Synthesis and biological evaluation of new conformationally biased integrin ligands based on a tetrahydroazoninone scaffold

Article abstract of DOI:10.1016/j.bmcl.2006.11.085

The synthesis of new conformationally biased cyclic pentapeptides, incorporating the RGD sequence, and built around a tetrahydroazoninone scaffold, is reported. They exhibit interesting activity towards integrin α_Vβ₃ and a remarkable

Full text of DOI:10.1016/j.bmcl.2006.11.085

Bioorganic & Medicinal Chemistry Letters 17 (2007) 1341–1345
Synthesis and biological evaluation of new conformationally biased
integrin ligands based on a tetrahydroazoninone scaffold
Luca Banfi,a,* Andrea Basso,a Gianluca Damonte,b Federico De Pellegrini,a
Andrea Galatini,a Giuseppe Guanti,a Ilaria Monfardini,a
Renata Rivaa and Carlo Scapollab
aDepartment of Chemistry and Industrial Chemistry, via Dodecaneso 31, 16146 Genova, Italy
bDepartment of Experimental Medicine, Biochemistry section c/o Center of Excellence for Biomedical Research,
University of Genova, v.le Benedetto XV, 16132 Genova, Italy
Received 6 November 2006; revised 27 November 2006; accepted 30 November 2006
Available online 2 December 2006
Abstract—The synthesis of new conformationally biased cyclic pentapeptides, incorporating the RGD sequence, and built around a
tetrahydroazoninone scaffold, is reported. They exhibit interesting activity towards integrin aVb3 and a remarkable selectivity in
comparison with integrin aVb5.
Ó 2006 Elsevier Ltd. All rights reserved.
A successful strategy for the design of small molecule
inhibitors of protein–protein interactions1,2 relies in
the rational transformation of a biologically important
peptide ligand into a peptidomimetic. Since the spatial
position of the amino acid side chains involved in the
interaction is of paramount importance, peptidomimet-
ics should be best constructed around a rigid or semirig-
id scaffold that displays these side chains in the correct
orientation.3 Modification of the scaffold nature may
be valuable as a chemical tool for better understanding
the right topology that leads to potent and selective
binders.
Cyclic pentapeptides containing this motif have been
synthesized by the Kessler’s group as constrained highly
active ligands for integrin aVb3.6 In particular c[f(N-
MeV)RGD] (Cilengitide) is currently under clinical
evaluation.
Later, various groups have substituted the two addition-
al aminoacid units with a rigid monocyclic7,8 or bicyclic
scaffold,9,10 finding, in some cases, novel very efficient
ligands.
Z-Tetrahydroazoninones 1 (Fig. 1) are a new family of
mesocyclic scaffolds corresponding to a constrained
dipeptide.11,12 We have previously shown that this struc-
ture can be accessed in just two synthetic steps, by cou-
Integrins are large heterodimeric surface receptors,
which regulate cell–cell and cell–matrix interactions.
They are classified according to the nature of their a
and b subunits. In the course of an inter-university
national project, we are currently interested especially
in integrins aVb3 and aVb5, which are involved in impor-
tant physiological processes, including tumour induced
angiogenesis.4,5 These two receptors are known to inter-
act with the recognition motif RGD, formed by the
three aminoacids arginine, glycine and aspartic acid.
pling an Ugi multicomponent reaction with
a
subsequent ring closing metathesis.11 An advantage
implied in this synthetic approach is that diverse
R1
N
H
N
2 R = Bn
3 R = H
H
N
R2
O
*
R
O
*
N
O
CO2R4
O
O
H
N
R3
HN
1
NH2
NH2
NH
H
N
O2C
Keywords: Integrin
Peptidomimetics.
ligands;
Azoninones;
Angiogenesis;
O
O
*
Corresponding author. Tel.: +390103536119; fax: +390103536118;
e-mail: banfi@chimica.unige.it
Figure 1.
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.11.085

1342
L. Banfi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1341–1345
substituents can be placed at will as R1 and R2. There-
fore, a ‘fine tuning’ of the structural and conformational
properties of 1 can be achieved in a straightforward way,
without the need to design each time a de novo synthe-
sis. Moreover, the multicomponent reaction allows the
introduction of a first aminoacid (R3CO) making the
overall cyclopeptide synthesis more convergent.
A particular feature of scaffolds 12a,b is the presence of
the N-benzyl group which of course prevents one of the
hydrogen bonds possible in Kessler fVRGD and f(N-
MeV)RGD cyclopentapeptides. In order to check the
importance of the free NH group in this position, we
synthesized also the related scaffolds 14a,b, where a free
NH group is bound to aspartic acid. For this purpose,
benzylamine was substituted with ammonia in the Ugi
reaction.
We have already described the application of this chem-
istry to the synthesis of a first cyclic pentapeptide where
a GGG triad was anchored to the scaffold.13 We now re-
port the synthesis of both cis diastereoisomers of penta-
peptides 2 and 3, containing the RGD recognition
sequence (Fig. 1).
Although examples of Ugi reaction with ketones and
ammonia are very rare in the literature,19,20 we were
able, after some optimization, to obtain adducts 7a,b
in satisfactory yield. In this case separation of the two
diastereoisomers turned out to be impossible neither at
this stage nor later and therefore we carried out the syn-
thesis up to the final product 3 on this 50:50 mixture.
The route from 7a,b to 14a,b was similar to the one al-
ready developed for 12a,b, but was in general less effi-
cient. For example, the RCM was in this case less
clean, giving also by-products derived from intermolec-
ular reactions. As a result, the yield of 9a,b was lower.
Moreover, the stereochemical course of the decarboxyl-
ation reaction was in this case unfavourable, affording
preferentially the unwanted trans isomers in a 58:42 ra-
tio. However, chromatographic separation of cis ad-
ducts 14a,b from their trans isomers 15a,b was very
easy.17
The synthetic routes are depicted in Scheme 1. For the
preparation of the N-benzylated scaffolds 12a,b, we react-
ed isocyanide 413 with commercially available protected
L-aspartic acid 5 and with the preformed imine of 5-hex-
en-2-one and benzylamine.13 When the reaction was car-
ried out under the typical conditions used before (1 M in
EtOH at rt), it was rather sluggish and substantial depro-
tection of the Fmoc group was observed. After a thorough
investigation, we found out that the use of a higher dilu-
tion and of CF3CH2OH/EtOH 2:1 as solvent completely
suppressed Fmoc cleavage, and allowed the obtainment
of the diastereoisomeric mixture of 6a and 6b in good
(77%) yield. As expected, the Ugi reaction gave nearly
no stereoselection at all. However, this fact did not repre-
sent a problem for our purposes (we were interested to test
both stereoisomers). Moreover, we found that the two
diastereoisomers could be very easily separated through
a crystallization from Et2O/petroleum ether. While the
crystalline isomer had a mp of 148.3 °C, the other one
was a foam that resisted all crystallization attempts.14
Having obtained in good or satisfactory overall yields
and in just four steps the enantiomerically pure key scaf-
folds 12a,b and 14a,b, we then converted them into the
RGD containing cyclic peptides 2 and 3 (Scheme 1).21
In the case of 12a and 12b we used a solution-phase pro-
tocol and the Mtr group for protection of the arginine
side chain. Saponification and coupling with the protect-
ed Arg-Gly dipeptide 20,22 followed by sequential
removal of the Alloc group and of the methyl ester, gave
the open pentapeptides 16a and 16b. On the contrary,
for the preparation of 17a,b, we chose a solid-phase ap-
proach, employing, as linker, the acid-labile 2-chlorotri-
tyl group.23 We also used in this case the more easily
removable Pbf group as the guanidine protection.
The ensuing ring closing metathesis, carried out sepa-
rately on the two isomers with Grubbs 1st generation
catalyst,15 turned out to be exceptionally efficient. No
intermolecular side-products were detected and the iso-
lated yields approached 90% (that became 95% taking
into account the recovered starting diene). We think that
the combined bulkiness of the malonate and of the
Fmoc group prevents complexation of the metal by
the amide groups, that is known to be deleterious for
these reactions.13,16 As usual in this series, the double
bond was formed only in the Z configuration.
The stage was now set for the cyclization, that turned
out to be relatively efficient for all the cis compounds
16a, 16b and 17a,b. On the contrary, the trans epimer
of 16b gave the cyclic pentapeptide in very low yield,
facilitating separation of the cis isomer 18b from its
trans counterpart.24
Monosaponification of the malonate in 8a,b caused also
Fmoc deblocking. So we preferred to change at this
stage the amine protecting group shifting to the allyloxy-
carbonyl (Alloc). This was performed in two simple and
high yielding steps. Monosaponification of 10a,b, fol-
lowed by decarboxylation, afforded the key intermedi-
ates 12a,b together with their trans counterparts
13a,b.17 To our pleasure this reaction was somehow
selective favouring the desired cis compounds 12.18
Removal of the two side-chain protection afforded the
final RGD containing peptidomimetics 2a, 2b and
3a,b, which were purified through preparative reverse-
phase HPLC and evaluated in vitro for their ability to
compete with [125I]-echistatin for binding to aVb3 and
aVb5 integrins.25 The results are shown in Table 1. The
two diastereoisomers of 2 showed a very similar activity
at 1 lM and so we decided to study more deeply just one
of them (2b). It was shown to inhibit aVb3 with IC50 of
about 100 nM. Surprisingly, at concentrations between
100 nM and 10 lM, this compound and its isomer 2a
did not show the usual dose-dependent activity,
In the case of 12a separation from 13a was easy, either
via chromatography or crystallization. On the contrary,
we were not able to separate efficiently 12b from 13b.
This turned out not to be a problem, since only 12b cyc-
lized efficiently to the final cyclopeptide (vide infra).

L. Banfi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1341–1345
1343
CO2Et
H
H
N
R1
O
N
R1
O
a (6a,b)
b (7a,b)
CO2Et
H
N
CO2Et
CO2Et
c
d,e
CO2Et
CO2Et
N
R1
O
N
NC
4
CO2Et
CO2Et
O
N
O
O
O
tBuO2C
OH
Alloc NH
CO2tBu
Fmoc NH
CO2tBu
5
NHFmoc
Fmoc NH
CO2tBu
8a,b: R1 = Bn
9a,b: R1 = H
10a,b: R1 = Bn
11a,b: R1 = H
6a,b: R1 = Bn
7a,b: R1 = H
H
N
H
N
H
H
R1
R1
O
R1
O
N
R1
O
N
f,g
N
N
N
N
CO2Et
CO2Et
CO2Et
CO2Et
O
N
O
O
O
O
Alloc NH
Alloc NH
CO2tBu
CO2tBu
CO2tBu
CO2tBu
Alloc NH
Alloc NH
12a,b: R1 = Bn
13a,b: R1 = Bn
14a,b: R1 = H
15a,b: R1 = H
h (from 12a,b)
i (from 14a,b)
H
N
H
N
R1
R1
O
N
O
O
O
O
O
H
H
N
j
k
HN
HN
N
NHR2
NH
NHR2
NH
2a,b
3a,b
NH2
NH
H
N
H
N
tBuO2C
tBuO2C
O
16a,b: R1 = Bn
18a,b: R1 = Bn
O
O
17a,b: R1 = H
19a,b: R1 = H
O
OH
Scheme 1. Reagents and conditions: (a) N-(Hex-5-en-2-ylidene)benzylamine, CF3CH2OH/EtOH 2:1 (0.17 M), 55 °C, 7 days, 77% (6a:6b
ratio = 47:53); (b) 5-hexen-2-one, NH3 (1.5 equiv), NH4Cl (0.5 equiv), EtOH–CF3CH2OH 1:3, 50 °C, 48 h, 47% (7a:7b ratio = 50:50); (c) Grubbs 1st
gen. cat., CH2Cl2 [6 mM (6a,b) or 3 mM (7a,b)], reflux, 96 h, 88% (8a), 85% (8b), 59% (9a,b); (d) Et2NH, CH2Cl2; (e) Allyl chloroformate, NaHCO3,
dioxane-H2O 84% (10a), 84% (10b), 83% (11a,b); (f) NaOH, EtOH; (g) dioxane, reflux, 83% (12–13a, 12:13 ratio = 70:30), 78% (12–13b, 12:13
ratio = 66:34), 79% (14–15a,b, 14:15 ratio = 42:58); (h) 1—NaOH, EtOH; 2—Mtr-L-Arg(NH2)Gly-OMe (20), Bop, Et3N, CH2Cl2, 85% (diast. a),
88% (diast. b); 3—Pd(PPh3)4, dimedone, THF; 4—LiOH, THF-H2O; (i) 1—NaOH, EtOH; 2—Pbf-L-Arg(NH2)Gly-O-linker-resin (see text), TBTU,
HOBT, EtN(iPr)2; 3—Pd(PPh3)4, dimedone, THF; 4—AcOH, CF3CO2H, CH2Cl2 1:1:3; (j) HATU, EtNiPr2, DMF, 10 mM, 53% (16a) 58% (16b),
32% (17a,b) (from 14a,b); (k) CF3CO2H, tioanisole, 43% (2a), 50% (2b), 43% (3a,b).
Table 1. Inhibition of the binding of integrins with echistatin by compounds 2,3a
Compound
Integrin
Concentration (lM)
% inhibition
ST-1646b
2a
2a
aVb3
aVb3
aVb3
aVb3
aVb3
aVb3
aVb3
aVb3
aVb3
aVb3
aVb5
aVb5
aVb5
aVb5
0.001
10
85
48
49
52
50
53
15
5
1
10
2b
2b
1
0.1
2b
2b
0.01
0.001
10
2b
3a,b
3a,b
2a
62
30
17
13
15
13
1
10000
1000
10000
1000
2a
2b
2b
a Values are means of six experiments. Echistatin concentration was 0.04–0.08 nM.
b For the formula of ST-1646 see Refs. 7, 10b.
reaching a plateau at about 50% displacement of echist-
atin, instead of about 100%. We do not know yet the
reason for this anomalous behaviour. An hypothesis
could be a strong allosteric binding to the receptor that
decreases the affinity towards the natural ligand. The
debenzylated adducts 3a,b were found to be less active,
and did not show a plateau. 2b and 2a were also exam-
ined as potential ligands for integrin aVb5. The binding
was in this case weaker indicating a certain degree of
selectivity among the two receptors.
In conclusion, compound 2b shows a potency as aVb3
ligand (IC50 = 100 nM) slightly higher than that of the
first developed Kessler pentapeptide [fVRGD]26
(195 nM when assayed under the same conditions)7,27
and of recently reported compounds where the RGD

1344
L. Banfi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1341–1345
sequence was bonded to an eight-membered lactam.8
However, the potency of 2a,b is inferior to that of
Cilengitide (18.9 nM)6,7,10a,27 and, most of all, of other
recently developed integrin aVb3 ligands,7,10 such as
ST-1646,10 that have IC50 values in the lower nanomolar
range. The higher activity of c[f(N-MeV)RGD] as well
as of the bicyclic bonded RGD compounds10 has been
explained by a shift from a conformation characterized
by a b-turn formed by H-bond between the C@O of
Asp and the NH of Arg, to one possessing two inverse
c-turns instead, leading to a less ‘kinked’ RGD arrange-
ment. One of these c-turns is characterized by H-bond
between the NH of D-Phe (or that pointing out from
the bicyclic scaffold) and the C@O of Gly. In com-
pounds 2a,b the presence of the N-benzyl group in the
position corresponding to D-Phe obviously precludes
this c-turn. Actually, we have previously demonstrated,
in related system,13 that the scaffold favours the b-turn
and a strong H-bond involving NH of R. In 3a,b the
NH pointing out from the scaffold is free, and H-bond
with glycine C@O is now possible. However, activity to-
wards aVb3 is even diminished. This result suggests that
the scaffold itself tends to favour a b-turn and a more
‘kinked’ arrangement.
9. Belvisi, L.; Bernardi, A.; Manzoni, L.; Potenza, D.;
Scolastico, C. Eur. J. Org. Chem. 2000, 2563.
10. (a) Belvisi, L.; Bernardi, A.; Checchia, A.; Manzoni, L.;
Potenza, D.; Scolastico, C.; Castorina, M.; Cupelli, A.;
Giannini, G.; Carminati, P.; Pisano, C. Org. Lett. 2001, 3,
1001; (b) Belvisi, L.; Bernardi, A.; Colombo, M.; Manz-
oni, L.; Potenza, D.; Scolastico, C.; Giannini, G.;
Marcellini, M.; Riccioni, T.; Castorina, M.; LoGiudice,
P.; Pisano, C. Bioorg. Med. Chem. 2006, 14, 169.
11. Banfi, L.; Basso, A.; Guanti, G.; Riva, R. Tetrahedron
Lett. 2003, 44, 7655.
12. Surprenant, S.; Lubell, W. D. Org. Lett. 2006, 8, 2851.
13. Anthoine-Dietrich, S.; Banfi, L.; Basso, A.; Damonte, G.;
Guanti, G.; Riva, R. Org. Biomol. Chem. 2005, 3, 97.
14. We have not been able to determine the relative config-
urations of these two isomers. In this paper, we will
designate with a the isomers derived from the crystalline
Ugi adduct.
15. Abbreviations used in this paper: Grubbs 1st gen. cat.:
benzylidene-bis(tricyclohexylphosphine)ruthenium dichlo-
ride.
phosphonium hexafluorophosphate. HOBT = N-Hydrox-
ybenzotriazole.
TBTU = O-(Benzotriazol-1-yl)-N,N,N0,
N0-tetramethyluronium
tetrafluoroborate. HATU =
Bop = (benzotriazol-1-yloxy)tris(dimeth-ylamino)
O-(7-Azabenzotriazol-1-yl)-N,N,N0,N0-tetramethyluronium
hexafluorophosphate. Mtr = 4-methoxy-2,3,6-trimethyl-
benzene-1-sulfonyl. Pbf = 2,2,4,6,7-pentamethyldihydro-
benzofuran-5-sulfonyl.
Relatively new integrins are now appearing on the
scene28 and are expected to become in the near future
very promising targets for cancer therapy. Therefore,
also semirigid scaffolds like ours may be precious in
assessing the best conditions for high activity and selec-
tivity. The synthetic strategy employed implies that
slight modifications of the scaffold may be realized by
simply changing the inputs of the initial Ugi MCR.
For example, the N-benzyl group in 2a,b represents a
diversity element that can be easily varied, also including
handles for conjugation with cytotoxic drugs.
16. Hebach, C.; Kazmaier, U. Chem. Commun. 2003, 596.
17. The cis compounds were clearly recognized as such on the
basis of NOE experiments.11,13
18. We had indeed already established that the trans isomers
were not well suited for the assembly of a pentapeptide
due to the excessively high distance between the two
handles (Ref. 13).
19. Simoneau, C. A.; George, E. A.; Ganem, B. Tetrahedron
Lett. 2006, 47, 1205.
20. Pick, R.; Bauer, M.; Kazmaier, U.; Hebach, C. Synlett
2005, 757.
21. In the case of 12b we carried out the synthesis up to 18b on
the 66:34 cis:trans mixture of 12b and 13b. During the final
cyclization, the trans isomer gave the cyclopeptide in less
than 10% yield and it could be separated at this stage by
chromatography. The yield reported in the Scheme is
calculated on the cis isomer only. In the case of 14a,b we
carried out the synthesis up to 3a,b without separating the
two diastereoisomers.
Acknowledgments
We thank MIUR and University of Genova (PRIN
2002 and PRIN 2004) for financial assistance.
22. 20 was obtained from L-Fmoc-Arg(Mtr)OH by: (a)
coupling with glycine methyl ester hydrochloride with
EDCI (N-ethyl-N0-dimethylaminopropyl carbodiimide),
HOBt (N-hydroxybenzotriazole), N-methylmorpholine,
DMF, 91%; (b) Et2NH, CH2Cl2; (c) Cbz-Cl, NaHCO3,
dioxane-H2O, 52%; (d) H2, Pd–C, MeOH, 85%. Direct use
of 20 derived from step b was also possible, but the
presence of small amounts of Et2NH lowered the yields of
coupling with the scaffold.
23. In this series, the conditions used for methyl ester
saponification led to partial decomposition. Thus, we
preferred the mild conditions of deblocking of the
2-chlorotrityl linker.
References and notes
1. Gadek, T. R.; Nicholas, J. B. Biochem. Pharmacol. 2003,
65, 1.
2. Yin, H.; Hamilton, A. D. Angew. Chem. Int. Ed. Engl.
2005, 44, 4130.
3. Souers, A. J.; Ellman, J. A. Tetrahedron 2001, 57, 7431.
4. Marinelli, L.; Lavecchia, A.; Gottschalk, K.-E.; Novelli-
no, E.; Kessler, H. J. Med. Chem. 2003, 46, 4393.
5. Gottschalk, K.-E.; Kessler, H. Angew. Chem. Int. Ed.
Engl. 2002, 41, 3767.
24. While 18b and each isomer of 19a,b gave a single set of
signals at NMR at room temperature, in 18a the presence
of three discrete conformations was evident. We think that
they may arise from interconversion of the two possible
rotamers at the tertiary amide as well as of the two
conformations of the ring.13
25. These tests were performed by Dr. Nicoletta Cini and
Prof. Alberto Pupi at the Department of Clinical Physio-
pathology of the University of Florence. We also thank
6. Dechantsreiter, M. A.; Planker, E.; Matha, B.; Lohof, E.;
Hoelzemann, G.; Jonczyk, A.; Goodman, S. L.; Kessler,
H. J. Med. Chem. 1999, 42, 3033.
7. Casiraghi, G.; Rassu, G.; Auzzas, L.; Burreddu, P.;
Gaetani, E.; Battistini, L.; Zanardi, F.; Curti, C.; Nicastro,
G.; Belvisi, L.; Motto, I.; Castorina, M.; Giannini, G.;
Pisano, C. J. Med. Chem. 2005, 48, 7675.
8. Creighton, C. J.; Du, Y.; Santulli, R. J.; Tounge, B. A.;
Reitz, A. B. Bioorg. Med. Chem. Lett. 2006, 16, 3971.

L. Banfi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1341–1345
1345
Prof. Gloria Menchi for her collaboration in these assays.
As a standard in these tests we used ST-1646 (compound
30 of Ref. 10a ), which has IC50 = 1.4 nM under these
conditions. The assays were done essentially as described
in Refs. 7 or 10a.
used echistatin as competitor, instead of vitronectin.
However, an indirect comparison to c[fVRGD] or cilen-
gitide can be made thanks to the experiment reported in
Refs. 7 and 10a. It should be noted that IC50 values (for
aVb3) determined in competitive binding with echistatin
are in general higher than IC50 values for competitive
binding for vitronectin.
26. Wermuth, J.; Goodman, S. L.; Jonczyk, A.; Kessler, H.
J. Am. Chem. Soc. 1997, 119, 1328.
27. A direct comparison between our results and those
reported in Refs. 26, 6, and 8 is not possible since we
28. Marinelli, L.; Meyer, A.; Heckmann, D.; Lavecchia, A.;
Novellino, E.; Kessler, H. J. Med. Chem. 2005, 48, 4204.