Interactions of HLA-DR and Topoisomerase I Epitope Modulated Genetic Risk for Systemic Sclerosis

Bioinformatics

Dynamics behavior of HLA-DR complexed to Top1 peptide

Root-mean-square displacement (RMSD) calculations were performed to monitor the conformational stability in the overall MD simulation, as shown in Fig. 2. RMSD along simulated time was evaluated from the geometrical coordinates of protein backbone (N-Cα-C-O) with respect to those of the initial structure. For HLA-DRs without Top1, RMSDs for the binding cleft and the whole protein were plotted in Fig. 2A. In the first 2-ns, RMSD value rapidly reaches up to 2 Å for HLA-DR binding cleft and whole protein in every system, and consequently establishes dynamics equilibrium. For the complex form (Fig. 2B), HLA-DR/Top1 complex, HLA-DR, binding cleft and nonameric core sequence of Top1 peptide are separately considered by averaging 10 trajectories. Similar curves are observed for the complex and the HLA-DR protein (black and red in Fig. 2B), indicating that the principle dynamics in the system largely depend on HLA-DR part. The systems of HLA-DR/Top1 complex show the RMSD values of ~2–3 Å with ~0.3 Å of fluctuation. This is due to a large size of HLA-DR molecules composing two major domains, which are an antigenic binding cleft and the cleft-distal domain interacting with a CD4+ T-cell receptor. Structural flexibilities result from peptide flanking regions and cleft-distal domain while the binding cleft maintains the movement within ~0.2 Å. From the RMSDs of binding cleft (blue) and nonameric core sequence of Top1 (green), HLA-DR/Top1 complex and HLA-DR free form are likely stable after 50 ns and 55 ns, respectively. Besides, the distance between the centers of gravity of 9-mer core Top1 peptide and HLA binding cleft is also monitored. It can be seen from the Supplemental Fig. S1 that the distance of all complexes has no significant change after 55 ns. From RMSD and the distance plots, the equilibrium phase is appropriately considered after the simulation reaches 10-ns for HLA-DR/Top1 complex and HLA-DR free form. Several sections during equilibrium phase of HLA-DR and HLA-DR/Top1 systems are rechecked by the RMSD frequency. The high frequency of distribution displacement within 1 Å is properly examined to resolve the orientation for productive sampling, especially for HLA-DR binding cleft and Top1 peptide. The conformational selections within equilibrium production are investigated for binding interaction of HLA-DR/Top1 complexes and dynamical behavior of free HLA-DRs.

Figure 2
Figure 2

RMSD on time series of HLA-DRs. (A) A variety of HLA-DRs without and (B) with Top1 binding are separately plotted for RMSD of the backbone atoms for HLA-DR/Top1 complex (black), HLA-DR protein (red), binding cleft (blue) and nonameric core sequence of Top1 peptide (green). Distributions of RMSD distance during productive selection of 90–100 ns for HLA free form and complex form are collected as the relative frequency.

To consider flexibility of the binding cleft in cases of Top1 bound and unbound, the set of 250 snapshots is superimposed within their simulated systems. In Fig. 3A,B, the binding cleft consists of the α-chain (residues 5–80 shown in tan ribbon) and the β-chain (residues 5–93 in pink ribbon), while the cleft-distal domains are not displayed. Each chain is constructed by four antiparallel sheets and one long helix. Empty HLA-DR clefts are depicted in the first column (Fig. 3A), and the next column represents individual HLA-DR of the same row binding with Top1 (Fig. 3B). For the molecular set of HLA-DR types, the superimposition of empty clefts exhibits open shape and high flexibility, especially at the helix and turn/coil parts on α-chain. The binding cleft manifests lower flexibility at the turn and coil conformations on α-chain when binding with Top1 peptide due to an induced-fit mechanism. For HLA-DR types with ATASSc association, the core length (p1–p9) of Top1 peptide orderly lies on the binding cleft, while two ends extend out of the cleft with a free motion. Unsurprisingly, Top1 peptide shows the highest flexibility in HLA-DRB1*01:01 complex (non-associated case).

Figure 3
Figure 3

Comparison of HLA-DR binding clefts without and with Top1 peptide (cyan-blue multicolor) in columns (A and B), respectively. Column (C) represents distribution of the distance between α-chain and β-chain of the binding clefts. The distance distribution of HLA-DR in free form is shown as a red peak and that for the binding peptide is shown as a blue peak.

To support this finding, the distance between the two centers of gravity (CG) of α-helixes are measured to represent a cleft size of HLA-DR with and without Top1 peptide bound. Overview of average distance is estimated to be in a range of 21.8–23.4 Å without peptide and becomes somewhat shorten to 18.6–20.9 Å with peptide binding. The effect of induced-fit with Top1 peptide leads to two α-helixes approaching each other only found in HLA-DRs associated and suspected ATASSc. By a comparison within identical HLA-DR molecules, the difference in the average distances of the binding cleft with and without Top1 are indicated in the parentheses as follows: HLA-DRB1*08:02 (−3.3 Å), HLA-DRB1*11:01 (−3.5 Å), HLA-DRB1*11:04 (−3.4 Å), HLA-DRB5*01:02 (−1.3 Å) and HLA-DRB1*01:01 (−0.8 Å), where the negative values are referred to smaller distances of binding cleft upon Top1 binding, respectively. In Fig. 3C, the profile of normal distribution of distance is quite similar for almost all HLA-DRs that are a broad peak for free protein and a narrow peak shifting to the shorter distance for the Top1-bound cases. According to the high protein flexibility, this could explain why the peak is spread out in the free form of HLA-DRs. The binding clefts become more stable by complexation with Top1, which is illustrated by a narrow peak of the distance distribution. More oscillatory configuration and the peak of distance distribution shifting to the extended distance suggest the less-fit Top1 in non-ATASSc binding cleft for HLA-DRB1*01:01.

Binding free energy calculations

MM-PB(GB)SA and QM/MM-GBSA binding free energy calculations were applied on the HLA-DR/Top1 complexes to quantify the antigenic binding strength in diverse antigen presenting types using 250 equilibrated frames for each replica. Binding free energy (∆Gbind) based on MM-PB(GB)SA is contributed by enthalpy and entropic terms at a constant temperature (-T∆S). The enthalpy term contains molecular mechanics energy (ΔEMM), and the solvation term (∆Gsol). The latter one comprises of polar and nonpolar solvation energies. The energy components are enumerated in Table 1. In gas phase, the electrostatic (∆Eele) energy mainly contributes for the binding interaction between HLA-DR and Top1 peptide, which is over four times stronger than the van der Waals (∆EvdW) attraction. The most stable complex is HLA-DRB5*01:02/Top1 (the suspect system with MM energy of −906.9 kcal/mol), while three HLA-DRs with ATASSc-association systems show such ∆EMM values ranging from −733.4 to −715.6 kcal/mol. The non-associated ATASSc protein, HLA-DRB1*01:01, has the lowest protein-protein interaction according to the ∆EMM term of −630.7 kcal/mol. Cooperation of enthalpy and entropy invoked from binding free energy clearly exposes that Top1 peptide had strong binding with ATASSc-associated HLA-DRB1*08:02 (−52.7 kcal/mol), HLA-DRB1*11:01 (−47.0 kcal/mol), HLA-DRB1*11:04 (−47.8 kcal/mol), ATASSc-suspect HLA-DRB5*01:02 (−51.2 kcal/mol) and rather weak binding with ATASSc-unassociated HLA-DRB1*01:01 (−40.9 kcal/mol), according to MM-PBSA approach. MM-GBSA binding free energies reveal similar tendency as stated in Table 1. Surprisingly for ATASSc-suspect, the HLA-DRB5*01:02/Top1 has the tightest binding interaction among HLA-DRs studied here. However, the self-antigen is only slightly bound to non-associated with ATASSc type, HLA-DRB1*01:01, as hypothesized. From the obtained results, it can be implied that not only HLA-DRB1*15:02 is the known gene expression in Thai-SSc patients but the strong linkage disequilibrium of HLA-DRB5*01:02 also is related to SSc disease. Works on the clinical data to confirm the relation between the genes linkage disequilibrium and the severity of the SSc disease are in progress.

Table 1 The relative binding free energy and energy components (kcal/mol) for the five HLA-DR/Top1 complexes predicted by the MM-PB(GB)SA and QM/MM-GBSA methods.

To strengthen the validity of the calculated binding energies for HLA-DR/Top1 complexes, the higher-level computation was applied using QM/MM-GBSA method. In QM region, the nonameric core sequence of Top1 peptide was treated with SCC-DFTB implementation. The ∆Gbind(QM/MM-GBSA) confirmed that the major HLA-DR binding cleft interacts within the nine core residues (p1 to p9) of Top1 peptide. Nonameric residues, p1 to p9, have conserved interactions with the HLA-DR cleft of associated (in a range of −33.2 to −36.5 kcal/mol) and suspect ATASSc (−37.5 kcal/mol) types; in contrast, they are barely attracted on the non-associated allele at −21.2 kcal/mol. The relative binding free energies (∆∆Gbind) are calculated with regard to HLA-DRB1*01:01 as the lowest value of 0 kcal/mol, as shown in Table 1. Evidently, the other HLA-DR systems are justified into the same group with close relative binding energy. Explanation for HLA-DR/Top1 binding is going to be more enlightened by per-residue energy decomposition and hydrogen bonding across protein-protein interface.

Per-residue energy decomposition

According to the structural database, the peptide motifs complexed to HLA class II were generally believed to have the same burial pattern with the key residue contacts at p1, p4, p6 and p933. To determine this point, the structural superimposition over all studied systems taken from the last MD snapshot was carried out. By a comparison in the side chain directions, i.e. carbon beta (CB) atom in Fig. 4A, five HLA-DR molecules are predicted for possible key anchors of Top1 peptide. The CB atoms of p1, p4, p6 and p9 are likely to face the binding cleft.

Figure 4
Figure 4

Top1 peptide binding with various HLA-DR molecules. (A) A superimposition of representative Top1 in five MD systems showing a side view of HLA-DR binding cleft. The Top1 binding to HLA-DRB1*08:02 is shown in black, HLA-DRB1*11:01 in red, HLA-DRB1*11:04 in blue, HLA-DRB5*01:02 in green and HLA-DRB1*01:01 in magenta. The Top1 peptide is depicted by the backbone chain and CB atoms of the side chains as a stick and a ball, respectively. (B) Per-residue binding free energy of Top1 peptide for all five HLA-DRs.

The relative binding affinity of Top1 in various HLA-DR types is examined by per-residue energy decomposition using implicit solvent model. The total binding free energy excluding entropic contribution is plotted per-residue in Fig. 4B to illustrate the Top1 binding pattern. The binding free energy of individual amino acids within the core 9-mer is compared at identical position among all five systems. Energetic plot per residue shows similar tendency of Top1 binding interaction on diverse HLA-DR molecules (Fig. 4B). Phenylalanine at the p1 and p9 positions clearly shows a strong interaction with HLA-DR with the binding free energy of <−9 and −6 kcal/mol, respectively. On the other hand, the p9 phenylalanine exhibits a weaker binding (~−5 kcal/mol) to HLA-DRB1*01:01 insusceptible to ATASSc than the other antigenic presenter molecules. Decreasing interaction of non-association complex is shown for p5, p6 and p9 (Fig. 4B). The residues concerning HLA-DRs’ sub-pockets confer a significant effect to Top1 interaction; therefore, the residues within 5 Å of core 9-mer bound region of Top1 are focused in which the residues with total decomposition energy ≤−1 kcal/mol are compared in Supplemental Fig. S2. With energy subdivision, the total energy is decomposed into two terms, polar (∆Eele + ∆Gpolar) and nonpolar (∆EvdW + ∆Gnonpolar) contributions. Overall Top1’s residues are mainly stabilized by HLA-DR protein through nonpolar interactions.

In Fig. 5, energetic fingerprints of difference (ΔΔGresidue) are represented as color-coded surface models of various HLA-DR binding clefts interacting with Top1 in comparison to the HLA-DRB1*01:01/Top1 (({rm{Delta }}{G}_{residue}^{HLA-DR}-{rm{Delta }}{G}_{residue}^{HLA-DRB{1}^{ast }01:01})). The HLA-DRs equally interact to p1 by hydrophobic residue cluster comprising of F24α, A52α and F54α. A hydrophilic S53α on α-chain and V85β on β-chain of HLA-DRB1*01:01 exert tighter binding than HLA-DRB1*08:02, HLA-DRB1*11:04 and HLA-DRB5*01:02 which exhibit adversative representation for HLA-DRB1*11:01 surface. Pi-pi stacking and pi-alkyl interactions are main stabilizations for the p1(F) side chain. The p2(K) and p3(I) are maintained on the cleft supported by N82β and pi-alkyl interaction with F54α, respectively. For some HLA-DR cases, the glutamate group of p4(E) interacted competitively between a positive charge attraction of R71β side chain and a negative charge repulsion of neighboring D70β. Repulsive force on p4(E) is observed in HLA-DRB1*11:04 associated and HLA-DRB5*01:02 suspected ATASSc (Fig. 4B), whereas in the other cases the electrostatic repulsion is lessen owing to Q9α stabilization (Supplementary Fig. S2). The middle cleft region surrounding p4(E) for HLA-DRB1*08:02, HLA-DRB1*11:01, HLA-DRB1*11:04 and HLA-DRB5*01:02 have lower or a little bit better binding interaction than HLA-DRB1*01:01 as respectively presented in blue or light yellow colored surface in Fig. 5. Besides, auxiliary interactions come from the adjacent residues of p4(E) like p3(I) and p5(P). The HLA-DRB1*01:01 carries Q70β that can accommodate p4(E) without electrostatic repulsion. The pyrrolidine side chain, a nonpolar group, of p6(P) on Top1 is mostly buried within the pocket with a total free energy in the range from −5 to −4 kcal/mol, while only the HLA-DRB1*01:01 case shows the free energy around −2.7 kcal/mol (Fig. 4B). Energetic contributions for p6(P) are partially participated by the residues 62α, 13β and 30β depending on the protein type. One of polymorphic residues is G13β for HLA-DRB1*08:02, S13β for HLA-DRB1*11:01 and HLA-DRB1*11:04, Y13β for HLA-DRB5*01:02, and F13β for HLA-DRB1*01:01. Sharing contribution of 30β interaction between p6 and p9 is verified on the protein surface (Fig. 5). Residue 30 partially constructing the p9-pocket plays an important role in efficient p9 holding of each protein. This pocket consists of nonpolar or aromatic residues and could contribute to the p9(F) anchor interaction with strong binding level, reducing in the non-association case (Fig. 4B). By exerting comparable nonpolar/aromatic effects, Y30β is presented for three HLA-DRs associated ATASSc, which differed to the small size G30β for ATASSc suspect and the polar C30β for non-ATASSc association. This leads to elude p9(F)-interaction in HLA-DRB1*01:01. The remaining residues of Top1 (p5, p7 and p8) within HLA-DRB1*01:01 complex is estimated to have small binding interaction with similar trend of the other four HLA-DRs, as revealed in Fig. 4B. Overall energetic fingerprints normalized with HLA-DRB1*01:01/Top1 are displayed in yellow to red tone at the pocket surrounding p9 of four HLA-DRs, especially at M73α deep inside the pocket. Although the ATASSc-suspect system, HLA-DRB5*01:02/Top1 has quite unlike amino acids from ATASSc association type, it nevertheless presents a resembling energy contribution to them.

Figure 5
Figure 5

Energetic fingerprints of difference (ΔΔGresidue) in per-residue free energy decomposition values of HLA-DR/Top1 complexes compared to non-associated with ATASSc HLA-DRB1*01:01. Energy contribution of HLA-DR residues for Top1 binding is illustrated by surface coloring. By comparing to per-residue decomposition energy of HLA-DRB1*01:01, the stronger binding is colored from yellow to red gradient and in contrast the weaker binding is shown in blue gradient.

Hydrogen bonds across protein-protein interface

A significant stabilizing contribution for HLA-DR/Top1 recognition occurrences is by salt bridge (SB) and hydrogen bond (HB) interactions at the protein-protein interface as seen in the gas phase energetic components for total binding free energy (Table 1) that the electrostatic interaction is highly contributed for HLA-DR/Top1 binding. Stabilization by hydrogen bonding interactions is determined by the percentage of occupation based on the criteria for the distance of hydrogen bond donor and acceptor atoms (HD…A) of ≤3.5 Å and the D-H…A angle of >120°. HB occupations between Top1 and sub-pocket residues are exhibited by grid cells in Fig. 6. The hydrogen bond strengths are divided into 3 levels: low (10–39%), moderate (40–69%), and strong (70–100%) interactions represented by the gradient of greenish, bluish and reddish grid cells, respectively. The nonameric core peptide consisting of –FKIEPPGLF– is buried in the antigenic binding cleft, while the rest of the residues on –NH3+ and –COO termini stay away from the cleft.

Figure 6
Figure 6

Hydrogen bond (HB) interactions of HLA-DR/Top1 complexes. HBs between HLA-DR (vertical axis) and Top1 (horizontal axis) residues are presented by grid map. Atom types are specified by tleap AMBER conversion. HB strength is defined by label and color in the grid cells, while salt bridge (SB) interaction is framed by a black border.

A similar HB pattern across protein-protein interface is found in all complexes by a strong interaction between the backbone atoms of p1(F) to S53α and p2(K) to the N82β amide group. In HLA-DRB1*08:02/Top1 complex, the acidic side chain of p4(E) is repulsed by the negative charge from D70β; meanwhile, a salt bridge is found with the R71β guanidinium group. The charge-charge repulsion of D70β-p4(E) leads to a decreased HLA-DR recognition according to the per-residue energy decomposition (Fig. 4). However, the p4(E) backbone atoms form strong hydrogen bonds with the Q9α and/or N62α amide groups in the binding cleft which compensates Coulomb repulsion for p4(E) in case of HLA-DRB1*08:02, HLA-DRB1*11:01 and HLA-DRB1*01:01. To assist conservative flanking by neighboring p4(E), p5(P) are stabilized by R71β for HLA-DRB1*11:01, HLA-DRB1*11:04 and HLA-DRB5*01:02. Since the amino acids at 28β, 30β and 70β in the focused systems are distinct, the detected hydrogen bond strengths are different. The Y30β hydroxyl group in HLA-DRs with ATASSc-association strongly stabilizes p7(G), because it has a HB acceptor group and a longer side chain than the other two residues: G30β for HLA-DRB5*01:02 and C30β for HLA-DRB1*01:01, whose side chains are unable to form a HB. p7(G) is found to be weakly stabilized by H28β instead for the suspect system. Hydrogen bond networks on the N69α amide group and W61β are linked to the backbone of p7(G), p8(L) and p9(F). To examine the overall hydrogen bond interactions for different HLA-DR in recognizing Top1 peptide, the hydrogen bonds formed across the protein-protein interface in individual systems are counted and showed strong occupation ranging from 71 to 100%, as illustrated in Fig. 6 with the numbers of reddish cells of 7 (HLA-DRB1*08:02), 7 (HLA-DRB1*11:01), 7 (HLA-DRB1*11:04), 8 (HLA-DRB5*01:02) and 6 (HLA-DRB1*01:01). Interestingly, the hydrogen bond interaction between the peptide and N69α is absent in the case of the insusceptible HLA-DRB1*01:01, making the overall number of hydrogen bond pairs lowest among all systems. However, even though majority of Top1 from p1 to p9 are nonpolar residues, hydrogen bond and salt bridge interactions still appeared as additionally supporting interactions. Polar energy contribution is in agreement with strong hydrogen bond network and rather appeared in protein complexes of systematic ATASSc-suspect. Expansion beyond the nonameric core peptide at two sides mostly exhibits temporary occupation of hydrogen bonds with unexpectedly strong occupancy in a few bonds (Supplementary Fig. S3).

Water solvation effect for self-peptide binding

Distribution of water accessibility at HLA-DR/Top1 interface was examined by 3D-RISM-KH molecular theory of solvation35,36 with TIP3P model using the rism3d.snglpnt module. Based on like-dissolve-like principle, a hydrophilic protein surface should find a higher amount of water than hydrophobic zone. As already pointed out before, the p1(F), p3(I)/p4(E), p6(P) and p9(F) of Top1 peptide are anchored on HLA-DR binding cleft. Water molecules accessibly inserted between these putative anchor residues and their surrounding residues of HLA in the binding cleft likely affect protein-protein binding strength. To predict the hydration sites around the antigenic binding, the stripped snapshots of peptide-protein complexes are inspected by 3D-RISM model. In 3D maps, the oxygen atoms of water molecules (in red flakes) are represented by 3D isosurfaces with a value >3 of g(r) level (Fig. 7), while hydrogen atoms are not shown. Water occupation is illustrated within HLA-DR sub-pockets around the nonameric core of Top1 peptide. Binding surface of HLAs class II is widely known to consist of flat and shallow sub-pockets, making it easy to be accessible by water molecules. Among all sub-pockets for Top1 binding in Fig. 7, there are two rather deep sub-pockets supporting p1(F) and p9(F) anchors. Their phenyl ring is perfectly buried in the sub-pockets without water molecules accessibility. In contrast for non-associated ATASSc, HLA-DRB1*01:01 complex has a high possibility of accessible waters inside the p9-pocket to compete with the peptide-protein interaction that agrees well with a lesser binding affinity of p9 in Fig. 4B. This site of the ATASSc-insusceptible protein is constructed of specific hydrophilic residues C30β and S37β, which differed from the other HLA-DR molecules. Even though the residue 37 on the β-chain of HLA-DRB5*01:02 is asparagine, its side chain points away from p9 making the hydrophobic pocket. For the three ATASSc-susceptible proteins, the –OH tail of Y37β also aligns away from p9(F).

Figure 7
Figure 7

3D distribution function of water oxygen atoms in the cross-sectional plane of HLA-DR/Top1 with g(r) >3. HLA-DRs are sliced off at the binding cleft by a red plane perpendicular to a top view. A zoom-in of a side view represented by a tan surface is the sites specifically supporting p1 and p9 anchors. The core-nonameric residues of Top1 peptide embedded in the binding cleft is represented by a blue ribbon for the backbone and by a stick model for the side chains.

Articles You May Like

New discovery makes fast-charging, better performing lithium-ion batteries possible
Novel antibody may suppress HIV for up to four months
How inflammation causes gastric cancer
What makes a jellyfish?
Fossils found in museum drawer in Kenya belong to gigantic carnivore

Leave a Reply

Your email address will not be published. Required fields are marked *