Cyclic RGD-peptidomimetics containing bifunctional diketopiperazine scaffolds as new potent integrin ligands

Article abstract of DOI:10.1002/chem.200902398

A perfect fit: The nanomolar affinity of a diketopiperazine-derived cyclic RGD-peptidomimetic for the integrin αvβ 3 receptor can be attributed to its high structural pre-organization. The transdiketopiperazine scaffold induces an extended RGD disposition permitting all important interactions with the extracellular domain of the αvβ 3 integrin. The metal ion in the metaliondependent adhesion site region is represented by a magenta sphere.

Full text of DOI:10.1002/chem.200902398

DOI: 10.1002/chem.200902398
Cyclic RGD-Peptidomimetics Containing Bifunctional Diketopiperazine
Scaffolds as New Potent Integrin Ligands
Ana Sofia M. da Ressurreiżo,[a] Anamaria Vidu,[a, b] Monica Civera,[c] Laura Belvisi,[c]
Donatella Potenza,[c] Leonardo Manzoni,[d] Sandrine Ongeri,[b] Cesare Gennari,*[c] and
Umberto Piarulli*[a]
The tripeptide sequence arginine-glycine-aspartate
(RGD) has been identified as the common motif used by
several endogenous ligands to recognise and bind a group of
integrins,[1] including aVb3, aVb5, a5b1, which play key roles
in angiogenesis, tumor progression and metastasis.[2] The
context of the ligand RGD sequence (flanking residues,
three-dimensional presentation) and individual features of
the integrin binding pockets determine the recognition spe-
cificity and discrimination ability, that is, whether a produc-
tive interaction occurs. A major breakthrough for under-
standing this interaction came in 2002 from the X-ray struc-
ture determination of the complex of integrin aVb3 with
cyclo-[Arg-Gly-Asp-d-Phe-N(Me)-Val] (Cilengitide).[3] This
potent aVb3 ligand was developed by Kessler and co-work-
ers,[4] and is currently in phase III clinical trials for patients
with glioblastoma multiforme as an angiogenesisis inhibitor.
The high activity and selectivity of this derivative has been
attributed to an extended conformation of the RGD motif
showing a distance of about 9 ꢀ between the Cb atoms of
Asp and Arg.[3,5] These observations prompted many other
research groups to investigate the use of conformationally
constrained cyclic RGD peptides and peptidomimetics as
active and selective integrin ligands, encompassing a wide
variety of rigid scaffolds and featuring 14-, 15- and even 16-
membered rings. Among the successful approaches, we like
to mention the g-amino acid RGD-peptidomimetics, con-
taining a g-aminocyclopentane carboxylic acid or a 4-amino-
proline residue (14-membered ring),[6] the azabicyclic lactam
RGD-peptidomimetics (15-membered ring, with the scaffold
mimicking a constrained dipeptide, see for example ST1646
in Table 1),[7] and the b-amino acid RGD-peptidomimetics,
embodying b3-homoamino acids[8] or a cis-b-aminocyclopro-
pane carboxylic acid (16-membered ring).[9]
In a recent paper, we have reported the synthesis of a
new bifunctional diketopiperazine scaffold (DKP-1,
Figure 1), formally derived from l-aspartic acid and (S)-2,3-
diaminopropionic acid, and bearing a carboxylic acid and an
amino functionalities.[10] When inserted into an oligopeptide
sequence, the DKP scaffold acts as a reverse turn inducer.
In addition, DKP-1, while being derived from a-amino
acids, can be seen as a conformationally constrained dipep-
tide formed by two b-amino acids and in particular a (S)-b2-
and a (S)-b3-amino acids (following Seebachꢁs nomencla-
ture).[10]
In the frame of a program aimed at the rational design of
peptidomimetics with a defined secondary structure, we en-
visioned the synthesis, conformational analysis and investi-
gation of the biological activity of cyclic RGD-peptidomi-
metics 3 and 4 (Figure 2), containing the bifunctional diketo-
piperazine scaffolds DKP-1 (cis) and DKP-2 (trans).
[a] Dr. A. S. M. da Ressurreiżo, Dr. A. Vidu, Prof. Dr. U. Piarulli
Universitꢂ degli Studi dell’Insubria
Dipartimento di Scienze Chimiche e Ambientali
Via Valleggio 11, 22100 Como (Italy)
Fax : (+39)031-238-6449
E-mail: umberto.piarulli@uninsubria.it
[b] Dr. A. Vidu, Dr. S. Ongeri
Universitꢃ Paris-Sud, Facultꢃ de Pharmacie
UMR-CNRS 8076, Molꢃcules Fluorꢃes et Chimie Mꢃdicinale
IFR 141, 5, rue J.B. Clꢃment
92296 Chꢄtenay-Malabry cedex (France)
[c] Dr. M. Civera, Dr. L. Belvisi, Dr. D. Potenza, Prof. Dr. C. Gennari
Universitꢂ degli Studi di Milano
Dipartimento di Chimica Organica e Industriale
Centro Interdipartimentale C.I.S.I., Via Venezian 21
20133 Milano (Italy)
Fax : (+39)02-5031-4072
E-mail: cesare.gennari@unimi.it
[d] Dr. L. Manzoni
Consiglio Nazionale delle Ricerche
Istituto di Scienze e Tecnologie Molecolari, Via Fantoli 16/15
20138 Milano (Italy)
The linear pentapeptide mimics were prepared by solu-
tion-phase synthesis using a Boc protection strategy and
tBu/Mtr in the side chains. The dipeptide ArgACHTUNGTRENNUNG(Mtr)-
GlyOMe was coupled to either DKP-1 or DKP-2 and the
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200902398.
12184
ꢅ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 12184 – 12188

COMMUNICATION
Table 1. Inhibition of biotinylated vitronectin binding to avb3 and avb5 re-
ceptors.
Compound
avb3 IC50 [nm][a]
avb5 IC50 [nm][a]
3
4
3898Æ418
3.2Æ2.7
3.2Æ1.3
> 104
114Æ99
cACTHNUTRGNE(NUG RGDfV)
7.5Æ4.8
Figure 1. Structure of the bifunctional diketopiperazine scaffolds DKP-1
and DKP-2, highlighting the conformationally constrained b2–b3 dipep-
tide sequence.
1.0Æ0.5
1.4Æ0.8
[a] IC50 values were calculated as the concentration of compound re-
quired for 50% inhibition of biotinylated vitronectin binding as estimat-
ed by GraphPad Prism software; all values are the arithmetic mean Æ
SD of triplicate determinations.
ST1646, the RGD-peptidomimetic 4 was approximately 40-
fold more selective for the avb3 integrin with respect to the
avb5, in this kind of assay.
Two other cyclic RGD ligands containing a bifunctional
diketopiperazine scaffold were reported in the literature,[14]
namely cyclo-[aspartate-(2S,4S)-4-aminoproline] by Robin-
son and co-workers,[14a] and cyclo-(Asp-Lys) by Albericio
and co-workers.[14b] In both cases the diketopiperazine sub-
stituents in the 3,6-positions were cis. The activities of these
compounds in competitive binding assays towards the avb3
integrin receptor were only modest (in the micromolar
range), and this was attributed to the loose conformational
constraint exerted by the scaffold.
Figure 2. Structure of the cyclic RGD-peptidomimetics 3 and 4 contain-
ing the bifunctional diketopiperazine scaffolds DKP-1 (cis) and DKP-2
(trans), respectively.
resulting tetrapeptide mimics, after Boc cleavage, were at-
tached to Cbz-AspACHTUNGTRENNUNG(OtBu)-OH. Transesterification of the
glycine methyl ester to the corresponding benzyl ester[11]
and hydrogenolysis afforded the linear pentapeptide precur-
sors which were subjected to macrolactamization (HATU,[12]
HOAT, collidine, DMF). The cis-macrolactam (containing
DKP-1) was obtained in good yield (61%) under moderate
dilution conditions (20 mm), while the trans-macrolactam
(containing DKP-2) required high dilution (2 mm) and was
obtained in a passable 41–45% yield. The cis-substituted
linear precursor is apparently better pre-organized for the
intramolecular cyclization than the corresponding trans-
isomer. Global deprotection gave cyclic RGD-peptidomi-
metics 3 and 4 (Figure 2; for details of the synthetic se-
quence, see the Supporting Information).
To investigate the origin of the strikingly different behav-
ior of the RGD-peptidomimetics 3 and 4, a conformational
study was performed by 1H NMR spectroscopy of dilute
H2O/D2O 9:1 solutions and by computational methods.
Both compounds were fully characterized by 1H- and
13C NMR spectroscopy (the NMR data are reported in the
1
Supporting Information). One-dimensional H NMR experi-
ments were conducted to detect intramolecular hydrogen
bonds, by measuring the chemical shift of the N-H protons
and their temperature coefficients (Dd/DT). NOESY spectra
were recorded to investigate both sequential and long-range
NOE signals that provide evidences of preferred conforma-
tions. Three-dimensional structures satisfying long-range
NOE contacts were generated by restrained mixed-mode
Metropolis Monte Carlo/Stochastic Dynamics (MC/SD) sim-
ulations,[15] using the implicit water GB/SA solvation
model.[16]
The RGD-peptidomimetic 3 exists in two different pre-
ferred conformations as reported in Figure 3 (A and B).
The NOESY spectrum of 3 shows two mutually exclusive
long-range NOE contacts. The cross peak between DKP-
NH10 and Asp-NH (strong) is indicative of a b-turn confor-
mation stabilized by a hydrogen bond between DKP-NH10
and Arg-C=O. The chemical shift value (7.46 ppm) and the
Dd/DT value (À2 ppbKÀ1) of the amide proton DKP-NH10
indicate that this proton is strongly locked in an intramolec-
ularly H-bonded state (Figure 3A). The cross peak between
Gly-NH and Asp-NH (medium) is indicative of a b-turn
The cyclic RGD peptidomimetics were examined in vitro
for their abilities to compete with biotinylated vitronectin
for binding to the purified avb3 and avb5 receptors (Table 1).
Screening assays were performed by measuring the effect of
the RGD macrolactams 3 and 4 on the interaction between
immobilized integrin receptors and biotinylated soluble li-
gands. The ability of the new compounds to inhibit the bind-
ing of vitronectin to the isolated avb3 and avb5 receptors was
compared with that of the standard compound c
(RGDfV)[13]
ACHTUNGTRENNUNG
and compound ST1646[7] (Table 1).
Interestingly, the cyclic RGD-peptidomimetic 4, contain-
ing the trans-diketopiperazine scaffold DKP-2, exhibited a
nanomolar affinity for the avb3 integrin, comparable to c-
ACHTUNGTRENNUNG(RGDfV) and ST1646. On the contrary, RGD-peptidomi-
metic 3, bearing the cis-diketopiperazine DKP-1, showed
only a modest micromolar affinity towards the avb3 integrin.
Moreover, unlike reference compounds cACTHNURGTNE(NUG RGDfV) and
Chem. Eur. J. 2009, 15, 12184 – 12188
ꢅ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12185

C. Gennari, U. Piarulli et al.
Figure 3. Preferred intramolecular hydrogen-bonded patterns proposed
for compound 3 on the basis of spectroscopic data. A) b turn at Gly-Asp.
B) b turn at Arg-Gly. The arrows indicate significant NOE contacts.
Figure 4. Structures of 3 as obtained by restrained MC/SD simulations
based on experimental distance information. A) g turn at Gly and bII’-
turn at Gly-Asp [C
N
bACHTUNGTRNE(NUNG Asp)=7.9 ꢀ]. B) g turn at Arg and bII’
turn at Arg-Gly [CbA(Arg)–C CHTUGNTRENNUNG
C
bA(Asp)=5.9 ꢀ].
conformation stabilized by a hydrogen bond between Asp-
NH and C(8)=O (Figure 3B).
Two 10 ns MC/SD simulations were performed for com-
pound 3 applying the DKP-NH10/Asp-NH or the Asp-NH/
Gly-NH distance restraint derived from NOESY spectra.
Approximately 70% of the conformations sampled during
the first simulation adopted a non-extended arrangement of
the RGD sequence characterized by a b turn at Gly-Asp
and the presence of the corresponding hydrogen bond be-
tween DKP-NH10 and Arg-C=O. In addition, the formation
of a g turn at Gly stabilized by the hydrogen bond between
Asp-NH and Arg-C=O was observed for 50% of the simula-
between DKP-NH10 and Gly-C=O was observed for 23%
of the simulation. A CbA(Arg)–CbA(Asp) average distance of
9.3 ꢀ was obtained during this MC/SD calculation. A repre-
sentative energy minimized conformation selected by cluster
analysis and featuring the H-bond between the Gly-NH and
C(5)=O is shown in Figure 5.
G
CHTUNGTRENNUNG
tion. A CbACHTUNGTRENNUNG(Arg)–CbACHTUNGTRENNUNG(Asp) average distance of 7.4 ꢀ was ob-
tained during this MC/SD calculation. A representative
energy minimized conformation selected by cluster analysis
and featuring both H-bonds is shown in Figure 4A. More
than 90% of the conformations sampled during the simula-
tion of 3 featuring the Asp-NH/Gly-NH distance restraint,
adopted a non-extended arrangement of the RGD sequence
characterized by a b turn at Arg-Gly and the corresponding
hydrogen bond between Asp-NH and C(8)=O. In addition,
the formation of a g turn at Arg stabilized by the hydrogen
bond between Gly-NH and C(8)=O was observed for 50%
Figure 5. Left: Preferred intramolecular hydrogen-bonded pattern pro-
posed for compound 4 on the basis of spectroscopic data. The arrow indi-
cates a significant NOE contact. Right: Structure of 4 as obtained by re-
strained MC/SD simulations based on experimental distance information
(distorted inverse g turn at Asp and pseudo b turn at DKP-Arg, Cb-
of the simulation. The CbACTHNUTRGNEN(UG Arg)–CbACHTUGNTREN(NUGN Asp) average distance in
this MC/SD calculation was 6.5 ꢀ. A representative energy
minimized conformation selected by cluster analysis and
featuring both H-bonds is shown in Figure 4B.
ACHUTNGERN(NUG Arg)–CbAHCTUNGTREN(NGUN Asp)=9.4 ꢀ).
On the contrary, the NOESY spectrum of compound 4
shows only one relevant long-range interaction between
Gly-NH and Arg-NH: this NOE is indicative of a b-turn
motif stabilized by a hydrogen bond between Gly-NH and
C(5)=O (confirmed by the chemical shift and the Dd/DT
value of Gly-NH) (Figure 5). The distance restraint corre-
sponding to the NOE contact between Gly-NH and Arg-
NH was applied in the 10 ns MC/SD simulation of com-
pound 4. More than 90% of the conformations sampled
during the simulation adopted an extended arrangement of
the RGD sequence characterized by a pseudo b turn at
DKP-Arg and the formation of the corresponding hydrogen
bond between the Gly-NH and C(5)=O. In addition, the for-
mation of a g turn at Asp stabilized by the hydrogen bond
In order to rationalize, on a molecular basis, the affinity
of compounds 3 and 4 for the avb3 receptor, docking studies
were performed starting from the representative conforma-
tions obtained from the restrained MC/SD simulations. The
crystal structure of the extracellular segment of integrin avb3
complexed with the cyclic pentapeptide Cilengitide (1L5G,
pdb code) was taken as a reference model for the interpre-
tation of the docking results in terms of ligand-protein inter-
actions.[3] In the X-ray complex, Cilengitide binds to the in-
terface of the a and b units forming specific electrostatic in-
teractions. The acid and basic pharmacophoric groups and
their orientation are essential for binding to avb3 because
12186
www.chemeurj.org
ꢅ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 12184 – 12188

RGD Peptidomimetics
COMMUNICATION
they act like an electrostatic clamp, interacting with charged
regions of the receptor binding site.[3]
For peptidomimetic 3, docking calculations starting from
both the representative geometries of Figure 4 produced
top-ranked poses conserving optimal interactions only with
the a subunit of the avb3 receptor. Probably, the short Cb-
not able to properly fit into the avb3 receptor. On the con-
trary, the nanomolar affinity of macrolactam 4 for avb3 can
be attributed to its high structural pre-organization. The
trans DKP-2 scaffold induces a preferred conformation with
Arg in the i+1 position of a pseudo b turn which deter-
mines an extended RGD disposition similar to the RGD
bound conformation of Cilengitide.
ACHTUNGTRENNUNG(Arg)–CbACHTUNGTRENNUNG(Asp) distances of these geometries prevent the
guanidine and carboxylic groups from achieving the re-
quired separation for binding to the avb3 integrin.[17] On the
other hand, docking calculations starting from the extended
geometry of peptidomimetic 4 reported in Figure 5 pro-
duced top-ranked poses conserving all the important inter-
actions of the X-ray complex (see Figure 6). The positively
Acknowledgements
We thank the European Commission (Marie Curie Early Stage Research
Training Fellowship “Foldamers” MEST-CT-2004-515968) for financial
support and for PhD Fellowships (to A.S.M.R. and A.V.). We also grate-
fully acknowledge Ministero dell’Universitꢂ e della Ricerca for financial
support.
Keywords: conformation analysis
· diketopiperazines ·
integrin ligands · molecular modeling · peptides
[1] a) R. O. Hynes, Cell 2002, 110, 673–687; b) Y. Takada, X. Ye, S.
Simon, GenomeBiology 2007, 8, 215.1–215.9.
[2] J. F. Hood, D. A. Cheresh, Nat. Rev. Cancer 2002, 2, 91–100.
[3] J.-P. Xiong, T. Stehle, R. Zhang, A. Joachimiak, M. Frech, S. L.
Goodman, M. A. Arnaout, Science 2002, 296, 151–155.
[4] M. A. Dechantsreiter, E. Planker, B. Mathꢆ, E. Lohof, G. Hçlze-
mann, A. Jonczyk, S. L. Goodman, H. Kessler, J. Med. Chem. 1999,
42, 3033–3040.
[5] K.-E. Gottschalk, H. Kessler, Angew. Chem. 2002, 114, 3919–3927;
Angew. Chem. Int. Ed. 2002, 41, 3767–3774.
[6] a) G. Casiraghi, G. Rassu, L. Auzzas, P. Bureddu, E. Gaetani, L.
Battistini, F. Zanardi, C. Curti, G. Nicastro, L. Belvisi, I. Motto, M.
Castorina, G. Giannini, C. Pisano, J. Med. Chem. 2005, 48, 7675–
7687; b) F. Zanardi, P. Bureddu, G. Rassu, L. Auzzas, L. Battistini,
C. Curti, A. Sartori, G. Nicastro, G. Menchi, N. Cini, A. Bottonocet-
ti, S. Raspanti, G. Casiraghi, C. Pisano, J. Med. Chem. 2008, 51,
1771–1782.
[7] a) L. Belvisi, A. Bernardi, M. Colombo, L. Manzoni, D. Potenza, C.
Scolastico, G. Giannini, M. Marcellini, T. Riccioni, M. Castorina, P.
LoGiudice, C. Pisano, Bioorg. Med. Chem. 2006, 14, 169–180; b) L.
Manzoni, L. Belvisi, D. Arosio, M. Civera, M. Pilkington-Miksa, D.
Potenza, A. Caprini, E. M. V. Araldi, E. Monferrini, M. Mancino, F.
Podestꢂ, C. Scolastico, ChemMedChem 2009, 4, 615–632.
[8] F. Schumann, A. Mꢇller, M. Koksch, G. Mꢇller, N. Sewald, J. Am.
Chem. Soc. 2000, 122, 12009–12010.
[9] S. Urman, K. Gaus, Y. Yang, U. Strijowski, N. Sewald, S. De Pol, O.
Reiser, Angew. Chem. 2007, 119, 4050–4053; Angew. Chem. Int. Ed.
2007, 46, 3976–3978.
[10] A. S. M. Ressurreiżo, A. Bordessa, M. Civera, L. Belvisi, C. Gen-
nari, U. Piarulli, J. Org. Chem. 2008, 73, 652–660.
[11] R. Imwinkelried, M. Schiess, D. Seebach, Org. Synth. 1987, 65, 230–
233.
[12] L. A. Carpino, A. El-Faham, F. Albericio, J. Org. Chem. 1995, 60,
3561–3564.
[13] a) G. Mꢇller, M. Gurrath, H. Kessler, J. Comput.-Aided Mol. Des.
1994, 8, 709–730; b) R. Haubner, R. Gratias, B. Diefenbach, S. L.
Goodman, A. Jonczyk, H. Kessler, J. Am. Chem. Soc. 1996, 118,
7461–7472; c) R. Haubner, W. Schmitt, G. Hçlzemann, S. L. Good-
man, A. Jonczyk, H. Kessler, J. Am. Chem. Soc. 1996, 118, 7881–
7891.
[14] a) C. Bisang, C. Weber, J. A. Robinson, Helv. Chim. Acta 1996, 79,
1825–1842; b) M. Royo, W. Van den Nest, M. del Fresno, A. Frie-
den, D. Yahalom, M. Rosenblatt, M. Chorev, F. Albericio, Tetrahe-
dron Lett. 2001, 42, 7387–7391.
Figure 6. Best pose of compound 4 (atom colour tube representation)
into the crystal structure of the extracellular domain of avb3 integrin (a
unit red and b unit blue wire representation) overlaid on the bound con-
formation of Cilengitide (green tube representation). Only selected integ-
rin residues involved in the interactions with the ligand are shown. The
Mn2+ ion at MIDAS is shown as a magenta CPK sphere. Non-polar hy-
drogens were removed for clarity.
charged Arg guanidinium group of the ligand interacts with
the negatively charged side chains of Asp218 and Asp150 in
the a unit, one carboxylate oxygen of the ligand Asp side
chain is coordinated to the metal cation in the metal-ion-de-
pendent adhesion site (MIDAS) region of the b unit, while
the second carboxylate oxygen forms hydrogen bonds with
the backbone amides of Asn215 and Tyr122 in the b unit.
Further stabilizing interaction involves the formation of a
hydrogen bond between the ligand backbone NH of the
Asp residue and the backbone carbonyl group of Arg216 in
the b unit.
In light of all these considerations, the micromolar affinity
of macrolactam 3 for avb3 (Table 1) can be explained in
terms of its low pre-organization for binding. In fact, as de-
termined by the computational and NMR studies, com-
pound 3 in solution mainly features non-extended RGD
conformations which, according to the docking results, are
Chem. Eur. J. 2009, 15, 12184 – 12188
ꢅ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12187

C. Gennari, U. Piarulli et al.
[15] F. Guarnieri, W. C. Still, J. Comput. Chem. 1994, 15, 1302–1310.
[16] W. C. Still, A. Tempczyk, R. C. Hawley, T. Hendrickson, J. Am.
Chem. Soc. 1990, 112, 6127–6129.
pound 3, probably also adopts more extended RGD arrangements
which are able to properly fit into the aVb3 receptor; this would ex-
plain its nanomolar affinity. A more thorough discussion will be re-
ported in a forthcoming full paper.
[17] As pointed out by a referee, a non-extended RGD conformation
with a relatively short distance (6.7–7.4 ꢀ) between C
(Asp) has been proposed for c(RGDfV), see reference [13]. c-
(RGDfV), which is conformationally much more flexible than com-
bACHTUNGTERN(NUNG Arg)–Cb-
G
E
Received: August 31, 2009
Published online: October 14, 2009
R
12188
www.chemeurj.org
ꢅ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 12184 – 12188