Toll-like receptors (TLRs) play key roles in sensing wide array of microbial signatures and induction of innate immunity. TLR2 in fish resembles higher eukaryotes by sensing peptidoglycan (PGN) and lipoteichoic acid (LTA) of bacterial cell wall and zymosan of yeasts. However, in fish TLR2, no study yet describes the ligand binding motifs in the leucine rich repeat regions (LRRs) of the extracellular domain (ECD) and important amino acids in TLR2-TIR (toll/interleukin-1 receptor) domain that could be engaged in transmitting downstream signaling. We predicted these in a commercially important freshwater fish species rohu (
The innate immune response elicited by a variety of pattern recognition receptors (PRRs) is an immediate nonspecific and first line of defense of the host against invading various pathogens [
Toll-like receptor 2 (TLR2) was shown to be the principal mediator of macrophages activation. It functions as homodimer [
PGN is a highly complex structural component and an important derivative of both Gram-positive and Gram-negative bacterial cell wall, and is the target of the innate immune system [
Various studies on TLR2 have also been reported in zebrafish [
India is the major supplier of fish in the world and ranks 3rd in freshwater fish production (FAO). Among various freshwater fishes, rohu (
To elucidate the structural scaffold in rohu TLR2, we report the 3D-model of extracellular domain of rohu TLR2 along with its key domains that are predicted to be involved in recognizing PGN, LTA and zymosan, and the critical region of interaction between TIR domains of TLR2 and MyD88. This is the first report across the fish species.
Rohu TLR2 protein (GenBank ID: ADQ74644) with N-terminal extracellular domain (ECD), transmembrane domain, and C-terminal cytoplasmic TIR domain [
Amino acid sequence of rohu TLR2 was aligned by MegAlign [
Template search for TLR2-ECD (561 aa), TLR2-TIR (146 aa), and MyD88-TIR (137 aa) domains in PDB database identified mouse TLR1-TLR2 heterodimer (PDB ID: 2Z81), TIR domain of human TLR2 (PDB ID: 1O77), and TIR domain of human MyD88 (PDB ID: 2Z5V) as the best homologous structures with top identity score. To ascertain the sensitivity and accuracy of the selected templates, FUGUE [
Molecular dynamics (MD) simulations were carried out for the modeled systems using the GROMACS 4.5.5 program [
The final snapshot obtained at the end of each MDS was considered to represent the structures of the TLR2-ECD, TLR2-TIR, and MyD88-TIR models. These simulated models were set for validation by SAVES (
Three different 2D structures of PGN [
Rohu and common carp (
The best protein-ligand complexes obtained from docking studies of PGN, LTA, and zymosan with TLR2-ECD were subjected to MDS using the previously defined parameters in GROMACS. To gain insight into the structural stability of the protein-ligand and protein-protein complexes, MD simulations were performed for PGN-I-TLR2-ECD, PGN-II-TLR2-ECD, PGN-DAP-TLR2-ECD, LTA-TLR2-ECD, zymosan-TLR2-ECD, and TLR2-TIR-MyD88-TIR complex for different time periods of MDS. A production MD run for 10 ns was carried out for TLR2-ECD ligand complexes and protein-protein complex. The existence of H-bonds in the complex in different periods of MDS was analyzed.
To identify the key amino acids among interacting amino acid residues in TLR2-ECD, TLR2-TIR, and MyD88-TIR domains, site-directed mutagenesis was carried out in DS 2.5 under build mutant protocol. Redocking was performed to calculate the fitness score in GOLD after mutation, and docking scores were cross-checked with previous fitness scores. Protein-protein interaction hot spots were predicted after mutagenesis by HADDOCK.
The full length TLR2 protein is constituted of 792 amino acids including a signal peptide of 30 amino acids (1–30 aa). The mature TLR2 protein ECD, trans-membrane (TM) and TIR domain constituted of 34–590, 595–612, and 645–790 amino acids respectively. The alignment of TLR2 amino acids with other species revealed their good conservation across the species (See Figure S1 in supplementary material available online at doi:
Multiple sequence alignment and secondary structure prediction of TLR2-TIR domain. (a) Multiple sequence alignment of TLR2-TIR domain of rohu with others by MegAlign program. Conserved residues were shown in yellow. Consensus residues are shown in the majority axis. (b) Secondary structure representation of TLR2-TIR domain by PSIPRED. Helices denoted as “H,” beta strands as “E,” and loops as “C.”
Multiple sequence alignment and secondary structure prediction of MyD88-TIR domain. (a) Multiple sequence alignment of MyD88-TIR domain of common carp with others by MegAlign program. Conserved residues were shown in yellow. Consensus residues are shown in the majority axis. (b) Secondary structure representation of MyD88-TIR domain by PSIPRED. Helices denoted as “H,” beta strands as “E,” and loops as “C.”
The BLAST search analysis showed that the ligand recognizing LRR regions in TLR2-ECD shared the close structural relationship with mouse TLR1-TLR2 heterodimer (PDB ID: 2Z81) having 35% and 52% sequence identities and similarities, respectively. The TLR2-TIR and MyD88-TIR domains shared 71% and 78% sequence identities with their respective templates (PDB ID: 1O77 and 2Z5V). The sequence-structure alignment by FUGUE revealed a good conservation of secondary structures (
Sequence identities between rohu TLR2-ECD (target) and mouse TLR2-ECD (template).
LRR | Identity (%) | LRR | Identity (%) |
---|---|---|---|
1 | 47.82 | 12 | 27.77 |
2 | 29.16 | 13 | 20 |
3 | 63.63 | 14 | 33.33 |
4 | 52 | 15 | 37.93 |
5 | 34.46 | 16 | 34.78 |
6 | 35.57 | 17 | 50 |
7 | 9.52 | 18 | 52.38 |
8 | 15.38 | 19 | 50 |
9 | 28.57 | 20 | 40.9 |
10 | 26.31 | 21 | 29.16 |
11 | 28 |
Sequence identities between rohu TLR2-TIR (target) and human TLR2-TIR (template).
Position | Identity (%) | Position | Identity (%) |
---|---|---|---|
|
100 | CD loop | 46.66 |
AA loop | 66.67 |
|
60 |
|
58.33 | DD loop | 69.23 |
|
42.85 |
|
88.88 |
BB Loop | 100 | DE loop | 50 |
|
90.9 |
|
100 |
|
75 | EE loop | 50 |
|
69.23 |
|
50 |
Sequence identities between common carp MyD88-TIR (target) and human MyD88-TIR (template).
Position | Identity (%) | Position | Identity (%) |
---|---|---|---|
|
100 | CC loop | 66.66 |
AA loop | 80 |
|
60 |
|
69.23 | DD loop | 70.58 |
|
100 |
|
100 |
BB Loop | 86.66 | EE loop | 66.66 |
|
100 |
|
68.75 |
|
85.71 |
The stability and MD properties were observed up to 40 ns for TLR2-ECD and up to 20 ns for TLR2-TIR and MyD88-TIR domains, and the RMSD values over time were shown in Figure
Root mean square deviation (RMSD) analysis. RMSD of (a) TLR2-ECD up to 40 ns and (b) MyD88-TIR and TLR2-TIR domains up to 20 ns MD simulation.
Root mean square fluctuation (RMSF) analysis for homology models. RMSF per residue over the dynamics was shown in graph. (a) TLR2-ECD; (b) TLR2-TIR; (c) MyD88-TIR.
The PROCHECK analysis at SAVES of three models (TLR2-ECD, TLR2-TIR, and MyD88-TIR) showed that the phi-psi angles of most of the residues were in the allowed regions of Ramachandran plot (Figure S2). The SAVES results (Table
Validation report for TLR2-ECD, TLR2-TIR, and MyD88-TIR homology models.
Validation by SAVES server
Ramachandran plot (PROCHECK) | TLR2-ECD | TLR2-TIR | MyD88-TIR |
---|---|---|---|
Residue (%) | Residue (%) | Residue (%) | |
Most favored regions | 66.7 | 78.7 | 73.2 |
Additionally allowed regions | 30.0 | 20.6 | 24.4 |
Generously allowed regions | 1.8 | 0.0 | 1.6 |
Disallowed regions | 1.6 | 0.7 | 0.8 |
Verify3D score | 95.37 | 97.28 | 87.6 |
ERRAT | 61.059 | 86.364 | 86.325 |
PROVE (mean |
1.609 | 1.48 | 1.63 |
Stereochemical quality of homology models by ProQ, ModFOLD, and MetaMQAP server
TLR2-ECD | TLR2-TIR | MyD88-TIR | |
---|---|---|---|
ProQ (LG/MX) | 7.062/0.432 | 6.401/0.772 | 7.067/0.847 |
ModFOLD (Q/P) | 0.6326/0.0065 | 0.7473/0.00038 | 0.5787/0.0022 |
MetaMQAP (GDT/RMSD) | 68.93/2.137 | 78.767/1.523 | 72.628/2.319 |
For docking analysis, the predicted top seven (B1 to B7) probable ligand binding pockets in TLR2-ECD (close to the N-glycosylation sites) were considered (Table S2) including previously reported LTA binding site in mouse TLR2 [
Molecular interaction of rohu TLR2-ECD with ligands.
Docking analysis of TLR2-ECD with PGN, LTA, and zymosan by AutoDock 4.0
Grid centre and ligand | Interacting residues | Binding energy kcal/mol | No. of H-bondsa |
---|---|---|---|
PGN-I (B5) | Tyr366, Thr395, Asn397, Ser399, Tyr421, Asn423, Ser425, His426, Asn446, Ser448, Ser449, Asp467, Glu470, Thr497 | −4.29 | 6 |
PGN-I (B6) | Ser259, Ser285, Tyr286, His312, Thr313, Arg342, Ser344, Tyr366 | −4.33 | 6 |
PGN-DAP (B5) | Asp368, Leu369, Ser370, Gln371, Asn397, Ser399, Gln400, Tyr421, Asp423, Ser425, His426, Asn446, Ser448, Ser449, Asp467, Ser469, Glu470, Thr489 | −3.19 | 11 |
PGN-DAP (B6) | Thr246, Glu254, Gly255, Lys258, Leu276, Thr277, Met279, Asp280, Gly281, Ser282, Ser283, Leu284, Ser304, Tyr305, Thr306, His307, Tyr308, Glu309 | −3.38 | 7 |
LTA sites | Glu323, Phe324, Phe325, Glu326, Met330, Met331, Phe335, Thr349, Val350, Phe351, Val352, Ile353, Pro354, Pro355, Ile356, Leu360, Asn372, Leu373, Leu374, Pro381 | −1.92 | 1 |
Zymosan | Leu473, Thr474, Val475, Phe476, Asn477, Thr495, Leu496, Pro497, His498, Gly499, Glu500, Leu501, Ser520, Ser521, Asp522, Arg525 | −7.55 | 13 |
Docking analysis of TLR2-ECD with PGN, LTA, and zymosan by FlexX 2.1
Ligands | Interacting residues | Binding energy kcal/mol | No. of H-Bondsa |
---|---|---|---|
PGN-I | Asn397, Ser399, Gln400, Asn401, His426, Ser428, Lys451, Glu470 | −18.85 | 18 |
PGN-II | Ser428, Phe429, Val430, Ser448, Lys451, Arg453, Lys454, Asp472, Ser469, Gln470 | −12.80 | 14 |
PGN-DAP | Gln400, Asn401, His426, Ser428, Ser449, Lys451 | −15.47 | 15 |
LTA | Asn318, Leu319, Asp320, Ile321, Phe324, Asn347, Gly348, Thr349, Val350, Gln371 | −6.92 | 13 |
Zymosan | Arg492, Leu493, Met494, Leu496, Arg516, Met517, Ser520, Asp522 | −13.81 | 10 |
Docking analysis of TLR2-ECD with PGN, LTA, and zymosan by GOLD 4.1
Ligands | Interacting residues | GOLD fitness score | No. of H-Bondsa |
---|---|---|---|
PGN-I (B5) | Ile394, Asn397, Ser399, Gln400, Tyr421, Asp423, Ser425, His426, Asn446, Ser448, Ser449, Asp467, Ser469, Glu470, Thr489, Gly490, Glu511, Arg512 | 42.38 | 17 |
PGN-II (B5) | Tyr366, Asp368, Ser370, Gln371, Asn397, Ser399, Gln400, Tyr421, Ser425, Asn446, Ser448, Val465, Asp467, Ser469, Glu470, Thr489, Gly490, Arg512 | 44.01 | 20 |
PGN-DAP (B5) | Asn446, Ser448, Ser449, Asp467, Ser469, Glu470, Ile487, Thr489, Gly490, Gln511, Arg512 | 40.55 | 13 |
PGN-II (B6) | Thr246, Glu247, Pro248, Phe249, Lys250, Thr252, Thr277, Asp280, Ser304, Tyr305, Thr306, His307, Tyr308, | 23.00 | 8 |
LTA | Leu319, Asp320, Ile321, Phe324, Phe327, Met330, Met331, Phe335, Gly348, Thr349, Val350, Phe351, Glu380, Pro381 | 44.65 | 4 |
Zymosan | Arg492, Leu493, Met494, Thr495, Leu496, Ala514, Leu515, Arg516, Met517, Phe518, Asn519, Ser520, Ser521, Asp522, Arg525 | 39.71 | 8 |
Illustration of the interaction of PGN, LTA, and zymosan with the modeled 3D structure of rohu TLR2-ECD by AutoDock 4.0 program. At B5 region, interaction of PGN-I with TLR2-ECD (a) and PGN-DAP with TLR2-ECD (b); at B6 region, interaction of PGN-I with TLR2-ECD (c) and PGN-DAP with TLR2-ECD (d); interaction of LTA with TLR2-ECD (e) and zymosan with TLR2-ECD (f). The TLR2-ECD was shown in ribbon and ligands (PGN, LTA, and zymosan) were shown in solid form. Amino acid number depicted in the figure was shown as per the matured protein (after removal of the signal peptide).
In FlexX docking, the binding sites B3 and B4 revealed high positive FlexX score for all ligands and, hence, excluded from further studies. The B5 site (LRR16–19) resulted in good FlexX score −18.85, −12.80, and −15.47 for PGN-I, PGN-II, and PGN-DAP, respectively. Docking at B6 site (LRR8-10) also resulted a satisfactory FlexX score for PGN-II (−13.23) and comparatively a low FlexX score was predicted for PGN-I and PGN-DAP. The rest of the binding pockets were found to be irrelevant for PGN interaction with very little positive and negative FlexX score. The interaction of LTA with TLR2-ECD with highest FlexX score (−6.92) was obtained at Bmouse region (LRR12–14). Interaction of LTA at other binding sites yielded irrelevant FlexX score. Zymosan interacted with TLR2-ECD at B7 (LRR20-CT) region effectively with a FlexX score of (−13.81), and other binding sites were found to be very less interactive. The FlexX scores at different binding regions were presented in Table
The GOLD scores for PGN-I, PGN-II, and PGN-DAP at B5 and B6 sites, for LTA at Bmouse and for zymosan at B7 site were given in Table
Illustration of the interaction of PGN, LTA, and zymosan with the modeled 3D structure of rohu TLR2-ECD by GOLD 4.1 program. Interaction of (a) PGN-I and TLR2-ECD at LRR16-19; (b) PGN-II and TLR2-ECD at LRR16-19; (c) PGN-DAP and TLR2-ECD at LRR16-19; (d) PGN-II and TLR2-ECD at LRR8-10; (e) zymosan and TLR2-ECD at LRR20-CT; (f) LTA and TLR2-ECD at LRR12-14. Rohu TLR2-ECD model was shown as line and ligands (PGN, LTA, and zymosan) are shown as stick.
The MyD88 functions as an adaptor molecule that transmits signal to downstream molecules from ligand activated TLRs by interacting with the TIR domains. The predicted interface residues in cons-PPISP, InterProSurf, and PatchDock were observed to be present in the most flexible regions of TLR2-TIR and MyD88-TIR domains (Table S3). Majority of the interacting amino acids were distributed in BB loop,
List of hydrogen bond forming and hydrophobic interacting residues between TLR2-TIR and MyD88-TIR domains.
Hydrogen bonds | Hydrophobic interactions | |||
---|---|---|---|---|
Donor | Acceptor | Length (Å) | Donor | Acceptor |
Trp689-NE1 | Leu84-O | 1.97 | Ala73 | Gln654 |
Ser715-O | Ser50-N | 2.34 | Gln77 | His655 |
Ser715-O | Ile48-O | 2.84 | Tyr8 | Asp656 |
Ser715-OG | Asp36-OD2 | 2.78 | Ser71 | His680 |
Asp723-OD2 | Lys55-NZ | 2.70 | Asp72 | Lys681 |
Thr714-N | Asp36-OD2 | 2.62 | Cys9 | Arg682 |
Thr714-O | Asp36-O | 2.72 | Phe76 | Phe684 |
Ser83-OG | Gly687-N | 2.94 | Lys79 | Pro686 |
Gln654-NE2 | Asp38-OD2 | 2.72 | Leu82 | Gly687 |
Ser11-SG | Asp656-OD1 | 3.15 | Ser83 | Trp689 |
Cys9-NZ | Asp656-OD1 | 2.70 | Phe80 | Ile690 |
Asp72-N | Asp656-OD2 | 2.56 | Leu84 | Glu710 |
Cys9-NZ | His655-O | 2.61 | Cys85 | His711 |
Cys9-NZ | His655-ND1 | 2.78 | Pro86 | Val713 |
Pro686-O | Ser50-OG | 2.14 | Asp36 | Thr714 |
Arg688-NH2 | Glu51-OE2 | 2.75 | Ala73 | Gln654 |
Asp692-OD1 | Lys10-NZ | 2.20 | Gln77 | His655 |
Lys681-NZ | Asp72-OD2 | 2.73 | Tyr8 | Asp656 |
His655-NE2 | Asp75-OD2 | 1.91 | Ser71 | His680 |
Interaction of TLR2-TIR and MyD88-TIR in Discovery Studio 2.5. (a) TLR2-TIR model is labeled as chain “A” and MyD88-TIR model is labeled as chain “B.” Interface residues are shown inside a rectangle box in ball and stick representation. (b) Clustering of interface residues between TLR2-TIR and MyD88-TIR domain in tree format. Residues of TLR2-TIR are marked as chain “A” and MyD88-TIR as chain “B.” Strongly interacting residues are highlighted with different colors.
Previously it was reported that binding site B5 (LRR16–19) (corresponding to human) had the highest affinity for PGN recognition, and binding site B6 (LRR8–10) had a low potential for PGN recognition [
Hydrogen bond analysis of protein-ligand complexes. H-bond analysis of (a) PGN-I and TLR2-ECD complex, (b) PGN-II and TLR2-ECD complex, (c) PGN-DAP and TLR2-ECD complex, (d) PGN-II and TLR2-ECD complex at LRR8-10 region (B6), (e) LTA and TLR2-ECD complex at LRR20-CT, (f) and zymosan and TLR2-ECD complex at LRR12-14.
Alanine scanning of B5 regions of PGN binding domains showed loss of interaction due to absence of donors or acceptors in the active sites. But no single mutation of alanine for the interacting residues at this region showed a complete loss of docking. Proline and aspartic acid scanning of B5 region also ensued very low fitness score (9.6 and 12.8) in GOLD. Analysis of B5 region by mutating residues in pair, triplet, and quadruple combinations at a time indicated the viable importance of Asp394, Ser396, Asn397, and Ser399 as the fitness score attained a very minimum value in comparison to other GOLD runs.
Mutagenesis study at B7 residues in TLR2-ECD that formed H-bond with zymosan revealed a good fitness score for alanine scanning. However, proline scanning of all residues revealed loss of zymosan interaction with TLR2-ECD. Single proline mutation and acidic to basic mutation of different interacting residues followed by individual GOLD runs explored the importance of Ser520 and Asp522. A single mutation of Ser520-Pro520 and Asp522-Gln522 resulted in complete loss of zymosan interaction. Mutation of other residues altered the GOLD fitness score; however, in none of the cases docking loss was ensued. Alanine and proline scanning of LTA binding site resulted in low docking score emphasizing the role of key residues Asp320 and Phe324 in LTA recognition.
The proposed 3D model of rohu TLR2 describes the protein features and its important domains. It also accentuates the importance of predicting key amino acids and LRR regions that are responsible for the specific ligand interaction and TLR2 signaling in fish and depicts a residue-detailed structural theoretical model. In the absence of crystal structure of TLR2 in any fish, this study provides structural insight into the TLR2 domains architecture. In rohu fish, the peptides at LRR16–19 (at B5), LRR12-14 (at Bmouse) and LRR20-CT (at B7) are predicted to be the most important interacting regions for PGN, LTA, and zymosan interactions, respectively. The structural organization of TIR domains in fish TLR2 and adapter molecule MyD88 have also been described. The interaction between TLR2-TIR and MyD88-TIR domain highlighted the contribution of BB loop,
This work was financially supported by the grant of National Agricultural Innovation Project (NAIP), Indian Council of Agricultural Research (ICAR) (Project Code C4-C30018). The authors express their gratitude to Dr. A. Bandyopadhyay and Dr. S. Kochar, National Coordinator, NAIP-Comp-4, for their valuable suggestions and help. The authors would like to express their deepest gratitude to Dr. G. C. Sahoo, HOD, Bioinformatics Division, RMRIMS, Patna, India, for his valuable suggestion. Part of the work related to Discovery Studio, FlexX, and GOLD was carried out at RMRIMS, Patna.The authors would like to thank Mr. Roman Laskowski for providing the academic license of Ligplot. They thank Mr. Mahesh Patra (Research Associate) and Mr. Sukanta Kumar Pradhan, HOD, Department of Bioinformatics, Orissa University of Agriculture and Technology, Bhubaneswar, Odisha, India, for their support and constructive suggestions in data analysis. They gratefully acknowledge the Bioinformatics Resources and Applications Facility (BRAF) at the Center for Development of Advanced Computing (C-DAC), Pune, India.