Optimization of a pipemidic acid autotaxin inhibitor

Article abstract of DOI:10.1021/jm9012328

Autotaxin (ATX, NPP2) has recently been shown to be the lysophospholipase D responsible for synthesis of the bioactive lipid lysophosphatidic acid (LPA). LPA has a well-established role in cancer, and the production of LPA is consistent with the cancer-promoting actions of ATX. Increased ATX and LPA receptor expression have been found in numerous cancer cell types. The current study has combined ligand-based computational approaches (binary quantitative structure-activity relationship), medicinal chemistry, and experimental enzymatic assays to optimize a previously identified small molecule ATX inhibitor, H2L 7905958 (1). Seventy prospective analogs were analyzed via computational screening, from which 30 promising compounds were synthesized and screened to assess efficacy, potency, and mechanism of inhibition. This approach has identified four analogs as potent as or more potent than the lead. The most potent analog displayed an IC₅₀ of 900 nM with respect to ATX-mediated FS-3 hydrolysis with a K_i of 700 nM, making this compound approximately 3-fold more potent than the previously described lead.

Full text of DOI:10.1021/jm9012328

1056 J. Med. Chem. 2010, 53, 1056–1066
DOI: 10.1021/jm9012328
Optimization of a Pipemidic Acid Autotaxin Inhibitor
Adrienne B. Hoeglund,†,‡ Heidi E. Bostic,§ Angela L. Howard,†,‡ Irene W. Wanjala,†,‡
Michael D. Best,§ Daniel L. Baker,*,† and Abby L. Parrill*,†,‡
†Department of Chemistry, and ‡Computational Research on Materials Institute, The University of Memphis, Memphis,
Tennessee 38152, and §Department of Chemistry, The University of Tennessee, Knoxville, Knoxville, Tennessee 37996
Received August 18, 2009
Autotaxin (ATX, NPP2) has recently been shown to be the lysophospholipase D responsible for
synthesis of the bioactive lipid lysophosphatidic acid (LPA). LPA has a well-established role in cancer,
and the production of LPA is consistent with the cancer-promoting actions of ATX. Increased ATX and
LPA receptor expression have been found in numerous cancer cell types. The current study has
combined ligand-based computational approaches (binary quantitative structure-activity relation-
ship), medicinal chemistry, and experimental enzymatic assays to optimize a previously identified small
molecule ATX inhibitor, H2L 7905958 (1). Seventy prospective analogs were analyzed via computa-
tional screening, from which 30 promising compounds were synthesized and screened to assess efficacy,
potency, and mechanism of inhibition. This approach has identified four analogs as potent as or more
potent than the lead. The most potent analog displayed an IC50 of 900 nM with respect to ATX-
mediated FS-3 hydrolysis with a Ki of 700 nM, making this compound approximately 3-fold more
potent than the previously described lead.
Introduction
drugs,41,42 and collectively they lack significant structural
diversity. Finally, the third category of reported ATX inhibi-
tors consists of nonlipid small molecules that collectively
extend structural diversity and in general possess physico-
chemical characteristics more closely related to orally bioa-
vailable drugs. Our group was the first to report the identi-
ficationof smallmoleculeATX inhibitorsusingvirtualscreen-
ing (binary QSAR and homology modeling) approaches.43
The most efficacious structures identified from that work are
shown in Figure 2A.43 H2L 7905958 (1) was the most effica-
cious compound from that initial single concentration screen
(at 10 μM, compound 1 fully inhibited ATX-catalyzed hydro-
lysis of 1 μM FS-3).43 Recently, Saunders and colleagues
identified additional small molecule inhibitors of ATX
(Figure 2B,C) as metastasis blockers.44 One compound
showed essentially 100% inhibition of melanoma metastasis
at micromolar concentrations.44
In this study, we investigated the structure-activity rela-
tionships of our most efficacious previously published small
molecule ATX inhibitor, compound 1 (Figure 3A, Chem-
Bridge, San Diego, CA).43 Using a combination of binary
QSAR modeling and synthetic optimization, we prepared 30
analogs of compound 1 and subjected them to pharmacolo-
gical characterization using the FRET-based, synthetic ATX
substrate FS-3.45 Four of these 30 compounds exhibited IC50
values less than or equal to the lead. Despite their structural
similarity, three of four reversibleclassesofinhibition(compe-
titive, mixed-mode, and noncompetitive) were observed with-
in this series. Inhibitors in these three classes share the ability
to bind to the enzyme in the absence of substrate, reflected by
Ki. Mixed-mode and noncompetitive inhibitors additionally
have the ability to bind to the enzyme-substrate complex,
reflected by Ki0. The Ki values, due to their independence of
substrate identity, suggest that the inhibition observed will
Autotaxin (ATX, NPP2)a was originally identified as an
autocrine motility factor in the conditioned media of A2058
melanoma cells.1 Subsequently, ATX was shown to be the
lysophospholipase D enzyme responsible for synthesis of the
bioactive lipid lysophosphatidic acid (LPA) in vivo.2,3 Recent
work has shown that ATX plays critical roles in normal
human development4-7 and disease. ATX has been linked
to cancer progression,8-20 multiple sclerosis,21 obesity,22-26
diabetes,22 Alzheimer’s disease,27 and chronic pain28-30
through the production of LPA. Specific, potent inhibitors
of ATX are therefore desirable as novel therapeutic
leads.
Examples of metal chelators, lipid analogs, and nonlipid
small molecules have all been identified as autotaxin inhibi-
tors. Metal chelators such as EDTA,31 phenanthroline,31 and
L-histidine32 have been shown to inhibit ATX activity, pre-
sumably via interactions with active site divalent metal ions
required for function. Lipid analogs represent the largest
group of reported ATX inhibitors (see Figure 1 for struc-
tures). LPA and the related bioactive lipid sphingosine 1-
phosphate (S1P) were previously shown to function as feed-
back inhibitors of ATX.33 This discovery led to the analysis of
several LPA and S1P analogs as ATX inhibitors. Reported
LPA analogs include fatty alcohol phosphates,34 Darmstoff
analogs,35 cyclic phosphatidic acid analogs,36 and phos-
phonates.37-39 One reported S1P analog, FTY720-phosphate
(32), has also been examined as an ATX inhibitor.40 While
many of these lipid analogs are potent ATX inhibitors, they
lack many characteristics seen in 90% of orally bioavailable
*Corresponding Authors: Phone: 901-678-2638 (ALP), 901-678-4178
(DLB). Fax: 901-678-3447. E-mail addresses: aparrill@memphis.edu,
dlbaker@memphis.edu.
r
pubs.acs.org/jmc
Published on Web 12/30/2009
2009 American Chemical Society

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3 1057
Figure 1. Structures of published lipid-like ATX inhibitors.33-38,40,54,55
transfer to inhibition of ATX-catalyzed inhibition of its
natural substrate, lysophosphatidylcholine (LPC).
(inactive) and 60 showed ATX inhibition g50% at 10 μM
(active). An initial set of nine compounds, containing either
ortho or para substituents (Figure 5B), was synthesized based
on these predicted activities. Experimental assays using a
single dose of 10 μM demonstrated that all nine compounds
were less efficacious than the parent (Table 1) and only two
showed g50% inhibition. This data suggests that model A,
based on a diverse training set, was unable to accurately
predict activity within this narrow structural series.
A more class-specific binary QSAR model (model B) was
constructed for use in lead optimization. This refined model
was trained on 248 compounds of which 204 were inactive and
44 were active. The training set for model B differed from that
used for model A in three ways. First, it lacked lipid-like ATX
inhibitors (Figure 2). Second, the nine initial analogs utilized
as the test set for model A were added (Figure 5B). Third, it
was augmented with an additional 12 analogs of compound 1
(Figure 5C) synthesized to better represent diversity within
this SAR series. The total internal accuracy of the model was
86%, with 59% accuracy for actives and 91% accuracy for
inactives.
Results
In 2008, we identified compound 1 (Figure 3A) as a potent
ATX inhibitor.43 This compound(10 μM) completelyblocked
ATX-mediated hydrolysis of the synthetic FRET reagent
FS-3 (1 μM). Following this initial report, we determined that
compound 1 inhibited ATX with an IC50 of 1.6 ( 0.4 μM
(Figure 3B). Simultaneous nonlinear regression of the inhi-
bition kinetics data shown in Figure 3C, using an average Km
value of 3.7 ( 1.9 μM (n = 22) produced the best fit to a
competitive inhibitionmechanism, with a Ki of 1.9 μM. With a
low micromolar inhibitor lead in hand, we next sought a
convenient diversification strategy to generate structure-
-activity relationship (SAR) data in hopes of better under-
standing ATX inhibitor recognition and identifying more
potent analogs. Here, the derivatization of pipemidic acid
with commercially available isothiocyanate building blocks
was identified as a facile route to a range of analogs of
compound 1 bearing various substituents on the phe-
nylthiourea motif (Figure 4).
The predictive accuracy of models A and B were examined
using an additional nine compound test set (Figure 5D). Both
models predicted four of the nine compounds (with only one
of the four being the same analog) to be active. Screening
results showed that model A was 67% accurate in predicting
active compounds and 33% accurate on both inactive com-
pounds and total accuracy. These predictions fell short of
expectations based on the training set results. Overall, model
Using this diversification approach, 70 proposed synthetic
analogs of compound 1 (Figure 5 and Supplemental Figure 1,
Supporting Information) were evaluated and prioritized using
a general binary QSAR model43 (model A) before they were
synthesized. This general model was trained on a diverse set of
compounds (298 total compounds) including both lipid and
nonlipid analogs. Of these, 238 lacked ATX inhibition

1058 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3
Hoeglund et al.
Figure 2. Structures of published nonlipid ATX inhibitors: (A) H2L compounds;43 (B) structures of hexachlorophene, merbromine, bithionol,
and their analogs;44 (C) ATX inhibitors identified from the NCI database.44
A failed to be predictive for this closely related SAR series.
Model B was shown to be 75% accurate in predicting active
compounds and 40% accurate on inactive compounds, with a
total accuracy of 56%. These predictions were consistent with
expectations based on the training set results.
Of the original 70 proposed analogs (Figure 5 and Supple-
mental Figure S1, Supporting Information), 30 compounds
were synthesized (Figure 5) based on computational results.
These analogs provided diverse substituent types and substi-
tution patterns (Table 1) to serve as a training set suitable for
specific activity prediction within this SAR series. The result-
ing analogs were examined for their ability to inhibit ATX-
mediated hydrolysis of FS-3 at 10 μM. Likewise, each analog
was tested in the presence of carboxyfluorescein (the product
of ATX-mediated hydrolysis of FS-3) to identify false positive
or negative results. None of the analogs tested signifi-
cantly altered carboxyfluorescein fluorescence. Compounds
containing meta substituents demonstrated ATX inhibition in
the range of 60-100%. Compound 11 (Figure 5C), with a
trifluoromethyl group in the meta position, yielded 100%
inhibition. A comparative assay of the 15 analogs showing
>50% inhibition was performed against both CCM and
purified ATX (8.3 nM enzyme concentration) to confirm that
relative potencies were independent of enzyme source. The
results showed that the potency rank of the inhibitors was
unchanged and that the results were within (5% (data not
shown).
The selectivities ofATX inhibitors 3, 7, 11, 13, 14, 18, 19, 20,
22, 23, 25, 27, 28, 29, and 31, all of which inhibited ATX-
mediated hydrolysis of FS-3 by g50% at single doses of
10 μM, were examined against NPP6 and NPP7, the only
other known lipid preferring NPP isoforms. None of the 15
compounds effected the activity of NPP6 or NPP7 (Supple-
mental Figure S2, Supporting Information). Likewise, each

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3 1059
Figure 3. Structure (panel A) and pharmacology of 1: (B) dose-response determined using 1 μM FS-3 gives IC50 = 1.6 μM (IC50 values were
within (3%, determination was based on average of two runs); (C) inhibitor effect on kinetics of FS-3 hydrolysis by ATX best fits competitive
inhibition (Ki = 1900 nM).
Figure 4. Synthetic scheme for the synthesis of analogs of 1 using commercially available pipemidic acid and isothiocyanates.
analog was tested in the presence of p-nitrophenol (the
product of ATX-mediated hydrolysis of p-nitrophenylpho-
sphorylcholine (pNPPC)) to eliminate the chance of false
positive or false negative results. None of the analogs tested
had a significant effect on the p-nitrophenol absorbance.
Dose-response curves were determined for all 15 analogs
that inhibited ATX-mediated hydrolysis of FS-3 by g50% at
10 μM. IC50 values in a range of 900 nM to 17.7 μM were
observed (Figure 6 and Table 1). Compounds demonstrating
an IC50 less than 15 μM were further examined to determine
their mechanism of inhibition by simultaneous nonlinear
regression of kinetic data to competitive, uncompetitive,
mixed, and noncompetitive kinetic inhibition equations. The
mechanism giving the lowest average percent residual errors
was selected. Ki (competitive, mixed, or noncompetitive) and
Ki0 (mixed, noncompetitive, or uncompetitive) values, repre-
senting affinity for the enzyme or the enzyme-substrate
complex, respectively, were obtained from the simultaneous
nonlinear regression (Table 1). None of the inhibitors proved
to act by uncompetitive inhibition; therefore each was capable
of binding to the enzyme in the absence of substrate. Com-
pounds 7, 11, 13, 18, 22, and 28 were competitive ATX
inhibitors with Ki values ranging from 700 to 7100 nM.
Analog 11 has 3 times greater affinity for the enzyme than
the lead. Compounds 3, 14, 20, and 29 were mixed-mode ATX
inhibitors with Ki values ranging from 2.6 to 13.2 μM. These
compounds showed reduced affinity for the enzyme-sub-
strate complex compared with their affinity for the enzyme as
reflected in higher Ki0 values ranging from 11.0 to 47.3 μM.
Compound31showednoncompetitiveinhibition withaKi/Ki0
value of 13.2 μM. Themixedandnoncompetitiveinhibitors,3,
14, 20, 29, and 31, have the ability to bind to the enzyme-sub-
strate complex as well as the enzyme. The noncompetitive
inhibitor, 31, binds with identical affinity to both the enzyme
and enzyme-substrate complex.
Table 2 provides a summary of single-dose data that
provides SAR insights. These data clearly demonstrate that
inhibition is greatest when substituents appear at the meta
position, followed by para with the weakest inhibition at the
ortho position. This trend is consistent for all substituents
examined.
Discussion
Interest in ATX as a therapeutic target is growing due to its
association with a number of human diseases including
various cancers.8-30 This interest is reflected in a rising
number of reported ATX inhibitors with potencies in the
nanomolar range (Figure 1). Carbacyclic analogs of cyclic
phosphatidic acid (3ccPA) were demonstrated to be potent

1060 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3
Hoeglund et al.
Figure 5. Structures of the 30 synthetic analogs of compound 1: (A) common structural core where R represents the components seen in panels
B-D; (B) compounds synthesized based on initial predictions using binary QSAR model A; (C) compounds synthesized to augment binary
QSAR training set for model B; (D) compounds synthesized to serve as the binary QSAR test set for model B.
inhibitors of ATX that failed to activate LPA GPCR.36 ATX
inhibition by 3ccPA translated to reduced cancer cell invasion
in vitro and reduction of cancer cell metastasis in vivo.36 The
LPA analog, fluoromethylene phosphonate, was shown to be
an effective ATX inhibitor, while also a modest LPA1/3
antagonist.39 Darmstoff analogs of LPA were examined for
their potential as LPA GPCR agonists and were shown to
inhibit ATX in the nanomolar range.35 Compound 32, a
phosphorylated immunomodulator, was shown to inhibit
ATX-mediated production of LPA in vivo.40 Another lipid
phosphonate analog, S32826 (Figure 1), was reported to have
potency in the nanomolar range.38 β-hydroxy and β-keto
phosphonate derivatives of LPA were studied as potential
ATX inhibitors. VPC8a202 (Figure 1), a β-hydroxy deriva-
tive, was identified as a potent ATX inhibitor.37 A common
structural theme among these potent ATX inhibitors is their
overalllipidnature, withpolar headgroupsandnonpolartails.
The development of highly potent nonlipid inhibitors has
beenhamperedby limited information about the three-dimen-
sional structure of ATX. Nevertheless, several groups have
reported nonlipid inhibitors,44,46 albeit with much poorer
potencies than the current lipid-like inhibitors. The current
work focused on elaborating the SAR for compound 1, a
previously identified, nonlipid ATX inhibitor, in our efforts to
better understand the structure of the binding pocket, as well
as to improve the potency of a known nonlipid inhibitor.43
For this study, 70 potential analogs of 1 were designed based
on a one-step synthetic scheme benefiting from commercially
available pipemidic acid and substituted phenyl isothiocya-
nates. In total, 30 compounds were synthesized and evaluated.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3 1061
Table 1. SAR Elaboration and Optimization of Lead Compound 1a
substitution on D ring
% response at 10 μM (FS-3)
IC50 (μM)
inhibition mechanism
Ki (μM)
Ki0 (μM)
lead - dichloro (R2, R4) (1)
trifluoromethyl (meta - R2) (11)
dichloro (R1, R4) (20)
0.0 ( 4.0
0.3 ( 0.6
5.4 ( 0.4
8.6 ( 0.9
9.7 ( 0.9
1.6 ( 0.4
0.9 ( 0.3
1.3 ( 0.2
1.6
competitive
competitive
mixed
1.9
0.70
3.4
22.0
11.0
47.3
iodo (meta - R2) (29)
mixed
mixed
7.4
2.6
chloro (meta - R2) (14)
thiadiazole (R2,R3) (7)
dimethyl (R2, R4) (22)
iodo (para - R3) (3)
2.5 ( 0.2
12.1 ( 2.4
9.0
28.6 ( 2.0
30.4 ( 1.3
32.7 ( 1.3
34.1 ( 3.0
36.2 ( 0.9
40.5 ( 1.0
40.9 ( 1.3
41.9 ( 0.8
42.4 ( 0.9
48.9 ( 1.1
49.9 ( 1.5
51.2 ( 1.3
53.5 ( 1.6
56.4 ( 1.0
59.9 ( 1.5
63.2 ( 2.8
64.1 ( 2.8
68.4 ( 1.3
68.6 ( 1.1
72.3 ( 0.4
72.7 ( 2.2
77.3 ( 1.6
79.7 ( 0.9
83.5 ( 8.2
97.6 ( 1.8
100.8 ( 1.9
competitive
competitive
mixed
7.1
4.2
5.6 ( 0.2
4.4 ( 1.5
1.5 ( 0.1
16.1 ( 4.5
17.7 ( 0.3
13.5 ( 1.3
10.7
13.2
4.0
34.1
ditrifluoromethyl (R2, R4) (18)
trifluoromethyl (para - R3) (31)
fluoro (meta - R2) (25)
methoxy (meta - R2) (19)
methyl (meta - R2) (13)
chloro (para - R3) (28)
fluoro (para - R3) (23)
methyl (para - R3) (27)
o-methyl ester (5)
competitive
noncompetitive
13.2
13.2
competitive
competitive
4.5
5.7
17.7
16.1 ( 4
trimethyl (R1, R3, R5) (8)
H (16)
4-dimethylaminonapthyl (2)
thiadiazole (R1,R2) (10)
sulfonylmorpholine (para - R3) (6)
iodo (ortho - R1) (26)
pyrazole (para - R3) (9)
dichloro (R1, R2) (15)
p-carboxyl (4)
fluoro (ortho - R1) (30)
methoxy (para - R3) (24)
chloro (ortho - R1) (17)
Methoxy (ortho - R1) (21)
trifluoromethyl (ortho - R1) (12)
a Compounds showing equal or improved IC50 (bold) or Ki (italic) values over the lead are emphasized. Values shown in parentheses refer to the
compound numbers (Figure 5). Errors where indicated are standard deviations of at least three independent experiments. IC50 values lacking standard
deviations are averages of two independent experiments that were less than 3% different in value.
Table 2. Single-Dose Response Comparison by Aromatic Substituent
and Positiona
meta-R2
para-R3
ortho-R1
trifluoromethyl
iodo
chloro
0.3 ( 0.6 (11)
8.6 ( 0.9 (29)
9.7 ( 0.9 (14)
40.5 ( 1.0 (25)
40.9 ( 1.3 (19)
36.2 ( 0.9 (31)
32.7 ( 1.3 (3)
42.4 ( 0.9 (28)
48.9 ( 1.1 (23)
79.7 ( 0.9 (24)
100.8 ( 1.9 (12)
68.4 ( 1.3 (26)
83.5 ( 8.2 (17)
77.3 ( 1.6 (30)
97.6 ( 1.8 (21)
fluoro
methoxy
a Values shown in parentheses refer to the compound numbers
(Figure 5).
Table 2 summarizes two important trends that we observed
with these analogs. First, inhibition improves going from
ortho to para to meta substituent positions regardless of the
substituent type. This suggests that meta substituents are
preferred for steric or conformational, rather than electronic
reasons. Second, in the meta-substituted series inhibition
improves in the order of methoxy, fluoro, chloro, iodo, and
trifluoromethyl. This order reflects neither size nor electronic
trends among these substituents. Four compounds (11, 20, 29,
Figure 6. Dose-response plots of the four compounds (11, 20, 29,
and 31) equivalent to or more potent than the lead compound.

1062 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3
Hoeglund et al.
and 31) were at least as potent as the lead compound (1). A
single meta-trifluoromethyl (11, Ki=0.70 μM) was preferred
over two (18, Ki=4.0 μM), Table 1. In fact, meta-trifluoro-
methyl (11) showed three times higher affinity for the enzyme
than the lead compound, while the disubstituted trifluoro-
methyl (18) exhibited 2-fold less enzyme affinity than the lead.
This suggests that one of the two meta substituents is located
in a substantially larger pocket than the other.
These SAR trends inform the development of computa-
tional tools aimed toward the identification and optimization
of ATX inhibitors such as our previously reported binary
QSAR model.43 An updated binary QSAR model was deve-
loped for this work by the addition of more nonlipid and lipid-
like compounds (published since our initial report) in an
attempt to improve the accuracy of prediction (model A).
We found this model unable to accurately predict activity
differences due to the very small structural changes examined
in this work. This is in large part due to its originally intended
use to screen diverse chemical collections for leads with varied
structures. To circumvent this issue, a more class-specific
model for lead optimization was designed (model B). When
predicted against the training set, model B showed an 11%
improvement over model A in predicting active compounds,
18% in inactive prediction, and an overall 41% improvement
in total prediction. Further improvements are expected, how-
ever, when training sets can be segregated on the basis of
mechanism of inhibition.
Previously published research defines inhibition mechan-
isms for a relatively small proportion of the known inhibitors.
Inhibitors with reported mechanism of inhibition include
L-histidine,32 LPA and S1P,33 32,40 and nonlipid small mole-
cules.44 L-Histidine demonstrates noncompetitive inhibition.
LPA and S1P demonstrated mixed-mode inhibition with LPA
having a Ki of ∼0.1 μM.33 van Meeteren and colleagues also
reported that 32 inhibited ATX in the low micromolar range
by a competitive mechanism with a Ki of 0.2 μM.40 Finally,
Saunders and colleagues demonstrated nanomolar potency
for a nonlipid small molecule competitive ATX inhibitor
(Figure 2C).44 Within the series of analogs synthesized here,
we observed competitive, mixed mode, and noncompetitive
inhibition mechanisms. All inhibitors tested showed affinity
for the enzyme in the absence of substrate, so the results
suggest that these inhibitors will be effective in assays reflect-
ing inhibition of LPC hydrolysis. Future computational tools
to be developed using the Ki values reported here include
traditional QSAR models, which not only will be able to
predict likelihood that candidates will exhibit inhibition better
than a threshold value but will actually be able to predict the
affinity for the enzyme.
Inhibitor specificity in the NPP family has previously not
been examined. Three members of the NPP enzyme family,
NPP2(ATX), NPP6, andNPP7, share the ability tohydrolyze
lysophospholipids47 and therefore might be expected to have
some overlap in their recognition of inhibitors. Supplemental
Figure 2, Supporting Information, illustrates the selectivity of
compounds inhibiting at least 50% of ATX activity (Table 1)
over NPP6 and NPP7. These analogs clearly do not inhibit
NPP6 or NPP7 at a single 10 μM dose. These inhibitors are
therefore selective for ATX among the NPP isoforms with
demonstrated preference for phospholipid substrates.
analogs were characterized as ATX inhibitors using single-
dose screening, analysis of selectivity against other lipid
preferring NPP isoforms, dose-response determinations,
and mechanism of inhibition analysis (Ki determination).
The 15 analogs that inhibited ATX activity g50% at 10 μM
were shown to be ATX selective. Eight analogs exhibited IC50
for ATX-mediated FS-3 hydrolysis less than 10 μM. Six of
these analogs exhibited competitive ATX inhibition with Ki
values ranging from 700 nM to 7.1 μM. The best of these
compounds has a potency (Ki =700 nM) approaching that
reported for some of the most active published nonlipid ATX
inhibitors.
Experimental Section
Binary QSAR Model Development. Binary QSAR model A
was developed from the expansion of model 1 in ref 43. Model A
was trained on a diverse set of 298 compounds (60 active and 238
inactive) consisting of both lipid-like analogs and nonlipids.
Model B was trained on 248 compounds of which 204 were
inactive and 44 were active. This training set did not contain the
lipid-like compounds present in model A. The compounds in
models A and B were not corrected for explicit hydrogen atoms
or charges on ionizable groups. Compounds were classified
active if at least 50% inhibition was observed at a concentration
of 10 μM. Over 220 structural descriptors, available in the
current version of the MOE software, were calculated for each
compound (both models). Descriptors that were dependent
upon the orientation of the compound in a coordinate system
and those involving semiempirical quantum mechanics were not
included in the analysis. The individual descriptors were reduced
to a set of orthogonal principal components. The binary models
were fitted using seven principal components.
Determination of ATX Inhibition. ATX inhibition was deter-
mined using the synthetic FRET-based substrate FS-3 (Echelon
Biosciences, Inc., Salt Lake City, UT). Concentrated (∼10ꢀ)
conditioned medium (CCM) from MDA-MB-435 cells was used
as the source of ATX. The final volume (60 μL) included 20 μL
of CCM, 20 μL of inhibitor (10 μM for single point experiments
or 30 nM to 30 μM for dose-response experiments, including
1% DMSO) in assay buffer (1 mM each CaCl2 and MgCl2,
5 mM KCl, 140 mM NaCl, 50 mM Tris, pH 8.0), and 20 μL of
FS-3, 1 μM in assay buffer containing charcoal-stripped,48 fatty
acid free BSA (30 μM) (Sigma Aldrich, St. Louis, MO). All
assays were performed in 96-well, half-area plates (Corning Inc.,
Lowell, MA) at 37 °C with data collected at 2 min intervals using
a Synergy2 fluorescence plate reader (BioTek, Winooski, VT).
Fluorescence resulting from the hydrolysis of FS-3 was mon-
itored using excitation at 485 nm and emission at 528 nm.45
Results were derived from the 1 h time point, where all fluore-
scence changes were linear as a function of time. All readings
were normalized to vehicle control after subtraction of fluores-
cence in the absence of CCM. Carboxyfluoroscein (Sigma-
Aldrich, St. Louis, MO) in the absence of ATX was used as an
analog of the FS-3 hydrolytic fluorophore in experiments to
confirm that the analogs did not provide false positive (signal
quenching) or false negative (signal enhancing) results, respec-
tively. Final volumes (60 μL) consisted of 20 μL of carboxy-
fluoroscein (200 nM) with 20 μL of each analog (10 μM), and
20 μL of charcoal-stripped BSA (30 μM) in assay buffer as
described. Data were collected in a single point analysis using
excitation at 485 nm and emission at 528 nm. Data are shown as
the mean ( SD of at least three wells. All experiments were
repeated at least twice, and representative results are shown.
Mechanism of Inhibition. ATX kinetics assays were performed
using substrate concentrations from 300 nM to 20 μM and three
different concentrations of inhibitor (corresponding to 0, 0.5,
and 2 times the IC50 for each analog). Normalized fluorescence
results were plotted as a function of time to determine initial
Conclusions
Thirty out of 70 potential analogs of compound 1 were
synthesized based on binary QSAR predictions. These 30

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3 1063
rates. These initial rates for the zero inhibitor concentra-
tion were plotted against the substrate concentration and a
rectangular hyperbolic curve was fitted to the data using
KaleidaGraph (version 4.03, Synergy Software, Reading, PA)
to determine Km. Data for all three inhibitor concentra-
tions were simultaneously fitted in the Michealis-Menton
equations for competitive, uncompetitive, mixed-mode, and
noncompetitive inhibition (shown below) using simultaneous
nonlinear regression for each model using WinNonLin 6.1
(Pharsight, Mountain View, CA). Mechanism of inhibition
was assigned to the model giving the lowest averaged percent
residuals. Ki and Ki0 values represent compound affinity for
free enzyme and the enzyme-substrate complex, respectively.
(NPP6 and NPP7) and concentrated, ∼10ꢀ (NPP7), using
10 kDa molecular weight cutoff filters (Millipore, Beverly,
MA) in an Amicon pressure cell (Millipore, Beverly, MA).
General Procedure for the Synthesis of Prospective Inhibitors
Shown in Figure 4. To a solution of pipemidic acid in N,N-
dimethylformamide (1 mL) was added the corresponding sub-
stituted phenyl isothiocyanate. In some cases, triethylamine was
added to speed the reaction, while in others, the addition of this
base led to the formation of byproducts and was thus avoided.
The mixture was then allowed to stir at room temperature under
nitrogen overnight. The solvent was then removed in vacuo, and
the residue was extracted twice with methylene chloride from
aqueous saturated ammonium chloride. The organic layers were
combined, dried with magnesium sulfate, filtered, and concen-
trated. The resulting crude product was then purified by flash
chromatography on silica gel eluting with methanol in methy-
lene chloride (0-20%) to yield the corresponding thiourea
products. All synthetic compounds were determined to be
>95% pure by HPLC.
Vmax½Sꢁ
½Iꢁ
For competitive inhibition : V0
¼
,
where R ¼ 1 þ
RKm þ ½Sꢁ
Ki
Vmax½Sꢁ
½Iꢁ
For uncompetitive inhibition : V0
¼
¼
þ R0½Sꢁ, where R0 ¼ 1 þ
Ki0
Km
Vmax½Sꢁ
For mixed-mode inhibition : V0
RKm þ R0½Sꢁ
Vmax½Sꢁ
¼
RðKm þ ½SꢁÞ
For noncompetitive inhibition; R ¼ R0; therefore : V0
2-(4-{[(4-Dimethylaminonapthylphenyl)amino]carbonothioyl}-
1-piperazinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-
6-carboxylic Acid (2). Pipemidic acid (25 mg, 0.0824 mmol),
4-dimethylamino-1-napthylisothiocyanate (18.8 mg, 0.0824
mmol), and triethylamine (23 μL, 0.165 mmol) were used.
ATX Expression. MDA-MB-435 cells were cultured at 37 °C
under a humidified atmosphere containing 5% CO2 in Dulbec-
co’s modified Eagle medium (DMEM) (MediaTech, Herndon,
VA) containing 100 U/mL penicillin, 100 μg/mL streptomycin
(Hyclone, Logan, UT), 5% fetal bovine serum (FBS) (Hyclone,
Logan, UT), and 2 mM L-glutamine (Hyclone, Logan, UT).
Upon reaching 80-95% confluency, the cells were washed twice
with sterile phosphate-buffered saline. Serum-free DMEM con-
taining L-glutamine and antibiotics was then added to the cells.
Conditioned media was collected after 36-48 h of incubation
with serum-free DMEM. The media was concentrated, ∼10ꢀ,
and buffer-exchanged into Tris (50 mM, pH 7.4) containing
20% ethylene glycol using 30 kDa molecular weight cutoff filters
(Millipore, Beverly, MA) in an Amicon pressure cell (Millipore,
Beverly, MA). Aliquots of 10ꢀ conditioned media were stored
at 4 °C.
Determination of NPP6 and NPP7 Inhibition. Inhibitor selecti-
vity was assayed using p-nitrophenylphosphoryl-choline (pNPPC)
(Sigma-Aldrich, St. Louis, MO) as substrate for NPP6 and NPP7.
The final volume (60 μL) included 20 μL of CM, NPP6, or CCM,
NPP7, 20 μL of inhibitor (10 μM including 1% DMSO) in assay
buffer, and 20 μL of pNPPC (10 mM for NPP6 or 1 μM for NPP7)
in assay buffer. NPP6 assay buffer contained 500 mM NaCl,
0.05% Triton X-100, and 100 mM Tris-HCl (pH 9.0),49 whereas
NPP7 assay buffer consisted of 50 mM Tris HCl, pH 8.5, 150 mM
NaCl, and 10 mM taurocholic acid.50-53 However, unlike the
published methods, EDTA was not added to the NPP7 assay
buffer. All assays were performed in 96-well, half-area plates
(Corning Inc., Lowell, MA) at 37 °C with data collected at
2 min intervals using a Synergy2 absorbance plate reader
(BioTek, Winooski, VT). Results are shown at the 1 h time point,
when all absorbance changes as a function of time were linear.
Readings were normalized to vehicle control after subtraction of
absorbance in the absence of CM or CCM. Data are shown as the
mean ( SD of at least three wells. All experiments were repeated
twice, and representative results are shown.
1
Purification on silica yielded 2 (23 mg, 53%). H NMR (300
MHz, CDCl3) δ 9.29 (s, 1H), 8.63 (s, 1H), 8.32-8.24 (m, 1H),
8.02-7.93 (m, 1H), 7.62-7.46 (m, 3H), 7.02 (d, J = 7.95 Hz,
1H), 4.29 (q, J = 6.90 Hz, 2H), 4.16-3.85 (m, 8H), 2.90 (s, 6H),
1.45 (t, J = 7.08 Hz, 3H) ppm.
2-(4-{[(4-Iodophenyl)amino]carbonothioyl}-1-piperazinyl)-8-
ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (3). Pipemidic acid (25 mg, 0.0824 mmol), 4-iodophenyl
isothiocyanate (21.5 mg, 0.0824 mmol), and triethylamine
(23 μL, 0.165 mmol) were used. Purification on silica yielded 3
(36 mg, 77%). 1H NMR (300 MHz, CDCl3) δ 9.32 (s, 1H), 8.66
(s, 1H), 7.66 (d, J=6.94 Hz, 1H), 7.37 (s, 1H), 7.27 (s, 1H), 7.00
(d, J=7.44 Hz, 1H), 4.33 (q, J= 6.90 Hz, 2H), 4.21-3.99 (m, 8H),
1.49 (t, J = 7.14 Hz, 3H) ppm.
2-(4-{[(4-Carboxylphenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (4). Pipemidic acid (25 mg, 0.0824 mmol), 4-isothiocyana-
tobenzoic acid (14.8 mg, 0.0824 mmol), and triethylamine (23 μL,
0.165 mmol) were used. Purification on silica yielded 4 (23 mg,
58%). 1H NMR (300 MHz, DMSO) δ 9.71 (s, 1H), 9.24 (s, 1H),
8.98 (s, 1H), 7.87 (d, J = 8.53 Hz, 2H), 7.49 (d, J = 8.43 Hz, 2H),
4.45-4.39 (m, 2H), 4.10-4.03 (m, 8H), 1.38 (t, J = 6.94 Hz, 3H)
ppm.
2-(4-{[(2-Methylesterphenyl)amino]carbonothioyl}-1-pipera-
zinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-car-
boxylic Acid (5). Pipemidic acid (25 mg, 0.0824 mmol), methyl
2-isothiocyanatobenzoate (16 mg, 0.0824 mmol), and triethyla-
mine (23 μL, 0.165 mmol) were used. Purification on silica
1
yielded 5 (24 mg, 59%). H NMR (300 MHz, CDCl3) δ 11.14
(s, 1H), 9.35 (d, J = 3.14 Hz, 1H), 8.77 (d, J = 7.80 Hz, 1H), 8.69
(d, J = 2.99 Hz, 1H), 8.01 (d, J = 6.95 Hz, 1H), 7.55 (t, J = 7.72
Hz, 1H), 7.12 (t, J = 7.99 Hz, 1H), 4.38-4.12 (m, 10H), 3.92 (s,
3H), 1.51 (m, 3H) ppm.
Assays were performed with CM or CCM from HEK293 cells
transiently transfected with NPP6 and -7 expression plasmids,
respectively. HEK293 cells were seeded in Dulbecco’s modified
Eagle’s medium (DMEM) containing 10% fetal bovine serum,
100 U/mL penicillin, 100 μg/mL streptomycin (Hyclone, Logan,
UT), and 2 mM L-glutamine. HEK293 cells were grown over-
night at 37 °C under 5% CO2 to 80% confluence. The cells were
then transfected with human NPP6ex49 or human NPP7ex52,53
expression plasmids in the pcDNA3.1(þ) mammalian vector in
the presence of Polyfect transfection reagent (Hyclone, Logan,
UT). Six hours after transfection, the culture medium was
changed to serum-free DMEM containing L-glutamine, and
cells were incubated for 48 h. Expressed protein was collected
2-(4-{[(4-Morpholinosulfonylphenyl)amino]carbonothioyl}-1-
piperazinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-
6-carboxylic Acid (6). Pipemidic acid (14.2 mg, 0.0468 mmol),
4-(morpholinosulfonyl)phenyl isothiocyanate (14 mg, 0.0468
mmol), and triethylamine (13 μL, 0.0935 mmol) were used.
1
Purification on silica yielded 6 (24 mg, 85%). H NMR (300
MHz, CDCl3) δ 9.28 (s, 1H), 8.65 (s, 1H), 8.09 (s, 1H), 7.64 (d,
J = 8.71 Hz, 2H), 7.51 (d, J = 8.61 Hz, 2H), 4.43-4.01 (m,
10H), 3.80-3.70 (m, 4H), 3.04-2.98 (m, 4H), 1.49 (t, J = 7.17
Hz, 3H) ppm.
2-(4-{[(3-Benzothiadiazolphenyl)amino]carbonothioyl}-1-pipe-
razinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-car-
boxylic Acid (7). Pipemidic acid (25 mg, 0.0824 mmol),

1064 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3
Hoeglund et al.
2,1,3-benzothiadiazol-5-yl isothiocyanate (15.8 mg, 0.0824
mmol), and triethylamine (23 μL, 0.165 mmol) were used.
Purification on silica yielded 7 (24 mg, 58%). H NMR (300
2-(4-{[(Phenyl)amino]carbonothioyl}-1-piperazinyl)-8-ethyl-
5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic Acid phenyl
(16). Pipemidic acid (25 mg, 0.0824 mmol), phenyl isothiocya-
nate (9.8 μL, 0.0824 mmol), and triethylamine (23 μL, 0.165
mmol) were used. Purification on silica yielded 16 (16 mg, 43%).
1H NMR (300 MHz, CDCl3) δ 9.33 (s, 1H), 8.68 (s, 1H),
7.41-7.33 (m, 3H), 7.22-7.15 (m, 2H), 4.38-4.28 (m, 2H),
4.24-3.89 (m, 8H), 1.49 (t, J = 7.57 Hz, 3H) ppm.
1
MHz, DMSO) δ 9.26 (s, 1H), 8.98 (s, 1H), 7.98-7.82 (m, 3H),
4.51-4.33 (m, 2H), 4.24-3.99 (m, 8H), 1.42-1.32 (m, 3H) ppm.
2-(4-{[(2,4,6-Mesitylphenyl)amino]carbonothioyl}-1-piperazi-
nyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carbo-
xylic Acid (8). Pipemidic acid (25 mg, 0.0824 mmol), mesityl
isothiocyanate (14.6 mg, 0.0824 mmol), and triethylamine
(23 μL, 0.165 mmol) were used. Purification on silica yielded 8
(36.6 mg, 92%). 1H NMR (300 MHz, CDCl3) δ 9.32 (s, 1H), 8.64
(s, 1H), 7.05 (s, 1H), 6.92 (s, 2H), 4.41-3.93 (m, 10H), 2.28 (s,
3H), 2.23 (s, 6H), 1.50 (t, J = 6.96 Hz, 3H) ppm.
2-(4-{[(2-Chlorophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (17). Pipemidic acid (56 mg, 0.184 mmol) and 2-chloro-
phenylisothiocyanate (20 μL, 0.153 mmol) were used. Purifica-
1
tion on silica yielded 17 (72 mg, 99%). H NMR (300 MHz,
2-(4-{[(4-pyrazolephenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (9). Pipemidic acid (27 mg, 0.0894 mmol) and 4-(1H-
pyrazol-1-yl)phenyl isothiocyanate (15 mg, 0.0745 mmol) were
used. Purification on silica yielded 9 (35.5 mg, 94%). 1H NMR
(250 MHz, CDCl3) δ 9.30 (s, 1H), 8.69 (s, 1H), 7.91 (s, 1H),
7.72-7.56 (m, 3H), 7.46-7.30 (m, 2H), 6.46 (s, 1H), 4.36 (q, J =
7.25 Hz, 2H), 4.28-3.96 (m, 8H), 1.48 (t, J = 7.04 Hz, 3H) ppm.
2-(4-{[(2-Benzothiadiazolphenyl)amino]carbonothioyl}-1-pipe-
razinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-car-
boxylic Acid (10). Pipemidic acid (25 mg, 0.0824 mmol), 2,1,3-
benzothiadiazol-4-yl isothiocyanate (16 mg, 0.0824 mmol), and
triethylamine (23 μL, 0.165 mmol) were used. Purification on
silica yielded 10 (10.8 mg, 26%). 1H NMR (300 MHz, DMSO) δ
9.81 (s, 1H), 9.27 (s, 1H), 9.00 (s, 1H), 7.93 (d, J = 8.40 Hz, 1H),
7.77-7.68 (m, 1H), 7.63 (d, J = 7.23 Hz, 1H), 4.48-4.41 (m,
2H), 4.24-3.99 (m, 8H), 1.39 (t, J = 6.52 3H) ppm.
2-(4-{[(3-Trifluoromethylphenyl)amino]carbonothioyl}-1-pipe-
razinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-car-
boxylic Acid (11). Pipemidic acid (25 mg, 0.0824 mmol), 3-(tri-
fluoromethyl)phenyl isothiocyanate (12.5 μL, 0.0824 mmol),
and triethylamine (23 μL, 0.165 mmol) were used. Purification
on silica yielded 11 (31 mg, 75%). 1H NMR (300 MHz, CDCl3) δ
9.32 (s, 1H), 8.72 (s, 1H), 7.41-7.35 (m, 1H), 7.29-7.21 (m, 2H),
7.21-7.16 (m, 1H), 4.37 (q, J = 7.10 Hz, 2H), 4.27-3.93 (m,
8H), 1.50 (t, J = 7.04 Hz, 3H) ppm.
CDCl3) δ 9.34 (s, 1H), 8.68 (s, 1H), 7.75 (d, J = 8.10 Hz, 1H),
7.44 (d, J = 8.00 Hz, 1H), 7.37-7.27 (m, 2H), 7.15 (t, J = 7.67
Hz, 1H), 4.35 (q, J = 7.15 Hz, 2H), 4.29-4.01 (m, 8H), 1.50 (t,
J = 7.20 Hz, 3H) ppm.
2-(4-{[(3,5-Bis(trifluoromethyl)phenyl)amino]carbonothioyl}-
1-piperazinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-
6-carboxylic Acid (18). Pipemidic acid (40 mg, 0.132 mmol) and
3,5-bis(trifluoromethyl)phenyl isothiocyanate (20 μL, 0.110
mmol) were used. Purification on silica yielded 18 (61 mg,
97%). 1H NMR (300 MHz, CDCl3) δ 9.30 (s, 1H), 8.68 (s,
1H), 8.62 (s, 1H), 7.94 (s, 2H), 7.62 (s, 1H), 4.42-4.05 (m, 10H),
1.51 (t, J = 7.07 Hz, 3H) ppm.
2-(4-{[(3-Methoxyphenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (19). Pipemidic acid (25 mg, 0.0824 mmol), 3-methoxyphe-
nyl isothiocyanate (11.6 μL, 0.0824 mmol), and triethylamine
(23 μL, 0.165 mmol) were used. Purification on silica yielded 19
(21.4 mg, 55%). 1H NMR (300 MHz, CDCl3) δ 9.31 (s, 1H), 8.67
(s, 1H), 7.47 (s, 1H), 7.30-7.21 (m, 1H), 6.80-6.68 (m, 3H), 4.32
(q, J = 6.90 Hz, 2H), 4.23-3.89 (m, 8H), 3.79 (s, 3H), 1.48 (t,
J = 7.16 Hz, 3H) ppm.
2-(4-{[(2,5-Dichlorophenyl)amino]carbonothioyl}-1-piperazi-
nyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carbo-
xylic Acid (20). Pipemidic acid (51 mg, 0.168 mmol) and 2,5-
dichlorophenyl isothiocyanate (20 μL, 0.140 mmol) were used.
Purification on silica yielded 20 (42.7 mg, 60%). 1H NMR (300
MHz, CDCl3) δ 9.39 (s, 1H), 9.06 (s, 1H), 7.57 (d, J = 2.39 Hz,
1H), 7.42-7.36 (m, 1H), 7.20 (d, J = 7.35 Hz, 1H), 4.55-4.39
(m, 2H), 4.31-4.02 (m, 8H), 1.52 (t, J = 6.88 Hz, 3H) ppm.
2-(4-{[(2-Methoxyphenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (21). Pipemidic acid (53 mg, 0.175 mmol) and 2-methox-
yphenyl isothiocyanate (20 μL, 0.145 mmol) were used. Purifi-
cation on silica yielded 21 (36 mg, 53%). 1H NMR (250 MHz,
CDCl3) δ 9.33 (s, 1H), 8.68 (s, 1H), 7.84 (d, J = 7.30 Hz, 1H),
7.54 (s, 1H), 7.20-7.04 (m, 1H), 7.03-6.85 (m, 2H), 4.44-3.73
(m, 13H), 1.49 (t, J = 6.60 Hz, 3H) ppm.
2-(4-{[(2-Trifluoromethylphenyl)amino]carbonothioyl}-1-pipe-
razinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-car-
boxylic Acid (12). Pipemidic acid (48 mg, 0.159 mmol) and
2-(trifluoromethyl)phenyl isothiocyanate (20 μL, 0.132 mmol)
1
were used. Purification on silica yielded 12 (55 mg, 82%). H
NMR (300 MHz, CDCl3) δ 9.32 (s, 1H), 8.69 (s, 1H), 7.74-7.64
(m, 1H), 7.63-7.51 (m, 2H), 7.43-7.33 (m, 1H), 4.47-3.90 (m,
10H), 1.50 (t, J = 6.86 Hz, 3H) ppm.
2-(4-{[(3-Methylphenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (13). Pipemidic acid (54 mg, 0.177 mmol) and m-tolyli-
sothiocyanate (20 μL, 0.148 mmol) were used. Purification on
silica yielded 13 (67 mg, 100%).1H NMR (300 MHz, CDCl3) δ
9.32 (s, 1H), 8.68 (s, 1H), 7.34-7.32 (m, 1H), 7.05-6.94 (m, 3H),
4.33 (q, J = 7.10 Hz, 2H), 4.24-3.87 (m, 8H), 2.35 (s, 3H), 1.49
(t, J = 7.19 Hz, 3H) ppm.
2-(4-{[(3,5-Dimethylphenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic Acid
(22). Pipemidic acid (45 mg, 0.148 mmol), 3,5-dimethylphenyl
isothiocyanate (24 mg, 0.148 mmol), and triethylamine (41 μL,
0.296 mmol) were used. Purification on silica yielded 22 (41.6
mg, 67%). 1H NMR (300 MHz, CDCl3) δ 9.30 (s, 1H), 8.66 (s,
1H), 7.48 (s, 1H), 6.86-6.75 (m, 3H), 4.33 (q, J = 7.10 Hz, 2H),
4.22-3.91 (m, 8H), 2.30 (s, 6H), 1.48 (t, J = 7.11 Hz, 3H) ppm.
2-(4-{[(4-Fluorophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (23). Pipemidic acid (35 mg, 0.1154 mmol) and 4-fluor-
ophenyl isothiocyanate (17.7 mg, 0.1154 mmol) were used.
2-(4-{[(3-Chlorophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (14). Pipemidic acid (25 mg, 0.0824 mmol), 3-chlorophenyl
isothiocyanate (11 μL, 0.0824 mmol), and triethylamine (23 μL,
0.165 mmol) were used. Purification on silica yielded 14 (8.2 mg,
21%). 1H NMR (300 MHz, CDCl3) δ 9.03 (s, 1H), 8.51 (s, 1H),
7.10 (s, 1H), 7.04-6.93 (m, 2H), 6.90-6.84 (m, 1H), 4.14 (q, J =
6.90 Hz, 2H), 4.00-3.74 (m, 8H), 1.23 (t, J = 7.01 Hz, 3H) ppm.
2-(4-{[(2,3-Dichlorophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (15). Pipemidic acid (51 mg, 0.168 mmol) and 2,3-dichlor-
ophenyl isothiocyanate (20 μL, 0.141 mmol) were used. Purifica-
1
Purification on silica yielded 23 (30 mg, 57%). H NMR (300
MHz, CDCl3) δ 9.34 (s, 1H), 8.69 (s, 1H), 7.33 (s, 1H), 7.25-7.16
(m, 2H), 7.13-7.03 (m, 2H), 4.34 (q, J = 7.00 Hz, 2H),
4.28-3.96 (m, 8H), 1.50 (t, J = 7.36 Hz, 3H) ppm.
2-(4-{[(4-Methoxyphenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (24). Pipemidic acid (30 mg, 0.0989 mmol) and 4-methoxy-
phenyl isothiocyanate (13.7 μL, 0.0989 mmol) were used.
1
tion on silica yielded 15 (15.5 mg, 22%). H NMR (300 MHz,
DMSO) δ 9.03 (s, 1H), 8.78 (s, 1H), 7.53 (d, J = 7.43 Hz, 1H),
7.41-7.26 (m, 2H), 4.51-3.90 (m, 10H), 1.43-1.27 (m, 3H) ppm.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3 1065
1
Purification on silica yielded 24 (37 mg, 80%). H NMR (300
MHz, CDCl3) δ 9.33 (s, 1H), 8.67 (s, 1H), 7.30 (s, 1H), 7.15 (d,
J = 8.76 Hz, 2H), 6.90 (d, J = 8.82 Hz, 2H), 4.34 (q, J = 7.10
Hz, 2H), 4.25-3.92 (m, 8H), 3.81 (s, 3H), 1.49 (t, J = 7.12 Hz,
3H) ppm.
Supporting Information Available: Proposed ATX analogs
for lead optimization and NPP6 and NPP7 selectivity results.
This material is available free of charge via the Internet at http://
pubs.acs.org.
2-(4-{[(3-Fluorophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (25). Pipemidic acid (25 mg, 0.0824 mmol) and 3-fluor-
ophenyl isothiocyanate (9.9 μL, 0.0824 mmol) were used.
Purification on silica yielded 25 (22 mg, 59%). 1H NMR
(300 MHz, CDCl3) δ 9.32 (s, 1H), 8.66 (s, 1H), 7.41 (s, 1H),
7.37-7.27 (m, 1H), 6.96 (t, J=7.39, 2H), 6.88 (t, J=7.67 Hz,
1H), 4.38-4.27 (m, 2H), 4.27-3.91 (m, 8H), 1.54-1.43
(m, 3H) ppm.
References
˚
(1) 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-stimu-
lating protein. J. Biol. Chem. 1992, 267, 2524–2529.
(2) 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.
(3) Tokumura, A.; Majima, E.; Kariya, Y.; Tominaga, K.; Kogure, K.;
Yasuda, K.; Fukuzawa, K. Identification of human plasma lyso-
phospholipase D, a lysophosphatidic acid-producing enzyme, as
autotaxin, a multifunctional phosphodiesterase. J. Biol. Chem.
2002, 277, 39436–39442.
(4) Tanaka, M.; Okudaira, S.; Kishi, Y.; Ohkawa, R.; Iseki, S.; Ota,
M.; Noji, S.; Yatomi, Y.; Aoki, J.; Arai, H. Autotaxin stabilizes
blood vessels and is required for embryonic vasculature by produ-
cing lysophosphatidic acid. J. Biol. Chem. 2006, 281, 25822–25830.
(5) Singer, S. J.; Kupfer, A. The directed migration of eukaryotic cells.
Annu. Rev. Cell Biol. 1986, 2, 337–365.
(6) van Meeteren, L. A.; Ruurs, P.; Stortelers, C.; Bouwman, P.; van
Rooijen, M. A.; Pradere, J. P.; Pettit, T. R.; Wakelam, M. J.;
Saulnier-Blache, J. S.; Mummery, C. L.; Moolenaar, W. H.;
Jonkers, J. Autotaxin, a secreted lysophospholipase D, is essential
for blood vessel formation during development. Mol. Cell. Biol.
2006, 26, 5015–5022.
(7) Tokumura, A.; Kanaya, Y.; Miyake, M.; Yamano, S.; Irahara, M.;
Fukuzawa, K. Increased production of bioactive lysophosphatidic
acid by serum lysophospholipase D in human pregnancy. Biol.
Reprod. 2002, 67, 1386–1392.
(8) Baumforth, K. R.; Flavell, J. R.; Reynolds, G. M.; Davies, G.;
Pettit, T. R.; Wei, W.; Morgan, S.; Stankovic, T.; Kishi, Y.; Arai,
H.; Nowakova, M.; Pratt, G.; Aoki, J.; Wakelam, M. J.; Young, L.
S.; Murray, P. G. Induction of autotaxin by the Epstein-Barr virus
promotes the growth and survival of Hodgkin lymphoma cells.
Blood 2005, 106, 2138–2146.
2-(4-{[(2-Iodophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (26). Pipemidic acid (25 mg, 0.0824 mmol) and 2-iodophe-
nyl isothiocyanate (21.5 mg, 0.0824 mmol) were used. Purifica-
1
tion on silica yielded 26 (35 mg, 75%). H NMR (300 MHz,
CDCl3) δ 9.36-9.30 (m, 1H), 8.69-8.65 (m, 1H), 7.89-7.82 (m,
1H), 7.62-7.55 (m, 1H), 7.41-7.32 (m, 1H), 7.00-6.92 (m, 1H),
4.40-4.01 (m, 10H), 1.53-1.44 (m, 3H) ppm.
2-(4-{[(4-Methylphenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (27). Pipemidic acid (89 mg, 0.293 mmol) and p-tolyl
isothiocyanate (40 mg, 0.268 mmol) were used. Purification on
silica yielded 27 (82 mg, 68%). 1H NMR (300 MHz, CDCl3) δ
9.32 (s, 1H), 8.67 (s, 1H), 7.40 (s, 1H), 7.17 (d, J = 7.60 Hz, 2H),
7.09 (d, J = 7.44 Hz, 2H), 4.33 (q, J = 6.40 Hz, 2H), 4.23-3.91
(m, 8H), 2.34 (s, 3H), 1.49 (t, J = 7.03 Hz, 3H) ppm.
2-(4-{[(4-Chlorophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (28). Pipemidic acid (58 mg, 0.191 mmol) and 4-chloro-
phenyl isothiocyanate (27 mg, 0.159 mmol) were used. Purifica-
1
tion on silica yielded 28 (50 mg, 67%). H NMR (300 MHz,
CDCl3) δ 9.34 (s, 1H), 8.68 (s, 1H), 7.36-7.34 (m, 2H), 7.17 (d,
J = 8.70 Hz, 2H), 4.39-4.28 (m, 2H), 4.28-3.93 (m, 8H), 1.49
(t, J = 6.68 Hz, 3H) ppm.
(9) Chen, M.; O’Connor, K. L. Integrin alpha6beta4 promotes expres-
sion of autotaxin/ENPP2 autocrine motility factor in breast carci-
noma cells. Oncogene 2005, 24, 5125–5130.
(10) Euer, N.; Schwirzke, M.; Evtimova, V.; Burtscher, H.; Jarsch, M.;
Tarin, D.; Weidle, U. H. Identification of genes associated with
metastasis of mammary carcinoma in metastatic versus non-meta-
static cell lines. Anticancer Res. 2002, 22, 733–740.
(11) Hoelzinger, D. B.; Mariani, L.; Weis, J.; Woyke, T.; Berens, T. J.;
McDonough, W. S.; Sloan, A.; Coons, S. W.; Berens, M. E. Gene
expression profile of glioblastoma multiforme invasive phenotype
points to new therapeutic targets. Neoplasia 2005, 7, 7–16.
(12) Kawagoe, H.; Stracke, M. L.; Nakamura, H.; Sano, K. Expression
and transcriptional regulation of the PD-Ialpha/autotaxin gene in
neuroblastoma. Cancer Res. 1997, 57, 2516–2521.
(13) Kehlen, A.; Englert, N.; Seifert, A.; Klonisch, T.; Dralle, H.;
Langner, J.; Hoang-Vu, C. Expression, regulation and function
of autotaxin in thyroid carcinomas. Int. J. Cancer 2004, 109, 833–
838.
(14) Koh, E.; Bandle, R. W.; Roberts, D. D.; Stracke, M. L.; Clair, T.
Novel point mutations attenuate autotaxin activity. Lipids Health
Dis. 2009, 8, 4.
(15) Lee, H. Y.; Murata, J.; Clair, T.; Polymeropoulos, M. H.; Torres,
R.; Manrow, R. E.; Liotta, L. A.; Stracke, M. L. Cloning,
chromosomal localization, and tissue expression of autotaxin from
human teratocarcinoma cells. Biochem. Biophys. Res. Commun.
1996, 218, 714–719.
(16) Masuda, A.; Nakamura, K.; Izutsu, K.; Igarashi, K.; Ohkawa, R.;
Jona, M.; Higashi, K.; Yokota, H.; Okudaira, S.; Kishimoto, T.;
Watanabe, T.; Koike, Y.; Ikeda, H.; Kozai, Y.; Kurokawa, M.;
Aoki, J.; Yatomi, Y. Serum autotaxin measurement in haemato-
logical malignancies: A promising marker for follicular lymphoma.
Br. J. Hamaetol. 2008, 143, 60–70.
2-(4-{[(3-Iodophenyl)amino]carbonothioyl}-1-piperazinyl)-8-
ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (29). Pipemidic acid (40 mg, 0.133 mmol) and 3-iodophenyl
isothiocyanate (29 mg, 0.111 mmol) were used. Purification on
silica yielded 29 (57 mg, 90%). 1H NMR (250 MHz, CDCl3) δ
9.33 (s, 1H), 8.73 (s, 1H), 7.66-7.64 (m, 1H), 7.54-7.51 (m, 1H),
7.40-7.24 (m, 1H), 7.12-7.06 (m, 1H), 4.48-4.31 (m, 2H),
4.27-3.93 (m, 8H), 1.34-1.17 (m, 3H) ppm.
2-(4-{[(2-Fluorophenyl)amino]carbonothioyl}-1-piperazinyl)-
8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-carboxylic
Acid (30). Pipemidic acid (25 mg, 0.0824 mmol) and 2-fluor-
ophenyl isothiocyanate (10 μL, 0.0824 mmol) were used. Pur-
ification on silica yielded 30 (24 mg, 64%). 1H NMR (300 MHz,
CDCl3) δ 9.35 (s, 1H), 8.69 (s, 1H), 7.64 (t, J = 7.34 Hz, 1H),
7.21-7.12 (m, 3H), 4.35 (q, J = 6.90 Hz, 2H), 4.28-4.02 (m,
8H), 1.50 (t, J = 7.09 Hz, 3H) ppm.
2-(4-{[(4-Trifluoromethylphenyl)amino]carbonothioyl}-1-pipe-
razinyl)-8-ethyl-5-oxo-5,8-dihydropyrido[2,3-d]pyrimidine-6-car-
boxylic Acid (31). Pipemidic acid (30 mg, 0.0989 mmol) and
4-trifluoromethylphenyl isothiocyanate (16.7 mg, 0.0824 mmol)
were used. Purification on silica yielded 31 (31 mg, 77%).
1H NMR (300 MHz, CDCl3) δ 9.34 (s, 1H), 8.69 (s, 1H),
7.65-7.57 (m, 2H), 7.51 (s, 1H), 7.37-7.31 (m, 2H), 4.34 (q,
J = 6.80 Hz, 2H), 4.27-3.97 (m, 8H), 1.49 (t, J = 7.25, 7.25 Hz,
3H) ppm.
Acknowledgment. The authors are grateful to the Chemi-
cal Computing Group for free access to the Molecular
Operating Environment (MOE) software. This work was
supported, in part, by a research grant from the Elsa U.
Pardee Foundation (A.L.P. and D.L.B.).
(17) Song, J.; Clair, T.; Noh, J. H.; Eun, J. W.; Ryu, S. Y.; Lee, S. N.;
Ahn, Y. M.; Kim, S. Y.; Lee, S. H.; Park, W. S.; Yoo, N. J.; Lee, J.
Y.; Nam, S. W. Autotaxin (lysoPLD/NPP2) protects fibroblasts
from apoptosis through its enzymatic product, lysophosphatidic
acid, utilizing albumin-bound substrate. Biochem. Biophys. Res.
Commun. 2005, 337, 967–975.

1066 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 3
Hoeglund et al.
(18) Stassar, M. J.; Devitt, G.; Brosius, M.; Rinnab, L.; Prang, J.;
Schradin, T.; Simon, J.; Petersen, S.; Kopp-Schneider, A.; Zoller,
M. Identification of human renal cell carcinoma associated genes
by suppression subtractive hybridization. Br. J. Cancer 2001, 85,
1372–1382.
(19) Zeng, Y.; Kakehi, Y.; Nouh, M. A.; Tsunemori, H.; Sugimoto, M.;
Wu, X. X. Gene expression profiles of lysophosphatidic acid-
related molecules in the prostate: relevance to prostate cancer
and benign hyperplasia. Prostate 2009, 69, 283–292.
(20) Zhang, G.; Zhao, Z.; Xu, S.; Ni, L.; Wang, X. Expression of
autotaxin mRNA in human hepatocellular carcinoma. Chin.
Med. J. (Engl.) 1999, 112, 330–332.
lysophosphatidic acid receptors. Bioorg. Med. Chem. Lett. 2006,
16, 451–456.
(36) Baker, D. L.; Fujiwara, Y.; Pigg, K. R.; Tsukahara, R.; Kobayashi,
S.; Murofushi, H.; Uchiyama, A.; Murakami-Murofushi, K.; Koh,
E.; Bandle, R. W.; Byun, H.-S.; Bittman, R.; Fan, D.; Murph, M.;
Mills, G. B.; Tigyi, G. Carba analogs of cyclic phosphatidic acid are
selective inhibitors of autotaxin and cancer cell invasion and
metastasis. J. Biol. Chem. 2005, 281, 22786–22793.
(37) Cui, P.; Tonsig, J. L.; McCalmont, W. F.; Lee, S.; Becker, C. J.;
Lynch, K. R.; Macdonald, T. L. Synthesis and biological evalua-
tion of phosphonate derivatives as autotaxin (ATX) inhibitors.
Bioorg. Med. Chem. Lett. 2007, 17, 1634–1640.
(21) Hammack, B. N.; Fung, K. Y.; Hunsucker, S. W.; Duncan, M. W.;
Burgoon, M. P.; Owens, G. P.; Gilden, D. H. Proteomic analysis of
multiple sclerosis cerebrospinal fluid. Mult. Scler. 2004, 10, 245–
260.
(38) Ferry, G.; Moulharat, N.; Pradere, J. P.; Desos, P.; Try, A.; Genton,
A.; Giganti, A.; Beucher-Gaudin, M.; Lonchampt, M.; Bertrand,
M.; Saulnier-Blache, J. S.; Tucker, G. C.; Cordi, A.; Boutin, J. A.
S32826, a nanomolar inhibitor of autotaxin: discovery, synthesis
and applications as a pharmacological tool. J. Pharmacol. Exp.
Ther. 2008, 327, 809–819.
(39) Prestwich, G. D.; Gajewiak, J.; Zhang, H.; Xu, X.; Yang, G.;
Serban, M. Phosphatase-resistant analogues of lysophosphatidic
acid: Agonists promote healing, antagonists and autotaxin inhibi-
tors treat cancer. Biochim. Biophys. Acta 2008, 1781, 588–594.
(40) van Meeteren, L. A.; Brinkmann, V.; Saulnier-Blache, J. S.; Lynch,
K. R.; Moolenaar, W. H. Anticancer activity of FTY720: Phos-
phorylated FTY720 inhibits autotaxin, a metastasis-enhancing and
angiogenic lysophospholipase D. Cancer Lett. 2008, 266, 203–208.
(41) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.
Experimental and computational approaches to estimate solubility
and permeability in drug discovery and development settings. Adv.
Drug Delivery Rev. 2001, 46, 3–26.
(22) Boucher, J.; Quilliot, D.; Praderes, J. P.; Simon, M. F.; Gres, S.;
Guigne, C.; Prevot, D.; Ferry, G.; Boutin, J. A.; Carpene, C.; Valet,
P.; Saulnier-Blache, J. S. Potential involvement of adipocyte insulin
resistance in obesity-associated up-regulation of adipocyte lyso-
phospholipase D/autotaxin expression. Diabetologia 2005, 48, 569–
577.
ꢀ
(23) Ferry, G.; Tellier, E.; Try, A.; Gres, S.; Naime, I.; Simon, M. F.;
Rodriguez, M.; Boucher, J.; Tack, I.; Gesta, S.; Chomarat,
P.; Dieu, M.; Raes, M.; Galizzi, J. P.; Valet, P.; Boutin, J. A.;
Saulnier-Blache, J. S. Autotaxin is released from adipocytes,
catalyzes lysophosphatidic acid synthesis, and activates preadipo-
cyte proliferation. J. Biol. Chem. 2003, 278, 18162–18169.
(24) Gesta, S.; Simon, M. F.; Rey, A.; Sibrac, D.; Girard, A.; Lafontan,
M.; Valet, P.; Saulnier-Blache, J. S. Secretion of a lysophospholi-
pase D activity by adipocytes: involvement in lysophosphatidic
acid synthesis. J. Lipid Res. 2002, 43, 904–910.
(42) Keller, T. H.; Pichota, A.; Yin, Z. A practical view of ’druggability’.
Curr. Opin. Chem. Biol. 2006, 10, 357–361.
(25) Simon, M. F.; Daviaud, D.; Pradere, J. P.; Gres, S.; Guigne, C.;
Wabitsch, M.; Chun, J.; Valet, P.; Saulnier-Blache, J. S. Lysopho-
sphatidic acid inhibits adipocyte differentiation via lysophospha-
tidic acid 1 receptor-dependent down-regulation of peroxisome
proliferator-activated receptor gamma2. J. Biol. Chem. 2005, 280,
14656–14662.
(26) Simon, M. F.; Rey, A.; Castan-Laurel, I.; Gres, S.; Sibrac, D.;
Valet, P.; Saulnier-Blache, J. S. Expression of ectolipid phosphate
phosphohydrolases in 3T3F442A preadipocytes and adipocytes.
Involvement in the control of lysophosphatidic acid production.
J. Biol. Chem. 2002, 277, 23131–23136.
(27) Umemura, K.; Yamashita, N.; Yu, X.; Arima, K.; Asada, T.;
Makifuchi, T.; Murayama, S.; Saito, Y.; Kanamaru, K.; Goto,
Y.; Kohsaka, S.; Kanazawa, I.; Kimura, H. Autotaxinexpression is
enhanced in frontal cortex of Alzheimer-type dementia patients.
Neurosci. Lett. 2006, 400, 97–100.
(28) Inoue, M.; Ma, L.; Aoki, J.; Chun, J.; Ueda, H. Autotaxin, a
synthetic enzyme of lysophosphatidic acid (LPA), mediates the
induction of nerve-injured neuropathic pain. Mol. Pain 2008, 4, 6.
(29) Inoue, M.; Ma, L.; Aoki, J.; Ueda, H. Simultaneous stimulation of
spinal NK1 and NMDA receptors produces LPC which undergoes
ATX-mediated conversion to LPA, an initiator of neuropathic
pain. J. Neurochem. 2008, 107, 1556–1565.
(43) Parrill, A. L.; Echols, U.; Nguyen, T.; Pham, T. C.; Hoeglund, A.;
Baker, D. L. Virtual screening approaches for the identification of
non-lipid autotaxin inhibitors. Bioorg. Med. Chem. 2008, 16, 1784–
1795.
(44) Saunders, L. P.; Ouellette, A.; Bandle, R.; Chang, W. C.; Zhou, H.;
Misra, R. N.; De La Cruz, E. M.; Braddock, D. T. Identification of
small-molecule inhibitors of autotaxin that inhibit melanoma cell
migration and invasion. Mol. Cancer Ther. 2008, 7, 3352–3362.
(45) Ferguson, C. G.; Bigman, C. S.; Richardson, R. D.; van Meeteren,
L. A.; Moolenaar, W. H.; Prestwich, G. D. Fluorogenic phospho-
lipid substrate to detect lysophospholipase D/autotaxin activity.
Org. Lett. 2006, 8, 2023–2026.
(46) Moulharat, N.; Fould, B.; Giganti, A.; Boutin, J. A.; Ferry, G.
Molecular pharmacology of adipocyte-secreted autotaxin. Chem.-
Biol. Interact. 2008, 172, 115–124.
(47) Stefan, C.; Jansen, S.; Bollen, M. NPP-type ectophosphodies-
terases: unity in diversity. Trends Biochem. Sci. 2005, 30, 542–550.
(48) Lee, M. J.; Van Brocklyn, J. R.; Thangada, S.; Liu, C. H.; Hand, A.
R.; Menzeleev, R.; Spiegel, S.; Hla, T. Sphingosine-1-phosphate as
a ligand for the G protein-coupled receptor EDG-1. Science 1998,
279, 1552–1555.
(49) Sakagami, H.; Aoki, J.; Natori, Y.; Nishikawa, K.; Kakehi, Y.;
Natori, Y.; Arai, H. Biochemical and molecular characterization of
a novel choline-specific glycerophosphodiester phosphodiesterase
belonging to the nucleotide pyrophosphatase/phosphodiesterase
family. J. Biol. Chem. 2005, 280, 23084–23093.
(50) Duan, R. D.; Bergman, T.; Xu, N.; Wu, J.; Cheng, Y.; Duan, J.;
Nelander, S.; Palmberg, C.; Nilsson, A. Identification of human
intestinal alkaline sphingomyelinase as a novel ecto-enzyme related
to the nucleotide phosphodiesterase family. J. Biol. Chem. 2003,
278, 38528–38536.
(30) Inoue, M.; Xie, W.; Matsushita, Y.; Chun, J.; Aoki, J.; Ueda, H.
Lysophosphatidylcholine induces neuropathic pain through
an action of autotaxin to generate lysophosphatidic acid.
Neuroscience 2008, 152, 296–298.
(31) Tokumura, A.; Miyake, M.; Yoshimoto, O.; Shimizu, M.; Fuku-
zawa, K. Metal-ion stimulation and inhibition of lysophospholi-
pase D which generates bioactive lysophosphatidic acid in rat
plasma. Lipids 1998, 33, 1009–1015.
(32) Clair, T.; Koh, E.; Ptaszynska, M.; Bandle, R. W.; Liotta, L. A.;
Schiffmann, E.; Stracke, M. L. L-histidine inhibits production of
lysophosphatidic acid by the tumor-associated cytokine, auto-
taxin. Lipids Health Dis. 2005, 4, 5.
(33) van Meeteren, L. A.; Ruurs, P.; Christodoulou, E.; Goding, J. W.;
Takakusa, H.; Kikuchi, K.; Perrakis, A.; Nagano, T.; Moolenaar,
W. H. Inhibition of autotaxin by lysophosphatidic acid and
sphingosine 1-phosphate. J. Biol. Chem. 2005, 280, 21155–
21161.
(34) Durgam, G. G.; Virag, T.; Walker, M. D.; Tsukahara, R.; Yasuda,
S.; Liliom, K.; van Meeteren, L. A.; Moolenaar, W. H.; Wilke, N.;
Siess, W.; Tigyi, G.; Miller, D. D. Synthesis, structure-activity
relationships, and biological evaluation of fattyalcohol phosphates
as lysophosphatidic acid receptor ligands, activators of PPAR-
gamma, and inhibitors of autotaxin. J. Med. Chem. 2005, 48, 4919–
4930.
(35) Gududuru, V.; Zeng, K.; Tsukahara, R.; Makarova, N.; Fujiwara,
Y.; Pigg, K. R.; Baker, D. L.; Tigyi, G.; Miller, D. D. Identification
of Darmstoff analogs as selective agonists and antagonists of
(51) Duan, R. D.; Cheng, Y.; Hansen, G.; Hertervig, E.; Liu, J. J.; Syk,
I.; Sjostrom, H.; Nilsson, A. Purification, localization, and expres-
sion of human intestinal alkaline sphingomyelinase. J. Lipid Res.
2003, 44, 1241–1250.
(52) Wu, J.; Nilsson, A.; Jonsson, B. A.; Stenstad, H.; Agace, W.;
Cheng, Y.; Duan, R. D. Intestinal alkaline sphingomyelinase
hydrolyses and inactivates platelet-activating factor by a phospho-
lipase C activity. Biochem. J. 2006, 394, 299–308.
(53) Wu, J.; Hansen, G. H.; Nilsson, A.; Duan, R. D. Functional studies
of human intestinal alkaline sphingomyelinase by deglycosylation
and mutagenesis. Biochem. J. 2005, 386, 153–160.
(54) Cui, P.; McCalmont, W. F.; Tomsig, J. L.; Lynch, K. R.; Macdo-
nald, T. L. alpha- and beta-substituted phosphonate analogs of
LPA as autotaxin inhibitors. Bioorg. Med. Chem. 2008, 16, 2212–
2225.
(55) Jiang, G.; Xu, Y.; Fujiwara, Y.; Tsukahara, T.; Tsukahara, R.;
Gajewiak, J.; Tigyi, G.; Prestwich, G. D. R-substituted phospho-
nate analogues of lysophosphatidic acid (LPA) selectively inhibit
production and action of LPA. ChemMedChem 2007, 2, 679–690.