On-resin cyclization of peptide ligands of the Vascular Endothelial Growth Factor Receptor 1 by copper(I)-catalyzed 1,3-dipolar azide-alkyne cycloaddition

Article abstract of DOI:10.1016/j.bmcl.2007.07.087

Cyclic peptides were obtained, on-resin, by the copper (I) catalysed 1,3-dipolar cycloaddition of azides and alkynes. The reaction led exclusively to the formation of the expected cyclomonomeric products which acted as ligands of the Vascular Endothelial Growth Factor receptor 1.

Full text of DOI:10.1016/j.bmcl.2007.07.087

Bioorganic & Medicinal Chemistry Letters 17 (2007) 5590–5594
On-resin cyclization of peptide ligands of the Vascular
Endothelial Growth Factor Receptor 1 by copper(I)-catalyzed
1,3-dipolar azide–alkyne cycloaddition
Victor Goncalves,a,b Benoit Gautier,a,b Anne Regazzetti,c Pascale Coric,d,e
a,b,
Serge Bouaziz,d,e Christiane Garbay,a,b Michel Vidala,b, and Nicolas Inguimbert
*
*
aUniversite´ Paris Descartes, UFR biome´dicale, Laboratoire de Pharmacochimie Mole´culaire et Cellulaire,
`
45 rue des saints Peres, F-75006 Paris, France
`
b
INSERM U648,45 rue des saints Peres, F-75006 Paris, France
cUniversite´ Paris Descartes, UFR des Sciences Pharmaceutiques et Biologiques, Laboratoire de Chimie Analytique,
4 avenue de l’observatoire, F-75006 Paris, France
dUniversite´ Paris Descartes, UFR des Sciences Pharmaceutiques et Biologiques, Laboratoire de pharmacologie chimique et ge´ne´tique,
4 avenue de l’observatoire, F-75006 Paris, France
eINSERM U640, 4 avenue de l’observatoire, F-75006 Paris, France
Received 15 June 2007; revised 26 July 2007; accepted 28 July 2007
Available online 22 August 2007
Abstract—Cyclic peptides were obtained, on-resin, by the copper (I) catalysed 1,3-dipolar cycloaddition of azides and alkynes. The
reaction led exclusively to the formation of the expected cyclomonomeric products which acted as ligands of the Vascular Endothe-
lial Growth Factor receptor 1.
Ó 2007 Elsevier Ltd. All rights reserved.
In the few years since its discovery, copper (I)-promoted
[3+2] Huisgen cycloaddition of azides with terminal
alkynes has proved to be one of the most efficient ‘click
reactions’ with widespread applications in organic
chemistry and drug discovery.1 It has been notably used
with interest in the field of peptide chemistry to produce
1,2,3-triazoles as peptide bond isosteres2–4 and to syn-
thesize cyclic peptides. Up to now, two strategies have
emerged to perform cyclization of such peptides. The
first one was the solution phase method which led to
monomeric,3 dimeric2 or mixtures of monomeric and
dimeric cyclic peptides depending on the nature of ami-
no acids introduced5 and substrate concentrations.6 The
second strategy was the on-resin cyclization of peptides,
as performed by Punna et al.7 Unexpectedly, this meth-
od led to the formation of cyclodimeric products despite
the pseudo-dilution effect of the resin.
Herein, we describe the application of copper(I)-cata-
lyzed 1,3-dipolar azide–alkyne cycloaddition to the
solid-phase synthesis of cyclic peptides targeting the
Vascular Endothelial Growth Factor Receptor 1 (VEG-
FR1). Interestingly, this method specifically allowed the
formation of monomeric peptides.
An increasing number of reports tend to indicate that
VEGFR1 is strongly implicated in pathological angiogen-
esis, a biomarker of cancer, and that its inhibition may
constitute an attractive strategy for stopping tumour
growth and metastasis.8–10 VEGFR1 is stimulated by
the binding of its pro-angiogenic ligands, the Vascular
Endothelial Growth Factor (VEGF) and the Placenta
Growth Factor (PlGF). An interesting approach for
inhibiting the receptor activation consists in developing
peptides, or small molecules, that act as ligands of the
receptor extra-cellular domain and therefore are able to
displace VEGF/PlGF binding to VEGFR1.11 Based on
the structural12,13 and mutagenesis14,15 data available,
several amino acids of the VEGF, essentials for VEGF-
VEGFR1 interaction were identified (Fig. 1). Considering
the spatial proximity existing between some of these
Keywords: Click chemistry; Huisgen reaction; VEGF receptors; Anta-
gonist; Resin supported synthesis; Cyclic peptide.
*
Corresponding authors. Tel.: +33 01 42 86 21 26; fax: +33 01 42 86 40
82; e-mail: nicolas.inguimbert@univ-paris5.fr
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.07.087

V. Goncalves et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5590–5594
5591
For each cyclic peptide, the corresponding linear prod-
uct was prepared.
First, the linear peptide 1a (N3-GHQMFYYPra-NH2)
was synthesized by standard Na-Fmoc chemistry on
Rink amide MBHA resin16 (substitution: 0.62 mmol/g)
in 0.25 mmol scale. Couplings were carried out with
in situ-activating reagents (HBTU, HOBt in the pres-
ence of DIPEA) to generate HOBt esters.
The N-terminal a-azido glycine was synthesized accord-
ing to the method developed by Lundquist et al.17 and
the alkyne moiety was introduced as a L-propargylgly-
cine introduced in the C-terminal position. This amino
acid, commercially available, presents the same stereo-
chemistry as natural amino acids and permits the side-
chain cyclization of the peptide. After the elongation
completed, a third of the resin was removed, cleaved
and deprotected with triisopropylsilane as scavenger,
leading to the linear azido-peptide 1a. The remaining
peptidyl-resin was cyclized by exposure to 0.5 equiva-
lents of copper (I) iodide, in presence of sodium ascor-
bate and 2,6-lutidine in NMP/DCM (Scheme 1,
method A). The reaction progress was monitored every
24 h by analyzing a sample of the resin by IR-spectros-
copy (disappearance of the 2100 cmÀ1 band characteris-
tic of the azide) and by the modified Kaiser test
described by Punna et al.18
Figure 1. Surface rendering of VEGF165 based on the crystal structure
of VEGF–VEGFR1d2 complex.12 Residues involved in VEGF–VEG-
FR1 interaction are colored and labeled. The encircled areas were
mimicked by the different synthesized peptides.
residues at the surface of the VEGF, we designed small
VEGF mimics as peptides containing these amino acids.
These peptides were cyclised in order to obtain the de-
sired circular arrangements of amino acids. Following
this approach, three families of peptides were synthe-
sized (Table 1).
After 48 h at room temperature, the conversion of the
azide and terminal alkyne in 1,2,3-triazole was complete.
The b-(1H-[1,2,3]triazol-4-yl)alanine amino acid gener-
ated from the propargylglycine is abbreviated as btA
in the following cyclopeptides. The peptide was then
cleaved from the resin, deprotected and purified by
HPLC to a purity of 95%. After these treatments, the
cyclomonomeric peptide 1b (cyclo[GHQMFYYbtA]-
NH2) was exclusively isolated. The linear and cyclic
peptides 2–6 were synthesized according to the same
The cyclic peptide 1b mimics the VEGF residues H27,
Q22, M18, F17, Y21, and Y25 (Fig. 1, green line).
The cyclopeptide 2b mimics L66, D63, N62, K48, M81,
I91, F17, W21 and Y25 amino acids (Fig. 1, red line).
A series of peptides (3b, 4b, 5b, 6b, 7b, 8b, 9b) imitates
the area covered by the VEGF residues Y21, Y25,
D63, E64, G65, L66 and E67 (Fig. 1, blue line).
Table 1.
Compound
Peptide sequencea
Activity at 100 lM (%)b
IC50 (lM)
1a
1b
2a
2b
3a
3b
4a
4b
5a
5b
6a
6b
7a
7b
8a
8b
9a
9b
SP5.2
N3-GHQMFYYPra-NH2
Cyclo[GHQMFYYbtA]-NH2
N3-GLDNKMIFWYPra-NH2
Cyclo[GLDNKMIFWYbtA]-NH2
N3-GhFDEGLEPra-NH2
Cyclo[GhFDEGLEbtA]-NH2
N3-GhFDEPLEPra-NH2
Cyclo[GhFDEPLEbtA]-NH2
N3-YYDEPLEPra-NH2
67
35
95
83
0
19.4 2.2
121 29
22.9 4.2
31.1 1.2
ND
10
0
ND
ND
29
0
ND
ND
Cyclo[YYDEPLEPra]-NH2
N3-FYDEPLEPra-NH2
0
ND
55
61
0
96 14
93 18
ND
Cyclo[FYDEPLEbtA]-NH2
N3-QYDEPLEPra-NH2
Cyclo[QYDEPLEbtA]-NH2
N3-EYDEPLEPra-NH2
0
ND
ND
6
Cyclo[EYDEPLEbtA]-NH2
N3-KYDEPLEPra-NH2
7
ND
151 29
162 24
27
28
99
Cyclo[KYDEPLEbtA]-NH2
NGYEIEWYSWVTHGMY-NH2
28
7
ND, not determined; In this set of experiments, recombinant human VEGF165 displayed an IC50 of 387 60 pM.
a hF, homophenylalanine; Pra, propargylglycine; btA, b-(1H-[1,2,3]triazol-4-yl)alanine.
b The inhibitory activity corresponds to the percentage of biotinylated VEGF165 displaced by 100 lM of peptide on VEGFR1.

5592
V. Goncalves et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5590–5594
Scheme 1. Synthesis of peptide 5b. Method A: Azido-acid coupling followed by Cu(I)-promoted on-resin cyclization. Method B: Direct solid-phase
conversion of the N-terminal amine into azide by diazo-tranfer followed by the cyclization of the peptide.
protocol. For all of the studied peptides, the comparison
of the IR spectra in KBr tablets allowed clearly to distin-
guish the linear peptides, which exhibited a strong
absorption at 2100 cmÀ1, from the cyclic peptide in
which such band was absent. Furthermore all the pep-
tides were purified by HPLC to a purity of at least
95% and their structure confirmed by ESI-mass and
NMR spectroscopy. Comparison of the 1H NMR spec-
tra of the linear and cyclic forms for the compounds 1–9
allowed the characterization of the triazole system. In-
deed all the cyclic peptides spectra show a particular res-
onance around 8 ppm corresponding to the HA proton
of the triazole. Furthermore the a-proton Ha of the ami-
no acid engaged in the triazole ring resonates at 5.5 ppm
and is low field shiefted by around 1.5 ppm when com-
pared to the same proton in the linear form. The assign-
ment of these signals was accomplished after 2D NMR
experiments performed on compounds 8a and 8b. The
observed NOE interaction between the triazole hydro-
gen HA and the proton Ha of the peptide sequence
attested for the ring closure. Moreover, many inter-res-
idue NOE were observed between residues 1 and 8, in
particular between proton Hb of residue 8 and protons
Hb* and Hc* of residue 1 (See: supporting information).
facilitate the cyclization which, indeed, appeared easier
with a shortening of the reaction time (72 h) and a crude
product with higher purity. Therefore, we chose to con-
serve the proline in the following peptides mimicking
this sequence of amino acids.
Subsequently, we proceeded to a structure–activity study
by evaluating the influence of the nature of the N-termi-
nal residue on both cyclization efficiency and VEGFR1
binding affinity of the peptides. First, the peptides 5b
(cyclo[YYDEPLE btA]-NH2) and 6b (cyclo[FYDEPLE
btA]-NH2) were synthesized. For these experiments,
N3a-Tyr(O-tBu)-OH and N3a-Phe-OH were prepared in
respectively 90% and 77% yields following the Lundquist
method. In the peptide 5b the tyrosine are expected to mi-
mic the Y21 and Y25, present in the VEGF while, in 6b,
the first tyrosine was replaced by a phenylalanine. These
two peptides were obtained, respectively, in 8.4% and
7.0% yields and only compound 6b behaved as a VEGFR
antagonist. Furthermore, the cyclization did not lead to
the expected increase of activity when compared to the
linear compound 6a.
Alternatively, we used a more direct method for synthe-
sizing the peptide 5b, i.e. the direct, on-resin, conversion
of the N-terminal amine into azide by diazo-transfer,19
followed by the Cu(I)-promoted cyclization of the pep-
tide (Scheme 1, method B). After the solid-phase elonga-
tion of the peptide, the N-terminal Fmoc protecting
group was cleaved by 20% piperidine in NMP. Then,
an excess of trifluoromethanesulfonic azide in DCM
was added to the resin and the suspension was stirred
overnight at room temperature, leading to the complete
conversion of the amine in the corresponding azide.
Finally, the cyclization was realized following the meth-
od described previously and the peptide was deprotected
The first peptide produced to mimic the VEGF residues
Y21, Y25, D63, E64, G65, L66 and E67 was the peptide
3b (cyclo[GhFDEGLE btA]-NH2. The homophenyl-ala-
nine (hF) was introduced in the sequence in order to al-
low its aromatic ring to mimic the position occupied by
the tyrosine 25 aromatic ring. Compared to the previous
peptides, 1b and 2b, cyclization of 3b proved to be slug-
gish (96 h) and gave raise to unidentified by-products
observed by HPLC analysis of the crude reaction
mixture. Thus, we attempted to replace the glycine
contained in this peptide by a proline (peptide 4b) to

V. Goncalves et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5590–5594
5593
and cleaved by acidic treatment. This direct method
proved to be as efficient as the previous one leading to
the formation of the crude peptide with the same yield
and level of purity.
that might correspond to the formation of a cyclohete-
rodimeric product has been detected. Given the fact that
we used almost the same reaction conditions than Punna
et al. for performing the on-resin cyclization of peptides,
we suggest that the difference of reactivity observed
might be explained by the peptide sequences. Indeed,
the linear peptides synthesized by Punna et al.7 were
characterized by the presence of an amino acid residue
separating the propargylglycine from the resin, and an
azide moiety supported by a relatively long alkyl chain.
As a result, the linear peptide may be sufficiently long
and flexible to allow the preferential formation of
cyclodimeric products following the mechanism they
proposed. By contrast, in our sequences, the propargyl-
glycine was directly linked to the resin and the azide was
supported by the N-terminal carbon-a. Thus, in this
case, the cyclization process may truly benefit from the
pseudo-dilution effect of the resin, and consequently
promote efficiently the formation of cyclomonomeric
products. In addition, the introduction of a proline in
the peptide sequence may facilitate the intramolecular
triazole formation by bringing the azide and alkyne moi-
eties in close proximity.
The second method was applied to the synthesis of the
linear peptides 7a–9a and of their corresponding cyclic
peptides. Overall, the cyclic peptides were obtained after
RP-HPLC purification with 4–9% yields while linear
peptides were recovered with 10–30% yields. The elonga-
tion, cyclization and deprotection steps allowed the for-
mation of the peptides in crude form with a yield of 50%
and a purity of 70%. The cyclization reaction was quan-
titative, as confirmed by IR spectroscopy and the mod-
ified Kaiser test. In order to dispose of pure peptides for
biological studies, a RP-HPLC purification was per-
formed. This purification step was low yielding and ex-
plains the low quantity of the isolated cyclic peptides.
The most interesting aspect of this study is that this so-
lid-phase cyclization approach led exclusively to the for-
mation of cyclomonomeric products, as observed by
HPLC and mass spectroscopy.
In order to experimentally confirm the exclusive forma-
tion of monomeric cyclic peptides, we proceeded to the
cyclization of two peptides differing by the nature of
their N-terminal azido-acid N3FYDEPLEPra-NH2 6a
and N3GYDEPLEPra-NH2 bound on a same resin.
Briefly, the first seven amino acids were coupled on a
rink amide resin and the last residue was introduced as
an equimolar mixture of tyrosine and glycine whose
amine moieties were converted to an azide by on-resin
diazo-transfer. Then we performed as previously the
on resin cyclization by exposure to 0.5 equivalent of
Cu(I). After cleavage and deprotection of a sample of re-
sin, an HPLC analysis pursued before or after cycliza-
tion showed a relatively clean profile presenting two
major peaks corresponding respectively to the two linear
azido-peptides synthesized or to the cyclic ones (Fig. 2).
The peptides affinity for VEGFR1 was determined
thanks to a competition assay previously described.20
Briefly, biotinylated VEGF165 (131 pM) was incubated
with the tested compound in presence of recombinant
human VEGFR1 adsorbed on a microtiterplate. The
biotinylated VEGF165 remaining after wash steps was
detected by chemiluminescence thanks to HRP-conju-
gated streptavidin. Table 1 summarizes the obtained re-
sults. As an internal control, we tested the peptide
SP5.2, identified by phage-display screening and de-
scribed as a VEGFR1-specific antagonist able to inhibit
a broad-range of VEGF-mediated events.21 Evaluated in
our assay, SP5.2 presented an IC50 of 28 lM. The linear
peptides 1a and 2a proved to be potent ligands of the
receptor 1 with respectively, an IC50 of 19 and 23 lM.
The series of linear peptides mimicking the VEGF resi-
dues Y21, Y25, D63, E64, G65, L66 and E67 displayed
strong differences of activity depending on the nature of
the N-terminal azido-acids introduced. Indeed, only
peptides containing a phenylalanine (6a) or a lysine
(9a) exhibited significant activity at 100 lM.
Therefore, as revealed by HPLC traces and MS analysis
the major products formed correspond to the
cyclomonomeric products and no other significant peak
Unfortunately, the corresponding cyclic peptides ap-
peared either as active (2b, 6b, 9b), or less active (1b)
than the corresponding linear peptides. We assume that
the cyclization of these short peptides limited strongly
their backbone flexibility and therefore, prevented the
peptides from adapting to the VEGFR1 surface or that
the selected sequence did not allow a proper alignment
on the receptor surface.
In summary, the similar activities on VEGFR1 of the
peptides 1a and 2a compared to SP5.2 is encouraging
and they could be useful as a template for designing
new VEGF antagonist. The evaluation of their biologi-
cal activities will be investigated. Furthermore, we have
demonstrated, for the first time, that CuI-catalysed
1,3-dipolar cycloaddition of azides and alkynes can be
Figure 2. RP-HPLC analysis of crude products before (A) and after
(B) cyclization of the peptide 6a and its analogue bearing an N-
terminal glycine residue, supported on a same resin.

5594
V. Goncalves et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5590–5594
6. Choi, W. J.; Shi, Z. D.; Worthy, K. M.; Bindu, L.; Karki,
R. G.; Nicklaus, M. C.; Fisher, R. J.; Burke, T. R. Bioorg.
Med. Chem. Lett. 2006, 16, 5265.
applied to the specific synthesis of cyclomonomeric
peptides.
7. Punna, S.; Kuzelka, J.; Wang, Q.; Finn, M. G. Angew.
Chem. Int. Ed. 2005, 44, 2215.
8. Ferrara, N.; Kerbel, R. S. Nature 2005, 438, 967.
9. Takahashi, H.; Shibuya, M. Clin. Sci. (Lond.) 2005, 109,
227.
Acknowledgments
Financial support for this work was provided by the Li-
gue nationale contre le cancer and the University Paris
10. Luttun, A.; Autiero, M.; Tjwa, M.; Carmeliet, P. Biochim.
Biophys. Acta 2004, 1654, 79.
´
Descartes ‘Bonus-Qualite-Recherche’ grant. The
authors thank Dr. Wang-Qing Liu for mass spectral
data.
11. Goncalves, V.; Gautier, B.; Lenoir, C.; Garbay, C.; Vidal,
M.; Inguimbert, N. Pharmachem 2006, 15.
12. Wiesmann, C.; Fuh, G.; Christinger, H. W.; Eigenbrot, C.;
Wells, J. A.; de Vos, A. M. Cell 1997, 91, 695.
13. Christinger, H. W.; Fuh, G.; de Vos, A. M.; Wiesmann, C.
J. Biol. Chem. 2004, 279, 10382.
Supplementary data
14. Muller, Y. A.; Li, B.; Christinger, H. W.; Wells, J. A.;
Cunningham, B. C.; de Vos, A. M. Proc. Natl. Acad. Sci.
U.S.A. 1997, 94, 7192.
15. Keyt, B. A.; Nguyen, H. V.; Berleau, L. T.; Duarte, C. M.;
Park, J.; Chen, H.; Ferrara, N. J. Biol. Chem. 1996, 271,
5638.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2007.07.087.
16. Chang, C. D.; Meienhofer, J. Int. J. Pept. Prot. Res. 1978,
11, 246.
References and notes
17. Lundquist, J. T.; Pelletier, J. C. Org. Lett. 2001, 3, 781.
18. Punna, S.; Finn, M. G. Synlett 2004, 99.
19. Rijkers, D. T. S.; van Vugt, H. H. R.; Jacobs, H. J. F.;
Liskamp, R. M. J. Tetrahedron Lett. 2002, 43, 3657.
20. Goncalves, V.; Gautier, B.; Garbay, C.; Vidal, M.;
Inguimbert, N. Anal. Biochem. 2007, 366, 108.
21. El-Mousawi, M.; Tchistiakova, L.; Yurchenko, L.; Pie-
trzynski, G.; Moreno, M.; Stanimirovic, D.; Ahmad, D.;
Alakhov, V. J. Biol. Chem. 2003, 278, 46681.
1. Wu, P.; Fokin, V. V. Aldrichim. Acta 2007, 40, 7.
2. van Maarseveen, J. H.; Horne, W. S.; Ghadiri, M. R. Org.
Lett. 2005, 7, 4503.
3. Bock, V. D.; Speijer, D.; Hiemstra, H.; van Maarseveen, J.
H. Org. Biomol. Chem. 2007, 5, 971.
4. Brik, A.; Alexandratos, J.; Lin, Y. C.; Elder, J. H.; Olson,
A. J.; Wlodawer, A.; Goodsell, D. S.; Wong, C. H.
Chembiochem 2005, 6, 1167.
5. Angell, Y.; Burgess, K. J. Org. Chem. 2005, 70, 9595.