A computational model of ESAT-6 complex in membrane

Abstract

One quarter of the world’s population are infected by Mycobacterium tuberculosis (Mtb), which is a leading death-causing bacterial pathogen. Recent evidence has demonstrated that two virulence factors, ESAT-6 and CFP-10, play crucial roles in Mtb’s cytosolic translocation. Many efforts have been made to study the ESAT-6 and CFP-10 proteins. Some studies have shown that ESAT-6 has an essential role in rupturing phagosome. However, the mechanisms of how ESAT-6 interacts with the membrane have not yet been fully understood. Recent studies indicate that the ESAT-6 disassociates with CFP-10 upon their interaction with phagosome membrane, forming a membrane-spanning pore. Based on these observations, as well as the available structure of ESAT-6, ESAT-6 is hypothesized to form an oligomer for membrane insertion as well as rupturing. Such an ESAT-6 oligomer may play a significant role in the tuberculosis infection. Therefore, deeper understanding of the oligomerization of ESAT-6 will establish new directions for tuberculosis treatment. However, the structure of the oligomer of ESAT-6 is not known. Here, we proposed a comprehensive approach to model the complex structures of ESAT-6 oligomer inside a membrane. Several computational tools, including MD simulation, symmetrical docking, MM/PBSA, are used to obtain and characterize such a complex structure. Results from our studies lead to a well-supported hypothesis of the ESAT-6 oligomerization as well as the identification of essential residues in stabilizing the ESAT-6 oligomer which provide useful insights for future drug design targeting tuberculosis. The approach in this research can also be used to model and study other cross-membrane complex structures.

1. Introduction

One quarter of the world’s population are infected by Mycobacterium tuberculosis (Mtb), which has been a leading death-causing bacterial pathogen. Recent evidences have demonstrated that two virulence factors, ESAT-6 and CFP-10, play crucial roles in cytosolic translocation.^1,2 Understanding the roles of ESAT-6 and CFP-10 will provide novel directions for future treatment of tuberculosis. Therefore, many efforts have been made to study the ESAT-6 and CFP-10 proteins. But the mechanism of how ESAT-6 and CFP-10 contribute to Mtb cytosolic translocation and virulence is still poorly understood.

Some recent research indicates that the ESAT-6 first disassociates with CFP-10 and then is inserted into the membrane to create a transmembrane pore.^3–5 The membrane rupturing activity of ESAT-6 was initially observed in both the conditions with and without CFP-10.⁶ Further studies from De Jonge et al. show that the interaction of CFP-10 with membranes is significantly weaker than that of ESAT-6. Furthermore, by employing electron microscopy, they demonstrated that ESAT-6 destabilized and lysed liposomal membrane, whereas CFP-10 did not in Ref. 7. More recently, direct interactions of ESAT-6 with membrane were observed,^3–5 demonstrating that ESAT-6 can be inserted into the membranes and then create membrane-spanning pores, which are essential to Mtb’s cytosolic translocation. In the investigation done by Ma et al. in 2015,⁵ the transmembrane region of ESAT-6 was mapped by labeling several residues with NBD (Nitro-2-1, 3-BenzoxaDiazol-4-yl). Among the labeled residues, residues Ser-35 and Ala-60 showed rapid and strong liposome dependent emission, which advocated that these residues are facing towards the lipid. In the complex form of ESAT-6/CFP-10, the residue Ser-35 and Ala-60 are at the contact interface between ESAT-6 and CFP-10 and in experiment by Ma et al.⁵ in 2015 they are inside the membrane and facing towards the membrane. It indicated the dissociation of the ESAT-6/CFP-10 complex and that only ESAT-6 is involved in the membrane lysis activity.⁵

The length of helixes of ESAT-6 is approximately 50 Å, equivalent to the depth of the typical lipid bilayer. Due to the small molecular size (6-kDa) of ESAT-6 monomer, insertion of a single ESAT-6 inside membrane is insufficient for the bacterial translocation or the exposure of the bacterial DNA to the DNA sensory mechanism in the cytosol of the host cell.⁵ While there are rarely any results describing the structure of ESAT-6 when inserted in the membrane, significant evidences have shown that ESAT-6 can form stable homo-complexes.^8–10 These factors have led many scientists to believe that the ESAT-6 forms oligomers for the purpose of creating the transmembrane pores, which are capable of rupturing the phagosome membrane and hence contributing to the virulence of Mtb. However, the oligomeric states of such an ESAT-6 transmembrane pore are unknown.

It is extremely challenging to determine the number of ESAT-6 monomers in the oligomer experimentally. Because the ESAT-6 proteins interact with membranes in a complicated manner and hence is hard to purify and crystallize them.^4,5,7,10,11 Fortunately, due to the rapid development of new algorithms, many computational approaches have been approved to be successful at biochemistry studies, such as protein–protein docking,^12,13 proteins assembly,^14,15 protein–membrane interactions,¹⁶ protein–protein binding energy^17,18 and force calculations,^19,20 etc. Combination of such computational approaches can solve complicated biology problems.^21,22 In this regard, computational approaches can aid the experimental techniques to model the structure of the oligomer of ESAT-6. Therefore, in this work, we adopted comprehensive computational techniques to model the structure of putative ESAT-6 oligomer as well as characterize their interaction with membrane.

As most transmembrane protein channels have cylindrical or barrel shape, one can hypothesize that the oligomer of ESAT-6 is also cylindrical in shape. Therefore, to imitate the process of oligomerization of ESAT-6 in a symmetrical manner, the docking algorithm, SymmDock,²³ was utilized when the docking symmetry was set as rotational symmetry from C3 to C8 as the number of involved ESAT-6 monomers increase from 3 to 8 (Fig. 1). The predicted structures were then screened based on cylindrical shape and configurations of residues Ser-35 and Ala-60 (Figs. 2(a) and 2(b)) as discussed above. The screened structures were then clustered to obtain the best and representative structures. These screening process resulted in six of the mostly reasonable structure ESAT-6 oligomers by SymmDock (Fig. 1), in which different numbers of ESAT-6 monomers involved. In order to improve the docking accuracy, the oligomerization process was predicted in parallel using another algorithm, M-ZDOCK.²⁴ A similar screening process was done for structures predicted by M-ZDOCK, resulting in six representative structures of oligomers. The obtained 12 structures were then subjected to molecular dynamic (MD)²⁵ simulations after these oligomers were inserted into the membrane (type POPC: Phosphatidylcholine) to further study their stabilities and to investigate the interaction between ESAT-6 oligomers and the membrane. The MM/PBSA²⁶ analysis was then carried out to calculate the binding free energies based on the trajectories of MD simulations, which shows that the oligomer formed by four ESAT-6 monomers has the lowest normalized binding energy, suggesting that the ESAT-6 is likely to be assembled into a cylindrical tetramer, which is responsible for creating the transmembrane pores. Furthermore, our salt bridge analysis, long-range electrostatic interaction and H-bond analysis identified several essential residues that are essential in stabilizing the tetramer and its interactions with the membrane.

Fig. 1.

(Color online) The best predicted structure of the ESAT-6 oligomers by SymmDock. (a) ESAT-6 trimer. (b) ESAT-6 tetramer. (c) ESAT-6 pentamer. (d) ESAT-6 hexamer. (e) ESAT-6 heptamer. (f) ESAT-6 Octamer. In all panels, both side view and the top view of the oligomers are presented, where the ESAT-6 monomers are shown in various color.

Fig. 2.

(Color online) Truncation and reconstruction of flexible N- and C-terminal arms of ESAT-6. (a) A best-predicted docking structure based on flexible terminals truncated ESAT-6 tetramer. Four monomers involved in the assembly of the oligomer and are colored differently. The lipid facing residues Ser-35 and Ala-60 are leveled and colored yellow. (b) Reconstruction of the flexible terminals to the predicted structure of ESAT-6 tetramer after the filtering process. The reconstructed arms are labeled and highlighted with green color. (c) Insertion of the ESAT-6 tetramer inside the membrane with proper orientation and removal of overlapped lipids. (d) Solvation and ionization of the system and its dimension along the X- and Y-axis. (e) 90° clockwise rotation of (d) and dimension along the Z-axis. (f) Simulation of protein-membrane systems resulting in an optimal insertion of ESTA-6 tetramer as the gaps between ESAT-6 tetramer and membrane (seen in panel (c)) was refilled with lipids. Ions and water molecules are removed for clearer observation.

In conclusion, with a rigorous and comprehensive computational technique, we successfully predicted the best model of ESAT-6 oligomer that is the most energetically favorable and stable while interacting with the membrane. Based on these results and various observations from previous experiments, we hypothesize that after disassociated from CFP-10, ESAT-6 forms a cylindrical tetramer that can span a pore on the membrane, which may contribute to the virulence of tuberculosis. Future experiments would further verify this prediction.

2. Methods

Preparing the structure of ESAT-6 for symmetric docking

The structure of ESAT-6 is obtained from the complex structure of ESAT-6/CFP-10, which was downloaded from the Protein Data Bank (PDB ID: 1WA8).¹¹ We employed symmetrical docking algorithms including SymmDock and M-ZDOCK which predict the cyclically symmetric oligomers based on the given symmetric operators. No experiments have revealed the number of ESAT-6 monomers in an ESAT-6 oligomer. Therefore, in the docking programs, the symmetry operator was changed from C3 to C8, as the number of the involved monomer increased from 3 to 8. The N- and C-terminal arms of ESAT-6 are highly flexible. However, these docking algorithms treat proteins as rigid bodies and may discard those predictions in which these arms are overlapped, and those predictions could be the good candidates which fill our requirement. To avoid such situations both flexible arms were truncated prior to performing protein docking algorithms, Fig. 2(a) as an example. The arms were assembled back using Chimera,²⁷ after the docking and some post docking processes, and simulations were performed to obtain the best structures for these flexible terminals, Fig. 2(b). The truncated flexible N-terminal arm includes residues 1 to 10 and C-terminal arm contains residues 79 to 95.

Screening the predicted structure

Biology information can improve the docking prediction significantly.^28,29 Therefore, some experimental observations are utilized in this stage to improve the oligomer structure predictions. The SymmDock predicted 3624, 4271, 3020, 2502, 1808, 1504 structures for symmetries C3, C4, C5, C6, C7, and C8, respectively. On the other hand, M-ZDOCK predicted 1500 structures of oligomers for each cyclic symmetry from C3 to C8. For screening of the most trustworthy model from the predicted structures of oligomers, we used two criteria: (1) like most of the transmembrane helixes, the oligomer should be cylindrical; (2) the residues Ser-35 and Ala-60 should face towards the membrane based on the experimental determination,⁵ Fig. 2(a). Chimera was used for the visualization and the screening process. Furthermore, a clustering analysis was then performed based on the root mean squared deviation (RMSD) of the predicted structures from docking algorithms. RMSD trajectory tool from VMD was utilized to perform the clustering process. Two structures with RMSD less than 10 Å are considered to be in the same cluster. Here, 10 Å was chosen as the threshold based on the standard of CAPRI,³⁰ a protein–protein docking competition. After the screening and clustering, for each cyclic symmetry, the best structure from SymmDock and that from the M-ZDOCK are selected, resulting in 12 most representative predictions with the number of involved monomers increasing from 3 to 8. The truncated flexible N- and C- arms were then reconstructed back to the final 12 oligomer structures by aligning the full-length structure with the truncated structure using Chimera (Figs. 1, 2(a) and 2(b)). The error in these reconstruction processes introduced by such alignment was minor and would diminish during the following energy minimization and simulation.

Preparation of the system for MD Simulation

The interactions of the oligomers of different cyclic symmetry with the membrane were then studied using MD simulations. The bi-lipid membrane of type POPC (Phosphatidylcholine) with hydrated headgroups was generated using Charmm-27 force field³¹ in Visual Molecular Dynamics (VMD)³² with dimensions 200, 200 and 60 Å along X-, Y - and Z- axis, respectively. The predicted structures of oligomers were then embedded at the center of the membrane with the helixes inserted perpendicularly to the membrane along Z-axis (Figs. 2(c)–2(e)). The lipids residues located within 20 Å from Z-axis in the X-Y plane or within a distance less than 0.9 Å to the ESAT-6 oligomers were removed, resulting in a channel formed in the membrane. Due to the removal of lipids overlapped with the oligomers, certain gaps were generated between the outer surface of the oligomer and the membrane. However, such gaps will be reoccupied by lipids during MD simulation (Fig. 2(f)). Then the system was solvated with TIP3³³ water model and ionized with Sodium Chloride (NaCl) with concentration 0.15 mol/L to mimic the native environment. The dimension of the final system after the solvation and ionization was 200 Å, 200 Å, 100 Å along X-, Y - and Z-axis, respectively (Fig. 2(d)). The thickness of the water layer was kept more on the side of the membrane were the flexible arms of ESAT-6 are present to avoid the situation were during the simulation the flexible arms stretch out of the water box (Fig. 2(e)).

MD simulation

The simulations were performed with the MD simulation package NAMD 2.12.³⁴ The simulations were performed in four stages. In stage one, the water molecules, salt ions, the oligomers and the head groups of lipids were constrained in space while tails of the bi-lipid membrane were simulated to obtain the fluid-like bi-layer. The system was minimized for 20,000 steps followed by a 2 ns simulation. In the second stage, the constrained atoms in stage one were released. However, a harmonic constraint was applied on the protein so that the proper packing of protein with membrane could be achieved. The membrane shrank a little bit to fulfill the gaps which were created by the removal of the overlapped lipids within and around the proteins in previous steps of preparation of the system. For simulation, see Fig. 1(f). This stage has 20,000 steps of minimization and 1 ns of the simulation was performed at this stage. Besides, a force on the water molecules was applied to avoid them entering the hydrophobic regions of the membrane. This force is the user-defined force in NAMD which can be defined in the NAMD Tcl Forces modules.³⁴ In the third stage, the harmonic constraints on the protein as well as the force on water molecules are released. Instead, a restraint on the secondary structures of ESAT-6 oligomers was applied to preserve the secondary structures. This stage contains 5 ns of simulation and no steps of minimization as all the components are already minimized in the previous stages. After the system was fully equilibrated through three stages of simulation, the fourth stage is the production run where the system was simulated for 20 ns. The area of the membrane along the x-y plane of the membrane was kept constant in this stage as the well-equilibrated protein packed against the membrane is already achieved. The steps in the preparation of the system for simulations and the different stages of simulations were performed as guided by the Membrane Proteins Tutorial.³⁵

All 12 predicted oligomers were subjected to MD Simulation package NAMD version 2.12 in the same manner. The first stage of the simulation was performed with the NVT ensemble and the other stages were performed with the NPT ensemble. The temperature and pressure were kept 300 K and 1 atm, respectively. The periodic boundary conditions were applied to the solvation box. The long-range electrostatic interactions which are beyond the cutoffs were calculated with Particle Mesh Ewald³⁶ method. The cutoff was defined at 12 Å.

MM/PBSA Analysis

To quantify the binding free energy of the oligomers while inserted inside the membrane, MM/PBSA analysis was carried out. Two hundred snapshots were taken from the final 10 ns of the production run. Water molecules and salt ions were removed before conducting the MM/PBSA analysis. According to this method, the total binding free energy is given as

$$E=E_{\text{coul }}+E_{\text{vwd}}+E_{\text{pol}}+E_{\text{np}},$$ (1)

where E_coul and E_pol are Coulombic and Polar Solvation Energies which were calculated employing DelPhi.^37–39 For DelPhi calculations, the dielectric for protein and lipids was set as 2 and 80 for water. The protein and the lipids occupied 70% volume of the DelPhi calculation box. The concentration of salt was set as 0.15 mM and the resolution for the grid was set to be 2 grids/Å. The Van der Waals energy (E_vwd) was calculated with NAMD 2.12 subjecting the structures to 1 step of equilibration. The Non-polar Solvation Energy (E_np) was calculated based on the solvent accessible surface area (SASA), where SASA was calculated via NACCESS⁴⁰ and energy associated is given as

$$E_{\text{np}}=\alpha \cdot \text{SASA}+\beta ,$$ (2)

where α = 0054 and β = 0.92 kcal/mol.

The binding free energies of the oligomers inside the membranes were calculated with the formula

$$\Delta E_{\text{binding }}=E_{PL}-E_L-E_{P1}-E_{P2}-E_{P3}-\cdots -E_{Pn},$$ (3)

where P is the Protein, L is the Lipid, n is the number of monomers in the corresponding oligomer, and P1, P2, P3,… , Pn, are the corresponding ESAT-6 monomers in the oligomer.

Salt bridge analysis

The salt bridges formed by differently charged residues at the binding interface of the neighboring ESAT-6 monomers play significant roles in stabilizing the structure of the oligomers inside the membrane. The salt bridge analysis was performed using VMD, with a cutoff at 4 Å between the Oxygen and Nitrogen atoms of the oppositely charged residues, based on the final 10 ns simulation of production run. The long-range electrostatic interactions were also studied with a 10 Å cutoff.

Hydrogen bond (H-bond) analysis

The network of H-bond between the neighboring helixes in membrane proteins is important in their formation, structure, stability and functionality. Therefore, to observe the network of H-bonds among the helixes at the binding interfaces of the neighboring ESAT-6 monomer in the oligomers, H-bond analysis was performed with HBond Plugin in VMD. The final 10 ns simulation of the production run was subjected to the H-bonds analysis. The cutoff distance between the acceptor and donor atoms was set to 6 Å. Similarly, the cutoff angle between the receptor-hydrogen-donor was set to 20°.

Residues contacting the lipids

Based on the previous experiments, certain residues such as Ser-35 and Ala-60 are oriented towards the membrane.⁵ To verify this, the contacts of the protein residues with lipids were counted with threshold 3 Å based on the last 10 ns simulation of the production run. Herein, the contact between protein and lipid is defined as the distance between the atoms of the lipids and the protein atoms are less than 3 Å. Total contacts were counted for the respective residues and divided with the total number of frames and the number of monomers involved to calculate the contact made by individual residues in each frame (Fig. 5(b)).

Fig. 5.

H-bond formation and number of residues-lipid contacts in ESAT-6 tetramer. (a) The population of H-bonds on every binding interface of the ESAT-6 tetramer from M-ZDOCK, presented by a number of counts in every fame (count per frame) during the last 10ns simulation. (b) The occurrence of contact between lipid and residues Ser-35/Ala-60, presented count per frame during the last 10 ns simulation.

3. Results and Discussions

The calculation of binding free energy of the oligomers in the membrane with a different number of symmetries and from different docking algorithms is shown in Table 1. Based on binding free energy calculations, no matter which docking algorithm was used, the contribution of the single ESAT-6 to the total binding energy is lowest for the ESAT-6 tetramer (Table 1 and Fig. 3), suggesting that the ESAT-6 tetramer could be the best model for ESAT-6 oligomerization and membrane interaction. The ESAT-6 tetramer predicted by M-ZDOCK has the least normalized binding energy, which is −266.87 kcal/mol. Thus, in this work, the further investigations of ESAT-6 oligomerization are focused on the ESAT-6 tetramer from M-ZDOCK to identify the mechanisms responsible for making this model energetically favorable and most stable among all other forms of oligomers.

Table 1.

Binding energy profile of the oligomers from MM/PBSA analysis for oligomers with different symmetries.

		Polar solvation energy (kcal/mol)		Non-polar solvation energy (kcal/mol)		Van der waals energy (kcal/mol)		Coulombic energy (kcal/mol)		Total binding energy (kcal/mol)
Docking	#of ESAT-6	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Normalized binding energy (kcal/mol)
	3	117.25	67.9	−112.12	3.12	−790.87	22.95	191.66	69.39	−594.02	26.1	−198.01
	4	−95.51	39.89	−142.61	2.72	−1112.28	27.28	282.94	37.42	−1067.46	24.35	−266.87
	5	−636.3	141.22	−168.67	2.51	−1276.42	29.53	1032.95	155.33	−1048.44	29.44	−209.68
M-ZDOCK	6	−616.35	31.65	−172.62	2.21	−1328.07	26.05	822.92	36.8	−1294.11	25.43	−215.69
	7	−1592.07	103.29	−205.37	1.95	−1550.19	32.08	2071.43	106.96	−1276.22	34.61	−182.31
	8	−1147.99	42.12	−227.53	2.12	−1734.44	29.61	1468.71	43.36	−1641.25	29.5	−205.16
	3	249.34	45.61	−106.82	1.7	−814.53	18.8	−38.29	48.7	−710.3	20.64	−236.77
	4	−21.22	33.32	−134.81	1.65	−1048.01	19.11	178.79	39.18	−1025.25	20.28	−256.31
	5	−609.85	85.73	−161.27	3.27	−1220.23	35.3	997.58	88.98	−993.76	22.21	−198.75
SymmDock	6	−551.5	41.24	−190.49	2.22	−1448.27	36.33	810.56	39.06	−1379.71	41.71	−229.95
	7	−2090.36	86.41	−206.71	2.48	−1584.25	26.3	2580.17	89.81	−1301.13	28.36	−185.88
	8	−1158.56	47.56	−225.28	3.08	−1729.19	28.98	1487.39	49.46	−1625.64	28.52	−203.21

Fig. 3.

(Color online) Binding energy contribution of an individual ESAT-6 in oligomers with different symmetries. Blue and orange lines indicate normalized binding energies from M-ZDOCK and SymmDock respectively corresponding to Table 1.

Identification of key residues from the salt bridge and long-range Electrostatic interactions analysis

From the salt bridge analysis with cutoff 4 Å, we identified two salt bridges ASP59 A – LYS38 D and GLU49 D – LYS38 C (Table 2). The letters A, C, D following the residue number indicate the monomer ID in ESAT-6 tetramer as indicated in Figs. 4(a) and 4(b). The residue LYS38 is associated with both the salt bridges, and these salt bridges are not formed coherently but are exclusive to each other (Figs. 4(a), 4(b), 4(e) and 4(f)).

Table 2.

Salt bridges (4 Å)/long-range electrostatic interactions (10 Å) and their cyclic occurrence.

Cutoff	Salt bridge/long-range interaction	Total	%
	ASP59 A – LYS38 D	631	31.55
4 Å	GLU49 D – LYS38 C	170	8.8
	GLU631 C – LYS57 D	497	24.85
	ASP59 A – LYS38 D*	1895	94.75
	ASP59 B – LYS38 A*	222	11.10
10 Å	GLU12 C – ARG74 D	1294	64.7
	GLU49 D – LYS38 C	1354	67.7
	GLU64 C – LYS38 B**	232	11.6
	GLU64 A – LYS38 D**	1017	50.85
	GLU64 B – LYS38 A**	207	10.35

Note:

^*repetition of ASP59 – LYS38,

^**repetition of GLU64 – LYS38.

Fig. 4.

Example of relative motion between the ESAT-6 in oligomer and salt bridges with a cutoff of 4 Å. (a) Formation of salt bridge GLU49 D – LYS38 C when monomer D is close to monomer C. (b) Formation of salt bridge ASP59 A – LYS38 D when monomer D is close to monomer A. (c) and (d) Magnified interfaces of the salt bridges corresponding to (a) and (b). (e) and (f) Population of salt bridges corresponding to salt bridge GLU49 D – LYS38 C, and ASP59 A – LYS38 D, respectively. (g) Measures of bond length salt bridges while monomer D swings between monomer C and A. Black dashed line is the cutoff line. Red and yellow lines are the regions where salt bridge GLU49 D – LYS38 C and ASP59 A – LYS38 D is observed and are corresponding to (e) and (f).

The salt bridge ASP59 A – LYS38 D mainly exists in the second half of the last 10 ns simulation of production run (Fig. 4(f)) whereas the salt bridge GLU49 D – LYS38 C mainly exists in the first half (Fig. 4(e)). Suggesting there is a switch between these two salt bridges during the simulation. To understand the mechanism behind it, the bond lengths of these salt bridges were measured (Fig. 4(g)). The ESAT-6 (D) swung towards the ESAT-6 (A) in the second half of the simulation (Figs. 4(a) and 4(b)), and as a result, the relative distance between ASP59 A and LYS38 D decreased and the distance between GLU49 D and LYS38 C increased (Fig. 4(g)). Therefore, the formation of salt bridge is more favorable for ASP59 A and LYS38 D when compared to the GLU49 D and LYS38 C in the second half, which is the latest stage of the MD simulation, suggesting that the salt bridge ASP59 A – LYS38 D may be more stable than GLU49 D – LYS38 C. This prediction is consistent with their occurrence (Table 2).

Though the ESAT-6 tetramer has a C4 symmetry resulted from the docking process (Figs. 1(a) and 1(b)), the symmetric patterns were not observed in the formation of salt bridges (Table 2). For example, the salt bridge ASP59 – LYS38 is observed only in the interface of ESAT-6 (A) and ESAT-6 (D) but remains absent in another interface of ESAT-6 (between monomer A and B, B and C, C and D). Initially, the symmetry C4 from the docking had perfect symmetry but once the structure is subjected to MD simulation with membrane, the dynamics are random due to the random forces, which lead to the distortion in the symmetry. The swing of ESAT-6 (D) is an example and there could be other different possible motions like sliding up or down, shifting, rotation, etc. which explains the relative motions between the neighboring monomers and the absence of the cyclic salt bridges in each binding interface of the neighboring ESAT-6. The salt bridges were observed with a high occurrence at the cutoff 4 Å, which once again suggested that they are very strong and stable in the ESAT-6 tetramer. Furthermore, when the analyzing cutoff was increased to 10 Å to account for long-range electrostatic interaction, the similar oppositely charged residues pairs were found in different interfaces as denoted by * and ** in Table 2. The long-range electrostatic interactions found in all the binding interfaces between the ESAT-6 monomers, aids the stability of the ESAT-6 tetramer in the membrane. It is noteworthy that the residues LYS38 in all four monomers are involved in long-range electrostatic interaction with the neighboring GLU residues (Table 2, and Figs. 4(a))–4(d)). Also, LYS38 is presented in both the salt bridges observed for cutoff 4 Å suggesting that residue LYS38 is likely to play a key role in stabilizing the ESAT-6 tetramer, especially, when inserted in the membrane.

H-bonds

The H-bonds per frame for every binding interface of the ESAT-6 tetramer is depicted in Fig. 5(a). The presence of H-bonds in every interface indicates that the H-bonds have significant roles in stabilizing the tetramer inside the membrane. Nevertheless, the interface BC has the least H-bonds per frame. Table 2 shows that the residues pairs at the interface AB and BC have the least occurrences compared to others. In the interface AB, there are two residue pairs: ASP59 B-LYS38 A and GLU64 B - LYS38 A and has greater H-bond per frame compared to the interface BC, Fig. 5(a). Furthermore, the interface BC has only one residue pair, i.e. GLU64 C – LYS38 B with less occurrence and has the least H-bonds per frame and no salt bridges with cutoff 4 Å. So, the interface BC can be considered as the weakest interface, where the breaking of tetramer can be initiated.

Conformation of Ala-60 and Ser-35 facing membrane

Based on the experiments performed by Yue Ma et al., the residues Ala-60 and Ser-35 face towards the membrane. This observation indicates that the lipids of the membrane and the residues should be in close vicinity. In this work, we defined a contact formed by these residues with lipid residues as mentioned in the method part and quantified their number of contacts per frame (Fig. 5(b)). Consistent with the experimental observations, both residues Ala-60 and Ser-35 have close contacts with lipids. Where the residue Ser-35 has more significant contacts than residue Ala-60, it means that it has a stronger interaction with a membrane in comparison to residue Ala-60.

4. Conclusion

The interaction of the ESAT-6 with the membrane is considered as a key step for the virulence of tuberculosis. A deep understanding of the molecular mechanism of the interaction will be exceptionally helpful for developing the treatment of tuberculosis. In this work, using two different docking algorithms, 12 putative ESAT-6 oligomer structures were obtained, which were then subjected for a thorough analysis with several computational approaches. Experiments have observed that residues Ser-35 and Ala-60 are facing the lipid. Thus, during the structure prediction, such experimental observations have been utilized to improve the predictions. Among all predicted oligomers, the ESAT-6 tetramer, with C4 symmetry, was shown to have the lowest binding energy, suggesting that the ESAT-6 oligomerization is very likely to result in a tetramer, which functions as a transmembrane protein and creats pores on the membrane with the purpose of Mtb infection. To better understand the stability of the ESAT-6 tetramer and the interactions among the ESAT-6 monomers, the salt bridges, long-range electrostatic interactions and H-bonds were investigated and was shown to contribute significantly to its stability. Several key residues that form salt bridges and have long-range electrostatic interaction between two ESAT-6 monomers to stabilize the tetramer structure are identified, among which, the LYS38 is the most important one and served as a target of high potential for designing drugs against ESAT-6 oligomerization.

This work mainly predicted the ESAT-6 oligomer structure in the membrane. Then the stability of the oligomer structure was analyzed, and the key residues are identified which contribute to the stability significantly. It’s the first work on structural study on ESAT-6, oligomer which will pave the way for future studies on tuberculosis treatment. The approach of predicting the structure of the transmembrane oligomers can also be utilized in future studies for other transmembrane structures.