Pyrimidinopyrimidine inhibitors of ketohexokinase: Exploring the ring C2 group that interacts with Asp-27B in the ligand binding pocket

Article abstract of DOI:10.1016/j.bmcl.2012.06.008

Inhibitors of ketohexokinase (KHK) have potential for the treatment of diabetes and obesity. We have continued studies on a pyrimidinopyrimidine series of potent KHK inhibitors by exploring the 2-position substituent (R ³) that interacts with Asp-27B in the ATP-binding region of KHK (viz. 1, 2; Table 1). We found that increased spacing between the terminal ammonium group and the heterocyclic scaffold (viz. 16-20), such that interaction with Asp-27B is not possible, still results in potent KHK inhibition (IC₅₀ = 15-50 nM). We propose a new interaction with Asp-194, which serves to expand the pyrimidinopyrimidine pharmacophore.

Full text of DOI:10.1016/j.bmcl.2012.06.008

Bioorganic & Medicinal Chemistry Letters 22 (2012) 5326–5329
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Pyrimidinopyrimidine inhibitors of ketohexokinase: Exploring the ring C2
group that interacts with Asp-27B in the ligand binding pocket
⇑
,
Bruce E. Maryanoff , John C. O’Neill, David F. McComsey, Stephen C. Yabut, Diane K. Luci, Alan C. Gibbs,
⇑
Margery A. Connelly
Janssen Pharmaceutical Companies of Johnson & Johnson, Welsh & McKean Roads, Spring House, PA 19477-0776, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Inhibitors of ketohexokinase (KHK) have potential for the treatment of diabetes and obesity. We have
continued studies on a pyrimidinopyrimidine series of potent KHK inhibitors by exploring the 2-position
substituent (R3) that interacts with Asp-27B in the ATP-binding region of KHK (viz. 1, 2; Table 1). We
found that increased spacing between the terminal ammonium group and the heterocyclic scaffold
(viz. 16ꢀ20), such that interaction with Asp-27B is not possible, still results in potent KHK inhibition
(IC50 = 15ꢀ50 nM). We propose a new interaction with Asp-194, which serves to expand the pyrimidino-
pyrimidine pharmacophore.
Received 13 March 2012
Revised 1 June 2012
Accepted 4 June 2012
Available online 17 June 2012
Keywords:
Kinase
Diabetes
Ó 2012 Elsevier Ltd. All rights reserved.
Obesity
Fructose metabolism
Elevated fructose ingestion promotes various metabolic
disturbances in animal models, including weight gain, hyperlipid-
emia, hypertension, and insulin resistance.1–5 Furthermore, the
long-term consumption of fructose in overweight humans increases
energy intake, body weight, fat mass, blood pressure, and plasma
triglycerides.6,7 This situation exacerbates, and contributes to the
onset of, non-insulin-dependent (type 2) diabetes mellitus (NID-
DM), which is characterized by hyperglycemia and insulin resis-
tance. Indeed, a marked correlation exists between the prevalence
of NIDDM and increased energy intake from highly refined sugars.8
The enzyme ketohexokinase (KHK; EC 2.7.1.3), also known as
fructokinase, rapidly metabolizes fructose within the liver.9 KHK
catalyzes the phosphorylation of fructose on position C1 by trans-
ferring a phosphate group from adenosine 50-triphosphate (ATP)
to yield fructose-1-phosphate (F1P), which enters normal metabolic
pathways.10–12 Unfortunately, fructose metabolism lacks robust
control mechanisms, unlike the tight regulation of glucose path-
ways.13 Given that KHK has a high KM value, is not inhibited by prod-
uct (F1P), and is not regulated allosterically, high concentrations of
fructose can readily provide the components of triglycerides.14 Sup-
port for KHK as a therapeutic target emanates from human genetic
mutations that cause essential fructosuria.15 In this benign condi-
tion, individuals have inactive isoforms of hepatic KHK, such that
ingestion of fructose, sucrose, or sorbitol imparts elevated blood
fructose levels and increased urinary excretion of fructose. Conse-
quently, the targeting of fructose metabolism by inhibition of KHK
could offer a novel approach to treat NIDDM and obesity amidst
modern diets due to reduced body weight, free fatty acids, and tri-
glycerides, without a mechanism-based safety issue.
We recently reported a series of pyrimidinopyrimidine deriva-
tives that provide potent, selective inhibitors of human hepatic
KHK (KHK-C isoform).16 For example, 1 exhibited a KHK IC50 value
of 12 nM and showed good cellular KHK inhibition (IC50 <500 nM).
An X-ray cocrystal structure of 1ꢁKHK, a functional pseudo-
homodimer (a and b subunits) with each subunit occupied by the
ligand, revealed important interactions within the enzyme’s ATP-
binding pocket, one of which involved the side chain of Asp-27B
(Fig. 1).16 Because this ionic interaction seems particularly impor-
tant for anchoring the ligand and achieving high affinity, we have
explored it more thoroughly, as depicted by the red atoms in struc-
ture 2. The results of this investigation are reported herein.
SMe
NH
SMe
NH
⇑
Corresponding authors. Tel.: +1 267 980 3512 (B.E.M.); tel.: +1 215 628 5202
(M.A.C.).
7
N
N
1
N
N
E-mail addresses: bmaryano@scripps.edu (B.E. Maryanoff), mconnell@its.jnj.com
(M.A. Connelly).
Present addresses: Department of Chemistry, The Scripps Research Institute,
10550 North Torrey Pines Road, La Jolla, CA 92037, USA. Pennsylvania Drug Discovery
Institute, 3805 Old Easton Road, Doylestown, PA 18902, USA.
5
N
N
N
3
N
N
N
X
N
NH
NH
Z
H
HN
Y
1
2
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.06.008

B. E. Maryanoff et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5326–5329
5327
Table 1
Structures and KHK inhibition results for analogues of 1 with variation of so-called R3
SMe
NH
N
N
N
R3
N
N
HN
b
c
Compda
N-R3
IC50 (nM)
1
Piperazino
12
3
4
2,6-Diazaspiro[3.3]hept-2-yl
3-Me-piperazino
8.0
70
5
(R)-2-Me-piperazino
58
6
7
cis-2,6-diMe-piperazin-4-yl
3,3-di-Me-piperazino
110
40
8
(piperidin-4-yl)CH2NMe–
66
9
(R,R)-cis-H8-6H-Pyrrolo[3,4-b]pyridin-6-yl
(S,S)-cis-H8-6H-Pyrrolo[3,4-b]pyridin-6-yl
(S,S)-2,5-Diazabicyclo[2.2.1]hept-2-yl
(3-Amino-8-aza-bicyclo[3.2.1]octan)-8-yl
2,3,4,5-Tetrahydro-1H-1,4-benzodiazepin-4-yl
3,30-Spiro(bis-pyrrolidin)-1-yl
2,6-Diazaspiro[3.4]octan-6-yl
4,40-Spiro(bis-piperidin)-1-yl
4-(Piperidin-4-yl)piperazino
4-(Piperazin-1-yl)piperidino
4-(Piperidin-4-yl)piperidino
440
300
4600
800
6000
43
2.0
15
48
10
11
12
13
14
15
16
17
18
19
20
Figure 1. Crystal structure of the 1ꢁKHK complex. View of 1 (C, green; N, blue; S,
yellow) and neighboring KHK residues in subunit a (C, white; N, blue; O, red; S,
yellow) with Asp-27 in subunit b (‘Asp-27B’) (C, light blue; N, blue; O, red). The red
sphere represents a conserved water molecule. The H-bond between the piperazine
nitrogen and Asp-27B Od is 2.8 Å (PDB accession number: 3Q92).
28
50
30
4-(4-Me-piperazin-1-yl)piperidino
The 2-(methylthio)phenylamino group was found to be optimal
at position 8 (viz. 1) and cyclopropylmethylamino was among the
best groups at position 4.15 Since the Asp-27B residue of KHK is
located in the vicinity of the polar channel through which the
triphosphate moiety of ATP passes, we supposed that additional,
useful interactions could be obtained by modifying the substituent
at position 2 (R3), especially by extension into the ‘triphosphate
channel’ (Fig. 2). As reported earlier,15 the piperazino group can
be successfully replaced by 3-aminopiperidino, 4-(amino-
a
New compounds were purified by HPLC and characterized by ESI-MS and 1H
NMR.
b
The term ‘R3’ relates to the substituent on position 2 (i.e., C2) of the pyrimidi-
nopyrimidine scaffold, as used in Ref. 6.
c
Inhibition of recombinant human hepatic KHK in terms of IC50 values.
Further exploration of this portion of the ligand was carried out
with compounds 4ꢀ19 (Table 1; Chart 1). The compounds of inter-
est were synthesized by using key starting material 21,16 as exem-
plified for production of 14 (Scheme 1), and assayed for inhibition
of human hepatic KHK (KHK-C) according to the method described
earlier.16
methyl)piperidino,
methyl)azetidino, and 2,6-diazaspiro[3.3]hept-2-yl [3,30-spiro(bis-
azetidin)-1-yl; 3] groups, maintaining good potency (IC50
3-(aminomethyl)piperidino,
3-(amino-
=
8ꢀ30 nM); thus, the spacing between the Asp-27B interactive
nitrogen and the heterocyclic scaffold is not narrowly defined.
However, the acyclic R3 group MeNHCH2CH2(Me)N- was not as
effective, with an IC50 of 130 nM. An ammonium group bearing
more than one proton, that is NH2+ or NH3+, appeared to be optimal
for interaction with Asp-27B. For example, the N-methylpiperazino
analogue of 1 was 10-fold less potent than 1 (IC50 = 110 nM).
Analogues 4ꢀ7 were meant to test the effect of C-methylation
of the piperazine ring, the distal nitrogen of which binds to Asp-
27B. A single methyl group adjacent to the interacting piperazine
nitrogen resulted in a sixfold loss of potency (cf. 4 and 1) and a sin-
gle, remote methyl group resulted in fivefold loss in potency (cf. 5
and 1). Two methyl groups adjacent to the interacting piperazine
nitrogen, on opposite sides, resulted in a ninefold loss of potency
(cf. 6 and 1) and gem-dimethyl groups resulted in just a threefold
loss of potency (cf. 7 and 1). These data indicate that moderate ste-
ric hindrance in this part of the ligand is not a major impediment to
effective KHK binding. However, bicyclic piperazine 11 led to a
nearly 400-fold loss in potency (cf. 11 and 1) and bulky bicyclic
amine 12 was much less potent (cf. 12 and 1). A fused phenyl ring
that decreased the basicity of the interactive nitrogen was very dis-
ruptive (cf. 13 and 1).
As noted above, additional spacing between the Asp-27B inter-
active nitrogen and the pyrimidinopyrimidine scaffold, such as
with 4-(aminomethyl)piperidine (KHK IC50 = 10 nM), still afforded
potent KHK inhibition. Reversal of the connectivity, as in 8, was
also reasonably well tolerated (IC50 = 66 nM). However, conforma-
tional constraint, as with 9 and 10, resulted in a 30-fold reduction
in potency. Because the use of a spirocyclic piperazine-mimetic, as
in 3, provided very potent KHK inhibition (IC50 = 8 nM), we
explored this approach further with compounds 14ꢀ16, which
gave good KHK inhibition. Indeed, compound 15, with an IC50 value
of 2 nM, has the distinction of being the most potent substance yet
Figure 2. Crystal structure of the 1ꢁKHK complex (mainly subunit a) depicting the
D27B interaction (open ellipse) and the potentially useful channel domain (filled
ellipse). Color code: C, light blue; N, blue; O, red; S, yellow; C in subunit b, green.

5328
B. E. Maryanoff et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5326–5329
SMe
N
N
N
SMe
N
N
N
SMe
N
N
N
H
N
H
N
H
N
N
H
N
H
N
H
N
N
N
N
N
N
H
N
H
NH
H
NH2
9/10
11
12
SMe
N
N
SMe
H
N
N
N
N
SMe
N
N
H
H
N
N
N
N
H
N
H
N
N
H
N
N
N
N
N
N
N
NH
N
N
Figure 3. Computer docking model of 18 (mostly grayish white and yellow
Connolly surface; N, blue) complexed with KHK (mainly subunit a; stick model: C,
green; N, blue; O, red; C in subunit b, light blue) depicting the Asp-194 interaction
(yellow side chain; red, hashed Connolly surface). Acid residues Asp-27B (white
side chain with white arrow below it) and Glu-173 (yellow side chain), which are
H
H
15
13
16
SMe
SMe
SMe
N
N
N
N
N
N
H
N
H
N
H
N
+
N
N
out of reach of the piperazine NH2 group, are also labeled.
N
H
H
H
N
N
N
N
N
N
N
N
N
N
N
H
N
R
N
H
18
20
17
R = H
R = Me
19
Chart 1.
MeS
HN
MeS
HN
NH
N
Boc
N
N
CF3CO2H
i-Pr2EtN, THF, Δ, 24 h
N
N
22
N
N
Cl
N
N
N
then HPLC
HN
HN
HN
21
14
Scheme 1.
reported for inhibition of KHK (fourfold better than 3).16–18 The
high potency for 16 (IC50 = 15 nM) was surprising, since the inter-
active nitrogen atom is now positioned well beyond the region of
the active site that is dominated by Asp-27B. Additional studies
with 17ꢀ19 confirmed this point, with IC50 values ranging from
30 to 50 nM. Moreover, methylation of the terminal nitrogen of
18 did not cause a significant loss in potency (viz. 20), which differs
from the above-mentioned 10-fold loss for N-methylation of 1.
These results suggest that a different KHK carboxylate residue
may be involved.
We sought to explain the observations with 16ꢀ20 by obtaining
a ligand-KHK cocrystal structure. However, we were unable to
grow suitable crystals with 16, 17, or 18, perhaps due to conforma-
tional changes in the protein sufficient to alter the original crystal-
lization mode. Therefore, we resorted to a computational docking
study with 18 in the closed subunit (a) of KHK to acquire some
understanding (Fig. 3).17,19 Ligand 18 is oriented in the active site
of KHK is the usual manner,16 except that the terminal ammonium
group of the now-extended R3 group (i.e., the piperazine nitrogen)
is not in the vicinity of Asp-27B to interact with its carboxylate side
chain. By contrast, that ammonium group of 18 is poised to interact
Figure 4. Schematic diagrams showing the key pharmacophore requirements for
(a) 1 and (b) 18.
with the carboxylate side chain of Asp-194 within the triphosphate
channel. The carboxylate side chain of Glu-173, another potential
site for interaction, is too remote to be accessed. If the modeling

B. E. Maryanoff et al. / Bioorg. Med. Chem. Lett. 22 (2012) 5326–5329
5329
picture proposed for 18 is valid, then compounds 16, 17, 19, and 20
would presumably adopt this new interaction mode, as well.
An alternative explanation for the high potency of 16ꢀ20 could
derive from considering binding of the ligand in the active site of
KHK subunit b, which has a very ‘open’ conformation.16 However,
an inspection of this active-site cavity did not reveal an obvious
acid residue at a favorable location for enhancing ligand affinity.
Given the excellent KHK inhibition for 15, it was studied in a
HepG2 cellular assay that measures the level of fructose-1-phos-
phate in cell lysates by using LC–MS for quantification.16 Com-
pound 15 exhibited potent cellular KHK inhibition, with an IC50
value of 150 nM.
with Asp-27B is not possible; however, a new, putative interaction
with Asp-194 is introduced. This finding not only further defines
the pyrimidinopyrimidine pharmacophore for obtaining potent
KHK inhibition, but also suggests a potential interaction site for
other KHK inhibitor chemotypes.
Acknowledgments
We thank Dr. Marta Abad for helpful discussions; Dr. John Mas-
ucci and William Jones for LC–MS analysis.
References and notes
Compound 17 was examined in a diverse panel of 31 human
protein kinases, representing the different families, for off-target
1. Kasim-Karakas, S. E.; Vriend, H.; Almario, R.; Chow, L. C.; Goodman, M. N. J. Lab.
Clin. Med. 1996, 128, 208.
2. Hwang, I. S.; Ho, H.; Hoffman, B. B.; Reaven, G. M. Hypertension 1987, 10, 512.
3. Reiser, S.; Hallfrisch, J. J. Nutr. 1977, 107, 147.
4. Zavaroni, I.; Sander, S.; Scott, S.; Reaven, G. M. Metabolism 1980, 29, 970.
5. Martinez, F. J.; Rizza, R. A.; Romero, J. C. Hypertension 1994, 23, 456.
6. Raben, A.; Vasilaras, T. H.; Moller, A. C.; Astrup, A. Am. J. Clin. Nutr. 2002, 76,
721.
7. Teff, K. L.; Elliott, S. S.; Tschop, M.; Kieffer, T. J.; Rader, D.; Heiman, M.;
Townsens, R. R.; Keim, N. L.; D’Alessio,; Havel, P. J. J. Clin. Endocrinol. Metab.
2004, 89, 2963.
8. Gross, L. S.; Li, L.; Ford, E. S.; Liu, S. Am. J. Clin. Nutr. 2004, 79, 774.
9. Basciano, H.; Federico, L.; Adeli, K. Nutr. Metab. 2005, 2, 5. http://dx.doi.org/
10.1186/1743-7075-2-5.
10. Raushel, F. M.; Cleland, W. W. Biochemistry 1977, 16, 2169.
11. Raushel, F. M.; Cleland, W. W. Biochemistry 1977, 16, 2176.
12. Raushel, F. M.; Cleland, W. W. J. Biol. Chem. 1973, 248, 8174.
13. Mayes, P. A. Am. J. Clin. Nutr. 1993, 58, 754S.
14. Elliott, S. S.; Keim, N. L.; Stern, J. S.; Teff, K.; Havel, P. J. Am. J. Clin. Nutr. 2002, 76,
911.
15. Asipu, A.; Hayward, B. E.; O’Reilly, J.; Bonthron, D. T. Diabetes 2003, 52, 2426.
16. Maryanoff, B. E.; O’Neill, J. C.; McComsey, D. F.; Yabut, S. C.; Luci, D. K.; Jordan,
A. D.; Masucci, J. A.; Jones, W. J.; Abad, M. C.; Gibbs, A. C.; Petrounia, I. ACS Med.
Chem. Lett. 2011, 2, 538.
kinase inhibition at a concentration of 10
100
M ATP).20 None of the 31 kinases was inhibited >50% at this
high concentration of 17. CAMK2A was inhibited by 41% at
10 M, but the rest were inhibited <15% (IC50 <<10 M for 30 ki-
nases). Compound 3 did not inhibit any of these kinases by more
than 25% at 10 M.
lM (Invitrogen; with
l
l
l
l
Our original study on pyrimidinopyrimidines16 was the first
published report of potent KHK inhibitors (IC50 <100 nM). The
pharmacophore was established by X-ray data, such as for 1ꢁKHK:
in the a subunit of KHK, the ring nitrogen N3 of the pyrimidinopyr-
imidine was hydrogen bonded to the conserved water molecule,
the 2-(methylthio)phenyl group occupied a hydrophobic region
largely defined by Phe-260, and the piperazine NH2 (of R3) inter-
+
acted with the carboxylate side chain of Asp-27B (Fig. 4a). Our
present study suggests an expansion of the pharmacophore for po-
tent KHK inhibition, wherein the terminal ammonium group of
16ꢀ20 now extends beyond the reach of Asp-27B to interact with
Asp-194, as represented in Figure 4b. Thus, it would appear that
the R3 substituent can be substantially augmented, as in 16ꢀ20,
while retaining a high level of KHK affinity. Other KHK inhibitor li-
gands, identified through fragment-based methods, further illus-
trate the possible modes of interaction within the ATP-binding
pocket of KHK based on different chemotypes.17,18
17. Gibbs, A. C.; Abad, M. C.; Zhang, X.; Tounge, B. A.; Lewandowski, F. A.; Struble,
G. T.; Sun, W.; Sui, Z.; Kuo, L. C. J. Med. Chem. 2010, 53, 7979.
18. Zhang, X.; Song, F.; Kuo, G.-H.; Xiang, A.; Gibbs, A. C.; Abad, M. C.; Sun, W.; Kuo,
L. C.; Sui, Z. Bioorg. Med. Chem. Lett. 2011, 21, 4762.
19. KHK structure coordinates from in-house studies were used for Glide docking
studies. Details on this method are given in Ref. 17.
20. Kinase selectivity was assessed with a panel of 31 kinases, across the kinase
families, by using a FRET assay platform from Invitrogen. The kinases were
In summary, we have studied the pyrimidinopyrimidine series
of KHK inhibitors further by exploring the substituent at ring posi-
tion 2 (referred to as R3) that interacts with Asp-27B in the ligand
binding pocket. We found that the R3 group can be substantially
enlarged while retaining potent KHK inhibition, as with analogues
16ꢀ20. Given the extended spacing of the terminal ammonium
group away from the pyrimidinopyrimidine scaffold, interaction
tested with 10
ranked based on
l
M compound and 100
lM ATP. Inhibition activities were
%
inhibition at 10 M. Kinases examined: ABL1, ALK4
l
(ACVR1B), AKT1, AMPK A1/B1/G1, AURKA, CAMK1D, CAMK2A, CDK1/cyclin B,
CHEK1, CHEK2, CSNK1D (CK1d), DAPK3, EGFR, EPHB1, GSK3B, INSR, IRAK4,
JAK2, MAPK13 (p38d), MST4, NEK2, NTRK1 (TRKA), PAK3, PDGFRB, PIM2, PLK3,
PRKACA, PRKCQ (PKCh), ROCK1, RPS6KA3 (RSK2), SRC. Full names for the
kinases are given in the Supplementary data for Ref. 16.