The peptidoglycan of Staphylococcus aureus is highly amidated. Amidation of α-D-isoglutamic acid in position 2 of the stem peptide plays a decisive role in the polymerization of cell wall building blocks. S. aureus mutants with a reduced degree of amidation are less viable and show increased susceptibility to methicillin, indicating that targeting the amidation reaction could be a useful strategy to combat this pathogen. The enzyme complex that catalyzes the formation of α-D-isoglutamine in the Lipid II stem peptide was identified recently and shown to consist of two subunits, the glutamine amidotransferase-like protein GatD and the Mur ligase homolog MurT. We have solved the crystal structure of the GatD/MurT complex at high resolution, revealing an open, boomerang-shaped conformation in which GatD is docked onto one end of MurT. Putative active site residues cluster at the interface between GatD and MurT and are contributed by both proteins, thus explaining the requirement for the assembled complex to carry out the reaction. Site-directed mutagenesis experiments confirm the validity of the observed interactions. Small-angle X-ray scattering data show that the complex has a similar conformation in solution, although some movement at domain interfaces can occur, allowing the two proteins to approach each other during catalysis. Several other Gram-positive pathogens, including Streptococcus pneumoniae, Clostridium perfringens and Mycobacterium tuberculosis have homologous enzyme complexes. Combined with established biochemical assays, the structure of the GatD/MurT complex provides a solid basis for inhibitor screening in S. aureus and other pathogens.
Staphylococcus aureus is a frequent constituent of human nasal microflora and a major cause of severe endogenous infections1. Effective treatment of staphylococcal infections remains a worldwide challenge. In the United States alone, Staphylococci are responsible for about 19,000 deaths per year, a number that is higher than that associated with HIV2. Methicillin-resistant Staphylococcus aureus (MRSA) strains, which are resistant to many commonly used antibiotics including methicillin, amoxicillin, penicillin, and oxacillin, represent an increasing challenge to human health worldwide3.
Species-specific cell wall modifications impact on several key aspects of the infection process, including adherence1,4, immune recognition5, and resistance to host defenses6,7. In Gram-positive bacteria such as S. aureus, a thick multilayered peptidoglycan (PG) layer constitutes the major component of the cell wall. The PG is essential for survival and maintenance of cell shape and is crucial to resist osmotic pressure8. The PG heteropolymer consists of alternating disaccharide units composed of N-acetyl-glucosamine (GlcNAc) and N-acetyl-muramic acid (MurNAc), which are cross-linked by short peptides to generate a rigid network.
Assembly of PG is a multistep process that begins in the cytoplasm and terminates on the exterior of the cell (Supplementary Fig. S1). The process is initiated by the MurA-F ligases, which catalyze the formation of the soluble PG precursor UDP-MurNAc-pentapeptide in the cytoplasm9. The membrane-bound enzyme MraY then links this precursor to the membrane carrier undecaprenyl phosphate to yield Lipid I (undecaprenyl-phosphate-MurNAc-pentapeptide), which is then connected with UDP-GlcNAc to form Lipid II (undecaprenyl-phosphate-MurNAc-pentapeptide-GlcNAc) by the glycosyltransferase MurG10. In S. aureus, Lipid II is modified by a Gly5-interpeptide bridge attached by the FemXAB peptidyltransferases11,12, followed by translocation across the cytoplasmatic membrane likely facilitated by the flippases FtsW, RodA and MurJ13. Once it has reached the exterior cell surface, the modified Lipid II is assembled into the growing PG network by penicillin-binding proteins (PBPs) catalyzing transglycosylation and transpeptidation reactions14,15. In particular, transpeptidation has been proposed to require at least one amidated stem peptide16,17.
Amidation of the α-carboxyl group of the D-isoglutamate residue in Lipid II, resulting in the formation of D-isoglutamine16,18, is catalyzed by a recently identified enzyme complex19,20. This complex consists of two proteins, GatD and MurT, which assemble into a binary complex. GatD has sequence similarities to the catalytic domains of glutamine amidotransferases, while MurT is similar in sequence to the substrate-binding domains of Mur ligases. Together, these two proteins catalyze the amidation of α-D-isoglutamic acid of cell wall precursor stem peptides in an ATP-dependent reaction. Mutants that are deficient in Lipid II amidation show a reduction in PG cross-linking and are more susceptible to antibiotics21,22,23. Thus, intervening with the amidation reaction in S. aureus may represent a useful strategy to combat this pathogen.
In order to provide insight into the overall organization of this complex and to facilitate an understanding of the amidation mechanism, we have determined the crystal structure of the GatD/MurT complex. We find that the two proteins assemble into a curved, boomerang-shaped structure, with GatD docking to the C-terminal domain of MurT. Together with mutagenesis data and structural analysis of a complex with an ATP analog, our data provide an excellent foundation to understand the concerted activities of both proteins. Small-angle X-ray scattering (SAXS) experiments confirm that the complex has a similar open conformation in solution, and suggest that some flexibility between the domains exist. Structure-based sequence alignments demonstrate that several other pathogenic organisms have homologous enzyme complexes that likely function in the same manner. In combination with the established in vitro assays, our findings provide the basis for more directed inhibitor screenings.
Formation and characterization of the GatD/MurT complex
Full-length GatD and MurT were co-expressed as described19, and the complex was purified using nickel affinity chromatography. A final size exclusion chromatography step demonstrated that the two proteins elute together, forming a stable complex in solution. The elution volume in size exclusion chromatography corresponds to an estimated molecular weight of 72 kDa, which is consistent with the calculated molecular weight of 78.8 kDa for a binary GatD/MurT complex. The SAXS data (see below) also clearly indicate that one copy of GatD and one copy of MurT assemble into a stable heterodimer.
Overall structure of GatD/MurT
The native structure of the GatD/MurT complex was solved at a resolution of 2.04 Å using single isomorphous replacement with anomalous scattering (SIRAS). The refined structure has excellent statistics (Table 1) and includes all residues of the expressed proteins with the exception of MurT residues 1–35, 195–196 and 434–437. These regions are poorly visible in the electron density maps and therefore likely have multiple conformations and increased mobility. The GatD/MurT heterodimer adopts a boomerang-shaped conformation, with GatD packing against the C-terminal domain of MurT (Fig. 1). As previously postulated19 and recently shown24, GatD exhibits a class-I glutamine amidotransferase-like fold. A DALI search25 identifies the enzymes HisH, PdxT and PurQ from Thermotoga maritima as the closest structural homologs (Z-values of 17.4, 15.8 and 15.0, respectively). Superimposition of GatD with structures obtained from a secondary structure-based search using HHPRED26 reveals a well-conserved core architecture, with root-mean-square deviation (r.m.s.d.) values of 2.6 Å (all C-α atom pairs) over the entire length of GatD for the closest structural homolog, HisH. However, GatD distinguishes itself from other, homologous structures through the presence of an extended C-terminal helix, termed helix α7 (Fig. 1c). This helix mediates many of the contacts with MurT, explaining its presence in GatD.
MurT contains the Mur ligase middle and C-terminal domains typical for the Mur ligase family. The C-terminal domain is built around a central six-stranded, predominantly parallel β-sheet (Fig. 1c,e) that is sandwiched between four α-helices on one (αc1–4) and two α-helices (αc5-6) on the other side. The middle domain is constructed around a crescent-shaped, nine-stranded predominantly parallel β-sheet, which encloses a three-helix bundle (αm1, αm7-8). The ATP molecule required for catalysis, replaced here by the non-hydrolyzable analog β,γ-imidoadenosine 5′-triphosphate (AMPPNP), is bound at the base of helix αm1 with the adenine moiety bound in a cleft formed next to helix αm1 while the phosphates are contacted by residues from strands βm1 and βm3 (Fig. 1d). The central β-sheet is capped on the opposite side by five α-helices. A RanBP-type Zinc-ribbon27 is located on the side of the middle domain as an insertion between β-strands βm6 and βm9. This feature is missing in homologous proteins. A search with DALI25 yielded MurF enzymes from Pseudomonas aeruginosa, Escherichia coli and Acinetobacter baumannii as the closest structural homologs, with Z-values of 23.2, 22.6, and 21.7, respectively. Mur ligases typically contain a third, N-terminal domain, which is not present in MurT. Instead, MurT only contains a truncated and likely flexible N-terminus (residues 1–37), which is not visible in our electron density maps.
It can sometimes be challenging to assign correct physiologic contacts from crystal structures alone as crystal packing can offer alternative solutions for possible contacts between two subunits that are often not easy to distinguish from physiologic contacts28. In order to validate the observed interaction between GatD and MurT, all possible GatD/MurT interfaces in the crystal were analysed with the PISA28 and EPPIC29 servers. Both algorithms clearly classify only the interaction depicted in Fig. 1c as physiologic. Other possible interactions between GatD and MurT have much smaller buried surface areas and are classified as crystal packing artifacts. Moreover, they do not place critical catalytic residues (see below) in close proximity.
The overall structure of the GatD/MurT complex is also supported by solution SAXS experiments (Fig. 2). Scattering profiles derived from dilution series experiments were used to refine the crystallographic model. After rigid body refinement, the overall fit of a calculated scattering curve to the experimental data improved from an initial χ2 = 25.7 to χ2 = 2.2 (Fig. 2b). The refined model retains the overall characteristics of the crystal structure, but it has a more compact conformation with a closest distance between GatD and the MurT middle domain, measured at the top of the structure, of 22 Å instead of 32 Å (Fig. 2a). These experiments show that, while the conformation of the unliganded GatD/MurT complex is indeed open and elongated as observed in the crystal structure, the complex can breathe somewhat in solution. Unfortunately, SAXS data obtained from titrations with AMPPNP and a soluble Lipid II mimic (UDP-MurNAc-L-Ala-D-Glu-γ-L-Lys-D-Ala-D-Ala) resulted in moderate to severe aggregation of GatD/MurT (Supplementary Fig. S2) and could thus not be used for modelling. However, the observed change in protein solubility clearly suggests a considerable conformational rearrangement in the protein upon ATP and Lipid II binding.
Interactions between GatD and MurT
Extensive interactions between GatD and MurT establish a large, continuous interface that buries a total surface area of 940 Å2 from solvent (calculated with PISA28). Residues involved in interface formation are contributed by two consecutive MurT α-helices (αc2/3 and αc4), which tightly pack against two α-helices and a loop of GatD (helices α2 and α7, loop β13-α6 Figs 1 and 3). One of these α-helices is the C-terminal helix α7 absent in homologous class-I glutamine amidotransferases, which do not form similar complexes with MurT. The interface is entirely devoid of solvent molecules. Two large aromatic (GatD-F146, MurT-Y354) and several aliphatic side chains (I20, I24, P194, and V231 in GatD and I353, L381 and L385 in MurT) that would otherwise be solvent-exposed are buried upon formation of the GatD/MurT contact. Figure 3a shows interactions made by the MurT-Y354 and MurT-L381 side chains as an example. MurT-Y354 also contacts two prolines in the GatD β13-α6 loop (P191 and P194), with P191 being especially well conserved throughout GatD homologs but not across functionally unrelated glutamine amidotransferases. These more centrally located hydrophobic interactions are augmented by intermolecular salt bridges (GatD-K31/MurT-E387, GatD-E225/MurT-K384, GatD-K195/MurT-E378, GatD-R235/MurT-D342), which are located towards the edges of the interface, and by hydrogen bonds (Fig. 3b,c). Residues participating in interface formation are highly conserved across species that have homologous proteins, as demonstrated by the conservation analysis shown in Fig. 3d. The interface also contains several residues with particularly small side chains (G21, A25, G190, A224, and A228 in GatD and S351 in MurT), and in each case a larger side chain would be incompatible with the observed interface.
Catalytic site of GatD
Class-I glutamine amidotransferases contain a highly conserved catalytic triad in their active site, which consists of a cysteine responsible for the initial nucleophilic attack on the substrate and nearby histidine and glutamic acid residues to form a proton relay chain. Even remotely related members of this enzyme family superimpose very well for the active site (Fig. 4). Earlier studies suggested that GatD residues C94 and H189 might form part of a catalytic triad19. In homologous GATases, the third residue of this triad is typically a glutamic acid that is located two residues downstream of the histidine in a conserved HPE tripeptide sequence30. This glutamic acid is replaced with a proline (P191) in GatD and separated from H189 by a glycine, resulting in a HGP motif that is well conserved throughout putative GatD enzymes. However, as proline cannot function as a proton relay in a catalytic triad, GatD enzymes must either function with only a catalytic dyad, as recently suggested24, or have arrived at a different solution to establish such a triad.
The structurally homologous enzymes PdxT from Bacillus subtilis (PDB ID 2nv0), two homoserine O-acetyltransferases from Bacillus cereus and Thermotoga maritima (PDB IDs 2vfj and 2h2w, respectively), and the glutamine amidotransferase Mfla_0438 from Methylobacillus flagellatus (PDB ID 3m3p), obtained from a HHPRED search, align well with the overall structure of GatD. A closer inspection of the active site architecture reveals that GatD-C94 superimposes well with the active site cysteine residues from these enzymes (Fig. 4a). The sulfhydryl groups in particular are located in almost identical positions. However, the side chain of GatD-H189 does not overlay well with the histidines from the related enzymes (Fig. 4a). Instead, GatD-H189 is shifted towards the heterodimer interface, where it contacts an aspartic acid (D349) of MurT, thus establishing a possible alternative catalytic triad (GatD-C94, GatD-H189, MurT-D349) involving residues from both GatD and MurT. A superimposition of our structure with the recently published structure of monomeric GatD24 (PDB ID 5n9m) shows that the interaction with MurT is required to fully position the loop containing GatD-H189 (Supplementary Fig. S3). The close proximity of the side chains of GatD-C94 and GatD-H189 (3.7 Å) and GatD-His189 and MurT-D349 (3.1 Å) as well as their relative orientation suggest that these three residues might indeed function as a proton relay system in catalysis. GatD-H189 is followed by a short hydrophobic sequence framed by two proline residues, GatD-P191 and GatD-P194, that are involved in contacts with MurT (Fig. 3a). The arrangement of the putative catalytic triad is therefore dependent on correct assembly of the binary complex. As shown in Fig. 4, sequence comparison with confirmed31 and predicted homologs19 as well as a sequence- and structure based conservation analysis revealed the combination of GatD-H189 and the subsequent hydrophobic stretch to be highly conserved, suggesting that the GatD homologs possess similar catalytic triads and interfaces with corresponding MurT homologs. Similarly, the sequence context of MurT-D349 is conserved within MurT homologs, but not across other Mur ligases.
In order to investigate whether the spatial proximity of MurT-D349 to GatD-C94 and GatD-H189 is indicative of a possible involvement in catalysis, a MurT-D349N mutant was generated. In vitro amidation assays of Lipid II revealed severely reduced in vitro activity of this mutant compared to the wildtype GatD/MurT complex (Fig. 4, Supplementary Fig. S4). A conservative mutation of the active site cysteine (GatD-C94S) to serine completely abolished activity in vitro, similarly to the previously characterized, less conservative GatD-C94G mutant19. The folds and thermal stabilities of the mutants were assessed with circular dichroism (CD) spectroscopy and thermal shift assay (TSA), respectively. Both MurT-D349N and GatD-C94S possess a fold indistingushable from that of the wildtype protein (Supplementary Fig. S5) as well as virtually identical melting temperatures (Supplementary Fig. S6, Table S1) that lie well above the temperature used for activity assays. Taken together, the mutagenesis experiments demonstrate that both GatD-C94 and MurT-D349 are likely relevant for catalytic activity of the complex, and they therefore also provide additional support for the physiologic nature of the observed GatD/MurT interface.
The nucleotide binding site of MurT
To obtain insight into interactions of the GatD/MurT complex with ATP, we prepared a ternary complex by soaking crystals with 2.1 mM of AMPPNP, an ATP analog. The structure of this complex was solved at 2.14 Å resolution using molecular replacement (Table 1), and the corresponding electron density allowed us to unambiguously model AMPPNP into a binding site in the middle domain of MurT (Figs 1d and 5, Supplementary Fig. S7). The binding site contains the consensus sequence GTNGKT19, with residues K59 stabilizing the β and γ phosphate groups of AMPPNP and T60 coordinating a magnesium ion that is also contacted by the side chain of the conserved MurT-E108 residue. The magnesium ion helps in positioning the AMPPNP β and γ phosphates from the opposite side (Fig. 5a). The position of the observed ATP binding pocket is strictly conserved throughout Mur ligases as structural superposition of the middle domain of MurT with the middle domains of close sequence homologues MurE from S. aureus32 (PDB ID 4c12) and MurF from Acinetobacter baumannii33 (PDB ID 4qdi) revealed nearly identical modes of nucleotide-binding (Fig. 5b). The observed binding mode is in excellent agreement with our mutagenesis data, as single or double mutants of nucleotide-binding site residues (T60A, E108A and N267Y) have completely lost their enzymatic activity (Fig. 5c, Supplementary Fig. S8). As with the previously described mutants, protein fold and stability were confirmed via CD spectroscopy and TSA (Supplementary Figs S5 and S6, Table S1).
Conservation analysis based on a redundance-corrected alignment of 145 homologous sequences, as automatically performed on the ConSurf server34 and projected onto the surface of the GatD/MurT model, revealed a high level of conservation at the inside of the crescent-shaped binary complex. While the outwards-facing surface of the protein, including that of the Zinc finger, is variable in sequence, the inner surface residues, which contain the ligand-binding regions of other Mur ligases, are strongly conserved. This includes the aforementioned components of the putative catalytic triad composed of GatD-C94, GatD-H189 and MurT-D349, as well as the MurT-GTNGKT nucleotide binding motif and MurT-E108 (Fig. 6), which coordinates the magnesium ion that lies in proximity to the bound ATP analog. The consensus sequence D(D,N)P(N,A) in the amino acid‐binding pocket of MurE from Thermotoga maritima was found to mediate the recognition of L-lysine35. Structural work by Ruane and colleagues revealed that the Staphylococcus aureus MurE protein contains additional, often charged residues that are important for the interaction with L-lysine, e. g. Asp-406 and Glu-46032.
The MurT sequence DNAAD containing the conserved aspartate residue was previously suggested to be involved in interactions with the lysine residue of the Lipid II stem peptide19. Conservation mapping revealed that this motif is conserved and located in a large, concave area between the GatD catalytic triad and the ATP binding site in the middle domain of MurT (Fig. 7). Thus, it seems reasonable that this region is functionally relevant, either by engaging parts of Lipid II or by mediating conformational changes.
The recently characterized GatD/MurT complex is responsible for the amidation of glutamic acid in the peptide stem structure of the S. aureus PG19. Inhibition of this reaction leads to lower growth rate, reduced resistance to beta-lactam antibiotics, and increased sensitivity to lysozyme in MRSA strains19,20. Strategies to interfere with peptide stem amidation could therefore have potential for combatting S. aureus, including strains that are resistant to currently available antibiotics. Precise knowledge about the three-dimensional structure, organization and catalytic mechanism of the GatD/MurT complex is invaluable for fully exploiting this potential.
The crystal structure of GatD bound to MurT establishes an initial framework for probing the function of this enzyme complex and for developing strategies to intervene with amidation. The two proteins assemble into an elongated complex, with a large open space separating GatD from the middle domain of MurT. The previously identified catalytically active cysteine (GatD-C94)19 is located near the GatD/MurT interface, and is part of a catalytic triad that includes additional residues from both proteins (GatD-H189, MurT-D349). As our mutagenesis data confirm the importance of both GatD-C94 and MurT-D349 for catalysis while recent findings by Leisico and coworkers24 corroborate the crucial role of GatD-H189, the location of this putative catalytic triad nicely explains the requirement of the assembled complex for activity, and it is in agreement with the observed interface between the two proteins. Complex formation with an ATP analog revealed a nucleotide-binding site in MurT that is similar to those found in other Mur ligases. It is conceivable that the large open space separating the GatD/MurT catalytic triad and ATP binding site may accommodate a portion of the bulky Lipid II substrate. The SAXS experiments demonstrate that the GatD/MurT complex has a similar open conformation in solution, and they also show that small movements of the terminal regions towards each other are possible. Such movements might be more pronounced in the presence of Lipid II, as the enzyme might partially close around it during catalysis.
The hydrophobicity of Lipid II and the lack of suitable, commercially available less hydrophobic Lipid II alternatives did not allow us to effecitively pursue a complex of GatD/MurT bound to Lipid II. Soaking or co-crystallization experiments with soluble mimics of the peptide stem alone (e.g. glutamate, or commercially available peptides based on the S. aureus peptide stem sequence) did not result in complex formation, suggesting that additional portions of the Lipid II molecule are required for binding.
The S. aureus enzyme GatD/MurT likely belongs to a distinct group of enzymes that is strongly conserved throughout a subset of bacteria exhibiting a high degree of peptidoglycan amidation16, as suggested by conservation analysis19. The organisms that express GatD/MurT homologs include a number of established human pathogens, such as Streptococcus pneumoniae, for which the GatD/MurT enzyme complex was recently characterized31, and, likely, Mycobacterium tuberculosis. Knowledge about the structure and catalytic mechanism of GatD/MurT will therefore transcend the S. aureus field.
Structurally, GatD belongs to the family of glutamine amidotransferases (GATases), whose members are involved in a multitude of biosynthetical processes. While GATases exhibit a wide variety of substructure variants due to differing substrates or co-substrates, their core architecture is highly conserved, including that of GatD. The elongated C-terminal loop leading to an additional C-terminal α-helix (helix α7) is, however, unique to GatD. This segment contributes 36% of the dimer interface with MurT, and the length of the linker connecting it to the rest of the protein suggests that it may allow for a certain degree of flexibility. Thus, the linker could help mediate breathing of the complex, which would explain the observed differences in the overall GatD/MurT conformation between the crystal structure and the solution scattering data (Fig. 2).
While MurT lacks the N-terminal domain typically found in Mur ligases, the overall folds of the middle and C-terminal MurT domains agree very well with the equivalent domains in published structures of different Mur ligases from various organisms. Although the position of the observed ATP-binding pocket is strictly conserved throughout Mur ligases (Fig. 5), the angle between the middle and C-terminal domains varies significantly across different enzymes, indicating an inherent flexibility of this feature. The almost linear arrangement between the two MurT domains appears to be unique to MurT as all other known conformations of this two-domain segment are more closed in Mur ligases. The two most striking structural differences between MurT and related enzymes are the truncated N-terminal domain and the Cys4 Zinc-ribbon situated at the side of the middle domain. While the role of this last feature is unclear, it forms a very exposed structure protruding from the main bulk of the MurT middle domain. Zinc fingers of this class have been previously shown to provide a platform for protein-protein interactions36. The zinc finger is less conserved compared with many other features of GatD/MurT, and it appears to be present only in a subset of bacteria with putative gatD/murT genes.
The truncated N-terminus (with respect to other Mur ligases), on the other hand, is a well-conserved feature of MurT in different organisms. The closely related MurC-F enzymes catalyze the consecutive addition of amino acids of the peptide stem to UDP-activated MurNAc in the initial phase of peptidoglycan biosynthesis. The roughly 90 residues long N-terminal domain present in these enzymes plays an important role in substrate recognition by interacting with its UDP moiety. In contrast, MurT has a much shorter N-terminus of only 35 residues, which appears to be structurally flexible and poorly ordered as it could not be observed in the crystal structure. Furthermore, the positioning of the pyrophosphate linker in the UDP-conjugated substrates of ligand-bound structures of S. aureus MurE (PDB ID 4c12) and Pseudomonas aeruginosa MurF (PDB ID 4cvk) coincides with the boundary between their N-terminal and middle domains. In the GatD/MurT substrate Lipid II, the pyrophosphate linkage connects the sugar moiety to the undecaprenyl membrane anchor, possibly marking the position at which Lipid II emerges from the membrane. Hence, a UDP-MurNAc-binding N-terminal domain is not only unnecessary, it would also likely clash with the plasma membrane.
How does Lipid II bind to GatD/MurT, and how does catalysis proceed? Prompted by the drastically altered solubility of the protein observed by SAXS upon minimal substrate mimic addition (Supplementary Fig. S2) and based on previously published data revealing that MurT is able to replace MurE in vitro19 as well as on alignments with the substrate-bound structures of MurE and MurF, which consistently show closed conformations, an analogous structural rearrangement upon substrate binding can be modeled for MurT. To generate this model, the C-terminal domains of MurE or MurF were superimposed with that of MurT, and the MurT middle domain was then moved to fit those of MurE or MurF, producing a more closed conformation of MurT that mimics those present in the other two enzymes (Fig. 7). If the GatD structure is then added to the modified MurT using the known GatD/MurT interface, it becomes clear that the distance between the MurT middle domain and GatD is drastically reduced. The gap between GatD-C94 and the γ-phosphate of AMPPNP diminishes from over 40 Å to just over 20 Å. In such a putative more closed conformation of GatD/MurT, the peptide stem of Lipid II would be sandwiched between the MurT middle and C-terminal domains in linear continuation of the triphosphate axis of the nucleotide (Fig. 7b). It is conceivable that channeling of nascent ammonia from the active site of GatD to the muropeptide substrate in MurT, which would then lie in close proximity to the nucleotide, may be enabled by such a closed conformation. The DNADD motif, which was previously suggested to be involved in the engagement of L-lysine in the muropeptide sequence19, indeed comes into close proximity with the modeled MurE product, which was also extracted from PDB ID 4c12. Due to the direct backbone connection with MurT-D349, the relative position of the residues forming the putative catalytic triad may be influenced by Lipid II binding to the enzyme. Thus, hydrolysis of glutamine in the absence of an acceptor substrate and hence escape and accumulation of ammonia in the cell could be avoided.
The model presented here was obtained by domain superimposition with known substrate-bound structures of homologous proteins, and it is only meant to approximate the possible conformations of ligand-bound GatD/MurT and the Lipid II substrate. While the superimposition predicts a closed conformation of substrate-bound GatD/MurT, some residues located at the newly formed interface between the MurT C-terminal and middle domains clash both with the respective other domain and with the modeled MurE ligand, and this demonstrates that additional movements in the protein are required to properly accomodate the substrate. As the physiologic substrate of MurT, Lipid II, differs from that of MurE, it would not be surprising if the actual active conformation of GatD/MurT may be a few degrees more open to allow accommodation of the large Lipid II molecule. The precise conformation of a Lipid II-bound complex of GatD/MurT is still unknown and likely somewhat different from the one shown in Fig. 7. We therefore believe that performing in-depth molecular docking calculations would be inappropriate at this time.
Our model suggests that the mechanism of Lipid II amidation likely starts with an ATP-bound, open conformation of GatD/MurT similar to that observed in our crystal structure. A domain rearrangement between the two MurT domains to generate a more closed conformation could be triggered by the binding of the substrate peptide, and this would enable both the hydrolysis of glutamine by GatD through movement of the triad residues and the correct positioning of the free carboxylate of D-iso-glutamate in the peptide stem of Lipid II with respect to ATP, thus allowing for a concerted amidation reaction. Of course, this hypothesis needs to be tested in future experiments.
Peptidoglycan is characteristically modified in many Gram-positive pathogens and amidation is known to affect the level of methicillin resistance21,37 and to contribute to vancomycin susceptibility in S. aureus38, and thus inhibition of the GatD/MurT complex may represent a strategy to combat S. aureus and perhaps also other related bacterial pathogens that amidate their peptide stem in a similar manner. The nucleotide-binding region of MurT may provide a good base for developing multiple-target inhibitors, owing to the similarity of this site with those of other Mur ligases. However, as most monomeric Mur ligases are also present in beneficial bacteria such as E. coli or other components of our gut microbiota, it may be desirable to develop drugs more specifically aimed at GatD/MurT. As the GatD/MurT interface is highly conserved and likely unique to complexes between these two proteins, interfering with the interaction between GatD and MurT could selectively inactivate the amidation reaction. It is conceivable that small molecules that mimic components of the interface could serve to block the reaction carried out by the GatD/MurT complex.
Materials and Methods
Expression and purification
A pET21b vector (Novagen) containing the gatD/murT operon for co-expression of untagged MurT with GatD as a C-terminal His-tag fusion protein  was introduced into E. coli BL21 (Promega) cells. The bacteria were grown in LB-medium (75 µg/mL ampicillin) at 37 °C. At an OD600 of 0.6, the cultures were cooled to 30 °C and 0.25 mM IPTG was added for induction. Cells were harvested by centrifugation after 5 hours and stored at −20 °C until purification.
Frozen pellets were resuspended in lysis buffer (50 mM Tris pH 8.5 at 4 °C, 300 mM NaCl, 10 mM imidazole, 1 mM DTT). Next, 200 µg/mL lysozyme, 1:10000 DNAse I (ThermoFisher) and 1 mg RNAse A (Sigma) were added and cells were incubated on ice for 30 min. Following lysis by sonication, the lysate was clarified by centrifugation and added to a Ni-NTA-agarose slurry (Novagen). The mixture was incubated on a tilt shaker at low rpm for 18 h at 4 °C and subsequently transferred to a EconoPac gravity flow column (BioRad). Following a wash with lysis buffer, weakly bound material was removed by washing with lysis buffer containing 50 mM imidazole, and the target protein was then eluted with lysis buffer containing 350 mM imidazole. Buffer exchange to SEC I buffer (50 mM Tris pH 8.2 @ 4 °C, 500 mM NaCl, 50 mM MgCl2, 5 mM KCl, 5 mM DTT) was carried out directly afterwards using PD-10 desalting columns (GE). Sample homogeneity was achieved by means of two subsequent iterations of preparative size exclusion chromatography using a HiLoad 26/60 column packed with Superdex 200 column material (GE Healthcare). In the second iteration, 5 mM TCEP was substituted for DTT as a reducing agent (Buffer SEC II). Throughout the purification, sample purity was assessed by SDS-PAGE. The protein was freshly prepared for each analysis.
Crystallization and structure determination
Crystals of GatD/MurT were grown in a sitting-drop vapor diffusion setup using protein concentrations of 2–5 mg/mL in a broad range of Tris-buffered conditions containing PEG8000, MgCl2 and glycerol. The conditions from which the crystals for this study were obtained are 0.1 M Tris pH 8.6, 40% (w/v) PEG8000, 0.35 M MgCl2 for the native data, 0.1 M Tris pH 9.0, 26.7% (w/v) PEG8000, 0.35 M MgCl2 for the thiomersal derivative, and 0.1 M Tris pH 9.1, 18% (w/v) PEG8000, 14% glycerol, 0.35 M MgCl2 for the AMPPNP soak, respectively. Cryoprotection was performed by adding 20% MPD or 25% glycerol to the reservoir solution, and crystals were flash-cooled in liquid nitrogen. X-ray data were collected at beamlines X10SA and X06DA of the Swiss Light Source (SLS) in Villigen (Switzerland). Data were integrated and reduced using the XDS program package39.
Crystals were derivatized with 10 mM thiomersal for phase determination and a mixture of 2.1 mM of the ATP analogue β,γ-imidoadenosine 5′-triphosphate (AMPPNP), 2.1 mM of the glutamine analogue 6-Diazo-5-oxo-L-norleucine (DON), and 2.5 mM UDP-MurNAc-L-Ala-D-Glu-γ-L-Lys-D-Ala-D-Ala for evaluating complex formation. Of the latter three compounds, only AMPPNP was found to bind to the crystals based on the inspection of difference electron density maps. Derivatization was performed by transferring native crystals into drops of otherwise equal composition containing the compound to be soaked. After soaking times of 10 to 30 min, crystals were cryoprotected and flash-cooled in liquid nitrogen.
Phasing and initial model building
A dataset obtained from a crystal derivatized with 10 mM thiomersal was used to extract initial phase information using the single isomorphous replacement with anomalous scattering (SIRAS) approach. The derivative data were scaled against the native data using CAD and Scaleit of the CCP4 program suite40. Using the program suite autoSHARP41,42, a substructure of 13 mercury sites was determined, followed by heavy atom refinement and density modification. Initial automated model building was performed using the AutoBuild Wizard43 in the PHENIX program suite prior to transfer to the native data set.
Model building and refinement
Manual model building was carried out in Coot44 and alternated with reciprocal space refinement using the programs REFMAC545 and PHENIX.REFINE46. The data set collected from crystals soaked with AMPPNP was phased by molecular replacement using the program PHASER47. Prior to model building, simulated annealing was performed using PHENIX.REFINE in order to avoid model bias from the original dataset.
Small-angle X-ray scattering (SAXS) data were collected at beamline P12 at the German Electron Synchrotron (DESY) in Hamburg, Germany. Triplicates of a two-fold dilution series of GatD/MurT in SEC II buffer ranging from 0.5 to 8 mg/mL were recorded and extrapolated to zero concentration to remove possible concentration effects using the PRIMUS interface of the ATSAS program suite48. Rigid-body refinement of the native crystal structure was carried out with the program SREFLEX49. Synthetic scattering profiles for comparison with the experimental data were generated using CRYSOL50. SAXS experiments were repeated in the presence of increasing concentrations of the ATP analog AMPPNP (10 µM–5 mM) and the soluble Lipid II mimic UDP-MurNAc-L-Ala-D-Glu-γ-L-Lys-D-Ala-D-Ala (8 µM–1 mM).
Mutagenesis and in vitro amidation assay
Site-directed mutagenesis was performed according to the manufacturer´s instructions using plasmid pET21-murT/gatD as the template to generate active site mutants GatD-C94S and MurT-D349N (QuikChange Lightning Site-Directed Mutagenesis Kit, Agilent) and to generate mutations in the ATP binding site (MurT mutants T60A, E108A, N267Y and double mutant T60A E108A; Q5 site-directed mutagenesis kit, New England Biolabs). Proteins were expressed and purified and impact of mutations on enzyme activity were tested as previously described19.
Circular dichroism (CD) spectroscopy
CD spectroscopy experiments were performed on a JASCO J-720 spectrometer under nitrogen flow. Protein concentration was set to 0.3 mg/mL in ten-fold diluted SEC II buffer. The resulting spectra were corrected for concentration, protein size and cuvette thickness. Mutant protein spectra were compared with the wildtype.
Thermal shift assay (TSA)
TSA was performed on a QuantStudio 5 real-time PCR cycler (Applied Biosystems, Thermo Fisher Scientific). Wildtype and mutant GatD/MurT were supplied at 0.3 mg/mL and Protein Thermal Shift DyeTM (Applied Biosystems, Thermo Fisher Scientific) was added as a TSA fluorophore. The samples were equilibrated at 4 °C and gradually heated to 95 °C over 30 min while monitoring dye fluorescence. The averaged protein melting temperature for the wildtype and each mutant was derived from the melting curve inflection point using seven technical replicates and the Protein Thermal ShiftTM Software v1.3 (Applied Biosystems, Thermo Fisher Scientific).
Homology searches, conservation analysis and interface mapping
Sequence-based conservation analysis was performed using ConSurf34. Herein, an initial search was automatically performed separately for GatD and MurT using BLAST (https://blast.ncbi.nlm.nih.gov/) and a multiple sequence alignment (MSA) of 145 sequences each was generated using MAFFT51. The resulting conservation scores were then projected onto the GatD/MurT crystal structure. Color coding of conservation was done by setting the average conservation score of the alignment to zero. More or less conserved residues were then colored according to a color ramping scheme, with dark blue and red indicating highest and lowest conservation scores, respectively. Structural homology searches were conducted at a secondary structure level using the HHPRED tool implemented in the bioinformatics toolkit of the Max-Planck-Institute for Developmental Biology in Tübingen52.
Finally, in order to identify related proteins based on their entire structure and quantify the structural similarities, a DALI search25 was performed with the final crystal structure against the entire Protein Data Bank (PDB) repository at www.rcsb.org.
The GatD/MurT structures described here have been deposited with the Protein Data Bank (www.rcsb.org) with PDB IDs 6GS2 (unliganded GatD/MurT) and 6H5E (complex with AMPPNP).
Foster, T. J. The Staphylococcus aureus “superbug”. J Clin Invest 114, 1693–1696 (2004).
Zeller, J. L., Burke, A. E. & Glass, R. M. JAMA patient page. MRSA infections. JAMA 298, 1826 (2007).
Roemer, T., Schneider, T. & Pinho, M. G. Auxiliary factors: a chink in the armor of MRSA resistance to beta-lactam antibiotics. Curr Opin Microbiol 16, 538–548 (2013).
Weidenmaier, C. et al. Role of teichoic acids in Staphylococcus aureus nasal colonization, a major risk factor in nosocomial infections. Nat Med 10, 243–245 (2004).
Fournier, B. & Philpott, D. J. Recognition of Staphylococcus aureus by the innate immune system. Clin Microbiol Rev 18, 521–540 (2005).
Kraus, D. & Peschel, A. Molecular mechanisms of bacterial resistance to antimicrobial peptides. Curr Top Microbiol Immunol 306, 231–250 (2006).
Peschel, A. & Sahl, H. G. The co-evolution of host cationic antimicrobial peptides and microbial resistance. Nature reviews. Microbiology 4, 529–536 (2006).
Vollmer, W., Blanot, D. & de Pedro, M. A. Peptidoglycan structure and architecture. FEMS Microbiol Rev 32, 149–167 (2008).
Barreteau, H. et al. Cytoplasmic steps of peptidoglycan biosynthesis. FEMS Microbiol Rev 32, 168–207 (2008).
Bouhss, A., Trunkfield, A. E., Bugg, T. D. & Mengin-Lecreulx, D. The biosynthesis of peptidoglycan lipid-linked intermediates. FEMS Microbiol Rev 32, 208–233 (2008).
Hegde, S. S. & Shrader, T. E. FemABX family members are novel nonribosomal peptidyltransferases and important pathogen-specific drug targets. J Biol Chem 276, 6998–7003 (2001).
Schneider, T. et al. In vitro assembly of a complete, pentaglycine interpeptide bridge containing cell wall precursor (lipid II-Gly5) of Staphylococcus aureus. Mol Microbiol 53, 675–685 (2004).
Mohammadi, T. et al. Identification of FtsW as a transporter of lipid-linked cell wall precursors across the membrane. Embo J 30, 1425–1432 (2011).
Scheffers, D. J. & Pinho, M. G. Bacterial cell wall synthesis: new insights from localization studies. Microbiol Mol Biol Rev 69, 585–607 (2005).
van Heijenoort, J. Lipid intermediates in the biosynthesis of bacterial peptidoglycan. Microbiol Mol Biol Rev 71, 620–635 (2007).
Schleifer, K. H. & Kandler, O. Peptidoglycan types of bacterial cell walls and their taxonomic implications. Bacteriol Rev 36, 407–477 (1972).
Nakel, M., Ghuysen, J. M. & Kandler, O. Wall peptidoglycan in Aerococcus viridans strains 201 Evans and ATCC 11563 and in Gaffkya homari strain ATCC 10400. Biochemistry 10, 2170–2175 (1971).
Gotz, F., Bannerman, T. & Schleifer, K. H. The Genera Staphylococcus and Macrococcus. Prokaryotes: A Handbook on the Biology of Bacteria, Vol 4, Third Edition, 5–75 (2006).
Munch, D. et al. Identification and in vitro analysis of the GatD/MurT enzyme-complex catalyzing lipid II amidation in Staphylococcus aureus. PLoS Pathog 8, e1002509 (2012).
Figueiredo, T. A. et al. Identification of genetic determinants and enzymes involved with the amidation of glutamic acid residues in the peptidoglycan of Staphylococcus aureus. PLoS Pathog 8, e1002508 (2012).
Gustafson, J., Strassle, A., Hachler, H., Kayser, F. H. & Berger-Bachi, B. The femC locus of Staphylococcus aureus required for methicillin resistance includes the glutamine synthetase operon. J Bacteriol 176, 1460–1467 (1994).
Stranden, A. M., Ehlert, K., Labischinski, H. & Berger-Bachi, B. Cell wall monoglycine cross-bridges and methicillin hypersusceptibility in a femAB null mutant of methicillin-resistant Staphylococcus aureus. J Bacteriol 179, 9–16 (1997).
Boyle-Vavra, S., Labischinski, H., Ebert, C. C., Ehlert, K. & Daum, R. S. A spectrum of changes occurs in peptidoglycan composition of glycopeptide-intermediate clinical Staphylococcus aureus isolates. Antimicrob Agents Chemother 45, 280–287 (2001).
Leisico, F. et al. First insights of peptidoglycan amidation in Gram-positive bacteria - the high-resolution crystal structure of Staphylococcus aureus glutamine amidotransferase GatD. Sci Rep 8, 5313 (2018).
Holm, L. & Laakso, L. M. Dali server update. Nucleic Acids Res 44, W351–355 (2016).
Zimmermann, L. et al. A Completely Reimplemented MPI Bioinformatics Toolkit with a New HHpred Server at its Core. J Mol Biol (2017).
Malgieri, G. et al. The prokaryotic zinc-finger: structure, function and comparison with the eukaryotic counterpart. The FEBS journal 282, 4480–4496 (2015).
Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from crystalline state. J Mol Biol 372, 774–797 (2007).
Duarte, J. M., Srebniak, A., Scharer, M. A. & Capitani, G. Protein interface classification by evolutionary analysis. BMC Bioinformatics 13, 334 (2012).
Massiere, F. & Badet-Denisot, M. A. The mechanism of glutamine-dependent amidotransferases. Cell Mol Life Sci 54, 205–222 (1998).
Liu, X. et al. High-throughput CRISPRi phenotyping identifies new essential genes in Streptococcus pneumoniae. Mol Syst Biol 13, 931 (2017).
Ruane, K. M. et al. Specificity determinants for lysine incorporation in Staphylococcus aureus peptidoglycan as revealed by the structure of a MurE enzyme ternary complex. J Biol Chem 288, 33439–33448 (2013).
Cha, S. S., An, Y. J., Jeong, C. S., Yu, J. H. & Chung, K. M. ATP-binding mode including a carbamoylated lysine and two Mg2+ ions, and substrate-binding mode in Acinetobacter baumannii MurF. Biochemical and biophysical research communications 450, 1045–1050 (2014).
Ashkenazy, H. et al. ConSurf 2016: an improved methodology to estimate and visualize evolutionary conservation in macromolecules. Nucleic Acids Res 44, W344–350 (2016).
Boniface, A., Bouhss, A., Mengin-Lecreulx, D. & Blanot, D. The MurE synthetase from Thermotoga maritima is endowed with an unusual D-lysine adding activity. J Biol Chem 281, 15680–15686 (2006).
Gamsjaeger, R., Liew, C. K., Loughlin, F. E., Crossley, M. & Mackay, J. P. Sticky fingers: zinc-fingers as protein-recognition motifs. Trends Biochem Sci 32, 63–70 (2007).
Berger-Bachi, B., Strassle, A., Gustafson, J. E. & Kayser, F. H. Mapping and characterization of multiple chromosomal factors involved in methicillin resistance in Staphylococcus aureus. Antimicrob Agents Chemother 36, 1367–1373 (1992).
Cui, L., Murakami, H., Kuwahara-Arai, K., Hanaki, H. & Hiramatsu, K. Contribution of a thickened cell wall and its glutamine nonamidated component to the vancomycin resistance expressed by Staphylococcus aureus Mu50. Antimicrob Agents Chemother 44, 2276–2285 (2000).
Kabsch, W. Xds. Acta crystallographica. Section D, Biological crystallography 66, 125–132 (2010).
CCP4. The CCP4 suite: programs for protein crystallography. Acta Crystallographica D50, 760–763 (1994).
Bricogne, G., Vonrhein, C., Flensburg, C., Schiltz, M. & Paciorek, W. Generation, representation and flow of phase information in structure determination: recent developments in and around SHARP 2.0. Acta crystallographica. Section D, Biological crystallography 59, 2023–2030 (2003).
Vonrhein, C., Blanc, E., Roversi, P. & Bricogne, G. Automated structure solution with autoSHARP. Methods Mol Biol 364, 215–230 (2007).
Terwilliger, T. C. et al. Iterative model building, structure refinement and density modification with the PHENIX AutoBuild wizard. Acta crystallographica. Section D, Biological crystallography 64, 61–69 (2008).
Emsley, P. & Cowtan, K. Coot: model building tools for molecular graphics. Acta Crystallographica D60, 2126–2132 (2004).
Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of Macromolecular Structures by the Maximum-Likelihood Method. Acta Crystallographica Section D 53, 240–255 (1997).
Afonine, P. V. et al. Towards automated crystallographic structure refinement with phenix.refine. Acta crystallographica. Section D, Biological crystallography 68, 352–367 (2012).
McCoy, A. J. et al. Phaser crystallographic software. J Appl Crystallogr 40, 658–674 (2007).
Franke, D. et al. ATSAS 2.8: a comprehensive data analysis suite for small-angle scattering from macromolecular solutions. J Appl Crystallogr 50, 1212–1225 (2017).
Panjkovich, A. & Svergun, D. I. Deciphering conformational transitions of proteins by small angle X-ray scattering and normal mode analysis. Phys Chem Chem Phys 18, 5707–5719 (2016).
Svergun, D., Barberato, C. & Koch, M. H. J. CRYSOL - A program to evaluate x-ray solution scattering of biological macromolecules from atomic coordinates. Journal of Applied Crystallography 28, 768–773 (1995).
Katoh, K., Misawa, K., Kuma, K. & Miyata, T. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res 30, 3059–3066 (2002).
Soding, J., Biegert, A. & Lupas, A. N. The HHpred interactive server for protein homology detection and structure prediction. Nucleic Acids Res 33, W244–248 (2005).
Sievers, F. et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol 7, 539 (2011).
Bond, C. S. TopDraw: a sketchpad for protein structure topology cartoons. Bioinformatics 19, 311–312 (2003).
We thank members of the Stehle and Schneider laboratories for critical reading of the manuscript and for discussion. We thank Lukas Schönberger and Astrid Illigmann for technical assistance. We gratefully acknowledge the Swiss Light Source (Villigen, Switzerland) for beamtime and the staff at beamline X06DA and X06SA for assistance during data collection. We also thank members of the Deutsches Elektronen Synchrotron (DESY) in Hamburg, especially Martin Schroer, for assistance with SAXS data collection and interpretation.
The authors declare no competing interests as defined by Nature Research, or other interests that might be perceived to influence the results and/or discussion reported in this paper.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
About this article
Cell wall peptidoglycan in Mycobacterium tuberculosis: An Achilles’ heel for the TB-causing pathogen
FEMS Microbiology Reviews (2019)
Diversity of peptidoglycan structure—Modifications and their physiological role in resistance in antibiotic producers
International Journal of Medical Microbiology (2019)
Unraveling the mechanism of peptidoglycan amidation by the bifunctional enzyme complex GatD/MurT: A comparative structural approach
International Journal of Medical Microbiology (2019)
Antimicrobial Agents and Chemotherapy (2019)