Antiangiogenic effect of dual/selective α5 β1/αvβ3 integrin antagonists designed on partially modified retro-inverso cyclotetrapeptide mimetics

Article abstract of DOI:10.1021/jm9013532

Recent evidence highlighted the role of α₅β ₁ integrin in angiogenesis and in regulating α _vβ₃ integrin function. As a consequence, selective α₅β₁ integrin inhibitors or dual α₅β₁/α_vβ₃ integrin inhibitors are considered promising candidates for the development of cancer therapeutic agents. In this paper, we describe the synthesis and pharmacological characterization of a minilibrary of cyclotetrapeptide mimetics containing a PMRI Arg-Gly-Asp sequence. In particular, c[(R)- βPheψ(NHCO)Aspψ-(NHCO)Gly-Arg] (3) displayed a good activity in inhibiting the α_vβ₃ integrin-mediated cell adhesion of fibronectin or vitronectin, as well as the adhesion of fibronectin to the α₅β₁ integrin. Interestingly, the diastereomeric compound c[(S)-βPheψ(NHCO)Aspψ(NHCO)Gly-Arg] (2) maintained a good efficacy in inhibiting α₅β₁ integrin while gaining a certain selectivity over α_vβ ₃ integrin. These two integrin antagonists significantly inhibited bFGF-induced human endothelial cell tube formation at submicromolar concentrations. Conformational analysis and Molecular Docking calculations suggest that the different α₅β₁ versus α_vβ₃ selectivity of 2 and 3 can be rationalized on the basis of the alternative display of the aromatic side chain adjacent to Asp.

Full text of DOI:10.1021/jm9013532

106 J. Med. Chem. 2010, 53, 106–118
DOI: 10.1021/jm9013532
Antiangiogenic Effect of Dual/Selective r5 β1/rvβ3 Integrin Antagonists Designed on Partially
Modified Retro-Inverso Cyclotetrapeptide Mimetics
Luca Gentilucci,*,† Giuliana Cardillo,† Santi Spampinato,‡ Alessandra Tolomelli,† Federico Squassabia,§ Rossella De Marco,†
Andrea Bedini,‡ Monica Baiula,‡ Laura Belvisi, and Monica Civera
†Department of Chemistry “G. Ciamician”, Universitaꢀ degli Studi di Bologna, via Selmi 2, 40126 Bologna, Italy, ‡Department of Pharmacology,
Universitaꢀ degli Studi di Bologna, via Irnerio 48, 40126 Bologna, Italy, §Stepbio srl, Bologna, Via P. Nanni Costa 12/3/e-f, 40133 Bologna, Italy,
and Department of Organic and Industrial Chemistry and CISI, University of Milano, Via Venezian 21, Milano, Italy
Received May 20, 2009
Recent evidence highlighted the role of R5β1 integrin in angiogenesis and in regulating Rvβ3 integrin
function. As a consequence, selective R5β1 integrin inhibitors or dual R5β1/Rvβ3 integrin inhibitors are
considered promising candidates for the development of cancer therapeutic agents. In this paper, we
describe the synthesis and pharmacological characterization of a minilibrary of cyclotetrapeptide
mimetics containing a PMRI Arg-Gly-Asp sequence. In particular, c[(R)-βPheψ(NHCO)Aspψ-
(NHCO)Gly-Arg] (3) displayed a good activity in inhibiting the Rvβ3 integrin-mediated cell adhesion
of fibronectin or vitronectin, as well as the adhesion of fibronectin to the R5β1 integrin. Interestingly, the
diastereomeric compound c[(S)-βPheψ(NHCO)Aspψ(NHCO)Gly-Arg] (2) maintained a good efficacy
in inhibiting R5β1 integrin while gaining a certain selectivity over Rvβ3 integrin. These two integrin
antagonists significantly inhibited bFGF-induced human endothelial cell tube formation at submicro-
molar concentrations. Conformational analysis and Molecular Docking calculations suggest that the
different R5β1 versus Rvβ3 selectivity of 2 and 3 can be rationalized on the basis of the alternative display
of the aromatic side chain adjacent to Asp.
Introduction
The endogenous extracellular matrix ligands vitronectin and
fibronectinbindtotheRvβ3 integrin, and fibronectin binds also
to R5β1 integrin by recognition of the acidic and guanidino side
chain groups of the same tripeptide motif, the Arg-Gly-Asp
(RGD)a sequence.14 Therefore, small RGD-mimetic mole-
cules that are able to interfere with the R5β1 and/or Rvβ3
integrins are currently considered of interest for the angiostatic
therapy of cancer.15-17 The RGD sequence is the recognition
site for various integrin receptors, and the conformation of the
sequence in the individual adhesion protein or peptide is
critical for the specificity of this recognition.18,19
Among the different members of the integrin receptor
family, Rvβ3 integrin is generally considered a privileged target
for antiangiogenic therapy.1-6 At the same time, R5β1 integrin
has emerged as a new promising target for the development of
cancer therapeutic agents. Besides an important function in
the migration of activated lymphocytes during the immune
response and the involvement in diabetes and inflammatory
diseaseslikerheumatoidarthritis,7 the proangiogenicfunction
of the R5β1 receptor has been clearly demonstrated.8
Interestingly, the R5β1 integrin has been shown to affect
Rvβ3-mediated endothelial cell migration and angiogenesis via
regulation of Rvβ3 integrin function.9 A study with Rvβ3 and
R5β1 specific antibodies demonstrated that integrin R5β1
regulates the function of integrin Rvβ3 on endothelial cells
during their migration in vitro or angiogenesis in vivo.10
Likewise, it has been reported that the combined antagonism
of both Rvβ3 and R5β1, as opposed to Rvβ3 alone, induces
apoptosis of angiogenic endothelial cells cultured on type I
collagen.11
a Abbreviations: bFGF, basic fibroblast growth factor; RGD, Arg-
Gly-Asp; CTP, cyclotetrapeptide; PMRI, partially modified retro-in-
verso; 2D and 3D, two- and three-dimensional, respectively; Cbz,
benzyloxycarbonyl; Fmoc, fluorenylmethyloxycarbonyl; Bz, benzyl;
DPPA, diphenylphosphoryl azide; TFA, trifluoroacetic acid; EDCI,
N-[3-(dimethylamino)propyl]-N0-ethylcarbodiimide; HOBt, 1-hydroxy-
benzotriazole; TEA, triethylamine; DCM, dichloromethane; DMF, N,
N-dimethylformamide; DMA, dimethylamine; THF, tetrahydrofuran;
DMSO, dimethyl sulfoxide; K562, human erythroleukemic cells; SK-
MEL-24, human malignant melanoma cells; HUVEC, human umbilical
vein endothelial cells; BSA, bovine serum albumin; ECM, extracellular
matrix; SEM, standard error of the mean; ANOVA, analysis of variance
test; MD, molecular dynamics; VT, variable temperature; rmsd, root-
mean-square deviation; AMBER, assisted model building with energy
refinement; TIP3P, transferrable intermolecular potential three point;
MIDAS, metal ion-dependent adhesion site; CPK, Corey, Pauling, and
Koltun; DAD, diode array detector; ODS, octadecyl silane; EDTA,
ethylenediaminetetraacetic acid; MEM, minimum essential medium;
RPMI, Roswell Park Memorial Institute; FBS, fetal bovine serum;
PBS, phosphate-buffered saline; PMA, phorbol 12-myristate 13-acetate;
PI, propidium iodide; HMBC, heteronuclear multiple-bond coherence;
HSQC, heteronuclear single-quantum coherence; rt, room temperature.
Another approach to developing integrin inhibitors is to
modify extracellular matrix proteins like angiostatin, tamu-
statin, arresten, and endostatin capable of antagonizing Rvβ3
and R5β1.12 A laminin-1-derived peptide that similarly blocks
Rvβ3 and R5β1 has been shown to inhibit growth of melanoma
cells in vivo.13
*To whom correspondence should be addressed. E-mail: luca.
gentilucci@unibo.it. Phone: þ39 0512099462. Fax: þ39 0512099456.
r
pubs.acs.org/jmc
Published on Web 11/30/2009
2009 American Chemical Society

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1 107
Figure 1. General structure of the PMRI-cyclotetrapeptide models c[(S/R)-βPheψ(NHCO)(S/R)-Alaψ(NHCO)Gly-(S/R)-Phe] 1 and
structures of stereoisomeric PMRI RGD mimetics c[(S/R)-βPheψ(NHCO)(S/R)-Aspψ(NHCO)Gly-(S/R)-Arg] 2-9.
With regard to the Rvβ3 integrin receptor, a large number of
peptidic, peptidomimetic,20,21 and nonpeptidomimetic22 scaf-
folds have been successfully employed to provide the desired
conformational constraint to maintain the acidic and basic
side chains at the appropriate distance and in a conformation
suitable for binding. The cyclopentapeptide c[Arg-Gly-Asp-
(R)-Phe-Val] has represented the first highly active and
selective Rvβ3 integrin antagonist23-25 and has served as a
lead structure for the development by Kessler et al. of the
cyclic pentapeptide cilengitide, c[Arg-Gly-Asp-(R)-Phe-
NMeVal] (EMD121974),26 currently in clinical trials for
antiangiogenic cancer therapy.27,28 Furthermore, the crystal
structure of the extracellular segment of the Rvβ3 integrin with
cilengitide bound to the active site has been disclosed pre-
viously,29 giving the opportunity to design novel integrin
antagonists based on the receptor-bound conformation of
the RGD tripeptide.30
On the other hand, the experimental 3D structure of
R5β1 integrin is not available, and few structural details about
ligand-receptor interactions have been obtained until now.31,32
As a consequence, only some potent and selective antagonists of
R5β1 or dual R5β1/Rvβ3 antagonists have been described.33-36
Recently, a homology model of the receptor has been pub-
lished,37 which has successfully been used to optimize ligand
activity and selectivity.38
In essence, the criteria for designing integrin antagonists
showing selectivity for a certain RGD-binding integrin type
over the others seem to reside for the most part in subtle
differences in the reciprocal orientation and distance between
the Asp and Arg side chains, and in the disposition of an
aromatic side chain adjacent to Asp. In particular, the pre-
sence of definite secondary structure elements in RGD linear
or cyclic peptides has often been correlated to the ligand’s
specificity.20,30,31,38 Therefore, the synthesis of novel, confor-
mationally definite RGD-containing scaffolds to mimic im-
portant secondary structure elements such as γ- or β-turns is
of great interest.
Very recently,39,40 we designed a new kind of cyclotetra-
peptide (CTP) scaffold based on a partially modified retro-
inverso (PMRI) variant of CTP containing a β-amino acid,
forming 13-membered rings.41 Retro-inverso and PMRI ana-
logues have been widely utilized to improve the performance of
a bioactive parent peptide.42-46 We envisaged the opportunity
to utilize these scaffolds for designing peptidomimetic mole-
cules with
a well-defined 3D display of the RGD
sequence and aimed to discover selective R5β1 integrin anta-
gonists or dual Rvβ3/R5β1 integrin antagonists, whose activity
could be synergistically effective in preventing angiogenesis.
Results
Design and Synthesis. Compared to CTPs composed of all
R-amino acids, which can be considered the smallest turn-
mimetic structure, 13-membered analogues incorporating a
β3- or β2-amino acid proved to be easier to synthesize and
conformationally more defined.41,47-49 The eight stereoisomeric
PMRI-CTP models (1) of the general sequence c[(S/R)-βPheψ-
(NHCO)(S/R)-Alaψ(NHCO)Gly-(S/R)-Phe] (Figure 1) con-
tain a 1,2-diamine to replace the β-amino acid, and a diacid as
a Gly mimetic. Conformational analysis revealed that these
13-membered scaffolds manifest specific 3D geometries in com-
parison to normal 13-membered CTPs39 and have a strong
tendency to adopt turn structures. This preference can be
ascribed to the propensity of the PMRI-CTP structures to
stabilize H-bonded conformations involving the diamine amide
protons.50
On the basis of the structures of the scaffolds 1, we
designed a minilibrary of PMRI RGD-mimetic com-
pounds,21 aiming to obtain integrin antagonists having the
pharmacophoric side chains in well-defined, predictable
spatial dispositions. For this purpose, we prepared the
stereoisomeric compounds c[(S/R)-βPheψ(NHCO)(S/R)-
Aspψ(NHCO)Gly-(S/R)-Arg] 2-9 by introducing (S)- or
(R)-Asp and (S)- or (R)-Arg in place of (S)- or (R)-Ala and
(S)- or (R)-Phe, respectively (some preliminary results of the

108 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1
Scheme 1. Synthesis of 2a
Gentilucci et al.
a (i) EDCI/HOBt/TEA, DCM/DMF; (ii) DMA/THF; (iii) CH3CN, 70 °C; (iv) H2, Pd/C, EtOH; (v) DPPA/NaHCO3, DMF; (vi) 94:2:1:2:1 TFA/
PhOH/PhSH/H2O/Et2S.
Table 1. Analytical Characterization of 2-9 and Inhibition of Rvβ3 and R5β1 Integrin-Mediated Cell Adhesion to Fibronectin (FN) and Vitronectin
(VN) in the Presence of 2-9, or Reference Cyclic Peptide c[Arg-Gly-Asp-(R)-Phe-Val] and Negative Control Peptide c[Arg-Ala-Asp-(R)-Phe-Val],
c[RGDfV] and c[RADfV], Respectively, in Brief
Rvβ3 vs FN
Rvβ3 vs VN
R5β1 vs FN
compd
purity (%)
MS [M þ 1]a
(SK-MEL-24) IC50 ( μM)b
(HUVEC) IC50 ( μM)b
(K562) IC50 ( μM)b
2
3
4
5
6
7
8
9
97
96
95
95
95
95
96
96
-
490.2
490.3
490.3
490.4
490.2
490.1
490.2
490.3
-
11 ( 6
25.2 ( 5.2
0.24 ( 0.08
1.5 ( 0.4
>100
0.52 ( 0.04
0.024 ( 0.003
4.7 ( 0.4
>100
0.18 ( 0.05
2.3 ( 0.3
>100
>100
>100
>100
>100
>100
>100
2.1 ( 0.3
>100
0.92 ( 0.09
>100
1.3 ( 0.1
>100
c[RGDfV]c
c[RADfV]c
7.1 ( 2.7
>100
9.6 ( 2.9
>100
34.7 ( 8.4
>100
-
-
a Calculated [M þ 1] of 490.2. b Values are means ( the standard error of three experiments conducted in quadruplicate. c Purchased from Bachem.
Rvβ3 integrin-mediated cell adhesion inhibition in the pre-
sence of 2-4 and 8, at a single concentration, have been
reported in a previous paper50).
DPPA, gave cyclopeptide 14 in good yield and purity after
preparative RP-HPLC. The final deprotection of Asp and
Arg side chains was performed with TFA in the presence of a
mixture of scavengers. PMRI-CTP 2 was isolated by pre-
parative RP-HPLC.
Accordingly, we prepared the remaining PMRI RGD
analogues 3-9; purities and mass characterizations are
reported in Table 1.
Inhibition of Cell Adhesion. The ability of compounds 2-9
to inhibit the adhesion of K562 (human erythroleukemic
cells, expressing R5β1 integrin) or SK-MEL-24 (human
malignant melanoma cells, expressing Rvβ3 integrin) to im-
mobilized fibronectin and the adhesion of HUVEC (human
umbilical vein endothelial cells, expressing Rvβ3 integrin) to
immobilized vitronectin was evaluated. These cell models are
widely used to investigate potential antagonists of Rvβ3
integrin (HUVEC and SK-MEL-24)53,54or of R5β1 integrin
(K562).55 In these experiments, the cells were seeded onto
plates coated with different substrata and allowed to adhere
before quantitation of the number of adherent cells. Under
these conditions, no significant cell adhesion was observed
for BSA-coated plates (negative control) or nonspecific
The cyclic PMRI RGD mimetics were easily obtained
from the corresponding linear precursors. As a prototypic
example, the synthesis of 2, the first member of the mini-
library, is shown in Scheme 1, the syntheses of the other
stereoisomers (3-9) being the same.
The retrosynthetic analysis of the protected linear precur-
sor 13 was thought to proceed by in-solution coupling of
51
fragments 11 and 12. Reduction of Cbz-Phe-NH2 with
BH352 gave chiral Cbz-1,2-diamine 10 in excellent yield. The
coupling of 10 under standard conditions with Fmoc-Arg-
(Mtr)-OH afforded dipeptide 11, whose deprotection was
performed by treatment with 2 M DMA in THF.
The second fragment was straightforwardly prepared
from Meldrum’s acid and H-Asp(Ot-Bu)-OBz, giving dipep-
tide acid 12, not needing further deprotection steps. The
standard coupling of dipeptides 11 and 12 gave fully pro-
tected linear tetrapeptide 13. The removal of the Cbz and
benzyl ester protecting groups at the N- and C-termini,
respectively, by hydrogenolysis, followed by cyclization with

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1 109
Figure 2. Rvβ3/R5β1 integrin antagonists inhibit bFGF-induced tube formation in HUVEC on Matrigel. (A) Representative pictures of
capillary-like structures (the black bar in the pictures corresponds to 100 μm): (CTRL) HUVEC plated on Matrigel in the absence of bFGF,
(bFGF) HUVEC plated on Matrigel in the presence of bFGF (30 ng/mL) for 16 h, (Vehicle) HUVEC treated with vehicle alone (containing up
to 1% DMSO dissolved in cell culture medium), (Compd2) HUVEC treated with compound 2 (1 μM), (Compd3) HUVEC treated with
compound 3 (1 μM), and (Compd8) HUVEC treated with compound 8 (10 μM). (B) The ability to form capillary-like sproutings on Matrigel
was significantly diminished when HUVEC were treated with compounds 2 and 3. Data are presented as means ( SEM of the number of tubes
per square millimeter. These experiments were conducted in quadruplicate and repeated three times. One asterisk indicates P < 0.01, and two
asterisks indicate P < 0.05 vs control (Newman-Keuls test after ANOVA).
substrate-coated plates (i.e., collagen I for SK-MEL-24 and
HUVEC expressing Rvβ3 integrin and vitronectin for K562
expressing R5β1 integrin) (data not shown). Results are
summarized in Table 1. The compound c[Arg-Gly-Asp-
(R)-Phe-Val] was included as reference cyclic peptide, being
a potent inhibitor of cell adhesion,23,56,57 and the peptide
c[Arg-Ala-Asp-(R)-Phe-Val] was used as a negative con-
trol.58
(employed in SK-MEL-24) and vitronectin (employed in
HUVEC).59,60
In agreement with previous studies,23,56,57,61 c[Arg-Gly-
Asp-(R)-Phe-Val] displayed a micromolar inhibitory activity
on the cell adhesive capacity driven by Rvβ3 and R5β1
integrin, whereas the cyclic peptide c[Arg-Ala-Asp-(R)-
Phe-Val] had IC50 values of >100 μM. Interestingly, com-
pound 3 was more potent than reference compound c[Arg-
Gly-Asp-(R)-Phe-Val] as it inhibited cell adhesion mediated
by Rvβ3 and R5β1 integrin possessing IC50 values 40- and
1445-fold less, respectively.
Compound 3, c[(R)-βPheψ(NHCO)Aspψ(NHCO)Gly-
Arg], containing (S)-Arg and (S)-Asp, exhibited the highest
potency as an inhibitor of cell adhesion mediated by Rvβ3 and
R5β1 integrins. The IC50 value versus R5β1 integrin-mediated
cell adhesion was 10 times lower than that versus Rvβ3
integrin (Table 1). The diastereoisomer 2, c[(S)-βPheψ-
(NHCO)Aspψ(NHCO)Gly-Arg], which differs from 3 ex-
clusively for the stereochemistry of the diamine moiety,
maintained a notable efficacy as R5β1 integrin inhibitor
and gained selectivity versus Rvβ3 integrin. Indeed, although
2 exhibited an IC50 of 0.52 μM as an R5β1 integrin antagonist,
the activity toward Rvβ3 integrin dropped to an IC50 value of
11 μM against fibronectin and of 25.2 μM against vitronec-
tin. PMRI RGD mimetic 4 containing (S)-Asp and (R)-Arg,
and mimetic 8 containing (R)-Asp and (S)-Arg, gave com-
paratively inferior results with respect to 2 and 3. Both
compounds exhibited modest IC50 values, in the micromolar
range (Table 1). The rest of the compounds poorly affected
adhesion of cells to immobilized fibronectin or vitronectin,
with an IC50 of >100 μM.
Effect of Integrin Antagonists on in Vitro Angiogenesis
Elicited by Basic Fibroblast Growth Factor (bFGF). The
formation of capillary-like tube structures by HUVEC in the
extracellular matrix (ECM) is the pivotal step in angiogenesis
and is also involved in cell migration and invasion.62-64 To
evaluate any potential antiangiogenic activity of these novel
Rvβ3/R5β1 integrin antagonists, in vitro angiogenesis assays
were conducted by evaluating bFGF-induced angiogenesis of
HUVEC cultured in a 3D gel consisting of Matrigel. As shown
in Figure 2, when HUVEC were plated on wells coated with
Matrigel without the addition of the growth factor, they
showed only a few spontaneous tube formations (taken as an
index of neo-angiogenesis), and most of them were still in a
highly proliferating state with a cobblestone shape. On the
other hand, when HUVEC were plated on Matrigel with
addition of bFGF (30 ng/mL), cells formed a capillary-like
network within 16 h (Figure 2). In the presence of compounds
2 and 3 (1 μM), the extent of tube formation induced by bFGF
was significantly reduced (Figure 2) in comparison to that in
cells treated with the vehicle alone (containing up to 1%
Compounds 2-4 and 8 possess IC50 values comparable to
those of Rvβ3 integrin inhibitors in the two cell models
expressing this integrin toward the two ligands fibronectin

110 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1
Gentilucci et al.
Table 2. Δδ/Δt Values (ppb/K) of Amide Protons for 2 and 3, Deter-
mined by VT H NMR Analysis in an 8:2 DMSO-d6/H2O Mixture at
1
400 MHz over the Range of 298-348 K [NHa and NHb (see Figures 1
and 3-6)]
compd
AspNH
ArgNH
NHa
NHb
2
3
-6.1
-6.3
-5.8
-5.0
1.4
-4.0
-0.6
-1.0
DMSO dissolved in cell culture medium). Interestingly, 3
exhibited the greatest inhibitory effect in blocking bFGF-
induced angiogenesis. The number of tube branches per square
millimeter was reduced from 10 ( 3 (bFGF-treated HUVEC)
to 5.1 ( 0.7 (3-treated HUVEC), and 6.8 ( 0.9 (2-treated
HUVEC). The minimal concentration of these compounds
yielding a complete inhibition of endothelial morphogenesis
on Matrigel was 1 μM; lower concentrations (100 and 500 nM)
were less effective or (10 nM) ineffective (data not shown),
whereas compound 8 (10 μM) did not cause any significant
reduction in the level of angiogenesis (Figure 2). Similarly,
compounds 4-7 and 9, added to the cells to a final concentra-
tion of up to 10 μM, were not effective as angiogenesis
inhibitors (see the Supporting Information).
Figure 3. Representative structure of 2 consistent with ROESY
analysis.
r5β1/rvβ3 Integrin Antagonists Do Not Affect Endothelial
Cell Viability. To determine whether R5β1 and Rvβ3 integrin
antagonists could alter endothelial cell viability, morphology
and flow cytometric analysis were evaluated in HUVEC after
treatment for 16 h with 1 and 10 μM RGD cyclopeptides.
These data indicate that neither significant changes in the
morphology (data not shown) nor significant increases in the
level of apoptosis or necrosis (Supporting Information) were
observed in endothelial cells, suggesting that these com-
pounds do not display any relevant cytotoxicity in HUVEC.
Conformational Analysis of 2 and 3 in Solution. The
biological activities of some of the compounds vary radically
(Table 1), although the sequence of the peptides is identical.
This observation confirms the anticipation that RGD-
mimetic compounds based on the PMRI CTPs behave as
topologically distinct structures. It is accepted that the
conformation of a cyclic peptide is controlled by the stere-
ochemistry array of the residues, while the precise nature of
the residues has a minor impact.21,65,66 In this section, we
discuss the in-solution structural features of 2 and 3, aiming
to deduce useful clues about the biologically active structures
of this class of pseudopeptides, while the conformations of
the less active or inactive compounds 4 and 5 are discussed in
the Supporting Information. The conformations of 6-9, the
mirror images of 2-5, respectively, were not re-examined.
The conformational analyses were performed by NMR
spectroscopy and molecular dynamics (MD) simulations.
The NMR analyses of cyclopeptides 2 and 3 were conducted
using standard techniques in the 8:2 DMSO-d6/H2O biomi-
metic medium.67 We could not perform experiments in water,
Figure 4. Representative, minimized structures of 2a (top) and 2b
˚
˚
˚
(bottom) calculated by unrestrained MD in a 30 A ꢀ 30 A ꢀ 30 A
box of standard TIP3P water molecules, characterized by explicit
H-bonds and secondary structural elements.
1
since the peptides were very poorly soluble. The H NMR
analyses of 2 and 3 revealed a single set of resonances, suggest-
ing conformational homogeneity or a fast equilibrium.21,65
VT 1H NMR experiments were utilized to deduce the
presence of H-bonds (Table 2). The analyses indicated that in
the two cyclopeptides, AspNH and ArgNH do not seem to be
involved in intramolecular H-bonds. On the other hand, for
both compounds, diamineNHb [NHa, NHb, and Hc (see
Figures 1 and 3-6)] very likely participates in a strong H-bond
(|Δδ/Δt| e 1 ppb/K).65,68 With regard to diamineNHa, while
in 2 the low Δδ/Δt value is suggestive of the involvement in
H-bonds (Δδ/Δt = 1.4 ppb/K), the comparatively higher
Figure 5. Representative structure of 3a consistent with ROESY
analysis.
|Δδ/Δt| value observed in 3 accounts for solvent-exposed pro-
tons (Δδ/Δt=-4.0 ppb/K).
2D ROESY data were utilized to investigate the spatial
disposition of molecular backbones (Tables 3 and 4). For the
absence of HRi-HRiþ1 cross-peaks, indicative of a cis pep-
tide bond conformation, all of the ω bonds were set to 180°.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1 111
These structures were clustered by the rmsd analysis of the
backbone atoms. For both peptides, one major cluster
comprising more than 90% of the structures was obtained;
the lowest-energy structures of the major clusters are shown
in Figures 3 and 5.
The ROESY-derived structure of 2 (Figure 3) does not
show explicit H-bonds, which were predicted on the basis of
the VT NMR analysis. Apparently, the conformations de-
termined by ROESY analyses represent the average on the
NMR time scale of different geometries in equilibrium.21,65
To estimate the residual flexibility of the cyclopeptide, the
dynamic behavior of the structure derived from ROESY was
analyzed by unrestrained MD for 10 ns in a box of explicit
water. During the simulation, the main structural features
described on the basis of ROESY were maintained. In
addition, the examination of the trajectories revealed the
occurrence of explicit H-bonds in agreement with VT NMR
analysis.
Figure 6. Representative, minimized structure of 3b calculated by
˚
˚
˚
unrestrained MD in a 30 A ꢀ 30 A ꢀ 30 A box of standard TIP3P
water molecules, characterized by secondary structural elements.
Table 3. Nonobvious ROESY Cross-Peaks Observed for 2a
cross-peak
intensity
cross-peak
intensity
Indeed, the trajectories of 2 clearly revealed the occurrence
of two slightly different backbone structures. One is char-
acterized by a type I β-turn centered on Arg-diamine and
stabilized by a H-bond between malonylCO and diamine-
NHa (Figure 4, top). The second one is characterized by a
type I β-turn centered on Asp-diamine, with a H-bond
between the other malonylCO and diamineNHb (Figure 4,
bottom).
AspNH-NHa
AspNH-COCH2COu
ArgNH-NHb
ArgNH-COCH2COd
ArH-NHa
ArH-diamHβ
NHb-ArgHR
NHa-AspHR
NHa-diamHβ
AspHR-AspHβu
ArgHR-ArgHγ
diamHR-Hc
ArgHδ-ArgHβ
ArgHβ-ArgHγ
s
AspNH-AspHR
AspNH-AspHβd
ArgNH-ArgHR
ArH-NHb
ArH-diamHR
NHb-NHa
m
m
m
m
m
m
m
s
vs
s
vs
s
vs
m
m
s
NHb-Hc
NHa-diamHR
AspHR-AspHβd
ArgHR-ArgHβ
ArgHR-ArgHδ
diamHR-diamHβ
ArgHδ-ArgHγ
m
s
As for compound 3, the ROESY-derived conformation
shows a distinct H-bond between NHb and malonylCO, as
anticipated on the basis of VT NMR, and an overall type II
β-turn conformation on Asp-diamine (Figure 5). Unrest-
rained MD simulation was performed as reported for 2.
In addition to the conformation of 3a shown in Figure 5,
the analysis of the trajectories of 3 gave a second conformer,
3b, whose structure is in agreement with a pseudoinverse
γ-turn centered on Asp, and an inverse γ-turn on Arg.71 This
alternative secondary structure (Figure 6) is still compatible
with the involvement of the NHb proton in a H-bond as
predicted by VT NMR analysis. Despite the different sec-
ondary structures, the two conformers, 3a and 3b, still
maintain a very similar display of the pharmacophores.
However, the two situations slightly differ for the distance
w
w
s
m
s
w
s
s
a Stereochemistry has beenomitted. For Ha, Hb, and Hc, see Figure1.
Abbreviations: u, upfield; d, downfield; vs, very strong; s, strong; m,
medium; w, weak.
Table 4. Nonobvious ROESY Cross-Peaks Observed for 3a
cross-peak
intensity
cross-peak
intensity
AspNH-NHa
AspNH-COCH2COu
ArgNH-NHb
ArgNH-COCH2COd
ArH-NHa
ArH-diamHβd
NHb-ArgHR
NHb-Hcu
w
vs
m
vs
m
s
AspNH-AspHR
ArgNH-ArgHR
ArH-diamHR
ArH-diamHβu
NHb-NHa
m
m
s
˚
vs
m
m
m
m
vs
m
m
s
between the Cβ atoms of Asp and Arg: ∼8.3 A in 3a and
˚
∼8.8 A in 3b.
NHb-diamHR
NHb-Hcd
The structures of 2 and 3 share a certain similarity in the
RGD-mimetic region. The main difference among the two
structures resides in the different orientations of the diamine
aromatic side chain and of NHa. In 2, the aromatic side chain
and NHa point above the molecular plane; in 3, the situation
is reversed. In particular, in 2, the diamine phenyl side chain
adopts a pseudoaxial disposition; in 3, it adopts a pseudo-
equatorial disposition.
s
vs
vs
m
s
NHa-diamHR
AspHR-AspHβu
ArgHR-ArgHβu
ArgHR-ArgHδ
diamHR-Hcu
NHa-AspHR
NHa-diamHβ
ArgHR-ArgHγ
diamHR-Hcd
diamHR-diamHβd
ArgHβ-ArgHγ
m
s
diamHR-diamHβu
ArgHδ-ArgHγ
s
vs
s
a Stereochemistry has beenomitted. For Ha, Hb, and Hc, see Figure1.
Abbreviations: u, upfield; d, downfield; vs, very strong; s, strong; m,
medium; w, weak.
Molecular Docking. To interpret on a molecular basis the
different affinities of compounds 2 and 3 for the RVβ3
receptor, docking studies were performed using Glide
(version 4.5)72 by starting from the representative macro-
cycle conformations sampled during the unrestrained MD
simulations. Molecular docking of compounds 4-9 is dis-
cussed in the Supporting Information. The protein binding
site was derived from the X-ray crystal structure of the
extracellular segment of integrin RVβ3 in complex with the
cyclic pentapeptide ligand cilengitide [Protein Data Bank
(PDB) entry 1L5G].29 In this X-ray structure, the potent
RVβ3 antagonist cilengitide, bound to the headgroup of the
integrin, features an extended conformation of the RGD
Conformations consistent with the spectroscopic analysis
were obtained by restrained MD in a box of explicit,
equilibrated water molecules,69 using the distances deduced
from ROESY as constraints and minimized with the AM-
BER70 force field. A set of 50 random structures generated
by means of a unrestrained high-temperature MD simula-
tion was subjected to restrained MD with a scaled force field,
followed by a high-temperature simulation with full
restraints, after which the system was gradually cooled.

112 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1
Gentilucci et al.
Figure 7. Top-ranked docking poses of (a) ligand 3 (atom color tube representation, macrocycle geometry inverse γ/inverse γ 3b) and (b)
ligand 2 (atom color tube representation, macrocycle geometry β-turn 2b) into the crystal structure of the extracellular domain of RVβ3 integrin
(R unit colored cyan and β unit colored orange in a wire representation) overlaid on the bound conformation of cilengitide (green tube
representation). Only selected integrin residues involved in the interactions with cilengitide are shown. The Mn2þ ion at MIDAS is shown as a
green CPK sphere. Nonpolar hydrogen atoms were removed for the sake of clarity.
˚
sequence with a distance of ∼9 A between the Cβ atoms of
Asp and Arg. The crystal complex interaction pattern in-
volves the formation of an electrostatic clamp between the
guanidinium group of the ligand and the negatively charged
side chains of Asp218 and Asp150 in the R unit and between
the carboxylic group of cilengitide and the metal cation in the
metal ion-dependent adhesion site (MIDAS) region of the
β unit. Moreover, further stabilization occurs through hy-
drogen bonds between the NH group of the ligand Asp
residue and the carbonyl oxygen atom of Arg216 in the
β subunit as well as between the ligand carboxylate oxygen
not coordinated to MIDAS and the backbone amides of
Asn215 and Tyr122 in the β unit (Figure 7).
The experimentally observed binding mode of cilengitide
with RVβ3 integrin was taken as a reference model for the
interpretation of the docking results in terms of the ligand-
Figure 8. Top-ranked docking poses of mimics 3b (green tube
representation) and 2b (purple tube representation) in the crystal
bound conformation and ligand-protein interactions.
The models built for the interaction of compounds 2 and 3
with RVβ3 integrin through docking studies showed that
suitable macrocycle conformations of these ligands enable
them to fit properly in the shallow cleft of the receptor,
sharing the binding features of the crystal structure of the
cilengitide-RVβ3 complex, especially those governing the
recognition process. Among the conformations selected
from the MD simulations of 2 and 3, only the macrocycle
geometries “inverse γ/inverse γ” 3b and “β-turn” 2b generate
top-ranked docking poses conserving the relevant ligand-
protein interactions observed in the crystal structure of the
cilengitide-RVβ3 complex. However, the inverse γ/inverse
γ geometry of 3b reproduces the RGD backbone of the X-ray
ligand better than the β-turn geometry of 2b (Figure 7).
Moreover, the pseudoequatorial benzyl group in 3b points to
the same direction of the Phe side chain of the reference
ligand, that is, toward the outside of the integrin binding site,
allowing the aromatic ring to fit unhindered and to form a
T-shaped interaction with the Tyr122 side chain of the
β subunit (Figures 7 and 8).
structure of the extracellular domain of Rvβ3 integrin represented as
a molecular surface (R unit colored cyan, β unit colored orange).
Mn2þ ions are represented as yellow CPK spheres.
the T-shaped interaction with the aromatic ring of the Tyr122
side chain. Accordingly, the Glide score values suggest a
slightly better binding for epimer 3.
Discussion
In this study, a selected minilibrary of stereoisomeric, modi-
fied retro-inverso cyclotetrapeptides 2-9 having the general
structure c[(S/R)-βPheψ(NHCO)(S/R)-Aspψ(NHCO)Gly-
(S/R)-Arg] inhibited adhesion of cells to vitronectin- or fibro-
nectin-coated surfaces and inhibited angiogenesis. The differ-
ent compounds displayed variable activities as Rvβ3 or R5β1
integrin receptor antagonists, as determined by testing the
inhibition of adhesion of fibronectin and vitronectin to Rvβ3
or R5β1 integrin receptor-expressing cell lines. Interest-
ingly, c[(S)-βPheψ(NHCO)Aspψ(NHCO)Gly-Arg] (2) dis-
played a submicromolar activity for R5β1 integrin, and a cer-
tain selectivity over Rvβ3 integrin, while c[(R)-βPheψ-
(NHCO)Aspψ(NHCO)Gly-Arg] (3) was demonstrated to be
a potent dual antagonist, with 10-8 and 10-7 activities for R5β1
and Rvβ3 integrins, respectively. On the other hand,
c[(S)-βPheψ(NHCO)(S)-Aspψ(NHCO)Gly-(R)-Arg] (4) and
The β-turn structure of 2b properly drives the pharmaco-
phoric groups within the binding site to form the key
electrostatic interactions with the receptor. Nevertheless,
the pseudoaxial benzyl group forces the entire molecule to
enter the receptor lopsided. The benzyl group fails in forming

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1 113
c[(R)-βPheψ(NHCO)(R)-Aspψ(NHCO)Gly-(S)-Arg] (8) re-
vealed moderate micromolar inhibitory activities, while the
remaining compounds, 5-7 and 9, were ineffective. These
results demonstrate that compound 2 is more specific as an
R5β1 integrin antagonist whereas compounds 3, 4, and 8
nonselectively antagonized Rvβ3 and R5β1 integrins, albeit with
different efficacies, 3 being the most potent.
Moreover, we showed that R5β1 and Rvβ3 integrin anta-
gonists inhibited bFGF-induced human endothelial cell tube
formation at submicromolar concentrations as examined
using the Matrigel assay. Among them, compound 3 showed
themost potentantiadhesionand antiangiogenic effects;itcan
directly interact with Rvβ3 integrin expressed on melanoma
and endothelial cells and can prevent adhesion of endothelial
cells to vitronectin displaying higher binding selectivity for
Rvβ3 integrin in cell culture.
Blockade of Rvβ3 integrin-mediated functions by antibodies
or RGD peptides disrupts blood vessel formation in various
animal models; Rvβ3 antagonists may perturb the growth and/
or maturation of blood vessels without detectable alteration
on the preexisting blood vessels.73
Angiogenesis is also directly regulated by binding of fibro-
nectin to its receptor, the R5β1 integrin. Specific antibodies,
peptides, and novel nonpeptide antagonists of R5β1 integrin
can block angiogenesis induced by several growth factors in
both chick embryo and murine models. In fact, these R5β1
antagonists inhibited tumor angiogenesis, thereby causing
regression of human tumors in animal models.74 Antagonists
of Rvβ3 and R5β1 integrins substantially prevent angiogenesis
induced by bFGF, suggesting that these integrins regulate
similar pathways of angiogenesis.
The results of this study suggest that the partially retro-
inverso RGD motif in the cyclotetrapeptide retains the Rvβ3
and R5β1 integrin antagonist activity. Therefore, we propose
that the potent antiangiogenic effect of 3 in cell adhesion and
tube formation assays is probably due to the interference of
the Rvβ3 and R5β1 integrin-mediated interactions between
endothelial cells and ECM protein. Indeed, 3 significantly
decreased the level of bFGF-induced angiogenesis in an in
vitro model. Therefore, the mechanism of action of these
compounds in suppressing bFGF-elicited angiogenesis could
be mediated by the specific blockade of Rvβ3 ligation. Finally,
these integrin antagonists did not cause any cytotoxic effect as
evidenced by flow cytometric analysis.
The sequence of peptides 2-9 is identical, the only differ-
ence being the stereochemistry array. The comparatively
higher inhibitory efficacy toward R5β1 integrin observed for
compounds 2 and 3, the most potent of the minilibrary, can
be correlated to the specific (S,S) stereochemistry of the
retro-inverso RGD sequence, ψ(NHCO)(S)-Asp-ψ(NHCO)-
Gly-(S)-Arg. Indeed, the remaining compounds possess (S,
R), (R,R), or (R,S) (Arg,Asp) configurations.
It is worth mentioning that, due to the partial retro-inverso
nature of the RGD sequence, the (S,S) stereochemistry of Arg
and Asp in 2 and 3 leads to overall conformations showing the
Arg and Asp pharmacophoric side chains on the opposite
sides of the molecular scaffold, an unusual biologically active
conformation in comparison to other well-known RGD
cyclopentapeptide integrin antagonists, such as c[Arg-
Gly-Asp-(R)-Phe-Val] or c[Arg-Gly-Asp-(R)-Phe-NMeVal]
(cilengitide), having the pharmacophoric side chains placed
on the same side. Apparently, in the smaller cyclotetrapeptide
structures, an opposite orientation allows for a correct dis-
tance between the pharmacophores. For both 2 and 3, during
Figure 9. Overlap of the backbone structures of 2 (2b, white) and 3
(3b, black) determined by ROESY and MD analysis. Side chains
have been positioned in the trans extended conformation.
the unrestrained MD simulations, the average distance bet-
˚
ween Asp and Arg Cβ was ∼8.5 A, and that between the
˚
carboxylate and guanidino functions was ∼13 A, very similar
to that of the R5β1 or Rvβ3 selective cyclopentapeptides.29,30,38
As revealed by conformational analysis (Results), the two
compounds show a very similar display of the retro-inverso
RGD sequence. On the other hand, the different R5β1 versus
Rvβ3 selectivity displayed by 2 and 3, sharing the same (S,S)
stereochemistry of the RGD-mimetic sequence, can be attri-
buted to the presence of an (S)-diamine in 2 and an (R)-
diaminein 3. The mostrelevant 3Ddifference canbe identified
in the disposition of the diamine portion. Apart from the
different location of diamine NHa and NHb, in 2 the diamine
phenyl side chain adopts an axial position while in 3 it is
equatorial (Figure 9).
The experimental 3D structure of R5β1 integrin is not yet
available. As a consequence, the comparison of the structures
of 2 and 3 with the structures of a few ligands, and with a in
silico model, previously reported in the literature, could
furnish some useful information about the biologically active
conformation at this receptor. Goodman et al. described
the receptor-bound conformation of the RGD sequence
in the partially 15N-labeled R5β1 integrin weak antagonist
c[Mpa15N-Arg-15N-Gly-15N-Asp-15N-Asp-15N-Val-Cys]-NH2
(IC50 = 1.2 μM), determined by 15N-edited 2D transferred
nuclear Overhauser effects.31 Data were indicative of a tilted
conformation, leading to a very short average distance bet-
˚
ween Arg and Asp Cβ of ∼5.6 A. More recently, Sewald et al.
reported for the R5β1 integrin cyclohexapeptide ligand c[Arg-
Gly-Asp-D-Phe-Val-β-Ala] (IC50=2 ( 1 μM in K562 adhe-
sion assays) a preferred extended RGD conformation show-
ing a quite longer average distance between Arg Cβ and Asp
Cβ of ∼9.3 A.32 Apparently, the average Cβ-Cβ distance of
˚
˚
8.5 A observed for 2 and 3 is somewhatbetween the two values
reported above.
More recently, a homology model of the R5β1 integrin has
been reported by Kessler, Novellino, and co-workers.37 The
model, obtained by comparative protein modeling and vali-
dated with experimental data of nonpeptide ligands, allowed
identification of a distance between the carboxylate function
˚
and the basic moiety of ∼13 A. This value is very close to the
˚
distance of ∼13.8 A between the carboxylate and basic groups
determined for Rvβ3 integrin-bound ligands.20,29,30 As a con-
sequence, the distance between the basic and acid moieties
does not constitute a discriminant for Rvβ3 versus R5β1
selectivity. Therefore, the different pharmacological profile

114 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1
Gentilucci et al.
Figure 10. Schematic representation of the proposed disposition of 3 (left, black) and 2 (right, white) within R5β1 (black sticks) and Rvβ3 (gray
sticks) integrin binding site sketches. Unfavorable interactions are also shown (transparent gray flashes).
of 2 and 3 toward Rvβ3 and R5β1 integrins cannot be explained
on the basis of the Cβ-Cβ distance alone.
Conclusions
In summary, we have reported the synthesis of a selected
small library of RGD mimetics as Rvβ3 and/or R5β1 integrin
antagonists. The compounds have been designed starting
from 13-membered CTP scaffolds based on a partially retro-
inverso structure, containing (S)- or (R)-Asp, (S)- or (R)-Arg,
and an (S)- or (R)-diamine as a β-Phe surrogate.
However, the comparison of R5β1 and Rvβ3 binding pockets
revealed the occurrence of mutated residues in the β sub-
unit. Indeed, (β3)-Arg214 is replaced with (β1)-Gly217, and
(β3)-Arg216 is mutated to (β1)-Leu219. The substitution of both
the bulky Arg residues expands the site of the R5β1 binding
pocket in comparison to the Rvβ3 integrin. The authors
concluded that ligands carrying bulky aromatic substituents
in the proximity of the Asp (or Asp mimetic) residue can still
fit the larger R5β1 binding pocket, but not the cramped Rvβ3
binding pocket, therefore showing a certain selectivity for
R5β1 over Rvβ3 integrins.
The model highlights the fundamental role of the aromatic
residue flanking Asp in determining selectivity. This rationa-
lization is corroborated by the molecular docking computa-
tions performed by us on compounds 2 and 3 positioned
within the Rvβ3 integrin receptor binding site (see Results).
The results lead to conjecture that the different disposition of
diamine side chain is responsible of the different activity of 2
and 3 toward Rvβ3 integrin.
Molecular modeling studies showed that the pseudoequa-
torial benzyl group of 3 points toward the outside of the
integrin binding site, allowing the aromatic ring to fit unhin-
dered and reproducing the orientation of cilengitide (Results),
matching side by side the Tyr122 side chain. Conversely, the
axial disposition of the benzyl group of 2 gives rise to less
favorable interactions, forcing the entire molecule to go into
the receptor lopsided.
With regard to the R5β1 integrin, on the basis of the
homology model discussed above, the R5β1 integrin seems to
be more tolerant in hosting the aromatic side chain of both 2
and 3. In particular, its larger binding site would more easily
accommodate the phenyl ring of 2 with respect to the Rvβ3
integrin, leading to a pronounced R5β1 selectivity, while the
disposition of 3 would be adequate for both integrins, leading
to a generally good affinity (Figure 10).
The different compounds exhibited variable activities as
Rvβ3 or R5β1 integrin receptor antagonists, as determined by
testing the inhibition of adhesion of fibronectin or vitronectin
to Rvβ3 or R5β1 integrin receptor-expressing cell lines. c[(S)-
βPheψ(NHCO)Aspψ(NHCO)Gly-(R)-Arg] (4) and c[(R)-
βPheψ(NHCO)(R)-Aspψ(NHCO)Gly-Arg] (8) revealed
moderate micromolar inhibitory activities; c[(S)-βPheψ-
(NHCO)Aspψ(NHCO)Gly-Arg] (2) displayed a submicro-
molar activity for R5β1 integrins, and a certain selectivity over
Rvβ3 integrins, while c[(R)-βPheψ(NHCO)Aspψ(NHCO)-
Gly-Arg] (3) was demonstrated to be a potent dual antagonist,
with activities of 10-8 and 10-7 M for R5β1 and Rvβ3 integrins,
respectively, and significantly inhibited bFGF-induced angio-
genesis of HUVEC cultured in vitro. The different R5β1 versus
Rvβ3 selectivity of 2 and 3 has been explained on the basis of
their alternative secondary structures, discussed on the basis
of NMR, ROESY, MD, and molecular docking analyses. In
particular, the 3D display of the phenyl substituent of the
diamine, pseudoaxial in 2 and pseudoequatorial in 3, seems to
be a major contributor to receptor selectivity.
These Rvβ3 and R5β1 antagonists may be potentially effec-
tiveand safe for therapeutic purposes. Small RGD-containing
peptidomimetics may be used as lead compounds for devel-
oping the potential therapeutic agents for angiogenesis-
related diseases, including cancer.
Experimental Section
General Methods. Unless stated otherwise, standard chemi-
cals were obtained from commercial sources and used without
further purification. Flash chromatography was performed on
silica gel (230-400 mesh), using mixtures of distilled EtOAc and
MeOH. Semipreparative RP-HPLC was performed on a C18
column (7 μm particle size, 21.2 mm ꢀ 150 mm, from a 7:3 H2O/
CH3CN mixture to 100% CH3CN in 15 min) at a flow rate of
12 mL/min. The purity (g95%) of tested compounds was
assessed by analytical RP-HPLC, on an ODS column (4.6 μm
Despite the moderate or absent activity as adhesion inhibi-
tors, compounds 4-9 can be utilized to confirm the validity of
the model proposed above for the more active compounds, 2
and 3. Indeed, the structures of 4-9 obtained by conforma-
tional analysis (see the Supporting Information) in general
correlate well with the activity profile of the compounds
toward R5β1 and Rvβ3 integrins. Finally, the different affinities
of 4-9 for the RVβ3 receptor can be explained on the basis of
docking analyses performed by Glide (see the Supporting
Information).
˚
particle size, 100 A pore diameter, 250 mm, DAD 210 nm, from
a 9:1 H2O/CH3CN mixture to a 2:8 H2O/CH3CN mixture in
20 min) at a flow rate of 1.0 mL/min, followed by 10 min at the
same composition. 1H NMR spectra were recorded using 5 mm
tubes, using 0.01 M peptide at 400 MHz at room temperature.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1 115
Solvent suppression was performed by the solvent presaturation
procedure implemented in varian (presat). Chemical shifts are
reported as δ values relative to the solvent peak. The unambig-
uous assignment of 1H NMR resonances was performed by
gCOSY, HMBC, and HSQC. gCOSY experiments were con-
ducted with a proton spectral width of 3103 Hz. VT 1H NMR
experiments were performed over the range of 298-348 K.
Peaks were calibrated on DMSO. Conformational rearrange-
ment was excluded since signal broadening was absent. 2D
spectra were recorded in the phase sensitive mode and processed
using a 90°-shifted, squared sine-bell apodization. ROESY
experiments were recorded with a 300 or 350 ms mixing time
with a proton spectral width of 3088 Hz.
Representative Synthetic Procedures and Analytical Charac-
terization of PMRI RGD Mimetics 2 and 3. 11. HOBt (0.16 g,
1.2 mmol) was added to a stirred solution of Fmoc-Asp(Mtr)-
OH (0.61 g, 1.0 mmol) in a 9:1 DCM/DMF mixture (15 mL) at
rt. After 10 min, 10 (0.28 g, 1.0 mmol), EDCI-HCl salt (0.24 g,
1.2 mmol), and TEA (0.40 mL, 3.0 mmol) were added while the
mixture was stirred at rt. After 4 h, the mixture was diluted with
DCM, and the solution was washed with 0.5 M HCl and
saturated Na2CO3. The organic layer was dried over Na2SO4,
and the solvent was removed under reduced pressure. The Fmoc
dipeptide 11 was isolated by crystallization from a DCM/Et2O
mixture [0.74 g, 85%, 88% pure by RP-HPLC (see General
Methods); Rf=11.7 min]: ES-MS m/z 875.3 (M þ 1), calcd 875.4;
1H NMR (200 MHz, CDCl3) δ 1.40-1.60 (m, 2H), 1.60-1.80
(m, 2H), 2.10 (s, 3H), 2.60 (s, 3H), 2.68 (s, 3H), 2.70-2.78 (m,
3H), 3.00-3.15 (m, 2H), 3.15-3.22 (m, 2H), 3.78 (s, 3H), 4.05
(m, 1H), 4.20 (m, 1H), 4.30-4.50 (m, 2H), 4.99 (s, 2H), 5.45 (br
d, 1H), 5.50 (br d, 1H), 5.70 (br s, 1H), 6.10 (br s, 2H), 6.50 (s,
1H), 7.05-7.42 (m, 15H), 7.55 (br d, 2H), 7.75 (br d, 2H).
Fmoc group deprotection was performed by treatment with
2 M DMA in THF (6 mL) at rt. After 30 min, the solution was
evaporated at reduced pressure, and the treatment was repeated.
After final evaporation of the solution, the residue was tritu-
rated twice in n-pentane. The deprotected dipeptide [0.50 g,
90%, 86% pure by RP-HPLC (see General Methods); Rf=8.8
min] was used without further purification: ES-MS m/z 653.3
(M þ 1), calcd 653.3.
7.18-7.42 (m, 15H), 7.78 (br t, 1H), 7.97 (br d, 1H), 8.10 (br d,
1H); 13C NMR (300 MHz, 9:1 CDCl3/DMSO-d6) δ 14.5, 17.1,
23.6, 27.3, 29.0, 29.2, 35.5, 37.9, 38.5, 39.8, 48.7, 49.2, 53.3, 55.0,
56.1, 69.6, 73.7, 111.0, 121.3, 125.8, 127.3, 127.3, 127.5, 128.3,
128.3, 128.4, 128.4, 128.6, 128.6, 128.7, 128.7, 130.1, 130.1,
130.6, 132.3, 132.9, 133.8, 136.5, 138.8, 150.2, 157.2, 162.6,
163, 165.7, 167.6, 170.9, 170.9, 172.0, 174.7.
14 {c[βPheψ(NHCO)Asp(Ot-Bu)ψ(NHCO)Gly-Arg(Mtr)]}.
Removal of the protecting group from 13 (0.32 g, 0.32 mmol)
was performed by treatment with H2 and catalytic Pd/C in
EtOH (15 mL) at rt. After 6 h, the mixture was filtered over
Celite, and the solvent was evaporated at reduced pressure,
giving the linear tetrapeptide H-βPheψ(NHCO)Asp(Ot-Bu)ψ-
(NHCO)Gly-Arg(Mtr)-OH [0.24 g, 96%, 84% pure by RP-
HPLC (see General Methods); Rf = 5.9 min], used without
further purification: ES-MS m/z 776.3 (M þ 1), calcd 776.4.
A mixture of the deprotected tetrapeptide (0.24 g, 0.31 mmol),
DPPA (0.15 mL, 0.62 mmol), and NaHCO3 (0.42 g, 5.0 mmol) in
DMF (70 mL) was stirred at rt. After 72 h, the solvent was
distilled under reduced pressure, the residue was diluted with
water, and the mixture was extracted three times with DCM.
The solution was evaporated under reduced pressure, and the
residue was precipitated from a DCM/Et2O mixture. Semipre-
parative RP-HPLC (General Methods) gave 14 [0.16 g, 69%,
96% pure by RP-HPLC (see General Methods); Rf=8.3 min]:
1
ES-MS m/z 758.5 (M þ 1), calcd 758.4; H NMR (400 MHz,
DMSO-d6) δ 1.39 (s, 9H, t-Bu), 1.40-1.60 (m, 3H, ArgHγ þ
ArgHβ), 1.70 (m, 1H, ArgHβ), 2.07 (s, 3H, CH3), 2.29 (dd, J=
10.0, 16.5 Hz, 1H, AspHβ), 2.49 (dd, J = 4.1, 16.5 Hz, 1H,
AspHβ), 2.52 (s, 3H, CH3), 2.62 (s, 3H, CH3), 2.62-2.85 (m, 2H,
diamHβ), 2.95-3.18 (m, 4H, Hc þ ArgHδ), 3.18 (d, J=11.0 Hz,
1H, COCH2CO), 3.26 (d, J=11.0 Hz, 1H, COCH2CO), 3.81 (s,
3H, OCH3), 3.85 (m, 1H, diamHR), 4.05 (m, 1H, ArgHR), 4.25
(m, 1H, AspHR), 6.30-6.50 (m, 2H, ArgNHη), 6.46 (d, J=8.4
Hz, 1H, NHa), 6.70 (s, 1H, 50-ArH), 6.82 (t, J=5.4 Hz, 1H,
NHb), 6.92 (m, 1H, ArgNHε), 7.15-7.35 (m, 5H, ArH), 8.88 (d,
J=6.6 Hz, 1H, ArgNH), 9.04 (d, J=6.6 Hz, 1H, AspNH); 13C
NMR (400 MHz, DMSO-d6) δ 18.5, 19.1, 24.0, 25.9, 29.1, 29.3,
36.0, 37.5, 39.1, 39.8, 48.8, 50.4, 51.8, 55.0, 56.3, 73.2, 112.2,
121.3, 125.8, 128.3, 128.3, 128.6, 128.6, 132.6, 132.9, 136.3,
139.2, 162.9, 165.8, 170.9, 170.9, 172.2, 174.7, 175.0.
12. A solution of Meldrum’s acid (0.85 g, 6 mmol) and
H-Asp(Ot-Bu)-OBz (1.4 g, 5 mmol) in CH3CN (15 mL) was
warmed to 70 °C under an inert atmosphere. After 3 h, a 4:1
cyclohexane/Et2O mixture (40 mL) was added, and the oily
residue that precipitated was separated. This residue was
washed twice with a 4:1 hexane/Et2O mixture (20 mL), and
the resulting dense oil was dissolved in EtOAc (40 mL) and
washed with 0.1 M HCl (5 mL). The organic layer was dried
over Na2SO4, and solvent was evaporated at reduced pressure
(<40 °C), giving 12 as a waxy solid, which was used for the
following step without further purification (1.2 g, 68%, 84%
pure by NMR analysis): 1H NMR (300 MHz, CDCl3) δ 1.31 (s,
9H), 2.72 (dd, J=5.1, 17.2 Hz, 1H), 2.86 (dd, J=4.8, 17.2 Hz,
1H), 3.28 (br s, 2H), 4.82 (m, 1H), 5.05 (d, J=12.1 Hz, 1H), 5.14
(d, J = 12.1 Hz, 1H), 7.25-7.40 (m, 5H), 7.86 (br d, 1H),
9.70-10.4 (br s, 1H).
13. Dipeptides 11 (0.33 g, 0.5 mmol) and 12 (0.18 g, 0.5 mmol)
were coupled under the same conditions used for the synthesis
of 11 with HOBt (0.082 g, 0.6 mmol), EDCI-HCl salt (0.12 g,
0.6 mmol), and TEA (0.20 mL, 1.5 mmol) in a 9:1 DCM/DMF
mixture (10 mL) at rt. After 5 h, the usual workup afforded the
fully protected tetrapeptide 13, isolated by crystallization from a
DCM/Et2O mixture [0.32 g, 65%, 86% pure by RP-HPLC (see
General Methods); Rf=11.2 min]: ES-MS m/z 1001.3 (M þ 1),
calcd 1001.4; 1H NMR (200 MHz, CDCl3) δ 1.48 (s, 9H)
1.54-1.70 (m, 2H), 1.77 (m, 1H), 1.92 (m, 1H), 2.10 (s, 3H),
2.50 (br s, 1H), 2.70 (s, 3H), 2.76 (s, 3H), 2.74-2.85 (m, 3H),
3.18-3.38 (m, 3H), 3.39-3.50 (m, 3H), 3.90 (s, 3H), 4.10 (m,
1H), 4.60 (m, 1H), 4.93 (m, 1H), 5.03-5.25 (m, 4H), 6.01 (d, J=
8.0 Hz, 1H), 6.20-6.40 (br s, 1H), 6.42 (br s, 2H), 6.60 (s, 1H),
2 {c[βPheψ(NHCO)Aspψ(NHCO)Gly-Arg]}. The protected
cyclotetrapeptide 14 (0.16 g, 0.21 mmol) was treated with a
94:2:1:2:1 mixture of TFA and scavengers [TFA/PhOH/PhSH/
H2O/Et2S (5 mL)] at rt for 30 min. The mixture was distilled
under reduced pressure, and the treatment was repeated. The
residue was suspended in Et2O, and the precipitate was centri-
fuged. The resulting crude residue was purified by semiprepara-
tive RP-HPLC (General Methods), giving 2 [0.077 g, 76%, 97%
pure by analytical RP-HPLC (see General Methods); Rf=1.8
1
min]: ES-MS m/z 490.2 (M þ 1), calcd 490.2; H NMR (400
MHz, 8:2 DMSO-d6/H2O) δ 1.45-1.60 (m, 3H, ArgHγ þ
ArgHβ), 1.74 (m, 1H, ArgHβ), 2.30 (dd, J=9.5, 16.0 Hz, 1H,
AspHβ), 2.47 (dd, J=4.0, 16.0 Hz, 1H, AspHβ), 2.60-2.86 (m,
2H, diamHβ), 2.92-3.20 (m, 4H, Hc þ ArgHδ), 3.19 (d, J=10.8
Hz, 1H, COCH2CO), 3.24 (d, J=10.8 Hz, 1H, COCH2CO), 3.68
(m, 1H, diamHR), 3.96 (m, 1H, ArgHR), 4.20 (m, 1H, AspHR),
6.70 (d, J=8.4 Hz, 1H, NHa), 6.90 (br t, 1H, NHb), 7.10-7.35
(m, 5H, ArH), 7.90 (br s, 1H, ArgNHε), 9.10 (br d, 1H, ArgNH),
9.28 (br d, 1H, AspNH); 13C NMR (400 MHz, 8:2 DMSO-d6/
H2O) δ 26.0, 29.2, 36.5, 37.5, 39.0, 39.8, 40.0, 48.8, 50.2, 51.8,
55.0, 126.0, 128.1, 128.3, 128.8, 128.7, 138.8, 163.9, 171.0, 171.2,
174.7, 175.1, 177.7.
(3) {c[(R)-βPheψ(NHCO)Aspψ(NHCO)Gly-Arg]}: 1H NMR
(400 MHz, 8:2 DMSO-d6/H2O) δ 1.42-1.50 (m, 2H, ArgHγ),
1.56 (m, 1H, ArgHβ), 1.74 (m, 1H, ArgHβ), 2.40 (dd, J=6.0,
16.8 Hz, 1H, AspHβ), 2.58 (dd, J=7.2, 16.8 Hz, 1H, AspHβ),
2.75 (dd, J=6.6, 16.0 Hz, 1H, diamHβ), 2.83 (dd, J=8.0, 16.0
Hz, 1H, diamHβ), 2.95-3.15 (m, 3H, Hc þ ArgHδ), 3.20 (d, J=
10.2 Hz, 1H, COCH2CO), 3.21 (d, J=10.2 Hz, 1H, COCH2CO),

116 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1
Gentilucci et al.
3.28 (m, 1H, Hc), 3.67 (m, 1H, diamHR), 3.95 (m, 1H, ArgHR),
4.26 (m, 1H, AspHR), 7.02 (br t, 1H, NHb), 7.10-7.35 (m, 5H,
ArH), 7.62 (d, J=6.9 Hz, 1H, NHa), 7.90 (br s, 1H, ArgNHε),
8.85 (br d, 1H, ArgNH), 9.22 (br d, 1H, AspNH); 13C NMR (400
MHz, 8:2 DMSO-d6/H2O) δ 26.0, 29.0, 34.5, 36.6, 39.3, 39.8,
39.9, 48.8, 50.4, 52.8, 56.6, 124.5, 127.9, 127.9, 129.0, 129.1,
138.8, 163.8, 171.3, 175.0, 175.1, 176.1.
necrotic, currently undergoing apoptosis from cells that have
died of necrosis. HUVEC were treated with different concentra-
tions of the assayed compounds for 30 min at 37 °C; the cells
were detached and fixed with 1% paraformaldheyde for 15 min
and then washed with PBS. Thereafter, HUVEC were resus-
pended in Annexin-V-Fluos incubation buffer [10 mM Hepes/
NaOH (pH 7.4), 140 mM NaCl, and 5 mM CaCl2] with
prediluted Annexin-V-Fluos and PI, according to the manufac-
turers’ instructions. After incubation for 15 min at room tem-
perature in the dark, the cells were analyzed with EPICS
XLMCL (Beckman Coulter).
(v) In Vitro Tubular Formation of HUVEC. The endothelial
tube formation assay was performed as described.78 Matrigel-
precoated 96-well plates were used, and HUVEC (5000 cells/
well) were seeded in the presence of different concentrations of
test compounds, in the absence (negative control) or presence of
bFGF (30 ng/mL). Cells were incubated for 16 h at 37 °C. After
incubation, these cells underwent differentiation into capillary-
like tube structures; tube formation was examined by a NIKON
microscope equipped with a camera. The experiments were
repeated three times.
Pharmacological Assays. (i) Materials for Bioassays. Tryp-
sin/EDTA, nonessential amino acids, minimum essential med-
ium (MEM), RPMI-1640 with L-glutamine, antibiotic and
antimycotic solution, and glycine were purchased from Invitro-
gen (Carlsbad, CA). Fetal bovine serum (FBS) and phosphate-
buffered saline (PBS) were from Cambrex (Walkersville, MD).
Citrate buffer solution, EDTA, DMSO, Triton X-100, 4-nitro-
phenyl N-acetyl-β-D-glucosaminide, phorbol 12-myristate 13-acet-
ate (PMA), pyruvic acid, fibronectin, and vitronectin, both from
human plasma, were obtained from Sigma-Aldrich SRL (Milan,
Italy). SK-MEL-24 (human malignant melanoma) and K-562
(human erythroleukemia) were obtained from American Tissue
Culture Collection (ATCC, Rockville, MD). HUVEC (human
umbilical vein endothelial cells) were obtained from Clonetics
(Cambrex). Matrigel-precoated 96-well plates and human re-
combinant bFGF were from BD Biosciences (Bedford, MA).
(ii) Cell Culture. SK-MEL-24 were routinely grown in MEM
supplemented with 10% FBS, nonessential amino acids, and
sodium pyruvate. K-562 were maintained as a stationary sus-
pension culture in RPMI-1640 and L-glutamine with 10% FBS.
HUVEC were cultured in endothelial growth medium (EGM,
Clonetics), containing fetal bovine serum, bovine brain extract,
human epidermal growth factor, hydrocortisone, gentamicin,
and amphotericin B; cells from passages 2-7 were used in this
study. Cells were kept at 37 °C in a 5% CO2 humidified atmo-
sphere. Forty hours before the experiment, K-562 were treated
with 25 ng/mL PMA to induce differentiation with an increased
level of expression of cell surface antigens.75
Conformational Analysis. ROESY intensities were classified
very strong, strong, medium, and weak and were associated with
˚
distances of 2.3, 2.6, 3.0, and 4.0 A, respectively. Geminal
couplings and other obvious correlations were discarded. For
the absence of HR-COCH2CO or HR-NHCH (i, i þ 1)
ROESY cross-peaks, the ω bonds were set at 180° (force
constant of 16 kcal mol-1
straints were included in the restrained MD. The restrained MD
A
-2). Only ROESY-derived con-
˚
simulations were conducted using the AMBER70 force field in a
˚
˚
˚
30 A ꢀ 30 A ꢀ 30 A box of standard TIP3P models of
equilibrated water. All water molecules with atoms that come
˚
closer to a solute atom than 2.3 A were eliminated. A 50 ps
simulation at 1200 K was used for generating 50 random
structures that were subsequently subjected to a 20 ps restrained
MD with a 50% scaled force field at the same temperature,
(iii) Cell Adhesion Assays.76 Plates (96 wells) (Corning, New
York, NY) were coated by passive adsorption with fibronectin
(10 μg/mL) or vitronectin (10 μg/mL) overnight at 4 °C. Cells
were counted and exposed to different concentrations of each
compound for 30 min at room temperature to allow the
ligand-receptor equilibrium to be reached. Stock solutions
(10-2 M) of the assayed compounds were prepared in 33%
DMSO in phosphate-buffered saline (v/v); further dilutions
were done in PBS alone. The highest concentration of DMSO
in the assays was 1% of the stock solution. Control cells were
exposed to the same concentration of DMSO. At the end of the
incubation time, the cells were plated (50000 cells/well) and
incubated at room temperature for 1 h. Then, all the wells were
washed with PBS to remove nonadherent cells, and 50 μL of
hexosaminidase [4-nitrophenyl N-acetyl-β-D-glucosaminide dis-
solved at a concentration of 7.5 mM in 0.09 M citrate buffer
solution (pH 5) and mixed with an equal volume of 0.5% Triton
X-100 in water] was added. This product is a chromogenic
substrate for β-N-acetylglucosaminidase that is transformed in
4-nitrophenol whose absorbance is measured at 405 nm. As
previously described,77 there is a linear correlation between
absorbance and enzymatic activity. Therefore, it is possible to
identify the number of adherent cells in treated wells, interpolat-
ing the absorbance values of the unknowns in a calibration
curve. The reaction was blocked by addition of 100 μL of a
stopping solution [50 μM glycine and 5 μM EDTA (pH 10.4)],
and the plates were read in a Victor2 Multilabel Counter
(Perkin-Elmer, Waltham, MA). Experiments were conducted
in quadruplicate and were repeated at least three times. Data
analysis and IC50 values were calculated using Graph-Pad Prism
3.0 (GraphPad Software, San Diego, CA).
followed by 20 ps with full restraints (distance force constant of
˚
7 kcal mol-1
A
-2), after which the system was cooled in 5 ps to
50 K. H-Bond interactions were not included, nor were torsion
angle restraints. The resulting structures were minimized with
3000 cycles of steepest descent and 3000 cycles of conjugated
gradient (convergence of 0.01 kcal A-1 mol-1). The backbones
˚
of the structures were clustered by the rmsd analysis module of
HyperChem. Unrestrained MD simulation in explicit water was
performed for 10 ns at 298 K, at constant temperature and
pressure (Berendsen scheme,79 bath relaxation constant of
0.2).80 For 1-4 scale factors, van der Waals and electrostatic
interactions are scaled in AMBER to half their nominal value.
The integration time step was set to 0.1 fs. Box equilibration was
set to 10 ps.
Molecular Docking. All calculations were run using the
€
Schrodinger suite of programs (http://www.schrodinger.com)
through the Maestro graphical interface.
(i) Protein Setup. The recently determined crystal structure of
the extracellular domain of the integrin RVβ3 receptor in com-
plex with cilengitide and in the presence of the proadhesive ion
Mn2þ (PDB entry 1L5G) was used for docking studies. Docking
was performed only on the globular head of the integrin because
the headgroup of integrin has been identified in the X-ray
structure as the ligand-binding region. The protein structure
was set up for docking as follows; the protein was truncated to
residues 41-342 for chain R and residues 114-347 for chain β.
Due to a lack of parameters, the Mn2þ ions in the experimental
protein structure were modeled via replacement with Ca2þ ions.
The resulting structure was prepared using the Protein Prepara-
tion Wizard of the graphical user interface Maestro and the
OPLSAA force field.
(iv) Flow Cytometry Assay. To assess the amount of intact,
apoptotic or necrotic cells, an Annexin V-Fluos (Roche)/propi-
dium iodide (PI) (Sigma-Aldrich) assay was performed. Double
staining against annexin V together with PI separates cells, not
(ii) Docking. The automated docking calculations were per-
formed using Glide72 (Grid-based Ligand Docking with
Energetics) within the framework of Impact version 4.5 in a

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1 117
standard precision mode (SP). The grid generation step started
from the extracellular fragment of the X-ray structure of the
RVβ3 complex with cilengitide, prepared as described in Protein
Setup. The center of the grid-enclosing box was defined by the
center of the bound ligand. The enclosing box dimensions,
which are automatically deduced from the ligand size, fit the
entire active site. For the docking step, the size of the bounding
(16) Thijssen, V. L. J. L.; van Beijnum, J. R.; Mayo, K. H.; Griffioen,
A. W. Identification of novel drug targets for angiostatic cancer
therapy; it takes two to tango. Curr. Pharm. Des. 2007, 13, 3576–
3583.
(17) Schmidmaier, R.; Baumann, P. Anti-adhesion evolves to a promis-
ing therapeutic concept in oncology. Curr. Med. Chem. 2008, 15,
978–990.
(18) Ruoslahti, E.; Pierschbacher, M. D. Arg-Gly-Asp: A versatile cell
˚
recognition signal. Cell 1986, 44, 517–518.
box for placing the ligand center was set to 12 A. No further
(19) DSouza, S. E.; Ginsberg, M. H.; Plow, E. F. Arginyl-glycyl-
aspartic acid (RGD): A cell-adhesion motif. Trends Biochem. Sci.
e
modifications were applied to the default settings. The Glide-
Score function was used to select 20 poses for each ligand. Glide
was initially tested for its ability to reproduce the crystallized
binding geometry of cilengitide. The program was successful in
reproducing the experimentally found binding mode of this
compound, as it corresponds to the best-scored pose.
1991, 16, 246–250.
(20) Henry, C.; Moitessier, N.; Chapleur, Y. Vitronectin receptor RVβ3
integrin antagonists: Chemical and structural requirements for
activity and selectivity. Mini-Rev. Med. Chem. 2002, 2, 531–42.
(21) Wermuth, J.; Goodman, S. L.; Jonczyk, A.; Kessler, H. Stereo-
isomerism and biological activity of the selective and superactive
Rvβ3 integrin inhibitor cyclo(-RGDfV-) and its retro-inverso pep-
tide. J. Am. Chem. Soc. 1997, 119, 1328–1335.
Acknowledgment. We thank MIUR (PRIN 2004) and
Bologna University (Funds for Selected Topics) for providing
financial support and Stepbio srl, Bologna (Bologna, Italy)
for technical support. We also thank Consorzio Spinner for
providing a grant to R.D.M. (Project 023/08).
(22) Cacciari, B.; Spallato, G. Non peptidic Rvβ3 antagonists: Recent
developments. Curr. Med. Chem. 2005, 12, 51–70.
€
€
(23) Pfaff, M.; Tangemann, K.; Muller, B.; Gurrath, M.; Muller, G.;
Kessler, H.; Timpl, R.; Engel, J. Selective recognition of cyclic
RGD peptides of NMR defined conformation by RIIbβ3, RVβ3,
and R5β1 integrins. J. Biol. Chem. 1994, 269, 20233–20238.
€
(24) Gurrath, M.; Muller, G.; Kessler, H.; Aumailley, M.; Timpl, R.
Supporting Information Available: Inhibition of tube forma-
tion of HUVEC on Matrigel (Figure S1), Rvβ3/R5β1 integrin
antagonists that did not induce apoptosis or necrosis in endo-
thelial cells (Figure S2), analytical characterization of PMRI
RGD mimetics 4 and 5, nonobvious ROESY cross-peaks
observed for 4 and 5 (Tables S1 and S2), conformational
analysis of 4 and 5 in solution, and molecular docking of 4-9.
This material is available free of charge via the Internet at http://
pubs.acs.org.
Conformation/activity studies of rationally designed anti-adhesive
RGD peptides. Eur. J. Biochem. 1992, 210, 911–921.
(25) Nisato, R. E.; Tille, J.-C.; Jonczyk, A.; Goodman, S. L.; Pepper,
M. S. Rvβ3 and Rvβ5 integrin antagonists inhibit angiogenesis in
vitro. Angiogenesis 2003, 6, 105–119.
€
(26) Dechantsreiter, M. A.; Planker, E.; Matha, B.; Lohof, E.;
€
Holzemann, G.; Jonczyk, A.; Goodman, S. L.; Kessler, H.
N-Metylated Cyclic RGD peptides as highly active and selective
Rvβ3 integrin antagonists. J. Med. Chem. 1999, 42, 3033–3040.
(27) Cai, W.; Chen, X. Anti-angiogenic cancer therapy based on
integrin Rvβ3 antagonism. Anticancer Agents Med. Chem. 2006,
6, 407–428.
(28) Heckmann, D.; Kessler, H. Design and chemical synthesis of
integrin ligands. Methods Enzymol. 2007, 426, 463–503.
(29) Xiong, J.-P.; Stehle, T.; Zhang, R.; Joachimiak, A.; Frech, M.;
Goodman, S. L.; Amin Arnaout, M. Crystal structure of
the extracellular segment of integrin Rvβ3 in complex with an
Arg-Gly-Asp ligand. Science 2002, 296, 151–155.
References
(1) Brower, V. Tumor angiogenesis, new drugs on the block. Nat.
Biotechnol. 1999, 17, 963–968.
(2) Arndt, T.; Arndt, U.; Reuning, U.; Kessler, H. Integrins in
angiogenesis: Implications for tumor therapy. In Cancer Therapy:
Molecular Targets in Tumor-Host Interactions; Weber, G. F., Ed.;
Horizon Bioscience: Norfolk, U.K., 2005.
(30) Marinelli, L.; Lavecchia, A.; Gottschalk, K.-E.; Novellino, E.;
Kessler, H. Docking studies on Rvβ3 integrin ligands: Pharmaco-
phore refinement and implications for drug design. J. Med. Chem.
2003, 46, 4393–4404.
(3) Curley, G. P.; Blum, H.; Humphries, M. J. Integrin antagonists.
Cell. Mol. Life Sci. 1999, 56, 427–441.
(4) Brooks, P. C.; Clark, R. A.; Cheresh, D. A. Requirement of
vascular integrin Rvβ3 for angiogenesis. Science 1994, 264, 569–
571.
(31) Zhang, L.; Mattern, R.-H.; Malaney, T. I.; Pierschbacher, M. D.;
Goodman, M. Receptor-bound conformation of the R5β1 integrin
antagonist by 15N-edited 2D transferred nuclear Overhauser
effects. J. Am. Chem. Soc. 2002, 124, 2862–2863.0
(5) Hynes, R. O. A reevaluation of integrins as regulators of angiogen-
esis. Nat. Med. 2002, 8, 918–921.
ꢁ
(32) Zimmermann, D.; Guthc-hrlein, E. W.; Malesevic , M.; Sewald, K.;
(6) Reynolds, L. E.; Wyder, L.; Lively, J. C.; Taverna, D.; Robinson,
S. D.; Huang, X.; Sheppard, D.; Hynes, R. O.; Hodivala-Dikle, K.
M. Enhanced pathological angiogenesis in mice lacking β3 integrin
or β3 and β5 integrins. Nat. Med. 2002, 8, 27–34.
(7) Cardarelli, P. M.; Lobl, T. J. Peptide inhibitors of β1 integrins. In
Leukocyte Recruitment in Inflammatory Diseases; Peltz, G., Ed.;
Springer: Heidelberg, Germany, 1996; pp 275-294.
(8) Ruoslahti, E. Integrin signaling and matrix assembly. Tumor Biol.
1996, 17, 117–124.
(9) Okada, N.; Watarai, M.; Ozeri, V.; Hanski, E.; Caparon, M.;
Sasakawa, C. A matrix form of fibronectin mediates enhanced
binding of Streptococcus pyogenes to host tissue. J. Biol. Chem.
1997, 272, 26978–26984.
(10) Kim, S.; Harris, M.; Varner, J. A. Regulation of integrin Rvβ3-
mediated endothelial cell migration and angiogenesis by integrin
R5β1 and protein kinase A. J. Biol. Chem. 2000, 275, 33920–33928.
(11) Ikari, Y.; Yee, K. O.; Schwartz, S. M. Role of R5β1 and Rvβ3
integrins on smooth muscle cell spreading and migration in fibrin
gels. Thromb. Haemostasis 2000, 84, 701–705.
Wobbe, L.; Heggemann, C.; Sewald, N. Integrin R5β1 ligands:
Biological evaluation and conformational analysis. ChemBioChem
2005, 6, 272–276.
(33) Benfatti, F.; Cardillo, G.; Fabbroni, S.; Galzerano, P.; Gentilucci,
L.; Juris, R.; Tolomelli, A.; Baiula, M.; Sparta’, A.; Spampinato, S.
Synthesis and biological evaluation of non-peptide Rvβ3/R5β1
integrin dual antagonists containing 5,6-dihydropyridin-2-one
scaffolds. Bioorg. Med. Chem. 2007, 15, 7380–7390.
(34) Takagi, J.; Strokovich, K.; Springer, T. A.; Walz, T. Structure of
integrin R5β1 in complex with fibronectin. EMBO J. 2003, 22,
4607–4615.
(35) Smallheer, J. M.; Weigelt, C. A.; Woerner, F. J.; Wells, J. S.;
Daneker, W. F.; Mousa, S. A.; Wexler, R. R.; Jadhav, P. K.
Synthesis and biological evaluation of nonpeptide integrin antago-
nists containing spirocyclic scaffolds. Bioorg. Med. Chem. Lett.
2004, 14, 383–387.
(36) Kumar, C. C.; Malkowski, M.; Yin, Z.; Tanghetti, E.; Yaremko,
B.; Nechuta, T.; Varner, J.; Liu, M.; Smith, E. M.; Neustadt, B.;
Presta, M.; Armstrong, L. Inhibition of angiogenesis and tumor
growth by SCH221153, a dual Rvβ3 and Rvβ5 integrin receptor
antagonist. Cancer Res. 2001, 61, 2232–2238.
(12) Bewick, M. A.; Lafrenie, R. M. Adhesion dependent signalling in
the tumour microenvironment: The future of drug targeting. Curr.
Pharm. Des. 2006, 12, 2833–2848.
(37) Marinelli, L.; Meyer, A.; Heckmann, D.; Lavecchia, A.; Novellino,
E.; Kessler, H. Ligand binding analysis for human R5β1 integrin:
Strategies for designing new a5b1 integrin antagonists. J. Med.
Chem. 2005, 48, 4204–4207.
(38) Heckmann, D.; Meyer, A.; Marinelli, L.; Zahn, G.; Stragies, R.;
Kessler, H. Probing integrin selectivity: Rational design of highly
active and selective ligands for the R5β1 and Rvβ3 integrin receptor.
Angew. Chem., Int. Ed. 2007, 46, 3571–3574.
(13) Mitjans, F.; Meyer, T.; Fittschen, C.; Goodman, S.; Jonczyk, A.;
Marshall, J. F.; Reyes, G.; Piulats, J. Use of gastrin-releasing
peptide promoter for specific expression of thymidine kinase gene
in small-cell lung carcinoma cell. Int. J. Cancer 2000, 87, 716–723.
(14) Plow, E. F.; Haas, T. A.; Zhang, L.; Loftus, J.; Smith, J. W. Ligand
binding integrins. J. Biol. Chem. 2000, 275, 21785–21788.
(15) Ruoslahti, E. RGD and other recognition sequences for integrins.
Annu. Rev. Cell Dev. Biol. 1996, 12, 697–715.

118 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 1
Gentilucci et al.
(39) Gentilucci, L.; Cardillo, G.; Tolomelli, A.; Spampinato, S.; Sparta,
A.; Squassabia, F. Cyclotetrapeptide mimics based on a 13-mem-
bered, partially modified retro-inverso structure. Eur. J. Org.
Chem. 2008, 729–735.
(40) Gentilucci, L.; Cardillo, G.; Tolomelli, A.; Squassabia, F.;
De Marco, R.; Chiriano, G. Cyclopeptide analogs for generating
new molecular and 3D diversity. Comb. Chem. High Throughput
Screening 2009, 12, 929–939.
(41) Schumann, F.; Muller, A.; Koksch, M.; Muller, G.; Sewald, N. Are
β-amino acids γ-turn mimetics? Exploring a new design principle
for bioactive cyclopeptides. J. Am. Chem. Soc. 2000, 122, 12009–
12010.
(60) Cheresh, D. A. Human endothelial cells synthesize and express an
Arg-Gly-Asp-directed adhesion receptor involved in attachment to
fibrinogen and von Willebrand factor. Proc. Natl. Acad. Sci. U.S.
A. 1987, 84, 6471–6475.
(61) Kawaguchi, M.; Hosotani, R.; Ohishi, S.; Fujii, N.; Singh Tulachan,
S.; Koizumi, M.; Toyoda, E.; Masui, T.; Nakajima, S.; Tsuji, S.; Ida,
J.; Fujimoto, K.; Wada, M.; Doi, R.;Imamura, M. Amovel synthetic
Arg-Gly-Asp-containing peptide cyclo(-RGDf[dbond]V-) is the
potent inhibitor of angiogenesis. Biochem. Biophys. Res. Commun.
2001, 288, 711–717.
(62) Carmelit, P.; Jain, R. K. Angiogenesis in cancer and other diseases.
Circ. Res. 2000, 87, 176–178.
(42) Chorev, M.; Goodman, M. A dozen years of retro-inverso pepti-
(63) Liotta, L.; Steeg, P.; Stetler-Stevenson, W. Cancer metastasis and
angiogenesis: An imbalance of positive and negative regulation.
Cell 1991, 64, 327–336.
(64) Hanahan, D.; Folkman, J. Patterns and emerging mechanisms of
the angiogenic switch during tumorigenesis. Cell 1996, 86, 353–364.
(65) Kessler, H. Conformation and biological activity of cyclic peptide.
Angew. Chem., Int. Ed. 1982, 21, 512–523.
(66) Stradley, S. J.; Rizo, J.; Bruch, M. D.; Stroup, A. N.; Gierasch,
L. M. Cyclic pentapeptides as models for reverse turns: Determina-
tion of the equilibrium distribution between type I and type II
conformations of Pro-Asn and Pro-Ala β-turns. Biopolymers 1990,
29, 263–287.
(67) For a leading reference on the use of a cryoprotective DMSO-d6/
H2O mixture as a biomimetic medium, see: Temussi, P. A.; Picone,
D.; Saviano, G.; Amodeo, P.; Motta, A.; Tancredi, T.; Salvadori,
S.; Tomatis, R. Conformational analysis of an opioid peptide in
solvent media that mimic cytoplasm viscosity. Biopolymers 1992,
32, 367–372 and references cited herein .
domimetics. Acc. Chem. Res. 1993, 26, 266–273.
(43) Chorev, M. The partial retro-inverso modification: A road tra-
veled together. Biopolymers 2005, 80, 67–84.
(44) Han, Y.; Giragossian, C.; Mierke, D. F.; Chorev, M. Ni-to-Niþ3-
ethylene-bridged partially modified retro-inverso tetrapeptide
β-turn mimetic: Design, synthesis, and structural characterization.
J. Org. Chem. 2002, 67, 5085–5097.
(45) Lee, Y. S.; Agnes, R. S.; Davis, P.; Ma, S.-w.; Badghisi, H.; Lai, J.;
Porreca, F.; Hruby, V, J. Partial retro-inverso, retro, and inverso
modifications of hydrazide linked bifunctional peptides for opioid
and cholecystokinin (CCK) receptors. J. Med. Chem. 2007, 50,
165–168.
(46) Fletcher, M. D.; Campbell, M. M. Partially modified retro-inverso
peptides: Development, synthesis, and conformational behavior.
Chem. Rev. 1998, 98, 763–795.
(47) Glenn, M. P.; Kelso, M. J.; Tyndall, J. D. A.; Fairlie, D. P.
Conformationally homogeneous cyclic tetrapeptides: Useful new
three-dimensional scaffolds. J. Am. Chem. Soc. 2003, 125, 640–641.
(48) Norgren, A. S.; Buttner, F.; Prabpai, S.; Kongsaeree, P.;
Arvidsson, P. I. β2-Amino acids in the design of conformationally
homogeneous cyclo-peptide scaffolds. J. Org. Chem. 2006, 71,
6814–6821.
(68) Toniolo, C. Intramolecularly hydrogen-bonded peptide conforma-
tions. CRC Crit. Rev. Biochem. 1980, 9, 1–44.
(69) HyperChem, release 8.0.3; Hypercube Inc.: Gainesville, FL, 2007.
(70) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.;
Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.;
Kollman, P. A. A second generation force field for the simulation
of proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc.
1995, 117, 5179–5197.
(71) In a preliminary investigation of the conformational features of the
PMRI RGD mimetics by ROESY and MD, we analyzed the in-
solution structural features of c[βPheψ(NHCO)Asp(Ot-Bu)ψ-
(NHCO)Gly-Arg(Mtr)] (14) and c[(R)-βPheψ(NHCO)Asp(Ot-
Bu)ψ(NHCO)Gly-Arg(Mtr)], the protected synthetic precursors
of 2 and 3, respectively. While the conformation of fully protected 2
(14) was practically coincident to the type I β-turn/ type I β-turn
structure of 2 (Figures 5 and 6), the predominant conformation of
fully protected 3 was compatible with the inverse γ/inverse γ
structure reported in Figure 8. Apparently, increasing the size of
the side chains of the PMRI CTPs tends to stabilize the inverse γ/
inverse γ structure over the type II β-turn structure (Figure 8).
(49) Maulucci, N.; Chini, M. G.; Di Micco, S.; Izzo, I.; Cafaro, E.;
Russo, A.; Gallinari, P.; Paolini, C.; Nardi, M. C.; Casapullo, A.;
Riccio, R.; Bifulco, G.; De Riccardis, F. Molecular insights into
Azumamide E histone deacetylases inhibitory activity. J. Am.
Chem. Soc. 2007, 129, 3007–3012.
(50) Gentilucci, L.; Cardillo, G.; Tolomelli, A.; De Marco, R.; Garelli,
ꢀ
A.; Spampinato, S.; Sparta, A.; Juaristi, E. Synthesis and confor-
mational analysis of cyclotetrapeptide mimetic β-turn templates
and validation as 3D scaffolds. ChemMedChem 2009, 4, 517–523.
(51) Nakabayashi, S.; Warren, C. D.; Jeanloz, R. W. The preparation of
a partially protected heptasaccharide-asparagine intermediate for
glycopeptide synthesis. Carbohydr. Res. 1988, 174, 279–289.
(52) Morie, T.; Kato, S.; Harada, H.; Fujiwara, I.; Watanabe, K.;
Matsumoto, J.-I. Ring expansion of nitrogen-containing chloro-
methylheteroalicycles via aziridinium intermediates. J. Chem. Soc.,
Perkin Trans. 1 1994, 2565–2569.
€
(72) Glide, version 4.5; Schrodinger, LLC: New York, 2007.
(53) Conner, S. R.; Scott, G.; Aplin, A. E. Adhesion-dependent activa-
tion of the ERK1/2 cascade is by-passed in melanoma cells. J. Biol.
Chem. 2003, 278, 34548–34554.
(54) Park, K.; Kim, Y.-S.; Lee, G. Y.; Parl, R.-W.; Kim, I.-S.; Kim,
S. Y.; Byun, Y. Tumor endothelial cell targeted cyclic RGD-
modified heparin derivative: Inhibition of angiogenesis and tumor
growth. Pharm. Res. 2008, 25, 2786–2798.
(55) Lundell, B. I.; McCarthy, J. B.; Kovach, N. L.; Verfaillie, C. M.
Activation-dependent R5β1 integrin-mediated adhesion to fibro-
nectin decreases proliferation of chronic myelogenous leukemia
progenitors and K562 cells. Blood 1996, 87, 2450–2458.
(56) Belvisi, L.; Riccioni, T.; Marcellini, M.; Vesci, L.; Chiarucci, I.;
Efrati, D.; Potenza, D.; Scolastico, C.; Manzoni, L.; Lombardo,
K.; Stasi, M. M.; Orlandi, A.; Ciucci, A.; Nico, B.; Ribatti, D.;
Giannini, G.; Presta, M.; Carminati, P.; Pisano, C. Biological and
molecular properties of a new Rvβ3/Rvβ5 integrin antagonist. Mol.
Cancer Ther. 2005, 4, 1670–1680.
(57) Del Gatto, A.; Zaccaro, L.; Grieco, P.; Novellino, E.; Zannetti, A.;
Del Vecchio, S.; Iommelli, F.; Salvatore, M.; Pedone, C.; Saviano,
M. Novel and selective Rvβ3 receptor peptide antagonist: Design,
synthesis, and biological behavior. J. Med. Chem. 2006, 49, 3416–
3420.
(58) Giannis, A.; Rubsam, F. Integrin antagonists and other low
molecular weight compounds as inhibitors of angiogenesis: New
drugs in cancer therapy. Angew. Chem., Int. Ed. 1997, 36, 588–590.
(59) Gentilucci, L.; Cardillo, G.; Squassabia, F.; Tolomelli, A.;
Spampinato, S.; Sparta, A.; Baiula, M. Inhibition of cancer cell
adhesion by heterochiral Pro-containing RGD mimetics. Bioorg.
Med. Chem. Lett. 2007, 17, 2329–2333.
(73) Brooks, P. C.; Montgomery, A. M. P.; Rosenfeld, M. Integrin Rvβ3
antagonists promote tumour regression by inducing apoptosis of
angiogenic blood vessels. Cell 1994, 79, 1157–1164.
(74) Kim, S.; Bell, K.; Mousa, S.; Varner, J. A. Regulation of angiogen-
esis in vivo by ligation of integrin R5β1 with the central cell-binding
domain of fibronectin. Am. J. Pathol. 2000, 156, 1345–1362.
(75) Hunakova, L.; Sedlak, J.; Klobusicka, M.; Sulikova, M.;
Chorvath, B. Phorbol ester (TPA)-induced differential modulation
of cell surface antigens in human pluripotential leukemia (K-562)
cell line: Effects of protein kinase inhibitors with broad and PKC
selective inhibitory activity. Neoplasma 1995, 42, 249–253.
(76) Caltabiano, S.; Hum, W. T.; Attwell, G. J.; Gralnick, D. N.;
Budman, L. J.; Cannistraci, A. M.; Bex, F. J. The integrin
specificity of human recombinant osteopontin. Biochem. Pharma-
col. 1999, 58, 1567–1578.
(77) Shibata, H.; Yagi, T. Rate assay of N-acetyl-β-D-hexosaminidase
with 4-nitrophenyl N-acetyl-β-D-glucosaminide as an artificial
substrate. Clin. Chim. Acta 1996, 251, 53–64.
(78) Maeshima, Y.; Manfredi, M.; Reimer, C.; Holthaus, K. A.;
Hopfer, H.; Chandamuri, B. R.; Kharbanda, S.; Kalluri, R.
Identification of the anti-angiogenic site within vascular basement
membrane-derived tumstatin. J. Biol. Chem. 2001, 276, 15240–
15248.
(79) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.;
DiNola, A.; Haak, J. R. Molecular dynamics with coupling to an
external bath. J. Chem. Phys. 1984, 81, 3684–3690.
(80) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J.; Impey, R. W.;
Klein, M. L. Comparison of simple potential functions for simulat-
ing liquid water. J. Chem. Phys. 1983, 79, 926–935.