1,4-Diazepane-2-ones as novel inhibitors of LFA-1

Article abstract of DOI:10.1016/S0960-894X(02)00991-5

The design, synthesis, and biological evaluation of 1,4-diazepane-2-ones as novel LFA-1 antagonists from a scaffold-based combinatorial library are described. Initial optimization of the library lead has resulted in high-affinity antagonists of the LFA-1/ICAM-1 interaction, such as compounds 18d and 18e with IC₅₀ values of 110 and 70 nM, respectively.

Full text of DOI:10.1016/S0960-894X(02)00991-5

Bioorganic & Medicinal Chemistry Letters 13 (2003) 499–502
1,4-Diazepane-2-ones as Novel Inhibitors of LFA-1
Sompong Wattanasin,a,* Rainer Albert,b Claus Ehrhardt,b Didier Roche,a
Michael Sabio,a Ulrich Hommel,b Karl Welzenbachb and Gabriele Weitz-Schmidtb
aNovartis Institute for Biomedical Research, Novartis Pharmaceuticals Corporation,
556 Morris Avenue, Summit, NJ 07901, USA
bNovartis Pharma A.G., Preclinical Research, Basel, Switzerland
Received 9 September 2002; accepted 8 October 2002
Abstract—The design, synthesis, and biological evaluation of 1,4-diazepane-2-ones as novel LFA-1 antagonists from a scaffold-
based combinatorial library are described. Initial optimization of the library lead has resulted in high-affinity antagonists of the
LFA-1/ICAM-1 interaction, such as compounds 18d and 18e with IC50 values of 110 and 70 nM, respectively.
# 2002 Elsevier Science Ltd. All rights reserved.
Antagonism of cell surface receptors belonging to the
integrin superfamily is a new paradigm for drug dis-
covery in transplantation and cancer, as well as in other
diseases.1 Lymphocyte function-associated antigen-1
(LFA-1, CD11a/CD18, aLb2) is a heterodimeric adhe-
sion receptor belonging to the b2 integrin family. LFA-1
is expressed on all leukocytes. Similar to other integrins,
LFA-1 must undergo affinity and/or avidity changes to
bind to its ligands of the immunoglobulin superfamily
ICAM-1 (CD54), -2 (CD102), -3 (CD50), -4, and -5.
Together with other adhesion receptors, LFA-1 med-
iates leukocyte migration to sites of inflammation and
antigen exposure.2À4 LFA-1 is involved in both firm
adhesion and locomotion of leukocytes.5 Moreover,
during an immune response, LFA-1 binding to ICAM-1
can enhance T-cell-receptor-dependent activation and
proliferation of T-cells.6 Recent studies employing
LFA-1 deficient mice or cell-based assay systems show a
requirement for LFA-1 in CD8+ T-cell activation and
effector function.7À9 LFA-1 is an attractive therapeutic
target: blockade of integrin LFA-1 by monoclonal anti-
bodies (mAbs) has shown efficacy in animal models of
inflammation and autoimmune disease, for example,
arthritis,10 ischemia/reperfusion injury11 and transplant
rejection.12 Clinical studies suggest that anti-LFA-1
therapy is beneficial in bone marrow and solid organ
transplantation.13,14 Moreover, recent data show that a
humanized anti-LFA-1 mAb is efficacious in patients
suffering from moderate to severe plaque psoriasis.15b
Because of the importance of this integrin and its
implication in various disease processes, efforts toward
the design and synthesis of orally bioavailable low
molecular weight (LMW) inhibitors have intensified in
recent years.15
We recently reported that lovastatin (mevinolin), a
known HMG-CoA (3-hydroxy-3-methylglutaryl coen-
zyme A) reductase inhibitor, also blocks LFA-1 func-
tion,16a and we have shown by NMR spectroscopy and
X-ray crystallography that lovastatin binds to a unique
site on the I-domain (inserted domain) of the LFA-1 a
chain (Fig. 1). This lovastatin-binding site (L-site)16b is
distant from the metal ion binding domain called metal
ion dependent adhesion site (MIDAS) and is situated at
the proposed I-domain allosteric site (IDAS).16a,17 Dis-
covery of this small molecule-binding pocket within the
I-domain of LFA-1 thus provides an opportunity for
structure-based design of LFA-1 antagonists, which
prompted us to employ
a combinatorial library
approach in identifying novel antagonists using small
molecule scaffolds. 1,4-Diazepane-2-one 1 (Fig. 2) is an
interesting, nonplanar library scaffold that has many
potential sites of diversity with which we selected to
explore interactions within the L-site pocket of the
LFA-1 I-domain. Herein, we report the discovery of
potent and selective LFA-1 antagonists from this diaze-
pane-based library.
The 1,4-diazepane-2-one library was generated on Rink-
amide MBHA resin (Scheme 1). Coupling of the Rink
*Corresponding author. Fax: +1-908-277-4885; e-mail: sompong.
wattanasin@pharma.novartis.com
0960-894X/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00991-5

500
S. Wattanasin et al. / Bioorg. Med. Chem. Lett. 13 (2003) 499–502
Figure 1. X-ray structure of the I-domain of LFA-1 with lovastatin.
Figure 2. 1,4-Diazepane-2-one scaffold, showing three sites of diver-
sity, for creating the combinatorial library of 90 compounds.
resin with amino acids using standard HOBt/DICD
conditions gave the polymer-supported amino acids 2.
The completion of the reaction can be monitored by
using the bromophenol blue test or by titration of the
FMOC residue. The aminoacids 2 were then depro-
tected with 20% piperidine in DMF and treated with a
0.5 M methyl vinylketone solution in DMF or CH2Cl2
to give 3. The conditions are applicable to a variety of
substituents including BOC-protected tryptophan. It is
interesting to note that even though a large excess of
methyl vinylketone was used (12–20 equiv), no Michael
di-adduct was detected after TFA cleavage from the
resin. Acylation of 3 with activated amino acids 4 using
the above HOBt/DICD/DMF conditions gave unsa-
tisfactory yields of 5. However, better yields could be
obtained by using a 0.5 M mixture of resin/aminoacid/
DICD (1:10:5) in DMF. After FMOC deprotection of
5, the diazepine ring cyclization was achieved by the
reductive amination (NaBH3CN in DMF buffered with
1% acetic acid). The resulting cyclic amines 6 were then
acylated with acids 7 using DICD in DMF. TFA clea-
vage in CH2Cl2 gave diazepane-2-ones 1.18 Only a single
diastereomer of 1, in which the methyl group on the 5-
position of the diazepane ring is in a syn relationship
with the substituent on the 3-position (i.e., the R2
group), was detected.19 It is conceivable that the reduc-
tive amination is selectively controlled by the chiral
center of the R2 group present in the ring.20 Minor
amounts of the corresponding acids 9 were also
observed.
protecting group, a 1,4-addition of the amino inter-
mediate to methyl vinylketone leads to 12. Coupling of
12 with an N-BOC protected amino acid gives 13, after
silica gel purification. Deprotection and intramolecular
reductive amination of 14 affords the seven-membered
ring intermediate 15. Acylation with carboxylic acid 16
leads to the desired diazepane 1. The overall yield of this
unoptimized sequence is in the range of 5–10%.
The design of the library was based on an idea to use the
diazepane ring as a scaffold to which various interacting
moieties could be attached. One of the R groups should
correspond to the decalin system of lovastatin and fill
most of the large hydrophobic pocket within the
L-site.21 For an exploration of possible substituents for
the R1, R2, and R3 groups on the diazepane template,
docking studies based on the X-ray crystal structure of
the LFA-1 I-domain in complex with lovastatin were
performed.22 It became obvious that the diazepane
scaffold would serve very well as a template, making
only a few contacts with the protein, while the binding
affinity would have to arise from interactions involving
appropriate hydrophobic groups in R1, R2, and R3.
Information from superposition analyses utilizing other
weak hits from HTS indicated that large hydrophobic
substituents on the R3 position might be necessary.23
Molecules with large hydrophobic groups in R3 were
then docked manually into the binding site and the
corresponding complexes were geometry optimized.24
Figure 3a shows such a structure in which the naphtha-
lenyl-methyl moiety in R3 points deeply into the binding
site, occupying an area close to that of the decalin sys-
tem in lovastatin. This binding mode would suggest
The solution phase synthesis was also developed for
selected 1,4-diazepane-2-ones (Scheme 2). The synthesis
starts with the conversion of BOC-amino acid 10 into
the corresponding amide 11. After removal of the BOC
Scheme 1. (a) FMOC-amino acid (6 equiv), HOBt (6 equiv), DICD (6 equiv), DIPEA (1 equiv), DMF, rt, 12 h; (b) 20% piperidine, DMF, rt, 20 min;
(c) 0.5 M methyl vinylketone, DMF, rt, 12 h; (d) 4 (8 equiv), DICD (4 equiv), DMF, rt, 48 h; (e) 0.5 M NaBH3CN, DMF, rt, 12 h; (f) 7 (10 equiv),
DICD (10 equiv), DMF, rt, 12 h; (g) TFA/CH2Cl2 1/1, rt, 60 min; (h) freeze dried in CH3CN/H2O.

S. Wattanasin et al. / Bioorg. Med. Chem. Lett. 13 (2003) 499–502
501
Scheme 2. Solution phase synthesis of 1,4-diazepane-2-ones: (a) DCCI (1.2 equiv), HOBt (1.2 equiv), 25% NH4OH (1.2 equiv), DMF, rt, 16 h; (b)
TFA (neat), 0 ꢀC, 15 min, then 3 N HCl/diethylether; (c) methyl vinylketone (1.5 equiv), dioxane, DIPEA (1.2 equiv), rt, 16 h; (d) BOC-amino acid
(1 equiv), DIPC (4 equiv), NMM (4 equiv), DMF, rt, 16 h; (e) sodium cyanoborohydride (2 equiv), dioxane–water (4:1), pH 5.4, 30 min; (f) DIPEA
(1 equiv), EADC (2 equiv), CH2Cl2, rt, 16 h.
relatively small hydrophobic groups in the R2 position,
but no conclusion could be drawn about the R1 group,
because it points out of the L-site. An alternative bind-
ing hypothesis achieved by additional docking (Fig. 3b)
revealed the possibility that the R1 naphthyl group of
this diazepane template and the decalin moiety of
lovastatin could occupy the same binding pocket in the
L-site and provided information about other sub-
stituents on this scaffold. Because we intended to cover
a broad range of substituents in our combinatorial
library to gain insights into binding requirements of this
scaffold, no attempts were made to determine the opti-
mal combination of substituent groups or to evaluate
other binding conformations in greater detail. The
above considerations led to the substituent selection on
the R1, R2, and R3 positions of the diazepane template
as shown in Figure 2 for the first library consisting of 90
compounds.
high throughput chemistry prompted us to use 17a as a
lead for further optimization. NMR investigations indi-
cated that 17a binds to the L-site of the LFA-1
I-domain25 suggesting that either the R1 or R2 naphthyl
group of 17a might occupy the binding site used by the
decalin portion of lovastatin. However, NMR analysis
of 17a was limited due to the presence of two almost
identical residues (i.e., two 2-naphthyl groups) on its R1
and R3 positions and suggested the need to differentiate
these two groups. The SAR trend from the above
library (Fig. 4) suggested the requirement of a large
hydrophobic group on the R1 position and a medium-
sized group on the R2 position in this series of com-
pounds. The preference for the (S)-isobutyl over the
(R)-isobutyl group at the R2 position was also indicated.
No information, however, was available for smaller R3
groups. Accordingly, our initial optimization effort was
focused primarily on reducing the size and lipophilicity
of the R3 group in 17a (Table 1). The total loss of
potency of compound 15a (Scheme 2) suggests the
importance of an acyl group in this R3 position.
Initial screening of the crude products in the 90-com-
pound diazepane library (Fig. 4) using an LFA-1/
ICAM-1 ELISA-type binding assay16b identified 17a (1;
R1, R3=e, a; R2=b) as the most active compound.
Compound 17a and its carboxylic acid 17b could be
purified directly from the crude library sample. 17a was
also independently synthesized by solution-phase chem-
istry (Scheme 2) and was shown to have an IC50 value
(2 mM) comparable to that of lovastatin in the cell-free
LFA-1/ICAM-1 adhesion assay. Its corresponding acid
17b is much less active (IC50=44 mM). Despite its rela-
tively weak activity, the structural novelty and available
Comparable activities were observed for 17a and the
4-bromophenyl (18b) replacement. A slight improve-
ment in potency was seen for the phenyl (18c) and 4-
pyridyl (18a) derivatives. A significant improvement in
potency, however, was observed for the quinoline deri-
vatives. Both 6-quinolyl (18d) and 3-quinolyl (18e)
Figure 3. Superposition of representative diazepane-2-ones [structure
1; R1=R2=Me in panel (a); R2=R3=Me in panel (b)] and lovastatin
in the L-site of LFA-1. Compound 1 is drawn in green. Lovastatin is
drawn in magenta.
Figure 4. The 90-compound diazepane library screened against the
LFA-1/ICAM-1 assay. Each compound was screened at 12 mM.

502
S. Wattanasin et al. / Bioorg. Med. Chem. Lett. 13 (2003) 499–502
Table 1. In vitro cell-free LFA-1 assay (LFA-1/ICAM-1 binding
assay)16b
7. Krensky, A. M.; Sanchez-Madrid, F.; Robbins, E.; Nagy,
J. A.; Springer, T. A.; Burakoff, S. J. J. Immunol. 1983, 131,
611.
8. Schmits, R.; Kundig, T. M.; Baker, D. M.; Shumaker, G.;
Simard, J. J.; Duncan, G.; Wakeham, A.; Shahinian, A.; van
der Heiden, A.; Bachmann, M. F.; Ohashi, P. S.; Makt, T. W.;
Hickstein, D. D. J. Exp. Med. 1996, 183, 1415.
9. Shier, P.; Ngo, K.; Fung-Leung, W. P. J. Immunol. 1999,
163, 4826.
10. Issekutz, A. C. Inflamm. Res. 1998, 47, S123.
11. Qubenaissa, A. O.; Mouas, C.; Bourgeous, F.; Le Deist,
F.; Alberici, G.; Moalic, J.-M.; Menasche, P. Circulation 1996,
9, II.
12. Nakakura, E. K.; Shorthouse, R. A.; Zheng, B.; McCabe,
S. M.; Jardieu, P. M.; Morris, R. E. Transplantation 1996, 62,
547.
13. Fischer, A.; Landais, P.; Friedrich, W.; Gerritsen, B.;
Fasth, A.; Porta, F.; Vellodi, A.; Benkerrou, M.; Jais, J. P.;
Cavazzana-Calvo, M.; Souillet, G.; Bordigoni, P.; Morgan,
G.; Van Dijken, P.; Vossen, J.; Locatelli, F.; di Bartolomeo, P.
Blood 1994, 83, 1149.
Compd
Substituenta
IC50 (mM)b
15a
17a
17b
18a
18b
18c
18d
18e
18f
R1=(S)-2-naphthyl; R2=(S)-i-Bu
R4=NH2
R4=OH
R3=4-pyridyl; R4=NH2
R3=4-bromophenyl; R4=NH2
R3=phenyl; R4=NH2
R3=6-quinolyl; R4=NH2
R3=3-quinolyl; R4=NH2
R3=6-quinolyl; R4=OMe
>20
2
>20
0.97
1.30
0.75
0.11
0.07
2.60
aAll compounds have been synthesized according to Scheme 2.
bValues are means of three experiments.
14. Hourmant, M.; Bedrossian, J.; Durand, D.; Lebranchu,
Y.; Renoult, E.; Caudrelier, P.; Buffet, R.; Soulillou, J.-P.
Transplantation 1996, 62, 1565.
15. (a) For recent reviews: Liu, G. Exp. Opin. Ther. Patents
2001, 11, 1383. (b) Liu, G. Drugs Future 2001, 26, 767.
16. (a) Kallen, J.; Welzenbach, K.; Ramage, P.; Geyl, D.;
Kriwacki, R.; Legge, G.; Cottens, S.; Weitz-Schmidt, G.;
Hommel, U. J. Mol. Biol. 1999, 292, 1. (b) Weitz-Schmidt, G.;
Welzenbach, K.; Brinkmann, V.; Kamata, T.; Kallen, J.;
Bruns, C.; Cottens, S.; Takada, Y.; Hommel, U. Nat. Med.
2001, 7, 687.
17. Huth, J. R.; Olejniczak, E. T.; Mendoza, R.; Liang, H.;
Harris, E. A.; Lupher, M. L., Jr.; Wilson, A. E.; Fesik,
S. W.; Staunton, D. E. Proc. Natl. Acad. Sci. U.S.A. 2000,
97, 5231.
18. The crude products 1 were shown to be in the range of 14–
72% purity by HPLC.
19. NOESY study clearly showed the presence of an NOE
effect between H3 and H5, suggesting a cis relationship
between these two protons.
derivatives showed more than an 18- and 28-fold
increase, respectively, in potency in comparison to that
of the library lead 17a. These binding improvements
suggested that the nitrogen atom in the quinoline ring of
18d and 18e might be involved in hydrogen-bond inter-
actions within the L-site pocket of the LFA-1 I-domain.
NMR studies confirm that 18d binds to the L-site
pocket. However, specific residues that could be
involved in hydrogen-bond formation were not revealed
and await further I-domain-ligand crystallographic evi-
dence. The methyl ester (18f) of 18d is less active.
In summary, we have discovered diazepanes as novel
antagonists of LFA-1 by a combinatorial library
approach. Preliminary optimization of the initial library
lead has resulted in high-affinity antagonists of the
LFA-1/ICAM-1 interaction, such as compounds 18d
and 18e with IC50 values of 110 and 70 nM, respectively.
Further exploration of this novel series is in progress
and will be reported in due course.
20. Compounds without the R2 substituents show a 1:1 mix-
ture of two diastereomers. These stereochemical results sup-
port the selectivity induction by the chiral center in the
diazepane ring: Wattanasin, S.; Roche, D. Novartis; unpub-
lished results.
21. Hydrophobic interactions between lovastatin’s decalin
ring system and the I-domain seem to be the major binding
determinants. For specific interactions of various moieties of
lovastatin with the protein residues in the I-domain/lovastatin
complex, please see ref 16a.
References and Notes
1. (a) For recent reviews: Heavner, G. A. Drug Discov. Today
1996, 1, 295. (b) Mousa, S. A.; Cheresh, D. A. Drug Discov.
Today 1997, 2, 187. (c) Bochner, B. S. Cell Adhes. Mol. Matrix
Proteins 1998, 113. (d) Elices, M. J. Curr. Opin. Anti-Inflam-
mat. Imunomodulat. Invest. Drugs 1999, 1, 14.
2. Gahmberg, C. G.; Valmu, L.; Fagerholm, S.; Kotovuori,
P.; Ihanus, E.; Tian, L.; Pessa-Morikawa, T. Cell. Mol. Life
Sci. 1998, 54, 549.
3. Carlos, T. M.; Harlan, J. M. Blood 1994, 84, 2068.
4. Hughes, P. E.; Pfaff, M. Trends Cell Biol. 1998, 8, 359.
5. Dustin, M. L.; Springer, T. A. Annu. Rev. Immunol. 1991,
9, 27.
6. Ni, H.-T.; Deeths, M. J.; Li, W.; Mueller, D. L.; Mescher,
M. F. J. Immunol. 1999, 162, 5183.
22. Lovastatin was removed from the structure and the dia-
zepane system was docked manually into the L-site using the
in-house program WitnotP written by: Widmer, A. Novartis;
unpublished results.
23. The HTS hits are Novartis proprietary structures.
24. (a) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.;
States, D. J.; Swaminathan, S.; Karplus, M. J. Comput.
Chem. 1983, 4, 187. (b) Accelrys (formerly Molecular Simu-
lations Inc.), 9685 Scranton Road, San Diego, CA 92121,
USA.
25. Binding of the compounds to the LFA-1 L-site was con-
firmed by the chemical shift changes shown by the 15N-HSQC
NMR spectra of the isolated LFA-1 I-domain described,
please see ref 16a.