Biochemical and crystallographic studies of l,d-transpeptidase 2 from Mycobacterium tuberculosis with its natural monomer substrate

The essential l,d-transpeptidase of Mycobacterium tuberculosis (LdtMt2) catalyses the formation of 3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\to$$\end{document}→3 cross-links in cell wall peptidoglycan and is a target for development of antituberculosis therapeutics. Efforts to inhibit LdtMt2 have been hampered by lack of knowledge of how it binds its substrate. To address this gap, we optimised the isolation of natural disaccharide tetrapeptide monomers from the Corynebacterium jeikeium bacterial cell wall through overproduction of the peptidoglycan sacculus. The tetrapeptides were used in binding / turnover assays and biophysical studies on LdtMt2. We determined a crystal structure of wild-type LdtMt2 reacted with its natural substrate, the tetrapeptide monomer of the peptidoglycan layer. This structure shows formation of a thioester linking the catalytic cysteine and the donor substrate, reflecting an intermediate in the transpeptidase reaction; it informs on the mode of entrance of the donor substrate into the LdtMt2 active site. The results will be useful in design of LdtMt2 inhibitors, including those based on substrate binding interactions, a strategy successfully employed for other nucleophilic cysteine enzymes.

The essential L,D-transpeptidase of Mycobacterium tuberculosis (Ldt Mt2 ) catalyses the formation of 3!3 cross-links in cell wall peptidoglycan and is a target for development of antituberculosis therapeutics.Efforts to inhibit Ldt Mt2 have been hampered by lack of knowledge of how it binds its substrate.To address this gap, we optimised the isolation of natural disaccharide tetrapeptide monomers from the Corynebacterium jeikeium bacterial cell wall through overproduction of the peptidoglycan sacculus.The tetrapeptides were used in binding / turnover assays and biophysical studies on Ldt Mt2.We determined a crystal structure of wild-type Ldt Mt2 reacted with its natural substrate, the tetrapeptide monomer of the peptidoglycan layer.This structure shows formation of a thioester linking the catalytic cysteine and the donor substrate, reflecting an intermediate in the transpeptidase reaction; it informs on the mode of entrance of the donor substrate into the Ldt Mt2 active site.The results will be useful in design of Ldt Mt2 inhibitors, including those based on substrate binding interactions, a strategy successfully employed for other nucleophilic cysteine enzymes.
Tuberculosis (TB) imposes a major impact on global health and causes an estimated 1.6 million deaths per year 1 .There is a limited set of drugs for TB treatment; however, the prolonged treatment regimens required for efficacy and increasing resistance mean there is a clear need for new therapeutics targeting Mycobacterium tuberculosis (Mtb), the causative agent of TB 2,3 .
Disruption of cell wall biosynthesis is arguably the most effective current approach for treatment of bacterial infections, with the cell wall biosynthesis targeting β-lactams being amongst the most successful of antibacterial drug classes 4 .β-Lactams are efficient and normally safe inhibitors of the penicillin binding proteins (PBPs), a class of nucleophilic serine enzymes that are involved in several of the final steps of peptidoglycan synthesis; PBPs include D,D-transpeptidases (some with bifunctional glycosyltransferase activity), D,D-carboxypeptidases, and endopeptidases (Fig. 1A) 5 .To date, the application of β-lactams for the treatment of TB has been limited, reflecting challenges including β-lactamase mediated resistance 6 .A further issue with β-lactam based TB treatment is that in addition to the nucleophilic serine PBPs, nucleophilic cysteine enzymes play a key role in Mtb cell wall biosynthesis 7 .
Whereas the D,D-transpeptidases catalyse the formation of 4!3 crosslinks between peptidoglycan pentapeptide chains, the functionally related L,D-transpeptidases (Ldts) catalyse the formation of 3!3 cross-links between tetrapeptide chains (Fig. 1) 8,9 .Despite being well conserved between bacterial species 10 , the low abundancy of 3!3 cross-links is proposed to render Ldts redundant in most bacteria, at least in controlled laboratory environments 11 .However, Mtb displays up to 80% of 3!3 cross-links during the stationary phase 7 .Hence, the Ldts have been identified as attractive targets for the development of anti-TB treatment 12 .In particular Ldt Mt2 , one of the five Ldts of Mtb, is recognised as being important for virulence [13][14][15] .
Whereas the D,D-carboxypeptidases and D,D-transpeptidases are nucleophilic serine enzymes (Fig. S1) 16,17 , the Ldts employ a catalytic triad involving the sidechains of a nucleophilic cysteine and a histidine residue, and the backbone carbonyl of a third residue in a characteristic Hxx 14-17 (S/T)HGCZN motif, in which Z represents a hydrophobic residue 18 .The Ldts, which do not manifest structural homology with the PBPs and likely originate from a different evolutionary precursor 18 , share a common YkuD fold (named after the first identified member of this class), distinguished by a β-sandwich with two mixed β-sheets and one helix, and a set of four loops (Fig. S1) [18][19][20] .One of these loops borders the Ldt Mt2 active site 19 , creating two entrances towards the nucleophilic cysteine: the inner cavity and the outer cavity (Fig. S1) 19 .
The established substrate for Ldt Mt2 is the disaccharide tetrapeptide monomer 1 (GlcNAc-MurNAc-L-Ala-D-iGlu NH2 -m-DAP NH2 -D-Ala; Fig. 2A) 7,13 .The formation of 3!3 crosslinks is proposed to occur through covalent reaction of the amide bond between the 3 rd and 4 th residue with the nucleophilic cysteine of Ldt Mt2 , i.e., Cys354, to give an acyl-enzyme complex 21 .The 3 rd residue of a second strand of the tetrapeptide chain then reacts with the thioester intermediate to give the 3!3 crosslinked product (Fig. 1B) 22 .
Though traditionally viewed as PBP inhibitors, selected β-lactams also inhibit the Ldts.In particular, carbapenems have been shown to inhibit Ldts [23][24][25] , and recent work has shown the potential of cephalosporins to inhibit Ldt Mt2

26
. Several other classes of covalently reacting electrophilic compounds, e.g., ebselen, nitriles and acrylamides, have also been reported to inhibit Ldt Mt2 26,27   .Structure-based design of Ldt inhibitors has, however, been limited by a lack of detailed knowledge of enzyme-substrate interactions.
Studies of Ldt Mt2 in complex with its substrate or acyl-enzyme intermediate have been hindered by the restricted availability of 1. Reported studies of the Ldt Mt2 -substrate complex have thus far relied on unnatural substrate analogues, e.g., a crystal structure of the peptidoglycan fragment D- γ-Glu-m-DAP bound to the Ldt Mt2 active site 22 .Efforts to chemically synthesise substrate 3 (lactoyl-L-Ala-D-iGlu NH2 -m-DAP NH2 -D-Ala) have yielded only low amounts of material 28 .The peptidoglycan substrates 1, 2 (GlcNAc-Muramitol-L-Ala-D-iGlu NH2 -m-DAP NH2 -D-Ala), and 3 can, however, be obtained through isolation from the cell well of Corynebacterium jeikeium, which shares an identical peptidoglycan monomer structure with Mtb 7,13,29 .Using a variation of this method, we obtained 1-3 from the cell wall of C. jeikeium in significantly improved yields, enabling detailed studies on their interaction with Ldt Mt2 .Kinetic and binding studies suggest that isolated Ldt Mt2 has a low affinity for the isolated tetrapeptide peptidoglycan monomer.X-ray crystallographic studies of Ldt Mt2 and substrate 3 yielded a structure of the proposed thioester intermediate of the transpeptidase reaction, which suggests the donor substrate enters through the inner cavity, but subsequently interacts with both the inner and outer cavities.

Results
Isolation of the tetrapeptide monomer Initially, we isolated the reduced disaccharide tetrapeptide monomer 2 from the cell wall of C. jeikeium using reported methods 29,30 .C. jeikeium cells were lysed via high-pressure homogenesis, boiled in 8% sodium dodecyl sulphate, then treated with trypsin, pronase, mutanolysin, and lysozyme to yield soluble peptidoglycan monomers (Figure S2).MurNAc moieties in peptidoglycan monomers were then reduced to muramitol using sodium borohydride, to improve resolution during HPLC separation (Fig. S3A) 30 .This method was scaled to a dried cell mass of 94.6 g C. jeikeium from 21.6 L of culture, which yielded ∼4.5 mg of 2 (Fig. S4A).Aiming to improve the yield, we attempted to bypass the reduction of MurNAc in 1. Decreased separation of peaks was observed by HPLC (Fig. S3B); however, highly purified (>95%) disaccharide-tetrapeptide 1 (Fig. S4B) was obtained (∼3.6 mg from 61.3 g dried cell mass).
Next, we assessed the structural requirements of the tetrapeptide residues that enable reaction with Ldt Mt2 .A previous study identified the importance of sidechain amidation of both the D-iGlu and m-DAP residues for turnover 28 .To further investigate the structural requirements for Ldt Mt2 substrates, we incubated Ldt Mt2 with tetrapeptide 10 (L-Ala-D-iGlu-L-Lys-D-Ala), a peptidoglycan monomer that is often observed in Gram-positive bacteria 11 .We observed no evidence for turnover of 10.The tetrapeptide 11 (L-Ala-D-iGlu NH2 -L-Lys-D-Ala), which contains an amidated D-iGlu residue, but wherein m-DAP NH2 is replaced with L-Lys, was also not a substrate (over 24 h), evidencing the importance of the m-DAP NH2 residue in peptidoglycan fragments for turnover with isolated Ldt Mt2 .
We were interested in assessing the transpeptidase activity of the active site cysteine to serine mutant of Ldt Mt2 (Ldt Mt2

C354S
), in part because previous reports have shown that in contrast to wild-type Ldt Mt2 13,21 , Ldt Mt2 C354S is unable to form a covalent complex with several penem and carbapenem derivatives 23,31 .When Ldt Mt2 C354S was incubated with 1, 2, 3 or 4, no product formation was observed, suggesting the active site cysteine is essential for transpeptidase activity, and cannot be replaced by a potentially nucleophilic serine residue.
Differential scanning fluorometric thermal shift assays 32 (Fig. S10 and Table S1) with a 100-fold excess (500 µM) of peptidoglycan fragments showed pentapeptide 4 induced a modest stabilising effect, as evidenced by a shift in the melting temperature (T m ) of ≥2 °C.With a 200-fold excess (1 mM), stabilisation was observed with tetrapeptides 1, 3 and 9, and pentapeptides 4 and 8, all manifesting a T m shift of 2-3 °C; no significant shift in T m was observed with 10 and 11.
Protein-observed solid-phase extraction mass spectrometric (SPE-MS) studies were performed with Ldt Mt2 (1 µM) in the presence of 1 and 3 (100 µM).While transpeptidase activity was confirmed with 1 and 3, as evidenced by observation of protein unbound species with masses of 1786 Da, and 974 Da, (correlating to 5 and 7, respectively), no covalent adducts of Ldt Mt2 were observed, suggesting that in solution the lifetime of the covalent thioester (predicted mass increase of 849 Da in the case of 1 and 443 Da in the case of 3, relative to unmodified Ldt Mt2 ) intermediate is short-lived (Fig. S11).

Kinetics of the Ldt Mt2 transpeptidase reaction with peptidoglycan monomers
Thus far, Ldt Mt2 transpeptidase activity has been measured by LC-MS analysis 13 , and Ldt Mt2 inhibition assays have relied on use of unnatural chromogenic or fluorogenic probe substrates 23,33,34 , as employed in an assay suitable for high-throughput screening 26 .These assays are, however, likely poor mimics of the natural reaction.We therefore developed an efficient, quantitative, and biologically relevant method of determining Ldt Mt2 activity and inhibition based on turnover of its natural substrate.A microtiter plate-based assay that determines PBP activity has been reported, based on detection of D-Ala, which is released on D,D-transpeptidase and D,D-carboxypeptidase activity (Fig. 3A) [35][36][37] .D-Ala is converted to pyruvate, ammonia and hydrogen peroxide by D-amino acid oxidase (DAAO).Hydrogen peroxide is then used by horseradish peroxidase (HRP) to oxidise Amplex Red (AR) forming fluorescent resorufin 35 .We considered that the transpeptidase assay could be extended to monitor Ldt Mt2 catalysis.
Optimisation initially involved variation of DAAO and HRP concentrations, aiming to minimise delay in conversion of D-Ala.When the assay mixture contained an excess of DAAO (2 U/mL) and HRP (2 U/mL), conversion of D-Ala was apparently near instantaneous (Fig. S12A, B).The assay performed better in sodium phosphate compared to Tris or HEPES buffers (Fig. S12C), and in the presence of 100 mM NaCl (Fig. S12D).A higher pH manifested increased Ldt Mt2 activity (tested pH range 7.0-8.5;Fig. S12E) and activity was higher at 37 °C than room temperature (Fig. S12F).Assays were therefore performed at 37 °C in 50 mM sodium phosphate pH 8.0, 100 mM NaCl and 0.01% (v/v) Triton X-100, with an Ldt Mt2 concentration of 500 nM and substrate (3) concentration of 35 µM (Fig. S12G).Under these conditions, a Z′ value of 0.88 and a signal to background value of 5.0 was obtained.The hydrogen peroxide produced has potential to oxidise the Ldt Mt2 catalytic cysteine, however, dose-response studies 34 manifested no evidence for interference with assay relevant concentrations (Fig. S12H).
The assay was applied to investigate the inhibition of reported Ldt Mt2 inhibitors (Fig. S13 and Table S2) 26,27,33 .The results showed good correlation with the reported Ldt Mt2 ligand binding assay utilising a fluorogenic thiol reactive probe (Fig. 3B) 34 .The DAAO/HRP method presents an improved assay for Ldt Mt2 inhibitor screening, as the assay more closely resembles the biological process.The possibility for interference through inhibition of DAAO and HRP necessitates use of appropriate controls; indeed, the nonspecific inhibitors ebselen 24 and ebsulfur analogue 23 were found to cause interference (Fig. S14 and Table S2).Note that the assay could be adapted to use L-lactate dehydrogenase to monitor reduction of NAD + in the presence of L-lactate.
With fixed concentrations of 1 and 3 (35 µM), a substantial difference in reaction rates was observed (120.0 ± 1.7 RFU min −1 and 224.2 ± 2.6 RFU min −1 for 1 and 3, respectively; Fig. S15), suggesting that interaction of the GlcNAc-MurNAc moiety of the substrate, at least with isolated components, has the potential to hinder the reaction rate.Increasing concentrations of 1 did not saturate the reaction (up to 20 mM; Fig. 3C and Table S3), hence the K m of 1 could not be determined.With 3, a K m of 6.2 mM was obtained, indicating low affinity; k cat was 2.6 × 10 2 min −1 , providing a k cat /K m of 4.2 min −1 M −1 (Fig. 3D and Table S3).Note, that the effective incorporation of fluorescent amino acids and peptidoglycan fragments into the Mtb cell wall suggests more efficient turnover occurs in a cellular context 38 .The difference between isolated activity and cellular activity may reflect a need for complex formation, missing cofactors, or for binding of full-length peptidoglycan for optimal Ldt Mt2 activity.
The non-substrate peptidoglycan fragments 4, 8, 10, 12 and 13 showed modest Ldt Mt2 inhibition (Fig. S16).Pentapeptides 8 and 10 and tripeptide 12 were the most potent (pIC 50 4.8), while dipeptide 13 manifested the lowest potency (pIC 50 4.0).The ability of structurally related non-substrate peptides to displace the substrate from the active site shows potential for the development of substrate-like inhibitors, although optimisation to improve affinity will be required.

X-ray crystallographic studies of the Ldt Mt2 -substrate complex
To investigate the structural interactions of Ldt Mt2 with its substrate, we carried out X-ray crystallographic studies with the peptidoglycan fragments, aiming to obtain a structure of an enzyme-substrate complex; we obtained structures of Ldt Mt2 in complex with 1 and 3.
While in the absence of 3, Ldt Mt2 crystallised in the reported P12 1 1 space group 26,27 , in the presence of 3 Ldt Mt2 crystallised in the P2 1 2 1 2 space group with a single protein chain in the asymmetric unit (1.98 Å, Fig. 4; PDB 8PXZ), using precipitation solutions with a pH between 7.0-7.5,yielding structures of Ldt Mt2 with a molecule of 3 bound to the Ldt Mt2 immunoglobulin-like domain 2 (IgD2).Notably, however, when Ldt Mt2 and 3 were co-crystallised using a precipitation solution at pH 7.0, we observed additional electron density extending from the sidechain of the nucleophilic residue Cys354.The results are in agreement with formation of the dipeptide (D-iGlu NH2 -m-DAP NH2 ), covalently linked to Ldt Mt2 via a thioester with Cys354, apparently reflecting a key intermediate of the transpeptidase reaction with the donor peptide (Fig. 4A).The thioester is observed in the (Z)-stereochemistry, consistent with prior studies on nucleophilic serineand cysteine-enzymes, including the SARS-CoV-2 main protease (M pro ; Fig. S17A) 39,40 and porcine pancreatic elastase 41 .
In the Ldt Mt2 -3 complex structure, the m-DAP NH2 residue sidechain protrudes into the outer cavity, while the D-iGlu NH2 sidechain is located in the inner cavity, suggesting that the donor substrate enters from the inner cavity (Fig. 4B, C).Note that no electron density was observed corresponding to the L-Ala residue of 3, which has previously been suggested not to play a significant role in substrate binding 42 .Additional electron density at the entrance of the outer cavity was observed, in which D-Ala, released during the formation of the thioester intermediate, could be modelled (Fig. 4A).We did not observe density that could reflect binding of an acceptor substrate.
The overall fold of the Ldt Mt2 -3 complex strongly resembles that observed for unmodified Ldt Mt2 (PDB 6RLG 27 ; main chain RMSD 0.65 Å).Similarly to unmodified Ldt Mt2   27,43   , the S γ atom of Cys354 is positioned 3.9 Å from the N ε2 atom of His336, and the N δ1 atom of His336 is positioned for polar interactions with the main-chain carbonyl O of Ser337 at a distance of 2.7 Å. Notably, no conformational changes relative to unmodified Ldt Mt2 are observed in loop II (Fig. S18), which encloses the active site (residues 300-323) and which has previously been observed to undergo conformational changes upon binding of various Ldt Mt2 inhibitors, and which has been proposed to undergo extensive conformational changes upon substrate binding 26,33,43 .
The thioester acyl group, as well as the m-DAP NH2 residue and the D-iGlu NH2 residue of 3 are positioned to engage in interactions with elements in the Ldt Mt2 -3 complex active site (Fig. 4, S19).The thioester acyl group is located in the proposed oxyanion hole (the backbone NH groups of His352, Gly353 and Cys354) at respective distances of 3.8 Å, 3. Crystallographic studies of Ldt Mt2 in the presence of 1 did not manifest evidence for the presence of substrate at the active site; similarly to 3, however, 1 was observed to bind to the surface of the IgD2 domain through various polar interactions, as observed with both wild-type Ldt Mt2 and with Ldt Mt2 C354S (PDB 8PXY, 2.15 Å; Fig. S20).The GlcNAc moiety of 1, which is absent in 3, was observed to interact with the surface the catalytic domain of a second symmetry related molecule of Ldt Mt2 , manifesting polar interactions with Asp367 (2.5 Å and 2.8 Å) and with Gln363 (3.0 Å), located ~20 Å away from the nucleophilic cysteine (or serine, in the case of Ldt Mt2

C354S
), potentially preventing binding of the second molecule of 1 at the Ldt Mt2 active side as observed in the structure of Ldt Mt2 co-crystallised with 3.

Discussion
Targeting bacterial cell wall transpeptidases is a validated method for treating bacterial infections, in particular via β-lactam mediated PBP inhibition 5 .Though a role for β-lactams in the widespread treatment of TB has not been established, promising activity against Mtb has been observed, particularly with carbapenems, which inhibit both PBPs and Ldts 44 .Ldts, and specifically Ldt Mt2 , are essential for Mtb 13 , but have been considerably less intensively studied than PBPs.Whilst several early-stage studies on inhibitors for Ldt Mt2 are reported 23,26 , the search for potent and clinically useful inhibitors is ongoing.Despite its biological and medicinal importance, the mechanism of the transpeptidase reaction of Ldt Mt2 has been the subject of limited studies, in part likely due to the lack of availability of the peptidoglycan monomers.Reported studies on Ldt Mt2 have thus mainly relied on the use of relatively simple substrate analogues 22,45 .Our modified procedure for the isolation of peptidoglycan monomers from the C. jeikeium cell wall enabled us to obtain a crystal structure of Ldt Mt2 reacted with 3 and to obtain insights into the binding and reaction kinetics of Ldt Mt2 .
Exposure of C. jeikeium cells to subinhibitory concentrations of ampicillin during growth, results in accumulation of peptidoglycan pentapeptide monomer 4. Ldt Mt2 substrate 1 can subsequently be obtained from 4 via enzymatic cleavage of the terminal D-Ala.Via this method, the tetrapeptide monomer was obtained in a yield of ∼21 mg from 56 g of dried cell mass.Presently, this approach is substantially more efficient than reported synthetic procedures 28 .
X-ray crystallographic studies of Ldt Mt2 with substrate 3 yielded a structure of the covalently linked thioester acyl-enzyme complex, likely substantially reflecting the transient intermediate during the Ldt Mt2 transpeptidation reaction.Many enzymes employing nucleophilic cysteine-or serine-residues proceed through covalent acyl-enzyme intermediates; commonly these intermediates are unstable, and hence structures of them in complex with substrates, including at the acyl-enzyme complex stage, can be challenging to obtain under catalytically relevant conditions 46 .Crystallographic studies of the (thio)ester intermediates frequently rely on active site mutations (e.g., ref. 47), the use of substrate mimics (e.g., ref. 48), the use of rate-reducing conditions, for example, non-optimal pH (e.g., refs.41,49), or use of an excess of a reversibly reacting product 41 .There is a particular lack of reported thioester, compared to ester, intermediate structures, perhaps reflecting the increased reactivity of thioesters with nucleophiles.To our knowledge, the only other nucleophilic cysteine enzyme where structural analysis of a thioester derived from a natural substrate has been achieved is SARS-CoV-2 M pro39,40 , where a thioester was observed to form in crystallo after being presented with and excess of an acid product, resulting in reverse reaction.
The observation of a thioester in our Ldt Mt2 -3 complex acyl-enzyme crystal structure could in part be due to the inability of the acceptor substrate to enter the active site in the crystalline state due to reduced flexibility that may be required for acceptor substrate binding.In addition to the use of an excess of substrate (and hence product) and lattice interactions, other factors that may affect the persistence of the thioester in crystallo are the reduced pH (7.0, compared to 7.5) and temperature (4 °C, compared to rt) with respect to the solution-based studies, both of which reduce the rate of the Ldt Mt2 reaction (Fig. S12D-E).It is also important to note that transpeptidation reactions are fundamentally different from hydrolysis reactions, in that during transpeptidation the acyl-enzyme complex is required to be sufficiently stable in order to allow for the m-DAP residue of the acceptor substrate to react with the (thio)ester of the donating substrate.
The relative positions of the catalytic residues and of the oxyanion hole residues in the Ldt Mt2 -3 complex are similar to those observed in the M pro thioester intermediate structure, and in structures of Ldts in complex βlactams, e.g., in a structure of the Enterococcus faecium Ldt fm complexed with ertapenem 50 , and in structures of Ldt Mt2 complexed with meropenem or biapenem (Fig. S17A-D) 21,51 .In all cases the thioester link has the (Z)stereochemistry that is commonly associated with acyl-enzyme intermediates 46,52 , including in a structure of PBP4a of B. subtilis in complex with a peptidoglycan mimetic (Fig. S17E) 53 .
The structure reported here informs on how Ldt Mt2 interacts with a substantial part of its donor substrate.In a structure of a non-substrate peptidoglycan fragment (13, D-γ-Glu-m-DAP) bound to the Ldt Mt2 active site 22 , the m-DAP residue faces into the inner cavity, contrasting with our structure, suggesting entrance from the outer cavity (Figs.S21A, B and  S22A).Molecular dynamics studies have suggested that entrance of the substrate from the inner cavity of Ldt Mt2 will not lead to acylation 42 .Our results suggest that during the initial steps of productive binding via the inner cavity, the substrate may project deeper into the active site than predicted by the reported calculations.In agreement with this proposal, a solid-state NMR structure of the B. subtilis L,D-transpeptidase (Ldt Bs ) in complex with intact peptidoglycan suggests binding of the donor substrate in the analogous inner cavity (Figs.S21C, S22B) 54 .
An allosteric binding pocket of Ldt Mt2 , named the S-pocket, is proposed to bind the sugar-moieties (Fig. S21E) 45 .We did not observe interaction of GlcNAc-MurNAc of 1 in this pocket.Nevertheless, extrapolation of the studies on Ldt Bs bound to intact peptidoglycan suggests that intact peptidoglycan likely binds in the S-pocket associated with the IgD2 domain of Ldt Mt2 54 .
A coupled assay for detecting D-Ala was optimised for assessing Ldt Mt2 activity.The assay represents an increased throughput method of assessing turnover in comparison with previous LC-MS studies 13,28 , and provides increased biologically relevance over the reported highthroughput assay for inhibition of Ldt Mt2 26 .However, due to the relatively low enzymatic activity, high concentrations of substrate are required, making the use of this assay in a high-throughput setting challenging without a well-established synthesis route.Additionally, the coupled enzymes DAAO and HRP increase the risk of assay interference.Based on these observations, we recommend that the fluorescence-based transpeptidase assay is perhaps most suited for use as a secondary assay following more high-throughput assays for Ldt Mt2 , testing potential inhibitors in a more biologically relevant manner.
Although our studies suggest that the isolated recombinant Ldt Mt2 has a relatively low affinity and catalytic activity with 1 and 3, under our conditions 3 is a somewhat more efficient substrate.Interestingly, evidence for efficient transpeptidase activity by Ldts has been observed in Mtb cells in a previous report 38 , suggesting the need for increased interactions than present in our assays; potentially these will be achieved with use of full-length peptidoglycan.Alternatively, complex formation with related proteins / enzymes may increase activity, as observed with the functionally related D,D- transpeptidases and glycosyltransferases PBP1A and PBP2B, which require interactions with lipoprotein cofactors LpoA and LpoB, both in isolated enzyme and cellular systems [55][56][57][58] .In E. coli, the Ldt YcbB was found to associate with PBP1b and PBP5 revealing links between the two types of peptidoglycan transpeptidases 59 .
Amidation of both D-iGlu and m-DAP residues is reported to be required for Ldt Mt2 catalysis, as well as for growth of Mtb 28 .We observed that 11, a close analogue of 3, containing D-iGlu NH2 but with the m-DAP NH2 substituted with L-Lys, is not a substrate for Ldt Mt2 .The Ldt Mt2 -3 complex structure suggests that the multiple polar interactions m-DAP NH2 makes with Asn356, His336, Thr320, and His352 are important, as L-Lys is likely only able to (directly) interact with Asn356 and His336.By contrast with the high substrate selectivity observed for Ldt Mt2 , PBPs have been observed to react with D-Ala-D-Ala peptidoglycan mimetics, suggesting that modified features of the second and third peptide chain residues may not be as important for PBP catalysis as they appear to be for Ldt Mt2 60,61 .Most reported Ldt Mt2 inhibitors interact with the catalytic cysteine (Cys354), and may fill either the inner or the outer active site cavities, e.g., in the Ldt Mt2 :ertapenem complex the inhibitor occupies the inner cavity, while in the Ldt Mt2 :biapenem complex the inhibitor occupies the outer cavity (Fig. S17C-D) 21,43,45,51 .Based on the structure of Ldt Mt2 reacted with 3, elements of which occupy both the inner-and outer-cavities, improvement of inhibitor potency might be achieved through interactions with both the inner-and outer-cavities.Design of inhibitors achieving this may be based on our tetrapeptide structure, but optimisation will be necessary, especially in light of the low affinity of Ldt Mt2 for 3 in solution.The structure of Ldt Mt2 reacted with 3 suggests that a focus for future optimisation studies may be on the D-iGlu NH2 -m-DAP NH2 -D-Ala residues, in particular the D-iGlu NH2 -m-DAP NH2 elements which are observed to interact with Ldt Mt2 in the intermediate structure.The addition of an electrophilic group to the m-DAP NH2 residue in lieu of the peptide bond with D-Ala may increase reactivity, e.g., a nitrile group, which has been shown to react with Cys354 of Ldt Mt2

26
. Optimisation of the substrate for inhibition, including the addition of a reversibly covalently reacting nitrile group has been a successful method in the inhibitor development of M pro , with one resulting inhibitor, nirmatrelvir, being approved for COVID-19 treatment 62,63 .Polar interactions with Tyr308 may be achieved through incorporation of a heteroatom in the D-iGlu NH2 carbon chain.While the carbon chain of the m-DAP NH2 does not undergo direct interactions with the active site of Ldt Mt2 , the flexibility of this residue may be required to allow for entrance into the active site and the polar interactions observed with His336, Tyr320, and Asn356, as well as the water-mediated interactions with His352.

Conclusion
Ldt Mt2 is a promising drug target for TB treatment, though little has been reported on the mechanism of the transpeptidase reaction that it catalyses.We optimised the isolation of the natural disaccharide tetrapeptide monomers 1-4 from the C. jeikeium bacterial cell wall through overproduction of peptidoglycan sacculus.We explored the interactions of Ldt Mt2 with the isolated tetrapeptide monomers 1 and 3 via X-ray crystallographic studies, obtaining a structure of the proposed thioester intermediate in Ldt Mt2 catalysis.The structure informs on the mode of entrance of the donor substrate into the Ldt Mt2 active site and active site interactions made by it.The combined results will be useful in the design of novel Ldt Mt2 inhibitors, including those based on substrate binding interactions, a strategy that has been successfully employed for other nucleophilic cysteine enzymes, including M pro62,63 .

Materials
Faropenem was from Fluorochem Ltd., FAD was from Alfa Aesar, AR was from Cambridge Bioscience Ltd., Fmoc-D-iGln-OH was from Ambeed, 12 and 13 were from InvivoGen.16-23 were obtained from the GSK HTS compound library.All other compounds were from Merck.

Protein production and purification
Recombinant Ldt Mt2 was produced in Escherichia coli BL21(DE3) cells transformed with pNIC28-Bsa4-Ldt Mt2 Δ1-55, in the presence of 50 μg/mL kanamycin 31 .Expression was induced with isopropyl β-D-thiogalactopyranoside (IPTG, 0.5 mM) at OD 600 of 0.6, with subsequent incubation overnight at 18 °C.Cells were lysed by sonication (SONIC Vibra-Cell, 60% amplitude) in the presence of DNase I; lysates were loaded onto a 5 mL HisTrap column (GE Life Sciences), pre-equilibrated in HisTrap buffer A (50 mM Tris, pH 8, 500 mM NaCl, 20 mM imidazole).The protein was eluted with HisTrap buffer B (50 mM this, pH 8, 500 mM NaCl, 500 mM imidazole) using a gradient from 0% to 100% buffer B. Protein-containing fractions were combined and the buffer was exchanged for HisTrap buffer A. The HisTag was cleaved using the TEV protease at 4 °C, over 12 h.The HisTag cleaved Ldt Mt2 was passed through a 5 mL HisTrap Column, and loaded onto a 300 mL Superdex 75 column (GE Life Sciences) preequilibrated in buffer C (50 mM Tris, pH 8, 500 mM NaCl).Ldt Mt2 was eluted with buffer C. The purity and identity of Ldt Mt2 was confirmed by SDS-PAGE (>95% purity) and MS (calculated 37975 Da, observed 37978 Da) analyses.Site-directed mutagenesis was carried out using the partial overlapping primer design method 64 , with primer sequences shown in Table 1 being used to achieve the C354S mutation of Ldt Mt2 .PCR was performed using Q5 High-Fidelity DNA Polymerase (New England Biolabs) following standard protocols.Ldt Mt2 C354S was produced and purified as described for the wild-type enzyme 31 .Purity and identity of Ldt Mt2 C354S was confirmed by SDS-PAGE (>95% purity) and MS (calculated 37959 Da, observed 37962 Da).Recombinant E. coli DacA was produced and purified as reported 65 .
Peptidoglycan fragment isolation C. jeikeium (National Collection of Type Cultures: NCTC 11913) was grown overnight at 37 °C on Columbia agar supplemented with 5% sheep blood (Scientific Laboratory Supplies Ltd).Cultures were used to inoculate brain heart infusion broth supplemented with 1% (v/v) Tween 80 and 2 μg/mL ampicillin 29 .The culture was incubated at 37 °C for 16 hours with shaking.
The disaccharide peptides were separated via reverse phase highperformance liquid chromatography (HPLC) on a C 18 column (10 µm, 21,2 × 250 mm, Avantor ACE AQ) at a flowrate of 15 mL/min.A gradient of 1-20% (v/v) of solvent B was applied between 0 and 20 min (solvent A: 0.05% (v/v) trifluoroacetic acid in water; solvent B: 0.035% (v/v) trifluoroacetic acid in acetonitrile).Peaks containing 4 were lyophilised and dissolved in 50 mM tris pH 7.5 and DacA (20 µM) was added.The mixture was incubated for 5 h at 37 °C, lyophilised and purified by HPLC on a C 18 column (10 µm, 21,2 × 250 mm, Avantor ACE AQ) at a flowrate of 15 mL/min.A gradient of 1-5% (v/v) of solvent B was applied between 0 and 7.5 min, followed by a gradient of 5-7% (v/v) of solvent B over 40 min to obtain 1 as a white solid.
To obtain 2, fragment 1 was dissolved in 0.5 M borate buffer pH 8.0 containing 10 mg/mL sodium borohydride.After 20 min the pH was adjusted to 2-4 with orthophosphoric acid.Purification by HPLC was performed as described above.
To obtain 3, fragment 1 was dissolved in 25% (v/v) NH 4 OH in H 2 O and stirred at 37 °C for 5 hours.The mixture was then neutralised via addition of acetic acid and the mixture was lyophilised.Purification by HPLC was performed as described above.

C354S -reverse CTAACATTCAGACTACCATGGCTGGTATTGGTATG
In bold is the nucleic acid that is changed from the wild type.
Compounds (in DMSO, with final concentrations ranging from 40 µM to 20.3 nM; 10 dilutions of factor 3) were added using a CyBio liquid handling system (Analytik Jena AG), with 4 replicates per inhibitor concentration.In the case of Control 1 (expected 0% inhibition; 100% fluorescence signal) and Control 2 (expected 100% inhibition; 0% fluorescence signal) the compound solution was substituted with neat DMSO.Ldt Mt2 (5 µL, 500 nM final concentration) was added to all wells using a MultiDrop Combi dispenser (Thermo Fisher Scientific).In the case of Control 2, Ldt Mt2 was substituted with the assay buffer (5 µL).This mixture was incubated for 30 minutes at rt.The substrate (1 or 3, as specified, 5 µL, 35 µM final concentration) was then added to all wells.The fluorescence signal was measured using a PHERAstar plate reader (BMG Labtech) with λ ex = 540 nm and λ em = 590 nm.Data were analysed using Prism 10 (GraphPad).
The transpeptidase interference assay was performed as described above, with the modification that the Ldt Mt2 mixture was replaced by 5 µL assay buffer, and the substrate was replaced by D-Ala (500 nM final concentration).Data were analysed using Prism 10 (GraphPad).

Crystallography
Ldt Mt2 wild type or Ldt Mt2 C354S (Δ1-55; with the N-terminal His Tag removed, in 50 mM tris, pH 8.0, 100 mM NaCl) was incubated with substrate 1 or 3 for 30 minutes on ice, and then crystallised using a well solution consisting of 0.1 M HEPES, pH 7.0, and 25% (v/v) Jeffamine ED-2001 pH 7.0 (identified from the JCSG Plus TM crystallisation screen, Hampton Research), using sitting drop vapour diffusion crystallisation plates (low reservoir Intelli-Plate 93-3), with 1 µL of protein-substrate solution and 1 µL of well solution.Note that crystallisation conditions previously reported by us 26,27 did not yield crystals with additional electron density in the active site upon either soaking existing crystals with substrates 1-3 or cocrystallisation with 1-3, neither did we obtain complexes with Ldt Mt2

C354S
. Crystallisation plates were stored at 4 °C, and crystals grew over 24 h.The crystals were mounted on nylon loops, cryocooled and stored in liquid nitrogen.Datasets were collected using the MX beamline i03 at Diamond Light Source at 100 K, at a wavelength of 0.97625 Å. Datasets were processed using the automated processing pipeline at Diamond Light Source, using xia2 66 .
Structures were solved by molecular replacement with Phaser 67 , using PDB entry 6RLG 27 as the search model.Ligand restrain files were created using Grade2 68 .Alternating cycles of refinement using PHENIX 69 and manual model building using COOT 70 were performed until no further significant improvement of R work and R free was achieved.Data collection and refinement statistics are described in Table 3.There were no Ramachandran outliers for the Ldt Mt2 C354S -1 (PDB 8PXY) and Ldt Mt2 -3 (PDB 8PXZ) structures, with 98% and 97% Ramachandran favoured, respectively.

Statistics and reproducibility
All experiments were performed 2-4 times, as indicated.Individual data points are plotted, and average values and standard deviation are presented in corresponding tables.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Fig. 1 |
Fig. 1 | Roles of glycosyltransferases, transpeptidases and carboxypeptidases in the final stages of bacterial peptidoglycan synthesis.A Penicillin binding proteins (PBPs) with glycosyltransferase activity link lipid II to peptidoglycan, PBPs with carboxypeptidase activity cleave peptide strands from pentapeptides to give tetrapeptides and PBPs with transpeptidase activity form 4!3 cross-links between pentapeptide strands.The L,D-transpeptidases (Ldts) form 3!3 cross-links between tetrapeptide strands.Endopeptidases reverse the transpeptidase activity, cleaving the peptide bonds between cross-links.Lcps cleave peptide strands from tetrapeptides to tripeptides.The figure was created using BioRender.com.GlcNAc is in green, MurNAc is in teal, amino acids are in light blue, cross-linked amino acids are in dark blue.B Mechanism for the cross-linking reaction between the disaccharide tetrapeptide monomers catalysed by Ldt Mt2 (adapted from de Munnik et al. 26 ).

Fig. 2 |
Fig. 2 | Structures of peptidoglycan fragments of Mycobacterium tuberculosis.A Structures of the tetrapeptide substrates of Ldt Mt2 1-3, obtained through isolation from the cell wall of C. jeikeium.B Structure of the pentapeptide monomer of peptidoglycan from Mtb and C. jeikeium 4, obtained through isolation from the cell wall of C. jeikeium.C The structures of the products of transpeptidase activity of Ldt Mt2 5-7, obtained through cross-linking of 1, 2 and 3, respectively.D Peptidoglycan fragments 8-13 that do not act as substrates for Ldt Mt2 , were obtained through solid-phase peptide synthesis (9 and 11) or commercial sources (8, 10, 12 and 13).GlcNAc stands for N-acetylglucosamine, MurNAc stands for Nacetylmuramic acid.
3 Å and 2.3 Å, suggesting that, at least in the crystal structure obtained, no strong interactions are made with the backbone NH groups of His352.Polar interactions of m-DAP NH2 with His336 (3.0 Å), Asn356 (3.4 Å) and Thr320 (2.8 Å), as well as water-mediated polar interactions with His352 (3.3 Å to water, 3.4 Å to His352) are observed.The m-DAP NH2 carbon chain is positioned for hydrophobic interactions with the sidechain of Tyr318.The D-iGlu NH2 residue is observed to engage in water-mediated polar interactions with His352 (2.7 Å to water and 3.3 Å to His352) and Ser331 (2.9 Å to water and 2.8 Å to Ser331), as well as hydrophobic interactions with Gly332.

Fig. 3 |
Fig. 3 | A fluorescence-based assay for Ldt Mt2 transpeptidase activity.A Principle of the Ldt Mt2 transpeptidase assay, employing D-amino acid oxidase (DAAO) and horseradish peroxidase (HRP) to report on the release of D-Ala.B Correlation between pIC 50 values obtained with the transpeptidase activity assay and fluorogenic thiol probe assay 26 .Individual data points are shown.C Michaelis-Menten graph of Ldt Mt2 in reaction with substrate 1.D Michaelis-Menten graph of Ldt Mt2 in reaction with substrate 3.

Fig. 4 |
Fig. 4 | Crystallographic studies of Ldt Mt2 with peptidoglycan fragment 3 show covalent thioester formation.A A structure of Ldt Mt2 reacted with 3 (PDB 8PXZ, 1.98 Å resolution) shows the covalent thioester intermediate structure, and a second molecule of 3 bound to the IgD2 of Ldt Mt2 .The mF 0 − DF c polder OMIT map is contoured at 3.0 σ and is carved around the 3-derived thioester; 3 is in grey mesh.Polar interactions are shown in grey dashes.B View of the inner cavity, which interacts with the D-iGlu NH2 residue.C View of the outer cavity, which interacts with the m-DAP NH2 residue.

Table 1 |
Primers used in the site-directed mutagenesis to achieve the C354S mutation of Ldt Mt2

Table 2 |
Master mixture for the DAAO assay

Table 3 |
Data collection and refinement statistics for the crystal structure of Ldt Mt2 C354S in complex with 1 and Ldt Mt2 in complex with 3 aA single crystal was used for each structure.Values in parentheses are for the highestresolution shell.