Nonpeptide RGD antagonists: A novel class of mimetics, the 5,8-disubstituted 1-azabicyclo[5.2.0]nonan-2-one lactam

Article abstract of DOI:10.1016/S0960-894X(03)00181-1

The 1-azabicyclo[5.2.0]nonan-2-one lactam 1 adequately substituted on both cycles A and B as scaffolds mimics the conformationally constrained β-turn of the tripeptide RGD signaling motif of fibronectin. Using an in vitro assay, we establish that trans di

Full text of DOI:10.1016/S0960-894X(03)00181-1

Bioorganic & Medicinal Chemistry Letters 13 (2003) 1561–1564
Nonpeptide RGD Antagonists: A Novel Class of Mimetics, the
5,8-Disubstituted 1-Azabicyclo[5.2.0]nonan-2-one Lactam
Erika Bourguet,a Jean-Louis Baneres,a Joseph Parello,a Xavier Lusinchi,b
Jean-Pierre Girarda,* and Jean-Pierre Vidala
a
´
´
Laboratoire de Chimie Biomoleculaire et Interactions Biologiques, Unite Mixte de Recherche CNRS 5074,
´
Universite Montpellier I, Faculte de Pharmacie, 15 Av. C. Flahault, BP 14491, 34093 Montpellier Cedex 5, France
´
bInstitut de Chimie des Substances Naturelles du CNRS, Gif-sur-Yvette, France
Received 18 September 2002; revised 16 December 2002; accepted 10 February 2003
Abstract—The 1-azabicyclo[5.2.0]nonan-2-one lactam 1 adequately substituted on both cycles A and B as scaffolds mimics the
conformationally constrained b-turn of the tripeptide RGD signaling motif of fibronectin. Using an in vitro assay, we establish that
trans diastereoisomer 1b dissociates a soluble fibronectin–integrin a5b1 complex at concentrations comparable to those of a linear
RGDS peptide as a competitor.
# 2003 Elsevier Science Ltd. All rights reserved.
Attempts to mimic b-turn topologies of signaling pep-
tide motifs, as is the case of the Arg-Gly-Asp (RGD)
tripeptide in a variety of extracellular matrix proteins,
have converged towards the synthesis of constrained
non-peptidic molecules, several of them including a
seven-membered benzodiazepine scaffold.1ꢀ5 The con-
formational adaptability of a 7-membered ring is cer-
tainly part of the ability of such benzodiazepine-type
compounds to interact efficiently with different cell
adhesion receptors (integrins or IN).
spatial relationship similar to that of the RGD motif in
the crystal structure of an RGD-containing Fn frag-
ment,7 a suitably substituted amide group was added at
position 5 of lactam 1.
The occurrence in these non-peptidic molecules of both
guanidinium and carboxylate groups are essential ele-
ments for mimicking the Arg and Asp side chains of
RGD, respectively.6
The 1-azabicyclo[5.2.0]nonan-2-one bicyclic skeleton is
not common and has been scarcely synthesized. The few
reported synthetic routes8 are not straightforward and
usually result in poor yields. The synthetic approach we
have developed (Scheme 1) takes advantage of several
aspects of our recently reported synthesis of 1-azabicy-
clo[5.2.0]nonan-2-one itself and derivatives.9 Among the
considerations that led to such a synthesis were: (a) the
lactam skeleton of compound 1 that is generated by
photochemical rearrangement of the hexahydroindole
oxaziridine 8 without interfering with the stereochemistry
at the substituted positions; (b) the easy availability of
such bicyclic structures from substituted indoles.
In this paper, we report on the synthesis of a new
bicyclic lactam 1 (with cycle A being a seven-membered
ring and B being a 4-membered ring) as a non-peptidic
scaffold mimicking the RGD b-turn topology. Lactam 1
is substituted on both cycles A and B with the former
bearing an analogue of the Arg side chain and the latter
bearing an analogue of the Asp side chain. Finally, the
CH2 in the azetidine cycle might well mimic the CH2 of
Gly in the RGD motif. To maintain an Arg-to-Asp
The non-peptide RGD mimetic 1 synthesized here was
designed to compete with fibronectin for binding to the
a5b1 integrin. Evidence has been provided that a5b1 and
*Corresponding author. Tel.: +33-4-6754-8620; fax: +33-4-6754-
8625; e-mail: girard@pharma.univ-montp1.fr
0960-894X/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00181-1

1562
E. Bourguet et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1561–1564
Scheme 1. Reagents and conditions (isolated yield): (a) CH2O, l-proline, Meldrum’s acid, CH3CN, 22 h, 20 ꢁC, 83%; (b) Cu, EtOH/pyridine, 16 h,
reflux, 88%; (c) H2, Pd/C (10%), AcOH/water, 70 kg/cm2, 70 ꢁC, 22 h, 91%; (d) Boc2O, NaOH, dioxane/water, 16 h, 20 ꢁC, 92%; (e) N-benzyloxy-
carbonyl-1,3-diaminopropane, DCC, DMAP, CH2Cl2, 21h, 20 ꢁC, 72%; (f) TFA, CH2Cl2, 1h, 20 ꢁC, 94%; (g) step 1: NCS, ether/CH3CN, 1h,
20 ꢁC; step 2: KO2, DMF, 4 h, 20 ꢁC, 58% over the two steps; (h) mCPBA, CH2Cl2, 30 min, 0 ꢁC, 86%; (i) hn, CH3CN, 8 h, 18 ꢁC, 45%; (j) H2,
Pd(OH)2/C, EtOH, 6 h, 20 ꢁC, 98%; (k) N,N0-bis-t-butoxycarbonylthiourea, HgCl2, Et3N, DMF, 30 min at 0 ꢁC, 48 h at 20 ꢁC, 79%; (l) NaOH,
dioxane/water, 30 min at 0 ꢁC, 20 h at 20 ꢁC, 45%; (m) CF3COOH, CH2Cl2, 4 h, 20 ꢁC, 98%.
its ligand fibronectin play critical roles in angiogenesis,
resulting in tumor growth in vivo.10 Importantly, anti-
bodies, peptide and non-peptide antagonists of integrin
a5b1 have been shown to block angiogenesis induced by
several growth factors in both chick embryo and murine
models.10 Non-peptide a5b1 antagonists could thus have
a therapeutic application as either angiogenesis-block-
ing compounds or tumor-targeting molecules.
Pd/C in acetic acid/water, resulting in the concomitant
deprotection of the benzyl group (90% yield; mixture of
cis and trans diastereoisomers). The amino group of 4
was protected by Boc, before introducing the Arg-pre-
cursor chain at position 5 in cycle A upon reaction of
the free carboxylic acid with N-benzyloxycarbonyl-1,3-
diaminopropane (DMAP-catalyzed DCC coupling).
Boc deprotection of 5, by treatment with TFA, gave the
octahydroindole 6 in 94% yield. To prepare oxaziridine
8, we have demonstrated9 that the most convenient
strategy corresponds to the oxidation of the octahy-
droindole imine 7. The chloramine, prepared by action
of NCS on 6, upon reaction with potassium superoxide
(KO2) in DMF, led to imine 7 in 58% yield. Oxidation
of imine 7 to oxaziridine 8 was achieved by m-chloro-
perbenzoic acid (mCPBA). The photochemical
rearrangement of oxaziridine 8 into lactam 9 is the cri-
tical step of the synthesis. As expected,9 this photore-
arrangement was not influenced by disubstitutions.
Upon UV irradiation (for conditions see ref 9), oxazir-
idine 8 afforded the expected 5,8-disubstituted 1-azabi-
cyclo[5.2.0]nonan-2-one 9 (45% yield of isolated pure
product).
Chemistry
The general synthetic route11 to compound 1 is illu-
strated in Scheme 1. Compound 1 was prepared starting
from the monosubstituted benzyl 5-indolecarboxylate
2,12 after introduction of a second substituent to afford
the disubstituted compound 3. The latter was obtained
by addition of Meldrum’s acid13-formaldehyde followed
by cleavage of the Meldrum’s group with copper powder.
The newly introduced ester chain in 3 mimics the aspartyl
side chain. The transformation of indole 3 into the octa-
hydroindole compound 4, was carried out by catalytic
reduction under pressure with H2 in the presence of 10%

E. Bourguet et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1561–1564
1563
Deprotection of the N-benzyloxycarbonyl chain by cat-
alytic hydrogenolysis of lactam 9 with Pd(OH)2/C
(Pearlmans catalyst) gave 10 in 98% yield. Coupling of
10 in the presence of HgCl2 with N,N0-bis-t-butoxy-
carbonylthiourea14 resulted in the formation of com-
pound 11 in 79% yield. Saponification of the ethyl ester
by sodium hydroxide followed by deprotection of N-t-
butoxycarbonyl by treatment with TFA gave 5-(guanidino
carboxamido propyl) 1-azabicyclo[5.2.0]nonan-2-one 8-
propionic acid 1, as a mixture of two diastereoisomers, a
major one 1a, 85%, and a minor one 1b, 1 5%. The two
isomers, 1a and 1b, were then separated by preparative
HPLC.15 1H NMR spectra15 provided clear evidence (J )
that the major 1a had the cis relative configuration and the
minor 1b the trans relative configuration.
two orders of magnitude in efficiency to compete with
Fn for binding to the a5b1 soluble receptor. We antici-
pate that one of the enantiomers of 1b might still display
an enhanced potency in releasing Fn from its receptor.
Exact knowledge of the positioning of 1b within the
RGD-binding site of the integrin receptor is of crucial
importance for adapting the syntheses of our non-pep-
tide RGD mimetics. Since our synthetic strategy is inti-
mately linked to the indole chemistry (Scheme 1) this
offers a powerful strategy for introducing a variety of
structural modulations in a regio- and stereoselective
manner that could be exploited for enhancing the affi-
nity, as well as the selectivity of recognition of the
RGD-type antagonist by a given integrin receptor. In
vivo degradation by proteolysis of a non-peptide
antagonist as 1b might be substantially reduced in
comparison to peptidyl RGD-analogues, thus offering a
real advantage in a therapeutical perspective. In con-
clusion, we have developed a versatile and novel route
to 5,8-disustituted 1-azabicyclo[5.2.0]nonan-2-one lac-
tams as compound 1b. trans diastereoisomer 1b diplays
an efficacy at dissociating the IN/Fn interactions in our
in vitro assay comparable to that observed for a linear
RGD peptide. This suggests that this bicyclic lactam is a
convenient framework as a non-peptidic compound
mimicking the b-turn topology of the RGD signal in
fibronectin, and likely other RGD-containing extra-cel-
lular matrix proteins involved in cell adhesion.
Biological Results
The capacity of compounds 1 to act as RGD mimetics
was evaluated using a miniaturized recombinant a5b1
mini-integrin16 that is soluble by only including part of
the extracellular domain with different fibronectin
recognition sites in both a and b subunits.
We previously described an in vitro assay to evaluate
the fibronectin-binding properties of our mini-integ-
rin.16 This assay17 was used here to determine the ability
of compounds 1 to dissociate the soluble a5b1/fibro-
nectin complex. Briefly stated, the 1:1 complex (formed
between the recombinant a5b1 mini-integrin and a
fibronectin 3Fn8-11 fragment that is produced as a GST
fusion protein and encompasses the 8th–11th type III
Fn modules with a single RGD motif in the 10th mod-
ule), or IN/Fn complex, was immobilized on a Ni-nitri-
loacetic agarose support (His-tagged mini-integrin16)
and the competitor compounds was added at increasing
concentrations. This linear tetrapeptide was used as a
reference compound since it has been previously shown
to efficiently dissociate the a5b1/fibronectin complex in
vivo.18 The dissociation of the IN:Fn complex was esti-
mated by determining (by UV spectrophotometry; see
ref 17) the amount of Fn eluted from the column at a
given competitor concentration.
Acknowledgements
This work was supported by ARC (grant No. 5151;
Villejuif, France).
References and Notes
1. McDowel, R. S.; Blackburn, B. K.; Gadek, T. R.; McGee,
L. R.; Rawson, T.; Reynolds, M. E.; Robarge, K. D.; Somers,
T. D.; Thorsett, E. D.; Tischler, M.; Webb, R. R.; Venuti,
M. C. J. Am. Chem. Soc. 1994, 116, 5077.
2. Ku, T. W.; Ali, F. E.; Barton, L. S.; Bean, J. W.; Bondinell,
W. E.; Burgess, J. L.; Callahan, J. F.; Calvo, R. R.; Chen, L.;
Eggleston, D. S.; Gleason, J. G.; Huffman, W. F.; Hwang,
S. M.; Jakas, D. R.; Karash, C. B.; Keeman, R. M.; Kopple,
K. D.; Miller, W. H.; Newlander, K. A.; Nichols, A.; Parker,
M. F.; Peishoff, C. E.; Samanen, J. M.; Uzinskas, I.; Ven-
slavsky, J. W. J. Am. Chem. Soc. 1993, 115, 8861.
The two diastereoisomers, 1a and 1b, were assayed for
their capacity to dissociate the IN:Fn complex. As shown
in Table 1, it appears that the trans diastereoisomer 1b
dissociates the IN/Fn complex at an IC50 value close to
that observed for RGDS. A remarkable observation is
that both cis 1a and trans 1b diastereoisomers, differ by
3. Keenan, R. M.; Miller, W. H.; Kwon, C.; Ali, F. E.; Call-
ahan, J. F.; Calvo, R. R.; Hwang, S. M.; Kopple, K. D.;
Peishoff, C. E.; Samanen, J. M.; Wong, A. S.; Yuan, C. K.;
Huffman, W. F. J. Med. Chem. 1997, 40, 2289.
4. Gadek, T. R.; McDowel, J. W.; Abstracts of Papers, 21th
ACS National Meeting, New Orleans, LA, March 1996;
MEDI 235.
Table 1. Competitiona between fibronectin (3Fn8-11 fragment) and
different RGD analogues for binding16 to soluble integrin a5b1 (IC50
values in M)
5. Callahan, J. F.; Bean, J. W.; Burgess, J. L.; Eggleston, D. S.;
Hwang, S. M.; Kopple, K. D.; Koster, P. F.; Nichols, A.;
Peishoff, C. E.; Samanen, J. M.; Vasko, J. A.; Wong, A.;
Huffman, W. F. J. Med. Chem. 1992, 35, 3970.
RGDS
Mimetic 1
Mimetic 1b
Mimetic 1a
2.2ꢂ0.3 ꢃ 10ꢀ6 1.3ꢂ0.4 ꢃ 10ꢀ3 6.5ꢂ0.5 ꢃ 10ꢀ5 3.1ꢂ0.4 ꢃ 10ꢀ3
6. (a) Ali, F. E.; Samanen, J. M.; Calvo, R.; Romoff, T.;
Yellin, T.; Vasko, J.; Powers, D.; Stadel, J.; Bennett, D.;
Berry, D.; Nichols, A. In Potent Fibrinogen Receptor Antago-
nists Bearing Conformational Constraints; Smith, J. A.; Rivier,
J. E.; Eds., ESCOM: Leiden, 1992; p 761. (b) Samanen, J.; Ali,
aCompetition between the RGD mimetics, the RGDS linear peptide and
fibronectin GST-3Fn8-11 fragment; (see text) for binding to the recom-
binant a5b1 mini-integrin. The stoichiometric ratio of the a5b1:GST-
3Fn8-11 complex is 1:1 while the stoichiometric ratio is 1:2 with RGDS
(Baneres et al., unpublished results). For Kd values see ref 17.

1564
E. Bourguet et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1561–1564
F.; Romoff, T.; Calvo, R.; Sorenson, E.; Vasko, J.; Storer, B.;
Berry, D.; Bennet, D.; Strohsaker, M.; Powers, D.; Stadel, J.;
Nichols, A. J. Med. Chem. 1991, 34, 3114.
3.99 (dd, 1H, H7, J78=0 Hz, J76a=2.4 Hz, J76b=3.5 Hz).
Trans isomer 1b: 1H NMR (360 MHz, CD3OD, d ppm):1.88
(ddd, 1H, H6b, J6b6a=14 Hz, J6b7=13 Hz, J6b5=8 Hz); 2.18
(dd, 1H, H6a, J6a5=0 Hz, J6a6b=14 Hz, J6a7=7.0 Hz); 2.26 (m,
1H, H8, J87=0 Hz, J89a=7.2 Hz, J89b=6.8 Hz); 2.45 (dd, 1H,
H5, J56a=0 Hz, J56b=6.5 Hz, J54a=12.5, J54b=0 Hz); 3.14
(dd, 1H, H9b, J9b9a=10.4 Hz, J9b8=6.8 Hz); 3.75 (dd, 1H, H9a,
J9a9b=10.4 Hz, J9a8=7.2 Hz); 3.96 (dd, 1H, H7, J78=0 Hz,
J76a=7.0 Hz, J76b=13 Hz).
16. (a) Baneres, J. L.; Roquet, F.; Green, M.; LeCalvez, H.;
Parello, J. J. Biol. Chem. 1998, 27, 24744. (b) Baneres, J. L.;
Roquet, F.; Martin, A.; Parello, J. J. Biol. Chem. 2000, 275,
588.
17. A standard competition assay was carried out as follows.
Step 1: a 400 mL volume of an equimolecular mixture (5 mM)
of the His-tagged mini-integrin a5b1 and of the human fibro-
nectin fragment 3Fn8-11 (see ref 15 for the definition of these
macromolecules) in buffer A (12.5 mM Na-borate, 150 mM
NaCl, 5 mM MgCl2 at pH 8.3) was added to a 400 mL volume
of Ni-nitriloacetic agarose resin in buffer A previously loaded
on a Pasteur pipet and stopped with glass wool; the total
mixture was incubated for 30 min at room temperature. Step 2:
elution is started and the resin is washed once with a 400 mL
volume of buffer A (no UV absorbance detected in the eluate).
Step 3: a 400 mL volume of a given RGD analogue at varying
concentration was loaded onto the column; after a 30 min
incubation, elution was started and the resin washed with an
additional 400 mL volume of buffer A. The UV absorbance of
the total volume eluted (2ꢃ400 mL) was measured at 276 nm
(thus giving the quantity of 3Fn8-11 eluted; a molar absorp-
tivity of 45,800 L molꢀ1 cmꢀ1 was used15), as well as at 320 nm
(no correction for scattering effects was necessary). Competi-
tion profiles included 10–12 points each and were analyzed by
a computer-assisted program (PRISM, San Diego, CA, USA)
with Kd(Fn)=2.5ꢃ10ꢀ7 M, n=1, Kd(RGDS)=1.7ꢃ10ꢀ5 M,
n=2 (thermodynamic data with n being the stoichiometric
ratios: Baneres et al., not yet published).
7. Leahy, D. J.; Aukhil, I.; Erickson, H. P. Cell 1996, 84, 155.
8. (a) Hanessian, S.; McNaughton-Smith, G.; Lombart, H. G.;
Lubell, W. D. Tetrahedron 1997, 53, 12789. (b) Milligan, G. L.;
Mossman, C. J.; Aube, J. J. Am. Chem. Soc. 1995, 117, 10449.
(c) Ouazzani-Chadi, L.; Quirion, J. C.; Troin, Y.; Gramain,
J. C. Tetrahedron 1990, 46, 7751. (d) Edwards, O. E.; Paton,
J. M.; Benn, M. H.; Mitchell, R. E.; Watanatada, C.; Vohra,
K. N. Canadian J. Chem. 1971, 49, 1648. (e) Angiolini, M.;
Araneo, S.; Belvisi, L.; Cesarotti, E.; Checchia, A.; Crippa, L.;
Manzoni, L.; Scolastico, C. Eur. J. Org. Chem. 2000, 14, 2571.
9. Bourguet, E.; Baneres, J. L.; Girard, J. P.; Parello, J.;
Vidal, J. P.; Lusinchi, X.; Declercq, J. P. Organic Lett. 2001, 3,
3067.
10. Kim, S.; Bell, K.; Mousa, S. A.; Varner, J. A. Am.
J. Pathol. 2000, 156, 1345.
1
11. All new compounds gave satisfactory 360 MHz H NMR,
mass spectral and microanalytical data.
12. Jacobs, R. T.; Brown, F. J.; Cronk, L. A.; Aharony, D.;
Buckner, C. K.; Kusner, E. J.; Kirkland, K. M.; Neilson, K. L.
J. Med. Chem. 1993, 36, 394.
13. Farlow, D. S.; Flaugh, M. E.; Horvarth, S. D. Organic
Prep. Proc. int. 1981, 13, 39.
14. Iwanowicz, E. J.; Poss, M. A.; Lin, J. Synth. Commun.
1993, 23, 1443.
15. The 5-[guanidinocarboxamidopropyl]-1-aza-bicyclo-[5.2.0]-
nonan-2-one 8-propanoic acids cis 1a and trans 1b were sepa-
rated by reverse phase HPLC Lichrospher 100 RP18, 5 mm,
250ꢃ10 mm. Elution with H2O/CH3CN/TFA 86:14:0,1 affor-
ded successively diastereoisomers 1a (7.5 min, 85%) and 1b
1
(10.5 min, 15%). cis isomer 1a: H NMR (360 MHz, CD3OD,
d ppm): 1.67 (ddd, 1H, H6b, J6b6a=15 Hz, J6b7=3.4 Hz,
J6b5=12 Hz); 2.26 (ddd, 1H; H5, J56a=0 Hz, J56b=12 Hz,
J54a=2.6 Hz, J54b=12 Hz); 2.31 (m, 1H, H8, J87=0 Hz,
J89a=10 Hz, J89b=7 Hz); 2.50 (dd, 1H, H6a, J6a,6b=15 Hz,
J6a7=2.6 Hz, J6a5=0 Hz); 3.39 (dd, 1H, H9b, J9b9a=10 Hz,
J9b8=7 Hz); 3.62 (dd, 1H, H9a, J9a9b=10 Hz, J9a8=10 Hz);
18. Pierschbacher, M. D.; Ruoslahti, E. Nature 1984, 309,
30.