Synthesis and structure-activity relationships of tyrosine-based inhibitors of autotaxin (ATX)

Article abstract of DOI:10.1016/j.bmcl.2010.09.030

Autotaxin (ATX) is a secreted soluble enzyme that generates lysophosphatidic acid (LPA) through its lysophospholipase D activity. Because of LPA's role in neoplastic diseases, ATX is an attractive therapeutic target due to its involvement in LPA biosynthe

Full text of DOI:10.1016/j.bmcl.2010.09.030

Bioorganic & Medicinal Chemistry Letters 20 (2010) 7132–7136
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and structure–activity relationships of tyrosine-based
inhibitors of autotaxin (ATX)
James E. East a, , Andrew J. Kennedy a, Jose L. Tomsig b, Alexandra R. De Leon b,
⇑
Kevin R. Lynch b, Timothy L. Macdonald a,b,
⇑
a University of Virginia, Department of Chemistry, PO Box 400319, McCormick Road, Charlottesville, VA 22904, United States
b University of Virginia, Department of Pharmacology, 1340 Jefferson Park Avenue, Charlottesville, VA 22908, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Autotaxin (ATX) is a secreted soluble enzyme that generates lysophosphatidic acid (LPA) through its lyso-
phospholipase D activity. Because of LPA’s role in neoplastic diseases, ATX is an attractive therapeutic tar-
get due to its involvement in LPA biosynthesis. Here we describe the SAR of ATX inhibitor, VPC8a202, and
apply this SAR knowledge towards developing a high potency inhibitor. We found that electron density in
the pyridine region greatly influences activity of our inhibitors at ATX.
Received 2 August 2010
Revised 3 September 2010
Accepted 7 September 2010
Available online 15 September 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
Lysophosphatidic acid
Autotaxin
Structure–activity relationships
Hansch linear free energy relationship
Autotaxin (ATX) is an enzyme in the nucleotide phosphatase/
pyrophosphatase 2 (NPP2) family of enzymes and it is the only
NPP2 with lysophospholipase D activity. ATX catalyzes the hydro-
lysis of lysophosphatidylcholine (LPC) to yield lysophosphatidic
acid (LPA) which signals cells via a set of five G-protein coupled
receptors (GPCRs). Originally identified as an autocrine growth fac-
tor,1 ATX is implicated in tumor progression because of LPA’s role
in cell migration and invasiveness.2 As a result there has been con-
siderable interest in targeting ATX with small molecules to affect
LPA driven neoplastic diseases.3,4
LPA has been demonstrated to be a feedback inhibitor of ATX.5
With this in mind our laboratories have set out to develop LPA ana-
logs as ATX inhibitors (Fig. 1). Our earliest series of LPA analogs
were made as probes of the various receptors. These molecules fea-
tured the N-acyl ethanolamide phosphoric acid (NAEPA) backbone,
which was a modification of the natural glycerol moiety in LPA.6
The NAEPA compounds eventually evolved into LPA analogs con-
taining various amino acids other than glycine7 with
L-tyrosine
analogs exhibiting the most useful activities.8 Phosphate head
group mimetics including phosphonic and thiophosphonic acid
derivatives9 and substituted phosphonates10,11 were also investi-
gated because of enhanced metabolic stability.
From these compounds, 5a(anti) (VPC8a202) (Fig. 1) was found
to be a potent inhibitor (Ki = 1 lM) of ATX. However, a clear
⇑
Corresponding authors. Tel.: +1 434 924 7718 (J.E.).
E-mail addresses: je5y@virginia.edu (J.E. East), tlm@virginia.edu (T.L. Macdonald).
Figure 1. Evolution of tyrosine-based LPA analog 5a(anti) (VPC8a202).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.09.030

J. E. East et al. / Bioorg. Med. Chem. Lett. 20 (2010) 7132–7136
7133
Scheme 1. Reagents and conditions: (a) palmitoyl Cl, DIEA, DCM, 0 °C; (b) ROMs (or RBr), K2CO3, 18-crown-6, acetone, 70 °C; (c) dimethyl methylphosphonate, nBuLi, À78 °C;
(d) CeCl3Á7H2O, NaBH4, EtOH:THF, 0 °C; (e) TMSBr, DCM 0 °C to rt.
structure–activity relationship of the pyridyl region remained
unexplored and this information is needed for our future efforts to-
wards more potent inhibitors of ATX. Therefore, we report herein
the synthesis and SAR studies of a series of tyrosine-based analogs
of 5a(anti) (VPC8a202).
Table 1
In vitro activities for compounds 5a–m
O
P
OH
H
N
HO
HO
C15H31
O
Synthesis of the b-hydroxy phosphonates (Scheme 1) began
with acid chloride coupling of palmitoyl chloride with L-tyrosine
BenzylO
methyl ester hydrochloride to give amide 1. Benzylic meslates or
bromides were coupled to the phenol of 1 in the presence of potas-
sium carbonate and 18-crown-6 in refluxing acetone to give com-
pounds 2a–m. Next, treatment of intermediates 2a–m with
dimethyl methylphosphonate and n-butyllithium in THF at
À78 °C afforded b-keto phosphonates 3a–m. Reduction of the ke-
tone under Luche conditions12 yielded protected b-hydroxy phos-
phonates 4a–m. The resulting diastereomers were separated by
flash chromatography. Designations of syn and anti were arrived
at through the use of stereochemical studies performed by Peng
et al.11 and homology modeling studies using ATX with our com-
pounds (vide infra). Finally, bromotrimethylsilane (TMSBr) treat-
ment with subsequent silyl cleavage in MeOH:H2O afforded the
deprotected phosphonates 5a–m, diastereomers syn and anti.
All compounds were tested for their inhibitory capability in a
serum based in vitro assay (see Supplementary data). The deletion
of the nitrogen heteroatom (Table 1) and methyl substituents
(5b(anti)/(syn)) from 5a(anti) showed a significant loss in activity.
Retention of the nitrogen heteroatom (5c(anti)), plus methoxy
substituents (5d(anti)/5d(syn)) or alkyl groups (5e(anti)) allowed
for near equipotent compounds in relation to 5a(anti). Elongation
of the methoxy groups to ethoxy (5f(anti)/(syn)) and propoxy
(5g(anti)/(syn)) led to a progressive loss in activity. 5i(anti)/
(syn), 5m(anti), and 5k(anti)/(syn) retained potency while con-
taining no nitrogen heteroatom yet retaining methoxy and/or
methyl groups. Interestingly, the dichlorinated aryl ethers,
5j(anti)/(syn), proved to be the least potent in this series of com-
pounds. This result is presumably from the electron withdrawing
capability of the chlorines.
A molecular docking analysis was performed with the utiliza-
tion of a previously reported homology model of the catalytic do-
main of ATX by Parrill et al.13 which was derived from the solved
Xac NPP crystal stucture (entry 2GSU14 in the Protein Data Base).
Docking studies with 5a(anti) showed a series of low energy pos-
sible binding conformations; all of which presented the phospho-
nate within 5 Å of Thr210 and the two zinc metal centers of the
active site (Fig. 2). These studies also revealed that the saturated
tail of 5a(anti) fills the large lipophilic pocket also found to bind
the lipid tail of LPC. The anti configuration of the beta alcohol

7134
J. E. East et al. / Bioorg. Med. Chem. Lett. 20 (2010) 7132–7136
preference for the anti beta alcohol over the syn relationship.
One interesting development was the evidence that the substituted
pyridine of 5a(anti) binds the hydrophilic leaving group pocket of
ATX, and the tyrosine core acts almost solely as a linker to facilitate
this interaction. Specifically, the model suggests an aromatic–gua-
nidine interaction with Arg456 and a weak hydrogen bond be-
tween the methoxy substituent at the 4-position and Lys209
(Fig. 2C). Docking analysis of the full library of aryl derivatives
(5a–m) showed similar results, with the degree of interaction
depending on aryl ring electron density and geometric fit in the
pocket.
For each member of the aryl derivative library (5a–m), three
docking scores (London dG, Affinity dG, and Alpha HB) available
in MOE15 were used to judge different binding conformation possi-
bilities. The long acyl chains (C15) were considered only for the
lead 5a and was not varied for the other library members. Even
though there were a number of similarly scored conformations
for the acyl chain of 5a(anti), they led to negligible differences in
the orientation of the tyrosine linker, phosphinated and pyridyl
group. One conformation of the acyl chain was then conserved
for every member of the library not for its accuracy, but to limit
the test set for docking analysis. In most cases, a single conforma-
tion scored best in two or three of the possible conformers, and in
fact all three scoring functions proved capable of producing similar
lowest energy binding conformations. However, in analyzing the li-
brary as a whole, Alpha HB was the only scoring function to corre-
late with statistical relevance to activity (Fig. 3A). All library
members preferably docked in a conformation similar to that of
5a(anti), where the tyrosine core acts as a linker to display the
benzyl group of the inhibitor to the leaving group pocket of ATX.
Alpha HB, a scoring function based on geometrical fit, was able
to accurately explain activity by how well the substituents filled
the pocket, as well as how closely the benzyl group interacted with
Arg456. Substituents that add electron density into the
p system of
the benzyl group had closer association with Arg456 (5a(anti) =
2.2 Å, 5j(anti) = 3.9 Å), and therefore scored better. There is also a
steric effect with substitution that if exceeded lead to dissociation
of the benzyl group from the pocket, explaining the linear decrease
in activity from 5a to 5f to 5g.
Figure 2. ATX homology model with LPC 18:1 and 5a(anti). (A) ATX shown as an
electrostatic surface (uncharged = white, positive = blue, and negative = red) with
LPC 18:1 shown as a stick model. (B) ATX shown as an electrostatic surface with
5a(anti) shown as a stick model. (C) Close up of ATX active site with important
residues and 5a(anti) shown as stick models. Hydrogen bonds and arene–cation
interactions shown as dotted white lines. Zinc cations shown as yellow spheres.
The homology model docking study predicts the interaction be-
tween the aromatic
p systems of 5(a–m) and the guanidine of
Arg456 to be a principle component in binding affinity of the aryl
system. Therefore, there should be a negative relationship between
Hammett substituent constants (r) and pKi, since the arene–cation
bond is dependant on aryl electron density. The system was eval-
uated using a Hansch linear free energy model,16 and the aromatic
derivatives were divided into individual substituents and each
present in 5a(anti) forms a weak hydrogen bond with the zinc
co-ordinating Asp312 residue,14 that may explain the stereo
Figure 3. Docking and Hansch model studies. (A) Experimental pKi—docking score plot (R2 = 0.72833) calculated with MOE Alpha HB scoring function for minimized docking
conformations (see Supplementary data for details). (B) Hansch linear free energy linear model (R2 = 0.99416) based on the electronic, steric and solubility effects of aryl
substitution described in Table 2.

J. E. East et al. / Bioorg. Med. Chem. Lett. 20 (2010) 7132–7136
7135
Table 2
Hansch linear free energy model data set
Compound
pKi
Prod. pKid
z-Score
Aromatic substitution
Hammell value (op)a
van der Walls volume (Å)b
nc
3
4
5
6
3
4
5
6
3
4
5
6
5a
5b
5c
5d
5e
5f
5g
5h
5i
6.00
4.55
5.36
5.82
5.41
5.33
5.06
4.94
5.74
3.23
5.80
4.05
5.37
6.00
4.58
5.63
5.86
5.34
5.37
4.97
5.01
5.77
3.23
5.76
4.05
5.46
0.00
0.38
1.01
0.63
1.19
0.74
1.40
1.35
0.51
0.00
1.65
0.00
1.65
Me
H
H
OMe
H
H
OMe
H
OEt
OPt
OCH2CF3
H
Me
H
H
H
H
H
H
H
H
H
H
H
H
Me
Me
Me
À0.17
0
0
À0.27
0
0
À0.17
0
0
0
0
0
0
0
0
0
0
0
0
31.5
10.4
1.04
37.1
10.4
10.4
37.1
10.4
55.9
73.5
64.8
10.4
10.4
31.5
31.5
10.4
31.5
10.4
10.4
10.4
31.5
31.5
31.5
31.5
31.5
28.5
37.12
10.4
37.13
0.70
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
10.4
31.5
31.5
31.5
À0.90
0.00
À1.27
H
H
0
À0.27
0
10.4
31.5
31.5
31.5
31.5
31.5
28.5
10.4
10.4
10.4
1.45
Me
Me
Me
Me
Me
Cl
H
H
H
Me
Me
Me
Me
Me
Cl
OMe
H
OMe
À0.17
À0.17
À0.17
À0.17
À0.17
0.23
0
0
0
0.60
0
À0.17
À0.17
À0.17
À0.17
À0.17
0.23
À0.27
0
À0.30
À0.56
0.06
À0.24
À0.25
À0.21
0
À0.13
0.97
5j
5k
5l
H
0
1.05
0.45
1.05
0.12
Me
Me
H
À0.17
À0.17
0
À0.17
À0.17
À0.17
0.00
5m
À0.27
Relative importance of descriptions
0.53
0.70
1.00
0.57
0.79
0.51
a
b
c
Ref. 17.
Ref. 18.
za = C Lop Pi À C Log P 5b.
d
Prod. pKi = (À3.34 (3op) À 2.94 (4op) À 3.36 (5op)) + (À0.0627 (3 vol) À 0.0170 (4 vol) + 0.0438 (5 vol) À 0.0579 (6 vol)) + 0.0261 (x) À 0.467 (Àx2) + 5.55.
substituent described by its Hammett value,17 van der Waals vol-
(VPC8a202). Our compounds are comparable to other reported po-
tent ATX inhibitors that were tested in our choline release assay.
These tyrosine derivatives share the common features of HA51,
HA130,19 S32826,20 and Br-LPA21 (Table 3) in that they have an
electrophilic head group and a hydrophobic tail region. Through
the use of classical SAR and QSAR we discovered that potency of
our compound library increased with increasing electron density
contained in the pyridine ring. Our use of homology modeling sug-
gests that this trend may be due to an interaction with the pyridine
group and Arg456. We hope to use these findings to aid us in our
work towards further validating the homology model and, ulti-
mately, developing more potent inhibitors of autotaxin.
ume,18 and difference in C log P (
p) from the unsubstituted benzyl
derivative 5b (Table 2). This Hansch linear free energy model
(Fig. 3B) showcases the relationship the electron density of the aro-
matic region of this class of inhibitors, when corrected for steric ef-
fects and C log P. There is a negative relationship between Hammet
value and pKi at the 3, 4, and 5 positions of the benzyl ring, which
shows the preference for electron rich aromatics and supports the
results from the docking study (Table 2). The Hansch analysis also
corrects for substituent steric effects, which were compared using
the van der Waals volumes, and indicated the substitution was
least tolerated at the 3-position and most tolerated at the 5-posi-
tion of the ring.
In summary, we set out to determine what elements were
important in the pyridine ring of the ATX inhibitor 5a(anti)
Acknowledgment
This work is supported by NIH grants R01 GM052722, R01
GM067958.
Table 3
Reported ATX inhibitors tested in choline release assay
Name
Structure
Ki (lM)
Supplementary data
O
HO
B
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2010.09.030.
N
S
OH
HA130
0.094
F
O
References and notes
O
N
S
COOH
1. Stracke, M. H.; Krutzsch, H. C.; Unsworth, E. J.; Arestad, A.; Cioce, V.;
Schiffmann, E. J. Biol. Chem. 1992, 267, 2524.
2. Mills, G. B.; Moolenaar, W. H. Nat. Rev. Cancer 2003, 3, 582.
3. Albers, H.; van Meeteren, L.; Egan, D.; van Tilburg, E.; Moolenaar, W.; Ovaa, H. J.
Med. Chem. 2010, 13, 4958.
4. North, E.; Howard, A.; Wanjala, I.; Pham, T.; Baker, D.; Parrill, A. J. Med. Chem.
2010, 53, 3095.
5. Meeteren, L.; Ruurs, P.; Christodoulou, E.; Goding, J.; Takakusa, H.; Kikuchi, K.;
Perrakia, A.; Nagano, T.; Moolenaar, W. J. Biol. Chem. 2005, 280, 21155.
6. Hook, S.; Ragan, S.; Hopper, D.; Honemann, C.; Durieux, M.; Macdonald, T.;
Lynch, K. Mol. Pharm. 1998, 53, 188.
HA51
0.187
0.367
F
O
OH
O
P
O
OH
S32826
Br-LPA
C13H27
N
H
OH Br
HO
O
C17H35
O
40.1
P
OH
7. Heasley, B.; Jarosz, R.; Lynch, K.; Macdonald, T. Bioorg. Med. Chem. Lett. 2004, 14,
2735.
O

7136
J. E. East et al. / Bioorg. Med. Chem. Lett. 20 (2010) 7132–7136
8. Heasley, B.; Jarosz, R.; Carter, K.; Van, S.; Lynch, K.; Macdonald, T. Bioorg. Med.
Chem. Lett. 2004, 14, 4069.
9. Santos, W.; Heasley, B.; Jarosz, R.; Lynch, K.; Macdonald, T. Bioorg. Med. Chem.
Lett. 2004, 14, 3473.
10. Cui, P.; Tomsig, J.; McCalmont, W.; Lee, S.; Becker, C.; Lynch, K.; Macdonald, T.
Bioorg. Med. Chem. Lett. 2007, 17, 1634.
11. Cui, P.; McCalmont, W.; Tomsig, J.; Lynch, K.; Macdonald, T. Bioorg. Med. Chem.
2008, 16, 2212.
16. (a) Hansch, C.; Muir, R. M.; Fujita, T.; Miloney, P. P.; Geiger, F.; Streich, M. J. Am.
Chem. Soc. 1963, 85, 2817; (b) Hansch, C.; Fukunaga, J. Y.; Jow, Y. C. J. Med.
Chem. 1977, 20, 96.
17. (a) Hammett, L. P. J. Am. Chem. Soc. 1937, 59, 96; (b) Charton, M. J. Org. Chem.
1963, 28, 3121.
18. Ertl, P.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 43, 3714.
19. Albers, H.; Dong, A.; van Meeteren, L. A.; Egan, D.; Sunkara, M.; van Tilburg, D.;
Schuurman, K.; van Tellingen, O.; Morris, A.; Smyth, S.; Moolenaar, W.; Ovaa, H.
Proc. Natl. Acad. Sci. U.S.A. 2010, 107(16), 7257.
12. Luche, J. L. J. Am. Chem. Soc. 1978, 100, 2226.
13. Parrill, A. L.; Echols, U.; Nguyen, T.; Pham, T.-C. T.; Hoeglund, A.; Baker, D. L.
Bioorg. Med. Chem. 2008, 16, 1784.
14. Zalatan, J. G.; Fenn, T. D.; Brunger, A. T.; Herschlag, D. Biochemistry 2006, 45, 9788.
15. Molecular Operating Environment (MOE 2009.10); C.C.G., Inc., 1010
Sherbrooke West, Suite 910, Montreal, Quebec, Canada H3A 2R7.
20. Ferry, G.; Moulharat, N.; Pradere, J.; Desos, P.; Try, A.; Genton, A.; Giganti, A.;
Beucher-Gaudin, M.; Lonchampt, M.; Bertrand, M.; Saulnier-Blache, J.; Tucker,
G.; Cordi, A.; Boutin, J. J. Pharmacol. Exp. Ther. 2008, 327, 809.
21. Zhang, H.; Xu, X.; Gajewiak, J.; Tsukahara, R.; Fujiwara, Y.; Liu, J.; Fells, J.;
Perygin, D.; Parrill, A.; Tigyi, G.; Prestwich, G. Cancer Res. 2009, 69, 5441.