Novel Autotaxin Inhibitors for the Treatment of Osteoarthritis Pain: Lead Optimization via Structure-Based Drug Design

Article abstract of DOI:10.1021/acsmedchemlett.6b00207

In an effort to develop a novel therapeutic agent aimed at addressing the unmet need of patients with osteoarthritis pain, we set out to develop an inhibitor for autotaxin with excellent potency and physical properties to allow for the clinical investigation of autotaxin-induced nociceptive and neuropathic pain. An initial hit identification campaign led to an aminopyrimidine series with an autotaxin IC₅₀ of 500 nM. X-ray crystallography enabled the optimization to a lead compound that demonstrated favorable potency (IC₅₀ = 2 nM), PK properties, and a robust PK/PD relationship.

Full text of DOI:10.1021/acsmedchemlett.6b00207

Letter
pubs.acs.org/acsmedchemlett
Novel Autotaxin Inhibitors for the Treatment of Osteoarthritis Pain:
Lead Optimization via Structure-Based Drug Design
Spencer B. Jones,* Lance A. Pfeifer, Thomas J. Bleisch, Thomas J. Beauchamp, Jim D. Durbin,
V. Joseph Klimkowski,† Norman E. Hughes, Christopher J. Rito, Yen Dao, Joseph M. Gruber,‡ Hai Bui,
Mark G. Chambers, Srinivasan Chandrasekhar,† Chaohua Lin, Denis J. McCann, Daniel R. Mudra,
Jennifer L. Oskins, Craig A. Swearingen, Kannan Thirunavukkarasu, and Bryan H. Norman
Lilly Research Laboratories, A Division of Eli Lilly and Company, Indianapolis, Indiana 46285, United States
S
* Supporting Information
ABSTRACT: In an effort to develop a novel therapeutic
agent aimed at addressing the unmet need of patients with
osteoarthritis pain, we set out to develop an inhibitor for
autotaxin with excellent potency and physical properties to
allow for the clinical investigation of autotaxin-induced
nociceptive and neuropathic pain. An initial hit identification campaign led to an aminopyrimidine series with an autotaxin
IC50 of 500 nM. X-ray crystallography enabled the optimization to a lead compound that demonstrated favorable potency (IC50 =
2 nM), PK properties, and a robust PK/PD relationship.
KEYWORDS: Autotaxin, tool molecule, osteoarthritis, LPA
steoarthritis (OA) is a highly prevalent disease affecting
Omany adults including more than one out of three
individuals aged 65 or older in the United States.1 In addition
to significant accompanying pain, OA frequently leads to
pronounced disability resulting in the loss of work, hospital-
ization, and joint replacement procedures.2 Current first-line
pharmacological treatment options for OA focus on reducing
inflammation and the associated pain. Nonsteroidal anti-
inflammatory drugs (NSAIDS) and selective COX-2 inhibitors
are among the most prescribed medications for OA pain but
unfortunately are also frequently accompanied by gastro-
intestinal, renal, and CV side effects, limiting their use.3
Recently, the role of lyosophosphatidic acid (LPA) in certain
inflammatory conditions has been studied.4 LPA exists as a
number of molecular species that have variable saturated and
unsaturated fatty acid chains.5 Signaling of LPA through six
GPCRs (LPA Receptors 1−6) has been shown to lead to the
upregulation of inflammatory cytokines and matrix metal-
loproteinases, which contribute to the pathogenesis of OA.6
LPA signaling has also been associated with many other
pathologies, such as pulmonary fibrosis and cancer.7
In vivo, the enzyme autotaxin (ATX), with lyosophosopholi-
pase D activity, is the primary source of extracellular LPA, which
results from the cleavage of choline from lysophosphatidylcho-
line (LPC) (Figure 1). LPA is also produced through action of
secreted phospholipases A2 (sPLA2) on phosphatidic acid (PA),
although this is believed to be a minor route of extracellular LPA
production in vivo.8,9
Figure 1. Two biosynthetic routes of lysophosphatidic acid (LPA)
production. R groups on lipids = various unsaturated and saturated fatty
acid chains.
lysophopholipase D enzyme, which catalyzed the conversion of
LPC to LPA.11 Autotaxin is a multidomain protein with two N-
terminal somatomedin B-like domains, a centrally located
phosphodiesterase domain, and a catalytically inactive nucle-
ase-like domain on the C-terminal region. It is expressed in four
main isoforms (ATXα−δ) with largely unknown differential
functionality in vivo.7 The catalytic domain of ATX comprises
two zinc ions coordinated with histidine and aspartic acid
residues with a threonine alcohol serving as the nucleophile. A
large hydrophobic pocket and hydrophilic groove accommodate
the various lipophilic LPC species and likely position the choline
group in the vicinity of the zinc ions and the phosphate group
proximal to the secondary alcohol of the threonine residue.
Autotaxin is an extracellular, 125 kDa protein that was
originally characterized in 1993 by Stracke et al. as a motility
stimulating protein.10 In 2002, Umezu-Goto and co-workers
demonstrated that ATX was the same protein as a known
Received: May 18, 2016
Accepted: July 25, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acsmedchemlett.6b00207
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
Structures of several LPA species with mouse ATX have been
published illustrating this interaction (Figure 2).12
Table 1. Zinc-Binding Group SAR
Figure 2. Cocrystal structure of LPA 22:6 bound to mouse autotaxin.12
In an effort to enable clinical investigation of the connectivity
between human OA pain and autotaxin induced LPA generation,
we set out to design a small molecule inhibitor of this enzyme.
Two biochemical assays were developed to enable character-
ization of the autotaxin inhibitors: an LPC coupled enzyme assay
and a human plasma assay. The LPC coupled assay uses an
enzyme cocktail consisting of autotaxin, choline oxidase, and
horseradish peroxidase to enable a fluorescent readout of LPA
production via oxidation of the choline byproduct. The human
plasma assay utilizes human donor plasma with endogenous
autotaxin and LPC substrates. The LPA formed during the
course of the assay is quantified by mass spectrometry. The
plasma assay has the advantage of accounting for plasma protein
binding. Indeed, a potency shift was observed when moving from
the LPC to plasma assay in a manner that correlated with plasma
protein binding.
Following a traditional high-throughput screen of the Lilly
compound collection, we identified the aminopyrimidine 1,
which demonstrated favorable starting potencies in the LPC
coupled enzyme assay and human plasma assay (Table 1). X-ray
crystallography studies with compound 1 revealed numerous
favorable interactions with the protein backbone (Figure 3).
Specifically, the indane group of 1 effectively occupies the
hydrophobic pocket and forms a T-shaped π−π interaction with
Phe273. In addition, donor and acceptor hydrogen bonds are
observed from the aminopyrimidine N,NH atoms to the Trp275
and Phe273 backbone residues. Finally, an additional π−π
interaction between the electron-rich phenol of Tyr306 and the
electron-deficient pyrimidine ring is observed.
a
Values are reported as arithmetic means with standard deviations.
b
Number of replicates in parentheses. Side chain pKa calculated using
Marvin pKa plugin (www.chemaxon.com, ChemAxon Kft, Budapest,
Hungary).
The crystal structure also revealed the terminal acetamide did
not interact with the protein backbone and was over 8 Å from the
catalytic zinc ion. Limited SAR within a closely related
aminopyrimidine series (data not shown) revealed that
substitution of either of the nitrogen atoms involved in hydrogen
bonding for carbon resulted in significantly diminished potency.
In 2010, researchers at Pfizer disclosed an inhibitor of
autotaxin known as PF-8380 (Chart 1). This inhibitor was
reported to have low nanomolar activity in an LPC enzyme assay
and 100 nM IC50 in a human whole blood assay.13 In our
laboratories, PF-8380 confirmed potency of 8.4 and 44 nM in
our LPC coupled enzyme and plasma assays, respectively. An
Figure 3. Cocrystal structure of 1 bound to rat autotaxin. Hydrogen
bond distances shown in Å.
internally solved crystal structure of PF-8380 with autotaxin is
shown in Figure 4. The lipophilic dichlorophenyl group of PF-
8380 anchors the molecule within the hydrophobic pocket with
B
DOI: 10.1021/acsmedchemlett.6b00207
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
Chart 1. Literature Autotaxin Inhibitor PF-8380
Figure 5. Cocrystal structure overlays of 1 and 2 bound to rat autotaxin.
Table 2. Hydrophobic Pocket SAR
Figure 4. Cocrystal structure of PF-8380 bound to rat autotaxin.
the carbamate carbonyl acting as a hydrogen bond acceptor to
Trp275 NH. Notably, PF-8380 extends through the hydrophilic
core and the benzoxazolone nitrogen engages the catalytic zinc
ion via an electrostatic interaction, presumably responsible for
the increased potency of PF-8380 relative to the amino-
pyrimidine 1.
In an effort to improve potency within our series, we also
recognized the potential to interact directly with the catalytic zinc
center. Indeed, coupling a propionylbenzoxazolone group onto
the core piperidine afforded compound 2, which demonstrated
2.5 nM IC50 in the LPC coupled enzyme assay but, similar to PF-
8380, notably weaker potency in the plasma assay (41 nM IC50).
A cocrystal structure of 2 with autotaxin demonstrated minimal
shift from the parent 1 with the flexible linker enabling the
benzoxazolone to interact with the catalytic zinc ion (Figure 5).
Despite achieving a significant increase in potency relative to
the ligand ID hit (1), compound 2 was plagued with poor
solubility, high microsomal metabolism, and very poor rat PK,
which precluded further development. Moreover, attempts to
improve physical properties and activity in a closely related series
through substitution of the indane for alternative hydrophobic
pocket binding groups was not successful as this led to
compounds with weaker plasma activity and higher cLogPs
(see Table 2).
However, encouraged by the potency breakthrough com-
pound 2 afforded, we examined a number of potential zinc-
binding groups as an alternative for the benzoxazolone with
varying linker lengths. Ultimately, a linker length of four atoms
proved ideal with longer linkers being equipotent and shorter
linker lengths being significantly less potent. Among those tested
(Table 1) were triazoles (3,4), tetrazole (5), imidazoles (6,7),
and pyrazole (8). After extensive exploration, the 1,2,3-triazoles
and 1- and 4-imidazoles emerged as the most potent compounds
a
Calcuated using Marvin pKa plugin (www.chemaxon.com, ChemAx-
on Kft, Budapest, Hungary). N = 1 for all LPC and Plasma IC50s.
with these alternative zinc-binding groups. While we were able to
obtain X-ray structures of these alkyl-linked heterocyclic
inhibitors, no electron density was ever observed for the linker
region, presumably due to the flexibility and mobility within this
section of the inhibitor.
A SAR analysis revealed a pronounced correlation between the
pKa of the zinc-binding group and the relative potency (LPC
IC50) with a maximum occurring at pKa ≈ 8.5−10.5 for acidic
heterocycles and pKa ≈ 6−8 for basic heterocycles. These
observations can potentially be explained using insights from
Duca and co-workers’ work on the metalloproteinase TNF-α
converting enzyme (TACE).14 They showed that binding of
hydroxamic and carboxylic acids to the Lewis-acidic zinc ion
within TACE results in a decrease in pKa of the binding group by
2−3 units. Extrapolation of these findings to our system led to
the following theory regarding optimal pKa for ATX inhibitors
(Figure 6). In the case of acidic heterocycles, in order to
maximize the binding affinity, the heterocycle would ideally
undergo deprotonation only upon coordination to the zinc ion
because prior deprotonation (e.g., pKa < ∼7.4) would impose a
higher desolvation penalty. Alternatively, if the pKa were too high
(e.g., pKa > ∼11), the Lewis acid assisted reduction in pKa would
C
DOI: 10.1021/acsmedchemlett.6b00207
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
Figure 6. Rationale of pKa dependency on autotaxin inhibitor potency.
be insufficient to enable deprotonation of the heterocycle at
physiological pH.
In the case of basic zinc-binders, the more basic heterocycles
are protonated and hence require greater desolvation energy at
physiologic pH. As a result, the ideal basic heterocycle would be
able to engage the zinc ion without significant energetic penalty
by being predominantly neutral at physiological pH (e.g., pKa <
∼8). However, as the pKBH decreases, binding to the zinc
becomes energetically less favorable due to an increasing
mismatch between the HOMO (nitrogen lone pair electrons)
and LUMO (zinc ion). Overall, this hypothesis correlates
empirically with the observed binding affinity for numerous
heterocycles.
Further optimization of this series culminated in the
identification of triazole 9. Indeed, substitution for the [6,5]-
pyrrolopyrimidine core ring system and inclusion of oxygen
within the alkyl linker improved the ADME properties and
afforded an inhibitor of autotaxin with 2 nM IC50 in both the
enzyme and plasma assays.
Figure 7. Rat PK/PD (10 mg/kg) experiment illustrates the correlation
between plasma drug concentration and reduction in plasma LPA.
a
Table 3. Pharmacokinetic Properties of 9 in Rat
parameters
compound 9
I.V. dose (mg/kg)
1.0
T
1/2, iv (h)
dss (L/kg)
1.14 0.12
1.4 0.37
10
V
PO dose (mg/kg)
max (nM)
C
660 120
3440 340
AUC (nM·h)
%F
23
4
a
Rat PK data is mean SD, n = 3. Formulation: PO, 40% HPBCD in
water; IV, DMA 10%/EtOH 15%/propylene glycol 30%/NaPO4
buffer (pH 8) 25 mM.
The synthesis of 9 was completed as shown in Scheme 1.
Commercially available chloropyrimidine 14 was coupled to 2-
aminoindane (15) via an SNAr reaction to give aminopyrimidine
16. Acid-mediated deprotection of the BOC group afforded the
dihydropyrrole 17, which was then acylated with 2-chloroacetyl
chloride to give the α-chloroacetamide 18. Finally, alkylation of
18 with 3-butyn-1-ol, gave the alkyne 19, which was converted to
triazole 9 via a copper(I)-catalyzed azide−alkyne cycloaddi-
tion.15
Encouraged by its in vitro profile, 9 was evaluated in a rat PK/
PD model that measured the reduction in plasma LPA vs drug
exposure (Figure 7, Table 3). We anticipated target engagement
and lower LPA levels when exposures exceeded the plasma IC80.
Due to the extremely short half-life of LPA in vivo,16 autotaxin
inhibition is accompanied by a nearly real-time reduction in
plasma LPA. After a single oral dose (10 mg/kg) in normal rats,
plasma concentrations of drug and LPA were monitored over 24
h. As shown in Figure 7, early time points demonstrated high
plasma concentrations of 9 and concomitant low concentrations
of LPA. As expected, as the plasma drug concentration decreased,
LPA levels returned to near normal levels. Plasma concentrations
of 9 exceeded the calculated IC80 for more than 12 h with
concurrent reduction in LPA over the same period. The robust
PK/PD relationship of 9 has translated into efficacy in a number
of animal models, the details of which have been reported in an
additional communication.17
In summary, we have utilized structural information to
facilitate the design of highly potent autotaxin inhibitors.
Through the course of this investigation, we identified a strong
correlation between the pKa of the zinc-binding group and the
a
Scheme 1. Synthesis of 9
a
Reagents and conditions: (a) iPr2EtN, NMP, 80 °C, 75%; (b) 5 N HCl(aq), THF, 50 °C, 80%; (c) 2-chloroacetyl chloride, Et3N, CH2Cl2, 23 °C,
84%; (d) 3-butyn-1-ol, NaH, THF, 0 °C, 23%; (e) CuSO4·5H2O, L-sodium ascorbate, TMSN3, DMF, H2O, 90 °C, 30%.
D
DOI: 10.1021/acsmedchemlett.6b00207
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
(2) Gabriel, S. E.; Crowson, C. S.; Campion, M. E.; O’Fallon, W. M.
Direct medical costs unique to people with arthritis. J. Rheumatol. 1997,
24, 719−25.
potency of the compounds. Further optimization of the core and
linker region of the inhibitors led to the identification of 9, a 2 nM
inhibitor of autotaxin in the human plasma assay with a robust
PK/PD relationship. Based partially on this data, compound 9
was identified as a valuable tool for investigating the role of
autotaxin in OA pain.
(3) Laufer, S. Osteoarthritis therapy-are there still unmet needs?
Rheumatology 2004, 43, i9−i15.
(4) Orosa, B.; Garcia, S.; Conde, C. The autotaxin-lysophosphatidic
acid pathway in pathogenesis of rheumatoid arthritis. Eur. J. Pharmacol.
2015, 765, 228−233.
ASSOCIATED CONTENT
* Supporting Information
(5) Bandoh, K.; Aoki, J.; Taira, A.; Tsujimoto, M.; Arai, H.; Inoue, K.
Lysophosphatidic acid (LPA) receptors of the EDG family are
differentially activated by LPA species. FEBS Lett. 2000, 478, 159−165.
(6) Mototani, H.; Lida, A.; Nakajima, M.; Furuichi, T.; Tsunoda, T.;
Sudo, A.; Kotani, A.; Uchida, A.; Ozaki, K.; Tanaka, Y.; Nakamura, Y.;
Tanaka, T.; Notoya, K.; Ikegawa, S. A functional SNP in EDG2 increases
susceptibility to knew osteoarthritis in Japanese. Hum. Mol. Genet. 2008,
17, 1790−1797.
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acsmedchem-
lett.6b00207.
Procedures for the preparation of 1−13, crystallization
and structural determination, and biological assays (PDF)
Autotaxin_1 (PDB)
Autotaxin_pf8380 (PDB)
Autotaxin_2 (PDB)
(7) Perrakis, A.; Moolenaar, W. H. Autotaxin: structure-function and
signaling. J. Lipid Res. 2014, 55, 1010−1018.
(8) Cai, Q.; Zhao, Z.; Antalis, C.; Yan, L.; Priore, G. D.; Hamed, A. H.;
Stehman, F. B.; Schilder, J. M.; Xu, Y. Elevated and secreted
phospholipase A2 activities as new potential therapeutic targets in
human epithelial ovarian cancer. FASEB J. 2012, 26, 3306−3320.
(9) Aoki, J.; Inoue, A.; Okudaira, S. Two pathways for lysophospha-
tidica acid production. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids 2008,
1781, 513−518.
Accession Codes
Coordinates and structure factors are available from the Protein
Data Bank with accession codes 5L0K, 5L0B, and 5L0E for
rATX/PF-8380, rATX/1, and rATX/2 binary complexes,
respectively.
(10) Stracke, M. L.; Krutzsch, H. C.; Unsworth, E. J.; Arestad, A.;
Cioce, V.; Schiffmann, E.; Liotta, L. A. Identification, purification, and
partial sequence analysis of autotaxin, a novel motility-stimulating
protein. J. Biol. Chem. 1992, 267, 2524−2529.
(11) Umezu-Goto, M.; Kishi, Y.; Taira, A.; Hama, K.; Dohmae, N.;
Takio, K.; Yamori, T.; Mills, G. B.; Inoue, K.; Aoki, J.; Arai, H. Autotaxin
has lysophospholipase D activity leading to tumor cell growth and
motility by lysophosphatidic acid production. J. Cell Biol. 2002, 158,
227−233.
(12) Nishimasu, H.; Okudaira, S.; Hama, K.; Mihara, E.; Dohmae, N.;
Inoue, A.; Ishitani, R.; Takagi, J.; Aoki, J.; Nureki, O. Crystal structure of
autotaxin an insight into GPCR activation by lipid mediators. Nat. Struct.
Mol. Biol. 2011, 18, 205−213.
AUTHOR INFORMATION
Corresponding Author
*E-mail: jonessp@lilly.com. Tel: +1-317-277-5543.
■
Present Address
‡Dow AgroSciences, Zionsville, Indiana 46268, United States.
Notes
The authors declare no competing financial interest.
†Retired.
(13) Gierse, J.; Thorarensen, A.; Beltey, K.; Bradshaw-Pierce, E.;
Cortes-Burgos, L.; Hall, T.; Johnston, A.; Murphy, M.; Nemirovskiy, O.;
Ogawa, S.; Pegg, L.; Pelc, M.; Prinsen, M.; Schnute, M.; Wendling, J.;
Wene, S.; Weinberg, R.; Wittwer, A.; Zweifel, B.; Masferrer, J. A novel
autotaxin inhibitor reduces lysophosphatidic acid levels in plasma and
the site of inflammation. J. Pharmacol. Exp. Ther. 2010, 334, 310−317.
(14) Cross, J. B.; Duca, J. S.; Kaminski, J. J.; Madison, V. S. The active
site of a zinc-dependent metalloproteinase influences the computed pKa
of ligands coordinated to the catalytic zinc ion. J. Am. Chem. Soc. 2002,
124, 11004−11007.
(15) Aher, N. G.; Pore, V. S.; Mishra, N. N.; Kumar, A.; Shukla, P. K.;
Sharma, A.; Bhat, M. K. Synthesis and antifungal activity of 1,2,3-triazole
containing fluconazole analogues. Bioorg. Med. Chem. Lett. 2009, 19,
759−763.
(16) Albers, H. M.; Dong, A.; van Meeteren, L. A.; Egan, D. A.;
Sunkara, M.; van Tilburg, E. W.; Schuurman, K.; van Tellingen, O.;
Morris, A. J.; Smyth, S. S.; Moolenar, W. H.; Ovaa, H. Boronic acid-
based inhibitor of autotaxin reveals rapid turnover of LPA in the
circulation. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 7257−7262.
(17) Thirunavukkarasu, K.; Swearingen, C.; Oskins, J.; Lin, C.; Bui, H.;
Jones, S.; Pfeifer, L.; Norman, B.; Mitchell, P.; Chambers, M. Manuscript
under review.
ACKNOWLEDGMENTS
■
Use of the Advanced Photon Source, an Office of Science User
Facility operated for the U.S. Department of Energy (DOE)
Office of Science by Argonne National Laboratory, was
supported by the U.S. DOE under Contract No. DE-AC02-
06CH11357. Use of the Lilly Research Laboratories Collabo-
rative Access Team (LRL-CAT) beamline at Sector 31 of the
Advanced Photon Source was provided by Eli Lilly and
Company, which operates the facility.
ABBREVIATIONS
■
OA, osteoarthritis; NSAID, nonsteroidal anti-inflamatory drug;
COX-2, cyclooxygenase-2; CV, cardiovascular; LPA, lysophos-
phatidic acid; GPCR, G protein-coupled receptor; ATX,
autotaxin; LPC, lysophosphatidylcholine; sPLA2, secreted
phospholipases A2; PA, phosphatidic acid; ID, identification;
compd, compound; SAR, structure−activity relationship; PK,
pharmacokinetics; TACE, TNF-a converting enzyme; HOMO,
highest occupied molecular orbital; LUMO, lowest unoccupied
molecular orbital; ADME, absorption distribution metabolism
excretion; PD, pharmacodynamics
REFERENCES
■
(1) Lawrence, R. C.; Felson, D. T.; Helmick, C. G.; Arnold, L. M.;
Choi, H.; Deyo, R. A.; Gabriel, S.; Hirsch, R.; Hochberg, M. C.; Hunder,
G. G.; Jordan, J. M.; Katz, J. N.; Kremers, H. M.; Wolfe, F. Estimates of
the prevalence of arthritis and other rheumatic conditions in the United
States Part II. Arthritis Rheum. 2008, 58, 26−35.
E
DOI: 10.1021/acsmedchemlett.6b00207
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX