International Journal of
Molecular Sciences Article
Interaction of Nevirapine with the Peptide Binding Groove of HLA-DRB1*01:01 and Its Effect on the Conformation of HLA-Peptide Complex Makoto Hirasawa 1, *, Katsunobu Hagihara 2 Noriaki Hirayama 3 ID 1
2 3
*
ID
, Koji Abe 1 , Osamu Ando 1 and
Drug Metabolism & Pharmacokinetics Research Laboratories, Daiichi Sankyo Co., Ltd., 1-2-58 Hiromachi, Shinagawa-ku, Tokyo 140-8710, Japan;
[email protected] (K.A.);
[email protected] (O.A.) Biomarker Department, Daiichi Sankyo Co., Ltd., 1-2-58 Hiromachi, Shinagawa-ku, Tokyo 140-8710, Japan;
[email protected] Institute of Advanced Biosciences, Tokai University, 4-1-1 Kitakaname, Hiratsuka-shi, Kanagawa 259-1292, Japan;
[email protected] Correspondence:
[email protected]; Tel.: +81-33-492-3131
Received: 14 May 2018; Accepted: 29 May 2018; Published: 4 June 2018
Abstract: Human leukocyte antigen (HLA)-DRB1*01:01 has been shown to be involved in nevirapine-induced hepatic hypersensitivity reactions. In the present study, in silico docking simulations and molecular dynamics simulations were performed to predict the interaction mode of nevirapine with the peptide binding groove of HLA-DRB1*01:01 and its possible effect on the position and orientation of the ligand peptide derived from hemagglutinin (HA). In silico analyses suggested that nevirapine interacts with HLA-DRB1*01:01 around the P4 pocket within the peptide binding groove and the HA peptide stably binds on top of nevirapine at the groove. The analyses also showed that binding of nevirapine at the groove will significantly change the inter-helical distances of the groove. An in vitro competitive assay showed that nevirapine (1000 µM) increases the binding of the HA peptide to HLA-DRB1*01:01 in an allele-specific manner. These results indicate that nevirapine might interact directly with the P4 pocket and modifies its structure, which could change the orientation of loaded peptides and the conformation of HLA-DRB1*01:01; these changes could be distinctively recognized by T-cell receptors. Through this molecular mechanism, nevirapine might stimulate the immune system, resulting in hepatic hypersensitivity reactions. Keywords: nevirapine; HLA (human leukocyte antigen); hepatic hypersensitivity reaction; idiosyncratic drug toxicity; MD (molecular dynamics) simulation
1. Introduction Nevirapine (Viramune® ) is an orally available antiretroviral agent, which inhibits reverse transcriptase. It is used in combination with other antiretroviral agents for the treatment of HIV/AIDS infections and is included in the WHO Model List of Essential Medicines. Although it is generally well-tolerated and effective, it has been associated with hypersensitivity reactions (HSRs) in approximately 5% of patients with various clinical phenotypes, including cutaneous, hepatic, and systemic symptoms [1–4]. Several features of nevirapine HSRs, including the delayed onset of reaction, more rapid and severe development of symptoms after rechallenge, and association with the percentage of pretreatment CD4 T-cells, suggest that nevirapine HSRs are immune-mediated [3,5]. Various genetic association studies have shown that cytochrome P450 (CYP) 2B6 genotypes and human leukocyte antigen (HLA) alleles are predisposed to the development of nevirapine HSRs [4–8]; Int. J. Mol. Sci. 2018, 19, 1660; doi:10.3390/ijms19061660
www.mdpi.com/journal/ijms
Int. J. Mol. Sci. 2018, 19, 1660
2 of 12
Int. J. Mol. Sci. 2018, 19, x
2 of 11
however, these associations were very complicated. 8]; however, these associations were very complicated.For Forexample, example,CYP2B6 CYP2B6 G516T G516T polymorphism polymorphism increased nevirapine plasma exposure [9] and was associated with its cutaneous HSRs [6], but increased nevirapine plasma exposure [9] and was associated with its cutaneous HSRs [6],not butwith not hepatic ones [6,10]. In addition, unlike the cases of HLA-B*57:01 in abacavir HSRs [11] and HLA-B*58:01 with hepatic ones [6,10]. In addition, unlike the cases of HLA-B*57:01 in abacavir HSRs [11] and HLAin allopurinol-induced Stevens-Johnson syndrome (SJS) and necrolysis (TEN) [12], B*58:01 in allopurinol-induced Stevens-Johnson syndrome (SJS)toxic and epidermal toxic epidermal necrolysis (TEN) multiple class I and class II HLA alleles, which have variable distribution and risk across ethnic groups, [12], multiple class I and class II HLA alleles, which have variable distribution and risk across ethnic have been associated with nevirapine HSRs [6,13,14]. associations with protective groups, have been associated with nevirapine HSRs Interestingly, [6,13,14]. Interestingly, associations with HLA alleles have also been reported [7,13,15]. Furthermore, the associations with predisposing protective HLA alleles have also been reported [7,13,15]. Furthermore, the associations HLA with alleles tend to HLA be phenotype-specific where HLA-C*04 has been associated with cutaneous predisposing alleles tend to [6,14,16], be phenotype-specific [6,14,16], where HLA-C*04 has been HSRs, in particular SJS/TEN [6–8,13,14], but HLA-DRB1*01 has been associated with hepatic associated with cutaneous HSRs, in particular SJS/TEN [6–8,13,14], but HLA-DRB1*01 hasHSRs. been The exclusion individuals cutaneous andwith hepatic HSRs resulted in aand significant associated withofhepatic HSRs.with Thecoincident exclusion of individuals coincident cutaneous hepatic increase of the in odds ratios [5,6]. These results suggest that[5,6]. cutaneous and hepatic HSRs have HSRs resulted a significant increase of the odds ratios These results suggest thatmight cutaneous fundamentally different mechanisms. and hepatic HSRs might have fundamentally different mechanisms. We previously We previously reported reported in in silico silico and and in in vitro vitro interactions interactions of of lapatinib lapatinib [17] [17] and and ximelagatran ximelagatran [18] [18] with HLA-DRB1*07:01, which predisposed to the development of idiosyncratic hepatotoxicity by both with HLA-DRB1*07:01, which predisposed to the development of idiosyncratic hepatotoxicity by drugs, and indicated that these might might triggertrigger immune responses, leadingleading to drug-related both drugs, and indicated that interactions these interactions immune responses, to drugtoxicity. In the present westudy, performed in silico docking to predict the interaction related toxicity. In the study, present we performed in silicosimulations docking simulations to predict the mode of nevirapine (Figure 1) with the peptide binding groove of HLA-DRB1*01:01, as well interaction mode of nevirapine (Figure 1) with the peptide binding groove of HLA-DRB1*01:01, as as molecular dynamics (MD) simulations to predictto thepredict possible of nevirapine the structure of well as molecular dynamics (MD) simulations theeffect possible effect of on nevirapine on the HLA-DRB1*01:01 and the position of orientation the bound peptide derived from hemagglutinin structure of HLA-DRB1*01:01 andand the orientation position and of the bound peptide derived from (HA). An in vitro competitive assay was performed to evaluate the effect of nevirapine binding hemagglutinin (HA). An in vitro competitive assay was performed to evaluate the effecton ofthe nevirapine of peptides to three HLA-DR onligand the binding of ligand peptides to alleles. three HLA-DR alleles.
Figure 1. Chemical nevirapine. Figure 1. Chemical structure structure of of nevirapine.
2. Results 2. Results 2.1. Docking 2.1. Docking Simulations Simulations In performed to predict the the interaction mode of nevirapine with In silico silicodocking dockingsimulations simulationswere were performed to predict interaction mode of nevirapine the peptide binding groove of HLA-DRB1*01:01. The binding affinity of nevirapine to HLAwith the peptide binding groove of HLA-DRB1*01:01. The binding affinity of nevirapine to DRB1*01:01 was judged by a scoring function of generalized-Born volumevolume integral/weighted surface HLA-DRB1*01:01 was judged by a scoring function of generalized-Born integral/weighted area (GBVI/WSA_dG). The lowest GBVI/WSA_dG value of the of complex between nevirapine and surface area (GBVI/WSA_dG). The lowest GBVI/WSA_dG value the complex between nevirapine HLA-DRB1*01:01 was −5.18 kcal/mol. This value is higher than that of the complex between HLAand HLA-DRB1*01:01 was −5.18 kcal/mol. This value is higher than that of the complex between B*14:02 and nevirapine (−6.40(− kcal/mol) simulated by theby same software system [19], [19], indicating that HLA-B*14:02 and nevirapine 6.40 kcal/mol) simulated the same software system indicating the interaction of nevirapine with with HLA-DRB1*01:01 was weaker than than that with HLA-B*14:02. The that the interaction of nevirapine HLA-DRB1*01:01 was weaker that with HLA-B*14:02. binding mode of nevirapine at the peptide binding groove of HLA-DRB1*01:01 with the lowest The binding mode of nevirapine at the peptide binding groove of HLA-DRB1*01:01 with the lowest GBVI/WSA_dG value GBVI/WSA_dG valueisisshown shownin inFigure Figure 2. 2. Nevirapine Nevirapinesprawls sprawls out out across across the the pocket pocket P4–P6. P4–P6.
Int. J. Mol. Sci. 2018, 19, 1660
3 of 12
Figure 2. Binding mode of nevirapine to human leukocyte antigen (HLA)-DRB1*01:01 in docking simulations, (a) side view and (b) top view. The structures of HLA-DRB1*01:01, nevirapine, and amino acid residues of HLA-DRB1*01:01 in the vicinity of nevirapine are depicted in cartoon mode, ball-and-stick model, and wire model, respectively.
2.2. Molecular Dynamics (MD) Simulations MD simulations were performed to evaluate the possible effect of nevirapine on the structure of the peptide binding groove of HLA-DRB1*01:01 and the position and orientation of bound HA peptide. Each of the total energies of MD simulations stabilized within approximately 0.5 ns (Figure 3a) indicating that the simulations were energetically stable. The flexibility of HLA-DRB1*01:01 molecule was very similar in all simulations (Figure 3b). As shown in Figure 3c, the root mean square deviation (RMSD) values stabilized after 1–2 ns in all simulations, which indicated that the systems relaxed quickly. The interaction of nevirapine with HLA-DRB1*01:01 resulted in the closing motion of the peptide binding groove (Figure 3d, Table 1), which differs from the tightly closed binding groove observed with lapatinib-DRB1*07:01 [17] and the open groove observed with lapatinib-DRB1*01:01, lapatinib-DRB1*15:01, [17] and ximelagatran-DRB1*07:01 [18]. The mean inter-helical distance of the peptide binding groove of HLA-DRB1*01:01 in the presence of nevirapine (13.1 ± 0.4 Å; Table 1) was very similar to that observed in apo simulation (13.0 ± 0.4 Å), which reflected the transition of HLA-DR molecules from a peptide-receptive conformation to a non-receptive one in the absence of any ligand molecules [20]. This is reasonable considering the much smaller molecular size of nevirapine than that of lapatinib and ximelagatran. Table 1. Mean molecular dynamics (MD) simulations parameters. Hemagglutinin Peptide Frame 2 Frame 3 Frame 2 Frame 3
Nevirapine
− + − − + +
Energies (kcal/mol)
RMSF (Å)
Inter-Helical Distance (Å)
Slope of Inter-Helical Distance (Å/ns)
−106715 ± 309.9 −106370 ± 533.5 −111392 ± 579.8 −106078 ± 323.1 −120550 ± 243.6 −112166 ± 230.5
1.0 ± 0.4 0.8 ± 0.3 0.9 ± 0.4 0.9 ± 0.3 0.8 ± 0.3 0.9 ± 0.3
13.0 ± 0.4 13.1 ± 0.4 12.7 ± 0.2 14.2 ± 0.3 13.5 ± 0.2 12.6 ± 0.2
−0.12 −0.14 −0.16 0.04 0.02 −0.13
The total energies, average root mean square fluctuation (RMSF), distance between the α-helices flanking the binding groove, and slope of inter-helical distance curve are listed. Values represent the mean ± standard deviation (SD) during 5-ns simulations.
1
Int. J. Mol. Sci. 2018, 19, x Int. J. Mol. Sci. 2018, 19, 1660
4 of 11 4 of 12
Figure 3. Parameter profiles of molecular dynamics (MD) simulations of nevirapine-bound HLA-DRB1*01:01. (a) Calculated vs.dynamics time plot;(MD) (b) RMSF values of of nevirapine-bound polypeptide backbone. Figure 3. Parameter profiles of energies molecular simulations HLAThe location of α and β chains and residues that comprise the peptide binding groove helices backbone. are indicated DRB1*01:01. (a) Calculated energies vs. time plot; (b) RMSF values of polypeptide The bylocation the solidofand dashed lines running just above the x-axis, respectively. The α chain (solid black), α chain α and β chains and residues that comprise the peptide binding groove helices are indicated helix (solid purple), β chain (dashed black), β chain helixrespectively. (dashed purple); (c)chain Root(solid mean black), square α by the solid and dashed lines running just and above the x-axis, The α deviation (RMSD) values of polypeptide backbone vs. time plot; and (d) The average distance between chain helix (solid purple), β chain (dashed black), and β chain helix (dashed purple); (c) Root mean each Cα in the helix of the α chain and the closest Cα atom in the helix of the β chain. square deviation (RMSD) values of polypeptide backbone vs. time plot; and (d) The average distance between each Cα in the helix of the α chain and the closest Cα atom in the helix of the β chain.
Figure 4 shows structural analyses of five sets of MD simulations presented in three panels for each. In the absence HA peptide, nevirapine HLA-DRB1*01:01 in or near the P4panels pocketfor Figure 4 showsofstructural analyses of fivebound sets oftoMD simulations presented in three ineach. a stable conformation (Figure 4c) and had a small impact on the overall configuration α In the absence of HA peptide, nevirapine bound to HLA-DRB1*01:01 in or near the of P4 the pocket and chainsconformation of HLA-DRB1*01:01, compared lapatinib [17].overall Regarding HA peptide-bound in aβ stable (Figure 4c) and hadtoa that smallofimpact on the configuration of the α and simulations, starting framescompared (2, 3, andto4)that wereofconsidered, and Regarding frame 3 was thepeptide-bound most likely β chains ofthree HLA-DRB1*01:01, lapatinib [17]. HA because it was three stablestarting and exhibited with considered, both sides ofand the frame groove. The conformation simulations, framesgood (2, 3,contacts and 4) were 3 was the most likely was the closest to stable the X-ray structure 2 was the trimer because it was andcrystal exhibited good (Figure contacts4d–f). with Frame both sides ofalso the considered groove. Thefor conformation (nevirapine, HLA-DRB1*01:01, and HA peptide) simulations because of its total energy, although the was the closest to the X-ray crystal structure (Figure 4d–f). Frame 2 was also considered for the trimer (nevirapine, HApeptide peptide) simulations of its total energy,(Figure although the binding mode HLA-DRB1*01:01, and closing motionand of the binding groovebecause were somehow unusual 4g–i, binding closing motion ofconformational the peptide binding groove unusual (Figure 4g–i, Table 1). mode In theand trimer simulations, changes of were both somehow nevirapine and HA peptide Table 5-ns 1). In the trimerwere simulations, changes ofseems both to nevirapine and HA peptide during simulations slight in conformational both frames. Nevirapine interact mainly with the duringbinding 5-ns simulations werethe slight in bothwhereas frames.HA Nevirapine seemson totop interact mainly with the peptide groove around P4 pocket, peptide bound of nevirapine. Slight peptide binding groove around the P4 whereas HAofpeptide bound on with top ofthe nevirapine. conformational changes were observed inpocket, the α and β chains HLA-DRB1*01:01 α chain Slight conformational were the observed in the α and β chains of HLA-DRB1*01:01 with the α helix being maintained. changes Considering high root mean square fluctuation (RMSF) values of HA chain helix being maintained. Considering the 3high root more meanstably square fluctuation (RMSF) valuesinof peptide ends in frame 2, the HA peptide in frame appears bound to HLA-DRB1*01:01 HA peptide ends in frame 2, the HA peptide in frame 3 appears more stably bound to HLAthe trimer simulations. DRB1*01:01 the trimer simulations. Figure 5 in shows the differences in the positions and orientations of ligand peptides between Figure 5structures shows the differences in the positions and orientations of ligandhad peptides between representative of trimer and HLA-peptide dimer simulations. Nevirapine much smaller representative structures of trimer of and simulations. Nevirapine had much effects on the position and orientation theHLA-peptide ligand peptidedimer than lapatinib and ximelagatran. smaller effects on the position and orientation of the ligand peptide than lapatinib and ximelagatran.
Int. J. Mol. Sci. 2018, 19, 1660 Int. J. Mol. Sci. 2018, 19, x
5 of 12 5 of 11
Simulated representative representative structures structures of of nevirapine-bound nevirapine-bound HLA-DRB1*01:01 HLA-DRB1*01:01 (a–c), (a–c), Figure 4. 4. Simulated hemagglutinin (HA) (HA) peptide-bound peptide-boundHLA-DRB1*01:01 HLA-DRB1*01:01in in frame 2 (d–f) frame 3 (g–i), hemagglutinin frame 2 (d–f) and and frame 3 (g–i), and and nevirapine-bound HLA-DRB1*01:01-HA peptide in frame 2 (j–l) andframe frame33(m–o). (m–o). (a,d,g,j,m): (a,d,g,j,m): nevirapine-bound HLA-DRB1*01:01-HA peptide in frame 2 (j–l) and Alignment of initial structure (red) and representative structure (blue); (b,e,h,k,n): Sausage Sausage plot of the structure, structure, where where the the color color and and thickness thickness of of HLA-DRB1*01:01 HLA-DRB1*01:01 are are proportional proportional to to the the RMSF RMSF of of Cα; Cα; the and (c,f,i,l,o): (c,f,i,l,o): The The volume volume occupied occupied by by nevirapine nevirapine and and HA HA peptide. The blue blue envelope envelope outlines outlines the the and peptide. The 50% occupancy occupancy volume volume of of nevirapine nevirapine (red) 50% (red) and and HA HA peptide peptide (green). (green).
Int. J. Mol. Sci. 2018, 19, 1660 Int. J. Mol. Sci. 2018, 19, x
6 of 12 6 of 11
Figure 5.5. Drug-induced Drug-induced peptide peptide displacements displacements in in (a) (a) nevirapine-bound nevirapine-bound HLA-DRB1*01:01-HA HLA-DRB1*01:01-HA Figure peptide in frame 2 and (b) frame 3; (c) lapatinib-bound HLA-DRB1*01:01-HA peptide in frame frame22(d) (d) peptide in frame 2 and (b) frame 3; (c) lapatinib-bound HLA-DRB1*01:01-HA peptide in and frame 3; (e) lapatinib-bound HLA-DRB1*07:01-peptide from tetanus toxoid (TT) in frame 3; and and frame 3; (e) lapatinib-bound HLA-DRB1*07:01-peptide from tetanus toxoid (TT) in frame 3; (f) ximelagatran-bound HLA-DRB1*07:01-TT peptide in frame 3. Each in the ligand and (f) ximelagatran-bound HLA-DRB1*07:01-TT peptide in frame 3. residue Each residue in thepeptides ligand is colored to the displacement of the Cαof inthe the Cα trimer simulations, relative torelative its position peptides is according colored according to the displacement in the trimer simulations, to itsin the HLA-peptide dimer simulations. The same color scale is used and the range is from 0 (blue) to 10 position in the HLA-peptide dimer simulations. The same color scale is used and the range is from 0 Å (red). Small values mean that the drug has little effect on the position of ligand peptides in the (blue) to 10 Å (red). Small values mean that the drug has little effect on the position of ligand peptides groove. DrugsDrugs are shown usingusing grey stick representation. inbinding the binding groove. are shown grey stick representation.
2.3.HLA HLAClass ClassIIIICompetitive CompetitiveAssay Assay 2.3. Toevaluate evaluate effect of nevirapine the binding ofpeptides ligand to peptides to threemolecules, HLA-DR To thethe effect of nevirapine on theon binding of ligand three HLA-DR molecules, a DELFIA-based competitive assay was conducted. Nevirapine increased the binding of a DELFIA-based competitive assay was conducted. Nevirapine increased the binding of HA peptide HA peptide to soluble HLA-DRB1*01:01 only at0.0001), 1000 μM (p < 0.0001), there were to soluble HLA-DRB1*01:01 molecule onlymolecule at 1000 µM (p < whereas there whereas were no significant no significant or concentration-dependent effects on the binding of the peptide derived from tetanus or concentration-dependent effects on the binding of the peptide derived from tetanus toxoid (TT) to toxoid (TT) to HLA-DRB1*07:01 and that derived from myelin basic protein (MBP) to HLA-DRB1*07:01 and that derived from myelin basic protein (MBP) to HLA-DRB1*15:01 (Table HLA2). DRB1*15:01 (Table 2). Table 2. The effect of nevirapine on the binding of ligand peptides to HLA-DR molecules. Table 2. The effect of nevirapine on the binding of ligand peptides to HLA-DR molecules. HLA Allele DRB1*01:01 DRB1*07:01 DRB1*15:01 HLA Allele DRB1*01:01 DRB1*07:01 DRB1*15:01 # 1000 62.6 3.9 101.3 14.9 1000 263.9 263.9±±15.5 15.5 # 62.6± ± 3.9 101.3± ± 14.9 200 99.7 ± 11.3 110.4 ± 5.5 94.6 ± 10.6 200 99.7 ± 11.3 110.4 ± 5.5 94.6 ± 10.6 Concentration of nevirapine (µM) Concentration of nevirapine (μM) 40 106.2 ± 6.5 76.9 ± 4.6 93 ± 10.4 40 106.2 ± 6.5 76.9 ± 4.6 93 ± 10.4 8 101.9 ± 8.1 132.2 ± 64 98.9 ± 3.4 8 101.9 ± 8.1 132.2 ± 64 98.9 ± 3.4
The effects of nevirapine on the binding of ligand peptides to three HLA-DRs are expressed as % binding of ligand The effects nevirapine on the binding ligandsulfoxide peptides(DMSO) to three HLA-DRs expressed % peptides in theofpresence of nevirapine to that in of dimethyl control (n = 8). are Values representasthe mean ± SDofofligand quadruplicate. p
