–	–/26.33	32.33/–
Water	45.08	45.25	36.85	43.00
RMSD
Bond length (Å)	0.006	0.006	0.003	0.002
Bond angle (°)	0.796	0.812	0.636	0.465
Rampage plot residues (%)
Favored	99.19	98.68	97.39	97.91
Allowed	0.81	1.06	2.35	2.09

^aValues in parentheses represent data for the highest-resolution shell. RMSD, root mean square deviation; –, not applicable or not determined.

Isothermal titration calorimetry.

ITC assays were carried out on a MicroCal PEAQ-ITC calorimeter at 293 K. The titration protocol consisted of a single initial injection of 1 µl, followed by 19 injections of 2-µl ligands (TDP, TDP-Rha, or EF-P) into the sample cell containing native P. aeruginosa EarP or its mutants in buffer B at a concentration of about 40 µM with or without 240 µM TDP or TDP-Rha. Thermodynamic data were analyzed with a single-site binding model using MicroCal PEAQ-ITC analysis software provided by the manufacturer and are summarized in Table 2 .

TABLE 2.

Thermodynamic parameters from ITC experiments^a

Proteinb	Ligand	Mean ΔH (kcal/mol) ± SD	TΔS (kcal/mol)	N value	Mean Kd (µM) ± SD
EarP-WT	TDP	6.13 ± 0.394	12.2	1.06	31.4 ± 2.38
EarP-D16N	TDP-Rha	9.13 ± 0.126	16.1	1.16	6.29 ± 0.28
EarP-D16N	EF-P	–5.83 ± 1.82	0.194	0.74	32.7 ± 12.3
EarP-D16N+TDP-Rha	EF-P	–8.55 ± 0.17	–1.43	0.89	4.97 ± 0.34
EarP+TDP	EF-P	15.4 ± 0.134	–6.53	1.03	0.24 ± 0.02
EarP+TDP	EF-P-K29A	17.0 ± 0.123	–8.69	1.01	0.61 ± 0.03
EarP+TDP	EF-P-T52V	–13.9 ± 0.112	–5.24	1.11	0.35 ± 0.03
EarP+TDP	EF-P-K55A	–9.85 ± 3.97	–4.11	1.12	52.6 ± 27.7

^aN, ΔH, and TΔS refer to binding stoichiometry, enthalpy, and entropy, respectively. K_d refers to dissociation constant.

^bThe term “EarP-D16N+TDP-Rha” indicates that 40 µM EarP-D16N and 240 µM TDP-Rha were added to the sample cell. “EarP+TDP” indicates that 40 µM EarP and 240 µM TDP were added to the sample cell.

In vitro rhamnosylation assay.

Purified P. aeruginosa native EarP protein or its mutants were mixed with purified unmodified EF-P protein to concentrations of 5 and 25 µM, respectively, and equilibrated to 37°C in buffer B. The reaction was started by the addition of TDP-Rha to a concentration of 75 µM and stopped after 5 min of incubation at 37°C by the addition of buffer C (0.2% formic acid in water) with an equal volume of the reaction mixture. A 5-µl volume of each reaction mixture containing about 1.25 µg of EF-P protein was injected and analyzed by liquid chromatography (1260 Infinity LC system; Agilent) coupled with mass spectrometry (6530 Accurate-Mass Q-TOF; Agilent). The mobile phase was a linear gradient of 0 to 90% mobile phase B over 10 min (mobile phase A, 0.1% formic acid in water; mobile phase B, 0.1% formic acid in acetonitrile) at a flow rate of 0.2 ml/min. MS spectra were acquired across the range 2,000 to 6,000 m/z and then deconvoluted using a peak modeling deconvolution algorithm in Agilent MassHunter BioConfirm software. Each assay was repeated three times.

Data availability.

The structure coordinates and structure factors of P. aeruginosa EarP in the apo form, in a binary complex with TDP or TDP-Rha, and in a ternary complex with both TDP and full-length EF-P have been deposited in the Protein Data Bank (PDB) under accession numbers 6J7J, 6J7L, 6J7K, and 6J7M, respectively.