Int J Pept Res Ther DOI 10.1007/s10989-016-9516-x

Rational Improvement of Peptide Affinity to Human PregnancyRelated Serine Protease HtrA3 PDZ Domain by Introducing a Halogen Bond to the Domain–Peptide Complex Interface Hong Liu1 • Shuo-Fen Dou2 • Xue Zhang1 • Yan Wang1 • Qing-Li Wen1 Ya-Nan Mu3

•

Accepted: 9 February 2016 Ó Springer Science+Business Media New York 2016

Abstract The pregnancy-related serine protease HtrA3 plays an important role in human placental development and has recently been recognized as a potential therapeutic target in the treatment of cancer. Previously, a C-terminal pentapeptide FGRWV–COOH was identified to bind at the PDZ domain of HtrA3 with a moderate affinity. Here, based on the high-resolution complex crystal structure of HtrA3 PDZ domain with the pentapeptide ligand we successfully introduced a rationally designed halogen bond to the complex interface by substituting R4-hydrogen atom of the indole moiety of peptide Trp-1 residue with a halogen atom. High-level theoretical calculations suggested that bromine is the best choice that can render strong interaction energy for the halogen bond and can confer high affinity to the PDZ–peptide complex. Fluorescence spectroscopy characterizations revealed that the resulting R4-brominated peptide (Kd = 0.15 ± 0.03 lM) exhibited 12-fold affinity improvement relative to its nonhalogenated counterpart (Kd = 1.8 ± 0.4 lM). In contrast, the PDZ-binding affinity of R6-brominated peptide (Kd = 1.2 ± 0.1 lM), a negative control that was unable to form the halogen bond according to theoretical investigations, did not change substantially as compared to the nonhalogenated peptide.

& Shuo-Fen Dou [email protected] 1

Department of Obstitrcs, Affiliated Hospital of Weifang Medical University, Weifang 261031, China

2

Catheter Room, Affiliated Hospital of Weifang Medical University, Weifang 261031, China

3

Department of Ophthalmology, Affiliated Hospital of Weifang Medical University, Weifang 261031, China

Keywords Pregnancy-related serine protease  Peptide  Halogen bond

Introduction Human HtrA3, also known as pregnancy-related serine protease (PRSP), is a member of the high temperature requirement A family of homo-oligomeric serine proteases; the family proteins are highly conserved in evolution and whose primary function is to maintain biological quality control (Nie et al. 2006). The HtrA3 was initially identified in the developing placenta both in the mouse and human as a serine protease associated with pregnancy (Nie et al. 2003) and later its diverse functions including cell proliferation, migration and apoptosis have been revealed (Singh et al. 2012). The protein is characterized by the presence of a catalytic serine protease domain (PD) followed by a regulatory post-synaptic density 95, Drosophila discs large, zona occludens-1 (PDZ) domain, which has been shown to act as substrate specificity determinant by binding to the C-terminal hydrophobic stretch of its protein partners, leading to structural change in PD domain and enzyme activation (Glaza et al. 2015). Disruption of the HtrA3 PDZ–partner interaction can inactivate the enzyme and thus would be a potential therapeutic strategy to a variety of relevant diseases such as cancer, arthritis, neurodegenerative and neuromuscular disorders, and age-related macular degeneration. Previously, Runyon et al. have identified a pentapeptide ligand of HtrA3 PDZ domain using phage display (Runyon et al. 2007). Crystallographic analysis revealed that the ligand can bind at the peptide-binding pocket of the PDZ domain to compete with the domain’s cognate partners for the pocket, thus competitively disrupting the HtrA3 PDZ–

123

Int J Pept Res Ther

partner interaction. However, the peptide can only bind weakly to the domain with a moderate affinity. Here, we reported a successful improvement of the peptide affinity by introducing a rationally designed halogen bond to the complex interface of HtrA3 PDZ domain with the pentapeptide ligand. In the procedure, the highlevel ab initio quantum chemistry theory and hybrid quantum mechanics/molecular mechanics (QM/MM) approach were employed to systematically and accurately characterize the structural basis and energetic property of isolated halogen bonds as well as them in the context of PDZ–peptide complexes. The binding affinity of several designed halogenated peptides to the HtrA3 PDZ domain was also determined in vitro using fluorescence spectroscopy.

Materials and Methods Crystal Structure of HtrA3 PDZ–Pentapeptide Complex The complex crystal structure of human HtrA3 PDZ domain with a pentapeptide ligand Phe-4Gly-3Arg-2Trp-1 Val-COOH was retrieved from the PDB database (Berman 0 et al. 2000) under the accession ID 2P3W. In the crystal structure, the pentapeptide is fused to the C-terminus of PDZ domain and a glycine triplet (GlyGlyGly) serves as the flexible linker between the domain and the self-fused

Fig. 1 a Crystallographic unit cell of the complex crystal structure of human HtrA3 PDZ domain with a pentapeptide self-fused to the C-terminus of the molecule in the nearby asymmetric unit (PDB: 2P3W). b An asymmetric unit showing the HtrA3 PDZ domain–

123

peptide. The PDZ domain binds to the C-terminally fused peptide of the molecule in the nearby asymmetric unit (Fig. 1a). The pentapeptide was identified from a random peptide library displayed in a high-valency format by fusion to the C or N terminus of the M13 major coat protein expressed on enterobacteria phage surface, which can bind to HtrA3 PDZ domain with affinity at micromolar level (Runyon et al. 2007). Hybrid Quantum Mechanics/Molecular Mechanics (QM/MM) A two-layered QM/MM scheme (Senn and Thiel 2009) was used to refine the complex structures of HtrA3 PDZ domain with halogenated peptides. The QM/MM calculations were carried out using ONIOM algorithm (Chung et al. 2015). A method described by Lu et al. (2009a) was employed to partition the protein–peptide system involving halogen bond, that is, the protein residue and the peptide residue that form halogen bond between them were in the QM layer, while the rest atoms were in the MM layer. Hydrogen atoms were used as link atoms to saturate the dangling bonds. The QM layer was modeled by the density functional theory (DFT) of MPWLYP (Adamo and Barone 1998) in conjunction with 6-31G(d) basis set, which has been testified to provide accuracy close to high-level methods in a benchmark study of halogen bonding (Lu et al. 2009b), and the MM layer was described by means of the molecular force field AMBER theory (Duan et al.

pentapeptide complex. The peptide is folded into a b-strand that adds at the edge of the PDZ b-sheet composed of three b1, b2 and b3 strands

Int J Pept Res Ther

2003). The restricted electrostatic-potential (RESP) fitting procedure (Bayly et al. 1993) and generalized amber force field (GAFF) (Wang et al. 2004) were utilized to derive partial atomic charges and force field parameters for halogenated amino acid residues, respectively. The intermolecular interaction energy (DEint) of protein–peptide complex was calculated with a strategy described previously (Guo et al. 2012). The PoissonBoltzmann/surface area (PB/SA) method (Fogolari et al. 2002) was employed to estimate solvent effect associated with protein–peptide binding. In the PB/SA procedure, total desolvation free energy (DGslv) was computed from the polar (electrostatic) (DGplr) and nonpolar (DGnplr) desolvation energies. The polar aspect was calculated by finite difference solutions to the nonlinear Poisson-Boltzmann equation in DELPHI program (Rocchia et al. 2001), while nonpolar contribution was determined by summing up the weighted surface area of solute molecule (Sitkoff et al. 1994). The total binding free energy of a peptide ligand to its protein receptor can be computed as follows: DGtotal ¼ DEint þ DGplr þ DGnplr

ð1Þ

Ab Initio Electron Correlation Calculations The protein residue and the peptide residue that form halogen bond between them were striped from the QM/ MM-refined PDZ–peptide complex system, which were then used to calculate the interaction energy (DEhb) of halogen bond at a high electron correlation theory level of Møller-Plesset second order perturbation (MP2) in conjunction with a Dunning’s correlation consistent basis set aug-cc-pVDZ (for fluorine, chlorine and bromine) or augmented Lanl2DZ basis set (for iodine). The basis set superposition error (BSSE) was eliminated by the standard counterpoise method of Boys and Bernardi (1970). Fluorescence Spectroscopy Assay The nonhalogenated, R4-brominated and R6-brominated peptides were synthesized using Fmoc-solid phase synthesis protocol; the Fmoc halogenated amino acids used in the synthesis were obtained from a library of unnatural amino acids. The GST-tagged recombinant protein of human HtrA3 PDZ domain (residues 354–453) was cloned, expressed and purified following a standard protocol. The binding affinity between the protein and peptide was determined using fluorescence spectroscopy as described previously (Liu et al. 2011). The fluorescence emission spectra of fluorescently active residues around the peptidebinding pocket of PDZ domain were used to monitor change in their environment upon peptide binding. Changes in fluorescence were measured upon titration of peptide

solution. Peptide affinity Kd was estimated using below equation: F ¼ ½F0 þ F1 ð½p=Kd Þ=½1 þ ð½p=Kd Þ

ð2Þ

where [p] is peptide concentration; F and F0 are the fluorescence intensities of solution with and without peptide, respectively; F? is the maximal intensity of solution with saturated peptide. Each assay was performed for duplicate.

Results and Discussion The complex crystal structure of human HtrA3 PDZ domain with a phage display-derived pentapeptide Phe-4 Gly-3Arg-2Trp-1Val–COOH was examined in detail. The 0 structured peptide ligand is folded into a b-strand that adds at the edge of PDZ b-sheet composed of three b1, b2 and b3 strands (Fig. 1b). In particular, a hydrogen bond was observed between the backbone carbonyl group =O of PDZ Ile359 residue and the backbone amide group >N–H of peptide C-terminal Val0 residue. A lone electron pair of the carbonyl oxygen atom is involved in the hydrogen bond, while another lone pair of the oxygen points to the hydrogen atom H at the R4 position of indole moiety of peptide Trp-1 residue (Fig. 2a, b). In crystal structure the inter-atomic distance between the oxygen and hydrogen is ˚ , which is just within the range (2.5–3.5 A ˚ ) of stan2.8 A dard halogen bond lengths (Lu et al. 2009b). Here, we proposed that the hydrogen atom at R4 position of peptide Trp-1 residue can be replaced by a halogen atom X (X = F, Cl, Br or I) to define a geometrically satisfactory halogen bond between the X and the carbonyl oxygen O of PDZ Ile359 residue. Previously, Voth et al. found that halogen bond is shown to be geometrically perpendicular to and energetically independent of hydrogen bond; they can share a common carbonyl oxygen acceptor in biological context (Voth et al. 2009). Thus, the halogen bonding with Trp-1 and the hydrogen bonding with Val0 co-define an orthogonal molecular interaction system that work synergistically to promote PDZ–peptide binding (Fig. 2c). Subsequently, the R4-hydrogen atom of co-crystallized peptide was manually replaced by F, Cl, Br and I, resulting in four coarse-grained complex structures of HtrA3 PDZ domain with, respectively, the R4-fluoridated, R4-chlorinated, R4-brominated and R4-iodizated peptides, which were then refined via QM/MM minimization approach. This is an exhaustive computational procedure but can properly reflect the conformational changes due to introducing halogen atoms. Based on QM/MM-refined structures the geometrical and energetic parameters of these designed halogen bonds were calculated and listed in Table 1. According to the theoretical calculations, halogen bonds can indeed be formed in these halogenated complex

123

Int J Pept Res Ther

Fig. 2 a HtrA3 PDZ–pentapeptide complex crystal structure. b Zoom up of the region where a halogen bond is expected to create. c Schematic representation of the orthogonal molecular interactions between the native hydrogen bond and the designed halogen bond

Table 1 Geometric and energetic parameters of rationally designed halogen bonding systems X

˚ )a Length d(XO) (A

DEhb (kcal/mol)b

DGtotal (kcal/mol)c

F

2.57

-1.8

-13.4

Cl

2.74

-2.9

-15.8

Br

2.92

-3.5

-18.7

I

3.15

-3.2

-16.9

a

The equilibrated distance between the X atom of peptide Trp-1 residue and the carbonyl O atom of HtrA3 PDZ Ile359 residue b

The interaction energy of halogen bonds calculated at electron correlation level

c

The QM/MM derived total binding free energy of peptide ligand to HtrA3 PDZ domain

systems, as equilibrated bond lengths d(XO) and calculated interaction energies DEhb are consistent with those biological halogen bonds characterized previously (Lu et al. 2009a). The bond length increases following the order I [ Br [ Cl [ F, while the interaction energy reaches at its maximum with the Br (DEhb = -3.5 kcal/mol). According to high-level calculations performed on small model system (Lu et al. 2009b), halogen bond involving I atom should be stronger than that involving Br under perfect conditions. However, the complicated biological context would address additional effect on these halogen bond systems; the introduction of a bulky I atom to the small R4 position of peptide Trp-1 residue would cause a significant motion in the residue’s indole moiety (Fig. 3), leading to a large conformation adjustment of bound peptide ligand to response the motion. Thus, this is a compromise between the local rearrangement of halogen bond involving I and the global optimization of PDZ–peptide

123

Fig. 3 Superposition of QM/MM-refined conformations of R4-fluoridated, R4-chlorinated, R4-brominated and R4-iodizated Trp-1 in the context of HtrA3 PDZ–pentapeptide complex

interaction. Other three halogen elements (F, Cl and Br) are volumetrically much smaller than I and do not cause significant steric hindrance in the complex. The total binding free energies DGtotal of four halogenated peptides as well as their nonhalogenated counterpart (i.e. the native pentapeptide) were calculated using QM/MM-PB/SA scheme and tabulated in Table 1. As can be seen, R4-brominated peptide was predicted to have the highest binding potential in all the five investigated systems, of which the binding free energy (DGtotal = -18.7 kcal/mol) is improved by *6 kcal/mol as compared to the nonhalogenated peptide (DGtotal = -12.5 kcal/mol). The R4-fluoridated peptide exhibits a binding free energy (DGtotal = -13.4 kcal/mol) similar to that of nonhalogenated version. This is expected because the fluorine atom cannot form halogen bond in most cases and thus

Int J Pept Res Ther Table 2 The calculated free energies and measured binding affinities of nonhalogenated, R4-brominated and R6-brominated peptides to HtrA3 PDZ domain

Fig. 4 Superposition of QM/MM-refined conformations of R4-fluoridated, R4-chlorinated, R4-brominated and R4-iodizated Trp-1 extending between the side chains of HtrA3 PDZ residues Glu390 and Ala392

substitution of the hydrogen atom at Trp-1 R4 position by F cannot establish an effective halogen bond in the complex system. However, considering that the fluorine is a small atom that does not influence the PDZ–peptide binding mode essentially, the total binding free energy DGtotal only changes very modestly due to introducing the F. In contrast, the I, as discussed above, can cause significant steric hindrance at the complex interface, thus unfavorable to peptide binding, although strong halogen bonding force is created by presence of the I (DEhb = -3.2 kcal/mol) that can overcome the unfavorable effect. In addition, the Cl seems also to be a potent enhancer for the PDZ–peptide system, which can improve peptide potency considerably; the effect of introducing Cl on peptide binding is between that of introducing Br and I. In addition to the pentapeptide FGRWV–COOH studied here, previously a number of peptide ligands such as RSWWV–COOH, FGRWI–COOH and FGAWV–COOH were also selected for the HtrA3 PDZ domain by using phage display technique (Runyon et al. 2007). The Trp-1 residue is highly conserved across these potent binders, indicating an important role of the residue in HtrA3 PDZ–peptide recognition and association. By examining crystal structure it is observed that the indole ring of Trp-1 extends between the side chains of HtrA3 PDZ residues Glu390 and Ala392 that may confer high stability to the domain–peptide complex. Here, we investigated the structural effect of halogenations on the complex system. As can be seen in Fig. 4, the R4-substituted halogen atoms do not point to and thus cannot interact directly with the two residues. The QM/MM refinement also supports that the halogenations can only change modestly the conformation and location of

Pentapeptide ligand

DGtotal (kcal/mol)a

Kd (lM)b

Nonhalogenated peptide

-12.5

1.8 ± 0.4

R4-brominated peptide

-18.7

0.15 ± 0.03

R6-brominated peptide

-13.3

1.2 ± 0.1

a

The QM/MM derived total binding free energy of peptide ligand to HtrA3 PDZ domain

b

The dissociation constant determined by fluorescence spectroscopy

Trp-1 side chain between the domain residues Glu390 and ˚ Ala392, with root-mean-square deviation (rmsd) \0.3 A among the different versions of halogenated indole rings. All these suggest that the halogen substitution does not influence the interaction behavior of Trp-1 with Glu390 and Ala392 substantially. In order to substantiate the computational designs and findings, three peptides as listed in Table 2 were tested by fluorescence spectroscopy to determine their binding affinity for HtrA3 PDZ domain. The nonhalogenated counterpart was considered as a reference for those halogenated peptides; the R4-brominated peptide was suggested to possess the highest binding free energy in all investigated halogenated peptides, which can form a strong halogen bond with the domain but does not cause significant steric hindrance in the PDZ–peptide system; the R6brominated peptide is a negative control because the R6 position of indole moiety of peptide Trp-1 residue points out of the peptide-binding pocket of PDZ domain and is far away from any PDZ residue in complex crystal structure, and thus substituting a Br atom at this position was expected not to influence PDZ–peptide binding substantially. The three peptides were synthesized and purified, and their binding affinity to the GST-tagged recombinant protein of human HtrA3 PDZ domain (residues 354–453)

123

Int J Pept Res Ther

was measured in vitro using fluorescence spectroscopy. The changes in fluorescence emission spectra upon peptide titration were fitted to Eq. (2) to measure the effect of peptide binding on domain structure, which reflects the binding strength of peptide ligands to domain receptor. The measured dissociation constants Kd of the three tested peptides are listed Table 2. As might be expected, the nonhalogenated peptide can bind to PDZ domain with a moderate affinity (Kd = 1.8 ± 0.4 lM), which is in line with a previously reported value (Kd = 1.1 lM) for the peptide measured using isothermal titration calorimetry (ITC) (Runyon et al. 2007). Substituting a Br atom at the R4 position of peptide Trp-1 residue, which results in the R4-brominated peptide, can considerably improve the peptide affinity (Kd = 0.15 ± 0.03 lM) by 12-fold as compared to the nonhalogenated counterpart, suggesting that the bromine atom should play a critical role in promoting PDZ–peptide binding. In contrast, the affinity of R6-brominated peptide (Kd = 1.2 ± 0.1 lM) does not change substantially relative to the nonhalogenated peptide. All these suggest that a halogen bond, as expected, is formed between the indole Br atom of R4-brominated peptide and the carbonyl O atom of PDZ Ile359 residue, which can confer potent affinity and specificity to the PDZ–peptide complex system. Acknowledgments This work was supported by the Weifang Medical University. Compliance with Ethical Standards Conflict of Interest

None.

Statement of Informed Consent Not applicable. Statement of Human and Animal Rights

Not applicable.

References Adamo C, Barone V (1998) Exchange functionals with improved longrange behavior and adiabatic connection methods without adjustable parameters: the mPW and mPW1PW models. J Chem Phys 108:664–675 Bayly CI, Cieplak P, Cornell WD, Kollman PA (1993) A wellbehaved electrostatic potential based method using charge restraints for deriving atomic charges: the RESP model. J Phys Chem 97:10269–10280 Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, Shindyalov IN, Bourne PE (2000) The protein data bank. Nucl Acids Res 28:235–242 Boys SF, Bernardi F (1970) The calculation of small molecular interactions by the differences of separate total energies. Some procedures with reduced errors. Mol Phys 19:553–566

123

Chung LW, Sameera WM, Ramozzi R, Page AJ, Hatanaka M, Petrova GP, Harris TV, Li X, Ke Z, Liu F, Li HB, Ding L, Morokuma K (2015) The ONIOM method and its applications. Chem Rev 115:5678–5796 Duan Y, Wu C, Chowdhury S, Lee MC, Xiong G, Zhang W, Yang R, Cieplak P, Luo R, Lee T, Caldwell J, Wang J, Kollman P (2003) A point-charge force field for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. J Comput Chem 24:1999–2012 Fogolari F, Brigo A, Molinari H (2002) The Poisson-Boltzmann equation for biomolecular electrostatics: a tool for structural biology. J Mol Recognit 15:377–392 Glaza P, Osipiuk J, Wenta T, Zurawa-Janicka D, Jarzab M, Lesner A, Banecki B, Skorko-Glonek J, Joachimiak A, Lipinska B (2015) Structural and functional analysis of human HtrA3 protease and its subdomains. PLoS One 10:e0131142 Guo X, He D, Huang L, Liu L, Liu L, Yang H (2012) Strain energy in enzyme-substrate binding: an energetic insight into the flexibility versus rigidity of enzyme active site. Comput Theor Chem 995:17–23 Liu W, Wen W, Wei Z, Yu J, Ye F, Liu CH, Hardie RC, Zhang M (2011) The INAD scaffold is a dynamic, redox-regulated modulator of signaling in the Drosophila eye. Cell 145:1088–1101 Lu Y, Shi T, Wang Y, Yang H, Yan X, Luo X, Jiang H, Zhu W (2009a) Halogen bonding: a novel interaction for rational drug design? J Med Chem 52:2854–2862 Lu Y, Zou J, Fan J, Zhao W, Jiang Y, Yu Q (2009b) Ab initio calculations on halogen-bonded complexes and comparison with density functional methods. J Comput Chem 30:725–732 Nie G, Li Y, Minoura H, Batten L, Ooi GT, Findlay JK, Salamonsen LA (2003) A novel serine protease of the mammalian HtrA family is up-regulated in mouse uterus coinciding with placentation. Mol Hum Reprod 9:279–290 Nie G, Li Y, Hale K, Okada H, Manuelpillai U, Wallace EM, Salamonsen LA (2006) Serine peptidase HTRA3 is closely associated with human placental development and is elevated in pregnancy serum. Biol Reprod 74:366–374 Rocchia W, Alexov E, Honig B (2001) Extending the applicability of the nonlinear Poisson-Boltzmann equation: multiple dielectric constants and multivalent ions. J Phys Chem 105:6507–6514 Runyon ST, Zhang Y, Appleton BA, Sazinsky SL, Wu P, Pan B, Wiesmann C, Skelton NJ, Sidhu SS (2007) Structural and functional analysis of the PDZ domains of human HtrA1 and HtrA3. Protein Sci 16:2454–2471 Senn HM, Thiel W (2009) QM/MM methods for biomolecular systems. Angew Chem Int Ed Engl 48:1198–1229 Singh H, Makino S, Endo Y, Li Y, Stephens AN, Nie G (2012) Application of the wheat-germ cell-free translation system to produce high temperature requirement A3 (HtrA3) proteases. Biotechniques 52:23–28 Sitkoff D, Sharp KM, Honig B (1994) Accurate calculation of hydration free energies using macroscopic solvent models. J Phys Chem 98:1978–1988 Voth AR, Khuu P, Oishi K, Ho PS (2009) Halogen bonds as orthogonal molecular interactions to hydrogen bonds. Nat Chem 1:74–79 Wang J, Wolf RM, Caldwell JW, Kollman PA, Case DA (2004) Developing and testing of a general amber force field. J Comput Chem 25:1157–1174