3.5 Trajectory analysis
3.5.1 Root-mean square deviation and root-mean square fluctuation
RMSD data show that each simulation equilibrated at different time points, except PPL-MAN, which equilibrated at around 40 ns, while the aforementioned PPL-MAN reached equilibrium quite fast at 10 ns, with values varying between 2.0 and 4.5 Å (Fig. 3A).
This suggests that the number of steps was adequate for the analysis and that the systems stably transitioned from an energy-minimized state to a fully solvated state. The average RMSD values for PBL, PBL-MAN, PPL and PPL-MAN are 3.56 Å, 3.08 Å, 3.13 Å, and 2.41 Å, respectively. These results demonstrate small non-significant deviations between bound and unbound PBL and PPL. They also indicate that the impact of binding to MAN is similar on both lectins in relation to the reduction of RMSD values. For PBL, presence of the ligand did not affect the time for the system to equilibrate, while for PPL, the complexed form equilibrated much faster than its unbound counterpart. Considering the similarity between these lectins, this is a surprising fact. Both lectins deviated less when complexed with MAN in comparison to their unbound form. It should be noted that this was only a slight change for PBL, whereas the variation was quite noticeable for PPL. Equilibration is highly variable protein- to-protein and simulation-to-simulation, but it is generally accepted that ligand binding usually helps in shifting the simulation to an equilibrated state in a smaller number of steps owing to CRD stabilization in a closed position, as observed by Nascimento and colleagues (2018) with lectins from Arachis [39] and Cavada et al. (2019) [40] for the lectins of Canavalia ensiformis and Canavalia brasiliensis.
Using theoretical B-factors, RMSF data (Fig. 3B) were applied to verify the flexibility of the proteins per residue during all simulations. As shown in the RMSF plot, fluctuation was dramatically different between PPL and PBL, both in unliganded and complexed forms. Overall, PBL residues fluctuated more than PPL residues in all cases. The binding with MAN also differentially affected both lectins. For PPL, the binding had very little effect on fluctuation; meanwhile, for PBL, the binding was very stabilizing, in particular the regions from 150–300 and 600–900, which include residues from PBL’s CRD.
The binding of lectins with their specific respective sugars tends to cause local stabilization with consequent reduction of fluctuation. This was observed for other legume lectins, such as Canavalia bonariensis [41], although exceptions can be pointed out, such as Arachis lectins for which specific ligand binding had no effect or increased fluctuation on specific residue ranges [39]. These results clearly show that very similar lectins can present dramatically different properties and that these differences are especially evident when dynamics properties are considered. The structures of Parkia lectins are dramatically different from those of other legume lectins, but their overall binding follows the same general rules.
3.5.2 Radius of gyration
Radius of gyration analysis evaluates the folding and compaction of the lectin. This analysis was applied in the current work to evaluate the differences caused by ligand binding. Results show that differences in RoG between lectins, both bound and unbound, were very small, suggesting that Parkia lectins do not change their compaction during binding. The average RoG values for PBL, PBL-MAN, PPL and PPL-MAN are 32.55 Å, 32.60 Å, 32.39 Å and 32.17 Å, respectively. The overall shallowness of Parkia lectin CRDs could explain these results. This property was also observed for the Dioclea lasiophylla lectin [19]. In this example, no significant changes in RoG could be observed after binding with monosaccharides or oligomannose N-glycans, suggesting that ligand binding is not necessarily a factor affecting the compaction of lectin structure. The energy x RoG graph (Fig. 4) demonstrates that PPL and PBL adopt conformations that vary between − 8500 and − 7250 kcal/mol, with little difference between their unbound and bound states.
3.5.3 H-bonds and contacts analysis
A few differences could be observed in the number of H-bonds between sites and lectins. Previous works have already reported differences in sites that could lead to variability in carbohydrate interactions [5] (Supplementary Fig. 1). Average H-bonds showed differences between lectins with PBL CRD 1 averaging 2.5 H-bonds with MAN, PBL CRD 2 averaging 4.5, and PBL CRD 3 averaging 3, while for PPL, averages were 2.5, 4 and 3, respectively. However, for all sites, one can see an overall stability with very few frames with 0 bonds. Also, at no point during the trajectory did any MAN leave the binding sites, which can happen, even when the ligand is specific to the lectin, as verified for Arachis lectins. Stability in interactions suggests a more favorable outcome.
Contact frequency analysis for PPL and PBL is depicted in Fig. 5. At first glance, it can be seen that the contacts with MAN are divided between very high contact residues ( > = 80%) and very low contact residues ( < = 20%), which are probably random interactions.
High contact residues include the following: For PBL CRD 1, Gly15, Gly16, Ile92, Gly136, Tyr137, Tyr138 and Asp140; for PBL CRD 2, Ala163, Gly164, Asp165, Phe238, Val240, Gly280, Gly281, Tyr282, Tyr283 and Asp285; for PBL CRD 3, Gly309, Gly310, Phe387, Gly430, Asp431, Tyr432 and Asp434. For PPL CRD 1, Gly15, Gly16, Val92, Gly136, Tyr137, Tyr138 and Asp140; for PPL CRD 2, Ala163, Gly164, Asp165, Phe239, Val241, Ser281, Gly282, Tyr283, Tyr284 and Asp286; for PPL CRD 3, Gly310, Gly311, Phe388, Gly431, Asp432, Tyr433 and Asp435. Despite small differences, the frequencies are very similar between the two lectins. When comparing this with the static analysis, qualitatively, almost all residues have been previously reported, but not their frequencies; we have now done this, and as a result, the CRDs of both Parkia lectins tested in the current work could be completely defined.
3.5.4 SASA analysis
SASA analysis of the systems was performed to understand the behavior of the protein solvent and evaluate the impact of interaction with carbohydrates on this behavior (Supplementary Fig. 2A-B). The solvent can interact with the protein by polar and nonpolar interactions, but the interaction with carbohydrates can change the conformation of the protein in a way that affects solvent accessibility. As expected, PBL and PPL undergo a reduction in SASA after interacting with ligands since carbohydrates interact with the surface of the molecule in the CRD cavity.
The average SASA values for the dimers of PBL and PBL-MAN are 35340 Å and 34005 Å, respectively (Supplementary Fig. 2C). According to the evaluation of each domain separately in the PBL system (Supplementary Fig. 3A-C), domain 1 presented an average value of 5327/5479 Å (monomer1/monomer2), domain 2 presented 6399/6193 Å, and domain 3 presented 6021/5919 Å. In the PBL-MAN system, the domains presented the following average values: domain 1 is 5291/5300 Å, domain 2 is 5968/5853 Å and domain 3 is 5761/5630 Å. As shown in Supplementary Fig. 4A-C, CRD1 (domain 1) had an average SASA value of 462/481 Å for PBL and 356/404 Å for PBL-MAN, CRD2 (domain 2) was 827/933 Å for PBL and 455/571 Å for PBL-MAN, and CRD3 (domain 3) was 748/472 Å for PBL and 547/281 Å for PBL-MAN.
For the PPL and PPL-MAN systems, the average SASA values for the dimers were 33493 Å and 32041 Å (Supplementary Fig. 2C), respectively. The average values of domains 1, 2 and 3 from the PPL system were 5428/5410 Å, 5432/5452 Å and 5809/5960 Å (Supplementary Fig. 3D-F), and their respective CRDs were 525/420 Å, 671/626 Å and 688/686 Å (Supplementary Fig. 4D-F). Regarding the PPL-MAN system, the average SASA value of domain 1 was 5191/5248 Å, domain 2 was 5123/5230 Å and domain 3 was 5565/5682 Å, while their respective CRDs were 361/262 Å, 461/462 Å and 551/544 Å.
PPL has a slightly smaller SASA area than PBL, as reflected at the dimer and domain level, but in domain 2, a more significant difference could be observed, even at the CRD level. In the unbound state, CRD2 of PBL shows a higher SASA area, followed by CRD3 and, with greater difference, CRD1. PPL, on the other hand, has less discrepant SASA values among its CRDs and relatively lower values than those of PBL. In the bound state with MAN, reductions in SASA values can be seen for both lectins, being more intense in PBL, specifically in CRD2. SASA values of both lectins in the bound state tend toward similarity.
Despite the high similarity between PBL and PPL, it was possible to observe that both molecules have structural differences in their unbound and MAN-bound state in relation to their global structure, which tend to decrease in their bound state at the CRD level. Some small differences in SASA values were observed between monomer 1 and monomer 2, which was more intense in CRD3 of PBL. These differences were not random since they were observed in the PBL and PBL-MAN systems, as well as in repetitions performed. Thus, the interaction between PBL monomers does not have the same impact on the corresponding domains of each protomer. This was not observed in PPL or in other evaluated lectins.
3.5.5 Binding free energy
The binding free energy of PPL and PBL with mannose molecules was calculated using MM/GBSA and MM/PBSA combined with Entropy Interaction (EI) for entropy calculation. The results obtained are presented in Table 2. The binding free energy of PBL is similar to that of PPL, but domain 3 of PBL binds more favorably than domain 3 of PPL. Despite this difference, both have a similar pattern of binding in that domain 1 has higher free energy, followed by domains 2 and 3. The energy decomposition analysis demonstrated the participation of each amino acid residue in the binding free energy (Figs. 6 and 7). For PPL, domain 1 residues Gly15, Gly16, Gly55, Gly56, Lys57, Val92, Val94, Arg134, Ala135, Gly136, Tyr137, Tyr138 and Leu139 contributed favorably to the binding energy, while Leu93 and Asp140 interfere energetically to the binding free energy. Domain 2 residues Ala163, Gly164, Gly204, Gly205, Gln206, Phe239, Val241, Lys280, Ser281, Gly282, Tyr283 and Tyr284 contribute favorably, while Glu240 and Asp286 interfere with the interaction. Domain 3 residues Gly310, Gly311, Gly350, Gly351, Val352, Phe388, Thr390, Arg429, Ala430, Gly431 and Tyr433 contribute to the interaction, while Asp312, Thr389 and Asp435 caused interference. It is interesting to note that some residues in contact with the ligand do not favor interaction, including Asp140, Asp286 and Asp435, all of which have a contact frequency of 100%. Other residues that have high frequency of contact with the ligand, but still do not favor interaction, include Gly55, Gly56, Val94, Lys57, Ala135, Leu139, Gly204, Gly205, Gln206, Lys280, Gly350, Gly351, Val352, Thr390, Arg429 and Ala430. Instead, these residues favor the creation of a hydrophobic or electrostatic environment that does favor interaction.
Table 2
Energy contributions for the interaction of PPL and PBL with D-mannose in kcal/mol.
MM/GBSA | |
PPL | PBL | |
Energy component | MN1 | MN2 | MN3 | MN1 | MN2 | MN3 |
ΔGgas | -97.70 ± 5.23 | -73.17 ± 4.98 | -65.00 ± 4.45 | -97.71 ± 5.22 | -71.10 ± 2.45 | -72.90 ± 5.02 |
ΔGsolv | 57.73 ± 2.73 | 43.52 ± 3.72 | 40.64 ± 4.14 | 57.73 ± 2.72 | 44.31 ± 2.27 | 43.38 ± 3.83 |
ΔGtotal | -39.97 ± 2.70 | -29.65 ± 2.73 | -24.35 ± 2.50 | -39.98 ± 2.70 | -26.79 ± 2.29 | -29.52 ± 2.70 |
Entropy | 9.94 | 8.45 | 6.43 | 9.89 | 3.08 | 8.22 |
ΔGfinal | -30.03 ± 2.70 | -21.20 ± 2.73 | -17.93 ± 2.50 | -30.09 ± 2.70 | -23.71 ± 2.29 | -21.30 ± 2.70 |
MM/PBSA | |
Energy component | MN1 | MN2 | MN3 | MN1 | MN2 | MN3 |
ΔGgas | -97.70 ± 5.23 | -73.17 ± 4.98 | -65.00 ± 4.45 | -97.71 ± 5.22 | -71.10 ± 2.45 | -72.90 ± 5.02 |
ΔGsolv | 67.24 ± 2.77 | 53.93 ± 4.18 | 56.49 ± 4.81 | 67.26 ± 2.76 | 53.18 ± 3.07 | 53.84 ± 4.26 |
ΔGtotal | -30.46 ± 3.49 | -19.24 ± 3.73 | -8.51 ± 3.29 | -30.45 ± 3.49 | -17.92 ± 2.79 | -19.06 ± 3.67 |
Entropy | 9.94 | 8.45 | 6.43 | 9.89 | 3.08 | 8.22 |
ΔGfinal | -20.52 ± 3.49 | -10.79 ± 3.73 | -2.08 ± 3.29 | -20.56 ± 3.49 | -14.84 ± 2.79 | -10.84 ± 3.67 |
For PBL, domain 1 residues that contributed favorably to binding were Gly15, Gly16, Gly55, Gly56, Lys57, Ile92, Val94, Arg134, Ala135, Gly136, Tyr137, Tyr138 and Asp140, while Asp17 and Leu93 interfere with binding. In domain 2. Ala 163, Gly164, Asp165, Gly203, Gly204, Phe238, Val240, Lys279, Gly280, Gly281, Tyr282 and Tyr283 favor binding, while Glu239 and Asp285 interfere with the interaction. Finally, in domain 3, Gly309, Gly310, Gly349, Gly350, Phe387, Thr389, Arg428, Ala429, Gly430, Asp431 and Tyr432 contribute to free binding energy, whereas Asp311, Thr388 and Asp434 interfere with interaction. Residues Gly56, Lys57, Arg134, Ala135, Gly203, Gly204, Lys279, Gly349, Gly350, Thr389, Arg428 and Ala429 lack high frequency of contact, but they still participate in the interaction. However, residues Asp285 and Asp434 also have a high frequency of contact, but they do not favor the binding energy.
Although the binding energies of PPL and PBL are similar in domain 1 and domain 2, it is possible to notice some differences in the contribution of some residues, mainly in the interference level of Asp140 (domain 1) and Asp286 (domain 2) in PPL, which is greater, but balanced by the level of contributions from favorable interactions to a more negative free energy. Domain 3 binding energy is much lower for PPL since Asp435 interference is 9.5x greater than that of Asp434 in PBL. Furthermore, the energetic contribution of Gly309, Gly310, Gly430, Asp431 and Tyr432 is higher in PBL than their equivalents in PPL.
The interference of Asp residues present in the domains is greater in PPL, possibly resulting from amino acid differences between their structures and the additional amino acid residue present in PPL structure, which changes the position and behavior of Asp residues, which disfavor interactions of PPL with mannose.
The information obtained from dynamics trajectories provided more details of the interaction of Parkia lectins with their specific ligands. As presented in this work, the level of structural data on the interaction with carbohydrates expanded the work of Bari et al. [15], who only evaluated static structures, resulting in an increase in the number of residues that compose, in part, the CRDs of these two lectins.