Nucleotide-binding oligomerization domain-containing protein 1 (NOD1) is an intracellular pattern recognition receptor that recognizes bacterial peptidoglycan (PG) containing meso-diaminopimelic acid (mesoDAP) and activates the innate immune system. Interestingly, a few pathogenic and commensal bacteria modify their PG stem peptide by amidation of mesoDAP (mesoDAPNH2). In the present study, NOD1 stimulation assays were performed using bacterial PG containing mesoDAP (PGDAP) and mesoDAPNH2 (PGDAPNH2) to understand the differences in their biomolecular recognition mechanism. PGDAP was effectively recognized, whereas PGDAPNH2 showed reduced recognition by the NOD1 receptor. Restimulation of the NOD1 receptor, which was initially stimulated with PGDAP using PGDAPNH2, did not show any further NOD1 activation levels than with PGDAP alone. But the NOD1 receptor initially stimulated with PGDAPNH2 responded effectively to restimulation with PGDAP. The biomolecular structure–recognition relationship of the ligand-sensing leucine-rich repeat (LRR) domain of human NOD1 (NOD1–LRR) with PGDAP and PGDAPNH2 was studied by different computational techniques to further understand the molecular basis of our experimental observations. The d-Glu–mesoDAP motif of GMTPDAP, which is the minimum essential motif for NOD1 activation, was found involved in specific interactions at the recognition site, but the interactions of the corresponding d-Glu–mesoDAP motif of PGDAPNH2 occur away from the recognition site of the NOD1 receptor. Hot-spot residues identified for effective PG recognition by NOD1–LRR include W820, G821, D826 and N850, which are evolutionarily conserved across different host species. These integrated results thus successfully provided the atomic level and biochemical insights on how PGs containing mesoDAPNH2 evade NOD1–LRR receptor recognition.
Nucleotide-binding oligomerization domain proteins-1 and -2 (NOD1 and NOD2) are two members of the ‘nucleotide-binding domain and leucine-rich repeat containing’ (NLR) protein family , which recognizes bacterial peptidoglycan (PG) ligands [2–4]. Both NOD1 and NOD2 receptors have a tripartite domain structure, with an amino-terminal caspase recruitment domain (CARD), a central nucleotide-binding domain (NBD) and carboxy-terminal domain of leucine-rich repeats (LRRs)  (Figure 1A). Binding of the PG fragments to the ligand-sensing LRR domains induces conformational changes and thereby exposes the central NBD. Oligomerization of the NBD leads to the exposure of the CARD, which in turn interacts with the CARD of receptor-interacting protein 2. Thus, NOD-mediated recognition of PG by ligand-sensing LRR domains triggers a downstream signaling pathway, leading to activation and nuclear translocation of nuclear factor κ-light-chain-enhancer of activated B cell (NF-κB) protein. This, in turn, activates several target genes inducing the innate immune responses by the secretion of pro-inflammatory cytokines, chemokines and antimicrobial peptides [6–8]. NOD1 and NOD2 receptors are believed to play a significant role in the host immune response of different types of bacterial infections, including Escherichia coli, Listeria monocytogenes, Shigella flexneri, Helicobacter pylori, Chlamydophila pneumoniae and Streptococcus pneumoniae . The biomolecular understanding of the structure–recognition relationship of microbial cell wall components, such as PG, and their innate and adaptive immune evasion strategies is very important [10,11].
PG, the ligand for host NOD1 and NOD2 receptors, is a conserved component of the bacterial cell wall. The structure of PG plays a critical role in its specific biomolecular recognition process by NOD1 and NOD2. Based on their stem peptide composition, different types of PG are recognized differentially by NOD receptors (NOD1 and NOD2). The disaccharide backbone of PG is made up of alternating subunits of N-acetylglucosamine (GlcNAc) linked to N-acetylmuramic acid (MurNAc) by a β-1,4-glycosidic bond. The MurNAc is lactyl-linked to the stem peptide which is made up of alternating l- and d-amino acids: l-alanine (l-Ala), d-glutamic acid (d-Glu), meso-diaminopimelic acid (mesoDAP) or l-lysine (l-Lys) and d-alanine (d-Ala). The vital role played by NOD1 and NOD2 receptors in the PG biomolecular recognition has been studied by different research groups using various biochemical techniques [12–15]. NOD1 receptor exclusively recognizes muramyl tripeptide containing mesoDAP (MTPDAP = MurNAc-l-Ala-d-Glu-mesoDAP), but not PG containing l-Lys. Therefore, all Gram-negative bacteria that possess PGDAP, including E. coli are detected by the NOD1 receptor [8,12], whereas NOD2, the general sensor for all bacterial PGs, recognizes muramyl dipeptide (MDP = MurNAc-l-Ala-d-Glu) and is a general sensor of PGs containing mesoDAP or l-Lys. PGs containing l-Lys, found in most Gram-positive bacteria, including Staphylococcus aureus and S. pneumoniae, only get recognized by the NOD2 receptor . Apart from the presence of mesoDAP and l-Lys, a few bacterial species, including Bacillus sp. , Mycobacterium sp. , Lactobacillus plantarum  and Leptospira sp. , contain amidated mesoDAP (PG containing mesoDAPNH2, PGDAPNH2), with an additional ε-carboxamide group. An earlier study  has shown that PGDAPNH2 from Bacillus subtilis was less recognized by the NOD1 receptor, and this observation is in good agreement with the present study. Previously, Ferwerda et al.  and Wang et al.  studied the role of human NOD1 and NOD2 receptors in the detection of mycobacterial PG fragments. Both studies reported reduced NOD1 recognition of mycobacterial PG fragments, suggesting that amidation of mesoDAP plays a crucial role in reduced recognition of PGDAPNH2 by NOD1. To explain this impaired NOD1 recognition of PGDAPNH2, we adopted an integrated approach of in vitro NOD1 stimulation assays and computational techniques for the comparative analysis of the interactions that lead to recognition of PGDAPNH2 and PGDAP by the ligand-sensing LRR domain of the NOD1 receptor.
Previous studies have demonstrated that NLR family proteins involve either their NBD or LRR domain to interact with bacterial ligands. For example, bacterial flagellin protein is recognized by the α-helical regions of the NBD of NAIP5 (NLR family, apoptosis inhibitory protein-5) and not by the LRR domain [22,23]. However, in the case of NOD1, the LRR domain is primarily involved in PG ligand recognition and deletion of the LRR domain resulted in a complete lack of NOD1 activation . Hence, a three-dimensional (3D) structure of this ligand-sensing domain, NOD1–LRR, would prove useful in ligand recognition studies. At present, the 3D structure of the LRR domains of human NOD1 is not available. So, the atomic-level structural details and molecular basis of the impaired recognition of the mesoDAPNH2-containing bacterial PGs by NOD1–LRR were not clearly understood.
The main objective of the present study was to understand how PGDAPNH2 is less recognized by the NOD1 receptor and its fundamental role in the innate immune evasion biomolecular sensing mechanisms using NOD1 stimulation assays and computational experiments. A homology model of the ligand-sensing LRR domain of the human NOD1–LRR receptor was developed, and binding of the bacterial PG monomer tripeptides containing mesoDAP (GMTPDAP = GlcNAc-MurNAc-l-Ala-d-Glu-mesoDAP) and its amidated form, mesoDAPNH2 (GMTPDAPNH2 = GlcNAc-MurNAc-l-Ala-d-Glu-mesoDAPNH2), was studied by different computational techniques. The contribution of the individual NOD1–LRR domains and its key residues involved in the atomic-level binding with the PG stem peptide was also explored. Thus, the present work highlights the differential recognition of the PGDAP versus mesoDAPNH2 by NOD1–LRR, and its biomolecular structure–recognition relationships were established using integrated experimental and computational techniques.
Reagents, bacterial strains and cell lines
Endotoxin-free MDP and MTPDAP were purchased from Invivogen, and PG monomer tripeptides, GMTPDAP (GlcNAc-MurNAc-l-ala-d-Glu-mesoDAP purified from E. coli) and GMTPDAPNH2 [GlcNAc-MurNAc-l-Ala-d-Glu-mesoDAPNH2 purified from Mycobacterium smegmatis] were obtained from CeCo Labs, Germany.
All bacterial strains, including E. coli (ATCC 25922), M. smegmatis strain mc2155 (ATCC700084) and B. subtilis (MTCC-121), were cultured in Luria Bertani (LB) broth at 37°C with 160 rpm shaking. Lp (ATCC10241) was grown in de Man–Rogosa–Sharpe broth under similar conditions . HEK-Blue-NOD1 cells [25,26], i.e. HEK293 cell lines co-transfected with plasmid pUNO1-hNOD1 carrying the human NOD1 gene and plasmid pDRIVE5-SEAP carrying the secreted embryonic alkaline phosphatase (SEAP) reporter gene, placed under the control of the IFN-β minimal promoter fused to five NF-κB and AP-1-binding sites (Supplementary Figure S1A), were purchased from Invivogen and maintained in Dulbecco's Modified Eagle Medium supplemented with 10% (v/v) fetal bovine serum, 50 U/ml penicillin, 50 μg/ml streptomycin, 30 µg/ml blasticidin and 100 µg/ml zeocin in a 5% CO2 atmosphere at 37°C.
Bacterial PG isolation and limulus amebocyte lysate assays
Bacterial cells from stationary phase cultures were harvested by centrifugation, boiled with 5% sodium dodecyl sulfate (SDS) for 30 min and were broken with glass beads. Insoluble oligomeric PG (OPG) was harvested by centrifugation and washed several times with hot water to remove SDS. Broken cell walls were suspended in PBS and treated with 10 µg/ml DNase, 50 µg/ml RNaseA and subsequently with 0.1 mg/ml trypsin for 16 h at 37°C. To remove cell wall-associated polysaccharides, the OPG preparations were incubated with 48% hydrofluoric acid (HFA) for 48 h at 4°C. For complete removal of HFA, OPG was harvested by centrifugation and washed several times with water. OPGs were further repeatedly washed with acetone to remove endotoxin contamination. OPGs were finally washed several times with endotoxin-free water and lyophilized. Purified lyophilized OPGs were suspended in endotoxin-free water at a concentration of 1 mg/ml and stored at −20°C [13,27,28]. To detect endotoxin contaminations in the OPG preparations, E-Toxate limulus amebocyte lysate assay kit (Sigma) was used, according to the manufacturer's instructions. OPGs with endotoxin levels below 0.03 EU/ml were used for NOD1 activation assays.
NOD1 activation assays using different PG ligands
HEK-Blue-NOD1 cells were seeded at a density of 5 × 104 cells per well in 96-well tissue culture plates. Different concentrations of MDP, MTPDAP, GMTPDAP, GMTPDAPNH2 (0, 0.001, 0.01, 0.1, 1 and 10 µg/ml) and insoluble OPGs (0, 0.01, 0.1, 1, 10, 25 and 50 µg/ml) from different bacteria, along with 25 µg/ml Polymyxin B [29,30], were added to the wells. The cells were incubated further for 24 h. At defined time intervals, 20 µl of the cell culture supernatants were collected and the presence of secreted SEAPs within supernatants were measured spectrophotometrically at 620 nm using the Quanti-Blue detection substrate (Invivogen), which turns from pink to purple/blue color in the presence of SEAP (Supplementary Figure S1A).
We also carried out PG-induced dose-dependent competition assays and restimulation assays. In PG-induced dose-dependent competition assays, a fixed concentration of one ligand was mixed with increasing concentrations of the other ligand and added to the cells. The ligands were mixed in different proportions as described below: (i) aliquots of 10 µg/ml oligomeric PG containing mesoDAP (OPGDAP) from E. coli were mixed with 0, 0.01, 0.1, 1, 10, 25 and 50 µg/ml of OPG containing mesoDAPNH2 (OPGDAPNH2) from M. smegmatis, L. plantarum and B. subtilis and vice versa, and (ii) aliquots of 0.1 µg/ml GMTPDAPNH2 were mixed with 0, 0.001, 0.01, 0.1, 1 and 10 µg/ml of GMTPDAP and vice versa. Cells were incubated with the PG ligands, along with 25 µg/ml Polymyxin B [29,30], for 24 h, and the amount of the secreted SEAPs in the cell culture supernatants was quantified as described above. In restimulation assays, sequential addition of the ligands was done, where cells were incubated with the first ligand (10 µg/ml OPGDAP from E. coli or OPGDAPNH2 from M. smegmatis, L. plantarum and B. subtilis or 0.1 µg/ml GMTPDAP or GMTPDAPNH2) for 6 h, following which the second ligand at the same concentration of the respective first ligand was added. In a few experiments, the cells were washed using PBS to remove any uninternalized first ligand prior to the addition of the second ligand. Cells were incubated with the PG ligands for an additional 24 h, and the amount of the secreted SEAPs in the cell culture supernatants was quantified as described above.
The descriptive statistical analysis was done using the GraphPad prism software (GraphPad Software, Inc., San Diego, CA). Results are expressed as means ± SD of three independent experiments. Statistical comparisons were performed using the unpaired Student's t-test. The groups between which the comparisons were performed are indicated under each figure legend. A value of P < 0.05 was considered significant.
Multiple sequence alignment (MSA), molecular modeling, molecular docking, molecular dynamic (MD) simulations and molecular electrostatic potential (MEP) mapping studies on human NOD1–LRR and its binding mechanisms toward bacterial PG monomer tripeptides were carried out on a Linux workstation using Clustal X, Schrödinger, AMBER and PyMol molecular modeling packages.
MSA and homology modeling of the ligand-sensing LRR domains of human NOD1
MSA of NOD1–LRR domains of human, mouse, rat, rabbit, rohu and zebrafish was performed using the Clustal X (v.1.83) program . The consensus ‘LxxLxLxxNxL’ motifs were identified from the MSA analysis of these domains.
A homology model of the LRR domain of human NOD1, involved in PG recognition, was developed for structure–recognition studies. The residues 676–941 of the human NOD1 sequence (GenBank Accession Number: AAH40339.1) comprising the LRR domain were used as the target sequence for building its homology model using BLASTp (basic local alignment search tool protein)  and Modeller9 v.9.0  programs. MSA of human NOD1–LRR with the sequences of two best templates, mouse ribonuclease inhibitor [protein databank (PDB) code: 3TSR] with a sequence identity of 28% and ribonuclease inhibitor–angiogenin complex (PDB code: 1A4Y) with a sequence identity of 29%, is obtained from BLASTp as shown in Supplementary Figure S2. Based on these template crystal structures, the NOD1–LRR structure was generated using the Modeller software and validated using Structural Analysis and Verification Server (SAVES) .
MD simulations of human NOD1–LRR and molecular docking simulations with GMTPDAP and GMTPDAPNH2
The human NOD1–LRR final model was subjected to MD simulation for a time period of 10 ns using the AMBER12 simulation software . Total potential energy information and trajectory co-ordinates [to measure the root-mean-square deviation (RMSD) of the protein backbone] were recorded at every 250 ps. Finally, this 3D model structure of human NOD1–LRR was used in the molecular docking simulations using the Grid-based Ligand Docking with Energetics (GLIDE) module of Schrödinger suite .
For docking study, 3D structures of GMTPDAP and GMTPDAPNH2 PG fragments were constructed. Initially, crystal structure of PG backbone, GM (N-acetylglucosamine-N-acetylmuramic acid; GlcNAc-MurNAc), was obtained from hen egg white lysozyme in complex with a trisaccharide MGM crystal structure having PDB identifier 9LYZ with a resolution of 2.5 Å . Furthermore, the crystal structure of MurNAc (or M) in complex with Mot-B-C protein having PDB identifier 3CYP  was taken for structural comparison in the M region of the GM moiety of the PG backbone, to add more confidence to the final generated PG structure (GMTPDAP or GMTPDAPNH2) in the present study. The next step involved adding the modeled stem peptide (l-ala-d-Glu-mesoDAP/mesoDAPNH2) to the GM crystal structure, and the complete structure of GMTPDAP and GMTPDAPNH2 was created using the ChemBioDraw software.
Using the Protein Preparation Wizard Panel in Schrödinger suite, the human NOD1–LRR model structure was preprocessed. A receptor grid box for docking simulations was generated using the receptor grid generation panel of the GLIDE module . The GLIDE score option was selected as the fitness function in order to understand the PG ligand (GMTPDAP and GMTPDAPNH2)-binding energy with human NOD1–LRR. Dimensions of the ligand diameter midpoint cubic box were set equal to 14 × 14 × 14 Å3 and docking ligand length was set to a maximum of 36 × 36 × 36 Å3, resulting in a grid box of radius 50 Å, which broadly covers the entire LRR regions of the human NOD1 receptor for computing its scoring grid for PG ligand recognition studies. The key amino acid residues occupy the cubic box, H788, K790, G792, E816, G818, W810 and W874, experimentally proved to be critical for PG ligand recognition . Furthermore, the docking simulations of NOD1–LRR with PG ligands were performed in a flexible and non-constrained manner, allowing the ligand to move freely over the entire volume of the grid box (100 Å). All other settings were set as default and docking simulations were performed in two steps that include standard precision (SP) followed by an extra precision (XP) algorithm to predict a more accurate binding mode in understanding biomolecular-level recognitions and interactions [15,36].
Molecular redocking of GMTPDAPNH2 with NOD1–LRR:GMTPDAP complex and GMTPDAP with NOD1–LRR:GMTPDAPNH2 complex systems
To justify the results of the restimulation assays, a new computational strategy was designed to understand GMTPDAP or GMTPDAPNH2 recognition by redocking these ligands to the docked complexes of NOD1–LRR, i.e. NOD1–LRR:GMTPDAPNH2 and NOD1–LRR:GMTPDAP. To perform redocking simulation experiments, the receptor cubic inner grid box size was increased from the default value of 14 × 14 × 14 Å3 to maximum 16 × 16 × 16 Å3 to accommodate both the docked and redocked ligands simultaneously, resulting in a grid box with 52 Å radius, occupying the entire NOD1–LRR regions. NOD1–LRR:GMTPDAP-docked complex was redocked with amidated ligand (GMTPDAPNH2) using a larger docking grid volume and the redocking simulations were performed in two steps which include SP docking followed by XP docking. Similarly, the NOD1–LRR:GMTPDAPNH2-docked complex system was redocked with the non-amidated ligand (GMTPDAP). Best poses of GMTPDAP and GMTPDAPNH2 with the highest docking score against docked complexes of NOD1–LRR:GMTPDAPNH2 and NOD1–LRR:GMTPDAP, respectively, were selected and analyzed using different molecular visualization and analysis programs for understanding its biomolecular-level recognition and interactions.
Molecular electrostatic surface potential mapping studies of docked and redocked complexes of NOD1–LRR:GMTPDAP and NOD1–LRR:GMTPDAPNH2 systems
The molecular surface of the NOD1–LRR receptor was generated using the Schrödinger program . We computed the MEP surface mapping of the NOD1–LRR receptor with solute dielectric constant of 1, solvent dielectric constant of 80 and solvent probe radius of 1.4 Å using an MEP module of Schrödinger suite. The range of MEP mapping was measured in the potential scale, with maximum positive potential (+10 kcal/mol) in red color and maximum negative potential (−10 kcal/mol) in blue color. The docked ligands (GMTPDAP and GMTPDAPNH2), as explained previously, shown in the ball and stick model were visualized along with an MEP surface map of the NOD1–LRR receptor to understand their structure–recognition relationships. A similar MEP protocol was adopted for redocked GMTPDAPNH2 and GMTPDAP ligands to analyze their structure–recognition relationships with their respective docked complexes (NOD1–LRR:GMTPDAP and NOD1–LRR:GMTPDAPNH2), as explained above.
Results and discussion
Impaired activation of the NOD1 receptor by OPG containing mesoDAPNH2
PG-induced NOD1 receptor activation led to the production of SEAP in HEK-Blue-NOD1 cells, which was measured using a Quanti-Blue substrate. We stimulated the HEK-Blue-NOD1 cells using OPGDAP from E. coli and OPG containing mesoDAPNH2 (OPGDAPNH2) from M. smegmatis, L. plantarum and B. subtilis, respectively. On a comparative analysis of OPGDAP- versus OPGDAPNH2-induced NOD1 activation levels, there was no significant difference observed at lower concentrations of 0.01 and 0.1 µg/ml (Figure 1B,C). However, a dose- and time-dependent increase in the NOD1 activation was observed when HEK-Blue-NOD1 cells were stimulated with 1, 10, 25 and 50 µg/ml OPGDAP from E. coli. These results further support the earlier findings of Girardin et al. . When HEK-Blue-NOD1 cells were treated with these same concentrations of OPGDAPNH2, the levels of NOD1 activation were significantly lower (Figure 1D–G). To confirm further that the amidation of mesoDAP is responsible for the lack of recognition of OPGDAPNH2 by the NOD1 receptor, we used OPGDAPNH2 from different bacterial species, namely M. smegmatis, L. plantarum and B. subtilis (Figure 1B–G). Furthermore, it was reported earlier that the OPGDAPNH2 from B. subtilis is poorly recognized by the NOD1 receptor . Also, Mycobacterium tuberculosis PG, which also contains mesoDAPNH2, was better recognized by the NOD2 and TLR2 receptors than by the NOD1 receptor [20,40]. Altogether, our activation bioassay results (Figure 1 and Supplementary Figure S2) confirmed that the OPGDAPNH2 were unable to produce effective NOD1 activation.
Dose-dependent competition assays reveal PGDAPNH2 to be a weak activator of NOD1 receptor
We performed a dose-dependent competition assay using increasing concentrations ranging from 0 to 50 µg/ml OPGDAP from E. coli in combination with a fixed concentration of 10 µg/ml OPGDAPNH2 and vice versa. A dose-dependent increase in NOD1 activation levels was observed with increasing concentrations of OPGDAP (Figure 2A). The same effect was observed when these same concentrations of OPGDAP from E. coli were mixed with 10 µg/ml OPGDAPNH2 from M. smegmatis (Figure 2B). However, though increasing concentrations of OPGDAPNH2 from M. smegmatis showed a slight increase in NOD1 activation, the increase was not dose-dependent (Figure 2C), in a manner comparable with OPGDAP (Figure 2A). When these concentrations of OPGDAPNH2 were mixed with 10 µg/ml OPGDAP from E. coli, they all showed the same NOD1 activation levels as the OPGDAP alone (Figure 2D). These results have been confirmed using OPGDAPNH2 from L. plantarum and B. subtilis as well (data not shown). Further validation of the results has been performed using HPLC-purified PG monomer tripeptides: GMTPDAP and GMTPDAPNH2 (Supplementary Figures S1B,C). These ligands were used for the NOD1 activation assay for two purposes: (i) OPGs are complex structures containing a mixture of PG fragments with varying lengths of stem peptides; (ii) PG tripeptides are better recognized by the NOD1 receptors than the OPGs. The monomer tripeptides also generated similar results (Figure 3). These results show that in both oligomeric forms (OPGDAP/DAPNH2) and monomer tripeptides (GMTPDAP/DAPNH2), PG ligands containing mesoDAPNH2 (PGDAPNH2), though capable of producing slight levels of NOD1 activation by themselves, do not appear to make any significant contribution when combined with PG ligands containing mesoDAP (PGDAP). We initially thought that this could be an effect of maximal activation achieved by OPGDAP, beyond which OPGDAPNH2 could not produce any further activation. But, it is clearly seen that even 0.1 µg/ml GMTPDAP does not show maximal NOD1 activation levels when compared with its higher concentrations (Figure 3A), and when combined with increasing concentrations of GMTPDAPNH2 does not show any difference compared with the NOD1 activation level produced by 0.1 µg/ml GMTPDAP alone (Figure 3D). So, in the absence of a synergistic or additive effect, we presume that in a scenario where both ligands are present, PGDAP either overcomes PGDAPNH2 or demonstrates some inhibitory action, which renders PGDAPNH2 unable to stimulate the cells. Another possibility was that PGDAP and PGDAPNH2 bind at different ligand-sensing sites on the LRR domain of the NOD1 receptor (NOD1–LRR), which may influence their stimulatory activity.
Dose-dependent competition assay on HEK-Blue-NOD1 cells with different concentrations of Ec and Ms.
Dose-dependent competition assay on HEK-Blue-NOD1 cells with different concentrations of GMTPDAP and GMTPDAPNH2.
Restimulation assays reveal PGDAP to be effective in restimulating the NOD1 receptor primarily stimulated with PGDAPNH2, while PGDAPNH2 does not contribute to any further activation of the NOD1 receptor, primarily stimulated with PGDAP
We performed restimulation assays, where HEK-Blue-NOD1 cells were exposed to PGs one after the other. The assays were performed in two ways: (i) with the sequential addition of the PGs and (ii) with the washing of the cells to remove any uninternalized extracellular primary PG, followed by an addition of the restimulant PG (Figure 4). In both these cases, it was observed that when cells were first stimulated with OPGDAP from E. coli, restimulation with OPGDAPNH2 from M. smegmatis was not able to enhance the NOD1 activation (Figure 4A). However, when cells were first stimulated with OPGDAPNH2 from Ms, the use of OPGDAP from E. coli as a restimulant showed enhanced NOD1 activation levels (Figure 4B). The same effect was observed with OPGDAPNH2 from L. plantarum and B. subtilis (data not shown). We further confirmed our observations using GMTPDAP and GMTPDAPNH2 ligands (Figure 4C,D). The NOD1 activation levels shown by cells where the washing step was employed appeared a bit low, compared with the cells where the sequential addition of PGs was performed. Moreover, this was not an effect of the removal of the first stimulant, but rather an effect of the decreased cell numbers due to loss of cells that occur during the washing steps, i.e. HEK-Blue-NOD1 cells are not very adherent in nature and are prone to get dislodged during the washing steps. It is important to note that only the PG that was not internalized during incubation with the primary PG stimulant will be removed during washing. The internalized primary PG stimulant will still be capable of contributing to the NOD1 activation.
Effect of PG restimulation on prestimulated NOD1 receptor.
Our results showed that the NOD1 receptor initially stimulated with PGDAP does not show any further activation when restimulated with PGDAPNH2 (Figure 4A,C). It was possible that the initial stimulus given by PGDAP induces maximum NOD1 activation, which was not further enhanced by PGDAPNH2 restimulation. But, we had observed higher levels of NOD1 activation in earlier experiments. This further supports that PGDAPNH2 does not make any significant contribution to NOD1 activation in a restimulation scenario.
Furthermore, when NOD1 receptor was initially stimulated with PGDAPNH2, the increased activation levels on restimulation with PGDAP can be explained as an effect of suboptimal activation by PGDAPNH2. This effect was enhanced by both the simultaneous presence of restimulant PGDAP and the addition of restimulant PGDAP following removal of uninternalized primary stimulant PGDAPNH2. The results of both dose-dependent competition assays and restimulation assays (with and without the washing step to remove uninternalized primary stimulant) show that the addition of PGDAPNH2 as a restimulant does not contribute to additional NOD1 activation; and the presence of the PGDAPNH2 as the primary stimulant or co-stimulant does not hinder NOD1 activation on restimulation or co-stimulation with PGDAP. However, even with the washing step (to remove extracellular uninternalized ligand) in our experiments, the primary ligand, which is already internalized, remains within the HEK-Blue-NOD1 cells and there is a possibility of the primary ligand being bound simultaneously to the NOD1 receptor, along with the restimulant ligand.
The experimental impaired activation of NOD1–LRR by GMTPDAPNH2 and the above-mentioned restimulation experimental proof of concepts were further confirmed using computational docking of GMTPDAPNH2 and GMTPDAP with NOD1–LRR, and redocking techniques to analyze the interactions of GMTPDAP and GMTPDAPNH2 with NOD1–LRR in complex with GMTPDAPNH2 and GMTPDAP, respectively.
Homology model of the ligand-sensing LRR domains of human NOD1 was constructed by advanced modeling, to reveal its biomolecular structure recognition toward PG monomer tripeptides
The NOD1 protein is known to have a tripartite domain structure, which includes CARD, NBD and a ligand-sensing carboxy-terminal domain of 10 leucine-rich repeats (NOD1–LRR1–LRR10)  (Figure 1A). The NOD1–LRR sequences from different host species, including human, mouse, rat, rabbit, rohu and zebrafish, were analyzed and characteristic consensus motifs ‘LxxLxLxxNxL’ were identified (Supplementary Figure S4) . Due to the non-availability of its 3D structure, we constructed the homology model of NOD1–LRR, consisting of LRR1–LRR10 regions of the human NOD1 receptor using the Advanced Modelling (Modeller9v9) program . The NOD1–LRR showed a right-handed curved solenoid 3D structure, and the amino acid residues critical for ligand recognition are in the center of the cleft of its concave surface .
Each repeat was found to have a characteristic consensus sequence ‘LxxLxLxxNxL’ and is composed of an inner β-sheet and an exposed α-helix, with interconnecting loops (loop-β-sheet-loop-α-helix-loop). In concordance with Girardin et al., our NOD1–LRR model showed LRR1–LRR10 (Figure 5A) . The sequence of NOD1–LRR with the consensus LRR motif of ‘LxxLxLxxNxL’ in each repeat and the regions of β-sheet and α-helix marked are shown in Figure 5B.
Initially, the NOD1–LRR model was subjected to iterative steps of model refinement, followed by rigorous evaluation after each step, to ensure maximum structural and stereochemical quality of the final models generated. The Ramachandran plots and ERRAT plots of the models demonstrate their stereochemical and structural quality, respectively. The final NOD1–LRR model showed an ERRAT structural quality factor of 90.7, with all the residues of the recognition site with error below the cutoff limit of 95% (Supplementary Figure S5A). The Ramachandran plot of the final NOD1–LRR model showed favorable statistics: 83% of the residues in most favored regions, 15.4% in the additionally allowed regions, 1.7% in the generously allowed regions and no residues in the disallowed regions (Supplementary Figure S5B).
To understand the mechanistic interaction of the NOD1 receptor and PG with mesoDAP or mesoDAPNH2, we performed MD and molecular docking simulations along with the MEP mapping studies of these complex systems. The NOD1–LRR system was found to be structurally and energetically stable during the entire 10 ns MD simulations, with an average RMSD of 1.55 Å (Supplementary Figure S5C) and potential energy of −100 050 kcal/mol (Supplementary Figure S5D).
PG monomer tripeptide GMTPDAP shows an energetically stable interaction with the residues in the recognition site region of NOD1–LRR
In an earlier site-directed mutagenesis study of different amino acids in the NOD1–LRR, the recognition site residues critical for ligand sensing were identified — H788, K790 and G792–LRR5; E816, G818 and W820–LRR6 and W874–LRR8 . These residues were at the center of the concavity of the NOD1–LRR model developed in the present study (Figure 5A,C). In a recent computational model of the NOD1–LRR domain of zebra fish, similar residues were predicted as mentioned above, contributing toward binding with the d-Glu-mesoDAP motif of PG .
Different molecular docking softwares (Glide, GOLD, FlexX and Autodock) were tested for docking of carbohydrate ligands to the crystal structure of antibodies by Agostino et al. , and Glide was predicted best among them. We also performed molecular docking of PG fragments with the model of the human NOD1–LRR receptor with the standard precision (SP) followed by an XP method, using the Glide software. We found that three poses per ligand (GMTPDAP) in the post-docking minimization and RMSD >0.5 Å are discarded. Finally, the best pose of GMTPDAP with the highest docking score against the NOD1–LRR was selected and analyzed for its molecular-level interactions.
GMTPDAP best pose showed energetically stable hydrogen bonding with an XP docking score of −8.82 kcal/mol. The NOD1–LRR residues, Y764–LRR4, K790–LRR5, G818, G821 and D826–LRR6, S846, S849 and N850–LRR7, were observed to interact with the GMTPDAP ligand. Most of these residues were in the recognition site region (LRR5–LRR8) of the NOD1–LRR receptor (Table 1 and Figure 6A). Furthermore, some of these residues contributing to the electrostatic and van der Waals interactions with GMTPDAP were also located in the LRR6 and LRR7 regions (Supplementary Table S1). It is experimentally known that the minimum essential motif for NOD1 stimulation is the d-Glu-mesoDAP motif of the PG stem peptide . Our docking simulations also predicted that this stem peptide motif of GMTPDAP (denoted by the yellow outlined box in Figure 6A) was anchored into the PG recognition site of NOD1–LRR with residues — W820, G821 and D826–LRR6, S849 and N850–LRR7, in agreement with the experimental results. The carbohydrate moiety (GM or GlcNAc-MurNAc) of the GMTPDAP also made a few weak interactions with the LRR4–LRR7 regions of the receptor.
|Ligand||Residue on ligand||Residue on NOD1–LRR||LRR region||Type of interactions|
|Ligand||Residue on ligand||Residue on NOD1–LRR||LRR region||Type of interactions|
The MEP map of NOD1–LRR surface docked to a GMTPDAP ligand (Figure 6B) showed a tunnel-like cavity in its ligand recognition site, through which the GMTPDAP ligand passed. Specifically, the d-Glu-mesoDAP motif of GMTPDAP was found to bind at the tunnel-like cavity at the ligand recognition site toward LRR7, as shown in Figure 6B. These stable interactions of GMTPDAP with the residues in the ligand recognition site of NOD1–LRR demonstrate that GMTPDAP was able to activate the NOD1 receptor effectively. Present computational results thus predict the NOD1 stimulatory activity of PGDAP, which is in excellent agreement with our experimental observations explained earlier.
Thus, analysis of the binding mechanism between human NOD1–LRR and GMTPDAP using docking and MEP mapping techniques revealed the following interesting insights: (i) recognition of GMTPDAP ligand was highly specific to the ligand recognition site of NOD1 (LRR5–LRR8), with the PG stem peptide, namely l-Ala-d-Glu-mesoDAP interacting broadly with residues from LRR4 to LRR7 regions ; (ii) l-Ala mainly interacted with W820 and G821–LRR6 and S850–LRR7 of NOD1–LRR by making C–H and hydrogen bond (H-bond)-type interactions (Table 1 and Figure 6A); (iii) d-Glu interacted with residues W820 and G821–LRR6 and N850–LRR7 by C–H and H-bond-type interactions; (iv) mesoDAP was found to bind with D826–LRR6 and S849 and N850–LRR7 regions by C–H and H-bond-type interactions; (v) the carbohydrate moieties, GlcNAc (G) interacted with S846–LRR7 by H-bond and MurNAc (M) interacted with Y764–LRR4, K790–LRR5, G818 and W820–LRR6 and S849–LRR7 by C–H and H-bond-type interactions, respectively (Table 1 and Figure 6A) and (vi) MEP surface analysis further provided confidence to our docking simulations and showed that the d-Glu-mesoDAP motif of GMTPDAP binds mainly at the tunnel region at the recognition site (LRR5–LRR7) of the NOD1–LRR (Figure 6B). The previous site-directed mutagenesis study revealed that residues belonging mainly to LRR5, LRR6 and LRR8 contribute to the NOD1 stimulatory activity. Furthermore, a perfect agreement was obtained by integrating the results of the mutagenesis experiment , with our PG stimulation assays and computational results, to explain an effective NOD1 recognition by PGDAP.
Interaction of amidated PG with human NOD1–LRR receptor deviates from its recognition site
Docking of GMTPDAPNH2 with the human NOD1–LRR ligand-sensing region (LRR5–LRR8) involved rigorous docking simulations to arrive at a correct pose, as explained earlier. Furthermore, a single pose per ligand was obtained from the post-docking minimization of GMTPDAPNH2 having a docking score of −9.38 kcal/mol against NOD1–LRR, which was further analyzed for its molecular-level interactions.
This analysis revealed a sophisticated biomolecular interaction network, which would further address the impaired recognition of PGDAPNH2 by the NOD1–LRR receptor. Based on the interacting residues of NOD1–LRR with GMTPDAP and GMTPDAPNH2 (Table 1 and Supplementary Table S1), it appeared that there was a partial overlap between the binding sites of the two ligands on NOD1–LRR. However, the interactions of the d-Glu-mesoDAPNH2 motif of GMTPDAPNH2 ligand with NOD1–LRR mostly deviated from the recognition site. The interactions of mesoDAPNH2 of GMTPDAPNH2 with NOD1–LRR were seen at LRR3–LRR4, away from its recognition site, as expected (Figure 6C,D and Table 1). The motif, d-Glu-mesoDAPNH2 (denoted by the green outlined box in Figure 6C) of GMTPDAPNH2, showed interactions with the surface residues Q739–LRR3, N765 and N766–LRR4, Q767 and K793–LRR5, which mostly occur away from the recognition site region at the NOD1–LRR surface, indicating its weak NOD1 stimulatory effect (Table 1). Furthermore, electrostatic and van der Waals interactions of GMTPDAPNH2 involved residues spanning the region of LRR1 to LRR7; of which, the recognition site residues at LRR5–LRR7 interacted with the GM disaccharide motif instead of stem peptide  and, hence, were not likely to influence NOD1 receptor recognition (Supplementary Table S1). In Figure 6, yellow and green boxes represent the d-Glu-mesoDAP motif of GMTPDAP and GMTPDAPNH2, respectively.
The MEP map of NOD1–LRR surface docked to GMTPDAPNH2 showed that it avoids entering the tunnel-like cavity at the d-Glu-mesoDAP motif recognition site (LRR5–LRR8), through which the GMTPDAP ligand made interactions . Instead, the d-Glu-mesoDAPNH2 motif of GMTPDAPNH2 moved away from this recognition site of NOD1–LRR, depicted in the MEP map (Figure 6D). This deviation from the recognition site further explains the impaired ability of PGDAPNH2 to stimulate the NOD1 receptor. The tunnel at the NOD1–LRR recognition site region is marked with a black-dotted circle for clarity in both the MEP maps (Figure 6C,D).
Computational validation of the restimulation experiments using the redocking simulation technique
We integrated the restimulation assay results (Figure 4) and our molecular docking analyses of NOD1–LRR with GMTPDAP versus GMTPDAPNH2 (Figure 6 and Table 1) to build a hypothesis and integrated it with a computational redocking approach to justify our experimental hypothesis. When HEK-Blue-NOD1 cells were stimulated with GMTPDAP and then restimulated with GMTPDAPNH2, we observed that the NOD1 activation was more or less the same, with respect to that with the GMTPDAP alone (Figure 4C). Here, we extended our computer simulations to hypothesize the above-mentioned biochemical phenomenon. With the molecular docking and MEP results suggestive of the recognition sites of the NOD1 receptor being occupied by the primary GMTPDAP ligand in the (NOD1–LRR:GMTPDAP) complex (Figure 6A,B), a restimulation with GMTPDAPNH2 would prove ineffective. However, when HEK-Blue-NOD1 cells were first stimulated with GMTPDAPNH2 and then restimulated with GMTPDAP, the NOD1 response was significantly higher, when compared with the response of GMTPDAPNH2 alone (Figure 4D). Our computational calculations based on docking results shown in Figure 6 were in excellent agreement with the above-mentioned phenomenon, so that the (NOD1–LRR:GMTPDAPNH2) interactions mostly occur away from the d-Glu-mesoDAP motif recognition site (Figure 6C,D), leaving this site freely accessible for restimulation. Thus, we hypothesized that in this scenario, restimulant GMTPDAP will be permitted to occupy the free d-Glu-mesoDAP recognition site of NOD1–LRR, broadly triggering an effective NOD1 response. This experimental proof of concept was further confirmed using a computational redocking strategy to analyze the interactions of GMTPDAP and GMTPDAPNH2 with the d-Glu-mesoDAP motif recognition site of complexes of NOD1–LRR initially docked with primary ligands [(NOD1–LRR:GMTPDAPNH2) or (NOD1–LRR:GMTPDAPNH2)], respectively.
In redocking of GMTPDAPNH2 with the ‘NOD1–LRR:GMTPDAP’ complex, we obtained an XP docking score of −8.38 kcal/mol. The predocked GMTPDAP was observed to lie at the d-Glu-mesoDAP motif recognition site of NOD1–LRR, and the d-Glu-mesoDAPNH2 motif of GMTPDAPNH2 used for redocking depicted interactions with D711 and N712–LRR2, R734 and S736–LRR3, Y760 and N765–LRR4, K790, K793 and K795–LRR5, mostly away from its mesoDAP recognition site (Table 2 and Figure 7A).
|Residue on GMTPDAPNH2||Residue on NOD1–LRR:GMTPDAP||LRR region||Type of interactions|
|Residue on GMTPDAPNH2||Residue on NOD1–LRR:GMTPDAP||LRR region||Type of interactions|
The MEP map at the molecular surface of NOD1–LRR showed that predocked GMTPDAP occupied the tunnel-like cavity at the d-Glu-mesoDAP motif recognition site (LRR5–LRR8), rendering the residues non-accessible to the redocked GMTPDAPNH2 ligand, which was found to bind toward the LRR3 region at its surface (Figure 7B). Additionally, electrostatic and van der Waals interactions occur mainly at the residues located in LRR1–LRR4, which again were deviated away from the d-Glu-mesoDAP motif recognition site of NOD1–LRR (Supplementary Table S2). This biomolecular conformational mechanism obtained by molecular docking and MEP mapping explains the non-stimulatory nature of the ‘NOD1–LRR:GMTPDAP’ complex on restimulation with GMTPDAPNH2.
Conversely, when the ‘NOD1–LRR:GMTPDAPNH2’ complex was redocked with the ligand GMTPDAP, it showed interactions with the residues spanning over LRR1–LRR7 regions, with an XP docking score of −7.36 kcal/mol. The interacting residues forming H-bonds were K681 and T683–LRR1, D709 and D711–LRR2, K793–LRR5, W820–LRR6 and S849–LRR7 (Figure 7C and Table 3). Among these, the d-Glu-mesoDAP motif of GMTPDAP interacted specifically with K793–LRR5, W820–LRR6 and S849–LRR7, which lie within the recognition site tunnel of NOD1–LRR (Figure 7D). Residues contributing to electrostatic interactions are in the LRR1–LRR7 regions, and van der Waals-type interacting residues are located in LRR1 (Supplementary Table S3). However, the residues away from the recognition site (K681, T683 and Y684–LRR1, D709 and D711–LRR2) were seen to interact with the GM (GlcNAc-MurNAc) disaccharide motif of GMTPDAP, and not the d-Glu-mesoDAP motif. Thus, these interactions will not affect the NOD1 recognition. On the other hand, d-Glu-mesoDAP motif of GMTPDAP showed specific interactions with the residues at the LRR5–LRR7 regions, which form part of the recognition site of ‘NOD1–LRR: GMTPDAPNH2’ complex.
|Residue on GMTPDAP||Residue on NOD1–LRR:GMTPDAPNH2||LRR region||Type of interaction|
|Residue on GMTPDAP||Residue on NOD1–LRR:GMTPDAPNH2||LRR region||Type of interaction|
Furthermore, the redocking simulation study reveals a possibility of steric hindrance, which forced the GMTPDAP to adopt overlapping conformation with respect to that of the GMTPDAPNH2-binding site, as shown in Figure 7C,D. The GMTPDAP best binding site is depicted in Figure 6A,B, which showed the proper recognition of the mesoDAP moiety to the NOD1–LRR activation site of interest. However, it is noted at this juncture that there is a clear possibility that GMTPDAP having higher NOD1 stimulation should have the affinity to dislodge GMTPDAPNH2 from its present overlapping binding site (Figure 7C,D). This complex biomolecular phenomenon would in turn affect its correct recognition site for GMTPDAP. To have better clarification of the present molecular recognition phenomenon, Figures 6 and 7 were modified to show PG recognition moiety regions with colored boxes.
The MEP map on the molecular surface of NOD1–LRR, predocked with GMTPDAPNH2, showed that the redocked GMTPDAP ligand, was found to position itself meticulously, so that the d-Glu-mesoDAP motif interacts at a recognition site region (LRR7), which is in close proximity with this motif recognition site surface tunnel of NOD1–LRR (Figure 7D). However, as mentioned above, there is a possibility of steric hindrance (or displacement) between the GMTPDAP and GMTPDAPNH2 ligands, forcing them to adopt overlapping conformations. This in turn prevents GMTPDAP from entering through the recognition tunnel (Figure 7D) obtained from our MEP studies for its proper stimulation. These computational results further confirm the biomolecular conformational binding mechanisms, to explain an effective response of the NOD1 receptor prestimulated with GMTPDAPNH2 to restimulation with GMTPDAP.
Pathogenic and commensal bacteria are reported to modify their PG to evade the host innate immune responses. Modifications of the PG glycan backbone by O-acetylation  or N-glycosylation  of muramic acid or by N-deacetylation  of GlcNAc help bacteria to resist the bacteriolytic activity of lysozyme. We demonstrated earlier by computational and experimental techniques that O-acetylation hinders the proper positioning of the PG glycan strands into the binding site cleft of the lysozyme and therefore helps S. aureus to gain resistance against lysozyme . Apart from the PG backbone modifications, the stem peptides of a few bacterial species are also modified by the amidation of mesoDAP [4,17,18]. Thus, from previous studies, we came to the broad conclusion that amidation affects the physicochemical properties of PG, since it removes the negative charge from the ε-carboxyl group of mesoDAP and thereby contributes to evading the NOD1 receptor biomolecular recognition. Studies on L. plantarum demonstrated that amidation of mesoDAP was essential for bacterial survival. Furthermore, mutant containing mesoDAP displayed severe growth deficiency and altered cell morphology in L. plantarum . Previous studies using B. subtilis PG also showed that PGDAPNH2 were less recognized by the NOD1 receptor . Our experimental results also demonstrated that NOD1 recognition of OPGDAPNH2 from M. smegmatis, L. plantarum and B. subtilis was impaired, when compared with OPGDAP from E. coli.
The present work shows the biomolecular recognition mechanism of PGDAP versus mesoDAPNH2 at the ligand-sensing LRR region of the human NOD1 receptor by integrating various biochemical and computational techniques. Experimentally, it was known that the minimum essential motif for NOD1–LRR receptor recognition was the d-Glu-mesoDAP motif of the GMTPDAP . Based on the homology modeling and MD simulation studies of human NOD1–LRR regions followed by docking analyses of interactions of GMTPDAP and GMTPDAPNH2 ligands with its recognition sites, we established a differential NOD1 recognition of PGDAP versus mesoDAPNH2.
In summary, using a combinatorial in vitro and in silico approach, the differential NOD1 recognition of PG ligands containing mesoDAP versus mesoDAPNH2 was established to understand the key structure–recognition relationships between NOD1–LRR and PG ligands: GMTPDAP is well recognized by the NOD1 receptor, and the d-Glu-mesoDAP motif interacts with the recognition site residues of NOD1–LRR obtained using molecular docking simulations and MEP results. We identified a few hot-spot residues for PG recognition in NOD1–LRR, which include W820, G821, N850, D826 and N850. These residues are evolutionarily conserved across different species (Supplementary Figure S3). Some of these residues were proved to be very critical for NOD1 activation using mutational analysis . However, NOD1 recognition of GMTPDAPNH2 was impaired, owing to the recognition of the d-Glu-mesoDAPNH2 motif away from the recognition site of the receptor revealed from our experimental and computational studies.
Furthermore, when the NOD1 receptor initially stimulated with GMTPDAP was restimulated with GMTPDAPNH2, we observed NOD1 activation to be more or less the same, as cells were stimulated by GMTPDAP alone. Conversely, when the NOD1 receptor primarily stimulated with GMTPDAPNH2 was restimulated with GMTPDAP, we observed a significantly higher NOD1 activation response, compared with that of GMTPDAPNH2 alone. To mimic this experimental work, we adopted a computational redocking strategy in which the primary stimulating ligand in complex with NOD1–LRR was redocked to the other restimulating ligand to understand its molecular-level recognitions on two complex systems. (i) The NOD1–LRR:GMTPDAP complex system was redocked to the GMTPDAPNH2 ligand; its d-Glu-mesoDAPNH2 motif interactions were seen away from the LRR recognition site, suggesting its inability to activate further the receptor. (ii) The NOD1–LRR:GMTPDAPNH2 complex system was redocked to the GMTPDAP ligand; its d-Glu-mesoDAP motif interactions were seen in the vicinity of the LRR receptor recognition site for its activation. Redocking experiments further revealed that there is a strong possibility of steric interactions between these overlapping PG ligands or higher affinity ligands can displace the lower affinity ligands at the LRR recognition site for its stimulation. Thus, these computational results further confirm our experimental observations and thereby establish the structure–recognition relationship of the NOD1–LRR receptor with PG ligands containing mesoDAP and mesoDAPNH2.
basic local alignment search tool protein
caspase recruitment domain
Grid-based ligand docking with energetics
molecular electrostatic potential
multiple sequence alignment
nuclear factor κ-light-chain-enhancer of activated B cells
nucleotide-binding domain and leucine-rich repeat
nucleotide-binding oligomerization domain protein-1
nucleotide-binding oligomerization domain protein-2
PG containing mesoDAP
PG containing mesoDAPNH2
Structural Analysis and Verification Server
sodium dodecyl sulfate
secreted embryonic alkaline phosphatase
S.V., R.B. and C.G.M. designed the study and wrote the paper. S.V. and A.V. performed all the in vitro NOD1 activation assays. S.V., A.B. and A.C.P. built the NOD1–LRR structure and performed all the in silico experiments. All authors analyzed the results and approved the final version of the manuscript.
This work was supported by a FAST track grant from the Department of Science and Technology [DST; SR/FT/LS-095/2008] to R.B. and S.V. is supported by INSPIRE fellowship [IF110641], DST. A.C.P. is supported by the Kerala State Council for Science, Technology and Environment (KSCSTE) with a Junior Research Fellowship [012-35/FSHP/2012/CSTE], India.
We thank the Amrita Center for Nanosciences and Molecular Medicine for infrastructural support.
The Authors declare that there are no competing interests associated with the manuscript.