NRP-1 complex

Bioorganic & Medicinal Chemistry Letters 29 (2019) 2493–2497

Contents lists available at ScienceDirect

Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl

Urea moiety as amide bond mimetic in peptide-like inhibitors of VEGF-A165/ NRP-1 complex

T

Anna K. Puszkoa, , Piotr Sosnowskib,c, Karolina Pułka-Ziacha, Olivier Hermined,e,f, ⁎ Gérard Hopfgartnerb, Yves Lepelletierd,e,f, Aleksandra Misickaa,c, ⁎

a

Faculty of Chemistry, University of Warsaw, Pasteura 1, 02-093 Warsaw, Poland Department of Inorganic and Analytical Chemistry, University of Geneva, 24 Quai Ernest Ansermet, CH-1211 Geneva 4, Switzerland c Department of Neuropeptides, Mossakowski Medical Research Centre, Polish Academy of Sciences, Pawinskiego 5, 02-106 Warsaw, Poland d Université de Paris, Imagine Institute, 24 Boulevard Montparnasse, 75015 Paris, France e INSERM UMR 1163, Laboratory of Cellular and Molecular Basis of Normal Hematopoiesis and Hematological Disorders: Therapeutical Implications, 24 Boulevard Montparnasse, 75015 Paris, France f CNRS ERL 8254, 24 Boulevard Montparnasse, 75015 Paris, France b

ARTICLE INFO

ABSTRACT

Keywords: Peptidomimetics Amide bond mimetic Neuropilin-1 VEGF-A165 Protein–ligand interaction

NRP-1 is an important co-receptor of vascular endothelial growth factor receptor-2 (VEGFR-2). Many reports suggested that NRP-1 might also serve as a separate receptor for VEGF-A165 causing stimulation of tumour growth and metastasis. Therefore, compounds interfering with VEGF-A165/NRP-1 complex triggered interest in the design of new molecules, including peptides, as anti-angiogenic and anti-tumour drugs. Here, we report the synthesis, affinity and stability evaluation of the urea-peptide hybrids, based on general Lys(hArg)-AA2-AA3-Arg sequence, where hArg residue was substituted by Arg urea unit. Such substitution does not substantially affected affinity of compounds for NRP-1 but significantly increased their proteolytic stability in plasma.

Neuropilin-1 (NRP-1) is a membrane receptor, which plays a crucial role in the development of axon guidance and in the physiological and pathological formation of capillaries in angiogenesis process.1–3 The key role of extracellular region of this receptor is the binding of several ligands, which modified intrinsic signalling due to the association of appropriate co-receptor. Vascular endothelial growth factor-A165 (VEGF-A165) is one of the most important NRP-1 ligand, acting as proangiogenic factor that interacts with b1 and b2 receptor subdomains.4–6 NRP-1 is a multifunctional co-receptor, especially for tyrosine kinase receptor VEGFR-2,7 which signalling is critical in VEGF-A165/VEGFR-2mediated angiogenesis.8 NRP-1 lacks catalytic activity, but can function as an independent receptor.9–13 NRP-1 overexpression increases tumour growth and is often associated with poor prognosis in various types of tumours, especially in their epithelial origin.14–18 Amino acids encoded by exons 7 and 8 of VEGF-A165 are crucial for protein interaction with NRP-1.19–22 The binding pocket of NRP-1 b1-

domain is specific and requires the presence of Arg from the ligands at the C-terminus.5,22 Screening of phage libraries, containing peptides that bind to prostate cancer cells expressing NRP-1, showed that most of the selected compounds contained the Arg/Lys-Xaa-Xbb-Arg/Lys fragment embedded in their terminal sequence.23 The absolute requirement of this structural motif in NRP-1 peptidic ligands is called C-end rule (CendR).23 The C-terminus of VEGF-A165 (CDKPRR) encoded by the exon 8 fulfils this condition. CendR motif occurs also in the structure of known active peptides, that have affinity for NRP-1, e.g. A7R and tuftsin.24–27 Searching for new VEGF-A165/NRP-1 complex inhibitors, including functional peptides, could play a crucial role in drug development. Peptides as intrinsic signalling molecules for many physiological functions, present an opportunity for therapeutics that closely mimics natural pathways. However, certain limitations of native peptides, such as a short plasma half-life, their interactions at multiple receptors and negligible oral bioavailability, are an obstacle. To

Abbreviations: ArgU, arginine urea derivative; Aze, azetidine-2-carboxylic acid; Dab, 2,4-diaminobutyric acid; Dhp, 3,4-dehydroproline; DIPEA, N,N-diisopropylethylamine; ELISA, enzyme-linked immunosorbent assay; Fmoc, fluorenylmethyloxycarbonyl; hArg, homoarginine; HATU, 1-[bis(dimethylamino)methylene]-1H1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate; NRP-1, Neuropilin-1; Oic, octahydroindole-2-carboxylic acid; Pbf, 2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl; SPS, Solid-Phase Synthesis; TIS, triisopropylosilane; VEGF, vascular endothelial growth factor; VEGF-R, vascular endothelial growth factor receptor; TFA, trifluoroacetic acid; TIS, triisopropylsilane ⁎ Corresponding authors at: Faculty of Chemistry, University of Warsaw, Pasteura 1, 02-093 Warsaw, Poland (A. Misicka). E-mail addresses: [email protected] (A.K. Puszko), [email protected] (A. Misicka). https://doi.org/10.1016/j.bmcl.2019.07.016 Received 21 May 2019; Received in revised form 7 July 2019; Accepted 8 July 2019 Available online 10 July 2019 0960-894X/ © 2019 Elsevier Ltd. All rights reserved.

Bioorganic & Medicinal Chemistry Letters 29 (2019) 2493–2497

A.K. Puszko, et al.

triazoles have been shown to be viable surrogates for cis- and transamide bonds, depending on their substitution pattern (1,4 or 1,5-disubstituted).33–35 However, it should be noted that amide bond modifications may also have some negative effects on the biophysical and biochemical properties of obtained peptidomimetics.36,37 The rational design of peptidomimetics involves identifying the sites for the hydrolysis of the compound by proteases, followed by the introduction of chemical modifications. The urea moiety might be a valuable peptide bond surrogate due to enhanced stability, comparing to the peptide bond. Moreover, urea moiety may increase water solubility and provide additional binding sites to interact with biological targets, by forming additional hydrogen bonds.38 Furthermore, replacement of susceptible to enzymatic cleavage amide bond might affect the properties of the peptidomimetic, including modification of affinity for receptor. Following this strategy, peptidomimetic with one urea unit based on the parent Lys(hArg)-AA2-AA3-Arg sequence39,40 were designed and synthesized. Our previous studies showed that Lys(hArg) fragment, crucial for high affinity, undergoes enzymatic hydrolysis primarily.39,40 Thus, we decided to create peptidomimetics in which amide bond between carboxylic group of hArg and ε-amine group of Lys would be replaced by urea moiety. Chosen modification was associated with the extension of the main chain with an additional –CH2 (Fig. 1) and NH groups, as the synthesis of such units is very well described in the literature (scheme of the synthesis and analytical data are presented in Supplementary material).41 New compounds were designed also to test if the introduced urea bond has the same effect on the affinity for NRP-1, with different substituents in the middle part (AA2-AA3) of the molecule. To avoid excessive branching and elongation of the chain of a peptidomimetics, hArg was changed for Arg urea derivative. To insert this urea unit into the peptidomimetic sequence, succinimidyl carbamate building block with amine group masked as azide group was used. Both, the synthesis of activated building blocks and solid phase synthesis (SPS) are well described in the literature.41 Using the HyperChem Professional 8.0 software, the lengths of individual bonds in the case of an amino acid and urea residues were calculated. The results showed that the Lys(ArgU) fragment chain is 2.84 Å longer than the original Lys (hArg) fragment (Fig. 1). To illustrate extended hArg reach, 3D models of synthesized compounds were overlaid on parent peptidomimetics.

Fig. 1. Comparison of Lys(hArg) and Lys(ArgU) branching with amide bond and urea moiety lengths [Å].

Fig. 2. Overlays of the 3D models of compound A) 1, B) 2, C) 3 and D) 4 with Lys(ArgU) (tan) and its parent peptides (blue) with Lys(hArg) branching.

overcome these problems, compounds that mimic peptides are designed. Replacement of amide bonds by peptide bond surrogates might be a solution, especially to improve the metabolic stability of the compound. One of the simplest modifications of the amide bond is the reduced peptide bond [CH2NH], which was widely used in the design of stable enzymatically active pseudopeptides.28,29 Thioamide [CSNH] is a subtle modification with better hydrogen-bond donors proprieties which makes it a useful tool in design of peptide analogues.30–32 1,2,3-

Scheme 1. General scheme of the SPS of urea-peptidomimetic hybrids. 2494

Bioorganic & Medicinal Chemistry Letters 29 (2019) 2493–2497

A.K. Puszko, et al.

Table 1 The general structure and VEGF-A165/NRP-1 complex inhibitory effect of H-Lys(ArgU)-AA2-AA3-Arg-OH branched urea-peptide hybrids. Modified fragment of the sequence is marked in blue and red (AA2-AA3).

logIC50 ± SEMa

IC50 [µM]

PP IC50 [µM]b

H-Lys(ArgU)-Pro-Pro-Arg-OH

−5.11 ± 0.03

7.7

21

2

H-Lys(ArgU)-Dab-Aze-Arg-OH

−4.6 ± 0.0

26

12

3

H-Lys(ArgU)-Dab-Dhp-Arg-OH

−5.53 ± 0.03

2.9

4

4

H-Lys(ArgU)-Dab-Oic-Arg-OH

−5.27 ± 0.03

5.4

2

Compound

Sequence

1

AA2-AA3 structure

a R2 ranged between 0.98–0.99. b PP – parent peptidomimetic with Lys(hArg) branching at N-terminus; data are already published.40 Compounds were tested in the concentrations range 0.5 – 100 µM (compounds 1, 3 and 4) or 5 – 100 µM (compound 2).

Geometry of compounds was optimized using molecular mechanics (Polak-Ribiere Algorithm, RMS gradient of 0.02 kcal/Å mol, in vacuo). 3D models were rendered and overlaid with UCSF Chimera.42 Further reach of guanidine group, different amine group position on hArg/ArgU residue and extra nitrogen present in urea moiety are visible (Fig. 2). These modifications may potentially enable additional interactions with NRP-1 side chains. The synthesis of urea-peptide hybrids was carried out manually on the pre-loaded Fmoc-Arg(Pbf)-Wang resin following the Fmoc chemistry (Scheme 1).43 For amino acids standard coupling HATU/DIPEA protocol was used.44 The deprotection of Fmoc group was performed using 20% piperidine in DMF. Active carbamate building block coupling and azide reduction was performed under microwave irradiation (Scheme 1).41 The obtained peptidomimetics were cleaved from the resin using TFA:H2O:TIS mixture (details of synthesis, purification and analytical data are presented in Supplementary material). The affinity of compounds for NRP-1 was determined using ELISAlike assay with chemiluminescence detection as previously described (detailed procedure is given in Supplementary material).40,45–47 First, to investigate whether the urea modification will not cause the loss of peptidomimetics affinity to NRP-1, urea moiety was introduced into previously described Lys(hArg)-Pro-Pro-Arg sequence with known inhibitory activity (IC50 = 21 µM).39,40 Urea-peptide hybrid 1 showed almost trice higher inhibitory effect on VEGF-A165/NPR-1 relative to

parent sequence (Table 1). Due to positive results of urea moiety introduction into the Lys (hArg)-Pro-Pro-Arg peptide, new peptidomimetics based on more active Lys(hArg)-Dab-AA3-Arg sequence were synthesised. The middle part of the molecule is suspected of being involved in directing the N-terminal Lys(hArg) branched fragment, which allows side chain to interact with the NRP-1.39,40 In Lys(hArg)-Dab-Pro-Arg peptide, contrary to the stable positioning of C-terminal Arg, the N-terminal branching was observed to switch the receptor binding partners during the molecular modelling.39 In our previous studies, it was demonstrated that replacing Pro with its more rigid mimetics in the AA3 position increased interactions with NRP-1.40 Thus, we decided to test how the urea moiety will influence the affinity of Lys(hArg)-Dab-AA3-Arg sequence, where AA3 is mimetic of Pro with various ring size and rigidity (Table 1).40 Ureapeptidomimetic 2 with the 4-membered azetidine ring (Aze) and 4 with octahydroindole ring (Oic) had about twice lower inhibitory activity compared to corresponding parent peptidomimetics with hArg. This is probably due to disadvantageous geometry of the molecule, which reduced the ability of the elongated N-terminal Lys(ArgU) fragment to interact with the surface of the receptor. Urea moiety in the analogue 3 containing 2,3-dehydroproline (Dhp) in the AA3 position caused very slight increase of the affinity for NRP-1. This new peptidomimetic with pyrrolidine ring containing a double bond seems to be the best among sequences presented in this paper. 2495

Bioorganic & Medicinal Chemistry Letters 29 (2019) 2493–2497

A.K. Puszko, et al.

Fig. 3. Comparison of the in vitro metabolic stability of parent peptide and its A) urea-peptidomimetic analogue 1 and B) urea-peptidomimetic analogue 3. C) Bonds cleaved by enzymes in parent peptides Lys(hArg) branching, hybrid 1 and 3 (scissors show identified proteases cleavage sites). Superimposed extracted ion chromatograms (XIC) of detected and potential metabolites of D) compound 1 and E) compound 2; only fragments corresponding to amide bond cleavage were found; fragments related to enzymatic hydrolysis of urea bond were not found.

The main purpose of the introduction of the urea moiety to parent sequences, was to increase the stability of the compounds in biological fluids without the loss of affinity. We performed in vitro enzymatic stability studies in human blood plasma to investigate whether the introduced changes affected the half-life of new compounds. Urea-peptidomimetics 1 and 3 with higher affinity for NRP-1 compared to parent sequences with Lys(hArg) fragment were tested. Compounds were incubated with human plasma at 37 °C and at selected time intervals. Samples of the mixture were taken and analysed using RP-HPLC and LC-MS as described previously (detailed procedure is described in Supplementary material).40 The analysis of urea-peptidomimetics showed that the exchange of the amide bond for the urea group had an expected positive effect on the stability of the tested compounds. Fig. 3A and B shows the concentration decline of each of the tested hybrids during plasma incubation, with respect to their parent sequences without an urea moiety. The exact half-life of urea-peptide hybrid 1 and 3 could not be calculated, as even after 96 h, 40–45% of substrate was still present in the plasma.

However, it seems that new compounds are approximately 2 times more resistant to enzymatic hydrolysis, compared to peptides with Lys (hArg) fragment (Fig. 3A and B). Analysis of the metabolites of compounds 1 and 3 using LC-MS showed that enzymatic cleavage of Lys-εArgU bond did not occur (Fig. 3C–E, detailed data can be found in Supplementary material). Obtained results show that the modified peptide bond plays a significant role not only in the interaction with the receptor, but also in the stability of the branched compounds and it is particularly sensitive to the action of enzymes. Modification of this fragment to eliminate enzymatic cleavage was therefore very desirable. There are two main factors that are responsible for elimination of peptides from the body – rapid renal clearance and enzymatic degradation. The renal clearance of peptide drugs might be reduced by coupling peptides to molecules that prevent their elimination by the kidneys.48,49 Extending the plasma half-life might help to extend the time between drug applications and thus reduce patient burden from additional doses. Some important proteins, i.e. IgGs (except IgG3) and serum albumin, demonstrate the exceptionally long plasma half-life in 2496

Bioorganic & Medicinal Chemistry Letters 29 (2019) 2493–2497

A.K. Puszko, et al.

the range of several weeks (19 days for albumin and approximately 21 days for IgGs).50 The extended half-life of peptides can be achieved by improving their proteolytic stability through chemical modification, including amide bonds alteration. The choice of peptide bond mimetic is a compromise between the positive effect on pharmacokinetics and the potential negative effect on the activity and specificity of the compounds. In our studies, we have synthesized a new group of ureapeptidomimetics based on previous active structures. Obtained results demonstrate that the substitution of Lys-ɛ-hArg amide bond by urea group and change of hArg side chain for Arg in Lys(hArg)-AA2-AA3-Arg sequence eliminated the enzymatic cleavage of peptidomimetic branching. Simultaneously, VEGF-A165/NRP-1 complex inhibition by urea-peptide hybrids depends in major on the residues at the AA2-AA3 fragment of the molecule, as the substitution of hArg by Arg urea unit (Argu) does not radically affect the affinity for NRP-1. Thus, the designed peptidomimetics, especially compound 3, might be the sequence for further modifications.

4. Mamluk R, Gechtman Z, Kutcher ME, Gasiunas N, Gallagher J, Klagsbrun M. J Biol Chem. 2002;277:24818–24825. 5. Hicklin DJ, Ellis LM. J Clin Oncol. 2005;23:1011–1027. 6. Guo HF, Vander Kooi CW. J Biol Chem. 2015;290:29120–29126. 7. Miao HQ, Lee P, Lin H, Soker S, Klagsbrun M. FASEB J. 2000;14:2532–2539. 8. Soker S, Takashima S, Miao HQ, Neufeld G, Klagsbrun M. Cell. 1998;92:735–745. 9. Fukasawa M, Matsushita A, Korc M. Cancer Biol Ther. 2007;6:1173–1180. 10. Robinson SD, Reynolds LE, Kostourou V, et al. J. Biol. Chem. 2009;284:33966–33981. 11. Valdembri D, Caswell PT, Anderson KI, et al. PLoS Biol. 2009;7 e1000025. 12. Grünewald FS, Prota AE, Giese A, Ballmer-Hofer K. Biochim Biophys Acta. 1804;2010:567–580. 13. Grun D, Adhikary G, Eckert R. Oncogene. 2016;35:4379–4387. 14. Handa A, Tokunaga T, Tsuchida T, et al. Int J Oncol. 2000;17:291–295. 15. Kawakami T, Tokunaga T, Hatanaka H, et al. Cancer. 2002;95:2196–2201. 16. Bagri A, Tessier-Lavigne M, Watts RJ. Clin Cancer Res. 2009;15:1860–1864. 17. Prud’homme GJ, Glinka Y. Oncotarget. 2012;9:921–939. 18. Wild JRL, Staton CA, Chapple K, Corfe BM. Int J Exp Pathol. 2012;93:81–103. 19. Keyt BA, Berleau LT, Nguyen HV, et al. J Biol Chem. 1996;271:7788–7795. 20. Soker S, Gollamudi-Payne S, Fidder H, Charmahelli H, Klagsbrun M. J Biol Chem. 1997;272:31582–31588. 21. Fairbrother WJ, Chample MA, Christinger HW, Keyt BA, Starovasnik MA. Structure. 1998;6:637–648. 22. Parker MW, Xu P, Li X, Vander Kooi CW. J Biol Chem. 2012;14:11082–11089. 23. Teesalu T, Sugahara KN, Kotamraju VR, Ruoslahti E. Proc Natl Acad Sci USA. 2009;106:16157–16162. 24. Binétruy-Tournaire R, Demangel C, Malavaud B, et al. EMBO J. 2000;19:1525–1533. 25. Perret GY, Starzec A, Hauet N, et al. Nucl Med Biol. 2004;31:575–581. 26. Starzec A, Vassy R, Martin A, et al. Life Sci. 2006;79:2370–2381. 27. von Wronski MA, Raju N, Pillai R, et al. J Biol Chem. 2006;281:5702–5710. 28. Iguchi S, Kawasaki K, Okamoto H, Umezawa C, Okada Y. Chem Pharm Bull. 1999;47:423–427. 29. Thomas N, Pernot M, Vanderesse R, et al. Biochem Pharmacol. 2010;80:226–235. 30. Jones WC, Nestor JJ, Vigneaud VD. J Am Chem Soc. 1973;95:5677–5679. 31. Lankiewicz L, Bowers CY, Reynolds GA, et al. Biochem Biophys Res Commun. 1992;184:359–366. 32. De Zotti M, Biondi B, Peggion C, et al. Beilstein J Org Chem. 2012;8:1161–1171. 33. Tam A, Arnold U, Soellner MB, Raines RT. J Am Chem Soc. 2007;129:12670–12671. 34. Nahrwold M, Bogner T, Eissler S, Verma S, Sewald N. Org Lett. 2010;12:1064–1067. 35. Tischler M, Nasu D, Empting M, et al. Angew Chem Int Ed Engl. 2012;51:3708–3712. 36. Guichard G, Calbo S, Muller S, Kourilsky P, Briand JP, Abastado JP. J Biol Chem. 1995;270:26057–26059. 37. Laurencin M, Legrand B, Duval E, et al. J Med Chem. 2012;55:2025–2034. 38. Tamilarasu N, Huq I, Rana TM. Bioorg Med Chem Lett. 2001;11:505–507. 39. Tymecka D, Puszko AK, Lipiński PF, et al. Eur J Med Chem. 2018;158:453–462. 40. Puszko AK, Sosnowski P, Tymecka D, et al. Med Chem Comm. 2019;10:332–340. 41. Douat-Casassus C, Pulka K, Claudon P, Guichard G. Org Lett. 2012;14:3130–3133. 42. Pettersen EF, Goddard TD, Huang CC, et al. J Comput Chem. 2004;25:1605–1612. 43. Carpino LA, Han G. J Org Chem. 1972;37:3404–3409. 44. Carpino LA. Am Chem Soc. 1993;115:4397–4398. 45. Liu WQ, Borriello L, Allain B, et al. Int J Pept Res Ther. 2015;21:117–124. 46. Allain B, Jarray R, Borriello L, et al. Cell Signal. 2012;24:214–223. 47. Borriello L, Montès M, Lepelletier Y, et al. Cancer Lett. 2014;349:120–127. 48. Pollaro L, Heinis C. Med Chem Commun. 2010;1:319–324. 49. Zorzi A, Middendorp SJ, Wilbs J, Deyle K, Heinis C. Nat Commun. 2017;8:16092. 50. Kontermann RE. Expert Opin Biol Ther. 2016;16:903–915.

Acknowledgments This work was financially supported by Grant No. 2017/27/N/NZ7/ 02473 from Narodowe Centrum Nauki (National Science Centre, Poland). We thank COST Action CA15135: Multi-target paradigm for innovative ligand identification in the drug discovery process (MuTaLig), supported by COST (European Cooperation in Science and Technology) for a STSM grant that enabled AKP to work for one month at Imagine Institute in Paris. LC and MS analysis were partially done in Laboratory of Chemical Synthesis, CePT, Mossakowski Medical Research Centre, co-financed by EU from the European Regional Development Fund under the Operational Programme Innovative Economy, 2007-2013. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.bmcl.2019.07.016. References 1. He Z, Tessier-Lavigne M. Cell. 1997;90:739–751. 2. Lee P, Goishi K, Davidson AJ, Mannix R, Zon L, Klagsbrun M. Proc Natl Acad Sci USA. 2002;99:10470–10475. 3. Staton CA, Kumar I, Reed MW, Brown NJ. J Pathol. 2007;212:237–248.

2497