Rational design of non-nucleoside, potent, and orally bioavailable adenosine deaminase inhibitors: Predicting enzyme conformational change and metabolism

Article abstract of DOI:10.1021/jm050413g

From metabolic considerations and prediction of an inhibitor-induced conformational change, novel adenosine deaminase (ADA) inhibitors with improved activities and oral bioavailability have been developed on the basis of our originally designed non-nucleo

Full text of DOI:10.1021/jm050413g

4750
J. Med. Chem. 2005, 48, 4750-4753
Although a number of ADA inhibitors are known (e.g.,
pentostatin,6 (+)-EHNA,7 and various derivatives8), they
have many problems such as poor pharmacokinetics9
and several toxicities.10 Pentostatin, which is the only
ADA inhibitor in clinical use, is only available via
intravenous administration,11 and moreover, use is
limited to the treatment of adult patients with hairy
cell leukemia. Therefore, we initiated a search for ADA
inhibitors with reduced toxicity and oral bioavailability
by changing the nucleoside framework of the known
inhibitors to a non-nucleoside framework. A recent
report from our laboratories described the discovery of
a good lead compound 1 (IC50 ) 570 nM, human ADA)
for optimization because of its simple, non-nucleoside
framework.12 Rational lead hybridization of 1 with the
structurally distinct random screening hit 2 (IC50 ) 700
nM)12 and further efforts using a structure-based ap-
proach resulted in the discovery of the novel non-
nucleoside ADA inhibitor 3 (Figure 1).13 Although it had
a high potency in terms of ADA inhibitory activity, the
Rational Design of Non-Nucleoside,
Potent, and Orally Bioavailable
Adenosine Deaminase Inhibitors:
Predicting Enzyme Conformational
Change and Metabolism
Tadashi Terasaka,*,† Kiyoshi Tsuji,† Takeshi Kato,†
Isao Nakanishi,‡,# Takayoshi Kinoshita,‡
Yasuko Kato,§ Masako Kuno,§ Takeshi Inoue,§
Kohichiro Tanaka,| and Katsuya Nakamura†
Medicinal Chemistry Research Laboratories, Basic Research
Laboratories, Medicinal Biology Research Laboratories, and
Biopharmaceutical and Pharmacokinetic Research
Laboratories, Fujisawa Pharmaceutical Co., Ltd.,
2-1-6, Kashima, Yodogawa-ku, Osaka 532-8514, Japan
Received May 2, 2005
Abstract: From metabolic considerations and prediction of
an inhibitor-induced conformational change, novel adenosine
deaminase (ADA) inhibitors with improved activities and oral
bioavailability have been developed on the basis of our
originally designed non-nucleoside ADA inhibitors. They dem-
onstrated in vivo efficacy in models of inflammation and
lymphoma. Furthermore, X-ray crystal structure analysis has
revealed a novel induced fit to ADA.
oral bioavailability was too poor for clinical use (IC50
)
6.7 nM, bioavailability BA ) 10%). On the other hand,
we also discovered potent and orally bioavailable ADA
inhibitors 4 (IC50 ) 16 nM, BA ) 30%) and 5 (IC50 ) 15
nM, BA ) 44%) by a structure-based and metabolism-
directed design approach based on 1.14 Furthermore, 5
demonstrated in vivo efficacy in models of inflammation
and lymphoma after oral administration (po) for the first
time. However, it was still not sufficient in terms of
inhibitory activity and pharmacokinetics because 5 at
32 mg/kg po twice daily dosing (b.i.d.) was needed to
show the same in vivo antitumor activity in a SCID
mouse model15 as intraperitoneal administration (ip) of
pentostatin at 2.5 mg/kg once daily dosing (u.i.d.).
Therefore, we attempted further design of more im-
proved ADA inhibitors based on structure-activity
relationships (SAR) and structural information of in-
hibitor/ADA complexes obtained in our original studies.
In this communication, we report the discovery of
improved potent and orally bioavailable ADA inhibitors
by a predicted induced fit and metabolism-directed
design approach and disclose a novel binding mode for
ADA by a conformational change at the active site.
Adenosine is an endogenous purine nucleoside re-
leased by cells and has a wide variety of biological
activities such as immunosuppressive properties, pro-
tective effects in cerebral and myocardial ischemia, and
inhibitory effects in lymphoproliferation. In recent
years, adenosine has also been considered as an impor-
tant factor in the attenuation of inflammation,1,2 which
is signaled through the A2a adenosine receptor. Ad-
enosine deaminase (ADA) (EC 3.5.4.4) is a ubiquitous
purine metabolic enzyme that catalyzes the irreversible
deamination of adenosine and 2′-deoxyadenosine to
inosine and 2′-deoxyinosine, respectively.3 ADA exists
as cytosolic and extracellular forms and has an impor-
tant role in regulating intra- and extracellular adenos-
ine concentrations. It is well-known that ADA deficiency
results in severe combined immunodeficiency by ac-
cumulation of lymphotoxic adenosine and 2′-deoxyad-
enosine.4 From a pharmacology viewpoint, ADA inhi-
bition has interest as potential therapy of malignant
leukemia and lymphomas. Furthermore, it is considered
that ADA inhibiton has great potential for anti-inflam-
matory drugs with few side effects by preventing ad-
enosine released specifically at inflamed sites from
metabolism by extracellular ADA, colocalized on the cell
surface with CD26, which is strongly up-regulated
following T-cell activation and is known as the T-cell
activation marker. 5
Compound 5 was designed on the basis of the crystal
structure complexed with ADA and metabolic study of
4 to retain high inhibitory activity and to improve
pharmacokinetics. That is, according to the crystal
structure of the 4/ADA complex, the enzyme had
undergone conformational change, the same as in the
2/ADA complex, and formed binding pockets (F1 and
F2) that were spatially distinct from the pockets utilized
by substrate-like inhibitors (Figure 3a-c).16 In consid-
eration of the narrow planar hydrophobic space avail-
able at F1 and protection from metabolism, a 2,3-
dichlorophenyl ring was introduced in place of the
naphthyl ring of 4.14 The resulting 5 showed potent
inhibitory activity similar to that of 4 (Figure 2, Table
1). Although the naphthyl and 2,3-dichlorophenyl rings
were bound to the F1 space, these moieties are located
at the entrance position of the enzyme and do not utilize
the F1 space sufficiently. Moreover, the hydrophobic
* To whom correspondence should be addressed. Phone: +81-6-
6390-1286. Fax: +81-6-6304-5414. E-mail: tadashi.terasaka@
jp.astellas.com.
† Medicinal Chemistry Research Laboratories.
‡ Basic Research Laboratories.
# Current address: Department of Theoretical Drug Design, Gradu-
ate School of Pharmaceutical Sciences, Kyoto University, Sakyo-ku,
Kyoto 606-8501, Japan.
§ Medicinal Biology Research Laboratories.
| Biopharmaceutical and Pharmacokinetic Research Laboratories.
10.1021/jm050413g CCC: $30.25 © 2005 American Chemical Society
Published on Web 06/24/2005

Letters
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 4751
(Ala183-Ile188) and two leucine side chains (Leu58 and
62) from an R-helix (Thr57-Ala73). It was considered
that the lid was closed to accommodate the substrates.
On the other hand, ADA performed an induced fit to
bind our inhibitors by opening the lid and formed the
hydrophobic binding pockets (F1 and F2).12-14 There-
fore, the shape of the F1 pocket is somewhat changeable
depending on the inhibitor structure. In our previously
disclosed crystal structures of inhibitor/ADA complexes,
although the wall formed by Leu62 and the loop in the
substrate-like inhibitor/ADA complex was opened, the
inner wall of the F1 pocket is still formed by Leu 58
and Asp 185. However, Asp 185 is flexible because it is
placed in the loop region of the enzyme. Moreover, it
was considered that the interaction between Leu 58 and
Asp 185 was weak (Figure 4).
Figure 1. Chemical structures of adenosine and known ADA
inhibitors.
Therefore, we considered that these residues can be
pushed out in order to bind a rigid structure such as 6
and 7. As a result, they may be able to not only
accommodate in the enzyme active site but also use
further hydrophobic space for binding. Thus, 6 and 7
were actually synthesized and evaluated. Consequently,
they showed 2-fold more potent inhibitory activities
than 4 and 5 (Table 1). A crystal structure of the 7/ADA
complex verified our prediction about inhibitor-induced
change of the binding site (Figure 3e,f). That is, the
benzoxazolyl ring was bound to the F1 space, the same
as the naphthyl ring of 4, and the inner wall of the F1
pocket expanded outside the active site to accommodate
the p-chlorophenyl ring of 7. Moreover, interestingly,
the F2 space was diminished compared to that of our
previously designed inhibitor/ADA complexes. Despite
the conformational changes, the other parts showed the
same binding mode as 4. This binding mode was
previously unknown and is spatially distinct from that
of substrate-like inhibitors (Figure 3a) and our previ-
ously reported inhibitors (Figure 3b,c).
Because the primary objective of this study was
discovery of ADA inhibitors with superior pharmacoki-
netics compared to 5, compounds 6 and 7 with higher
potency than 4 and 5 were subjected to pharmacokinetic
experiments (Table 1). They displayed improved and/
or the same metabolic stability to in vitro clearance in
liver microsomes compared to 4 and 5. Moreover, 6 and
7 had drastically improved pharmacokinetics and ex-
ceeded 5 in most pharmacokinetic parameters, as shown
in Table 1. Upon dosing of 6 (10 mg/kg) to rats, drastic
improvements in maximum plasma concentration (Cmax
) 14.72 µg/mL) and area under the blood concentra-
tion-time curve (AUC ) 56.23 µg‚h/mL) were observed.
Despite the same metabolic stability to in vitro clearance
as 5, dosing of 7 (10 mg/kg) to rats showed a substantial
improvement of Cmax (5.58 µg/mL) and plasma half-life
(t1/2 ) 3.45 h) and led to a drastic improvement of AUC
(29.56 µg‚h/mL) (BA ) 52%). In particular, 6 displays
excellent oral bioavailability in rats and dogs (BA ) 86%
and 81%, respectively).
Figure 2. Design process for discovery of 6 and 7.
space of F2, which is occupied by the benzimidazole ring
of 2, was also not utilized in the binding of 4 and 5 (parts
b and c of Figure 3). It was considered that utilizing
these hydrophobic spaces contributed to the enhance-
ment of the inhibitory activities by gaining further van
der Waals or hydrophobic interactions. However, we
have also experienced from a previous study that
utilizing the F2 space contributes to enhancement of the
inhibitory activities and also resulted in increased
molecular weight and disadvantages for pharmacoki-
netic improvement.12,13 For metabolic stability, it is
necessary to remove the easily metabolized parts from
the designed compounds.14 It was considered that utiliz-
ing F1 hydrophobic space in binding with ADA is
effective. Therefore, on the basis of our original infor-
mation, we next planned to make the best use of the
narrow planar hydrophobic F1 space and block metabo-
lism of the designed compounds to enhance activity and
improve pharmacokinetics. As a result, 6 and 7 were
designed, as shown in Figure 2. That is, in consideration
of the narrow planar hydrophobic space available at F1
and protection of the designed compound from metabo-
lism (oxidation of the naphthyl ring was predicted from
the metabolic study of 4), the naphthyl ring of 4 was
replaced with a benzoxazolyl ring. Furthermore, the
2-position of the benzoxazolyl ring was substituted with
a p-chlorophenyl ring directly because introduction of
a halogen atom at the para position of the phenyl ring
can protect from metabolism.17 According to replace-
ment of the naphthyl ring of 4 with 2-(4-chlorophenyl)-
benzoxazolyl rings of the designed compounds based on
the 4/ADA complex, they could not be accommodated
to the binding site because they have a rigid and planar
structure at the 2-(4-chlorophenyl)benzoxazolyl moiety
and thus the p-chlorophenyl ring clashes with the inner
wall of F1 pocket (Figure 3d). However, we predicted
from previously obtained information of crystal struc-
ture analyses of ADA complexes with our inhibitors that
the enzyme can change the shape of the F1 pocket. In
the substrate-like inhibitor/ADA complex, the active site
is perfectly enclosed by the lid consisting of the loop
The in vivo antitumor activity of our best compound
6 was next evaluated in a SCID mouse lymphoma
model15 (Table 2). Oral administration of 6 with the
antitumor 9-â-D-arabinofuranosyladenine (AraA) at 10
and 32 mg/kg u.i.d. resulted in prolonged survival with
a median survival time (MST) of 43.0 and 45.5 days,
respectively. The mice not only survived significantly
longer than control mice (MST of 22.0 days) but was

4752 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15
Letters
Figure 3. Binding mode of inhibitors at the ADA active site. Accessible surfaces of the carbon atoms at the active sites are
drawn by mesh. The upper portions of (a), (b), (d), and (e) are the active site entrance (solvent region), and the other nonmeshed
black regions are occupied by protein. (a) Binding mode of 1-deazaadenosine to the active site of the substrate analogue/ADA
complex. (b) Binding orientations of 2 and 4 superimposed onto the active site surface of the 4/ADA complex. (c) View from the
active site entrance of (b). (d) Binding orientations of 7 and 4 superimposed onto the active site surface of the 4/ADA complex. (e)
Binding orientation of 7 and 4 superimposed onto the active site surface of the 7/ADA complex. (f) View from the active site
entrance of (e).
Table 1. In Vitro Clearance with Liver Microsomes of Rat, Dog, and Human, and Pharmacokinetic Parameters of 3-7 after po
Administration to Rats and Dogs (10 mg/kg, n ) 2-3)
b
in vitro clearance (mL min-1 kg-1
)
pharmacokinetics parameters in ratsb
BA (%)c
compd
IC50 (nM)a
rat
dog
human
Cmax (µg/mL)
t1/2 (h)
AUC0-24h (µg‚h/mL)
rat
dog
3
4
6.7
16
643
307
111
50
ndd
ndd
0.21
1.25
1.63
14.72
5.58
1.09
0.66
1.66
1.87
3.45
0.26
10
30
44
86
52
ndd
38
43
81
50
30
45
1.37
5
15
21
6.2
6.0
4.5
12
34
2.11
56.23
29.56
6e
7
8.7
8.6
131
a IC50 values were measured with human ADA. Assays were performed in duplicate. b See Supporting Information. c BA ) bioavailability.
d nd ) not determined. e HCl salt.
better than 5 because 6 at 10 mg/kg po u.i.d. exhibited
antitumor activity against lymphoma in combination
with AraA comparable to that observed for 5 at 32 mg/
kg po b.i.d. and intraperitoneal administration of pen-
tostatin at 2.5 mg/kg.
Furthermore, 6 and 7 were evaluated in an in vivo
model of inflammation. In the adjuvant arthritis model
in rats, both showed anti-inflammatory effects orally
(ED30 ) 0.3 and 0.6 mg/kg po, respectively), and these
efficacies were improved compared to that of 5 (ED30
)
1.6 mg/kg po).
The syntheses of 6 and 7 are outlined in Scheme 1.
The Wittig-Horner reaction between phosphonate 818
and the suitable arylaldehyde gave pentenone 9. Ste-
reoselective reduction of the ketone moiety was ac-
complished with L-selectride in THF, followed by cata-
lytic hydrogenation with palladium on carbon (Pd-C)
in EtOAc to afford the corresponding secondary alcohols
11. 12 was prepared by procedures similar to that of 4
in our recent report.14 6 and 7 were obtained by removal
of the tert-butyldimethylsilyl (TBS) group of the corre-
sponding 12 with tetrabutylammonium fluoride (TBAF).
Figure 4. “Lid” of the active site of the 1/ADA complex (PDB
code, 1NDY). The backbone atoms of residues 50-73 and
residues 180-190 are represented by ribbons, and several
residues consisting of the lid are drawn in stick form. Com-
pound 1 is in ball-and-stick form. Residue numbers of both
ends of the backbones are also indicated.
In summary, we report further efforts starting from
the novel non-nucleoside ADA inhibitor 4 and discovered

Letters
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 15 4753
(4) Resta, R.; Thompson, L. F. SCID: the role of adenosine deami-
nase deficiency. Immunol. Today 1997, 18, 371-374.
Table 2. Comparative Antitumor Activity of Compounds 5 and
6 and Pentostatin in U-937 Lymphoma Model
(5) Kameoka, J.; Tanaka, T.; Nojima, Y.; Schlossman, S. F.; Morim-
oto, C. Direct association of adenosine deaminase with T cell
activation antigen, CD26. Science 1993, 261, 466-469.
(6) Agerwal, R. P.; Spector, T.; Parks, R. E. Tight-binding inhibitors.
IV. Inhibition of adenosine deaminase by various inhibitors.
Biochem. Pharmacol. 1977, 26, 359-367.
(7) (a) Baker, D. C.; Hanvey, J. C.; Hawkins, L. D.; Murphy, J.
Identification of the bioactive enantiomer of erythro-3-(adenine-
9-yl)-2-nonanol (EHNA), a semi-tight binding inhibitor of ad-
enosine deaminase. Biochem. Pharmacol. 1981, 30, 1159-1160.
(b) Bessodes, M.; Bastian, G.; Abushanab, E.; Panzica, R. P.;
Berman, S. F.; Marcaccio, E. J.; Chen, S. F.; Stoeckler, J. D.;
Parks, R. J. Effect of chirality in erythro-9-(2-hydroxy-3-nonyl)
adenine (EHNA) on adenosine deaminase inhibition. Biochem.
Pharmacol. 1982, 31, 879-882.
drug
dose (mg/kg)b
MST (day)
control
22.0
23.5
34.0
42.5
38.0
43.0
45.5
34.0
42.5
AraA
10
5a (po)
10 (b.i.d.)
32 (b.i.d.)
32 (u.i.d.)
10 (u.i.d.)
32 (u.i.d.)
0.25
6a (po)
pentostatin (ip)
2.5
a HCl salt. b b.i.d. ) twice daily dosing. u.i.d. ) once daily
dosing.
(8) (a) Cristalli, G.; Eleuteri, A.; Franchetti, P.; Grifantini, M.;
Vittori, S.; Lupidi, G. Adenosine deaminase inhibitors: Synthesis
and structure-activity relationships of imidazole analogue of
erythro-9-(2-hydroxy-3-nonyl)adenine. J. Med. Chem. 1991, 34,
1187-1192. (b) Pragnacharyulu, P. V. P.; Varkhedkar, V.;
Curtis, M. A.; Chang, I. F.; Abushanab, E. Adenosine deaminase
inhibitors: Synthesis and biological evaluation of unsaturated,
aromatic, and oxo derivatives of (+)-erythro-9-(2′S-hydroxy-3′R-
nonyl) adenine [(+)-EHNA]. J. Med. Chem. 2000, 43, 4694-4700.
(9) (a) McConnell, W. R.; Furner, R. L.; Hill, D. L. Pharmacokinetics
of 2′-deoxycoformycin in normal and L1210 leukemic mice. Drug
Metab. Dispos. 1979, 7, 11-13. (b) Lathia, C.; Fleming, G. F.;
Meyer, M.; Ratain, M. J.; Whitfield, L. Pentostatin pharmaco-
kinetics and dosing recommendations in patients with mild renal
impairment. Cancer Chemother. Pharmacol. 2002, 50, 121-126.
(c) McConnell, W. R.; El-Dareer, S. M.; Hill, D. L. Metabolism
and disposition of erythro-9-(2-hydroxy-3-nonyl)[14C] adenine in
the rhesus monkey. Drug Metab. Dispos. 1980, 8, 5-7. (d)
Lambe, C. U.; Nelson, D. J. Pharmacokinetics of inhibition of
adenosine deaminase by erythro-9-(2-hydroxy-3-nonyl) adenine
in CBA mice. Biochem. Pharmacol. 1982, 31, 535-539.
(10) Brogden, R. N.; Sorkin, E. M. Pentostatin. A review of its
pharmacodynamic and pharmacokinetic properties, and thera-
peutic potential in lymphoproliferative disorders. Drugs 1993,
46, 652-677.
Scheme 1a
a Reagents and conditions: (a) ArCHO, n-BuLi, THF, 0 °C to
room temp, overnight; (b) L-selectride, THF, -20 °C, 30 min; (c)
H2/Pd-C, EtOAc, room temp, 1 h; (d) MsCl, Et3N, CH2Cl2, 0 °C,
1 h; (e) 4-imidazolecarboxamide, NaH, DMF, 80 °C, 24 h to 3 days;
(f) TBAF, THF, 0 °C, 1 h.
two inhibitors with potent activities and drastically
improved oral bioavailability from metabolic consider-
ations and prediction of an inhibitor-induced conforma-
tional change. Consequently, they also demonstrated
improved in vivo efficacy in models of inflammation and
lymphoma after oral administration. In particular, in
the lymphoma model in mice, oral administration of 6
at 10 mg/kg once daily showed the same antitumor
activity as intraperitoneal administration of pentostatin
at 2.5 mg/kg. Moreover, the X-ray structure of the
7/ADA complex revealed a novel conformation, which
should be useful for the further design of novel ADA
inhibitors.
(11) Rafel, M.; Cervantes, F.; Beltran, J. M.; Zuazu, J.; Nieto, L. H.;
Rayon, C.; Talavera, J. G.; Montserrat, E. Deoxycoformycin in
the treatment of patients with hairy cell leukemia. Cancer 2000,
88, 352-357.
(12) Terasaka, T.; Kinoshita, T.; Kuno, M.; Nakanishi, I. A highly
potent non-nucleoside adenosine deaminase inhibitor: Efficient
drug discovery by intentional lead hybridization. J. Am. Chem.
Soc. 2004, 126, 34-35.
(13) Terasaka, T.; Kinoshita, T.; Kuno, M.; Seki, N.; Tanaka, K.;
Nakanishi, I. Structure-based design, synthesis, and structure-
activity relationship studies of novel non-nucleoside adenosine
deaminase inhibitors. J. Med. Chem. 2004, 47, 3730-3743.
(14) Terasaka, T.; Okumura, H.; Tsuji, K.; Kato, T.; Nakanishi, I.;
Kinoshita, T.; Kato, Y.; Kuno, M.; Seki, N.; Naoe, Y.; Inoue, T.;
Tanaka, K.; Nakamura, K. Structure-based design and synthesis
of non-nucleoside, potent, and orally bioavailable adenosine
deaminase inhibitors. J. Med. Chem. 2004, 47, 2728-2731.
(15) Niitsu, N.; Yamamoto-Yamaguchi, Y.; Kasukabe, T.; Okabe-
Kado, J. Umeda, M.; Honma, Y. Antileukemic efficacy of 2′-
deoxycoformycin in monocytic leukemia cells. Blood 2000, 96,
1512-1516.
(16) (a) Kinoshita, T.; Nishio, N.; Sato, A.; Murata, M. Crystallization
and preliminary analysis of bovine adenosine deaminase. Acta
Crystallogr. 1999, D55, 2031-2032. (b) Kinoshita, T.; Nishio,
N.; Nakanishi, I.; Sato, A.; Fujii, T. Crystal structure of bovine
adenosine deaminase complexed with 6-hydroxyl-1,6-dihydro-
purine riboside. Acta Crystallogr. 2003, D59, 299-303.
(17) Smith, D. A.; Jones, B. C.; Walker, D. K. Design of drugs
involving the concepts and theories of drug metabolism and
pharmacokinetics. Med. Res. Rev. 1996, 16, 243-266.
(18) Shapiro, G.; Buechler, D.; Hennet, S. Synthesis of enantiomeri-
cally pure muscarine analogs. Tetrahedron Lett. 1990, 31,
5733-5736.
Acknowledgment. The authors are grateful to The
Chemo-Sero-Therapeutic Research Institute for the gift
of pentostatin. We thank Dr. David Barrett for valuable
comments and help in the preparation of the manu-
script.
Supporting Information Available: Experimental de-
tails and analytical data of compounds. This material is
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Cronstein. B. N. Adenosine, an endogenous antiinflammatory
agent. J. Appl. Physiol. 1994, 76, 5-13.
(2) Ohta, A.; Sitkovsky, M. Role of G-protein-coupled adenosine
receptors in downregulation of inflammation and protection from
tissue damage. Nature 2001, 414, 916-920.
(3) Cristalli, G.; Costanzi, S.; Lambertucci, C.; Lupidi, G.; Vittori,
S.; Volpini, R.; Camaioni, E. Adenosine deaminase: Functional
implications and different classes of inhibitors. Med. Res. Rev.
2001, 21, 105-128.
JM050413G