Role of UDP‐N‐acetylmuramic acid in the regulation of MurA activity revealed by molecular dynamics simulations

Abstract

The peptidoglycan biosynthesis pathway plays a vital role in bacterial cells, and facilitates peptidoglycan layer formation, a fundamental structural component of the bacterial cell wall. The enzymes in this pathway are candidates for antibiotic development, as most do not have mammalian homologues. The UDP‐N‐acetylglucosamine (UNAG) enolpyruvyl transferase enzyme (MurA) in the peptidoglycan pathway cytoplasmic step is responsible for the phosphoenolpyruvate (PEP)–UNAG catalytic reaction, forming UNAG enolpyruvate and inorganic phosphate. Reportedly, UDP‐N‐acetylmuramic acid (UNAM) binds tightly to MurA forming a dormant UNAM–PEP–MurA complex and acting as a MurA feedback inhibitor. MurA inhibitors are complex, owing to competitive binding interactions with PEP, UNAM, and UNAG at the MurA active site. We used computational methods to explore UNAM and UNAG binding. UNAM showed stronger hydrogen‐bond interactions with the Arg120 and Arg91 residues, which help to stabilize the closed conformation of MurA, than UNAG. Binding free energy calculations using end‐point computational methods showed that UNAM has a higher binding affinity than UNAG, when PEP is attached to Cys115. The unbinding process, simulated using τ–random acceleration molecular dynamics, showed that UNAM has a longer relative residence time than UNAG, which is related to several complex dissociation pathways, each with multiple intermediate metastable states. This prevents the loop from opening and exposing the Arg120 residue to accommodate UNAG and potential new ligands. Moreover, we demonstrate the importance of Cys115‐linked PEP in closed‐state loop stabilization. We provide a basis for evaluating novel UNAM analogues as potential MurA inhibitors.

Public significance

MurA is a critical enzyme involved in bacterial cell wall biosynthesis and is involved in antibiotic resistance development. UNAM can remain in the target protein's active site for an extended time compared to its natural substrate, UNAG. The prolonged interaction of this highly stable complex known as the ‘dormant complex’ comprises UNAM–PEP–MurA and offers insights into antibiotic development, providing potential options against drug‐resistant bacteria and advancing our understanding of microbial biology.

1. INTRODUCTION

Peptidoglycan is a cross‐linked polymer that surrounds the cell walls of Gram‐positive and Gram‐negative bacteria. It is responsible for stabilizing the cell wall, thus providing the structural integrity for cell osmotic equilibrium (Auer & Weibel, 2017; Egan et al., ²⁰²⁰). Peptidoglycan biosynthesis occurs in three different cellular compartments: the cytoplasm, cell membrane, and periplasmic space (Egan et al., 2020). The MurA–F enzymes belong to the main family of proteins involved in the enzymatic catalysis of peptidoglycans in the cytoplasm (Egan et al., 2020; Barreteau et al., ²⁰⁰⁸; Kumar et al., ²⁰²⁰; Laddomada et al., ²⁰¹⁶).

UDP‐N‐acetylglucosamine (UNAG) enolpyruvyl transferase (MurA) and UDP‐N‐acetylene pyruvylglucosamine reductase (MurB) are involved in the limiting steps of peptidoglycan biosynthesis, making them interesting targets for the development of new drugs (Kumar et al., 2020; Kumari & Subbarao, ²⁰²¹; Sanad et al., ²⁰²⁰). The MurA catalytic reaction involves transfer of the enolpyruvate group from phosphoenolpyruvate (PEP) to UNAG to form UNAG enolpyruvate (UNAG‐EP) and inorganic phosphate (Mihalovits et al., 2019; Sonkar et al., ²⁰¹⁷). UNAG‐EP is used as a substrate in the reaction catalyzed by MurB, in which UNAG enolpyruvate is reduced to form UDP‐N‐acetylmuramic acid (UNAM) (Kumar et al., 2020; Sonkar et al., ²⁰¹⁷; Zhu et al., ²⁰¹²; Eniyan et al., ²⁰¹⁸). These reactions are illustrated in Figure 1.

FIGURE 1.

Scheme of reactions catalyzed by MurA and MurB in the cytoplasm.

Most studies consider the PEP structure to be a key scaffold for designing new inhibitors against various bacteria. This substrate is also used by other enzymes involved in bacterial survival (Egan et al., 2020; De Oliveira et al., ^{2020, 2022}; Mir et al., ²⁰¹⁵), such as 3‐deoxy‐D‐arabino‐heptulosonate‐7‐phosphate‐synthase (De Oliveira et al., 2020; Burschowsky et al., ²⁰¹⁸), 5‐enolpyruvylshikimate‐3‐phosphate synthase (De Oliveira et al., 2020, 2022), 3‐deoxy‐D‐manno‐2‐octulosonate‐8‐phosphate synthase (Araújo et al., 2019; Vainer et al., ²⁰⁰⁵), and N‐acetyl neuraminic acid synthetase (Popović et al., 2019). Currently, the PEP analogue fosfomycin is the only agent approved by the United States Food and Drug Administration for clinical use in the inhibition of MurA, primarily for the treatment of cystitis. Thus, it is important to highlight the increasing challenges posed by fosfomycin‐resistant microorganisms and the emergence of mutations that challenge fosfomycin effectiveness, demanding the use of covalent inhibitors of the MurA enzyme (Lima et al., 2021; Falagas et al., ²⁰¹⁹; Xin et al., ²⁰²²; De Oliveira et al., ²⁰²²).

One alternative solution is to investigate the formation of a dormant complex involving MurA as a potential target for the development of new antibiotics. According to Zhu et al. (2012). the dormant complex is a tightly locked complex of MurA and UNAM, which is the product of the MurB reaction, with PEP covalently attached to Cys115 (we refer to this as the QPA adduct). The dormancy complex is believed to be a regulatory mechanism that prevents peptidoglycan overproduction. In the presence of sufficient peptidoglycan, a dormant complex is formed, preventing MurA from catalyzing the reaction. This ensures that the cells produce only the required amount of peptidoglycan. Thus, the discovery of a dormant MurA complex suggests that UNAM inhibits MurA activity. Additionally, the dormant UNAM–PEP–MurA complex is believed to be sufficiently rigid to remain in the cytoplasm for an extended period, and the normal cellular concentration of inorganic phosphate is insufficient to open it. This could explain the failure of drug discovery efforts (Zhu et al., 2012).

To confirm these findings and to gain a better understanding of the molecular interactions that underlie the regulation of MurA activity by UNAM, computational simulations were performed to assess the binding free energies of UNAM and UNAG when bound to both PEP–MurA and MurA. In addition, τ‐random acceleration molecular dynamics (τRAMD) (Kokh et al., 2018) was used to estimate the residence time of UNAG and UNAM within the active site of the MurA enzyme. The binding mode description of UNAM in the enzyme active site sheds light on the design and development of novel UNAM analogues as potential MurA inhibitors.

2. RESULTS

2.1. Classical molecular dynamics

We used molecular dynamics (MD) simulations to capture the intricacies of the UNAG–PEP–MurA and UNAM–PEP–MurA complexes. These complexes are critical components in the biosynthesis of bacterial cell walls, making understanding their structure important in the context of antibiotic drug design and the study of resistance mechanisms. From these analyses, it was observed that the structural integrity of the UNAG–PEP–MurA and UNAM–PEP–MurA complexes was maintained throughout the 500 ns of the MD simulations. Systems in which PEP was not attached to Cys115 exhibited relatively low stability during the simulation, as shown in Figure 2a,b. The root mean square deviation values of the Pro112‐Pro121 loop (Figure 2c,d) demonstrate that the QPA adduct binding is essential for rapid complex system stabilization involving either UNAG or UNAM in the active site. Moreover, our findings demonstrate that the binding of PEP to Cys115 also induces a stabilizing effect on apo‐MurA, albeit to a lesser extent when compared to when UNAG or UNAM is present (Figure S1).

FIGURE 2.

Root mean square deviation (RMSD) plots obtained from (a) UNAG–PEP–MurA versus UNAG–MurA and (b) UNAM–PEP–MurA versus UNAM–MurA; and the loop region (residues 112–121) RMSD obtained from (c) UNAG–PEP–MurA versus UNAG–MurA and (b) UNAM–PEP–MurA versus UNAM–MurA.

The structural stability of the UNAM–PEP–MurA complex was greater than that of the UNAG–PEP–MurA complex, as UNAM formed hydrogen bonds with higher occupancy than its natural substrate, UNAG (Figure 3, Table S1). Particularly, UNAG acts as a hydrogen bond donor (HD) to Arg91 through the carboxylate lactyl moiety which is not present in UNAM. Furthermore, the occurrence of important interaction fingerprints including van der Waals (vdW), hydrophobic (HY), hydrogen bond acceptor (HA), and anionic character interactions (AN), are generally higher in UNAM than in UNAG, as observed for the residues Ala92, Trp95, Ile327, Phe328, and Arg371 (Figure 3a,b).

FIGURE 3.

Interaction fingerprints of MurA complexes, including van der Waals (vdW), hydrophobic (HY), hydrogen bond donor (HD), and acceptor (HA), and anionic (AN) character interactions, and the occurrence of each interaction. (a) UNAG–PEP–MurA versus UNAG–MurA (in red); (b) UNAM–PEP–MurA versus UNAM–MurA (in purple). The main HD interactions are detailed in panels (c) UNAG and (d) UNAM.

We clustered the 500 ns MD trajectories for the complexes using a hierarchical algorithm (Shao et al., 2007). Table S2 displays the results, highlighting the well‐populated cluster (Cluster 0) that was selected for further analysis. Representative structures from this cluster demonstrate that the UNAG–PEP–MurA and UNAM–PEP–MurA complexes both exist in a closed state, whereas the absence of PEP (covalently attached to Cys115) causes the active site of the enzyme to exhibit a semi‐open loop region (Figure 4).

FIGURE 4.

Centroid ID surface area demonstrates a difference in the loop region between residues 111 and 121. (a) UNAG–MurA (cyan) and UNAG–PEP–MurA (cyan); (b) UNAM–MurA (pink) and UNAM–PEP–MurA (pink). Loop region surfaces for (c) UNAG–PEP–MurA, (d) UNAG–MurA, (e) UNAM–PEP–MurA, and (f) UNAM–MurA.

In both UNAG– and UNAG–PEP–MurA complexes, the amino acid residue fluctuations were not especially high (≥3 Å) except in the C‐terminal region of the protein. However, in the UNAG–MurA and UNAM–MurA complexes (without PEP attached to Cys115) there was a single region that exhibited a fluctuation of >4.5 Å (Figure 5a,b). Particularly, in the root mean square fluctuation plots for the loop region (Figure 5c,d), the most pronounced fluctuations were observed around the polar residue, Cys115. The highlighted region is the semi‐opened loop of the enzyme, which contains the key residues, Cys115 and Arg120 (Figure 5e).

FIGURE 5.

Root mean square fluctuation (RMSF) plots for (a) UNAG–PEP–MurA and UNAG–MurA and (b) UNAM–PEP–MurA and UNAM–MurA; RMSF plot of the loop region (residues 110–123) for (c) UNAG–PEP–MurA and UNAG–MurA and (d) UNAM–PEP–MurA and UNAM–MurA; and (e) the residues involved in the loop region.

2.2. Binding free energy calculations

We performed binding free energy calculations to evaluate the affinity of the ligands for the PEP–MurA and MurA complexes using three methods: molecular mechanics with generalized Born and surface area solvation (MM/GBSA), molecular mechanics with Poisson–Boltzmann surface area solvation (MM/PBSA), and solvated interaction energy (SIE). Both the PEP–MurA and MurA complexes, showed predominantly favorable van der Waals interactions with the active‐site residues of the enzyme when associated with either UNAG or UNAM (Table 1 and Table S3). UNAM exhibited superior binding affinity for the enzyme compared to its native substrate, UNAG. The main favorable interacting residues for the UNAM system were Arg91, Arg120, Pro121, Val122, Asp123, Leu124, His115, Lys160, Ser162, Val163, Gly164, Ala165, Asp305, Ile327, Phe328, Arg331, Leu370, and Arg371 (Figure 6).

TABLE 1.

Binding free energy values for the PEP–MurA and MurA complexes obtained from MM/GBSA, MM/PBSA, and SIE methods.

Method	Energy (kcal/mol)	PEP–MurA complex		MurA complex
		UNAG	UNAM	UNAG	UNAM
∆G total	MM/GBSA	−45.25 (±6.30)	−54.09 (±10.59)	−110.08 (±14.83)	−74.56 (±15.96)
	MM/PBSA	−26.13 (±11.48)	−31.77 (±13.55)	−73.36 (±10.89)	−39.94 (±16.64)
	SIE	−7.55 (±0.60)	−8.27 (±0.84)	−11.21 (±0.57)	−8.55 (±1.17)

Abbreviations: MM/GBSA, molecular mechanics with generalized Born and surface area solvation; MM/PBSA, molecular mechanics with Poisson–Boltzmann surface area solvation; SIE, solvated interaction energy.

FIGURE 6.

Binding free energy per residue decomposition of ∆G _total (MM/GBSA) for the UNAG–PEP–MurA (black) and UNAM–PEP–MurA (blue) complexes.

2.3. Binding kinetic analyses using the τRAMD method

We computed relative residence times using the τRAMD protocol (Kokh et al., 2018, 2019). Four replicates with 20 sampling trajectories per replicate were used to calculate the dissociation time (DT, ns) for each trajectory; DT represents the time (ns) during which the ligand remains at the MurA active site. UNAM remained at the active site longer than UNAG in all trajectories. The Anderson–Darling, Shapiro–Wilk, and Kolmogorov–Smirnov normality tests demonstrated that there was no evidence to reject the null hypothesis that the sample had an approximately normal distribution (Table S5). For all replicates, the Q–Q plot (Figure S2) also showed a normal distribution.

The box plot shown in Figure 7 demonstrates the residence time (τ, ns) of each of the ligands at the binding site; the τ value is the average DT value for each trajectory in each replicate. The UNAM–PEP–MurA complex had a higher τ than the UNAG–PEP–MurA complex (22.1 ± 1.96 and 4.45 ± 0.77, respectively). In contrast to complexes with PEP, the absence of the covalent bonds destabilized the closed state of the Pro112–Pro121 loop. Consequently, the absence of the QPA adduct allowed the ligand to leave the binding site rapidly (UNAG: 2.98 ± 0.62; UNAM: 3.65 ± 0.82) (Figure S3).

FIGURE 7.

Results of the four replicas, box plots of the relative residence times obtained for each replica for (a) UNAG–PEP–MurA and (b) UNAM–PEP–MurA, obtained through τRAMD kinetic calculations. The median for each replica is represented in orange and the mean as a dashed red line. The average residence time (in ns) computed for all replicas is shown with its standard deviation above each box.

Equation (1) was used to derive k _off relative values for both complexes. UNAG–PEP–MurA and UNAM–PEP–MurA presented k _off relative values of 2.25 × 10⁸ s⁻¹ and 4.52 × 10⁷ s⁻¹, respectively.

$$\tau =\frac{1}{k_{\mathrm{off}}}\therefore k_{\mathrm{off}}=\frac{1}{\tau }$$ (1)

To calculate the protein–ligand interaction fingerprints (IFPs), we used k‐means to cluster the last 100 frames of the dissociation simulations of the 80 UNAG–MurA complex trajectories and of the 80 UNAM–PEP–MurA complex trajectories. We identified the metastable states involved in the dissociation pathways of UNAG/UNAM disassociating from PEP–MurA and their dissociation routes.

For the UNAG complexes, we observed three metastable states with large populations. Thus, these three states (Clusters 5–7) represent the flow of the dissociation route (gray arrows, Figure 8a) to the unbound final state. The differences between their ligand coordinate distances (∆COM) ranged between 5.5 and 7 Å (Clusters 5–7) and their unbounded state values were >13 Å away from their initial state values (Figure 8b). Each state has a ligand output direction (Figure 8c–e); Clusters 5 and 6 follow the same direction and Cluster 7 follows a ‘sideways’ direction. Note that all these directions pass below the Pro112–Pro121 loop region and the principal metastable states for UNAG dissociation are represented by Clusters 5–7.

FIGURE 8.

Clustering (k‐means) results for the analysis of the dissociation of the ligand UNAG from the PEP–MurA binding site. (a) Variation of the ligand center of mass (∆COM) from the initial equilibrium state through the initial clusters and metastable states to the unbound state, showing the nodes and their populations according to increasing ∆COM. (b) Metastable states corresponding to Clusters 5–7 shown as iso‐surfaces of the ligand center of mass population density. (c–e) Routes of dissociation in Clusters 5–7, respectively. (f) IFPs space for UNAG–PEP–MurA calculated for each node (Clusters 1–8) showing the population colored along a white–blue gradient according to increasing contribution. HD/HA, hydrogen‐bond donor/acceptor, respectively; HY, hydrophobic interactions; IN, salt bridge (negatively charged).

The protein–ligand IFPs demonstrated that the ligand moved from the binding site to an unbound state, forming HD/HA, HY, and salt bridge (negatively charged, IN) IFPs. Interactions with residues Arg120 (HD, HY) and Ser162 (HD) remained throughout the simulation; however, interactions with residues Asp123 (HA) and Leu124 (HD) stood out, with a population <7.0 from Cluster 2 onwards.

We identified several intermediate metastable states in the dissociation pathway (Figure 9a). These indole dissociation flows exhibited large variations (Clusters 1–5). In Clusters 4 and 5, the orientation of the route differed from that in the same ligand region during unbinding. Cluster 5 used a route heading toward the Pro112–Pro121 loop with the QPA residue, while Cluster 4 headed under and around this loop region (Figure 9d). Metastable Clusters 6 and 7 showed the same route orientation, next to the loop region. The main metastable states in the dissociation pathway to the UNAM unbound state were represented by Clusters 6 and 7.

FIGURE 9.

Clustering (k‐means) results for the analysis of the dissociation of the UNAM ligand from the PEP–MurA binding site. (a) Variation of the ligand center of mass (∆COM) from the initial equilibrium state, through the initial clusters and metastable states, to the unbound state showing the nodes and their populations according to increasing ∆COM. (b) Metastable states corresponding to Clusters 4–7 shown as iso‐surfaces of the ligand center of mass population density. (c–e) Routes of dissociation in Clusters 7, 4 and 5, and 6, respectively. (f) IFPs space for UNAM–PEP–MurA (ligand–protein contacts) calculated for each node (Cluster 1–8) showing the population color along a white–blue gradient according to increasing contribution.

The contact interactions in UNAM–PEP–MurA (Figure 9f) showed that the main residues Arg91 (IN), Arg120 (IN, HD), Arg123 (HA), Leu124 (HD), and Ser162 (HD) remained throughout the simulation; however, the main residues Trp95 (HY), Leu124 (HD), His125 (HD), Lys160 (HD), Val163 (HD), and Gly164 (HD) exhibited reduced interaction levels during the simulation.

3. DISCUSSION

3.1. UNAG and UNAM complexed with MurA binding affinities and PEP covalent binding importance in MurA

Our findings underscore the role of covalent interactions between PEP and Cys115 in ensuring protein stability. Cys115 is part of a crucial loop region (Pro112–Pro121) responsible for the induced‐fit mechanism that triggers the conformational change induced by the interaction of MurA with UNAG (Schönbrunn et al., 2000; Lima et al., ²⁰¹⁷). When attached to Cys115, PEP interacts with Arg91, Gly114, Ala116, Ile117, Gly118, Arg120, and Arg397 in the UNAG–PEP–MurA and UNAM–PEP–MurA complexes. In the absence of the covalent bond with Cys115, only interactions with Arg91, Arg120, or Arg397 were observed. Therefore, the absence of the rest of these interactions was directly associated with loop instability. Furthermore, our findings demonstrate that the covalent bond mediated by PEP is essential for stabilizing the loop region of the enzyme, even in the absence of UNAG/UNAM ligands at the active site (Figure S1).

A strong interaction between the ligands and Arg120 was observed in all studied MurA‐UNAG and MurA‐UNAM complexes with or without PEP (Figure 3). Both MurA‐UNAG and MurA‐UNAM complexes could be delineated as systems primarily engaged in HY and van der Waals interactions. Electrostatic interactions were relatively sparse in both systems; however, they were present, with key residues such as Arg91 and Arg120 playing pivotal roles in mediating these interactions.

It is worth noting that the distinguishing structural feature between UNAM and UNAG is the attachment of a lactyl group to the 3′‐hydroxy substituent on UNAG's UDP‐GlcNAc moiety (Figure 1). In the cytoplasmic phase of peptidoglycan biosynthesis, the enzyme MurA is responsible for initiating the Mur pathway by catalyzing the formation of UNAG‐EP (Egan et al., 2020; Barreteau et al., ²⁰⁰⁸). Subsequently, MurB utilizes NADPH to convert enolpyruvate into a lactyl group, thus facilitating the synthesis of UNAM. According to our results, the orientation of this lactyl group strongly favored interactions with Arg91 and Arg120 and promoted hydrogen bonding with the Leu124, Ser162, and Asp305 residues (Table S1). These results agree with previous experimental data, indicating that Arg91 and Arg120 are responsible for the tightly closed state of the MurA enzyme (Zhu et al., 2012). Additionally, it reinforces the idea that tightening the dormant complex with UNAM might be an efficient feedback regulation of murein biosynthesis. Potentially, disrupting the rapid interaction between PEP and UNAG, preventing their reaction.

Analysis of the main interactions in these systems highlights Asp305 interacting with O3′ and O4′ when UNAG is present (Figure 3). In UNAM, which contains a lactyl group attached to the O3′, Asp305 can only form a hydrogen bond with O4′. Nevertheless, our results show that systems that PEP is not bound to Cys115 are less likely to keep this interaction. Although computational predictions have outlined the initial framework of the MurA reaction mechanism, the precise interaction between Asp305 and its substrate, and its subsequent impact on enzymatic activity, are yet to be elucidated (Mihalovits et al., 2019). Asp305 has been shown to play an important role in the final proton abstraction from the C3 atom of the PEP moiety (Eschenburg et al., 2003), which is required for the elimination of inorganic phosphate.

Hence, we demonstrated that the covalent binding of PEP to Cys115 is pivotal for the structural conservation of the loop movement in this protein. Consequently, this binding directly correlates with the sustained closed state of MurA in the presence of UNAG/UNAM. The dynamic behavior of the residues comprising the loop region is intricately linked to the presence of the QPA adduct. Consequently, it is evident that UNAM–PEP–MurA systems exhibit minimal mobility and a propensity to maintain a closed‐state conformation.

We hypothesized that the higher occurrence of hydrogen bond interactions between UNAM and Arg120 in the UNAM–PEP–MurA complex than between UNAG and Arg120 in the UNAG–PEP–MurA complex could potentially explain the higher degree of loop movement conservation. This is because Arg120 is one of the key residues influencing the open–closed state of the protein (Mihalovits et al., 2019). Therefore, we demonstrated that UNAM– and PEP–binding together induced MurA to remain closed. This corroborated the hypothesis that UNAM can bind to the active site of the enzyme, thereby fostering a more extensive array of interactions.

We also emphasize that the QPA adduct is essential to the higher binding affinity displayed by UNAM. The absence of this covalent bond caused a change in the thermodynamics of the system, as demonstrated by the fact that UNAG alone had a higher affinity than UNAM alone. We observed favorable interactions with residues Arg120, Arg91, Ser162, Val163, and Gly164, which are conserved in the MurA of other organisms, such as Escherichia coli and Staphylococcus aureus and that are key residues for diterpene inhibitors (Funes Chabán et al., 2021). These residues play key roles in stabilizing the closed conformation of the MurA of Enterobacter cloacae (EcMurA). Therefore, in the UNAG–PEP–MurA complex, the phosphate group interacted with these residues.

3.2. Ligand unbinding mechanisms

Using the τRAMD computational method (Kokh et al., 2018), we estimated the relative residence time of UNAG and UNAM in the active site of the MurA enzyme and kinetically explained the phenomenon of UNAM forming a ‘dormant complex’. The residence time of the ligand is calculated as 1/k _off, as illustrated in Figure S4. The determination of the dissociation constant (k _off) poses a considerably more intricate computational challenge than the calculation of binding affinity. This complexity arises from the necessity of performing comprehensive sampling of transition states, which frequently occur owing to the multiple configurations of the macromolecule–ligand complex (Kokh et al., 2018). Freely accessible and highly accurate, the τRAMD method is effective in both the simulation process and data processing. It can provide relative dissociation rate constants, the dissociation route, and the transition metastable states. This method has been successfully applied in various research works (Kokh & Wade, 2021; Berger et al., ²⁰²¹; Nunes‐Alves et al., ²⁰²¹).

Through the assessment of the computed relative residence times, it was observed that UNAM remained in the active site of the MurA enzyme for a more extended period than its natural substrate UNAG. The primary intermolecular interactions within the UNAM–PEP–MurA complex persisted throughout the dissociation simulation. This preservation allowed the movement of the Pro112–Pro121 loop, which contains the QPA and Arg120 residues. These residues are responsible for the open–closed conformations of MurA in various organisms, such as Pseudomonas aeruginosa (Lima et al., 2017), producing an open state following ligand unbinding (Figure 10). Notably, the interactions observed in both the MD and τRAMD simulations maintain exceptional strength, persisting throughout the simulation period, even after the ligand leaves the binding site. This corroborates the thermodynamical idea that UNAM exhibits a significantly higher affinity for the MurA‐binding site than the native UNAG substrate.

FIGURE 10.

Bound and unbound states of UNAM–PEP–MurA, highlighting the Pro112–Pro121 loop region (blue color), UNAM, and the key residue Arg120.

The presence of multiple transition pathways within UNAM, each encompassing several intermediary states before reaching the final unbound state, led to a notable elongation in the dissociation duration. Specifically, each metastable state can be associated with subsequent transition barriers along the dissociation trajectory. These barriers extended the temporal span of the dissociation process. Consequently, the primary distinction between the ligand dissociation processes lies in the number of flow transitions and metastable states involved. As the dissociation process progresses, interactions involving key residues Leu124, Ser162, and Asp305 become weaker. Moreover, the increased intensity of interactions shown by UNAM, compared to UNAG, results in a longer residence time. This becomes particularly apparent when examining the IFP data for residues Arg91 and Arg120, as depicted in Figures 8f and 9f. Apart from Arg91, both analyses reveal that the interaction with Arg120 is sustained up to the point of complete ligand departure. This sustained interaction ultimately leads MurA to transition into the open conformation, as depicted in Figure 10.

We also emphasize that the formation of the QPA residue is fundamental to the ligand‐release mechanism. In the absence of covalently bound PEP, the ligands do not have well‐populated metastable states and, in addition to the absence of the QPA, this makes the loop region unstable. This allows the ligand to rapidly leave the active site in any possible direction, causing it to have more than two final exit states (linker regions after exiting the active site). However, we observed that the ligands still have the same characteristic of leaving the site “pulling” the loop, owing to their strong interaction with Arg120.

The crucial information regarding the interactions that UNAG and UNAM undergo at the active site of the MurA enzyme when PEP covalently binds to Cys115 paves the way for new studies on non‐covalent inhibitors. This provides alternative options to the use of the only commercially available medication, Fosfomycin. Understanding how the non‐covalent ligand UNAM induces the enzyme to remain tightly closed enables the design of new drugs based on these critical interactions.

4. CONCLUSIONS

We demonstrated the importance of kinetic and thermodynamic studies to understand the dynamics of MurA–ligand complexes. Using MD, we observed that UNAM interacted more strongly with the amino acid residues of the MurA active site than the natural substrate of the protein, UNAG. Strong interactions between UNAM and residues Arg120 and Arg91 play a key role in stabilizing the closed conformation of EcMurA. In terms of structure, UNAM distinguishes itself from UNAG through the inclusion of a lactyl moiety tethered to the 3′‐hydroxyl group. This structural difference between UNAG and UNAM is the main factor responsible for increased hydrogen interactions in the catalytic pocket of the protein. Furthermore, the UNAM–PEP–MurA complex had higher affinity than the UNAG–PEP–MurA complex. Thus, UNAM–PEP–MurA had a lower dissociation constant than UNAG–PEP–MurA. The higher value of the dissociation constant for UNAG is directly related to the fact that the primary pathway leads directly from the bound state to dissociation, or through an intermediate state located near the bound state. Therefore, the longer τ of UNAM is related to its complex dissociation pathways, each with multiple intermediate metastable states, as shown in previous research based on different systems (Berger et al., 2021; Nunes‐Alves et al., ²⁰²¹). Our calculations provide a clear rationale justifying the continued presence of UNAM in the active site of the protein for longer than its natural substrate UNAG, forming the strongly closed complex UNAM–PEP–MurA, known as the “dormant complex” (Zhu et al., 2012). Moreover, we demonstrate the importance of PEP–Cys115 covalent binding in stabilizing the closed conformation of the protein in complexes with both UNAG and UNAM. We consider the techniques used in this study particularly well suited for evaluating novel UNAM analogues as potential inhibitors of MurA activity.

5. MATERIALS AND METHODS

5.1. Molecular dynamics protocol

The three‐dimensional structure of the MurA of the Enterobacter cloacae (EcMurA) was obtained from the RSCB Protein Data Bank under the PDB code 3SU9 (Zhu et al., 2012) (X‐ray diffraction, resolution: 2.20 Å). This MurA structure is complexed with the substrates, UNAM and PEP, which are covalently linked to the residue Cys115.

The 3D structure of UNAG was constructed based on the crystallographic conformation of UNAM. General amber force field (Wang et al., 2004) parameters were used to describe UNAG and UNAM, employing the approach described in our previous study (Lima et al., 2017). The protonation states of the ionisable residues in the MurA structure were determined by pK _a calculations using the H⁺⁺ webserver (Anandakrishnan et al., 2012; Myers et al., ²⁰⁰⁶) at pH 7. The systems were solvated in the tLeap module using a cubic water box with the TIP3P explicit solvation model (Jorgensen et al., 1983) and a radius of 12.0 Å between the box and the protein surface. Na⁺ counter‐ions were added to maintain the electroneutrality of the system. We prepared four systems for the simulations: two with the PEP–Cys115 adduct (UNAG–PEP–MurA and UNAM–PEP–MurA), and two without it (UNAG–MurA and UNAM–MurA).

The energy minimization of the systems was performed in four steps. The first minimization step included only the counter‐ions and water molecules, the second step included only the hydrogen atoms of the protein structure, and the third step encompassed both the hydrogen atoms of the protein, ligand and the water molecules. Finally, a fourth minimization step was performed on the entire system.

The Amber18 software package (Case et al., 2018) was used to perform MD simulations. The system was heated from 0 to 300 K while performing 200 ps of MD in a constant volume with a restraint weight of 5.0 kcal.mol⁻¹ Ų at the positions of the atoms. Before carrying out the production step, all protein–ligand systems were equilibrated with 500 ps of MD with no restrictions at a constant pressure of 1 bar. The temperature was maintained at 300 K by coupling to a Langevin thermostat using a collision frequency of 2 ps⁻¹ and the isotropic constant pressure was maintained at 1 bar using a Berendsen barostat. A cutoff of 10 Å was used for unbound interactions, and the particle mesh Ewald method was used to calculate the long‐range electrostatic interactions. Finally, the DM simulation was performed with a simulation time of 500 ns at 300 K.

The profile of chemical interactions performed by the ligands in the active site of the enzyme during the MD simulation was determined using the ProLIF library (Bouysset & Fiorucci, 2021), which is responsible for performing fingerprint analyses of the interactions between proteins and ligands. After analyzing the four complexes (UNAG/UNAM–PEP–MurA and UNAG/UNAM–MurA), filtering was performed to select only those interactions that appeared with an occurrence greater than 50% during all simulation frames.

5.2. Binding free energy calculations

To perform binding free energy calculations, we employed the MM/GBSA method using the MM/PBSA.py (Miller et al., 2012) package in Amber18 (Case et al., 2018). The SIE method was employed using the Sietraj program (version 2.0) (Naïm et al., 2007). To perform these calculations, 10,000 frames of the MD trajectory, corresponding to the last 100 ns, were selected. We used MM/GBSA decomposition analysis to identify the main residues contributing to the binding affinity of the complex.

In MM/GBSA, the binding free energy value ($∆G_{\text{binding}}$) is calculated through Equation (2), where the terms $∆E_{\mathrm{MM}}$, $∆G_{\mathrm{sol}}$, and $T∆S$ correspond to the changes in the molecular mechanics energy of the gas phase (MM), free of solvation energy, and conformational entropy after ligand binding, respectively. Thus, the first term $∆E_{\mathrm{MM}}$ corresponds to the molecular mechanics energy, which is the sum of the intramolecular energy (internal, electrostatic, and van der Waals energy) (De Oliveira et al., 2020; Miller et al., ²⁰¹²; Genheden & Ryde, ²⁰¹⁵; Wang et al., ²⁰¹⁷).

$$∆G_{\text{binding}}=∆E_{\mathrm{MM}}+∆G_{\mathrm{sol}}-T∆S$$ (2)

The second term, $∆G_{\mathrm{sol}}$, is the sum of the electrostatic solvation energy (polar contribution) and the non‐polar contribution between the solute and solvent (Wang et al., 2019). Finally, in the $T∆S$ term, $T$ is the temperature and $∆S$ is the entropy of the system.

The SIE method belongs to a group of endpoint methods based on the force field and demonstrates a reasonable compromise between time, accuracy, and computational resources (Naïm et al., 2007). The SIE method uses an explicit solvation model to approximate the binding free energy value of a protein–ligand ($∆G_{\text{binding}}$) complex in an aqueous solution. The calculation of this value is based approximately on adding the contribution of the interaction energy to the contribution of the solvation energy (Naïm et al., 2007). Thus, the equation can be formally represented by the contributions of the electrostatic and non‐polar components (Naïm et al., 2007; Silva et al., ²⁰¹⁶), as shown in Equation (3).

$$\begin{array}{rcl}∆G_{\text{binding}}& \approx & E_{\text{inter}}+∆G_{\text{desolv}}\\ & =& {\left(E_{\text{inter}}^{\text{Coul}}+∆G_{\text{desolv}}^{\mathrm{R}}\right)}_{\text{electrostatic}}\\ & & +{\left(E_{\text{inter}}^{\mathrm{vdW}}+∆G_{\text{desolv}}^{\mathrm{np}}\right)}_{\text{non}-\text{polar}}\end{array}$$ (3)

where $E_{\text{inter}}^{\text{Coul}}$ represents the Coulombic contribution, $∆G_{\text{desolv}}^{\mathrm{R}}$ is the free energy of electrostatic desolvation (polar components), $E_{\text{inter}}^{\mathrm{vdW}}$ represents the van der Waals contribution, and $∆G_{\text{desolv}}^{\mathrm{np}}$ is the free energy of non‐polar desolvation (non‐polar component) (Naïm et al., 2007).

5.3. The τ‐random acceleration molecular dynamics protocol

The τRAMD technique is proficient in discerning the egress pathways of ligands from macromolecular binding sites. Consequently, this method enables the systematic exploration of ligand dissociation from the active site of an enzyme complex (Kokh et al., 2018, 2019; Nguyen et al., ²⁰²²). The RAMD method uses a randomly oriented force (f) of constant magnitude applied to the center of mass of the molecule for n simulation steps to accelerate the movement of the ligand from its position at the beginning of the n simulation steps (Kokh et al., 2018, 2019). The movement of the molecule is evaluated after these simulation steps through a distance calculation (r) between its center of mass in the new and old positions (Kokh et al., 2018, 2019).

When the change in distance r is less than the limit distance (r _min), a new direction of the applied force is chosen randomly. In contrast, when r is larger than r _min the simulation continues to the next n simulation steps adopting a force in the same direction (Kokh et al., 2018, 2019, 2020). The RAMD method is employed in the τRAMD protocol to calculate the relative residence time of a receptor–ligand complex (Figure S4). The residence time refers to the time during which the ligand remains in the active site of the enzyme, and until it reaches its unbound state. The τRAMD protocol aims to calculate the relative residence time (τ) of a molecular target, and it is the main determinant of the effectiveness of a drug. However, obtaining these data using computational methods remains highly challenging; thus, it is usually not considered in the early stages of drug design (Kokh et al., 2018).

The preparation of the substrates and proteins for the formation of the solvated complex followed the classical MD simulation protocol; although the distance between the edges of the box and the closest atom of the solute was 18 Å. Then, the atomic coordinates provided by the last equilibration step of Amber18 were used as the input in GROMACS–RAMD version 2.2, a version of GROMACS 2020 (Abraham et al., 2015) adapted according to Kokh et al. (2020) (Kokh et al., 2020) (available at https://github.com/HITS‐MCM/gromacs‐RAMD/releases/tag/gromacs‐2020.5‐RAMD‐2.0), to generate replicas in a new equilibration step.

During this equilibration step, each ligand‐protein complex was equilibrated four times. The first replica used the Berendsen thermostat and the second the Nose–Hoover thermostat. For the third and fourth replicas, we used the Parrinello–Rahman barostat and the Nose–Hoover thermostat with a simulation time of 20 ns. The τRAMD simulation procedure was performed with four replicas, using a random force with a magnitude of 14 kcal.mol⁻¹ Å⁻¹ for the simulations. For each replica (1–4), we simulated 20 dissociation trajectories using 40 ns for each trajectory to calculate the residence time required for ligand dissociation in 50% of the trajectories of the four ligand complexes (Kokh et al., 2018).

For statistical analysis, we used the Kolmogorov–Smirnov test (KS test), with $\alpha$ = 0.05, to analyze whether our samples had the same statistical distribution based on Equation (4), where $F\left(Y_i\right)$ is the theoretical cumulative distribution of the test distribution.

$$D=\underset{1\le i\le N}{\max }\left(F\left(Y_i\right)-\frac{i-1}{N},\frac{i}{N}-F\left(Y_i\right)\right)$$ (4)

To analyze the residence time from our four replicates (with 20 trajectories in each), we used Python script for MD‐IFP analysis (Van Rossum & Drake, 2009). Clustering of all trajectories was performed for the last 100 frames using a k‐means algorithm to identify metastable states and their routes to the unbinding state. We used this number of frames for both ligand systems because it contained both the binding state and the dissociation route for the unbinding state. The difference between the distance of the ligand coordinates (∆COM), in Å, was used to identify the unbinding state (>10 Å). Furthermore, we calculated the protein–ligand IFPs for each cluster until ligand dissociation, as calculated in previous studies (Berger et al., 2021; Nunes‐Alves et al., ²⁰²¹).

AUTHOR CONTRIBUTIONS

Anderson H. Lima: Conceptualization; formal analysis; writing – review and editing; supervision. Maycon Oliveira: Investigation; formal analysis; writing – original draft; methodology. Kauê S. da Costa: Writing – original draft; investigation; data curation; formal analysis. José Rogério Araújo Silva: Supervision; visualization; funding acquisition; formal analysis. Jerônimo Lameira: Supervision; writing – review and editing; formal analysis.

Supporting information

TABLE S1. Hydrogen interactions for the UNAG–PEP–MurA and UNAM–PEP–MurA complex.

TABLE S2. Results obtained in clustering for both complexes: cluster ID, frames, relative time (ns), centroid ID and occupancy.

TABLE S3. Binding free energy values were calculated for the analyzed PEP–MurA and MurA complexes using MM/GBSA, MM/PBSA, and SIE methods.

TABLE S4. Dissociation time on each of the 20 trajectories for each of the four replicas for the UNAG‐PEP‐MurA and UNAM‐PEP‐MurA complexes.

TABLE S5. Normality Tests for each replica in both complexes with PEP‐MurA using Anderson‐Darling, Shapiro–Wilk, and Kolmogorov–Smirnov tests, with $\alpha$ = 0.05 showing the value test and the p‐value.

FIGURE S1. (a) RMSF of the amino acid residues of MurA and MurA‐PEP (b) and the loop region. (c) RMSD of the MurA and MurA‐PEP systems and the (d) loop region.

FIGURE S2. Q‐Q plot for each replica for both complexes.

FIGURE S3. Clustering (k‐means) results for the analysis of dissociation of the ligand UNAG and UNAM from MurA binding site. Variation of the ligand center of mass (∆COM) from the initial equilibrium state through initial clusters and metastable states to the unbound state showing the nodes and their population according to increasing the ∆COM for (a) UNAG‐MurA and (d) UNAM‐MurA. Mean of the residence time (in ns) and its standard deviation (std) for complex (b) UNAG‐PEP‐MurA and (e) UNAM‐PEP‐MurA. IFPs space to (c) UNAG‐MurA and (f) UNAM‐MurA calculated by each node cluster 1–8 showing the population in color white‐blue pallet according with the increase contribution. AR, aromatic; HD/HA, hydrogen‐bond donor/acceptor, respectively; HY, hydrophobic interactions; IN, salt bridge (negative charge).

FIGURE S4. Scheme of MurA reaction constants in the unbound and bound states.

FIGURE S5. Clusters obtained from the MurA and MurA‐PEP complexes with emphasis on the movement of the loop region.

de Oliveira MVD, da Costa KS, Silva JRA, Lameira J, Lima AH. Role of UDP‐N‐acetylmuramic acid in the regulation of MurA activity revealed by molecular dynamics simulations. Protein Science. 2024;33(4):e4969. 10.1002/pro.4969