Biphenyls as potent vitronectin receptor antagonists. Part 2: Biphenylalanine ureas

Article abstract of DOI:10.1016/S0960-894X(03)00023-4

Vitronectin receptor (α_Vβ₃) antagonism has been implicated in a variety of disease states, like restenosis, osteoporosis and cancer. In this work, we present the development of a novel class of biphenyl vitronectin receptor antagonists. Identified from a focused combinatorial library based on para-bromo phenylalanine, these compounds show nanomolar affinity to the vitronectin receptor and display unprecedented SAR. Their binding mode can be rationalized by computational docking studies using the X-ray structure of α_Vβ₃.

Full text of DOI:10.1016/S0960-894X(03)00023-4

Bioorganic & Medicinal Chemistry Letters 13 (2003) 1071–1074
Biphenyls as Potent Vitronectin Receptor Antagonists.
Part 2: Biphenylalanine Ureas
Klaus Urbahns,a,* Michael Harter,a Andrea Vaupel,a Markus Albers,b Delf Schmidt,c
Ulf Bruggemeier,c Beatrix Stelte-Ludwig,d Christoph Gerdesd and Hideki Tsujishita,e
aInstitute of Medicinal Chemistry, Pharma Research Center, Bayer AG, D-42096Wuppertal, Germany
bCombinatorial Chemistry Group, Central Research, Bayer AG, D-51368 Leverkusen, Germany
cInstitute of Molecular Screening Technologies, Pharma Research Center, Bayer AG, D-42096Wuppertal, Germany
dInstitute of Cardiovascular Research, Pharma Research Center, Bayer AG, D-42096Wuppertal, Germany
eDepartment of Chemistry, Research Center Kyoto, Bayer Yakuhin, J-619-0126 Kyoto, Japan
Received 29 November 2002; revised 9 January 2003; accepted 9 January 2003
Abstract—Vitronectin receptor (aVb3) antagonism has been implicated in a variety of disease states, like restenosis, osteoporosis
and cancer. In this work, we present the development of a novel class of biphenyl vitronectin receptor antagonists. Identified from a
focused combinatorial library based on para-bromo phenylalanine, these compounds show nanomolar affinity to the vitronectin
receptor and display unprecedented SAR. Their binding mode can be rationalized by computational docking studies using the
X-ray structure of aVb3.
# 2003 Elsevier Science Ltd. All rights reserved.
The vitronectin receptor aVb3 is a member of the inte-
grin superfamily of cellular adhesion receptors. Small
molecule non-peptide antagonists of this receptor are of
continued interest for the oral treatment of restenosis,
cancer and osteoporosis.1
Here, we wish to report a novel class of biphenylala-
nine ureas with high binding affinity to aVb3. In con-
trast to 1, the compounds result from an alternative
distribution of functional groups around the common
(S)-phenylalanine core and display unprecedented
SAR.
Chemistry
In order to fully exploit the medicinal chemistry of
biphenylalanine derivatives we synthesized a targeted
combinatorial library of 500 members using p-bromo
phenylalanine (p-Br-F) as a central building block
(Scheme 1).3 We immobilized p-Br-F on solid support
via the acidic group, anticipating this moiety to be a
suitable mimic of aspartic acid, a known critical element
of the RGD pharmacophore. Immobilised p-Br-F (I)
was generated by coupling the fMOC-protected amino
acid to Wang resin followed by capping (2,6-dichloro-
benzoyl chloride, pyridine, DMF) and deprotection
(20% piperidine, DMF, rt). It has been demonstrated
earlier, that substituents in direct neighborhood to the
carboxylic group have significant impact on the potency
of b3-integrin antagonists.4 Therefore, substituents
attached to p-Br-F’s amino group were varied broadly,
In an earlier communication, we described potent
vitronectin receptor antagonists based on a biphenyl
moiety. Designed from Kessler’s peptide c-RGDfV, the
discovery of biphenyls such as 1 (Ki=30 nM) has illu-
strated the versatility of the biphenyl fragment as a beta
turn mimetic.2
*Corresponding author at current address: Bayer Yakuhin, Ltd.,
Department of Chemistry, Research Center Kyoto, 6-5-1-3 Kunimi-
dai, Kizu-cho, Soraku-gun, Kyoto 619-0216, Japan. Tel.:+81-774-75-
2485; fax:+81-774-75-2511; e-mail: klaus.urbahns.ku1@bayer.co.jp
0960-894X/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00023-4

1072
K. Urbahns et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1071–1074
cyclic substituents such as cyclopentyl (5), and phenyl
urethanes (6) showed comparable potency. All amides
prepared in the library (e.g., 8, further data not shown)
were inactive and ureas (9,10) were also considerably
less potent, indicating the importance of the urethane
oxygen for binding affinity in this series.
Whereas urethanes were the most potent derivatives in
the series of Table 1, none of the derivatives showed
activities below 100 nM. Only the replacement of the
urethane moiety in 2 by sulfonamides led to a first
example of double-digit nanomolar affinity (11) (Table
2). The substituted aryl sulfonamide 12 was almost
equally effective and exhibited clear SAR that led to the
identification of single digit nanomolar derivatives.
Omitting the para-Cl substituent in 12 led to loss of
activity (13), however potency could be retained by
ortho-methyl substitution (15). The lower activity of its
isomer (20) indicates the importance of a direct linkage
between the aromatic ring and the sulfonamide moiety.
The introduction of the 2,4,6-trimethylphenyl moiety
greatly enhanced the affinity of our leads (21).7
Scheme 1. Combinatorial synthesis of a biphenyl vitronectin receptor
antagonist library: (a) diisopropyl ethylamine/THF 1+10, sulfonyl
chloride or chloroformate reagent; for amide couplings: DIC, DMF,
rt, 3 h; (b) (1) 3-nitro-benzene boronic acid, Pd(PPh3)2Cl2, PPh3,
xylene/H2O, Na2CO3, 85 ꢀC; (2) SnCl2 dihydrate, N-methyl pyrroli-
done; (c) guanidinium group formation: MeS-C=N(BOC)-NH(BOC),
HgCl2, DMF, or: (1) diisopropyl ethylamine/THF 1+10, thio-
phosgene, then: ethylene diamine, DMF; (d) R-CHO, HC(OMe)3,
Bu4NBH4; (e) carboxylic acid, DIC, DMF, rt, 3 h; (f) diisopropyl
ethylamine 4-nitrophenylchloroformate, THF/dichloromethane (1:1),
then: R-NH2, diisopropyl ethylamine, DMF; (g) TFA/dichloro-
methane (1:1); (*=Wang resin).
Table 1. Structure–activity relationship of urethane, amide and urea
biphenyl vitronectin receptor antagonists
using amide, urea, urethane and sulfonamide coupling
chemistry. The bromo atom of p-Br-F was then exploi-
ted for Suzuki coupling reactions with amino-sub-
stituted phenylboronic acids that, in turn were used for
amide and urea formations.3
Compd
R
Ki (nM)
The serine ether 40 was prepared in solution from N-tri-
tylated serine methyl ester following Scheme 2. The
orthogonally protected intermediate VI allowed for late
stage introduction of amino group modifications as well
as for the synthesis of a variety of arginine mimics.5,6
2
3
4
5
6
7
8
9
10
O-Et
O-Neopentyl
O-Isobutyl
O-Cyclopentyl
O-CH2–C6H5
O-(4-MeO)–C6H4
4-MeO–C6H4
HN–CH2–C6H5
NH2
120
800
900
210
800
300
>10,000
850
560
Results
Ki values are means of three dose–response curves.
A first hit compound from the library was the ethoxy
urethane 2, with clear submicromolar binding affinity to
the vitronectin receptor. Whereas the introduction of
larger acyclic substituents (3,4) led to decreased activity,
Table 2. Exploration of the sulfonamide moiety of biphenyl vitro-
nectin receptor antagonists
Compd
R
Ki (nM)
11
12
13
14
15
16
17
18
19
20
21
22
Me
2-CF3-4-Cl–C6H3
2-CF3–C6H4
2-Me–SO2–C6H4
2-Me–C6H4
2,5-Dimethoxy-C6H3
3,4-Dimethoxy-C6H3
2-Methoxy-5-Cl–C6H3
2,3-Cl2–C6H3
CH2–C6H5
2,4,6-Trimethyl-C6H2
(S)-Camphor-10-yl
75
100
>10,000
>10,000
70
500
750
1700
170
300
4
1
Scheme 2. Synthesis of biphenyl ether vitronectin receptor antagonist
40: (a) 4-bromophenol, PPh3, DEAD, toluene, rt, 15 h, 70%; (b)
AsPh3, Pd(Ph3P)2Cl2, 3-amino phenyl boronic acid, Cs2CO3 (3 eq.),
DME, rt, 3 h (40%); (c) n-propyl isocyanate, DABCO, toluene, 14 h rt
(70%); (d) MeOH, HCl, dioxane, rt, 2 h, 30%; (e) mesitylsulfonyl
chloride, dichloromethane, diisopropyl ethylamine, rt, 1 day, 40%; (f)
NaOH (1 M), i-PrOH, 60 ꢀC, 4 h, 50%.
Ki values are means of three dose–response curves.

K. Urbahns et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1071–1074
1073
Although a camphor derivative (22)8 showed even
higher potency, 21 was selected for further SAR studies.
The camphor residue was expected to give rise to struc-
turally complicated metabolites in vivo and was not
investigated further.
as the direct comparison of 26 and 27 demonstrates,
potency was only marginally affected, indicating little
consequences for the active conformation.
Ureas, arising from secondary aliphatic amines were
generally inactive (data not shown), indicating the
importance of two hydrogen bond donors in this moi-
ety. Surprisingly, the introduction of 2-benzimidazolyl
ureas, which proved beneficial in 1 and its congeners,2
did not result in binding affinity at all (38).
The SAR of the urea moiety was also investigated
(Table 3). The simple ammonia derivative 23 showed
weaker activity than the corresponding guanidine deri-
vative 24. We expected this result, as the latter is struc-
turally close to the arginine side chain of the RGD
motif. However, it turned out, that the introduction of
small aliphatic substituents greatly improved activity of
the urea class with an optimum for n-propyl and cyclo-
propyl ureas (21, 26). Larger aliphatic substituents (30,
31) as well as fluorinated derivatives (32, 33, 34) showed
decreased activity. A methyl group was introduced
adjacent to the urea fragment in order to investigate
possible conformational effects of this moiety. However,
It has been repeatedly suggested that the correct spatial
arrangement, in particular the distance between the
carboxylic acid and the arginine group of the RGD
motif is critical for b3 integrin receptor affinity.9,10 In an
effort to investigate the effect of shortening this dis-
tance, the meta/meta isomer of 21 has been prepared
(39). This material was inactive, indicating the difference
to 1 and its active meta/meta derivatives.
In contrast, the introduction of an additional ether atom
(40), resulting in a prolonged RGD motif, did not affect
activity in comparison to 21, indicating the flexibility of
the vitronectin receptor pharmacophore (Table 4).
Table 3. Structure–activity relationship of substituted biphenyl ureas
Compd
X
R1
R2
Ki(nM)
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
O
NH
O
O
O
O
O
O
O
O
O
O
O
O
O
O
H
H
Me
c-Pr
c-Pr
i-Pr
i-Bu
2-Bu
c-Pent
H
H
H
H
Me
H
H
H
Me
H
H
80
11
160
2.5
10
14
10
74
100
20
8
80
F3C–CH2
F3CS–CH2–CH2
F3C–CF2–CH2
C6H5–CH2
o-Pyridyl–CH2
4- 3C-C6FH4–CH2
2-Benzimidazolyl
H
H
H
H
60
>10,000
1000
>10,000
H
Ki values are means of three dose–response curves.
Table 4. The effect of chain-length variation for biphenyl vitronectin
receptor antagonists.
Compd
Structure
Ki (nM)
>10,000
39
Figure 1. Surface representation of the vitronectin receptor ligand-
binding site, with ligands shown as stick models. The Mn2+ ion of
MIDAS is indicated as a yellow sphere. Only receptor amino acids
involved in ligand binding interactions are shown (aV: magenta, b3:
cyan): (a) cyclic Arg-Gly-Asp-DPhe-(N-methyl)-Val (green) and 21
(orange), yellow lines indicate hydrogen bonds between the receptor
and the ligands; (b) 21 (orange), 39 (red), and 40 (cyan).
40
4
Ki values are means of three dose–response curves

1074
K. Urbahns et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1071–1074
We used computational docking experiments to gain
further insight into the SAR trends of 21, 39 and 40.
After alignment with the recently described FAME
routine,11 the compounds were docked into the X-ray
structure of the vitronectin receptor bound to the ligand
c-RGDf(N-Me)V12,13 (Fig. 1a). The study suggests a
binding mode of 21 similar to that of the cyclic penta-
peptide. The carboxylic acid moiety would interact with
MIDAS in the b3 chain whereas both urea NH groups
form hydrogen bonds with Asp218 in the aV chain.
Further contacts include a hydrophobic biphenyl/
Ala218 interaction as well as a hydrogen bond from the
sulfonamide NH to Tyr122. Additionally, the 2,4,6-tri-
methylphenyl group interacts with Tyr122, presumably
via p-stacking.
3. All compounds synthesized on Wang resin (Rapp Poly-
mere, Tubingen, loading: 1.08 mmol/g) were obtained in good
yields (50–80%). The purity was >90% as determined by
HPLC (UV) in all cases. For further details of the syntheses
see: Urbahns, K.; Harter, M.; Albers, M.; Vaupel, A.;
Schmidt, D.; Stelte-Ludwig, B.; Gerdes, C.; Lustig, K.WO
0035864, 2000, Chem. Abstr. 2000, 133, 43809.
4. An accessory hydrophobic binding site (exosite) for the
RGD motif has been hypothesized to be located close to the
MIDAS site of aVb3 and related integrins. Hartman, G. D.;
Egbertson, M. S.; Halczenko, W.; Laswell, W.; Duggan, M. E.;
Smith, R. L.; Naylor, A. M.; Manno, P. D.; Lynch, R. J.;
Zhang, G.; Chang, C. T.-C.; Gould, R. J. Med. Chem. 1992,
35, 4640.
5. Baldwin, J. E.; Spivey, A. C.; Schofield, C. J.; Sweeney, J. B.
Tetrahedron 1993, 49, 6309.
6. Experimental details in: Vaupel, A. US 2002095050, 2002;
Chem. Abstr. 2002, 137, 94007.
Figure 1b shows the superposition of 39 and 40 in the
ligand-binding site of the receptor. Surprisingly, 39’s
urea and carboxylic acid moiety could well interact with
the receptor. The reduced distance between both motifs
was apparently not the reason for 39’s inability to bind
to aVb3. The model suggests however that one of 39’s
meta-substituted phenyl rings would clash sterically
with Ala218 from the beta 3 subunit.14 In addition, the
binding event would entail a reduction of the biphenyl
torsion angle from 44ꢀ (21) and 57ꢀ (40) towards a more
periplanar, higher energy conformation (27ꢀ). In contrast,
the more flexible 40 can easily adopt the correct binding
conformation, avoiding any steric contact with Ala218.
7. The 2,4,6-trimethyl phenyl sulfonyl residue has been incor-
porated into vitronectin receptor antagonists before: Voss,
E. J.; Jadhav, P. K.; Smallheer, J. M.; Batt, D. G.; Pitts, W. J.;
Wityak, J. WO 9637492, 1996; Chem. Abstr. 1997, 126, 89360.
8. (a) Incorporation of camphor residues into beta 3 integrin
antagonists: Diefenbach, B.; Fittschen, C.; Gante, J.; Good-
man, S.; Wiesner, M.; Rippmann, F. DE 95-195548709; Chem.
Abstr. 1997, 127, 95612. (b) Brashear, K. M.; Cook, J. J.;
Bednar, B.; Bednar, R. A.; Gould, R. J.; Halczenko, W.;
Holahan, M. A.; Lynch, R. J.; Hartman, G. D.; Hutchinson,
J. H. Bioorg. Med. Chem. Lett. 1997, 21, 2793.
9. Greenspoon, N.; Hershkoviz, R.; Alon, R.; Varon, D.;
Shenkman, G.; Marx, G.; Federman, S.; Kapustina, G.; Lider,
O. Biochemistry 1993, 32, 1001.
10. Integrin purification: Wong, A.; Hwang, S. M.; McDevitt,
P.; McNulty, D.; Stadel, J. M.; Johanson, K. Mol. Pharm 1996,
50, 529. (b) Echistatin as radioligand: Kumar, C. C.; Nie,
H. M.; Rogers, C. P.; Malkowski, M.; Maxwell, E.; Catino,
J. J.; Armstrong, L. J. Pharm. Exp. Ther. 1997, 283, 843.
11. Ullmann, G. M.; Hauswald, M.; Jensen, A.; Kostic, N. M.;
Knapp, E. W. Biochemistry 1997, 36, 16187.
12. (a) Xiong, J. P.; Stehle, T.; Diefenbach, B.; Zhang, R.;
Dunker, R.; Scott, D. L.; Joachimiak, A.; Goodman, S. L.;
Aranout, M. A. Science 2001, 294, 339. (b) Xiong, J. P.;
Stehle, T.; Zhang, R.; Joachimiak, A.; Frech, M.; Goodman,
S. L.; Aranout, M. A. Science 2002, 296, 151.
In conclusion, we have identified a novel, independent
class of biphenyl vitronectin receptor antagonists with
single digit nanomolar affinities from a combinatorial
library. Additionally, we demonstrated the usefulness of
docking simulations to the vitronectin receptor struc-
ture in rationalising SAR trends.
Acknowledgements
We would wish to thank Peter Reinemer for protein
support as well as Axel Jensen and Markus Hauswald
for providing us with the FAME program.
13. The X-ray structure of the receptor in complex with the
cyclic peptide was taken from the Protein Data Bank (PDB:
1L5G). The CONCORD 4.0.6 module in SYBYL 6.8 gener-
ated the 3-D structures of 21, 39 and 40. The alignment with
the cyclic peptide was generated using the FAME program.11
After removal of the peptide co-ordinates, these structures were
docked to the vitronectin receptor-binding pocket. The result-
ing complexes were then energy minimized using the
MMFF94s force field (e=1r) until criteria were reached
References and Notes
1. (a) Recent reviews: Samanen, J.; Jonak, Z.; Rieman, D.;
Yue, T. L. Current Pharm. Des. 1997, 3, 545. (b) Miller, W. H.;
Keenan, R. M.; Willette, R. N.; Lark, M. W. Drug Discov.
Today 2000, 5, 397. (c) Duggan, M. E.; Hutchinson, J. H. Exp.
Op. Ther. Pat. 2000, 10, 1367. (d) Hoelzemann, G. Idrugs
2001, 40, 72.
2. Urbahns, K.; Harter, M.; Albers, M.; Schmidt, D.; Stelte-
Ludwig, B.; Bruggemeier, U.; Vaupel, A.; Gerdes, C. Bioorg.
Med. Chem. Lett. 2002, 12, 205.
˚
(potential gradient RMS <0.05 kcal/(mol A). The co-ordinates
of the Mn2+ ions, the ion chelating amino acids as well as all
˚
distant residues (5 A) were kept constant. The docking proce-
dure was performed using SYBYL 6.8 (Tripos Inc., MO, USA).
14. A similar effect is known for c-RGDfV. Substitution of
the glycine residue by l-or d-alanine leads to loss of activity:
Haubner, R.; Gratias, R.; Diefenbach, B.; Goodman, S. L.;
Jonczyk, A.; Kessler, H. J. Am. Chem. Soc. 1996, 118, 7461.