http://informahealthcare.com/drt ISSN: 1061-186X (print), 1029-2330 (electronic) J Drug Target, Early Online: 1–7 ! 2014 Informa UK Ltd. DOI: 10.3109/1061186X.2014.959019
ORIGINAL ARTICLE
Targeting human secretory phospholipase A2 with designed peptide inhibitors for inflammatory therapy Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
Peng Wang1*, Yongtao Li2*, Qiuping Shao1, Wenqin Zhou1, and Kuifeng Wang2 1
Emergency Medical Center (Trauma Center), People’s Hospital affiliated to Jiangsu University, Zhenjiang, China and 2Shanghai GenHouse Technology Co., Ltd., Shanghai, China Abstract
Keywords
Phospholipase A2 (PLA2) is potentially an important target for anti-inflammatory therapeutics. Here, we described a systematic scheme that integrated protein docking and peptide redocking, molecular dynamics simulation, and binding affinity analysis to rationally design PLA2 inhibitory peptides based on a solved PLA2 crystal structure. The scheme employed protein docking to sample the interaction modes of PLA2 with its natural inhibitor Clara cell protein, from which a number of peptide fragments, including a pentapeptide LLLGS, were cut off and redocked to serve as the lead entities of PLA2 inhibitory peptides. In addition, a systematic mutation energy map that characterized the binding free energy changes DG upon mutations of each position of the putative pentapeptide to 20 amino acids was also profiled, which was subsequently used to guide peptide structure optimization. In order to solidify the computational findings, we performed kinetic and inhibition studies of few designed peptides against human secretory PLA2. Consequently, eight peptides were successfully identified to have potent inhibition potency, in which the LLAYK and AVFRS were found to suppress enzymatic activity significantly (Ki ¼ 0.75 ± 0.06 and 4.2 ± 0.3 mM, respectively). A further structure examination revealed that the designed peptides can form intensive nonpolar networks of van der Waals contacts and hydrophobic interactions at their complex interfaces with PLA2, conferring considerable stability and affinity for the formed complex systems.
Inflammation, peptide inhibitor, phospholipase A2, rational drug targeting
Introduction Inflammation is a pervasive phenomenon that operates during severe perturbations of homeostasis such as infection, injury, and exposure to contaminants, and is triggered by innate immune receptors that recognize pathogens and damaged cells. Among vertebrates, the inflammatory cascade is a complex network of immunological, physiological, and behavioral events that are coordinated by cytokines, immune signaling molecules [1]. A number of enzymes such as metalloproteases, serine proteases, and phospholipases A2 (PLA2) have been involved in the pathological process of inflammation. The PLA2 is a growing superfamily of 15 kDa, calcium-dependent, disulfide-linked, a-helical enzymes that catalyze hydrolysis of membrane phospholipids to result in diverse stoichiometric productions such as lysophospholipids and free fatty acids [2]. These phospholipid metabolites serve as precursors for inflammatory mediators such as eicosanoids or platelet-activating factors. Thus, the ability of PLA2 to produce substrates for the generation of *These authors contributed equally to this work. Address for correspondence: Peng Wang, Emergency Medical Center (Trauma Center), People’s Hospital affiliated to Jiangsu University, Zhenjiang 212002, China. E-mail: [email protected], zjwangpeng@ 163.com
History Received 19 May 2014 Revised 13 August 2014 Accepted 25 August 2014 Published online 19 September 2014
inflammatory lipid mediators in the process of tissue injury and rheumatoid arthritis makes this enzyme family as a therapeutic target of anti-inflammatory drugs [3]. Previously, animal experiments showed that some peptides and peptide analogues exhibited strong inhibitory capability to suppress the catalytic activity of PLA2 with low toxicity and high biocompatibility [4], suggesting that peptides can be considered as ideal lead entities to develop potent PLA2 inhibitors. Traditionally, bioactive peptides are primarily obtained by random approaches such as peptide library and phage display; these methods, however, are too timeconsuming and expensive to screen against all possible peptide combinations in a given length, largely limiting the practicability of therapeutic peptide development. Over the past decade, computational peptide design has spurred increasing interest in the medicinal community [5]. However, rational design of peptide ligands that can bind tightly to PLA2 receptor has long been a challenging task until the crystal structures of PLA2 as well as its complexes with diverse peptide ligands were solved at high-resolution level [6,7]. The important features of overall structure of PLA2 contain a N-terminal helix, an external loop, a calciumbinding loop, a second helix, a short anti-parallel b-wing, a third helix, and two single t-turns; such structure
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
2
P. Wang et al.
configuration defines a conserved active site within a hydrophobic channel lined by a number of invariant hydrophobic residues [8,9]; the wall of the channel contains several key catalytic residues such as His48 and Asp49 behind which a disulfide link involving Cys51–Cys93 stabilizes the channel architecture [10]. In the current study, based on the elucidated structure feature we performed systematic molecular modeling to investigate the interaction modes of human secretory PLA2 with a natural PLA2 inhibitor, Clara cell protein [11]. On this basis, various peptide fragments were derived from the putative protein interaction sites to define a distinct set of PLA2 inhibitory peptides, from which a number of candidates that were potential to target PLA2 were identified using a peptide redocking strategy with the guide of solved crystal structure data. Subsequently, enzymatic inhibition assay was conducted to determine the inhibitory capability of few highly promising peptides against PLA2. We also gave a systematic analysis of the structural basis, energetic property, and biological implication underlying the intermolecular interactions between the PLA2 catalytic domain and designed peptide inhibitors.
Materials and methods Crystal structure data The crystal structure of human secretory PLA2 was retrieved from the PDB database (PDB: 1KQU). The structure was solved in complex with a substrate analogue, the chiral precursor D-tyrosine, which is bound in the active site of the enzyme. In addition, two Ca2+ ions are chelated with the calcium-binding loop of PLA2 through a number of carboxylate and amide oxygen atoms, and multiple hydrophobic contacts and a T-shaped aromatic-group–His6 interaction are involved in the structure to stabilize PLA2 architecture [12] (Figure 1A). Previously, a variety of proteins have been reported to have inhibitory activity against PLA2, which could be considered as potential candidates to derive functional peptide fragments with PLA2 suppressing activity. For example, Thwin et al. [13] successfully designed a 17-residue PLA2 inhibitory peptide using the parent structure of a protein from Python serum.
J Drug Target, Early Online: 1–7
In this respect, we here attempted to exploit the crystal structure data of a natural PLA2 inhibitor Clara cell protein (CCP) [11], aiming to discover new potent PLA2 inhibitory peptides in this protein. The CCP is a 17-kDa protein secreted by Clara cells and other nonciliated cells of both the bronchiolar and the bronchial epithelium, which is a homodimer bound covalently by two disulfide bonds that joins the monomers together in an antiparallel manner such that the dimer encloses a large internal hydrophobic cavity (PDB: 1CCD) [14] (Figure 1B). Protein docking and peptide redocking Protein docking of CCP to PLA2 was carried out using ZDOCK method [15], which performed a full rigid-body search of docking orientations between two proteins in consideration of electrostatic interactions and solvent effect at the sampled protein–protein interface. Before performing the docking operation, the cocrystallized inhibitor ligand and calcium ions were removed from the PLA2 structure, and hydrogen atoms and protonation states were assigned for the CCP to PLA2 crystal structures using REDUCE [16] and PROPKA [17] programs, respectively. FlexPepDock server [18] was employed to redock isolated peptide fragments to PLA2. The method is a high-resolution peptide docking and refinement protocol implemented within the Rosetta framework, which has been shown to be able to accurately refine the peptide structure starting from the initial position nearby its native conformation, allowing full flexibility to the peptide and side-chain flexibility to protein receptor [19]. Molecular dynamics simulation and binding free energy analysis Docked complex systems (PLA2–CCP or PLA2–peptide complexes) were then subjected to molecular dynamics (MD) simulations using the AMBER03 force field [20]. In the MD simulation procedure, calcium ions were kept in PLA2 structure in order to stabilize the protein architecture. Complex was dissolved in a rectangular box full of TIP3P water molecules, and counterions of Na+ were placed to keep
Figure 1. The crystal structures of human secretory PLA2 (PDB: 1KQU) (A) and Clara cell protein (PDB: 1CCD) (B).
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
DOI: 10.3109/1061186X.2014.959019
the system neutral. The particle mesh Ewald (PME) strategy [21] and SHAKE algorithm [22] were used to treat the longrange electrostatic interactions and to constrain all hydrogen atoms, respectively. First, all hydrogen atoms, water molecules and counterions were minimized, and then the system was relaxed by 3000 cycles of minimization without constraints (500 steps of steepest gradient descent and 2500 steps of the conjugate gradient method). The subsequent MD simulations consisted of a gradual temperature increase to target 300 K over 100 ps and a 5-ns equilibration phase under the pressure of 1 atm for data collection, during which trajectory snapshots were saved every 100 ps. The binding free energy DG of CCP or peptide to PLA2 was calculated from the MD-derived snapshots of PLA2– CCP/peptide complex dynamics trajectory using the MM/ GBSA method [23], where the nonbonded interactions between PLA2 and CCP/peptide were computed with molecular mechanism (MM) approach, while the desolvation effect due to the PLA2–CCP/peptide binding was described by the generalization Born (GB) model (for polar contribution) and surface area (SA) strategy (for nonpolar contribu˚ , and the tion). The grid size for the GB calculations was 0.5 A values of interior and exterior dielectric constants were set to 1 and 80, respectively. Kinetic and inhibition studies Kinetic studies were performed to investigate the inhibition of PLA2 by designed peptides. The experimental protocol was described previously [10]. All assays were conducted in 20 mM glycylglycine buffer (pH 7.5, 30 C) containing 3 mM CaCl2 using 0.05 M cresol red as indicator. The enzyme fixed at concentration 1.5 mM was co-incubated with tested peptides for 2 h, and the amount of free enzyme was determined by the addition of phosphatidylcholine substrate. The volume of the
Targeting human secretory PLA2
3
substrate solution comprised 5% of the total reaction mixture. The initial velocity was calculated from the changes in an absorbance at 578 nm recorded with a Perkin–Elmer spectrophotometer.
Results and discussion PLA2–CCP docking sampling The ZDOCK method was employed to sample possible interaction modes between PLA2 and CCP [15]. The docking sampling procedure included a full rigid-body search phase of docking orientations between the two proteins and a refinement phase to optimize the structure models of complex architectures generated by the rigid-body docking. Consequently, various potential interactions were obtained directly from the ZDOCK server [24], and then we have oneby-one examined visually these docked modes in the PyMol graphics interface. It was found that most docked CCP locations were nearby the active site of PLA2, but the CCP protein can adopt multiple sites to touch on the PLA2 surface. Then, a number of modes were empirically identified as the promising interactions of PLA2 with CCP, where the sampled CCP conformations are clustered and superposed onto the PLA2 surface and shown in Figure 2(A). Since most of the existing PLA2 inhibitors are small peptides with short sequence lengths [25,26], we herein only considered CCP fragments of less than five amino acids derived from the interaction sites of docked PLA2–CCP complexes. The atomic contacts at the complex interface of PLA2 with each docked CCP conformation were analyzed using PROBE program [27] to determine the inter-residue interactions between PLA2 and CCP conformation. In this way, we extracted 11 CCP fragments that satisfied following criteria (Table 1): (i) their sequence lengths range between 3 and 5, (ii) all amino acid residues in a fragment should be
Figure 2. (A) Conformation cluster of docked CCP protein nearby the PLA2 active site. (B) An example of extracting peptide fragment LLLGS (13–17) from the interaction site of a docked CCP conformation with PLA2.
4
P. Wang et al.
J Drug Target, Early Online: 1–7
Table 1. The 11 peptide fragments extracting from the interaction sites of docked PLA2–CCP complexes.
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
Peptide fragment GFLQ LLLGS AALK PFNP NPAS NAGT LVDT LPQ IVK ILT PLCE
DG (kcal/mol)
Sequence position in CCP
PPI
PTI1
PTI2
6–8 13–17 23–26 27–30 29–32 36–39 44–47 48–50 56–58 63–65 67–70
22.8 34.5 28.7 46.4 28.0 41.4 34.7 35.6 28.6 21.3 25.5
8.6 12.4 14.5 5.6 6.8 7.6 12.0 7.8 12.6 11.3 9.7
12.3 17.4 19.7 11.0 10.6 12.2 15.9 14.5 16.2 15.3 13.7
PPI, the binding free energy of PLA2 to docked CCP conformation from which the corresponding peptide fragment was extracted. PTI1, the binding free energy of peptide fragment to PLA2 derived using the docked PLA2–CCP complex. PTI2, the binding free energy of peptide fragment to PLA2 derived using the redocked PLA2–peptide complex.
detected to be in atomic contact with the PLA2 partner, and (iii) the fragments should contain a considerable proportion of nonpolar residues due to the hydrophobic nature of PLA2 active site. An example of extracting peptide fragment LLLGS (13–17) from the interaction site of a docked CCP conformation with PLA2 is shown in Figure 2(B). Redocking of peptides into PLA2 active site Previously, Stein and Aloy [28] systematically examined various peptide-mediated protein interactions in three-dimensional crystal structures and found that the context of peptide motifs in protein–protein interfaces plays a crucial role in determining interaction stability and specificity. This finding suggested that the binding behavior of a peptide to its partner may differ significantly in isolated state and in protein context. In this respect, we performed peptide docking to separately redock all the 11 peptides into the active site of PLA2 by using the FlexPepDock server [18]; here the initial conformations of these peptides (the conformations that peptide fragments were cut out directly from the docked PLA2–CCP complexes) were used to guide the redocking procedure. Further, the 11 redocked peptide conformations were separately subjected to 5-ns MD simulations to equilibrate and refine their complex structures with PLA2. The redocked conformations of these peptides were observed to deviate considerably from their initial conformations, as the mean value of root-mean-square deviations (rmsd) of docked from initial conformations over the 11 ˚ . In contrast, MD simulations peptides is as much as 1.85 A can only address moderate or modest effect on peptide conformations in PLA2 active site, with the mean rmsd value ˚ between the 11 redocked and MD-equilibrated of 0.43 A conformations. For example, the initial, redocked and equilibrated conformations of pentapeptide LLLGS (corresponding to the sequence position 13–17 of CCP protein) bound with PLA2 are shown in Figure 3. It is seen that the redocked conformation has a large motion relative to initial position, but is very close to equilibrated conformation,
Figure 3. The initial, redocked, and MD-equilibrated conformations of peptide fragment LLLGS (13–17) in PLA2 active site. The initial conformation is that cut out directly from the interaction site of docked PLA2–CCP complex.
supporting that protein context contributes considerably to peptide binding. Next, the binding behavior of peptides to PLA2 was investigated at energetic level. The MM/GBSA method was employed to calculate the binding free energies between PLA2 and the 11 peptides using both initial and equilibrated conformations, as well as, for comparison purpose, to calculate the binding free energies of the docked CCP– PLA2 complexes that were used to derive the 11 peptides. The obtained free energy values DG are tabulated in Table 1. In energetic point of view, the extracted peptide fragments can only contribute about one-fifth of the total binding free energy to corresponding CCP–PLA2 complexes, ranging from 5 to 15 kcal/mol. However, after redocking and MD equilibration, the peptide binding affinities to PLA2 have a significant increase; their free energies were enhanced generally by two-fold. It is suggested that the redocked conformation, but not initial conformation, of peptides should be more close to their real conformation in isolated state; the MD simulations could further refine the atomic-level structures of peptide complexes with PLA2, which resulted in a significant improvement in peptide affinity. Thus, the MDequilibrated peptide conformations were adopted to perform subsequent structural and energetic analysis. Structure-based optimization of peptide inhibitors Three out of the 11 extracted peptides were selected to conduct kinetic studies; the three peptides (i.e. LLLGS, AALK, and IVK) were selected because they were calculated to have the highest theoretical affinities to PLA2 (DG ¼ 18.2, 19.7, and 16.2 kcal/mol, respectively). As can be seen, the three peptides share a common structure motif, that is, a polar or charged amino acid (S or K) occupied at their C-terminus, with a hydrophobic chain composed of two or three nonpolar amino acids (L, A, and V) to define their N-terminus. This structure motif is basically consistent
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
DOI: 10.3109/1061186X.2014.959019
with those known PLA2 inhibitory peptides reported previously [25,26]. As might be expected, the biological activities of the three peptides were determined to be moderate, with their inhibition constants Ki against PLA2 of 34.7 ± 3.1, 17.4 ± 1.5, and 152.5 ± 13.8 mM. The pentapeptide LLLGS and tetrapeptide AALK showed much higher activities as compared with the tripeptide IVK. This is anticipated because most known PLA2 inhibitory peptides are four or five amino acid residues long; too short sequences are unable to confer sufficient potency for peptide binding to PLA2. In order to further improve peptide activity, we here examined atomic interactions in the complex structure of PLA2 with LLLGS. The MD-equilibrated PLA2–LLLGS complex as well as nonbonded interactions across the complex interface are shown in Figure 4, from which it is evident that a number of van der Waals contacts and hydrophobic forces are formed between the nonpolar N-terminus of peptide ligand and the hydrophobic pocket
Targeting human secretory PLA2
5
around PLA2 active site, defining a stable complex architecture of PLA2 with the peptide ligand. However, there are no specific chemical forces such as hydrogen bonding and salt bridges observed at the complex interface, indicating that the recognition of peptide LLLGS by PLA2 is largely nonspecific that cannot be used to develop selective inhibitors. Thus, we performed a systematic amino acid scanning on each residue position of the pentapeptide ligand. The procedure is an extension of the widely used alanine scanning [29], where all the 20 amino acids, but not only the alanine, were in turn used to perform scanning. According to a previously described strategy [30], each residue of the pentapeptide was mutated to other 19 types of amino acids using SCAP program [31], followed by a AMBER force field minimization [32] to relax the mutated system. Consequently, we obtained 95 mutants as well as one template complex, based on which the mutation energy map of PLA2–LLLGS interaction is depicted in Figure 5, in which
Figure 4. (A) Stereoview of MD-equilibrated PLA2–LLLGS complex structure architecture. (B) Schematic representation of nonbonded interactions across the complex interface, generated by LIGPLOT program [34].
Figure 5. The mutation energy map of PLA2–LLLGS interaction. The mutation energies (DDG) are highlighted by binding energy changes upon systematic mutations of each position of pentapeptide LLLGS to other 19 amino acids.
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
6
P. Wang et al.
J Drug Target, Early Online: 1–7
Figure 6. The calculated binding free energy distribution of all the 243 combined pentapeptides to PLA2.
Table 2. The eight peptide candidates with binding free energies DG520 kcal/mol. Peptide
DG (kcal/mol)*
Ki (mM)y
VAYKS ALYYR LLAYK LIAFK AVFRS AVAFK LALYS LVAYS
22.4 20.9 23.8 22.9 20.3 22.1 22.4 23.8
NIz 107 ± 9 0.75 ± 0.06 NI 4.2 ± 0.3 NIz 18 ± 2 94 ± 8
NI, not inhibitory in the assay. *DG, calculated by MM/GBSA approach. yKi, measured by kinetic assay.
the mutation energies (DDG) are colored by binding energy changes upon systematic single-point mutations on each residue position. The mutation energy DDG for a specific peptide position upon a given mutation was defined as the difference between the PLA2 binding energies of mutant and the parent peptide (LLLGS) [33]. As can be seen from Figure 5, most mutations appear to reduce PLA2–peptide affinity (DDG40) – no matter the mutations were occurred at which positions, but this is not very significant for most mutants (DDG51 kcal/mol). By visually examining LLLGS and its mutants within the active pocket of PLA2, it was observed that these peptides are loosely bound with PLA2 to form a large nonpolar interface between them; mutations of just single residues seem not to be essential in affecting PLA2–peptide binding behavior, leading to only a slight affinity change upon the mutation. Here, we selected the top-3 favorable amino acids at each position of the pentapeptide ligand in terms of the mutation energy map, that is, P1: Ala, Leu, Val; P2: Val, Leu, Ile; P3: Ala, Leu, Val; P4: Tyr, Lys, Phe; P5: Lys, Arg, Ser. As can be seen, the N-terminal positions P1, P2, and P3 are favorable to bulk and/or hydrophobic amino acid types such as Leu, Ile, and Val, whereas the C-terminal positions P4 and P5 prefer to be occupied by polar or charged types such as Ser and Lys. This is in line with the basic structure feature of existing
PLA2 inhibitory peptides [25,26]. These favorable amino acids at each position were systematically combined to generate at most 35 ¼ 243 pentapeptides, and their complex structures with PLA2 were modeled from the PLA2–LLLGS complex using the SCAP virtual mutagenesis and AMBER minimization described above. Based on the modeled static structures, the binding free energies DG of the 243 pentapeptides to PLA2 were one-by-one calculated through the MM/GBSA approach [23], which are plotted in Figure 6. Evidently, only 11 out of 243 pentapeptides were estimated to bind more tightly than the parent LLLGS, with their DG518.2 kcal/mol, indicating that most mutations on LLLGS are unfavorable to peptide affinity. In order to select those highly promising candidates, we here only considered the peptides with DG520 kcal/mol and, consequently, eight samples satisfied this criterion and thus were selected, as tabulated in Table 2. The kinetic assay was performed to determine the inhibition potency of the eight peptides against PLA2, and obtained Ki values are also listed in Table 2. It is seen that five out of the eight peptides were measured to have high or moderate activity, with their Ki values ranging between 0.75 ± 0.06 and 107 ± 9 mM. Other three samples have no observable activity in the assay, which were supposed to possess relatively high hydrophobicity that may cause weak, non-specific interactions with PLA2 [35,36].
Declaration of interest The authors report that they have no conflicts of interest. This work was supported by the Social Development Project of Zhenjiang City (No. SH2013050).
References 1. Ashley NT, Weil ZM, Nelson RJ. Inflammation: mechanisms, costs, and natural variation. Annu Rev Ecol Evol Syst 2012;43: 385–406. 2. Burke JE, Dennis EA. Phospholipase A2 structure/function, mechanism, and signaling. J Lipid Res 2009;50:S237–42. 3. Meyer MC, Rastogi P, Beckett CS, McHowat J. Phospholipase A2 inhibitors as potential anti-inflammatory agents. Curr Pharm Des 2005;11:1301–12.
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
DOI: 10.3109/1061186X.2014.959019
4. Mahalka AK, Kinnunen PK. Class specific peptide inhibitors for secretory phospholipases A2. Biochem Biophys Res Commun 2013;436:349–53. 5. Zhou P, Wang C, Ren Y, et al. Computational peptidology: a new and promising approach to therapeutic peptide design. Curr Med Chem 2013;20:1985–96. 6. Chandra V, Kaur P, Srinivasan A, Singh TP. Three-dimensional structure of a presynaptic neurotoxic phospholipase A2 from ˚ resolution. J Mol Biol 2000;296: Daboia russelli pulchella at 2.4 A 1117–26. 7. Chandra V, Jasti J, Kaur P, et al. Design of specific peptide inhibitors of phospholipase A2: structure of a complex formed between Russell’s viper phospholipase A2 and a designed peptide Leu-Ala-Ile-Tyr-Ser (LAIYS). Acta Crystallogr D 2002;58: 1813–19. 8. Singh RK, Ethayathulla AS, Jabeen T, et al. Aspirin induces its anti-inflammatory effects through its specific binding to phospholipase A2: crystal structure of the complex formed between phospholipase A2 and aspirin at 1.9 angstroms resolution. J Drug Target 2005;13:113–19. 9. Singh RK, Somvanshi N, Sharma S, et al. Structural elements of ligand recognition site in secretory phospholipase A2 and structurebased design of specific inhibitors. Curr Top Med Chem 2007;7: 757–64. 10. Singh RK, Vikram P, Makker J, et al. Design of specific peptide inhibitors for group I phospholipase A2: structure of a complex formed between phospholipase A2 from Naja naja sagittifera (group I) and a designed peptide inhibitor Val–Ala–Phe–Arg–Ser (VAFRS) ˚ resolution reveals unique features. Biochemistry 2003;42: at 1.9 A 11701–6. 11. Jorens PG, Sibille Y, Goulding NJ, et al. Potential role of Clara cell protein, an endogenous phospholipase A2 inhibitor, in acute lung injury. Eur Respir J 1995;8:1647–53. 12. Hansford KA, Reid RC, Clark CI, et al. D-tyrosine as a chiral precusor to potent inhibitors of human nonpancreatic secretory phospholipase A2 (IIa) with antiinflammatory activity. Chembiochem 2003;4:181–5. 13. Thwin MM, Ong WY, Fong CW, et al. Secretory phospholipase A2 activity in the normal and kainate injected rat brain, and inhibition by a peptide derived from python serum. Exp Brain Res 2003;150: 427–33. 14. Umland TC, Swaminathan S, Furey W, et al. Refined structure of ˚ resolution. J Mol Biol 1992; rat Clara cell 17 kDa protein at 3.0 A 224:441–8. 15. Chen R, Li L, Weng Z. ZDOCK: an initial-stage protein-docking algorithm. Proteins 2003;52:80–7. 16. Word JM, Lovell SC, Richardson JS, Richardson DC. Asparagine and glutamine: using hydrogen atom contacts in the choice of sidechain amide orientation. J Mol Biol 1999a;285:1735–47. 17. Rostkowski M, Olsson MH, Søndergaard CR, Jensen JH. Graphical analysis of pH-dependent properties of proteins predicted using PROPKA. BMC Struct Biol 2011;11:6. 18. London N, Raveh B, Cohen E, et al. Rosetta FlexPepDock web server – high resolution modeling of peptide-protein interactions. Nucleic Acids Res 2011;39:W249–53.
Targeting human secretory PLA2
7
19. Raveh B, London N, Schueler-Furman O. Sub-angstrom modeling of complexes between flexible peptides and globular proteins. Proteins 2010;78:2029–40. 20. Duan Y, Wu C, Chowdhury S, et al. A point-charge force field for molecular mechanics simulations of proteins based on condensedphase quantum mechanical calculations. J Comput Chem 2003;24: 1999–2012. 21. Darden T, York D, Pedersen L. Particle mesh Ewald and N.log(N) method for Ewald sums in large systems. J Chem Phys 1993;98: 10089–92. 22. Ryckaert J, Ciccotti G, Berendsen HJC. Numerical-integration of Cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J Comput Phys 1977;23:327–41. 23. Kollman PA, Massova I, Reyes C, et al. Calculating structures and free energies of complex molecules: combining molecular mechanics and continuum models. Acc Chem Res 2000;33:889–97. 24. Pierce BG, Wiehe K, Hwang H, et al. ZDOCK server: interactive docking prediction of protein-protein complexes and symmetric multimers. Bioinformatics 2014;30:1771–3. 25. Church WB, Inglis AS, Tseng A, et al. A novel approach to the design of inhibitors of human secreted phospholipase A2 based on native peptide inhibition. J Biol Chem 2001;276:33156–64. 26. Thwin MM, Satyanarayanajois SD, Nagarajarao LM, et al. Novel peptide inhibitors of human secretory phospholipase A2 with antiinflammatory activity: solution structure and molecular modeling. J Med Chem 2007;50:5938–50. 27. Word JM, Lovell SC, LaBean TH, et al. Visualizing and quantifying molecular goodness-of-fit: small-probe contact dots with explicit hydrogen atoms. J Mol Biol 1999b;285:1711–33. 28. Stein A, Aloy P. Contextual specificity in peptide-mediated protein interactions. PLoS One 2008;3:e2524. 29. Morrison KL, Weiss GA. Combinatorial alanine-scanning. Curr Opin Chem Biol 2001;5:302–7. 30. Bi J, Yang H, Yan H, et al. Knowledge-based virtual screening of HLA-A*0201-restricted CD8+ T-cell epitope peptides from herpes simplex virus genome. J Theor Biol 2011;281:133–9. 31. Xiang Z, Honig B. Extending the accuracy limits of prediction of side-chain conformations. J Mol Biol 2001;311:421–30. 32. Hou T, Chen K, McLaughlin WA, et al. Computational analysis and prediction of the binding motif and protein interacting partners of the Abl SH3 domain. PLoS Comput Biol 2006;2:e1. 33. Ni Z, Chen H, Qi X, Jin R. Why is substrate peptide binding unsusceptible to multidrug-resistant mutations in HIV-1 protease? A structural and energetic analysis. Int J Pept Res Ther 2014;20: 43–51. 34. Wallace AC, Laskowski RA, Thornton JM. LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions. Protein Eng 1995;8:127–34. 35. Zhou P, Zou J, Tian F, Shang Z. Fluorine bonding – how does it work in protein-ligand interactions? J Chem Inf Model 2009;49: 2344–55. 36. Zhou P, Huang J, Tian F. Specific noncovalent interactions at protein-ligand interface: implications for rational drug design. Curr Med Chem 2012;19:226–38.
ORIGINAL ARTICLE
Targeting human secretory phospholipase A2 with designed peptide inhibitors for inflammatory therapy Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
Peng Wang1*, Yongtao Li2*, Qiuping Shao1, Wenqin Zhou1, and Kuifeng Wang2 1
Emergency Medical Center (Trauma Center), People’s Hospital affiliated to Jiangsu University, Zhenjiang, China and 2Shanghai GenHouse Technology Co., Ltd., Shanghai, China Abstract
Keywords
Phospholipase A2 (PLA2) is potentially an important target for anti-inflammatory therapeutics. Here, we described a systematic scheme that integrated protein docking and peptide redocking, molecular dynamics simulation, and binding affinity analysis to rationally design PLA2 inhibitory peptides based on a solved PLA2 crystal structure. The scheme employed protein docking to sample the interaction modes of PLA2 with its natural inhibitor Clara cell protein, from which a number of peptide fragments, including a pentapeptide LLLGS, were cut off and redocked to serve as the lead entities of PLA2 inhibitory peptides. In addition, a systematic mutation energy map that characterized the binding free energy changes DG upon mutations of each position of the putative pentapeptide to 20 amino acids was also profiled, which was subsequently used to guide peptide structure optimization. In order to solidify the computational findings, we performed kinetic and inhibition studies of few designed peptides against human secretory PLA2. Consequently, eight peptides were successfully identified to have potent inhibition potency, in which the LLAYK and AVFRS were found to suppress enzymatic activity significantly (Ki ¼ 0.75 ± 0.06 and 4.2 ± 0.3 mM, respectively). A further structure examination revealed that the designed peptides can form intensive nonpolar networks of van der Waals contacts and hydrophobic interactions at their complex interfaces with PLA2, conferring considerable stability and affinity for the formed complex systems.
Inflammation, peptide inhibitor, phospholipase A2, rational drug targeting
Introduction Inflammation is a pervasive phenomenon that operates during severe perturbations of homeostasis such as infection, injury, and exposure to contaminants, and is triggered by innate immune receptors that recognize pathogens and damaged cells. Among vertebrates, the inflammatory cascade is a complex network of immunological, physiological, and behavioral events that are coordinated by cytokines, immune signaling molecules [1]. A number of enzymes such as metalloproteases, serine proteases, and phospholipases A2 (PLA2) have been involved in the pathological process of inflammation. The PLA2 is a growing superfamily of 15 kDa, calcium-dependent, disulfide-linked, a-helical enzymes that catalyze hydrolysis of membrane phospholipids to result in diverse stoichiometric productions such as lysophospholipids and free fatty acids [2]. These phospholipid metabolites serve as precursors for inflammatory mediators such as eicosanoids or platelet-activating factors. Thus, the ability of PLA2 to produce substrates for the generation of *These authors contributed equally to this work. Address for correspondence: Peng Wang, Emergency Medical Center (Trauma Center), People’s Hospital affiliated to Jiangsu University, Zhenjiang 212002, China. E-mail: [email protected], zjwangpeng@ 163.com
History Received 19 May 2014 Revised 13 August 2014 Accepted 25 August 2014 Published online 19 September 2014
inflammatory lipid mediators in the process of tissue injury and rheumatoid arthritis makes this enzyme family as a therapeutic target of anti-inflammatory drugs [3]. Previously, animal experiments showed that some peptides and peptide analogues exhibited strong inhibitory capability to suppress the catalytic activity of PLA2 with low toxicity and high biocompatibility [4], suggesting that peptides can be considered as ideal lead entities to develop potent PLA2 inhibitors. Traditionally, bioactive peptides are primarily obtained by random approaches such as peptide library and phage display; these methods, however, are too timeconsuming and expensive to screen against all possible peptide combinations in a given length, largely limiting the practicability of therapeutic peptide development. Over the past decade, computational peptide design has spurred increasing interest in the medicinal community [5]. However, rational design of peptide ligands that can bind tightly to PLA2 receptor has long been a challenging task until the crystal structures of PLA2 as well as its complexes with diverse peptide ligands were solved at high-resolution level [6,7]. The important features of overall structure of PLA2 contain a N-terminal helix, an external loop, a calciumbinding loop, a second helix, a short anti-parallel b-wing, a third helix, and two single t-turns; such structure
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
2
P. Wang et al.
configuration defines a conserved active site within a hydrophobic channel lined by a number of invariant hydrophobic residues [8,9]; the wall of the channel contains several key catalytic residues such as His48 and Asp49 behind which a disulfide link involving Cys51–Cys93 stabilizes the channel architecture [10]. In the current study, based on the elucidated structure feature we performed systematic molecular modeling to investigate the interaction modes of human secretory PLA2 with a natural PLA2 inhibitor, Clara cell protein [11]. On this basis, various peptide fragments were derived from the putative protein interaction sites to define a distinct set of PLA2 inhibitory peptides, from which a number of candidates that were potential to target PLA2 were identified using a peptide redocking strategy with the guide of solved crystal structure data. Subsequently, enzymatic inhibition assay was conducted to determine the inhibitory capability of few highly promising peptides against PLA2. We also gave a systematic analysis of the structural basis, energetic property, and biological implication underlying the intermolecular interactions between the PLA2 catalytic domain and designed peptide inhibitors.
Materials and methods Crystal structure data The crystal structure of human secretory PLA2 was retrieved from the PDB database (PDB: 1KQU). The structure was solved in complex with a substrate analogue, the chiral precursor D-tyrosine, which is bound in the active site of the enzyme. In addition, two Ca2+ ions are chelated with the calcium-binding loop of PLA2 through a number of carboxylate and amide oxygen atoms, and multiple hydrophobic contacts and a T-shaped aromatic-group–His6 interaction are involved in the structure to stabilize PLA2 architecture [12] (Figure 1A). Previously, a variety of proteins have been reported to have inhibitory activity against PLA2, which could be considered as potential candidates to derive functional peptide fragments with PLA2 suppressing activity. For example, Thwin et al. [13] successfully designed a 17-residue PLA2 inhibitory peptide using the parent structure of a protein from Python serum.
J Drug Target, Early Online: 1–7
In this respect, we here attempted to exploit the crystal structure data of a natural PLA2 inhibitor Clara cell protein (CCP) [11], aiming to discover new potent PLA2 inhibitory peptides in this protein. The CCP is a 17-kDa protein secreted by Clara cells and other nonciliated cells of both the bronchiolar and the bronchial epithelium, which is a homodimer bound covalently by two disulfide bonds that joins the monomers together in an antiparallel manner such that the dimer encloses a large internal hydrophobic cavity (PDB: 1CCD) [14] (Figure 1B). Protein docking and peptide redocking Protein docking of CCP to PLA2 was carried out using ZDOCK method [15], which performed a full rigid-body search of docking orientations between two proteins in consideration of electrostatic interactions and solvent effect at the sampled protein–protein interface. Before performing the docking operation, the cocrystallized inhibitor ligand and calcium ions were removed from the PLA2 structure, and hydrogen atoms and protonation states were assigned for the CCP to PLA2 crystal structures using REDUCE [16] and PROPKA [17] programs, respectively. FlexPepDock server [18] was employed to redock isolated peptide fragments to PLA2. The method is a high-resolution peptide docking and refinement protocol implemented within the Rosetta framework, which has been shown to be able to accurately refine the peptide structure starting from the initial position nearby its native conformation, allowing full flexibility to the peptide and side-chain flexibility to protein receptor [19]. Molecular dynamics simulation and binding free energy analysis Docked complex systems (PLA2–CCP or PLA2–peptide complexes) were then subjected to molecular dynamics (MD) simulations using the AMBER03 force field [20]. In the MD simulation procedure, calcium ions were kept in PLA2 structure in order to stabilize the protein architecture. Complex was dissolved in a rectangular box full of TIP3P water molecules, and counterions of Na+ were placed to keep
Figure 1. The crystal structures of human secretory PLA2 (PDB: 1KQU) (A) and Clara cell protein (PDB: 1CCD) (B).
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
DOI: 10.3109/1061186X.2014.959019
the system neutral. The particle mesh Ewald (PME) strategy [21] and SHAKE algorithm [22] were used to treat the longrange electrostatic interactions and to constrain all hydrogen atoms, respectively. First, all hydrogen atoms, water molecules and counterions were minimized, and then the system was relaxed by 3000 cycles of minimization without constraints (500 steps of steepest gradient descent and 2500 steps of the conjugate gradient method). The subsequent MD simulations consisted of a gradual temperature increase to target 300 K over 100 ps and a 5-ns equilibration phase under the pressure of 1 atm for data collection, during which trajectory snapshots were saved every 100 ps. The binding free energy DG of CCP or peptide to PLA2 was calculated from the MD-derived snapshots of PLA2– CCP/peptide complex dynamics trajectory using the MM/ GBSA method [23], where the nonbonded interactions between PLA2 and CCP/peptide were computed with molecular mechanism (MM) approach, while the desolvation effect due to the PLA2–CCP/peptide binding was described by the generalization Born (GB) model (for polar contribution) and surface area (SA) strategy (for nonpolar contribu˚ , and the tion). The grid size for the GB calculations was 0.5 A values of interior and exterior dielectric constants were set to 1 and 80, respectively. Kinetic and inhibition studies Kinetic studies were performed to investigate the inhibition of PLA2 by designed peptides. The experimental protocol was described previously [10]. All assays were conducted in 20 mM glycylglycine buffer (pH 7.5, 30 C) containing 3 mM CaCl2 using 0.05 M cresol red as indicator. The enzyme fixed at concentration 1.5 mM was co-incubated with tested peptides for 2 h, and the amount of free enzyme was determined by the addition of phosphatidylcholine substrate. The volume of the
Targeting human secretory PLA2
3
substrate solution comprised 5% of the total reaction mixture. The initial velocity was calculated from the changes in an absorbance at 578 nm recorded with a Perkin–Elmer spectrophotometer.
Results and discussion PLA2–CCP docking sampling The ZDOCK method was employed to sample possible interaction modes between PLA2 and CCP [15]. The docking sampling procedure included a full rigid-body search phase of docking orientations between the two proteins and a refinement phase to optimize the structure models of complex architectures generated by the rigid-body docking. Consequently, various potential interactions were obtained directly from the ZDOCK server [24], and then we have oneby-one examined visually these docked modes in the PyMol graphics interface. It was found that most docked CCP locations were nearby the active site of PLA2, but the CCP protein can adopt multiple sites to touch on the PLA2 surface. Then, a number of modes were empirically identified as the promising interactions of PLA2 with CCP, where the sampled CCP conformations are clustered and superposed onto the PLA2 surface and shown in Figure 2(A). Since most of the existing PLA2 inhibitors are small peptides with short sequence lengths [25,26], we herein only considered CCP fragments of less than five amino acids derived from the interaction sites of docked PLA2–CCP complexes. The atomic contacts at the complex interface of PLA2 with each docked CCP conformation were analyzed using PROBE program [27] to determine the inter-residue interactions between PLA2 and CCP conformation. In this way, we extracted 11 CCP fragments that satisfied following criteria (Table 1): (i) their sequence lengths range between 3 and 5, (ii) all amino acid residues in a fragment should be
Figure 2. (A) Conformation cluster of docked CCP protein nearby the PLA2 active site. (B) An example of extracting peptide fragment LLLGS (13–17) from the interaction site of a docked CCP conformation with PLA2.
4
P. Wang et al.
J Drug Target, Early Online: 1–7
Table 1. The 11 peptide fragments extracting from the interaction sites of docked PLA2–CCP complexes.
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
Peptide fragment GFLQ LLLGS AALK PFNP NPAS NAGT LVDT LPQ IVK ILT PLCE
DG (kcal/mol)
Sequence position in CCP
PPI
PTI1
PTI2
6–8 13–17 23–26 27–30 29–32 36–39 44–47 48–50 56–58 63–65 67–70
22.8 34.5 28.7 46.4 28.0 41.4 34.7 35.6 28.6 21.3 25.5
8.6 12.4 14.5 5.6 6.8 7.6 12.0 7.8 12.6 11.3 9.7
12.3 17.4 19.7 11.0 10.6 12.2 15.9 14.5 16.2 15.3 13.7
PPI, the binding free energy of PLA2 to docked CCP conformation from which the corresponding peptide fragment was extracted. PTI1, the binding free energy of peptide fragment to PLA2 derived using the docked PLA2–CCP complex. PTI2, the binding free energy of peptide fragment to PLA2 derived using the redocked PLA2–peptide complex.
detected to be in atomic contact with the PLA2 partner, and (iii) the fragments should contain a considerable proportion of nonpolar residues due to the hydrophobic nature of PLA2 active site. An example of extracting peptide fragment LLLGS (13–17) from the interaction site of a docked CCP conformation with PLA2 is shown in Figure 2(B). Redocking of peptides into PLA2 active site Previously, Stein and Aloy [28] systematically examined various peptide-mediated protein interactions in three-dimensional crystal structures and found that the context of peptide motifs in protein–protein interfaces plays a crucial role in determining interaction stability and specificity. This finding suggested that the binding behavior of a peptide to its partner may differ significantly in isolated state and in protein context. In this respect, we performed peptide docking to separately redock all the 11 peptides into the active site of PLA2 by using the FlexPepDock server [18]; here the initial conformations of these peptides (the conformations that peptide fragments were cut out directly from the docked PLA2–CCP complexes) were used to guide the redocking procedure. Further, the 11 redocked peptide conformations were separately subjected to 5-ns MD simulations to equilibrate and refine their complex structures with PLA2. The redocked conformations of these peptides were observed to deviate considerably from their initial conformations, as the mean value of root-mean-square deviations (rmsd) of docked from initial conformations over the 11 ˚ . In contrast, MD simulations peptides is as much as 1.85 A can only address moderate or modest effect on peptide conformations in PLA2 active site, with the mean rmsd value ˚ between the 11 redocked and MD-equilibrated of 0.43 A conformations. For example, the initial, redocked and equilibrated conformations of pentapeptide LLLGS (corresponding to the sequence position 13–17 of CCP protein) bound with PLA2 are shown in Figure 3. It is seen that the redocked conformation has a large motion relative to initial position, but is very close to equilibrated conformation,
Figure 3. The initial, redocked, and MD-equilibrated conformations of peptide fragment LLLGS (13–17) in PLA2 active site. The initial conformation is that cut out directly from the interaction site of docked PLA2–CCP complex.
supporting that protein context contributes considerably to peptide binding. Next, the binding behavior of peptides to PLA2 was investigated at energetic level. The MM/GBSA method was employed to calculate the binding free energies between PLA2 and the 11 peptides using both initial and equilibrated conformations, as well as, for comparison purpose, to calculate the binding free energies of the docked CCP– PLA2 complexes that were used to derive the 11 peptides. The obtained free energy values DG are tabulated in Table 1. In energetic point of view, the extracted peptide fragments can only contribute about one-fifth of the total binding free energy to corresponding CCP–PLA2 complexes, ranging from 5 to 15 kcal/mol. However, after redocking and MD equilibration, the peptide binding affinities to PLA2 have a significant increase; their free energies were enhanced generally by two-fold. It is suggested that the redocked conformation, but not initial conformation, of peptides should be more close to their real conformation in isolated state; the MD simulations could further refine the atomic-level structures of peptide complexes with PLA2, which resulted in a significant improvement in peptide affinity. Thus, the MDequilibrated peptide conformations were adopted to perform subsequent structural and energetic analysis. Structure-based optimization of peptide inhibitors Three out of the 11 extracted peptides were selected to conduct kinetic studies; the three peptides (i.e. LLLGS, AALK, and IVK) were selected because they were calculated to have the highest theoretical affinities to PLA2 (DG ¼ 18.2, 19.7, and 16.2 kcal/mol, respectively). As can be seen, the three peptides share a common structure motif, that is, a polar or charged amino acid (S or K) occupied at their C-terminus, with a hydrophobic chain composed of two or three nonpolar amino acids (L, A, and V) to define their N-terminus. This structure motif is basically consistent
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
DOI: 10.3109/1061186X.2014.959019
with those known PLA2 inhibitory peptides reported previously [25,26]. As might be expected, the biological activities of the three peptides were determined to be moderate, with their inhibition constants Ki against PLA2 of 34.7 ± 3.1, 17.4 ± 1.5, and 152.5 ± 13.8 mM. The pentapeptide LLLGS and tetrapeptide AALK showed much higher activities as compared with the tripeptide IVK. This is anticipated because most known PLA2 inhibitory peptides are four or five amino acid residues long; too short sequences are unable to confer sufficient potency for peptide binding to PLA2. In order to further improve peptide activity, we here examined atomic interactions in the complex structure of PLA2 with LLLGS. The MD-equilibrated PLA2–LLLGS complex as well as nonbonded interactions across the complex interface are shown in Figure 4, from which it is evident that a number of van der Waals contacts and hydrophobic forces are formed between the nonpolar N-terminus of peptide ligand and the hydrophobic pocket
Targeting human secretory PLA2
5
around PLA2 active site, defining a stable complex architecture of PLA2 with the peptide ligand. However, there are no specific chemical forces such as hydrogen bonding and salt bridges observed at the complex interface, indicating that the recognition of peptide LLLGS by PLA2 is largely nonspecific that cannot be used to develop selective inhibitors. Thus, we performed a systematic amino acid scanning on each residue position of the pentapeptide ligand. The procedure is an extension of the widely used alanine scanning [29], where all the 20 amino acids, but not only the alanine, were in turn used to perform scanning. According to a previously described strategy [30], each residue of the pentapeptide was mutated to other 19 types of amino acids using SCAP program [31], followed by a AMBER force field minimization [32] to relax the mutated system. Consequently, we obtained 95 mutants as well as one template complex, based on which the mutation energy map of PLA2–LLLGS interaction is depicted in Figure 5, in which
Figure 4. (A) Stereoview of MD-equilibrated PLA2–LLLGS complex structure architecture. (B) Schematic representation of nonbonded interactions across the complex interface, generated by LIGPLOT program [34].
Figure 5. The mutation energy map of PLA2–LLLGS interaction. The mutation energies (DDG) are highlighted by binding energy changes upon systematic mutations of each position of pentapeptide LLLGS to other 19 amino acids.
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
6
P. Wang et al.
J Drug Target, Early Online: 1–7
Figure 6. The calculated binding free energy distribution of all the 243 combined pentapeptides to PLA2.
Table 2. The eight peptide candidates with binding free energies DG520 kcal/mol. Peptide
DG (kcal/mol)*
Ki (mM)y
VAYKS ALYYR LLAYK LIAFK AVFRS AVAFK LALYS LVAYS
22.4 20.9 23.8 22.9 20.3 22.1 22.4 23.8
NIz 107 ± 9 0.75 ± 0.06 NI 4.2 ± 0.3 NIz 18 ± 2 94 ± 8
NI, not inhibitory in the assay. *DG, calculated by MM/GBSA approach. yKi, measured by kinetic assay.
the mutation energies (DDG) are colored by binding energy changes upon systematic single-point mutations on each residue position. The mutation energy DDG for a specific peptide position upon a given mutation was defined as the difference between the PLA2 binding energies of mutant and the parent peptide (LLLGS) [33]. As can be seen from Figure 5, most mutations appear to reduce PLA2–peptide affinity (DDG40) – no matter the mutations were occurred at which positions, but this is not very significant for most mutants (DDG51 kcal/mol). By visually examining LLLGS and its mutants within the active pocket of PLA2, it was observed that these peptides are loosely bound with PLA2 to form a large nonpolar interface between them; mutations of just single residues seem not to be essential in affecting PLA2–peptide binding behavior, leading to only a slight affinity change upon the mutation. Here, we selected the top-3 favorable amino acids at each position of the pentapeptide ligand in terms of the mutation energy map, that is, P1: Ala, Leu, Val; P2: Val, Leu, Ile; P3: Ala, Leu, Val; P4: Tyr, Lys, Phe; P5: Lys, Arg, Ser. As can be seen, the N-terminal positions P1, P2, and P3 are favorable to bulk and/or hydrophobic amino acid types such as Leu, Ile, and Val, whereas the C-terminal positions P4 and P5 prefer to be occupied by polar or charged types such as Ser and Lys. This is in line with the basic structure feature of existing
PLA2 inhibitory peptides [25,26]. These favorable amino acids at each position were systematically combined to generate at most 35 ¼ 243 pentapeptides, and their complex structures with PLA2 were modeled from the PLA2–LLLGS complex using the SCAP virtual mutagenesis and AMBER minimization described above. Based on the modeled static structures, the binding free energies DG of the 243 pentapeptides to PLA2 were one-by-one calculated through the MM/GBSA approach [23], which are plotted in Figure 6. Evidently, only 11 out of 243 pentapeptides were estimated to bind more tightly than the parent LLLGS, with their DG518.2 kcal/mol, indicating that most mutations on LLLGS are unfavorable to peptide affinity. In order to select those highly promising candidates, we here only considered the peptides with DG520 kcal/mol and, consequently, eight samples satisfied this criterion and thus were selected, as tabulated in Table 2. The kinetic assay was performed to determine the inhibition potency of the eight peptides against PLA2, and obtained Ki values are also listed in Table 2. It is seen that five out of the eight peptides were measured to have high or moderate activity, with their Ki values ranging between 0.75 ± 0.06 and 107 ± 9 mM. Other three samples have no observable activity in the assay, which were supposed to possess relatively high hydrophobicity that may cause weak, non-specific interactions with PLA2 [35,36].
Declaration of interest The authors report that they have no conflicts of interest. This work was supported by the Social Development Project of Zhenjiang City (No. SH2013050).
References 1. Ashley NT, Weil ZM, Nelson RJ. Inflammation: mechanisms, costs, and natural variation. Annu Rev Ecol Evol Syst 2012;43: 385–406. 2. Burke JE, Dennis EA. Phospholipase A2 structure/function, mechanism, and signaling. J Lipid Res 2009;50:S237–42. 3. Meyer MC, Rastogi P, Beckett CS, McHowat J. Phospholipase A2 inhibitors as potential anti-inflammatory agents. Curr Pharm Des 2005;11:1301–12.
Journal of Drug Targeting Downloaded from informahealthcare.com by University of Hong Kong Libraries on 10/09/14 For personal use only.
DOI: 10.3109/1061186X.2014.959019
4. Mahalka AK, Kinnunen PK. Class specific peptide inhibitors for secretory phospholipases A2. Biochem Biophys Res Commun 2013;436:349–53. 5. Zhou P, Wang C, Ren Y, et al. Computational peptidology: a new and promising approach to therapeutic peptide design. Curr Med Chem 2013;20:1985–96. 6. Chandra V, Kaur P, Srinivasan A, Singh TP. Three-dimensional structure of a presynaptic neurotoxic phospholipase A2 from ˚ resolution. J Mol Biol 2000;296: Daboia russelli pulchella at 2.4 A 1117–26. 7. Chandra V, Jasti J, Kaur P, et al. Design of specific peptide inhibitors of phospholipase A2: structure of a complex formed between Russell’s viper phospholipase A2 and a designed peptide Leu-Ala-Ile-Tyr-Ser (LAIYS). Acta Crystallogr D 2002;58: 1813–19. 8. Singh RK, Ethayathulla AS, Jabeen T, et al. Aspirin induces its anti-inflammatory effects through its specific binding to phospholipase A2: crystal structure of the complex formed between phospholipase A2 and aspirin at 1.9 angstroms resolution. J Drug Target 2005;13:113–19. 9. Singh RK, Somvanshi N, Sharma S, et al. Structural elements of ligand recognition site in secretory phospholipase A2 and structurebased design of specific inhibitors. Curr Top Med Chem 2007;7: 757–64. 10. Singh RK, Vikram P, Makker J, et al. Design of specific peptide inhibitors for group I phospholipase A2: structure of a complex formed between phospholipase A2 from Naja naja sagittifera (group I) and a designed peptide inhibitor Val–Ala–Phe–Arg–Ser (VAFRS) ˚ resolution reveals unique features. Biochemistry 2003;42: at 1.9 A 11701–6. 11. Jorens PG, Sibille Y, Goulding NJ, et al. Potential role of Clara cell protein, an endogenous phospholipase A2 inhibitor, in acute lung injury. Eur Respir J 1995;8:1647–53. 12. Hansford KA, Reid RC, Clark CI, et al. D-tyrosine as a chiral precusor to potent inhibitors of human nonpancreatic secretory phospholipase A2 (IIa) with antiinflammatory activity. Chembiochem 2003;4:181–5. 13. Thwin MM, Ong WY, Fong CW, et al. Secretory phospholipase A2 activity in the normal and kainate injected rat brain, and inhibition by a peptide derived from python serum. Exp Brain Res 2003;150: 427–33. 14. Umland TC, Swaminathan S, Furey W, et al. Refined structure of ˚ resolution. J Mol Biol 1992; rat Clara cell 17 kDa protein at 3.0 A 224:441–8. 15. Chen R, Li L, Weng Z. ZDOCK: an initial-stage protein-docking algorithm. Proteins 2003;52:80–7. 16. Word JM, Lovell SC, Richardson JS, Richardson DC. Asparagine and glutamine: using hydrogen atom contacts in the choice of sidechain amide orientation. J Mol Biol 1999a;285:1735–47. 17. Rostkowski M, Olsson MH, Søndergaard CR, Jensen JH. Graphical analysis of pH-dependent properties of proteins predicted using PROPKA. BMC Struct Biol 2011;11:6. 18. London N, Raveh B, Cohen E, et al. Rosetta FlexPepDock web server – high resolution modeling of peptide-protein interactions. Nucleic Acids Res 2011;39:W249–53.
Targeting human secretory PLA2
7
19. Raveh B, London N, Schueler-Furman O. Sub-angstrom modeling of complexes between flexible peptides and globular proteins. Proteins 2010;78:2029–40. 20. Duan Y, Wu C, Chowdhury S, et al. A point-charge force field for molecular mechanics simulations of proteins based on condensedphase quantum mechanical calculations. J Comput Chem 2003;24: 1999–2012. 21. Darden T, York D, Pedersen L. Particle mesh Ewald and N.log(N) method for Ewald sums in large systems. J Chem Phys 1993;98: 10089–92. 22. Ryckaert J, Ciccotti G, Berendsen HJC. Numerical-integration of Cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J Comput Phys 1977;23:327–41. 23. Kollman PA, Massova I, Reyes C, et al. Calculating structures and free energies of complex molecules: combining molecular mechanics and continuum models. Acc Chem Res 2000;33:889–97. 24. Pierce BG, Wiehe K, Hwang H, et al. ZDOCK server: interactive docking prediction of protein-protein complexes and symmetric multimers. Bioinformatics 2014;30:1771–3. 25. Church WB, Inglis AS, Tseng A, et al. A novel approach to the design of inhibitors of human secreted phospholipase A2 based on native peptide inhibition. J Biol Chem 2001;276:33156–64. 26. Thwin MM, Satyanarayanajois SD, Nagarajarao LM, et al. Novel peptide inhibitors of human secretory phospholipase A2 with antiinflammatory activity: solution structure and molecular modeling. J Med Chem 2007;50:5938–50. 27. Word JM, Lovell SC, LaBean TH, et al. Visualizing and quantifying molecular goodness-of-fit: small-probe contact dots with explicit hydrogen atoms. J Mol Biol 1999b;285:1711–33. 28. Stein A, Aloy P. Contextual specificity in peptide-mediated protein interactions. PLoS One 2008;3:e2524. 29. Morrison KL, Weiss GA. Combinatorial alanine-scanning. Curr Opin Chem Biol 2001;5:302–7. 30. Bi J, Yang H, Yan H, et al. Knowledge-based virtual screening of HLA-A*0201-restricted CD8+ T-cell epitope peptides from herpes simplex virus genome. J Theor Biol 2011;281:133–9. 31. Xiang Z, Honig B. Extending the accuracy limits of prediction of side-chain conformations. J Mol Biol 2001;311:421–30. 32. Hou T, Chen K, McLaughlin WA, et al. Computational analysis and prediction of the binding motif and protein interacting partners of the Abl SH3 domain. PLoS Comput Biol 2006;2:e1. 33. Ni Z, Chen H, Qi X, Jin R. Why is substrate peptide binding unsusceptible to multidrug-resistant mutations in HIV-1 protease? A structural and energetic analysis. Int J Pept Res Ther 2014;20: 43–51. 34. Wallace AC, Laskowski RA, Thornton JM. LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions. Protein Eng 1995;8:127–34. 35. Zhou P, Zou J, Tian F, Shang Z. Fluorine bonding – how does it work in protein-ligand interactions? J Chem Inf Model 2009;49: 2344–55. 36. Zhou P, Huang J, Tian F. Specific noncovalent interactions at protein-ligand interface: implications for rational drug design. Curr Med Chem 2012;19:226–38.
