Highly constrained bicyclic VLA-4 antagonists

Article abstract of DOI:10.1016/j.bmcl.2006.11.011

VLA-4 is implicated in several inflammatory and autoimmune disease states. A series of cyclic β-amino acids (β-aa) was studied as VLA-4 antagonists. Binding affinity was highly dependent on the dihedral angle (φ{symbol}) between the amino and the carboxyl

Full text of DOI:10.1016/j.bmcl.2006.11.011

Bioorganic & Medicinal Chemistry Letters 17 (2007) 597–601
Highly constrained bicyclic VLA-4 antagonists
Linda L. Chang,a,* Quang Truong,a George A. Doss,c Malcolm MacCoss,a
Kathryn Lyons,c Ermengilda McCauley,b Richard Mumford,b Gail Forrest,b
Stella Vincent,c John A. Schmidtb and William K. Hagmanna
aDepartment of Medicinal Chemical Research, Merck Research Laboratories, Rahway, NJ 07065, USA
bDepartment of Immunology and Rheumatology, Merck Research Laboratories, Rahway, NJ 07065, USA
cDepartment of Drug Metabolism, Merck Research Laboratories, Rahway, NJ 07065, USA
Received 3 October 2006; revised 2 November 2006; accepted 3 November 2006
Available online 7 November 2006
Abstract—VLA-4 is implicated in several inflammatory and autoimmune disease states. A series of cyclic b-amino acids (b-aa) was
studied as VLA-4 antagonists. Binding affinity was highly dependent on the dihedral angle (/) between the amino and the carboxyl
termini of the b-aa. Compound 5m where the b-aa is embedded in a bicycle possesses the most preferred / (120°). It is a potent and
bioavailable VLA-4 antagonist (VCAM-Ig a4b1 IC50 = 54 nM, rat po F = 49%).
Ó 2006 Elsevier Ltd. All rights reserved.
The integrin VLA-4 (very late antigen-4, a4b1, CD49d/
CD29) is a heterodimeric cell surface glycoprotein trans-
membrane receptor.1 It is expressed on all leukocytes ex-
cept platelets and neutrophils, and is a key mediator in
cell–cell and cell–matrix interactions. It mediates
cell adhesion to vascular cell adhesion molecule-1
(VCAM-1), an immunoglobulin (IgG) expressed on
endothelial cells in response to proinflammatory cyto-
kines at sites of inflammation, and it binds extracellular-
ly to CS-1, an alternatively spliced form of fibronectin
(Fn).2 These cell adhesion interactions may be required
for the activation, migration, proliferation, and differen-
tiation of leukocytes during normal and/or pathophysi-
ological processes. Thus, inhibition of VLA-4 may
produce a reduction in the migration and/or activation
of cell types important to sustaining a prolonged inflam-
matory response.2
and low molecular weight antagonists.3 Thus, there is
a substantial interest in developing small-molecule
VLA-4 antagonists.4
Derivatization of VLA-4 antagonists, which emerged
from screening of a combinatorial library,5 led to more
potent compounds such as 1 (IC50 = 1.4 nM).6 Starting
from such capped dipeptides, our objective was to con-
ceive structures which would enhance the pharmacoki-
netic profile compared to reference compounds such as
1.6a One of the ideas followed was to replace the P3
phenylalanine in 1 with a non-natural-amino acid; for
example, aminobenzoic acid derivatives or their saturat-
ed counterparts, as depicted in 2. In this basic design,
the benzyl group found in P3 of 1 (or its equivalent),
which is also present in many VLA-4 antagonists,4b–f
is absent (Fig. 1).
Therapeutic efficacy in several animal models of inflam-
mation and autoimmune diseases, such as asthma, mul-
tiple sclerosis, and Crohn’s disease, via the inhibition of
the interaction between VLA-4 and its ligands has been
validated by both anti-VLA-4 monoclonal antibodies
H
O
N
P2
R
H
N
N
OH
O
SO2
Cl
N
CO2H
O
SO2
Cl
Cl
P3
Cl
Keywords: VLA-4 antagonists; Integrin; a4b1; Anti-inflammatory;
Autoimmune; b-Amino acid; Conformational restriction; Bioavailable;
Bicyclic.
R
=
P1
,
,
1
2
*
Corresponding author. Tel.: +1 732 594 6465; fax: +1 732 594
5350; e-mail: linda_chang@merck.com
Figure 1.
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.11.011

598
L. L. Chang et al. / Bioorg. Med. Chem. Lett. 17 (2007) 597–601
H7a
H7a
H7b
H2
NO2
H7b
OH
O-t-Bu
N
c, d
a, b
CO2Me
N
H
MeO2C
H3
O
SO2
Cl
O
a
+
O2N
H2
CO2Me
H3
3
5a-g
NO2
4
Cl
7a
7b
Scheme 1. Reagents and conditions: (a) (3,5-Cl2)C6H3SO2Cl, NEt(i-
Pr)2, DMAP, DCM, 94%; (b) 1:1 TFA/DCM, rt, 40–98%; (c)
R(NH2)(CO2Et), EDC, HOBt, NMM, DCM, 80–93%; (d) NaOH/
MeOH, rt, 4 h, 70–90%.
MeO2C
H
d, e
b, c
5m
H
NH2
Herein, we show the evolution of this approach, from
inactive entities to structurally novel, orally bioavailable
compounds with good potency for VLA-4.
1HNMR chemical shifts (δ) for H2, H3, H7a, and H7b in 7a and 7b.
cpd
7a
7b
H2
H3
H7a
H7b
3.08
3.72
5.43
4.73
1.65
1.74
1.72
1.92
Most of the compounds studied were prepared by sulf-
onylation of a protected P-2 amino acid such as 3, fol-
lowed by coupling of the resulting key intermediate 47
with an appropriate P3 amino acid. Subsequent base
hydrolysis furnished 5a–g. Compounds 6a, b were pre-
pared analogously, utilizing suitably protected P2-ami-
no acids.
Scheme 2. Reagents and conditions: (a) cyclopentadiene, Et2O, 0 °C,
40 min, 80%; (b) H2, PtO2, EtOAc, quant.; (c) HCO2NH4, 10% Pd/C,
MeOH, 30%; (d) 4, EDC, HOBt, NMM, DCM, 82%; (e) 1 equiv
NaOH/MeOH, rt, 2 h, 80%.
For 5h–k, n and o, the diastereomeric esters (Scheme 1,
step c) were hydrolyzed separately with excess base at
27 °C overnight. This resulted in partial racemization
of the a-carbon in each case. For 5j, the more active iso-
mer of 5j–k, these isomers were separated by preparative
HPLC. The structural assignments of the resulting iso-
Reduction of the double bond in 7a was followed by
transfer hydrogenation of the nitro group to provide
the aminoester which was coupled with 4. Hydrolysis
of the product provided the desired compound 5m, spec-
trally identical to 5m obtained previously, verifying the
initial structural assignment of 5m from HPLC separa-
tion. Compounds 5p, the bicyclo[2.2.2] analog of 5m,
and q, the oxygen isostere of 5m, were also synthesized
according to Scheme 2. For 5p, the Diels–Alder reaction
required prolonged heating with excess 1,3-cyclohexadi-
ene at 130 °C. Under these conditions, the endo-ni-
tro:exo-nitro stereoisomeric products (analogous to 7a
and b) were obtained in a 4:1 ratio.
1
mers 5l, m were made by H NMR spectral correlation
with spectra of isomerically pure intermediates and via
decoupling experiments. The more active isomer 5m, de-
rived from the epimerization, showed a W-coupling be-
tween the aforementioned a-proton (endo) and one of
the protons (anti) on the apical carbon of the bicy-
clo[2.2.1] system.
A stereochemically unambiguous synthesis of 5m was
undertaken to ascertain its stereochemical assignment.
As shown in Scheme 2, methyl (E)-3-nitroacrylate8 was
reacted with cyclopentadiene to provide the Diels–Alder
adducts 7a (endo-nitro) and 7b (exo-nitro) in a 6:1 ratio.9
The structure of the major component was deduced to
be 7a as described below. Based on literature prece-
dents9,10 the H3’s in 7a and b were assigned as the lower
field methyne protons (d 5.43 and 4.73), while the H2’s
were assigned as the higher field ones (d 3.08 and
3.72). Initial attempts to assign the stereochemistry at
C2 and C3 were based on long-range W-coupling in
The VLA-4 binding affinity of the compounds discussed
was assessed by measuring the reduction in binding of
125I-VCAM-Ig to a suspension of Jurkat cells (a human
a4b1 T cell line) in the presence of the test compound, as
previously described.6b All assays were run at least in
duplicate.
To begin, a series of aminobenzoic acids and (aminophe-
nyl)acetic acids was examined (Table 1, 5a–d). These
were inactive as VLA-4 antagonists. This lack of activity
compared to the parent 5e was thought to be due to the
rigidity of the aromatic ring. To increase the flexibility
of P3 and to diverge from a planar configuration be-
tween the amino and the carboxy moieties of P3, cyclo-
hexyl systems with the amino and the carboxyl groups in
a b-cis (5f) and b-trans (5g) relationship were evaluated.
1
the H NMR (500 MHz) spectra. In 7a, a definitive W
coupling is observed between H2 (d 3.08) and H7a (d
1.65), analogous to that observed in 5m earlier. For this
compound, no W coupling is observed between H3 (d
5.43) and H7a. The minor component 7b exhibits a W
coupling between H3 (d 4.73) and H7a (d 1.74) but not
between H2 (d 3.72) and H7a. Concurrent NOE studies
demonstrated that in 7a, an NOE is observed between
H3 and H7b, but not between H2 and H7b, while in 7b,
there is an NOE between H2 and H7b. but not between
H3 and H7b. In this manner, 7a was unambiguously as-
signed as the major product from the Diels–Alder
reaction.
Table 1 shows 5f, g have significantly improved binding
affinities over 5a. These results suggest that the dihedral
angle (/) between the amino and the carboxylic acid
groups in P3 could play an important role in optimizing
the interaction of the ligand with the receptor. In the
case of 5a, b, / = 0° and the compounds were inactive.
For 5c, d, the methylene spacer provided some flexibility
in /. For 5e, f, the outcome was much improved.

L. L. Chang et al. / Bioorg. Med. Chem. Lett. 17 (2007) 597–601
599
Table 1. VLA-4 binding affinities of (N-arylsulfonyl)prolyl amino
acids 5a–g
compounds was intriguing considering that in the chair
conformation, both sets of compounds have / of ꢀ60°.
We hypothesized that the difference in receptor binding
was due to either the availability of a trans-diaxial rela-
tionship (/ ꢁ 180°) between the amino and the carboxy
groups of 5g, or differences in the dihedral angles in the
boat and twist boat conformations between 5f and 5g, to
the extent that they are available. Considering the sterics
involved, the trans-compounds 5g would be expected to
achieve the boat/twist boat conformations (/ = 120–
150°) more easily than the cis-compounds 5f (/ ꢀ 0–
30°). This line of reasoning in combination with the data
for 5f, g suggests that the optimal dihedral angle lies be-
tween 60° and 180°.
H
Cl
N
SO
N
R
2 O
5
Cl
Compound
R
VLA-4 IC50 (nM)
5a
0% inhibition at 100 lM
HO2C
CO2H
5b
5c
2% inhibition at 100 lM
11% inhibition at 100 lM
A set of conformationally constrained bicyclic systems
was examined to further study the issue. The rigid bicy-
clo[2.2.1] system served as the starting point. As shown
in Table 2, the exo-norbornane system gave rise to two
epimeric pairs, 5h, i; likewise, the endo-norbornane sys-
tem gave rise to 5j, k. Pairwise, these bicyclo[2.2.1] com-
pounds were more active than their cyclohexyl
counterparts (5f, g). This could be due to conformation-
al constraint and/or the added hydrophobicity of the
bicyclo system. Compounds 5h, i, where the amino
group is exo on the bicyclic moiety, are less active than
5j, k, where the amino group is endo.
HO2C
CO2H
5d
5e
43% inhibition at 100 lM
1500
CO2H
HO2C
HO2C
CO2H
5f
9460
525
,
H
H
H
The most active pair of the 4 epimeric pairs, 5j, was sep-
arated into 5l (endo-acid) and 5m (exo-acid). The data
suggest that the endo-amino-exo-carboxylic acid combi-
nation (5m) represents the most preferred spatial
arrangement for receptor–ligand interaction among the
cyclic systems studied. The dihedral angle in this case
is close to 120°, within the range of our prediction.
CO2H
5g
H
The data from compounds 5n–q show that the integrin
binding affinity is affected substantially by steric changes
in this region and/or by small changes in the dihedral an-
gle (5n–p). The bicyclo[2.2.2] system (5p) is much more
flexible than the bicyclo[2.2.1] (norbornane) system.
Therefore, the operative dihedral angle is less well de-
fined in 5p compared to 5m, contributing to the decrease
in binding affinity. An exquisite sensitivity to increases
in polarity is demonstrated by 5q. Overall, data from
5m and the unsubstituted system 5e show that by judi-
ciously building conformational constraints into a linear
system via cyclic entities, a 30-fold increase in receptor–
ligand binding was achieved.
In these compounds, / = 60° assuming the chair confor-
mation predominates, implying that / = 60° is preferred
to / = 0°.
In an effort to optimize P2 in the presence of a cyclic P3,
we opened the pyrrolidyl ring in 5g, providing more
freedom in this putative b-turn region. The data suggest
that substitution at the a-position is critical for main-
taining binding (6a vs 6b, Fig. 2). A comparison of data
from 5g versus 6b shows that there is no significant
advantage in opening up the proline ring.
Meanwhile, we felt a closer examination of the data
from cyclohexane amino acids 5f and g was in order.
The 18-fold difference in binding affinity between these
Compound 5m is a selective VLA-4 antagonist versus
a4b7 (VCAM-Ig a4b7 IC50 = 13 lM) and exhibits an
excellent pharmacokinetic profile in Sprague–Dawley
rats with 49% oral bioavailability (using a cassette-dos-
ing regimen, Table 3).11
CO2H
CO2H
H
H
α
N
N
N
SO2
N
This study demonstrates the feasibility of designing non-
natural cyclic amino acids to fit into receptor–ligand
binding sites normally requiring L-amino acids for a ser-
ies of VLA-4 antagonists. Conformational analysis of
these cyclic b-amino acids was instrumental in arriving
at a preferred dihedral angle of ꢀ120° between the amino
and the carboxylic acid functionalities for optimal
O
O
SO2
Cl
Cl
Cl
6b: IC50 = 462 nM
Cl
6a: inactive @ 100 μM
Figure 2.

600
L. L. Chang et al. / Bioorg. Med. Chem. Lett. 17 (2007) 597–601
Table 2. VLA-4 binding affinities of (N-arylsulfonyl)prolyl amino acids 5g–q
H
Cl
N
SO
N
R
2 O
5
Cl
Compound
R
VLA-4 IC50 (nM)
525
H
H
HO2C
5g
HO2C
H
*
CO2H
HO2C
*
,
5h, i
299, 1490
H
H
CO2H
H
HO2C
H
*
*
5j, k
95, 322
1597
,
*
H
H
5l
CO2H
HO2C
H
*
,
5m
54
H
*
CO2H
H
HO2C
H
*
5n, o
657, 2160
*
HO2C
H
5p
5q
2740
H
O
*
HO2C
H
25,600
H
*The absolute stereochemical assignment is arbitrary.
Table 3. Pharmacokinetic data parameters of 5m
Acknowledgments
Routea Doseb CLp
(ml/min/kg)
t1/2 (h)c AUC
F (%)
We thank Dr. Linus Lin for the preparation of com-
pound 5q. We thank Ms. Amy Bernick for mass spectral
analysis.
(lM h kg/mg)
iv
po
1 mpk 3.3
2 mpk
1.9
1.7
9.1
5.5
—
49
—
a Sprague–Dawley rats.
b Cassette-dosing.
c Plasma half-life (0–8 h).
References and notes
1. (a) Butcher, E. C. Cell 1991, 67, 1033; (b) Springer, T. A.
Cell 1994, 76, 301; (c) Cox, D.; Aoki, T.; Seki, J.;
Motoyama, Y.; Yoshida, K. Med. Res. Rev. 1994, 14, 195.
2. (a) Bevilacqua, M. P. Annu. Rev. Immunol. 1993, 11, 767;
(b) Postigo, A. A.; Teixido, J.; Sanchez-Madrid, F. Res.
Immunol. 1993, 144, 723; (c) Elices, M. J. Curr. Opin. Anti-
Inflam. Immunol. 1999, 1, 15.
receptor–ligand interactions in this class of compounds. In
this manner, the b-alanine moiety of a micromolar com-
pound 5ewas transformed to a norbornane system, leading
to a series of compounds lacking the substituted phenyl or
benzyl group typically found in this class of VLA-4 antag-
onists.4b The best of these was 5m, a 54 nM orally bioavail-
able and selective VLA-4 receptor antagonist.
3. (a) Lobb, R. R.; Hemler, M. E. J. Clin. Invest. 1994, 94,
1722; (b) Lin, K.-C.; Castro, A. C. Curr. Opin. Chem. Biol.

L. L. Chang et al. / Bioorg. Med. Chem. Lett. 17 (2007) 597–601
601
1998, 2, 453; (c) Molossi, S.; Elices, M.; Arrhenius, T.;
Diaz, R.; Coulber, C.; Rabinovitch, M. J. Clin. Invest.
1995, 95, 2601; (d) Foster, C. A. J. Allergy Clin. Immunol.
1996, 98, S270; (e) Elices, M. Curr. Opin. Anti-Inflam.
Immunomod. Invest. Drugs 2000, 2, 228.
G.; McCauley, E.; Egger, L. E.; Kidambi, U.; Schmidt, J.
A.; Lyons, K.; Stearns, R.; Vincent, S.; Colletti, A.; Wang,
Z.; Tong, S.; Wang, J.; Zheng, S.; Owens, K.; Levorse, D.;
Hagmann, W. K. Bioorg. Med. Chem. Lett. 2002, 12, 637;
(b) Lin, L. S.; Kopka, I. E.; Mumford, R. A.; Magriotis, P.
A., ; Lanza, T., Jr.; Durette, P. L.; Kamenecka, T.; Young,
D. N.; de Laszlo, S. E.; McCauley, E.; Van Riper, G.;
Kidambi, U.; Egger, L. A.; Tong, X.; Lyons, K.; Vincent,
S.; Stearns, R.; Colletti, A.; Teffera, Y.; Fenyk-Melody, J.;
Schmidt, J. A.; MacCoss, M.; Hagmann, W. K. Bioorg.
Med. Chem. Lett. 2002, 12, 611.
4. (a)
www.fda.gov/cder/drug/infopage/natalizumab/de-
fault.htm+Tysabri& hl=en; Recent Reviews: (b) Hag-
mann, W. K. Curr. Top. Med. Chem. 2004, 1461; (c)
Huryn, D. M.; Konradi, A.; Kennedy, J. D. Curr. Top.
Med. Chem. 2004, 1473; (d) Sing, J.; Adams, S.; Carter, M.
B.; Cuervo, H.; Lee, W.-C.; Lobb, R. R.; Pepeinsky, B.;
Petter, R.; Scott, D. Curr. Top. Med. Chem. 2004, 1497; (e)
Tilley, J. W.; Chen, L.; Sidduri, A.; Fotouhi, N. Curr. Top.
Med. Chem. 2004, 1509; (f) Yang, G. X.; Hagmann, W. K.
Med. Res. Rev. 2003, 369; (g) Lobb, R. R.; Adams, S. P.
Expert Opin. Investig. Drugs 1999, 8, 935; (h) Adams, S.
P.; Lobb, R. R. Annu. Rep. Med. Chem. 1999, 34, 179; (i)
Engleman, V. W.; Kellogg, M. S.; Rogers, T. E. Annu.
Rep. Med. Chem. 1996, 31, 191.
7. This and all other compounds prepared were characterized
by mass spectrum (FAB or LC/MS), and 300 MHz,
400 MHz, or 500 MHz H NMR.
1
8. McMurray, J. E.; Musser, J. H.; Fleming, I.; Fortunak, J.;
Nubling, C.. In Organic Synthesis; Noland, W. E., Ed.;
Wiley & Sons, 1988; Coll. Vol. VI, p 799.
9. Blom, N. F.; Edwards, D. M. F.; Field, J. S.; Michael, J.
P. J. C. S. Chem. Commun. 1980, 1240.
5. Hagmann, W. K.; Durette, P. L.; Lanza, T.; Kevin, N. J.;
de Laszlo, S. E.; Kopka, I. E.; Yong, D.; Magriotis, P. A.;
Li, B.; Lin, L. S.; Yang, G.; Kamenecka, T.; Chang, L. L.;
Wilson, J.; MacCoss, M.; Mills, S. G.; Van Riper, G.;
McCauley, E.; Egger, L. A.; Kidambi, U.; Lyons, K.;
Vincent, S.; Stearns, R.; Colletti, A.; Teffera, J.; Tong, S.;
Fenyk-Melody, J.; Owens, K.; Levorse, D.; Kim, P.;
Schmidt, J. A.; Mumford, R. A. Bioorg. Med. Chem. Lett.
2001, 11, 2709.
10. Fresenius, W., Huber, J. F. K., Pungor, E., Rechnitz, G.
A., Simon, W., West, T. S. Eds.; Tables of Spectral Data
for Structure Determination of Organic Compounds; 1989;
Springer: Berlin, p H90.
11. The cassette dosing regimen that we used was
a
mixture of five compounds in equal amounts; each at
2 mpk po and 1 mpk iv. Whereas there is the possibi-
lity for drug–drug interactions in mixture dosing in
related structures, we did not observe significant
differences in PK parameters between single or mixture
dosing.
6. (a) Kopka, I. E.; Yong, D. N.; Lin, L. S.; Mumford, R. A.;
Magriotis, P. A.; MacCoss, M.; Mills, S. G.; Van Riper,