Lead optimization and structure-based design of potent and bioavailable deoxycytidine kinase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2009.09.081

A series of deoxycytidine kinase inhibitors was simultaneously optimized for potency and PK properties. A co-crystal structure then allowed merging this series with a high throughput screening hit to afford a highly potent, selective and orally bioavailab

Full text of DOI:10.1016/j.bmcl.2009.09.081

Bioorganic & Medicinal Chemistry Letters 19 (2009) 6784–6787
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Lead optimization and structure-based design of potent and bioavailable
deoxycytidine kinase inhibitors
a
a
a
a
Theodore C. Jessop a, , James E. Tarver , Marianne Carlsen , Amy Xu , Jason P. Healy ,
Alexander Heim-Riether a, Qinghong Fu a, Jerry A. Taylor a, David J. Augeri a, Min Shen a, Terry R. Stouch a,
Ronald V. Swanson c, Leslie W. Tari c, Michael Hunter c, Isaac Hoffman c, Philip E. Keyes a, Xuan-Chuan Yu b,
Maricar Miranda b, Qingyun Liu b, Jonathan C. Swaffield b, S. David Kimball a, Amr Nouraldeen b,
Alan G. E. Wilson b, Ann Marie DiGeorge Foushee b, Kanchan Jhaver b, Rick Finch b, Steve Anderson b,
Tamas Oravecz b, Kenneth G. Carson a
*
a Lexicon Pharmaceuticals, Princeton, NJ 08540, United States
b Lexicon Pharmaceuticals, The Woodlands, TX 77381, United States
c ActiveSite, San Diego, CA 92121, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of deoxycytidine kinase inhibitors was simultaneously optimized for potency and PK properties.
A co-crystal structure then allowed merging this series with a high throughput screening hit to afford a
highly potent, selective and orally bioavailable inhibitor, compound 10. This compound showed dose
dependent inhibition of deoxycytidine kinase in vivo.
Received 20 August 2009
Revised 21 September 2009
Accepted 22 September 2009
Available online 25 September 2009
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Immunology
Virology
Cancer
Deoxycytidine kinase
Pharmacokinetics
As part of our GENOME5000 program, we have investigated the
functions of nearly 5000 human genes by phenotypic analysis of
the corresponding knockout (KO) mice.1,2 The phenotype observed
in deoxycytidine kinase (dCK) KO mice identified dCK as a potential
drug target in multiple disease states within cancer, immunology
and virology.3 The connection between the nucleoside salvage path-
way and these disease states has precedent.4,5 Deoxycytidine kinase
catalyzes the phosphorylation of pyrimidine and purine deoxynu-
cleosides as well as various nucleoside analogs.6–8 Although ubiqui-
tously expressed, it is a nuclear enzyme with particularly high
expression in lymphocytes.7
Crystal structures of dCK bound to various substrates are
known.9 We obtained a co-crystal structure of 5-fluorodeoxycyti-
dine 1 in the substrate binding pocket of dCK (PDB code: 3IPX).
The binding is characterized by extensive hydrogen bonding to
the cytosine base and to the hydroxyl groups of the deoxysugar
(Fig. 1). In addition to the residues that contribute to the hydrogen
bonding network, the substrate binding pocket is further defined
NH
N
2
F
N
O
O
OH
OH
1
* Corresponding author. Tel.: +1 609 466 6076; fax: +1 609 466 3562.
E-mail address: tjessop@lexpharma.com (T.C. Jessop).
Figure 1. Schematic of the substrate binding site of dCK occupied by 1; PDB code
3IPX.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.09.081

T. C. Jessop et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6784–6787
6785
Table 1
by aromatic residues above and below the cytosine ring. Generally,
the enzymatic specificity of dCK allows for a wide range of nucleo-
sides to act as substrates.7 Compound 1 was found to be a compet-
itive inhibitor of dCK with an IC50 of 120 nM.
In designing new inhibitors of dCK, we sought to prepare non-
substrate derivatives that utilized the dense network of hydrogen
bonds available to the cytosine base. Towards this end, we synthe-
sized cyclopentane analogs of 1 as deoxyribose surrogates that dis-
played improved hydrolytic stability. One of the early molecules
that emerged from this approach was synthetic intermediate ester
5 (IC50 = 630 nM, Scheme 1) that prompted further investigation of
aryl substitution.
In vitro and cell potency of phenylsulfonamides 6a–f (see Scheme 1 for structures)
Compound no. R1 (4-position) R2 (3-position) IC50 (nM) EC50 (nM)
a
b
6a
6b
6c
6d
6e
6f
NO2
H
Br
Ph
H
H
H
H
H
Br
Ph
3200
3600
2700
350
1900
120
16,000
17,000
11,000
5500
2800
H
600
a
Inhibition of human dCK (IC50) was determined using a lysate filter-binding
assay.11
b
Cell-based assay was a rescue assay that utilized inhibition of dCK to avoid dCK-
mediated cytotoxicity of arabinoside C (Ara C).11
Ester 5 was prepared by a three step sequence beginning with the
palladium catalyzed
p-allyl reaction of allylic acetate 3 with the
sodium anion of 5-fluorocytosine to give an allylic alcohol in 80%
yield. Hydrogenation afforded saturated alcohol 4 (quantitative
yield) that was converted to 5 using the Martin Mitsunobuinversion.
Treatment of alcohol 4 with N-Boc-sulfonamides under similar
conditions followed by deprotection afforded phenyl sulfonamide
analogs such as 6a in low yields. Biphenylsulfonamides could be pre-
pared by reaction of a bromophenyl sulfonamide with boronic acids
under Suzuki conditions. Versatile intermediate amine 7, prepared
in 51% over three steps via a Boc-nosylate, was acylated with haloge-
nated aryl acids that were then further elaborated to biphenyl
analogs using Suzuki cross-couplings to give analogs 8a–e, 10.10
Replacement of the ester of 5 (IC50 = 630 nM) with a sulfon-
amide afforded 4-nitrophenylsulfonamide 6a with improved
chemical stability but approximately fivefold reduced activity.
Removal of the nitro group gave the nearly equipotent 6b. Biphenyl
analogs afforded potencies superior to our initial lead 5. Substitu-
tion at the 3-position was preferred to the 4-position (Table 1).
Although increasing the polarity of substituents generally had a
modest effect on the IC50, the effect on EC50 was dramatic (see
Table 2
In vitro and cell potency of m-biphenylsulfonamides 6g–s (see Scheme
structures)
1 for
Compound
no.
R1
R2 (3-position)
IC50
(nM)
EC50
(nM)
(4-position)
6g
6h
6i
6j
6k
6l
6m
6n
6o
6p
6q
H
H
H
H
H
H
H
H
H
H
H
4-Methylphenyl
4-Methoxyphenyl
4-Cyanophenyl
4-Methylsulfonyl
4-Aminocarboxy-phenyl
4-Carboxyphenyl
4-Chlorophenyl
3-Chlorophenyl
2-Chlorophenyl
2,4-Dichlorophenyl
4-Chloro-2-methylphenyl
110
64
49
63
89
15,300
51
79
45
21
500
300
390
1800
5900
n.d.
240
360
430
530
290
21
Table 2). For example, methyl substituted 6g and amide substituted
6k have similar IC50 values (110 nM and 89 nM, respectively),
however their EC50 values (500 nM and 5900 nM, respectively) are
separated by more than 10-fold. This trend is illustrated in the polar-
ities and potencies of 6g–k. Charged analogs (as in 6l) were not well
tolerated. There was not a strong preference for substitution at any
one position of the distal phenyl ring (as in 6m–o) and di-substitu-
tion appeared to be preferred (as in 6p, 6q).
NH2
N
NH2
N
NH2
N
F
F
F
N
O
N
O
c
N
O
a,b
H
2
When substituents from the biphenylsulfonamide series (6g–q)
were introduced onto a biphenylcarboxamide scaffold (see Table
3), the carboxamide analogs were generally more potent. Further-
more, frequent PK analysis of key analogs in mice allowed for
simultaneous optimization of PK and potency. A comparative PK
analysis was performed between the amide and sulfonamide series
(see Table 4). The amide analogs 8b, 8c and 8e were superior to the
corresponding sulfonamides 6h, 6m and 6q. In each instance, the
amide analogs had lower clearance and lower volume of distribu-
tion while maintaining greater exposure and oral bioavailability.
While discovery efforts focused mainly on the amide series, a
co-crystal structure was solved of dCK with the high-throughput
screening11 hit 9 (Fig. 2A; PDB code: 3IPY). Compound 9 bound
dCK in a new pocket, formed by reorienting four amino acid side
chains: Tyr 86, Tyr 204, Glu 196 and Glu 197. The biaryl region
of 9 occupies the newly formed pocket with one pyrimidine nitro-
+
OAc
HO
O
4
O
HO
F
3
d,e,f
g,h,i
O2N
5
NH2
NH2
N
F
N
NH2
N
F
N
O
N
O
j,k
N
O
HN
HN
O
O
S
O
H2N
HCl
X
7
R1
R2
6a-q
R
Table 3
8a-e, 10
In vitro and cell potency of m-biphenylamides 8a–e (see Scheme 1 for structures;
X = CH; R as shown below)
Scheme 1. Reagents and conditions: (a) (i) 2, NaOtBu, DMF; (ii) 3, Pd2(dba)3, PPh3,
THF, 80%; (b) 10% Pd/C, H2, MeOH (quant.); (c) PPh3, DEAD, 4-nitrophenylbenzoic
acid, THF, 66%; (d) N-Boc-sulfonamide, PPh3, DEAD, THF; (e) TFA, DCM (2–23%, two
steps); (f) boronic acid, Na2CO3, H2O, CH3CN, Pd(dppf)Cl2 (10–70%); (g) BocHN-Ns,
PPh3, DEAD, THF; (h) PhSH, K2CO3, CH3CN (51%, two steps); (i) HCl, dioxane, quant.;
(j) 3-bromobenzoic acid (X = CH) or 4-bromopyridine-2-carboxylic acid (X = N),
HATU, NMM, DMF or CH3CN (54–60%); (k) boronic acid, PdCl2(dppf), Na2CO3,
CH3CN, H2O (20–85%).
Ref no.
R
IC50 (nM)
EC50 (nM)
8a
8b
8c
8d
8e
Phenyl
260
30
42
23
21
910
100
270
180
170
4-Methoxyphenyl
4-Chlorophenyl
3-Chlorophenyl
4-Chloro-2-methylphenyl

6786
T. C. Jessop et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6784–6787
Table 4
Mouse pharmacokinetics of biphenyl sulfonamides compared to biphenyl amides
NH
2
F
N
R
N
O
Y* N
H
*
Compound no.
-R
Y
AUCinf (nM h)
Cmax (nM)
t1/2 (h)
tmax (h)
BA
Clearance (mL/min/kg)
Vz (L/kg)
IC50 (nM)
EC50 (nM)
6m
8c
4-Cl
4-Cl
4-OMe
4-OMe
4-Cl, 2-Me
4-Cl, 2-Me
SO2
C@O
SO2
C@O
SO2
C@O
5200
15,800
600
3800
2300
1000
16,200
300
3400
1500
10,100
7400
2.1
0.8
1.2
1.6
1.6
1.3
1.5
2
nd
12%
9%
17%
15%
58%
42%
140
3
16
0.2
4.4
1.3
3
51
42
64
30
21
21
1.7
240
270
310
100
290
170
17
0.3
0.8
0.3
0.3
0.3
0.6
6h
8b
6q
8e
10
63
18
23
20
7
10,800
22,000
1.6
0.8
See Figure 3
Mice were dosed at 10 mpk po and 1 mpk iv and blood sampled to determine compound levels.
Clearance and volume were determined from the iv dose; other pharmacokinetic parameters utilized po dosing data. AUCinf: Area under the curve (oral exposure) at infinite
time; Cmax: maximum concentration; t1/2: half-life; tmax: time of maximum concentration; BA: bioavailability; Vz: volume of distribution from the terminal phase.
gen forming a water mediated hydrogen bond to Tyr 86. The upper
portion of 9 sits in the pocket previously occupied by the cytosine
base of 1. The amide carbonyl of 9 appears to mimic the cytosine
carbonyl of dC while lacking the multiple hydrogen bond interac-
tions available with Gln 97, Asp 133 and Arg 104 (compare
Fig. 2A and Fig. 1).
Docking biphenyl amide 8a into this new X-ray crystal structure
suggested that both 8a and 9 might occupy common regions of the
binding pocket (Fig. 2B). Compound 8a appeared to form similar
hydrogen bonding and
p-stacking interactions as those seen
between dCK and the cytosine base. In addition, the biaryl portion
of 8a resided in the newly formed hydrophobic pocket.
Compounds 8a and 9 appeared to occupy the same region of the
enzyme and share a similar shape, degree of flexibility and relative
potency. Nonetheless, each compound seemed to derive much of
its binding affinity through interactions with different parts of
the enzyme. These characteristics compelled us to design a hybrid
molecule (see Fig. 3) that would contain the complementary
regions and binding features of the two molecules. Compound 8a
contained the highly functionalized cytosine base that could form
a dense network of hydrogen bonds (Fig. 2B, upper region). Com-
pound 9 lacked this hydrogen bonding network, yet retained good
affinity for the enzyme presumably due to interactions of the
pyrimidine and benzothiophene moieties with Pro 201, Ile 200,
Tyr 204 and Tyr 86 (see Fig. 2, panels A and B). The designed hybrid
10 incorporated the upper region of 8a and the lower region of 9 as
seen in Figure 3.
Gratifyingly, compound 10 had an IC50 of 1.7 nM and EC50 of
17 nM. Further evaluation showed it to be stable to both human
and mouse liver microsomes. Mouse pharmacokinetics of 10 were
also quite favourable (see Table 4). The activity of 10 was measured
in a panel of 75 in vitro assays. At a concentration of 10 lM, com-
NH2
F
NH2
N
N
F
N
O
N
O
N
O
N
x
HN
N
N
HN
O
O
N
S
S
N
Figure 2. Panel A: Co-crystal structure of dCK and screening hit 9; upper portion of
9 occupies some of the deoxycytidine binding pocket (compare Fig. 1) while the
lower biaryl portion occupies a newly formed pocket; the ATP binding pocket (not
shown) begins in the lower right corner of the panel; PDB code 3IPY. Panel B:
Docking of 8a from the m-biphenylamide series into the same binding site shown in
panel A.
8a
IC50= 260 nM
9
10
IC50= 510 nM
IC50= 1.7 nM
EC50= 17 nM
Figure 3. Design of hybrid molecule 10. Functionality shown in blue was utilized in
the design of hybrid molecule 10.

T. C. Jessop et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6784–6787
6787
1/dCK and 9/dCK allowed the design of hybrid molecule 10. Com-
pound 10 was a potent, selective, and orally bioavailable inhibitor
of dCK. Compound 10 also inhibited dCK both in primary T cells
and in vivo.
Acknowledgements
The authors thank Alan Main and Giovanni Cianchetta for help-
ful discussions and assistance in manuscript preparation.
References and notes
1. Zambrowicz, B. P.; Sands, A. T. Nat. Rev. Drug Disc. 2003, 2, 38.
2. Walke, D. W.; Han, C.; Shaw, J.; Wann, E.; Zambrowicz, B.; Sands, A. Curr. Opin.
Biotechnol. 2001, 12, 626.
Figure 4. Compound 10 shows dose-dependent (10, 30, 100 mpk) inhibition of dCK
activity in vivo. VC = vehicle control.
3. Anderson, S., in preparation.
4. Carson, D. A.; Kaye, J.; Seegmiller, J. E. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 5677.
5. Joachims, M. L.; Marble, P. A.; Laurent, A. B.; Pastuszko, P.; Paliotta, M.;
Blackburn, M. R.; Thompson, L. F. J. Immunol. 2008, 181, 8153.
6. Arner, E. S. J.; Eriksson, S. Pharmacol. Ther. 1995, 67, 155.
7. Bohman, C.; Eriksson, S. Biochemistry 1988, 27, 4258.
8. Gao, W. Y.; Shirasaka, T.; Johns, D. G.; Broder, S.; Mitsuya, H. J. Clin. Invest. 1993,
91, 2326.
9. Sabini, E.; Hazra, S.; Konrad, M.; Lavie, A. J. Med. Chem. 2007, 50, 3004. and
references cited therein.
10. Augeri, D. J.; Carlsen, M.; Carson, K. G.; Fu, Q.; Heim-Riether, A.; Jessop, T. C.;
Tarver, J.; Taylor, J. A. PCT Int. Appl. WO 2008076778 A1, 2008.
11. Yu, X. C.; Miranda, M.; Liu, Z.; Patel, S.; Nguyen, N.; Carson, K.; Liu, Q.;
Swaffield, J. C. J. Biomol. Screen., in press.
pound 10 showed less than 50% inhibition against all targets
tested, except human Beta-2 and A2A.
The potency of 10 was also evaluated in primary mouse T cells.
After stimulation of mouse splenic T cells with anti-CD3 and anti-
CD28 antibodies, incorporation of 3H-dC was measured.12 Com-
pound 10 inhibited the incorporation of 3H-dC with an EC50 of
90 nM (human T cells) and 17 nM (mouse T cells). As expected,
since deoxythymidine (dT) is not a substrate of dCK, incorporation
of dT was not reduced after similar administration of 10.
The ability of 10 to inhibit dCK in vivo was measured by reduc-
ing the incorporation of radiolabelled dC in mouse spleens after T
cell stimulation.13 Reduction of 3H-dC incorporation was dose
dependent and at a dose of 100 mpk, incorporation of 3H-dC was
reduced to the level observed without T cell stimulation (Fig. 4).
In summary, de novo design using 5-fluorodeoxycytidine 1
afforded submicromolar lead 5. Concurrent optimization of
potency and pharmacokinetics culminated in m-biphenylcarboxa-
mides 8a–e. Insights gained from X-ray co-crystal structures of
12. Radu, C. G.; Shu, C. J.; Nair-Gill, E.; Shelly, S. M.; Barrio, J. R.; Satyamurthy, N.;
Phelps, M. E.; Witte, O. N. Nature Med. 2008, 14, 783.
13. Mice were dosed with compound by oral gavage beginning on day À1 relative
to ip injection of anti-CD3 (0.5
lg/g body wt) antibodies. On day 0 mice
received the second dose of compound followed by injection of anti-CD3, and a
third dose of compound on day 1, followed by ip injection of 3H-dCTP (100 Ci/
animal). Spleens were harvested from the mice on day 2, dissociated into single
cell suspensions (based on spleen weights at 30 mg/ml), and a 25 l aliquot of
l
l
cell suspension harvested onto glass fibre filters and counted by liquid
scintillation. n = 6–8 mice per group. The group receiving no anti-CD3
represents background 3H-dC incorporation (n = 2). *p <0.05 compared to
vehicle control (VC).