Molecular modelling, synthesis and biological evaluation of peptide inhibitors as anti-angiogenic agent targeting neuropilin-1 for anticancer application

Vascular endothelial growth factor (VEGF) and its co-receptor neuropilin-1 (NRP-1) are important targets of many pro-angiogenic factors. In this study, nine peptides were synthesized and evaluated for their molecular interaction with NRP-1 and compared to our previous peptide ATWLPPR. Docking study showed that the investigated peptides shared the same binding region as shown by tuftsin known to bind selectively to NRP-1. Four pentapeptides (DKPPR, DKPRR, TKPPR and TKPRR) and a hexapeptide CDKPRR demonstrated good inhibitory activity against NRP-1. In contrast, peptides having arginine residue at sites other than the C-terminus exhibited low activity towards NRP-1 and this is confirmed by their inability to displace the VEGF165 binding to NRP-1. Docking study also revealed that replacement of carboxyl to amide group at the C-terminal arginine of the peptide did not affect significantly the binding interaction to NRP-1. However, the molecular affinity study showed that these peptides have marked reduction in the activity against NRP-1. Pentapeptides having C-terminal arginine showed strong interaction and good inhibitory activity with NRP thus may be a good template for anti-angiogenic targeting agent.


Introduction
Despite recent advances in surgery, chemotherapy and radiation treatment, survival of patients with advanced malignancy remains suboptimal. Photodynamic therapy (PDT) is now regarded as a promising clinical treatment modality in cancer. PDT involves the combined action of a photosensitizer, visible light of an appropriate wavelength and molecular oxygen to produce reactive oxygen species, particularly singlet oxygen that plays the main role in mediating cellular death. The effectiveness of PDT could be enhanced by combining peptide inhibitors that target angiogenesis with the treatment regimen.
Angiogenesis, the formation of new blood vessels from the pre-existing microvasculature is important in the pathogenesis of malignant, infectious, fibro-proliferative and inflammatory diseases. In cancer, angiogenesis occurs in the tumour surrounding tissues to form a novel vascular tube-like structure to supply nutrients and oxygen to maintain the viability of tumour progression and also to facilitate metastatic spreading into other parts of the body.
Tuftsin is an immunostimulatory peptide that can be used to enhance the immunogenicity of proteins (Gao et al., 2015). It is a tetrapeptide (Thr-Lys-Pro-Arg) produced by enzymatic cleavage of the Fc-domain of the heavy chain of immunoglobulin G. It stimulates the phagocytic activity of polymorphonuclear leukocytes. At this time, as far as we know, no clinical trials are performed with tuftsin. In the most recent research studies, tuftsin is used for the treatment of rheumatoid arthritis (Jain, Tran, & Amiji, 2015), for the modulation of murine lupus nephritis (Bashi et al., 2015) and a macrophage-targeting molecule (Horváti et al., 2014).
Tuftsin was shown to bind only at the b1 domain of NRP-1 in the conserved interstrand loops where it formed interactions with residues Y297 (of the first ligand binding loop, loop I), D320 (loop II), and S346, T349 and Y353 (loop III). These residues were also required for VEGF binding as shown by the overlapping binding competition study between tuftsin and mutational study of VEGF at the loop of residues S346, E348 and T349 (Vander Kooi et al., 2007).
Previously in our group, a peptide-conjugated photosensitizer (PS), TPC-Ahx-ATWLPPR targeting NRP-1 had showed an affinity for endothelial cells of tumour neovasculatures (Tirand et al., 2006). The molecule showed enhanced uptake and photodynamic properties compared to its nonconjugated counterpart. However, the peptide moiety of the conjugate degraded rapidly and was relatively unstable. Thus, the aim of the present study is to search for novel and stable peptide which could be conjugated to a photosensitizer and could act on NRP-1. A total of nine peptides of various sequence length (DKPRR, CDKPRR, TKPRR, TKPRR-NH 2 , APQPRPL, CPQPRPL, TKPPR, DKPPR, DKPPR-NH 2 ) were synthesized and evaluated for their binding affinity towards recombinant NRP-1 through in vitro molecular affinity study.
Among some peptides known to modulate VEGFdependent angiogenesis (Table 1), we chose some criteria: (1) the sequence (as short as possible); (2) the C-terminal residue generally involved in the recognition; (3) the hydrophilic/hydrophobic balance. Starting from tuftsin, we decided to change its C-terminal part by the PPR sequence of ATWLPPR. However, this PPR sequence being subject to turns mediated by cis/cis conformations, we also tested TKPRR (the choice of an additional arginine being related to potential additional hydrogen bonds). We also saw the option of a change at the C-terminus arginine either by amidification or coupling to another amino-acid. Concerning the N-terminus of the sequence, we chose an aspartate instead of threonine for hydrophilic and potential other hydrogen bonds.
Prior to that, the binding interaction of these peptides was investigated using molecular docking technique.  Giordano et al. (2005) Five peptides (DKPRR, DKPPR, TKPRR, TKPPR and CDKPRR) showed better affinity than ATWLPPR (Thomas et al., 2010;Tirand et al., 2006). Of these five peptides, two are novel i.e. DKPPR and TKPRR and were then chosen to be conjugated photosynthesizer molecules (Kamarulzaman et al., 2015). It is hoped that the information from this finding will provide deeper understanding towards the design of more active peptides as anti-angiogenic targeting agents.

Molecular docking
The three-dimensional (3D) structure of tuftsin (TKPR) was taken from the crystal structure along with the NRP-1 receptor (PDB ID: 2ORZ) (Vander Kooi et al., 2007). For the peptides (Figure 1), the structures were built using HyperChem Pro 6.0 (Hypercube Inc., Gainesville, USA) based on the 3D structure of tuftsin and their geometry was optimized with HyperChem using steepest descents and conjugate gradient algorithms (termination conditions set to a maximum of 500 cycles or .1 kcal/Å mol rms gradient). For example, for TKPRR, an additional of linear peptide chain was added to the structure at the C-terminal while for DKPRR, the tyrosine residue was mutated to an aspartic acid residue and the final structure was optimized as described above.
Both the ligands and NRP-1 receptor were prepared for docking using AutoDockTools (ADT) (Sanner, 2005). The protonation for the ligands and NRP-1 were assigned using ADT where polar hydrogen atoms were added to all ligands and receptor. The protonation states of an amino acid side chains assigned depend on its environment in the protein. The flexibility of peptides which is contributed to the calculation of the torsional free energy was assigned with AutoTors. Since Auto-Dock can only accommodate 32 torsions in each docking run, the number of active bonds was reduced. This was carried out by first examining the crystal structure of tuftsin-bound to NRP-1. The side chain and the functional group of tuftsin that contributed in the binding with NRP-1 were identified in order to decide which bond to be rotated. In general, all the amide bonds were assigned non-rotatable while the side chains of the peptides are kept flexible. The example of selection of active torsion is provided in the supplementary information (Supplementary Figure S1).
Default Kollman charges and solvation parameters were assigned to the protein atoms (Weiner et al., 1984) and Gasteiger charges (Gasteiger & Marsili, 1980) were added to each ligand atom. A grid box with the size of 70 × 70 × 70 points spaced .375 Å apart and centred on the tuftsin's centre of mass (−24.166, 7.136, −27.591) at the NRP-1 active site was used for all Autodock 3.0.5 docking runs (Morris et al., 1998b). The parameters of the Lamarckian Genetic Algorithm (LGA) were as follows: population size of 50, 1,500,000 energy evaluations, 200 search runs. The ligand pose with the lowest predicted free energy of binding, chosen from the most populated cluster, was used in subsequent analysis.

Chemicals and reagents
The Fmoc-amino acid-Wang resin and all other Fmocamino acid-OH were purchased from Senn Chemicals International (Gentilly, France). Reagents and solvents were purchased from Sigma-Aldrich (Saint-Quentin Fallavier, France). NRP-1 and KDR recombinant chimeric proteins and its reagents were purchased from R&D Systems, Lille, France.

Instruments
The peptides were synthesized using an automated ResPepXL peptide synthesizer (Intavis AG, Bioanalytical Instruments) and operated with a Multiple-Parallel Peptide Synthesis Program. The peptides were purified using Shimadzu LC-10ATvp, column C18 reverse phase Delta Pak (5 um; 150 mm × 21.1 mm). Mass spectra were recorded on a Bruker Reflex IV time-of-flight mass spectrometer (Bruker-Daltonic, Bremen, Germany) equipped with the SCOUT 384 probe ion source and electrospray on a Platform Micromass apparatus. Nuclear magnetic resonance (NMR) spectra for 1 H, TOCSY and COSY were recorded on Bruker Advance 300 (300 MHz) in DMSO-d 6 . The bioassays were measured using Microplate reader MCC/340 (Labsystems, Cergy-Pontoise, France).

Synthesis of peptides for biological test
The nine peptides (DKPRR, CDKPRR, TKPRR, TKPRR-NH 2 , APQPRPL, CPQPRPL, TKPPR, DKPPR, DKPPR-NH 2 ) were synthesized through Fmoc chemistry with HBTU activation via solid phase peptide synthesis (SPPS) on our peptide synthesizer. The Fmoc-amino acid-Wang resin was used as the starting material and swelled in dichloromethane (DCM). The functional groups at the building block of the resin were removed by piperidine (20% in dimethylformamide (DMF). The next amino acid was then grafted with the activation step where its carboxyl group was activated by adding a threefold excess of Fmoc-aminoacid-OH, 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU), 1-hydroxybenzo-triazole (HOBt) and N,N-isopropylethylamine (DIEA) in DMF. The process of deprotection, activation and coupling were repeated until the desired sequences of amino acids were grafted on the resin. At the end of each step, all soluble reagents on the resin and the protecting group at the N-terminus of the peptide were removed by filtration and washing process without disturbing or damaging the resin. Cleavage of the peptide from the resin was done using trifluoroacetic acid, triisopropylsilane and water. The resin was then washed, purified using high performance liquid chromatography (HPLC) and further lyophilized and stored at −20°C.
Biological testbinding study of the peptides on recombinant NRP-1 and VEGFR-2 (KDR) The affinity of the peptides towards NRP-1 was assessed through a competitive bioassay with biotinylated VEGF 165 . The surface of 96-well Maxisorp microplates (Dutscher) was coated with receptors NRP-1 and KDR recombinant chimeric proteins in phosphate-buffered saline (PBS) and left overnight at room temperature. After washing the wells with the wash buffer (PBS containing .05% Tween-20), the plates were then blocked with the blocking buffer containing PBS and .5% bovine serum albumin (BSA) for 1 h at 37°C to avoid nonspecific interactions. The biotinylated VEGF 165 was then added to put in contact with the recombinant chimeric proteins in the wells with or without peptide, together with the blocking buffer with additional of 2 μg/mL heparin. After 2 h of incubation at room temperature, the wells were rinsed with the wash buffer and the amount of bound biotinylated VEGF 165 was stained with streptavidin horseradish coupled to peroxide by adding this reagent to the wells for 20 min at room temperature. After rinsing, the colour reagent was added and the reaction was stopped after 30 min by the addition of the stop solution and the optical density was measured at 450 nm. The results were expressed as the relative absorbance from wells containing only biotinylated VEGF 165 . Three wells per condition were used in this test. The concentrations of receptors, biotinylated VEGF 165 and VEGF 165 competitor are summarized in Table 2.

Validation of docking calculations
Docking methods are typically validated by 'redocking' experiments, where a series of known complexes are separated and then redocked, to ensure that the docking algo-rithm can reproduce the observed experimental binding mode (Cosconati et al., 2010). In this study, the validation was done by performing a redocking run of tuftsin, the co-crystalized ligand into the binding pocket of NRP-1. Tuftsin, an immunostimulatory tetrapeptide (TKPR) shares a sequence motif similar to the C-terminal of VEGF 165 , i.e. CDKPRR, which has been tested to compete with VEGF 165 for binding to NRP-1 (Von Wronski et al., 2006). The re-docked tuftsin was found to be similar to the crystallographic pose with an RMSD of 1.94 Å (Figure 2). Although the RMSD value is relatively high, however, considering this peptide is highly flexible and has a sizeable number of atoms, this value is deemed acceptable. An RMSD of less than 2.0 Å is Focused view of interaction of tuftsin at the binding site. The binding site is represented as stick and coloured as velvet for Y or Tyr, pink (E or Glu), green (T or Thr), light grey (S or Ser) and purple (D or Asp). Tuftsin is rendered as ball and stick and coloured based on their atom type: carbon (grey), oxygen (red), nitrogen (blue) and hydrogen (white). Hydrogen bond and π-π interactions between the tuftsin and the NRP-1 are shown as black and red lines, respectively. generally accepted in the reproduction of experimentally determined structure (Jones, Willett, & Glen, 1995;Wahab, Choong, Ibrahim, Sadikun, & Scior, 2009).
Furthermore, it was further verified by visual inspection that the docked ligand bound to the protein at the same binding site as that of the crystal structure, at the conserved interstrand loops of the NRP-1 b1 domain (Figure 2(b)). Both the crystal and docked conformations bound in a region close to residues Tyr297 (loop I), Asp320 (loop II), and Ser346, Glu348, Thr349 and Tyr353 (loop III) (Figure 2(b)). It is hence confirmed that the experimental conformation was successfully reproduced by the docking protocol developed and thus the same docking procedure was applied to dock the other peptides on NRP-1.

Molecular docking
In this study, the computational docking between the peptide ligands and NRP-1 was calculated using a semiempirical free-energy force field method implemented in AutoDock. Although, it would be more preferable to calculate the free energy of binding using quantum chemical calculation, free energy perturbation method or full atomic molecular dynamic simulation, those methods are computationally demanding (Woo & Roux, 2005), (Huey, Morris, Olson, & Goodsell, 2007). In contrast, AutoDock provides rapid evaluation of the interaction energy between the ligand and its protein target using simpler force field but covering a wider region of conformational space (Morris et al., 1998a).
The free energy calculation in AutoDock utilizes a semi-empirical free-energy force field method which was parameterized using a large number of protein-inhibitor complexes for which both structure and inhibition constants, or Ki, are known. Figures 3 and 4 show the conformations of docked peptides with NRP-1. As expected, all the peptides bound to the same binding region as tuftsin at b1 domain of NRP-1 (Vander Kooi et al., 2007). The predicted binding free energy (ΔG bind ) and the inhibition constant (Ki) of tuftsin and the peptides are given in Table 3. Tuftsin showed the highest activity with the estimated ΔG bind of −12.28 kcal/mol followed by TKPRR and TKPPR (−11.01 and −10.49 kcal/mol, respectively). It is interesting to note that the longer the residue chain, the higher the binding energies as seen in the case of heptapeptides APQPRPL and CPQPRPL. In general, the peptides with arginine at the C-terminal showed higher activity than those with the other residues or with arginine in between the sequence. These findings demonstrated that the position of arginine and the length of the peptide sequence play an important role in their binding with NRP-1. Changing the carboxyl group with an amide did not have a consistent effect on the binding affinity towards NRP-1 as demonstrated by TKPRR-NH 2 and DKPPR-NH 2 . In the case of TKPPR, this replacement significantly reduced the activity in TKPRR-NH 2 while in the case of DKPPR, the replacement slightly increased the binding affinity towards NRP-1. NRP-1 is presented as ribbon and the binding site is represented as stick and coloured as velvet for Y or Tyr, pink (G or Glu), green (T or Thr), light grey (S or Ser), purple (D or Asp), cyan (G or Gly) and W or Trp (light blue). Peptide ligands are rendered as ball and stick and coloured based on their atom type: carbon (grey), oxygen (red), nitrogen (blue), hydrogen (white) and sulphur (yellow). Hydrogen bond and π-π interactions between the peptide ligands and the NRP-1 are shown as black and red lines, respectively. Similar to tuftsin, it was observed that the docked peptides bound in close proximity to Tyr297 (loop I), Asp320 (loop II), Ser346, Glu348, Thr349 and Tyr353 (loop III). In general, the peptides stacked in between the two tyrosine residues, i.e. Tyr297 and Tyr353 where they formed hydrogen bonding with these residues. However, not all of the peptide formed hydrogen bond with Y297 (i.e. TKPRR-NH 2 and CPQPRPL) and Y353 (i.e. DKPPR-NH 2 , CDKPRR, APQPRPL, CPQPRPL and ATWLPPR). The arginine C-terminal of each peptide was also held by a number of π-π interactions which mediated one, two or more interactions with NRP-1 through residues Y297, Y353 and W301 (see Figures 4 and 5). These interactions strengthen the binding of these peptides and increase their stability in the NRP-1 binding site. However, interestingly, DKPPR and TKPPR did not make any π-π interaction with NRP-1. These peptides also formed a network of hydrogen bonding with D320, S346, E348 and T349. These findings are in agreement with other studies (Starzec et al., 2007;Vander Kooi et al., 2007).
The distances of arginine of each peptide with those important residues were also examined and summarized in the Table 4. It was observed that most peptides bind closer to these side chains; Y297 (loop I), D320 (loop II), S346, E348, T349 and Y353 (loop III) compared to ATWLPPR, the peptide we previously demonstrated (Tirand et al., 2006) to inhibit NRP-1. This indicates that these peptides might have better interaction at the binding site of NRP-1 than ATWLPPR. Therefore, from all these observations, the investigated peptides (TKPRR, TKPPR, DKPPR-NH 2 , DKPPR, TKPPR-NH 2 , CDKPRR, APQPRPL and CPQPRPL) were then synthesized in order to confirm their binding affinity from the molecular affinity study.

Synthesis and characterization of the peptides
The peptides; DKPRR, DKPPR, DKPPR-NH 2 , TKPRR, TKPPR, TKPRR-NH 2 , CDKPRR, APQPRPL and CPQPRPL were synthesized through SPPS technique with a final purity obtained more than 95%, as assessed by HPLC. The identities of all peptides were confirmed by mass spectrometry and NMR (see supplementary material).  In vitro binding studies on NRP-1 (molecular affinity studies) The affinities of the different peptides against NRP-1 were evaluated by ELISA, a type of binding affinity test. It is well known that VEGF 165 binds specifically to VEGFR-1, VEGFR-2, NRP-1 and NRP-2, as well as nonspecifically to cell-surface and extra cellular matrix proteoglycans (Von Wronski et al., 2006). Hence, if the peptides do have affinities to NRP-1, they will compete with VEGF 165 in order to bind on the said receptor.
The replacement of carboxyl group by amide group at the C-terminal arginine of the peptides DKPPR and TKPRR showed lesser affinities for NRP-1 as compared to the acidic peptides. The APQPRPL and CPQPRPL (linear peptide) had no binding affinity towards NRP-1. Contrary to this finding, slightly different peptides that were synthesized by other group did show affinity towards NRP-1. Giordano and the team previously reported a sequence of PQPRPL (Giordano, Cardó-Vila, Lahdenranta, Pasqualini, & Arap, 2001) and CPQPRPLC in cyclic form CPQPRPLC (Giordano et al., 2001) which showed affinity towards NRP-1 receptor.

Further discussion
Protein-peptide interactions are involved in numerous cellular processes and therefore, it is not surprising that in the recent years, peptides have been attracting interests in drug discovery and development. Computational chemistry techniques have proven to successfully support the drug-discovery process; and computer simulations of peptide-protein binding, based on molecular docking and scoring, have been widely used in applications to computational structure prediction of ligand-protein complexes, virtual screening of large databases for active compounds and binding affinity prediction of inhibitorprotein complexes (Verkhivker et al., 2002). However, reliable prediction of protein-ligand docking structure remains a challenging problem. Peptide ligands present one of the more difficult cases because of their high flexibility, requiring extensive configurational sampling. Even for the rigid-protein model, it is demanding to handle flexible ligands with over 15-20 rotatable bonds, which is typical even for short peptides. This difficulty increases when the protein receptor is also flexible and a peptide ligand only binds to a small subset of the many possible conformations of the protein (Huang & Wong, 2009).
later superseded by Rosetta FlexPepDock which combines ab initio predictions (London, Raveh, Cohen, Fathi, & Schueler-Furman, 2011);HADDOCK (de Vries, van Dijk, & Bonvin, 2010), CABS-dock (Kurcinski, Jamroz, Blaszczyk, Kolinski, & Kmiecik, 2015) in addition to molecular dynamics (MD) simulation-based docking programme such DynaDock (Antes, 2010). Small molecules docking softwares such as AutoDock (Morris et al., 1998a), DOCK (Ewing, Makino, Skillman, & Kuntz, 2001), GOLD (Verdonk, Cole, Hartshorn, Murray, & Taylor, 2003) have also been applied for protein-peptide docking but they are generally more suitable for molecules with a limited number of rotatable bonds (Rubinstein & Niv, 2009). Nevertheless, Autodock has been shown to successfully dock short peptides up to four amino acids (Hetenyi & van der Spoel, 2002) and  up to seven amino acids (Arun Prasad & Gautham, 2008), peptide-antibody complexes (Chen, Simmonds, & Timkovich, 2013;Sotriffer et al., 2000), and even protein-protein interactions (Eyrisch & Helms, 2007). Based on these encouraging results, we are persuaded to apply AutoDock to predict the binding interaction of our protein-peptide complexes. In our previous study, a peptide-conjugated photosynthesizer, TPC-Ahx-ATWLPPR targeting NRP-1 had demonstrated an affinity for endothelial cells of tumour neovasculature (Tirand et al., 2006). However, the peptide moiety (ATWLPPR) of the conjugate degraded rapidly and was relatively unstable. The peptide ATWLPPR identified from phage epitope library was first reported to bind at VEGFR-2 (KDR) (Binetruy-Tournaire et al., 2000) but has been demonstrated to bind specifically to NRP-1 (Perret et al., 2004) and then selectively inhibits VEGF 165 binding to NRP-1 (Starzec et al., 2006). Besides ATWLPPR; PQPRPL (Giordano et al., 2001), CDKPRR (Jia et al., 2006) and TKPPR (Von Wronski et al., 2006) and two cyclic peptides CPQPRPLC (Giordano et al., 2005) and (Jia et al., 2006), have also been reported to inhibit NRP-1. The latter cyclic peptide was derived from exon 8 of VEGF (residues 138-165) (Jia et al., 2006). Thus, in this study, we evaluated the binding of the whole or modified sequence of these peptides onto NRP-1 and their affinity and molecular interactions were compared using molecular docking technique as well as in vitro molecular affinity study. The docked peptides were shown to bind at the b1 domain of NRP-1 and they were found to form a network of interactions with residues of Y297 (loop I), D320 (loop II) and S346, E348 and T349 (loop III). They were also found stacking in between the two tyrosines, Y297 and Y353 which could strengthen the binding of ligand/receptor complex. These findings are similar to what observed previously in the crystal structure of NRP-1/tuftsin co-complex (Von Wronski et al., 2006) as well as in the molecular dynamic simulation of NRP-1/ATWLPPR system (Starzec et al., 2007).
In general, peptides with C-terminal arginine showed a more negative predicted binding energy than those without, which agreed well with the in vitro molecular affinity results ( Figure 5). The APQPRPL and CPQPRPL, both which have a C-terminal leucine residue on the other hand failed to displace VEGF 165 binding on NRP-1. The distance analysis between the C-terminal residues with the important residues of NRP-1 binding site also correlates with this finding. In particular, C-terminal arginine showed the strongest interaction with Asp320, which is consistent with a molecular dynamic study of NRP-1 with a model peptide, RPAR (Haspel et al., 2011). This finding also corresponded well with the previous alanine-scanning and amino acid-deletion analysis which showed that the C-terminal arginine played an important role in the inhibitory activity of ATWLPPR on VEGF 165 binding to NRP-1 (Starzec et al., 2007). These findings are also consistent with the previous observation that the binding site for tuftsin (TKPR) in NRP-1 seems to optimally accommodate a C-terminal arginine (Vander Kooi et al., 2007;Von Wronski et al., 2006).
The results from molecular docking showed that changing the carboxyl group of the C-terminal arginine with a terminal amide group did not have consistent and significant effects on the binding capability of the peptide. From the in vitro binding, however, it was evidence that the peptides with a terminal amide group have much lower affinity to NRP-1 than the corresponding peptides with the carboxyl group. This result is in agreement with the previous studies which showed that the carboxyl group of the terminal arginine must be in its free acid form since inhibitory action drops dramatically for the peptide with a terminal amide group (Jia et al., 2006).
All the peptides used in this study had no affinity towards VEGFR-2 (KDR) which was evidence from their failure to displace the binding of biotinylated VEGF 165 on the KDR protein. DKPRR, DKPPR, TKPRR, TKPPR and CDKPRR showed good inhibitory activity on NRP-1. However, among these peptides, only DKPPR and TKPRR were chosen for further study (Kamarulzaman et al., 2015). These two peptides not only showed favourable binding interaction with NRP-1 as demonstrated by molecular docking but also from the in vitro binding affinity test in which both of them showing the highest displacement of VEGF 165 binding towards the recombinant NRP-1. DKPPR and TKPPR will hence be conjugated with a photosensitizer, and evaluated for their biological activity in vivo. This will be further discussed in our subsequent publication (Kamarulzaman et al., 2015).

Conclusion
The combination of computational docking study and biological evaluation in this study provide an insight into the design of peptide that might interact with NRP-1. Peptides having C-terminal arginine at position 4 and/or 5 i.e. DKPRR, DKPPR, TKPRR, TKPPR and CDKPRR made interaction with NRP-1 through hydrogen bond and π-π interaction. These peptides also demonstrated a good inhibition against NRP-1. On the contrary, peptides having seven amino acids and with arginine located at other than the C-terminal i.e. CPQPRPL and APQPRPL failed to displace the VEGF 165 binding to NRP-1. A minor replacement of amide group instead of carboxyl group at C-terminal domain i.e. DKPPR-NH 2 and TKPRR-NH 2 showed lower activity against NPR-1. Among the evaluated peptides, two are novel i.e. DKPPR and TKPRR and were then chosen to be conjugated photosynthesizer molecules (Kamarulzaman et al., 2015). It is hoped that the information from this finding will provide deeper understanding towards the design of more active peptides as anti-angiogenic targeting agent.

Disclosure statement
No potential conflict of interest was reported by the authors.