Amidines as amide bond replacements in VLA-4 antagonists

Article abstract of DOI:10.1016/j.bmcl.2004.01.100

VLA-4 (α₄β₁, very late activating antigen-4), a key cell surface integrin plays an important role in inflammation by promoting leukocyte attachment and extravasation from the vasculature into the peripheral tissues. As such, VLA-4 antagonists may be useful in the treatment, prevention, and suppression of diseases where cell adhesion and migration are important such as asthma, rheumatoid arthritis, and multiple sclerosis. Herein, we report on the discovery, synthesis, and biological evaluation of amidines as small molecule antagonists of VLA-4.

Full text of DOI:10.1016/j.bmcl.2004.01.100

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2323–2326
Amidines as amide bond replacements in VLA-4 antagonists
Theodore M. Kamenecka,a,* You-Jung Park,a Linus S. Lin,a Stephen de Laszlo,a
Ermengilda D. McCauley,b Gail Van Riper,b Linda Egger,c Usha Kidambi,c
Richard A. Mumford,b Sharon Tong,c Wei Tang,c Adria Colletti,c Yohannes Teffera,c
Ralph Stearns,c Malcolm MacCoss,a John A. Schmidtb and William K. Hagmanna
aDepartment of Medicinal Chemistry, Merck Research Laboratories, Rahway, NJ 07065, USA
bDepartment of Inflammation and Rheumatology Research, Merck Research Laboratories, Rahway, NJ 07065, USA
cDepartment of Pharmacology, Merck Research Laboratories, Rahway, NJ 07065, USA
Received 3 December 2003; revised 28 January 2004; accepted 29 January 2004
Abstract—VLA-4 (a4b1, very late activating antigen-4), a key cell surface integrin plays an important role in inflammation by
promoting leukocyte attachment and extravasation from the vasculature into the peripheral tissues. As such, VLA-4 antagonists
may be useful in the treatment, prevention, and suppression of diseases where cell adhesion and migration are important such as
asthma, rheumatoid arthritis, and multiple sclerosis. Herein, we report on the discovery, synthesis, and biological evaluation of
amidines as small molecule antagonists of VLA-4.
Ó 2004 Elsevier Ltd. All rights reserved.
The adhesion molecule VLA-4 (very late activating
antigen-4, a4b1, CD49d/CD29) is a member of the
integrin family that is expressed on all circulating leu-
kocytes except platelets.1 VLA-4 is a receptor for
VCAM-1 (vascular cell adhesion molecule-1) on the
endothelium and the alternatively spliced CS-1 domain
of fibronectin in the extracellular matrix. VCAM-1
expression is upregulated on endothelial cells by
inflammatory cytokines and is highly expressed at sites
of inflammation. The interaction between adhesion
molecules on the leukocyte surface and specific counter
receptors on the vascular endothelium may be important
for the activation, migration, proliferation, and differ-
entiation of leukocytes during normal and pathophysio-
logical processes.2 Monoclonal antibodies directed
against the a4-integrin have been shown to be effective in
animal models of autoimmune and allergic diseases such
as experimental autoimmune encephalomyelitis, colla-
gen-induced arthritis, and eosinophil recruitment.3 As
such, VLA-4 antagonists may be useful in the treatment,
prevention, and suppression of diseases where cell
adhesion and extravasation are important such as mul-
tiple sclerosis, rheumatoid arthritis, and allergic asthma.
Previously, we reported N-phenylsulfonylproline
biphenylalanine derivatives to be potent, specific VLA-4
antagonists.4 Characteristic of this class of compounds
are derivatives 1a,b (Fig. 1). Despite their desirable
potencies and specificities, compounds of this type
(sulfonylated dipeptides) suffer from generally poor
pharmacokinetic properties.
Typically, they have low oral bioavailability (F <10%)
and very high plasma clearance in the rat (Clp >90 mL/
H
H
N
CO2H
N
O
X
SO2
OMe
X
MeO
X = Cl (1a)
X = H (1b)
VLA-4: 0.08 nM
α4β7: 11 nM
Clp: 102 mL/min/kg
F%: 11
VLA-4: 0.10 nM
α4β7: 78 nM
Clp: 107 mL/min/kg
F%: 1.4
Keywords: VLA-4; Integrin; Asthma; Inflammation.
* Corresponding author. Tel.: +1-8582025449; fax: +1-8582025752;
e-mail: theodore_kamenecka@merck.com
log D: 4
log D: 2.6
Figure 1. Potent amide VLA-4 antagonists.
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.01.100

2324
T. M. Kamenecka et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2323–2326
Table 1. Inhibition of VLA-4a by amide bond replacements (IC50, nM)
H
a
CO2H
H
N
n
X
CO2H
N
X
SO2
Z
Z
Cl
SO2
OMe
X
Cl
MeO
H
H
N
CO2t-Bu
N
Number
n
X
Z
VLA-4a
a4b7
b
c
Y
X
SO2
1a
2
1
1
1
2
NH
O
O
O
O
0.08
15
11
OMe
3000
NDc
8400
3
NMe
NH
350
140
4
H,H
X
MeO
a VCAM-Ig IC50 nM.
b MAdCAM-Ig IC50 nM.
c Not determined.
Y = O (1a,b)
Y = S (5a,b)
b
H
N
min/kg). We wondered if the poor pharmacokinetic
properties of this class of compounds was the result of
the amide bond and made efforts to remove or modify
this group (Table 1). Replacing the amide bond with an
ester group (2) led to a 200-fold drop in potency, while
N-methylation (3) led to a 4000-fold drop. This clearly
indicates the necessity of the NH group for potency.
Removal of the amide carbonyl group as in 4 also leads
to a 2000-fold drop in potency. Thus, the NH bond must
also have the appropriate pKa. The secondary amine
may be too basic to engage in the necessary hydrogen
bonding interaction or perhaps a conformational change
in the molecule is responsible for the loss in potency.
Nonetheless, it was still not known if the amide bond
oxygen atom was needed, only that the amide NH was
required as a hydrogen bond donor. We envisioned
amidines as possible amide bond replacements with the
hope that attenuation of the basicity of the group on
nitrogen would have positive effects on potency and
pharmacokinetics.
CO2t-Bu
N
S
X
SO2
OMe
d,e
X
6a,b
MeO
H
H
N
CO2H
N
N
X
SO2
Z
OMe
X
MeO
7a-o
Scheme 1. (a) HATU, HOAt, 2,6-dimethoxybiphenyl alanine tert-
butyl ester, i-Pr2NEt, CH2Cl2; (b) LawessonÕs reagent, C6H5CH3,
70 °C; (c) K2CO3, Me2SO4, acetone, rt; (d) ZNH2, Et3N, MeOH,
50 °C; (e) TFA, CH2Cl2 [NB: for Z ¼ SO2Me (7n), amidine (7b) was
treated with MsCl,TEA].
The synthesis of amidine derivatives is outlined in
Schemes 1 and 2. Synthesis of 1a,b has already been
described.4 Conversion of the amides into thioamides
5a,b was accomplished with Lawessons reagent and
S-methylation was best carried out by stirring with an
excess of dimethyl sulfate in the presence of an inorganic
base. Gram quantities of thioimidates 6a,b were pre-
pared in this manner. Exposure of the thioimidate to an
amine nucleophile in the presence of base afforded the
amidine substitution product and ester cleavage pro-
vided final products (Table 2). If the amine nucleophile
contained another reactive site, side reactions were
sometimes observed. For instance, when hydrazine was
used, the cyclic product (7q) was obtained (Scheme 2).
This cyclization not only occurred in the substitution
reaction, but sometimes in the ester cleavage step, as
evidenced by the formation of 7p. The cyclized products
are much less active against VLA-4 as the acid moiety is
required for potency, however, this could potentially
lend itself to a prodrug approach if the N-acyl bond
were hydrolyzed in vivo.4 The binding results are sum-
marized in Table 2.
Interestingly, changing the oxygen atom in 1a to an NH
gives a compound (7a) with equal to better potency and
specificity. This substitution confirms that the amide
bond oxygen atom is not critical for potency, however, it
does not negate the fact that the amide oxygen or ami-
dine nitrogen may still act as a hydrogen bond acceptor.
Despite the slight loss in potency (7a,c vs 7b,d), the
chlorine atoms were removed from the aryl sulfonyl
group to reduce both molecular weight and lipophilicity,
which might hopefully improve pharmacokinetics. A
methyl group (7c,d) on nitrogen gives compounds
equipotent to 1a or 7a, but larger alkyl groups lead to a
slight loss in potency (7g–m). Regardless, it is remark-
able the range of substituents that is tolerated while still
maintaining sub-nanomolar potency. As the amide NH-
bond is critical for potency, it was postulated that a

T. M. Kamenecka et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2323–2326
2325
Table 2. Inhibition of VLA-4a by N-sulfonylated-proline biphenyl-
alanine derivatives: amidines (IC50, nM)
MeO
H
N
H
H
H
N
N
CO2H
MeO
O
N
MeO
N
SO2
N
X
SO2
N
X
OMe
Z
N
X =H (7p)
X
a, b
X
H
S
b
Number
X
Z
VLA-4a
a4b7
16
N
CO2t-Bu
N
7a
7b
7c
7d
7e
7f
Cl
H
Cl
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
0.05
0.10
0.05
0.12
0.14
0.17
0.13
0.21
0.28
0.17
X
SO2
19
21
40
98
35
27
120
65
78
43
35
69
OMe
Me
Me
OH
OMe
CH2CF3
X
6a,b
MeO
7g
7h
7i
c
CH2CH(CH3)3
CH2CH2OMe
CH2CO2H
7j
H
H
N
7k
7l
CH2CH(CH2CH2O)2 0.28
CH22-pyridyl
CH2CH21-triazole
SO2Me
0.15
0.16
OMe
N
7m
7n
7o
7p
7q
N
X
SO2
N
H
O
0.07
0.09
85% inhd
2.2
4.1
4090
NDe
CN
CH2-2-imidazole–c
NH–c
MeO
58
X
a VCAM-Ig ligand.
X =H (7q)
b MAdCAM-Ig ligand.
c Cyclized product.
d At 10 lM.
Scheme 2. (a) 2-Aminomethyl-1,3-imidazole, MeOH, Et3N; (b) TFA,
CH2Cl2; (c) hydrazine hydrochloride, MeOH, Et3N.
e ND ¼ not determined.
better hydrogen bond donor might provide for tighter
binding. Hence, an electron withdrawing group on
nitrogen should increase the acidity of the amidine NH
and thus, 7n,o were prepared. As anticipated, these were
the two most potent amidines prepared bearing the
simple phenyl sulfonyl residue. The amidines also
maintain their roughly 100-fold specificity for VLA-4
versus a4b7. The cyclized amidines (7p,q) were much less
potent as it was already established that the acid residue
is critical for potency.
Table 3. Physical properties and rat pharmacokinetics of amidines
Number
log D
(pH 7.3)
pKa
F a;b (%)
Clp (mL/
min/kg)
7a
7b
7c
7d
7e
7f
4.1
––
––
0.3
0.2
0.8
0.7
NDND
9.3
1.8
0
90
61
85
81
NDc
2.7
10.2
10.5
––
2.4
2.3
2.8
3.3
<4
––
79
45
64
7g
7h
7i
3.7
––
ND––
1.9
2.6
0
81
The physical properties of the amidines as a class are
interesting (Table 3). Despite the wide range of substi-
tution patterns, log D proved difficult to modify.5
Although relevant in the amide case (1a to 1b), removing
the two chlorine atoms from the phenyl sulfonyl ring in
the amidines (7c to 7d) did not grossly affect log D. Even
highly basic amidines such as 7c,d had similar log DÕs to
1b. This may be due to the fact that this small polar
residue (amidine group) is buried in hydrophobic resi-
dues on both sides. Only when large hydrophobic
groups are incorporated does the log D increase as
observed for 7g,h. Additionally, compounds with the
same log D values (7d vs 7n) could have very different
basicities as indicated by their respective pKa values. The
two most potent amidines (7n,o) are, in fact, not basic at
all based on their pKaÕs and are physically more like the
hydrophobic acid leads 1a,b.
7j
––
NDd
0.6
NDd
0
NDd
67
NDd
111
103
93
7k
7l
2.8
2.4
––
––
)2
)2
7m
7n
7o
7p
7q
2.4
2.5
1.6
3.0
382
3.3
ND––
3.2
0.6
––
86
a Sprague–Dawley rats.
b Dose: 1 mg/kg i.v.; 2 mg/kg p.o.
c ND ¼ not determined.
d Compound not stable.
e Cyclized product.
As a class, the amidines have poor pharmacokinetics in
the rat characterized by high plasma clearances (60–100

2326
T. M. Kamenecka et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2323–2326
mL/min/kg) and low oral bioavailability (0–10%), simi-
lar to the corresponding amides. Amidines 7g,h may
have reduced clearances due to their lipophilicity (higher
log DÕs) resulting from increased protein binding. Cyclic
amidines 7p,q also display poor pharmacokinetics.
Whether this is an inherent property of the cyclic sub-
strate or whether they are hydrolyzed to the acyclic acids
in vivo and consequently suffer from the same problems
as all the other amidines is uncertain.5 Studies were
undertaken to determine the reasons for the poor
pharmacokinetic profile of compounds of this type (i.e.,
8, 7a).6 In a study with Wistar rats dosed with 8 and 1-
aminobenzotriazole (1-ABT, a non-specific mechanism
based inhibitor of P-450Õs), no change in AUC (plasma
concentration vs time curves) or plasma clearance was
observed.7
log D and pKa. Compounds such as 7b and 7g are potent
examples, which have slightly better pharmacokinetic
profiles than 1a and 8. Additionally, the amidines as a
class appear not to suffer from the same clearance
mechanism (mrp-2) experienced by the amides. This
may have advantages in developing an oral VLA-4
antagonist when active transporters such as mrp-2 are
involved.
References and notes
1. Hemler, M. E.; Elices, M. J.; Parker, C.; Takada, Y.
Immunol. Rev. 1990, 114, 45.
2. Elices, M. J. Curr. Opin. Anti-inflamm. Immunol. 1999, 1,
15.
3. (a) Kent, S. J.; Karlik, S. J.; Cannon, C.; Hines, D. K.;
Yednock, T. A.; Fritz, L. C.; Horner, H. C. J. Neuroim-
munol. 1995, 58, 1; (b) Zeidler, A.; Brauer, R.; Thoss, K.;
Bahnsen, J.; Heinrichs, V.; Jablonski-Westrich, D.; Wrob-
lewski, M.; Rebstock, S.; Hamann, A. Autoimmunity 1995,
21, 245; (c) Pretolani, M.; Ruffie, C.; Lapa E Silva, J.-R.;
Joseph, D.; Lobb, R. R.; Vargaftig, B. B. J. Exp. Med.
1994, 180, 795.
Me
H
N
CO2H
N
O
Cl
SO2
VLA-4: 0.07 nM
α4β7: 14 nM
Clp: 110 mL/min/kg
OMe
Cl
8
MeO
4. Hagmann, W. K.; Mumford, R. A.; Durette, P. L.; Lanza,
T.; de Laszlo, S.; Kopka, I.; Lin, L.; Chang, L.; Truong,
Q.; Wilson, J.; Li, B.; Kamenecka, T.; Yang, G.; MacCoss,
M.; Mills, S.; Kevin, N. J.; Van Riper, G.; Egger, L.;
Schmidt, J. A.; Lyons, K.; Vincent, S.; Tong, S. Bioorg.
Med. Chem. Lett. 2001, 11, 2709.
5. Plasma levels of the corresponding acyclic acids were not
determined. Although acyl glucuronidation of some acids
was observed in vivo, this was not the major contributor
to the high plasma clearance rates.
6. The a-methyl group in 8 was initially installed to prevent
amide hydrolysis, but compounds lacking this steric
protection suffered relatively little hydrolysis anyways.
7. (a) Ortiz de Montellano, P. R.; Mico, B. A.; Mathews, J.
M.; Kunze, K. L.; Miwa, G. T.; Lu, A. Y. H. Arch.
Biochem. Biophys. 1981, 210(2), 717; (b) Mico, B. A.;
Federowitz, D. A.; Ripple, M. G.; Kerns, W. Biochem.
Pharmacol. 1988, 37(13), 2515.
8. For a review of P-glycoprotein and mrp-2 (multidrug
resistance associated protein-2), see: Van Zuylen, L.;
Nooter, K.; Sparreboom, A.; Verweij, J. Invest. New
Drugs 2000, 18(3), 205–220.
However, dosing 8 along with cyclosporin (a known
inhibitor of CYP3A, P-glycoprotein, and mrp-2)8 led to
a 5-fold drop in clearance rate as well as a 5-fold
increase in AUC.9 These results suggested that an active
transport mechanism may be responsible for the high
clearance rate of 8 and compounds in this class. In a
parallel study using a mutant strain of Wistar trans-
porter deficient rats (TRꢀ rats are deficient in the
organic ion transporter mrp-2) with 8, a similar 5-fold
drop in clearance and 5-fold increase in AUC was
observed supporting the cyclosporin results.10 Such
dramatic changes in pharmacokinetics were not only
observed for 8, but for other amides as well. Interest-
ingly, the pharmacokinetics of amidine 7a in Wistar rats
(Clp ¼ 49 mL/min/kg) were no different from that in TRꢀ
rats (Clp ¼ 58 mL/min/kg) implying that the organic ion
transporter mrp-2 may not be responsible for the
clearance of these amidines in rats.11 These results may
prove useful in the development of a compound with
improved and acceptable pharmacokinetics.
9. (a) Gutmann, H.; Fricker, G.; Torok, M.; Michael, S.;
Beglinger, C.; Drewe, J. Pharm. Res. 1999, 16(3), 402; (b)
Gutmann, H.; Fricker, G.; Drewe, J.; Toeroek, M.; Miller,
D . S.Mol. Pharmacol. 1999, 56(2), 383.
10. Paulusma, C. C.; Elferink, R. P. O. J. Mol. Med. 1997,
75(6), 420.
In summary, we have successfully identified potent and
specific non-amide containing VLA-4 antagonists.
While ester and amine substitutions failed, amidines
were equipotent if not more so than their corresponding
amides. A variety of substitution patterns are acceptable
allowing one to fine tune physical properties such as
11. Elferink, R. P. O.; Jansen, P. L. Pharmacol. Ther. 1994,
64(1), 77.