Discovery of hydroxamate bioisosteres as KAT II inhibitors with improved oral bioavailability and pharmacokinetics

Article abstract of DOI:10.1039/c2md20166f

A series of kynurenine aminotransferase II (KAT II) inhibitors has been developed replacing the hydroxamate motif with a bioisostere. Triazolinones or triazoles have proven to be effective replacements with significantly improved pharmacokinetics includin

Full text of DOI:10.1039/c2md20166f

MedChemComm
View Article Online
View Journal | View Issue
CONCISE ARTICLE
Discovery of hydroxamate bioisosteres as KAT II
inhibitors with improved oral bioavailability and
pharmacokinetics†‡
Cite this: Med. Chem. Commun., 2013,
4, 125
Jaclyn L. Henderson, Aarti Sawant-Basak, Jamison B. Tuttle, Amy B. Dounay,
Laura A. McAllister, Jayvardhan Pandit, Suobao Rong, Xinjun Hou, Bruce M. Bechle,
Ji-Young Kim, Vinod Parikh, Somraj Ghosh, Edelweiss Evrard, Laura E. Zawadzke,
Michelle A. Salafia, Brian Rago, Ronald S. Obach, Alan Clark, Kari R. Fonseca,
*
Cheng Chang and Patrick R. Verhoest
A series of kynurenine aminotransferase II (KAT II) inhibitors has been developed replacing the
hydroxamate motif with bioisostere. Triazolinones or triazoles have proven to be effective
a
Received 25th June 2012
Accepted 7th August 2012
replacements with significantly improved pharmacokinetics including reduced clearance and increased
bioavailability. An X-ray crystal structure of an inhibitor bound in KAT II confirms that the irreversible
binding to the co-factor is maintained and that the heterocycles make productive hydrogen bonds to
the arginine-399.
DOI: 10.1039/c2md20166f
www.rsc.org/medchemcomm
Kynurenic acid has a diverse and interesting pharmacology.
KYNA was rst reported as a native antagonist at the glycine site
Introduction
Treatment of the cognitive disruption in schizophrenia is a high
unmet medical need with no therapies approved for these
decits. Evidence in the literature has pointed to a disruption in
the tryptophan metabolism pathway with schizophrenic
patients showing an elevation in L-kynurenine and kynurenic
acid (KYNA) in cerebral spinal uid (CSF) and post mortem
prefrontal cortex.1,2 Kynurenic acid is one of the terminal
products in the pathway and is produced from L-kynurenine by
the kynurenine aminotransferase (KAT) family of enzymes. The
predominant isoform in human brain tissue is the KAT II
enzyme.3 Targeting inhibitors of this enzyme is a strategy to
reduce central KYNA burden in a disease state (Fig. 1).
of the N-methyl-D-aspartate (NMDA) receptor and later reported
to be an inhibitor of the alpha7 nicotinic ion channel.4,5 Inhi-
bition of both of these pathways in schizophrenics has been
hypothesized to contribute to the disease. More recently KYNA
has been shown to be a potent agonist at the arylhydrocarbon
(AhR) receptor. Endogenous brain levels of KYNA are consistent
with its AhR affinity making the AhR receptor and its pharma-
cology of signicant interest in understanding the biological
relevance of decreasing KYNA levels in disease.6
Recent literature reports have utilized selective tool
compounds (S)-ESBA (1) and BFF-122 (2) via central administra-
tion to explore the pharmacology of KAT II inhibition in vivo.7–9
We have reported the rst brain penetrant KAT II inhibitor PF-
04859989 (3), which can lower KYNA levels by 50% in rat
prefrontal cortex following subcutaneous administration.10 This
potent series of compounds inhibits KAT II irreversibly via
formation of an imine adduct with the aldehyde group of the
enzyme co-factor pyridoxal phosphate (PLP), followed by tauto-
merization to an enamine. This newly formed adduct with PLP
cannot exit the active site, resulting in irreversible inhibition of
the enzyme without covalent linkage to the protein (Fig. 2).
Fig. 1 Kynurenic acid biosynthesis.
Neuroscience Medicinal Chemistry, 700 Main Street, Cambridge, MA 12139, USA.
E-mail: patrick.r.verhoest@pzer.com
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c2md20166f
‡ This article is part of a MedChemComm ‘New Talents’ issue highlighting the
work of outstanding rising scientists in medicinal chemistry research.
Fig. 2 Known inhibitors of KAT II.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 125–129 | 125

View Article Online
MedChemComm
Concise Article
to be maintained were the primary amine–PLP adduct as well as
identication of a heterocycle that could hydrogen bond with
Arg-399 and Asp-202, as the hydroxamate efficiently does. Our
modelling suggested that a triazolinone or a triazole replace-
ment could make these desired interactions and maintain
selectivity and potency since the KAT II active site is slightly
larger than the other family members.11 While 3 had less
desirable pharmacokinetics due to glucuronidation, it is an
exquisitely efficient inhibitor which allowed for the addition of
some size and lipophilicity to replace the optimal hydroxamate
to gain potency through other interactions. We had previously
identied that incorporation of an O-aryl group para to the
hydroxamate had improved our potency and incorporated this
knowledge in the design of our bio-isostere replacements.12
Synthesis
Fig. 3 Triazolinone compound 4 (cyan) overlayed with 3 (orange) shows similar
As shown in Scheme 1, the triazolinone 4 could be accessed via a
short sequence from ortho-nitroaryl bromides. Aer installation
of the aryl ether moiety and reduction of the nitro group, a
Negishi reaction and concomitant cyclisation provided the key
lactam intermediate.13 This is in contrast to the synthesis of
hydroxamate derivatives in which the amino acid side chain was
installed prior to careful reduction of the nitro group.14
hydrogen bond to Arg-399 and Asn-202 as the hydroxamate (PDB 4GDY).
Strategy
Although PF-04859989 showed excellent brain penetration and
activity in vivo, it suffered from poor pharmacokinetic proper-
ties due to rapid O-glucuronidation of the hydroxamate. As
shown in the Table 2, PF-04859989 was metabolized to its
glucuronide as the major metabolic pathway in human hep-
atocyctes and human liver microsomes supplemented with
UDPGA. Following IV administration to rats, the clearance is
signicantly greater than blood ow (174 mL minꢀ1 kgꢀ1) with a
low oral bioavailability 2–18%. The short systemic half-life of
the molecule could be a benet of an irreversible inhibitor since
the pharmacodynamic effect is driven by the protein resyn-
thesis/clearance rate. However, improvement in bioavailability
and reduction in glucuronidation would result in less phar-
macokinetic interpatient variability and a reduced dose.
Our strategy to reduce the clearance and improve bioavail-
ability centered around identifying hydroxamate bioisosteres.
We utilized the knowledge from the X-ray crystal structure of 3
in the KAT II active site (Fig. 3). The key interactions that needed
Scheme 2 Synthesis of compound 7. (i) Phenol, Cs2CO3, MeCN, 60 ꢁC, 2 h, 99%;
(ii) Zn, I2, DMF then Pd(OAc)2, XPhos, 60 ꢁC, 12 h, 28%; (iii) Zn, NH4Cl, THF : MeOH
1 : 2, 60 ꢁC, 48 h, 73%; (iv) P2S5, Na2CO3, THF, reflux, 1 h, 76%; (v) acetic acid,
cyclohexanol, 160 ꢁC, 4 h, 66%; and (vi) HCl, 2-propanol, 25 ꢁC, 2 h, 68%.
Scheme 1 Synthesis of compound 5. (i) 3-Methoxyphenol, Cs2CO3, MeCN, 25 ꢁC, 12 h, 84%; (ii) Fe, NH4Cl, THF : MeOH : H2O 2 : 1 : 1, 70 ꢁC, 3 h, 90%; (iii) Zn, I2, DMF
then Pd(OAc)2, XPhos, 60 ꢁC, 12 h, 55%; (iv) P2S5, Na2CO3, THF, reflux, 12 h, 75%; (v) MeI, K2CO3, THF, 15 ꢁC, 20 h, 31%; (vi) ethyl hydrazinecarboxylate, EtOH, 70 ꢁC, 4 h,
44%; (vii) DMF, 150 ꢁC, 2 h, 53%; and (viii) HCl, Et2O, 0 ꢁC, 0.5 h, 69%.
126 | Med. Chem. Commun., 2013, 4, 125–129
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
Conversion of the lactam to the thio lactam and subsequent
methylation allowed for condensation with ethyl hydrazine
carboxylate. Unexpectedly, racemization of the amino-bearing
stereocenter occurred during this step. The synthesis was
completed by cyclisation and deprotection.
In order to access the triazole motif, a modied route was
employed, shown in Scheme 2. For the pyridyl containing core it
was necessary to perform the Negishi reaction prior to reduc-
tion of the nitro group, due to the low yields otherwise afforded
in this step.15 The nitro group could be rapidly reduced to the
amine using zinc, which cyclised spontaneously to the lactam in
situ, with none of the corresponding hydroxamate being
observed. As in the case of the triazolone, conversion to the thio-
lactam was necessary for efficient, acid catalyzed condensation
with formic hydrazide, with cyclisation to the triazole occurring
in the same step. Again, racemisation of the amine-bearing
center was observed.
Biological results and discussion
The strategy to maintain inhibition of the KAT II enzyme with
triazolinone and triazole hydroxamate replacements proved
fruitful. The IC50 for the for the triazolone analogues 4 and 5
showed activity less than 100 nM while the triazoles were
slightly less effective with an IC50 of 100–200 nM. The triazoles
had the advantage of reducing a hydrogen bond donor
compared to the hydroxamate and triazolinone which is a
desired improvement in a physicochemical property to increase
brain penetration.16 To thoroughly understand the SAR of the
irreversible inhibition, second order rate constants kinact/Ki were
determined to remove the time dependency portion of the
potency measure.17 The bioisosteres maintained good potency
with inhibition rate constants between 2000 and 5000 Mꢀ1 sꢀ1
(Table 1).
An X-ray crystal structure of 4 in the KAT II enzyme conrms
the hypothesis that a triazolinone enhances interaction with
Arg-399 and Asn-202 and explains the observed potency. As
designed, the triazolinone makes the desired hydrogen bond
interactions with Arg-399 and Asn-202, accepting three
hydrogen-bonds from Arg-399, and Asn-202. The structure
shows a slight backwards shi in the Arg-399 position versus the
crystal structure of 3 to accommodate the large heterocyclic
ring. We envision that the triazole isostere such as 7 also makes
similar hydrogen bonds to the key residues as a hydroxamate
replacement. This structure clearly shows that the amine moiety
forms a covalent bond with PLP in the enzyme without forming
a covalent bond to the protein. This is a desirable interaction
since a signicant potency gain is achieved without the poten-
tial for hapten mediated immune toxicity.18
Table 1 SAR of hydroxamate bioisosteresa
IC50
(nM)
kinact
Ki
(mM)
kinact/Ki
Entry
Structure
(minꢀ1
)
(Mꢀ1 sꢀ1
)
3
23
84
0.0156
n.d.
0.0142
n.d.
18 500
n.d.
4
5
6
65
0.0155
0.0141
0.0051
0.0815
5060
2915
To conrm the metabolic attributes of the heterocycles, 5
was incubated in human hepatocyctes (HHep) and human liver
microsomes (HLMs); metabolic prole was identied using
LC-MS/MS as shown in Fig. 4.
201
7
122
0.0203
0.0167
2160
As shown in the LC-MS/MS total ion chromatogram (bottom
panel), the peak eluting at 31.80 min corresponding to 5 was the
a
n.d. – not dened.
Fig. 4 Metabolic identification of 5.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 125–129 | 127

View Article Online
MedChemComm
Concise Article
Table 2 Comparison of pharmacokinetic properties of KAT II inhibitors from
hydroxamate to triazolone series; in vitro, in vivo, and predicted PK of leadsa
114 min, which predicts a low scaled blood clearance of 3.6 mL
minꢀ1 kgꢀ1, further conrming low clearance in humans. This
was strongly aligned with the metabolic prole of 5. In addition,
rat showed a strong in vitro to in vivo correlation in respect to
clearance and %F as shown in Table 2. All of the in vitro and in
vivo data gave high condence to the prediction human PK
attributes of 5 and 7 using HLMs. Thus, overall oral %F was
predicted to be 80–90% in humans contrasting to the low
predicted bioavailability of 3.19
To further characterize the clinical potential of this series of
KAT II inhibitors, it was paramount that brain penetration was
maintained. The physicochemical properties of 5 and 7 pre-
dicted that they would have good central penetration with each
having a CNS MPO scores of 5.17. While these compounds are
polar with a ClogP of 1.8(5) and 0.66(7) and a topological polar
surface area of 95(5) and 78(7), these are parameters that are in
line with marketed CNS drugs and do not impact the potential
to generate P-gp substrates. The molecular weight is low for
both compounds (324 and 279) and is the dominant physico-
chemical property predictor for central exposure.20 Compound 5
has an MDR efflux ratio of 1.51 and 7 showed an MDR efflux
ratio of 1.24 in a P-gp over expressing cell line suggesting that
these compounds would have no brain impairment in vivo.21 As
was predicted based on the physicochemical properties and in
vitro MDR efflux ratios, both compounds 5 and 7 displayed
excellent brain penetration with all three compartments, free
plasma, free brain and cerebral spinal uid essentially in
equilibrium (Fig. 5).22
Example 3
(PF-04859989) Example 5 Example 7
t
t
in HHep (min)
in HLM (min)
n.d.
n.d.
104
114
>104
n.d.
138
>120
n.d.
½
½
Clint in HLM
supplemented with UDPGA
Predicted Clb from HLM
Predicted Clb from HLM
Vdss
n.d.
n.d.
1.05
<1
3.6
1.9
1.4
10
1.6
1.4
12
Predicted plasma
elimination t (h)
½
Rat IV Clb
n.d.
n.d.
78.4
4.5
20.6
n.d.
(mL minꢀ1 kgꢀ1
Dog IV Clb
(mL minꢀ1 kgꢀ1
Rat %F
)
)
10.8
18.5
10.8
40
100
92
95
n.d.
91
Dog %F
Predicted %F
a
n.d. not determined.
major drug related moiety, suggesting that this compound has
low clearance, in in vitro human systems. It was also observed
that the major pathways of metabolism were mediated by phase
I oxidative mechanisms with direct phase II mediated conju-
gation i.e. N-glucuronidation at the triazolinone core being a
very small component of the total metabolic clearance. Due to
the increased condence from a PK perspective, 5 was proled
in rat and dog PK studies. The intrinsic clearance of 5 was also
studied in human, rat and dog liver microsomes.
With conrmation that phase II clearance was reduced from
the metabolism identication work with 5, incubation of 7 in
HLMs and HHeps demonstrated a prolonged in vitro t1/2 of >120
min. Using the well-stirred model, and incorporation of
appropriate binding factors, HLMs predicted a scaled blood
clearance of 1.6 mL minꢀ1 kgꢀ1, suggestive of very low turnover
in humans. In a similar manner, compound 5 was incubated
in HHeps and also demonstrated a prolonged in vitro t1/2 of
Ultimately, clinical candidate selection relies on predicted
efficacious dose, which represents the culmination of all phar-
macokinetic and pharmacodynamic properties. Based on
aforementioned favorable pharmacokinetic properties (Table
2), in vitro potency and pharmacodynamic understanding of the
KAT II system (to be published), 0.6 mg QD of 5 and 0.2 mg QD
dosing of 7 was predicted to be able to lower KYNA level by 50%
at steady state. This represented a signicant improvement over
compound 3 (corresponding dose: 40 mg), which can be
attributed to a projected clearance reduction and subsequently
improved oral bioavailability.
Fig. 5 Brain penetration data for compound 5 (left) and 7 (right). Cp ¼ total plasma, Cpu ¼ free plasma, Cb ¼ total brain, Cbu ¼ free brain and CCSF ¼ concentration
cerebral spinal fluid.
128 | Med. Chem. Commun., 2013, 4, 125–129
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
9 F. Rossi, C. Valentian, S. Garavaglia, K. Sathyasaikumar,
R. Schwarcz, K. Okuwaki, S. Ono, Y. Kajii and M. Rizzi, J.
Med. Chem., 2010, 52, 5684–5689.
Conclusion
In conclusion, the design to replace the hydroxamate motif
within the previously reported centrally penetrant KAT II
inhibitors proved successful. Both the triazolinone and triazole
proved to be competent bioisosteres for the hydroxamate. X-ray
crystal structure of 4 bound in KAT II conrms that these
compounds form a covalent adduct with PLP and are irrevers-
ible inhibitors. Additionally the triazolinone and triazole make
productive hydrogen bonds to Arg-399 and Asp-202. Metabolite
identication studies conrm that the use of the heterocycles
blocks the primary glucoronidation route of metabolism of 3.
This structural change results in excellent brain penetration
and signicantly improved projected human pharmacokinetic
properties. Ultimately this produces a signicantly lower dose
projection for 5 and 7 even though they have intrinsically less
potency at the KAT II enzyme than 3.
10 A. B. Dounay, M. Anderson, B. M. Bechle, B. M. Campbell,
M. M. Claffey, A. Evdokimov, E. Evrard, K. R. Fonseca,
X. Gan, S. Ghosh, M. M. Hayward, W. Horner, J.-Y. Kim,
L. A. McAllister, J. Pandit, V. Paradis, V. D. Parikh,
M. R. Reese, S. Rong, M. A. Salaa, K. Schuyten,
C. A. Strick, J. B. Tuttle, J. Valentine, H. Wang,
L. E. Zawadzke and P. R. Verhoest, ACS Med. Chem. Lett.,
2012, 3, 187–192.
11 F. Rossi, S. Garavaglla, V. Montalbano, M. Walsh and
M. Rizzi, J. Biol. Chem., 2008, 238, 3559–3566.
12 M. M. Claffey, A. B. Dounay, X. Gan, M. M. Hayward, S. Rong,
J. B. Tuttle and P. R. Verhoest, 2010, WO2010146488.
13 A. J. Ross, H. L. Lang and R. F. W. Jackson, J. Org. Chem.,
2010, 75, 245–248.
14 L. A. McAllister, B. M. Bechle, A. D. Dounay, E. Evrard,
X. Gan, S. Ghosh, J.-Y. Kim, V. D. Parikh, J. B. Tuttle and
P. R. Verhoest, J. Org. Chem., 2011, 76, 3484–3497.
15 J. B. Tuttle, J. M. Azzarelli, B. M. Bechle, A. B. Dounay,
E. Evrard, X. Gan, S. Ghosh, J. Henderson, J.-Y. Kim,
V. D. Parikh and P. R. Verhoest, Tetrahedron Lett., 2011, 52,
5211–5213.
References
1 K. R. Linderholm, E. Skogh, S. K. Olsson, M. Dahl, M. Holtze,
G. Engberg, M. Samuelsson and S. Erhardt, Schizophr. Bull.,
2010, 38, 426–432.
16 T. T. Wager, X. Hou, P. R. Verhoest and A. Villalobos, ACS
Chem. Neurosci., 2010, 1, 435–449.
17 R. A. Copeland, Evaluation of Enzyme Inhibitors in Drug
2 R. Schwarcz, A. Rassoulpour, H. Q. Wu, D. Medoff,
C. A. Tamminga and R. C. Roberts, Biol. Psychiatry, 2001,
50, 521–530.
Discovery:
A
Guide for Medicinal Chemists and
3 Q. Han, T. Cai, D. A. Tagle and J. Li, Cell. Mol. Life Sci., 2010,
67, 353–368.
4 M. Kessler, T. Terramani, G. Lynch and M. A. Baudry,
J. Neurochem., 1989, 52, 1319–1328.
5 C. Hilmas, E. F. Pereira, M. Alkondon, A. Rassoulpour,
R. Schwarcz and E. X. Albuquerque, J. Neurosci., 2001, 21,
7463–7473.
6 B. C. Natale, I. A. Murray, J. C. Schroeder, C. A. Flaveny,
T. S. Lahoti, E. M. Laurenzana, C. J. Omiecinski and
G. H. Perdew, Toxicol. Sci., 2010, 115, 89–97.
7 R. Pellicciari, R. C. Rizzo, G. Costantino, M. Marinozzi,
L. Amori, P. Guidetti, H. Q. Wu and R. Schwarcz,
ChemMedChem, 2006, 1, 528–531.
Pharmacologists, Wiley, Hoboken, NJ, 2005, pp. 1–265.
18 A detailed view of the N-terminal loop is in the ESI.†
19 S. R. Obach, J. G. Baxter, T. E. Liston, B. M. Silber, B. C. Jones,
F. MacIntyre, D. J. Rance and P. Wastall, J. Pharmacol. Exp.
Ther., 1997, 283, 46–58.
20 T. T. Wager, R. Y. Chandrasekaran, X. Hou, M. D. Troutman,
P. R. Verhoest, A. Villalobos and Y. Will, ACS Chem. Neurosci.,
2010, 1, 420–434.
21 B. Feng, J. B. Mills, R. E. Davidson, R. J. Mireles,
J. S. Janiszewski, M. D. Troutman and S. M. de Morais,
Drug Metab. Dispos., 2008, 36, 268–275.
22 J. C. Kalvass, E. R. Olson, M. P. Cassidy, D. E. Selley and
G. M. Pollack, J. Pharmacol. Exp. Ther., 2007, 323, 346–355.
8 R. Schwarcz, Y. Kajii and S. Ono, 2009, WO2009064836.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 125–129 | 129