Orally active 7-substituted (4-benzylphthalazin-1-yl)-2-methylpiperazin-1- yl]nicotinonitriles as active-site inhibitors of sphingosine 1-phosphate lyase for the treatment of multiple sclerosis

Article abstract of DOI:10.1021/jm500338n

Sphingosine 1-phosphate (S1P) lyase has recently been implicated as a therapeutic target for the treatment of multiple sclerosis (MS), based on studies in a genetic mouse model. Potent active site directed inhibitors of the enzyme are not known so far. Here we describe the discovery of (4-benzylphthalazin-1-yl)-2-methylpiperazin-1-yl]nicotinonitrile 5 in a high-throughput screen using a biochemical assay, and its further optimization. This class of compounds was found to inhibit catalytic activity of S1PL by binding to the active site of the enzyme, as seen in the cocrystal structure of derivative 31 with the homodimeric human S1P lyase. 31 induces profound reduction of peripheral T cell numbers after oral dosage and confers pronounced protection in a rat model of multiple sclerosis. In conclusion, this novel class of direct S1P lyase inhibitors provides excellent tools to further explore the therapeutic potential of T cell-targeted therapies in multiple sclerosis and other autoimmune and inflammatory diseases.

Full text of DOI:10.1021/jm500338n

Featured Article
pubs.acs.org/jmc
Orally Active 7‑Substituted (4-Benzylphthalazin-1-yl)-2-
methylpiperazin-1-yl]nicotinonitriles as Active-Site Inhibitors
of Sphingosine 1‑Phosphate Lyase for the Treatment of
Multiple Sclerosis
Sven Weiler,* Nadine Braendlin, Christian Beerli, Christian Bergsdorf, Anna Schubart, Honnappa Srinivas,
Berndt Oberhauser, and Andreas Billich
Novartis Institutes for BioMedical Research, Basel, CH-4002, Switzerland
ABSTRACT: Sphingosine 1-phosphate (S1P) lyase has recently been implicated as a thera-
peutic target for the treatment of multiple sclerosis (MS), based on studies in a genetic mouse
model. Potent active site directed inhibitors of the enzyme are not known so far. Here we
describe the discovery of (4-benzylphthalazin-1-yl)-2-methylpiperazin-1-yl]nicotinonitrile 5 in a
high-throughput screen using a biochemical assay, and its further optimization. This class of
compounds was found to inhibit catalytic activity of S1PL by binding to the active site of the
enzyme, as seen in the cocrystal structure of derivative 31 with the homodimeric human S1P
lyase. 31 induces profound reduction of peripheral T cell numbers after oral dosage and confers
pronounced protection in a rat model of multiple sclerosis. In conclusion, this novel class of
direct S1P lyase inhibitors provides excellent tools to further explore the therapeutic potential of
T cell-targeted therapies in multiple sclerosis and other autoimmune and inflammatory diseases.
rodents (Figure 1).7 More recently, derivatives of 3, such as 4
INTRODUCTION
■
(LX2931), have been synthesized.8,9 While these compounds do
Multiple sclerosis (MS) is a chronic, progressive inflammatory
not block S1PL activity in biochemical and cellular assays,9−11
disorder of the central nervous system (CNS). The disease is
they cause reduction of S1PL activity when administered in vivo,
mediated by pathogenic T cell responses against myelin antigens,
leading to elevated tissue S1P and T cell redistribution.4,9,12
leading to demyelination of axons in the central nervous system
(CNS) which is followed by neurodegeneration.1 Invasion of
Compounds that potently inhibit S1PL activity directly by
binding to the enzyme have not been reported so far, and as
pathogenic T cells into the CNS can be limited by reducing their
stated in a recent review,13 “the run is open to find the first-in-
egress from lymph nodes; this is achievable by down-modulation
of sphingosine 1-phosphate (S1P) receptor 1 (S1P1) in T cells.2 1
class direct S1P lyase inhibitor”. Here we report on novel
compounds that inhibit enzyme activity of purified S1PL and, as
(fingolimod, FTY720, Figure 1), a compound acting as functional
seen in the cocrystal structure with human S1PL, bind to the
active site of the enzyme. They cause an increase of S1P in
cultured cells and in rodents in vivo, where they rapidly induce
peripheral lymphopenia. Finally, for a selected derivative we
show a pronounced protective effect in rat EAE after oral dosage.
S1P1 antagonist after metabolic activation, is a disease modifying
therapy approved to treat relapsing forms of MS.3
An alternative pharmacological approach to prevent T cell
egress from the lymph nodes is to induce increased intracellular
S1P concentrations, which has been shown to ensue down-
modulation of S1P1 from the cell membrane of T cells.4 This can
be achieved by inhibiting the microsomal enzyme S1P lyase
(S1PL)5 which catalyzes the irreversible retro-aldol reaction of
S1P to 2-hexadecenal and phosphoethanolamine and tightly
controls intracellular S1P levels. We have shown that inducible
S1PL knockout mice featuring partial reduction of S1PL activity
are protected in experimental autoimmune encephalomyelitis
(EAE), a model of MS. In particular, T cell immigration into the
CNS was found to be profoundly reduced.6 Thus, S1PL may be a
therapeutic target for the treatment of MS and potentially other
T cell dependent autoimmune diseases.
RESULTS AND DISCUSSION
■
High-Throughput Screening. Mammalian S1PL is known
to be an intracellular protein, anchored via its N-terminus to
microsomal membranes with its catalytic site facing the
cytoplasm.14 The N-terminal region was shown to be dispensable
for enzyme activity.15 To provide a soluble form of the protein,
we expressed N-terminally shortened human S1PL (Δ1−61) in
insect cells and purified it to homogeneity. This protein was
catalytically active as shown by monitoring formation of the
labile cleavage product 2-hexadecenal from S1P as the natural
Functional S1PL inhibitors described so far include 2
(deoxypyridoxine, DOP4) and 3 (2-acetyl-4-tetrahydroxybutyl-
imidazole, THI), originally identified as a minor constituent of
ammonia caramel food coloring causing immunosuppression in
Received: March 3, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
Figure 1. Structures of 1, selected indirect S1PL inhibitors, and S1PL screening hit 5.
Figure 2. Reaction scheme of the assay to determine S1PL enzyme
activity.
Figure 3. Inhibition of S1PL (Δ1−61) catalytic activity by 5. See
Experimental Section for details of the assay method. The ratio of the
product peak area over the peak area of the internal standard after 1 h of
incubation with the enzyme is shown at graded concentrations of 5.
substrate; the aldehyde was trapped as a hydrazone that was then
quantified by LC/MS (Figure 2). Under the assay conditions
used (see Experimental Section for details), the Km value for S1P
cleavage by S1PL (Δ1−61) was determined to 5.2 0.9 μM.
The functional S1PL inhibitors 3 and 4 did not block catalytic
activity in the assay up to the highest concentration of 100 μM. 1,
described as inhibitor of S1PL with IC50 = 52 μM in assays using
microsomal preparations or tissue lysate as source of enzyme
activity,16,17 showed IC50 = 17 μM under our assay conditions. By
use of the biochemical assay in an HTS format to screen 250 000
compounds from the Novartis compound library, 5 was iden-
tified as a potent inhibitor, featuring IC50 = 2.4 0.2 μM in the
enzymatic assay (Figure 3).
The overarching goal of the optimization strategy was to
develop a potent and selective orally active S1P-lyase inhibitor
that would enable target validation experiments in vivo.
Chemistry. To prepare derivatives, the A- and B-rings of 5
were assembled via one of the three routes outlined in Scheme 1.
Route A started with a thermal coupling of (R)-2-methylpiper-
azine 7 and 1-benzyl-4-chlorophthalazine 6 to afford (R)-1-
benzyl-4-(3-methylpiperazin-1-yl)phthalazine 8 which was then
reacted with differently substituted 2-chloropyridines 9a−c to
give the final products 14, 15, and 16. Saponification of 14
yielded 17.
9d to afford differently substituted (pyridin-2-yl)-3-alkylpiper-
azine-1-carboxylic acid tert-butyl esters 11a−f. The BOC-group
was removed under acidic conditions to yield the corresponding
(pyridin-2-yl)-3-alkylpiperazines 12a−f which were coupled
without further purification with 6 under heating to yield the
desired products 5, 18, 19, 20, 21, and 22.
For variation in the D-region (route C) commercially available
1,4-dichlorophthalazine 13 was coupled with 12a to yield the
central building block 6-[(R)-4-(4-chlorophthalazin-1-yl)-2-
methylpiperazin-1-yl]nicotinonitrile 23. This was reacted to 24
and 25, respectively, using either phenylboronic acid or the cor-
responding phenylethylzinc reagent which was prepared in situ
from (2-bromoethyl)benzene under Ni catalysis.
For derivatization of the C-ring we started from commercially
available 1,4,6-trichlorophthalazine 26 (Scheme 2). Not unex-
pectedly, differentiation of the two chlorine atoms was not
observed and the reaction with 1 equiv of 12a at room temper-
ature led to the corresponding 4,6- and 4,7-dichlorophthalazine
derivatives 28 as an inseparable mixture. Further reaction of this
mixture with benzylzinc bromide under Pd catalysis led to the
final products 30 and 31 which could be separated using
In route B, commercially available BOC-3-alkylpiperazines
10a−f were coupled using Cu catalysis to 6-chloronicotinonitrile
B
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
a
Scheme 1. Synthetic Routes toward Derivatives Modified in the A, B, and D Regions
a
(a) Et3N, dioxane, 180 °C; (b) K2CO3, tetrakis(acetonitrile)copper(I) hexafluorophosphate, DMSO, 140 °C; (c) TFA, CH2Cl2, rt; (d) LiOH,
EtOH/H2O, 60 °C; (e) Et3N, DMAP, n-BuOH, 140 °C; (f) phenylboronic acid, Na2CO3, tetrakis(triphenylphosphine)palladium(0), DME/water,
150 °C; (g) (2-bromoethyl)benzene, 1,3-bis(dpp)propane-Ni(II), diethylzinc, dioxane, 90 °C.
preparative reverse phase HPLC. The corresponding 6- and
7-trifluoromethyl analogues 32 and 33 were synthesized in the
same way using 1,4-dichloro-6-trifluoromethylphthalazine 27 as
starting material. The chlorine substituent in 30 and 31 could
be replaced by cyanide with Zn cyanide under Pd catalysis
leading to the corresponding 6- and 7-cyano derivatives 34 and
35 (Scheme 2).
Structure−Activity Relationship (SAR). In order to
efficiently optimize 5, we explored the SAR of the four regions
separately. Substitution of the cyano group in the A-ring of 5 by
other residues markedly modulated the potency of S1PL
inhibition. While the ethyl ester 14 was equipotent to the parent
compound 5, the corresponding carboxylic acid 17 proved to be
inactive. The electronic contribution of 4-substituents on the
A-ring proved tobe an important factor, as replacing the CN group
by a CF3 group (15) led to a ∼10-fold reduction of potency
whereas a NO2 group (16) yielded a ∼3-fold increase (Table 1).
Potency of the derivatives was also very sensitive to changes at
the alkyl substituent in the B-ring. Inverting the stereochemistry
of the methyl substituent (18) or omitting it (19) led to inactive
compounds. While small changes in the alkyl chain were toler-
ated (20, 21), introduction of polar substituents, as exemplified
by 22, again yielded inactive derivatives. For the D-ring a benzylic
substitution proved to be optimal. Exchanging benzyl for phenyl
(24), phenylethyl (25), or chlorine (23) resulted in loss of
activity.
The part of the initial hit compound most sensitive to modu-
lation of activity turned out to be region C (Table 2). It became
clear that the 7-position was much preferred over the 6-position,
with 7-substituted compounds being at least 10-fold more potent
than their 6-substitued counterparts (31 vs 30; 33 vs 32; 35 vs
34). Replacing the hydrogen atom by small substituents yielded
potent derivatives with the chloro compound 31 giving an
approximately 10-fold increase in potency compared to 5. The
corresponding cyano derivative 35 was even 100-fold more
potent than the parent compound.
The initial hit compound 5 was originally described as a potent
antagonist of the receptor protein Smoothened (SMO).18 While
this activity was retained by 31, the most potent S1PL inhibitor
identified here, 35, featured no detectable activity in the Smooth-
ened assay (Table 3). Also, a selection of structurally diverse
inhibitors from the Novartis Smoothened project did not score
in the S1P lyase assays (data not shown). Thus, it was gratifying
to see that SAR for S1PL and Smoothened clearly diverged.
C
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
a
Scheme 2. Synthetic Route toward Derivatives Modified at the C Ring
a
(a) Et3N, NMP, rt; (b) benzylzinc bromide, tetrakis(triphenylphosphine)palladium(0), THF, 90 °C ; (c) zinc cyanide, Pd2(dba)3, X-Phos, DMF,
120 °C, microwave.
Table 1. Inhibition of Human S1PL by Derivatives of 5
Modified in Regions A, B, and D
Table 2. Inhibition of Human S1PL by Derivatives of 5
Modified in Region C
a
a
R1
R2
CN
R3
compd
IC50 [μM]
A Ring
Bn
R
compd
IC50 [μM]
(R)-Me
(R)-Me
(R)-Me
(R)-Me
(R)-Me
5
2.4 0.2
3.4 0.9
30 1.3
0.74 0.15
>30
H
5
2.4 0.2
COOEt
CF3
Bn
14
15
16
17
6-Cl
30
31
32
33
34
35
2.3 0.4
Bn
7-Cl
0.21 0.04
12 0.3
NO2
Bn
6-CF3
7-CF3
6-CN
7-CN
COOH
Bn
0.10 0.02
1.3 0.2
B Ring
Bn
(S)-Me
CN
CN
CN
CN
CN
CN
CN
CN
18
19
20
21
22
23
24
25
>30
0.024 0.005
H
Bn
>30
a
See Experimental Section for assay conditions.
(R)-Et
Bn
1.7 0.3
2.5 0.5
>30
(R)-isopropyl
(S)-CH2OH
(R)-Me
(R)-Me
(R)-Me
Bn
(4-fluoro-N-methyl-N-(1-(4-(1-methyl-1H-pyrazol-5-yl)-
phthalazin-1-yl)piperidin-4-yl)-2-(trifluoromethyl)benzamide)
shows that the phenyl part of the phthalazine core is placed in a
tight hydrophobic pocket. Therefore, a substituent in this region
is likely to diminish Smoothened binding; this might explain the
observed loss of activity when introducing a cyano group in the
7-position.
Bn
Cl
>30
Ph
>30
Ph(CH2)2-
30
a
See Experimental Section for assay conditions.
The recently published cocrystal structure of the Smooth-
ened receptor with the phthalazine antagonist19 LY2940680
One derivative, compound 31, was tested on the rat and human
enzyme in liver preparations and showed comparable inhibition
D
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
The inhibitor 31 is bound in the substrate binding site about
5 Å distal to the catalytic center at the branching point of a
Y-shaped channel which links the buried active site to the exterior.
The distal part of the channel is only partly resolved and con-
stitutes the binding region for the lipid chain of the substrate21
This rather narrow and hydrophobic channel is largely formed
by flexible hydrophobic side chains (Met, Leu, Ile, Phe, etc.)
and exhibits a complementary surface to the cyanopyridyl-
methylpiperazine side chain of the inhibitor. The methyl sub-
stituent at the piperazine is forced into an axial conformation by
the coplanar N-pyridyl substituent and induces a pronounced
left-handed twist of cyanopyridylpiperazine side chain, explain-
ing the >10-fold difference in potency of the enantiomers
(5 vs 18).
The position of the phthalazine core that extends into the side
channel is stabilized by π-stacking and an H-bridge of a backbone
NH to one of the ring nitrogens. This represents the single
canonical H-bridge in the protein−inhibitor complex. The chlo-
rine at the 7-position, which marks a critical area for potency, is
pointing toward an unresolved region and exerts a side-on
interaction with Asp176. The 4-benzyl substituent is entering the
central part of the binding site and reaches halfway to the
catalytic core without interacting with polar amino acids of
the catalytic machinery or the pyridoxal phosphate cofactor. In
the crystal, this polar site is occupied by water and a succinate
molecule originating from the crystallization buffer.
Table 3. Diverging SAR for Key Compounds with Respect to
Inhibition of S1PL and Smoothened Inhibition
inhibition of S1PL,
Smoothened antagonism,
a
R
compd
IC50 [μM]
IC50 [μM]
b
H
5
2.4 0.2
0.24 0.024
7-Cl
31
35
0.21 0.04
0.024 0.005
0.44 0.031
7-CN
>25
a
b
See Experimental Section for assay conditions. Data taken from ref 18.
(data not shown). 31 was also selective in a Novartis in-house
screening panel toward >50 receptors and kinases and therefore
suited as a tool compound for in vivo experiments.
Crystal Structure of Human S1PL Δ1−61 in Complex
with 31. In order to aid future optimization of S1PL inhibitors,
we cocrystallized soluble S1PL (residues 62−568) with the
potent inhibitor 31. The asymmetric unit of the crystal contains
two monomers, denoted A and B (Figure 4). These two subunits
form a tightly intertwined dimer with both chains contributing to
the catalytic cavity defined by the covalently bound cofactor
pyridoxal phosphate (PLP). In each subunit the N-terminal re-
gions spanning residues 62−109 and the C-terminal 18 residues
(554 to 568) were disordered, corresponding to absence or
poor electron density. In the folded core of S1PL Δ1−61, the
N-terminal domain spans residues 110−147, the central PLP
cofactor binding (at K353) domain extends from 148 to 424 and
the C-terminal domain from 425 to 553. The core exhibits the
typical type 1 PLP-dependent enzyme fold of the pyridoxal-
dependent decarboxylase subfamily.20 The overall structure,
cofactor binding mode, and the active site in general are well
conserved between human S1PL and Dpl1p (yeast S1P lyase21)
with monomer structural overlay rmsd of 0.8 Å over 400
common Cα atoms.
Pharmacodynamic Effect of Oral S1PL Inhibitor 31 in
the Rat. As we had reported earlier, 5 and 31 not only inhibit
purified S1PL but also act as inhibitors of the intrinsic S1PL
activity in mouse and human cells, dose-dependently inducing an
increase of intracellular S1P when added to cultured cells that are
supplied with sphingosine as precursor.10,11 To explore if 31, as
Figure 4. (a) Cartoon representation of S1PL (Δ1−59), with chain A in green and chain B in blue. The cofactor (PLP) in both chains, together with the
covalently bound side chain of K353, is in stick representation (magenta). The N and C termini of chain A are indicated. The inhibitor 31 is in stick
representation (orange). (b) Close-up view of the active sites and 31 binding site of S1PL. The inhibitor interaction with amino acid residues are shown
with dashed lines (H-bond, green; van der Waals, yellow; π-stacking, magenta).
E
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
Figure 5. Pharmacokinetics (PK) and pharmacodynamics (PD) of 31 in rats. (a) Single-dose PK study: 1 mg/kg iv (squares); 3 mg/kg po (triangles);
see Experimental Section for details. (b) Single-dose PK/PD study; CD4-positive T-cells in blood were measured by FACS at the indicated time points
after application of vehicle (open bars) or 31 (closed bars; 50 mg/kg po). Concentrations of 31 measured in the blood are indicated underneath. (c) 31
was given 4 times at 2 mg/kg po in 12 h intervals. At 12 h after the last dose, blood was analyzed by flow cytometry.
the more potent derivative, would also block S1PL in experimental
animals in vivo, we first determined the pharmacokinetics of the
compound after intravenous and oral dosing to male Sprague−
Dawley rats (Figure 5a). Following intravenous injection of a
solution at 1 mg/kg body weight, 31 showed low blood clearance
(23 mL min−1 kg−1) and a long half-life (9.1 h). After oral
dosage as a suspension at 3 mg/kg, good oral exposure AUC0−∞
(1.3 μM·h) with Cmax of 99 nM was reached after 5 h; oral
bioavailability was moderate at 26%. These data warranted
further investigation of 31 in vivo.
Next, we asked whether 31 would induce a reduction of
peripheral CD4+ T-cells. As shown in Figure 5b, when rats were
dosed at 50 mg/kg po, CD4+ T cell numbers were reduced by
>80% at 14 and 24 h after dosing; they were still markedly
reduced after 2 days but returned to normal values after 3 days.
This time course of lymphopenia was likewise observed for CD8+
T-cells (data not shown); it correlates well with the measured
exposure to 31 (Figure 5b).
In order to prepare for studies in a disease model with pro-
longed dosing, we asked whether pronounced lymphopenia
could also be achieved using multiple but significantly lower
doses of 31. Indeed, as shown in Figure 5c, after four doses of
2 mg/kg given at 12 h intervals we observed pronounced reduc-
tion of CD4- and CD8-positive T-cells (>95% reduction) at the
trough and a less complete effect on the B-cells (70% reduction);
exposure to the compound was 0.4 μM at the trough. Numbers of
neutrophils and monocytes were not changed at all (data not
shown), explaining the only partial reduction of total white blood
cells which is obviously solely attributable to the reduction of
lymphocytes.
Pharmacological Inhibition of S1PL Protects in
Experimental Autoimmune Encephalomyelitis (EAE) in
Rats. Previously, we demonstrated that partially S1PL-deficient
mice are fully protected in the model of MOG-induced experi-
mental autoimmune encephalomyelitis (EAE).6 Next we tested
31 for its efficacy to suppress EAE development in rats. Treat-
ment (2 mg/kg po, b.i.d.) was started 2 days before the induction
of EAE. While vehicle-treated rats developed EAE (Figure 6a), all
rats dosed with compound 31 were fully protected and did not
show any signs of EAE. Accordingly, spinal cord tissue from
compound-treated animals did not show any histopathological
changes related to EAE (Figure 6b and Figure 6c), including
absence of infiltration of H&E (hematoxylin and eosin) stained
lymphocytes. Concentration of 31 in blood and brain at the end
of the study was 0.84 and 0.61 μM, respectively. Concentrations
of S1P in the lymph nodes at the end of the study was 38
3.8 pmol/mg tissue in compound-treated rats, corresponding
to a ∼150-fold increase vs vehicle (0.25
0.07 pmol/mg),
while S1P in brain only increased from 0.4 0.04 pmol/mg to
2.4 0.2 pmol/mg.
SUMMARY AND CONCLUSIONS
■
We had previously demonstrated that inducible S1PL knockout
mice featuring partial deficiency of the enzyme activity show
peripheral lymphopenia and are protected in models of delayed-
type hypersensitivity and in EAE.6 These studies, together with
the reported efficacy on indirect S1PL inhibitors in models of
arthritis,8,9 identified S1PL as an attractive pharmacological target
for the treatment of inflammatory and autoimmune diseases.
Previous studies have measured human S1PL enzyme activity in
F
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
Figure 6. Pharmacological inhibition of S1PL protects rats from EAE. EAE was induced in female DA rats by MOG-immunization; 31 was dosed at
2 mg/kg b.i.d., starting 2 days before immunization: (a) clinical score; (b) histological analysis of thoracic sections of spinal cord tissue from vehicle and
compound-treated mice undergoing EAE (day 14; staining with H&E to visualize CNS-invading cells, scale bar 200 μm; arrows highlight areas of
inflammation); (c) semiquantative analysis of infiltration (H&E) in cervical and thoracic spinal cord.
cell lysates or suspensions of cell membranes,15,17 and attempts to
purify the enzyme failed.18 Our successful preparation of a soluble
form of human S1PL enabled screening for inhibitors in a
biochemical assay. A high-throughput screening campaign using a
diversity enriched library of 250 000 compounds led to the
discovery of 5, a 2.4 μM inhibitor from the phthalazine class,
originating from a program targeting the Smoothened receptor.
The screening hit not only showed activity in the biochemical
assay but also was active in the cellular assay in HEK293 cells
targeting endogenous SPL.
We rapidly identified the chiral piperazine side chain to be
essential for potency and the 7-position of the aminophthalazine
to control the overall pharmacological profile, including potency,
selectivity against the Smoothened pathway, and metabolic
stability. When a chloro or a cyano substituent (31 and 35) was
introduced at the 7-position, the biochemical potency was
improved 14- and 100-fold and the selectivity 4- and 20-fold,
respectively. In the case of 31, the in vitro clearance in rat
liver microsomes was also improved 2-fold (data not shown)
resulting in a PK profile suitable for peroral dosing in the rat
EAE model.
the CNS and may not involve, for example, down-regulation of
S1P1 on astrocytes as shown for 1.23
While our studies focused on MS as potential indication for
S1PL inhibitors, others have implicated rheumatoid arthritis,9
atherosclerosis,24 acute lung injury,25 and myopathy26 as con-
ditions where pharmacological inhibition of S1PL may translate
into therapeutic benefit. Our discovery of a potent class of oral
S1PL inhibitors will enable further assessment of these options in
preclinical models.
EXPERIMENTAL SECTION
■
Materials. 1 is an internal reference compound. 3 was obtained from
Bioduro, San Diego, CA. 4 was prepared at Novartis, following the
synthesis described.8 If not mentioned otherwise, all reagents were
purchased from Sigma-Aldrich. 6, 7, 9a−d, 10a−f, 13, 26, and 27 are
commercially available. 5, 8, 12a, 14, 18, and 19 were synthesized as
described earlier.18
Purity. Compound purity has been analyzed using analytical reverse
phase HPLC (Kinetex C18, 2.1 mm × 50 mm, 5 μm; solvent A, water +
0.1% TFA; solvent B, acetonitrile + 0.1% TFA; gradient, 0 min, 10% B;
2 min, 95% B; 3 min, 95% B). Purity was >95% for all described
compounds.
General Procedure A for Addition of Heteroaryl Chlorides to
1-Benzyl-4-piperazin-1-ylphthalazine. 8 (0.471 mmol, 1 equiv),
the heteroaryl chloride (9a−c, 0.942 mmol, 2 equiv), and triethylamine
(2.355 mmol, 5 equiv) were dissolved in 3 mL of dioxane, and the
resulting mixture was heated in a closed vial to 160 °C for 40 h.
Afterward the solvent was evaporated and the residual material was
purified via reverse phase HPLC.
Furthermore, the potent binding of 31 enabled cocrystalliza-
tion of the purified enzyme, providing the first structure of
human S1PL and thus paving the way toward future structure-
based optimization of inhibitors.
We showed for the first time that an active site directed S1PL
inhibitor induces an increase of S1P in cells and tissues, as pre-
dicted from genetic models and functional inhibitors.4,6,8,9,22
Most importantly, oral treatment of rats with inhibitor 31
induced peripheral lymphopenia, similar to what is seen in
murine S1PL deficiency, as a basis for its protective effect in EAE.
It is important to note that while the compound enters the CNS,
it did induce only a minor increase of S1P in the brain. Thus, as
previously shown for S1PL deficient mice,6 it appears that the
enzyme predominantly controls S1P levels in most tissues,
including the lymphoid organs but not in the CNS. Furthermore,
the data imply that the protective effect of 31 in EAE is
sufficiently explained by its inhibition of T cell immigration into
(R)-1-Benzyl-4-(3-methyl-4-(5-(trifluoromethyl)pyridin-2-yl)-
piperazin-1-yl)phthalazine (15) was synthesized from 8 and 9b in 24%
yield according to general procedure A. 1H NMR (360 MHz, DMSO-
d6) δ 8.49 (s, 1H), 8.24 (t, J = 6.44 Hz, 2H), 7.82−8.05 (m, 3H), 7.25−
7.37 (m, 4H), 7.16−7.23 (m, 1H), 7.02 (d, J = 9.09 Hz, 1H), 4.82−4.91
(m, 1H), 4.62 (s, 2H), 4.39 (d, J = 12.63 Hz, 1H), 3.91 (d, J = 11.37 Hz,
1H), 3.78 (d, J = 12.63 Hz, 1H), 3.48−3.63 (m, 1H), 3.00−3.18 (m,
1H), 1.43 (d, J = 6.57 Hz, 3H). LC/MS: t = 1.926, 100%, 464.2. HRMS
+
m/z 464.2066 [(M + H)+ calcd for C26H24F3N6 , 464.206 204].
(R)-1-Benzyl-4-(3-methyl-4-(5-nitropyridin-2-yl)piperazin-1-yl)-
phthalazine (16) was synthesized from 8 and 9c in 21% yield according
1
to general procedure A. H NMR (360 MHz, DMSO-d6) δ 9.05 J =
2.78 Hz, 1H), 8.19−8.36 (m, 3H), 7.96 (t, J = 8.84 Hz, 2H), 7.24−7.39
G
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
(m, 4H), 7.14−7.23 (m, 1H), 7.03 (d, J = 9.60 Hz, 1H), 4.62 (s, 2H),
4.54 (d, J = 13.64 Hz, 1H), 3.89−4.00 (m, 1H), 3.80 (d, J = 12.88 Hz,
1H), 3.63−3.75 (m, 1H), 3.25−3.35 (m, 2H), 3.13 (dt, J = 3.54, 12.25 Hz,
1H), 1.48 (d, J = 6.57 Hz, 3H). LC/MS: t = 1.837, 100%, 441.2. HRMS
(R)-6-(4-(4-Chlorophthalazin-1-yl)-2-methylpiperazin-1-yl)-
nicotinonitrile (23). 13 (500 mg, 2.51 mmol) and 12a (381 mg,
1.884 mmol) were suspended in n-BuOH (10 mL) at room temperature
under N2. Then Et3N (525 μL, 3.77 mmol) and DMAP (30.7 mg,
0.251 mmol) were added and the mixture was heated in a sealed tube at
140 °C. After 16 h the mixture was cooled to room temperature and the
solvent was removed in vacuo. Crystallization from i-PrOH yielded
310 mg of the desired material. 1H NMR (400 MHz, DMSO) δ 8.52 (d,
J = 2.34 Hz, 1H), 8.22−8.40 (m, 2H), 8.07−8.20 (m, 2H), 7.88 (dd, J =
2.54, 9.18 Hz, 1H), 6.94 (d, J = 8.98 Hz, 1H), 4.74−4.86 (m, 1H), 4.30
(d, J = 13.67 Hz, 1H), 4.08 (m, 1H), 3.85 (d, J = 12.89 Hz, 1H), 3.68 (m,
1H), 3.56 (dd, J = 3.51, 12.89 Hz, 1H), 3.33 (m, 1H), 1.36 (d, J =
6.64 Hz, 3H). LC/MS: t = 1.97, 97%, 365.8.
+
m/z 441.2043 [(M + H)+ calcd for C25H24N6O2 , 441.203 899].
(R)-6-(4-(4-Benzylphthalazin-1-yl)-2-methylpiperazin-1-yl)-
nicotinic Acid (17). To a solution of 14 (ref 18, 80 mg, 0.171 mmol)
dissolved in EtOH (4 mL) and H2O (3 mL) was added LiOH·H2O
(71.8 mg, 1.711 mmol), and the resulting mixture was heated to 60 °C
overnight. After the mixture was cooled to room temperature EtOH was
removed in vacuo. The residual aqueous layer was acidified to pH ≈ 3
with 5% citric acid to obtain a white suspension. The mixture was stirred
1 h at room temperature and was then filtered off. The filter cake was
washed with water and dried in high vacuum to obtain 39.4 mg (52%) of
off-white solid. 1H NMR (360 MHz, DMSO-d6) δ 8.70 (s, 1H), 8.24 (t,
J = 6.95 Hz, 2H), 7.88−8.07 (m, 3H), 7.11−7.39 (m, 5H), 6.92 (d, J =
9.09 Hz, 1H), 4.86 (m, 1H), 4.62 (s, 2H), 4.42 (d, J = 12.38 Hz, 1H),
3.91 (d, J = 12.38 Hz, 1H), 3.78 (d, J = 12.88 Hz, 1H), 3.53−3.64 (m,
1H), 3.22−3.37 (m, 2H), 3.06−3.19 (m, 1H), 1.43 (d, J = 6.57 Hz, 3H).
LC/MS: t = 1.506, 100%, 440.2. HRMS m/z 440.2091 [(M + H)+ calcd
(R)-6-(2-Methyl-4-(4-phenylphthalazin-1-yl)piperazin-1-yl)-
nicotinonitrile (24). To a solution of 23 (100 mg, 0.206 mmol),
phenylboronic acid (27.6 mg, 0.226 mmol), and Na2CO3 (65.4 mg,
0.617 mmol) in DME and water was added tetrakis(triphenyl-
phosphine)palladium(0) (11.88 mg, 10.28 μmol), and the mixture
was heated for 15 min at 150 °C in the microwave. The mixture was
poured onto aqueous NaHCO3 and extracted with AcOEt. The organic
layer was washed with brine, dried over MgSO4, and concentrated in
vacuo. The final product was crystallized from methanol in 43% yield as
a white powder. 1H NMR (360 MHz, DMSO-d6) δ 8.58 (d, J = 2.27 Hz,
1H), 8.33 (d, J = 7.83 Hz, 1H), 7.90−8.06 (m, 4H), 7.67−7.72 (m, 2H),
7.57−7.65 (m, 3H), 7.02 (d, J = 9.09 Hz, 1H), 4.90 (m, 1H), 4.45 (d, J =
13.39 Hz, 1H), 3.95−4.08 (m, 1H), 3.87 (d, J = 12.88 Hz, 1H), 3.64 (dd,
J = 9.73, 12.25 Hz, 1H), 3.38 (d, J = 3.54 Hz, 1H), 3.11−3.23 (m, 1H),
1.46 (d, J = 6.57 Hz, 3H). LC/MS: t = 1.727, 100%, 407.2. HRMS m/z
+
for C26H25N5O2 , 440.208 65].
General Route B for Synthesis of 6-(2-Alkylpiperazin-1-
yl)nicotinonitrile Derivatives 12a−f. 1-BOC-alkylpiperazine deriv-
atives (10a−f, 2.5 mmol, 1 equiv), 9d (2.75 mmol, 1.1 equiv), and K2CO3
(4.74 mmol, 1.9 equiv) were dissolved in 1 mL of DMSO. Tetrakis-
(acetonitrile)copper(I) hexafluorophosphate (0.022 mmol, 0.01 equiv)
was added, and the reaction mixture was heated to 140 °C. After 3 h the
reaction mixture was cooled to room temperature and was poured onto
AcOEt. The organic layer was washed with 0.1 M HCl (3×) followed by
aqueous NaHCO3 (2×). The organic layer was dried over MgSO4 and
concentrated in vacuo. Where necessary the residue (11a−f) was further
purified by flash chromatography.
BOC-protected 6-(2-alkylpiperazin-1-yl)nicotinonitrile deriva-
tives (11a−f, 0.75 mmol, 1 equiv) were dissolved at room temperature
in DCM (7.5 mL) before TFA (7.5 mmol, 10 equiv) was added. The
resulting mixture was stirred at room temperature. After 3.5 h the
mixture was neutralized by slow addition of solid Na2CO3. Layers
were separated, and the aquous layer was washed again 2× with DCM.
The combined organic layer was washed with brine, dried over MgSO4,
and concentrated in vacuum. The organic layer was dried over MgSO4
and concentrated in vacuo. The corresponding piperazine derivatives
12a−f were used in the next step without further purification.
+
407.1991 [(M + H)+ calcd for C25H22N6 , 407.198 42].
(R)-6-(2-Methyl-4-(4-phenethylphthalazin-1-yl)piperazin-1-
yl)nicotinonitrile (25). 1,3-Bis(dpp)propane-Ni(II) Cl (25 mg,
0.046 mmol) was dissolved in dioxane (5 mL) and flushed with
argon. 1 M diethylzinc in hexane (1.4 mL, 1.400 mmol) was added,
and the mixture was stirred for 10 min at room temperature. Then
(2-bromoethyl)benzene (0.2 mL, 1.478 mmol) was added. The resulting
mixture was refluxed for 1.5 h before adding 23 (80 mg, 0.219 mmol) as
a solution in dioxane. The reaction was continued for 2 h at 90 °C. The
mixture was purified by preparative HPLC to give the desired product in
16% yield. 1H NMR (600 MHz, DMSO-d6) d 8.57 (d, 3.0 Hz, 1H), 8.49
(d, J = 7.47 Hz, 1H), 8.39 (d, J = 8.28 Hz, 1H), 8.10−8.28 (m, 4H), 7.93
(dd, J = 1.92, 8.98 Hz, 1H), 7.14−7.36 (m, 3H), 6.97 (d, J = 9.08 Hz,
1H), 4.83 (br s, 1H), 4.35 (d, 12.9 Hz, 1H), 4.06 (d, 10.3 Hz, 1H), 3.86
(d, 12.7 Hz, 1H), 3.48−3.74 (m, 4H), 3.30 (br s, 1H), 3.12 (t, J =
8.07 Hz, 2H), 1.37 (d, J = 6.46 Hz, 3H). LC/MS: t = 1.845, 93%, 435.2.
(R)-6-(4-(4-Benzylphthalazin-1-yl)-2-ethylpiperazin-1-yl)nicotinonitrile
(20) was synthesized from 6 and 12d in 79% yield according to
the general procedure A described above. 1H NMR (360 MHz,
DMSO-d6) δ 8.53 (s, 1H), 8.23 (d, J = 8.34 Hz, 2H), 7.85−8.01
(m, 3H), 7.24−7.38 (m, 4H), 7.14−7.22 (m, 1H), 7.02 (d, J = 9.09
Hz, 1H), 4.58−4.73 (m, 3H), 4.50 (d, J = 11.12 Hz, 1H), 3.80−3.92
(m, 2H), 3.59 (t, J = 11.12 Hz, 1H), 3.24 (dd, J = 3.54, 12.88 Hz, 1H),
2.98−3.12 (m, 1H), 1.97 (q, J = 7.40 Hz, 2H), 0.87 (t, J = 7.33 Hz,
3H). LC/MS: t = 1.844, 100%, 435.2. HRMS m/z 435.2297 [(M + H)+
+
HRMS m/z 435.2298 [(M + H)+ calcd for C27H26N6 , 435.229 72].
(R)-6-(4-(4-Benzyl-6-chlorophthalazin-1-yl)-2-methylpiperazin-
1-yl)nicotinonitrile (30) and (R)-6-(4-(4-Benzyl-7-chlorophthalazin-
1-yl)-2-methylpiperazin-1-yl)nicotinonitrile (31). 1,4,6-Trichlor-
ophthalazine 26 (1.5 g, 6.42 mmol, 1 equiv) and 12a (1.293 g,
6.42 mmol, 1 equiv) were dissolved in NMP (11 mL) at room tem-
perature under N2, and the mixture was stirred at room temperature
overnight. The resulting suspension was poured onto water and filtered.
The filter cake was dried in high vacuum to obtain 1.939 g (76%) of
beige solid of a ∼1:1 mixture of the two regioisomeric chlorides 28
which was used without further purification in the next step.
An amount of 300 mg of that mixture was suspended in THF
(0.5 mL) at room temperature under argon before tetrakis(triphenyl-
phosphine)palladium(0) (43.4 mg, 0.038 mmol) was added. Benzylzinc
bromide solution (1954 μL, 0.977 mmol) was added over a period of
7 min. The resulting mixture was heated to 90 °C overnight. The
reaction mixture was cooled to room temperature and diluted with
AcOEt and aqueous NaHCO3. Layers were separated, and the aqueous
layer was washed again 2× with AcOEt. The combined organic layer
were washed with brine (2×), dried over MgSO4, and concentrated
in vacuum. The residue was purified by preparative HPLC to obtain
59.2 mg of 30 (17%) and 63.3 mg of 31 (17%) as beige solid.
+
calcd for C27H26N6 , 435.229 719].
(R)-6-(4-(4-Benzylphthalazin-1-yl)-2-isopropylpiperazin-1-yl)-
nicotinonitrile (21) was synthesized from 6 and 12e in 16% yield
according to the general procedure A. 1H NMR (360 MHz, DMSO-d6)
δ 8.42−8.54 (m, 3H), 8.23−8.12 (m, 2H), 7.90 (dd, 1H), 7.22−7.48 (m,
5H), 7.07 (d, J = 9.09 Hz, 1H), 4.79 (s, 2H), 4.11 (m, 2H), 3.68 (t, J =
11.1 Hz, 1H), 3.5 (d, J = 10.4 Hz, 1H), 3.35 (m, 1H), 0.93 (d, J =
7.33 Hz, 3H). LC/MS: t = 1.908, 100%, 449.3. HRMS m/z 449.2460
+
[(M + H)+ calcd for C28H28N6 , 449.245 37].
(S)-6-(4-(4-Benzylphthalazin-1-yl)-2-(hydroxymethyl)piperazin-
1-yl)nicotinonitrile (22) was synthesized from 6 and 12f in 14% yield
according to the general procedure A. 1H NMR (600 MHz, DMSO-d6)
δ 8.53 (d, J = 2.02 Hz, 1H), 8.32−8.39 (m, 1H), 8.15−8.23 (m, 1H),
7.84−7.94 (m, 3H), 7.21−7.34 (m, 5H), 7.14−7.19 (m, 1H), 6.99 (d, J =
9.08 Hz, 1H), 5.13 (dd, J = 4.64, 6.06 Hz, 1H), 4.39−4.64 (m, 3H), 4.11
(d, J = 12.72 Hz, 1H), 3.98−4.06 (m, 1H), 3.87 (d, J = 12.31 Hz, 1H),
3.57−3.62 (m, 1H), 3.43−3.52 (m, 1H), 3.25 (m, 1H), 3.05−3.14 (m,
2H). LC/MS: t = 1.643, 100%, 437.2. HRMS m/z 437.2096 [(M + H)+
(R)-6-(4-(4-Benzyl-6-chlorophthalazin-1-yl)-2-methylpiperazin-
1-yl)nicotinonitrile (30). 1H NMR (360 MHz, DMSO-d6) δ 8.57 (d,
J = 2.27 Hz, 1H), 8.19−8.33 (m, 2H), 7.87−8.04 (m, 2H), 7.11−7.43
(m, 5H), 7.00 (d, J = 9.09 Hz, 1H), 4.85 (m, 1H), 4.63 (s, 2H), 4.41
+
calcd for C27H24N6 , 437.208 98].
H
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
(d, J = 13.39 Hz, 1H), 3.91 (d, J = 11.87 Hz, 1H), 3.76 (d, J = 12.88 Hz,
1H), 3.50−3.65 (m, 1H), 3.28 (dd, J = 3.54, 1H), 3.04−3.16 (m, 1H),
1.41 (d, J = 6.57 Hz, 3H). LC/MS: t = 1.22, 100%, M = 455.2. HRMS
Virus stocks were generated by amplification in Sf9 cells and large-scale
expression was performed in shaking flasks, using optimized conditions
with a multiplicity of infection of 1 and a cell density of 1.82 × 106 cells/mL
for 72 h. Frozen cell pellets were homogenized in 50 mM Tris-HCl
(pH 8.0), 500 mM NaCl, 1 mM tris(2-carboxyethyl)phosphine
(TCEP), 10% glycerol, 3 mM methionine, 0.1 mM pyridoxal phosphate,
0.1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate
and the protease inhibitor cocktail Complete/EDTA free (Roche),
and cells were lysed in a Potter-Elvehjem tissue homogenizer. The su-
pernatant after centrifugation was loaded onto Ni-NTA Superflow resin
(Qiagen). The protein that was eluted with a gradient of 20−500 mM
imidazole was subjected to fusion tag removal by PreScission site
cleavage overnight. Finally, the S1PL was further purified by size
exclusion chromatography (Superdex SPX200/16/60; GE Healthcare),
using 20 mM Tris-HCl, 500 mM NaCl, 1 mM TCEP, 5% glycerol, 3 mM
methionine, 1 mM PLP, pH 7.4 as elution buffer.
+
m/z 455.1747 [(M + H)+ calcd for C26H23ClN6 , 455.175 10].
(R)-6-(4-(4-Benzyl-7-chlorophthalazin-1-yl)-2-methylpiperazin-
1-yl)nicotinonitrile (31). 1H NMR (360 MHz, DMSO-d6) δ 8.57 (d,
J = 2.02 Hz, 1H), 8.27 (d, J = 8.84 Hz, 1H), 8.18 (s, 1H), 7.90−8.01 (m,
2H), 7.25−7.35 (m, 4H), 7.15−7.23 (m, 1H), 7.00 (d, J = 9.09 Hz, 1H),
4.86 (m, 1H), 4.62 (s, 2H), 4.41 (d, J = 13.14 Hz, 1H), 3.89 (d, J = 12.63
Hz, 1H), 3.76 (d, J = 12.63 Hz, 1H), 3.53−3.65 (m, 1H), 3.28 (dd,
J = 3.28, 12.88 Hz, 1H), 3.07−3.19 (m, 1H), 1.43 (d, J = 6.32 Hz, 3H).
LC/MS: t = 1.39, 100%, 455.1. HRMS m/z 455.1751 [(M + H)+ calcd
+
for C26H23ClN6 , 455.175 10].
(R)-6-(4-(4-Benzyl-6-(trifluoromethyl)phthalazin-1-yl)-2-
methylpiperazin-1-yl)nicotinonitrile (32) and (R)-6-(4-(4-Benzyl-
7-(trifluoromethyl)phthalazin-1-yl)-2-methylpiperazin-1-yl)-
nicotinonitrile (33). 32 and 33 were synthesized from 1,4-dichloro-6-
trifluoromethylphthalazine 27 via 29 in analogy to the synthesis used for
30 and 31 in 11% and 22% yields, respectively.27
For crystallization purposes, the protein was concentrated to
10 mg/mL. The resulting protein was estimated to be >95% pure and
homogeneous by SDS−PAGE and reverse phase HPLC. The identity of
the protein was further confirmed by N-terminal sequencing and mass
spectrometry (Q-Tof, Micromass, Waters).
(R)-6-(4-(4-Benzyl-6-(trifluoromethyl)phthalazin-1-yl)-2-
1
methylpiperazin-1-yl)nicotinonitrile (32). H NMR (400 MHz,
DMSO-d6) δ 8.47−8.61 (m, 2H), 8.43 (d, J = 8.68 Hz, 1H), 8.25 (dd, J =
1.34, 8.68 Hz, 1H), 7.92 (dd, J = 2.38, 9.11 Hz, 1H), 7.22−7.38 (m, 4H),
7.13−7.23 (m, 1H), 7.00 (d, J = 9.17 Hz, 1H), 4.86 (m, 1H), 4.72 (s,
2H), 4.41 (d, J = 13.20 Hz, 1H), 3.94 (d, J = 12.10 Hz, 1H), 3.80 (d, J =
12.84 Hz, 1H), 3.53−3.65 (m, 1H), 3.00−3.17 (m, 2H), 1.41 (d, J =
6.60 Hz, 3H). LC/MS: t = 1.29, 100%, 489.3. HRMS m/z 489.200 90
Assay of S1PL Enzyme Activity. Recombinant SPL (final concen-
tration, 0.8 μg/mL) was added to a mixture of S1P (10 μM) and
inhibitors (at graded concentrations; added from stock solutions in
DMSO; final DMSO, 4%) in 100 mM 4-(2-hydroxyethyl)-1-piperazine-
ethanesulfonic acid (HEPES) buffer, pH 7.4, containing 0.1 mM EDTA,
0.05% Triton X-100, 0.01% Pluronic F127 (Biotium), and 0.1 mM
pyridoxal 5′-phosphate. The reaction mixture was incubated at ambient
temperature for 1 h which was in the linear phase of the reaction time
course. The reaction was stopped by addition of isonicotinylhydrazide
(final concentration, 3.1 mM) and 2-trans-pentadecenal (3.7 μM,
internal standard) in acetonitrile/0.3 M H2SO4 1:1.
A 10 μL aliquot of the reaction mixture was subjected to a liquid
chromatography system (Shimadzu) for separation on a C8 column
(Luna, 3 μm, 2 mm × 20 mm, Phenomenex) at 35 °C, eluted in 1 min
with a gradient starting at 50% of 0.2% (v/v) formic acid in water
ramping up to 100% of 0.2% (v/v) formic acid in acetonitrile at a flow
rate of 0.5 mL/min within 1 min and held constant for 0.25 min. The LC
system was interfaced with a Thermo LTQ linear ion trap mass
spectrometer equipped with a heated electrospray ion source. The
instrument was operated in positive MS/MS ion mode. Transitions of
the reaction product (358.39 → 123.06, 236.28, 331.40, 340.37) and the
internal standard (344.36 → 122.96, 222.21, 317.36, 326.36) were
monitored; fragments for product and standard were summed up,
respectively. As a measure of substrate turnover, the area under the curve
(AUC) was calculated using the Xcalibur processing method. The ratio
of the AUC of product and internal standard was calculated for
normalization.
IC50 values for inhibition of product formation were calculated by
fitting the data to a four-parameter logistic equation using the software
XLfit (IDBS). Reported values represent means of at least three
experiments standard deviation. The Km value for S1P was determined
by measuring initial velocities at substrate concentrations in the range of
0.08−10 μM; calculation was done using the Michaelis−Menten
equation in Prism 6 (GraphPad Software, La Jolla).
X-ray Crystallography. The S1PL was cocrystallized with 31
(1 mM) by mixing equal volumes of the protein−ligand/reservoir
solutions. Crystals were grown at 20 °C by the hanging drop method.
The data set was collected using crystals obtained with a reservoir of 20%
PEG 3350 and 8% Tacsimate (both from Hampton). Crystals were flash
frozen in liquid nitrogen for data collection using 20% ethylene glycol as
cryoprotectant. Synchrotron data sets were collected at the Swiss Light
Source (Villigen PSI, Switzerland) protein beamline X10SA on a pixel-
counting PILATUS 6M detector. X-ray diffraction data were collected
at 100 K. The S1PL−31 complex structure was solved by molecular
replacement (using PHASER) with the structure of plant glutamate
decarboxylase (PDB code 1HBX) as a search model. Data collection and
refinement statistics are given in Table 4. Figures displaying structures
were prepared with the program PyMOL (DeLano Scientific LLC, San
Carlos, CA, http://www.pymol.org).
+
[(M + H)+ calcd for C27H23F3N7 , 489.200 904].
(R)-6-(4-(4-Benzyl-7-(trifluoromethyl)phthalazin-1-yl)-2-
methylpiperazin-1-yl)nicotinonitrile (33). 1H NMR (400 MHz,
DMSO-d6) δ 8.54 (d, J = 2.35 Hz, 1H), 8.37−8.46 (m, 2H), 8.22 (d, J =
8.99 Hz, 1H), 7.90 (dd, J = 2.35, 8.99 Hz, 1H), 7.22−7.34 (m, 5H),
7.10−7.21 (m, 1H), 6.97 (d, J = 8.99 Hz, 1H), 4.79−4.87 (m, 1H), 4.65
(s, 2H), 4.41 (d, J = 13.29 Hz, 1H), 3.89 (d, J = 12.12 Hz, 1H), 3.77 (d,
J = 12.51 Hz, 1H), 3.50 (dt, J = 3.13, 12.71 Hz, 1H), 3.11−3.28 (m, 2H),
1.42 (d, J = 6.65 Hz, 3H). LC/MS: t = 1.31, 100%, 489.2. HRMS m/z
+
489.2010 [(M + H)+ calcd for C27H23F3N7 , 489.201 453].
(R)-4-Benzyl-1-(4-(5-cyanopyridin-2-yl)-3-methylpiperazin-1-
yl)phthalazine-6-carbonitrile (34) was synthesized from 30 in analogy
to the procedure described for 35. 1H NMR (600 MHz, DMSO-d6) δ
8.87 (s, 1H), 8.55 (s, 1H), 8.26−8.39 (m, 2H), 7.91 (d, J = 9.08 Hz, 1H),
7.35 (d, J = 7.67 Hz, 2H), 7.28 (t, J = 7.06 Hz, 2H), 7.16−7.20 (m, 1H),
6.98 (d, J = 9.08 Hz, 1H), 4.82−4.88 (m, 1H), 4.65 (s, 2H), 4.38 (d, J =
12.92 Hz, 1H), 3.90 (d, J = 12.31 Hz, 1H), 3.75 (d, J = 12.51 Hz, 1H),
3.59 (t, J = 11.91 Hz, 1H), 3.31 (br s, 1H), 3.09 (t, J = 11.71 Hz, 1H),
1.38 (d, J = 6.46 Hz, 3H). LC/MS: t = 1.924, 98%, M = 446.2. HRMS m/
+
z 446.2087 [(M + H)+ calcd for C27H23N7 , 446.208 77].
(R)-1-Benzyl-4-(4-(5-cyanopyridin-2-yl)-3-methylpiperazin-
1-yl)phthalazine-6-carbonitrile (35). 31 (200 mg, 0.440 mmol), zinc
cyanide (51.6 mg, 0.440 mmol), Pd2(dba)3 (20.13 mg, 0.022 mmol),
and X-Phos (20.96 mg, 0.044 mmol) were mixed together with DMF
(2.0 mL) in a microwave tube and heated for 45 min to 120 °C in the
microwave. Afterward the mixture was poured onto aqueous NaHCO3
and ethyl acetate. Layers were separated and the aqueous layer was
washed again 2× with AcOEt. The combined organic layer was washed
with brine, dried over MgSO4, and concentrated in vacuum. Flash
chromatography yielded 138 mg (70%) of an off-white solid. 1H NMR
(600 MHz, DMSO-d6) δ 8.68 (s, 1H), 8.55 (s, 1H), 8.35 (d, J = 8.68 Hz,
1H), 8.26 (d, J = 8.48 Hz, 1H), 7.91 (d, J = 8.88 Hz, 1H), 7.25−7.34 (m,
4H), 7.14−7.21 (m, 1H), 6.98 (d, J = 8.88 Hz, 1H), 4.80−4.87 (m, 1H),
4.63 (s, 2H), 4.34 (d, J = 12.92 Hz, 1H), 3.97 (d, J = 12.31 Hz, 1H), 3.81
(d, J = 12.72 Hz, 1H), 3.69 (t, J = 11.20 Hz, 1H), 3.35−3.43 (m, 1H),
3.09 (t, J = 11.30 Hz, 1H), 1.38 (d, J = 6.46 Hz, 3H). LC/MS: t = 1.940,
100%, 446.2. HRMS m/z 446.2097 [(M + H)+ calcd for C27H23N7+,
446.209 32].
Cloning, Expression, and Purification of Human S1PL. The
coding region for amino acid residues 62−568 of human S1PL was
cloned into a transfer vector (pFastBacDual) after an N-terminal His-tag
and a PreScission cleavage site. This construct was co-transfected with
pFastBac (Invitrogen) virus DNA into Sf9 insect cells using lipofection.
I
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
tone; 2, clear hind limb weakness; 3, complete hind-limb paralysis; 4,
moribund. Histology of spinal cords was performed as previously
described in detail.8
Table 4. Crystal Data Collection and Refinement Statistics
parameter
Histological Analysis. Animal were perfused with PBS followed by
perfusion with ice-cold PFA, and tissues were embedded in paraffin.
Inflammatory infiltrates were scored using the following scoring
scheme: 0.5, one to three perivascular or subpial lesions, localized one
area; 1, scattered perivascular or subpial lesions; 2, lesion that occupies
5−10% of the anatomical location or several perivascular lesions found
in gray and white matter; 3, as above but 10−50%; 4, as above but in
>50%. For each animal 3−6 sections from the cervical spinal cord and
6−10 sections of the thoracic spinal cord were analyzed and the average
infiltration in each spinal cord region was used for analysis.
Statistics. Bar graphs in the figures represent average values
standard error of the mean. Statistical significance was calculated using
Student’s t test, two-tailed for unequal variance, and is indicated in the
graphs as follows: ∗, P < 0.05; ∗∗, P < 0.01.
(A) Data Collection
X-ray source, wavelength (Å)
space group
SLS beamline X10SA, 0.9990
P21212
unit cell dimensions (Å)
resolution (Å)
a = 127.9, b = 130.5, c = 68.2
50.0−2.4
observations, unique reflections
completeness (%)
I/σ
303450, 45426
a
99.8 (99.9)
a
16.26 (3.39)
a
Rmerge
10.4 (59.9)
(B) Refinement Statistics
resolution (Å)
25−2.4
45369
reflections
R-factor, Rfree
17.9, 20.6
6859
total no. of protein atoms
water atoms, other heteroatoms
rms bonds (Å), angles (deg)
ASSOCIATED CONTENT
■
344, 133
0.01, 1.06
Accession Codes
Coordinates have been deposited with the RCSB Protein Data
Bank (PDB code 4Q6R).
a
Numbers in parentheses refer to the highest resolution shell (2.46−
2.4 Å).
AUTHOR INFORMATION
Corresponding Author
*E-mail: sven.weiler@novartis.com. Phone: 41-61-6962379.
Notes
■
Smoothened Assay and Cellular S1PL Assays. Antagonism on
the receptor protein Smoothened was assessed in the “Gli shift, 25 nM”
competition assay as described in detail in ref 18. The effect of inhibitors
on the intra- and extracellular S1P concentration in cultures of HEK293
cells supplemented with sphingosine as metabolic precursor was
determined as described previously in detail.10,11
The authors declare no competing financial interest.
PK Study of 31 in the Rat. 31 was administered to male Sprague
Dawley rats. Intravenous dosage was done using a compound solution in
N-methylpyrrolidone/polyethylene glycol 200 (30:70). For oral dosage
the compound was given via gavage as a suspension in carboxyme-
thylcellulose/water/Tween 80 (0.5:99:0.5). 31 concentrations in whole
blood were determined by LC/MS/MS analysis following protein
precipitation with acetonitrile.
ACKNOWLEDGMENTS
■
We thank Felix Freuler and Lukas Leder for cloning of S1PL,
Debra Burdick for initial purification of the protein, Shu Li and
Ji-Hu Zhang for performing the HTS, and Marc Bigaud for
helpful discussions on pharmacodynamics. The authors thank
Joe Kelleher and Stefan Peukert for providing Smoothened data,
Christian Guenat and Emine Sager for HR-MS and NMR
analysis of the final compounds. We thank Ophelia Ardayfio,
Pascal Bernet, Gina Cavelti, Celine Cojean, Martin Edelmann,
Jelena Filipova, Lea Haeberli, Qin Hao, Catherine Huck,
Raphaela Kutil, Angela Mackay, Bernard Mathis, Sebastien
Rieffel, and Beatrice Urban for excellent technical assistance.
PK/PD Studies of 31 in the Rat. In a single-dose study, 31 was
administered to male Wistar rats at a dose of 50 mg/kg po. At various
time points, blood was collected and analyzed by LC/MS to quantify
the inhibitor and by differential hematology and flow cytometry. In a
multiple-dose study, 31 was given 4 times at 2 mg/kg po in 12 h
intervals; at 12 h after the last dose, blood was analyzed as before.
Differential Hematology and Flow Cytometry. Hematology
analysis was performed on whole blood, using an Advia 120 instrument
(Siemens, Germany). For flow cytometry analysis of whole blood,
erythrocytes were lysed by hypotonic shock, washed once in FACS wash
buffer (PBS containing 1% FCS), blocked with mouse Fc block (BD
Biosciences), and stained for 30 min at 4 °C in the dark with the
indicated combination of fluorochrome-conjugated mAbs. After being
stained, the cells were washed twice with wash buffer and resuspended
in 200 μL of buffer. Samples were analyzed using a FACSCalibur
flow cytometer and CellQuest Pro software (BD BioSciences). The
following antibodies were used (all from BD BioSciences): anti-CD3-
PerCP (clone 145-2C11), anti-CD4-FITC (clone V4), CD8-FITC
(clone 53-6.7), and CD19-PerCP (clone 1D3). Absolute cell numbers
within each subset were calculated by multiplying their fractional
representation determined by FACS by the absolute number of white
blood cells measured by hematology analysis. T cells were identified as
CD3+, and B cells were identified as CD19+ mononuclear cells.
Determination of S1P. S1P was determined by LC/MS as
described in detail before.6
ABBREVIATIONS USED
■
CNS, central nervous system; EAE, experimental autoimmune
encephalomyelitis; H&E, hematoxylin and eosin; HEPES, 4-(2-
hydroxyethyl)-1-piperazineethanesulfonic acid; LC/MS, liquid
chromatography/mass spectrometry; MS, multiple sclerosis; PK,
pharmacokinetics; PD, pharmacodynamics; PLP, pyridoxal
phosphate; S1P, sphingosine 1-phosphate; S1P1, S1P receptor
1; SEM, standard error of the mean; S1PL, S1P lyase; TCEP,
tris(2-carboxyethyl)phosphine; THI, 2-acetyl-4(5)-(1(R),2(S),
3(R),4-tetrahydroxybutyl)imidazole
REFERENCES
■
(1) Fletcher, J. M.; Lalor, S. J.; Sweeney, C. M.; Tubridy, N.; Mills, K.
H. T cells in multiple sclerosis and experimental autoimmune
encephalomyelitis. Clin. Exp. Immunol. 2010, 162, 1−11.
(2) Matloubian, M.; Lo, C. G.; Cinamon, G.; Lesneski, M. J.; Xu, Y.;
Brinkmann, V.; Allende, M. L.; Proia, R. L.; Cyster, J. G. Lymphocyte
egress from thymus and peripheral lymphoid organs is dependent on
S1P receptor 1. Nature 2004, 427, 355−360.
(3) Brinkmann, V.; Billich, A.; Baumruker, T.; Heining, P.; Schmouder,
R.; Francis, G.; Aradhye, S.; Burtin, P. Fingolimod (FTY720): discovery
and development of an oral drug to treat multiple sclerosis. Nat. Rev.
Drug Discovery 2010, 9, 883−897.
EAE Model in Rats. Female DA rats were immunized with 75 μg of
MOG29−152 (Sino Biological) emulsified in incomplete Freund’s
adjuvant supplemented with 2 mg/mL heat-inactivated Mycobacterium
tuberculosis (both Difco Laboratories) at the base of the tail. Compound
treatment was started 2 days before EAE induction. 31 was dosed as a
solution in Capmul (Abitec) twice daily at 2 mg/kg, starting 2 days
before EAE induction. Animals were scored for neurological signs
according the following scale: 0, no symptoms; 1, complete loss of tail
J
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Featured Article
(4) Schwab, S. R.; Pereira, J. P.; Matloubian, M.; Xu, Y.; Huang, Y.;
Cyster, J. G. Lymphocyte sequestration through S1P lyase inhibition and
disruption of S1P gradients. Science 2005, 309, 1735−1739.
(5) Serra, M.; Saba, J. D. Sphingosine 1-phosphate lyase a key regulator
of sphingosine 1-phosphate signaling and function. Adv. Enzyme Regul.
2010, 50, 349−362.
(6) Billich, A.; Baumruker, T.; Beerli, C.; Bigaud, M.; Bruns, C.;
Calzascia, T.; Isken, A.; Kinzel, B.; Loetscher, E.; Metzler, B.; Mueller,
M.; Nuesslein-Hildesheim, B.; Kleylein-Sohn, B. Partial deficiency of
sphingosine-1-phosphate lyase confers protection in experimental
autoimmune encephalomyelitis. PLoS One 2013, 8, e59630.
(7) Gugasyan, R.; Coward, A.; O’Connor, L.; Shortman, K.; Scollay, R.
Emigration of mature T cells from the thymus is inhibited by the
imidazole-based compound 2-acetyl-4-tetrahydroxybutylimidazole. Im-
munology 1998, 93, 398−404.
(8) Bagdanoff, J. T.; Donoviel, M. S.; Nouraldeen, A.; Carlsen, M.;
Jessop, T. C.; Tarver, J.; Aleem, S.; Dong, L.; Zhang, H.; Boteju, L.;
Hazelwood, J.; Yan, J. J.; Bednarz, M.; Layek, S.; Owusu, I. B.;
Gopinathan, S.; Moran, L.; Lai, Z.; Kramer, J.; Kimball, S. D.;
Yalamanchili, P.; Heydorn, W. E.; Frazier, K. S.; Brooks, B.; Brown,
P.; Wilson, A.; Sonnenburg, W. K.; Main, A.; Carson, K. G.; Oravecz, T.;
Augeri, D. J. Inhibition of sphingosine 1-phosphate lyase for the
treatment of rheumatoid arthritis: discovery of (E)-1-(4-((1R,2S,3R)-
1,2,3,4-tetrahydroxybutyl)-1H-imidazol-2-yl)ethanone oxime
(LX2931) and (1R,2S,3R)-1-(2-(isoxazol-3-yl)-1H-imidazol-4-yl)-
butane-1,2,3,4-tetraol (LX2932). J. Med. Chem. 2010, 53, 8650−8662.
(9) Bagdanoff, J. T.; Donoviel, M. S.; Nouraldeen, A.; Tarver, J.; Fu, Q.;
Carlsen, M.; Jessop, T. C.; Zhang, H.; Hazelwood, J.; Nguyen, H.;
Baugh, S. D.; Gardyan, M.; Terranova, K. M.; Barbosa, J.; Yan, J.;
Bednarz, M.; Layek, S.; Courtney, L. F.; Taylor, J.; Digeorge-Foushee, A.
M.; Gopinathan, S.; Bruce, D.; Smith, T.; Moran, L.; O’Neill, E.; Kramer,
J.; Lai, Z.; Kimball, S. D.; Liu, Q.; Sun, W.; Yu, S.; Swaffield, J.; Wilson,
A.; Main, A.; Carson, K. G.; Oravecz, T.; Augeri, D. J. Inhibition of
sphingosine-1-phosphate lyase for the treatment of autoimmune
disorders. J. Med. Chem. 2009, 52, 3941−3953.
W.; Amaral, A.; Dorsch, M.; Kelleher, J. F., III. 1-Amino-4-
benzylphthalazines as orally bioavailable smoothened antagonists with
antitumor activity. J. Med. Chem. 2009, 52, 3954−3968.
(19) Wang, C.; Wu, H.; Katritch, V.; Han, G. W.; Huang, X.-P.; Liu, W.;
Siu, F. Y.; Roth, B. L.; Cherezov, V.; Stevens, R. C. Structure of the
human smoothened receptor bound to an antitumour agent. Nature
2013, 497, 338−343.
(20) Dutyshev, D. I.; Darii, E. L.; Fomenkova, N. P.; Pechik, I. V.;
Polyakov, K. M.; Nikonov, S. V.; Andreevaa, N. S.; Sukharevaa, B. S.
Structure of Escherichia coli glutamate decarboxylase (GAD) in complex
with glutarate at 2.05 Å resolution. Acta Crystallogr. 2005, D61, 231−
235.
(21) Bourquin, F.; Riezman, H.; Capitani, G.; Grutter, M. G. Structure
̈
and function of sphingosine-1-phosphate lyase, a key enzyme of
sphingolipid metabolism. Structure 2010, 18, 1054−1065.
(22) Vogel, P.; Donoviel, M. S.; Read, R.; Hansen, G. M.; Hazlewood,
J.; Anderson, S. J.; Sun, W.; Swaffield, J.; Oravecz, T. Incomplete
inhibition of sphingosine 1-phosphate lyase modulates immune system
function yet prevents early lethality and non-lymphoid lesions. PLoS
One 2009, 4, e4112.
(23) Choi, J. W.; Gardell, S. E.; Herr, D. R.; Rivera, R.; Lee, C. W.;
Noguchi, K.; Teo, S. T.; Yung, Y. C.; Lu, M.; Kennedy, G.; Chun, J.
FTY720 (fingolimod) efficacy in an animal model of multiple sclerosis
requires astrocyte sphingosine 1-phosphate receptor 1 (S1P1)
modulation. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 751−766.
(24) Bot, M.; Van Veldhoven, P. P.; de Jager, S. C.; Johnson, J.; Nijstad,
N.; Van Santbrink, P. J.; Westra, M. M.; Van Der Hoeven, G.; Gijbels, M.
J.; Muller-Tidow, C.; Varga, G.; Tietge, U. J.; Kuiper, J.; Van Berkel, T. J.;
̈
Nofer, J. R.; Bot, I.; Biessen, E. A. Hematopoietic sphingosine 1-
phosphate lyase deficiency decreases atherosclerotic lesion development
in LDL-receptor deficient mice. PLoS One 2013, 8, e63360.
(25) Zhao, Y.; Gorshkova, I. A.; Berdyshev, E.; He, D.; Fu, P.; Ma, W.;
Su, Y.; Usatyuk, P. V.; Pendyala, S.; Oskouian, B.; Saba, J. D.; Garcia, J.
G.; Natarajan, V. Protection of LPS-induced murine acute lung injury by
sphingosine-1-phosphate lyase suppression. Am. J. Respir. Cell Mol. Biol.
2011, 45, 426−435.
(10) Billich, A.; Beerli, C.; Bergmann, R.; Bruns, C.; Loetscher, E.
Cellular assay for the characterization of sphingosine-1-phosphate lyase
inhibitors. Anal. Biochem. 2013, 434, 247−253.
(11) Loetscher, E.; Schneider, K.; Beerli, C.; Billich, A. Assay to
measure the secretion of sphingosine-1-phosphate from cells induced by
S1P lyase inhibitors. Biochem. Biophys. Res. Commun. 2013, 433, 345−
348.
(26) Saba, J. D.; de la Garza-Rodea, A. S. S1P lyase in skeletal muscle
regeneration and satellite cell activation: exposing the hidden lyase.
Biochim. Biophys. Acta 2013, 1831, 167−175.
(27) Cavalleri, B.; Bellasio, E.; Sardi, A. Synthesis of 6-trifluoromethyl-
1,4-dihydrazinophthalazine. J. Med. Chem. 1970, 13, 148−149.
(12) Yu, X. Q.; Kramer, J.; Moran, L.; O’Neill, E.; Nouraldeen, A.;
Oravecz, T.; Wilson, A. G. Pharmacokinetic/pharmacodynamic
modelling of 2-acetyl-4(5)-tetrahydroxybutyl imidazole-induced pe-
ripheral lymphocyte sequestration through increasing lymphoid
sphingosine 1-phosphate. Xenobiotica 2010, 40, 350−356.
(13) Zu Heringdorf, D. M.; Ihlefeld, K.; Pfeilschifter, J. Pharmacology
of the sphingosine-1-phosphate signalling system. Handb. Exp.
Pharmacol. 2013, 215, 239−253.
(14) Van Veldhoven, P. P.; Mannaerts, G. P. Subcellular localization
and membrane topology of sphingosine-1-phosphate lyase in rat liver. J.
Biol. Chem. 1991, 266, 12502−12507.
(15) Van Veldhoven, P. P.; Gijsbers, S.; Mannaerts, G. P.; Vermeesch,
J. R.; Brys, V. Human sphingosine-1-phosphate lyase: cDNA cloning,
functional expression studies and mapping to chromosome 10q22(1).
Biochim. Biophys. Acta 2000, 1487, 128−134.
(16) Bandhuvula, P.; Tam, Y. Y.; Oskouian, B.; Saba, J. D. The immune
modulator FTY720 inhibits sphingosine-1-phosphate lyase activity. J.
Biol. Chem. 2005, 280, 33697−336700.
(17) Berdyshev, E. V.; Goya, J.; Gorshkova, I.; Prestwich, G. D.; Byun,
H. S.; Bittman, R.; Natarajan, V. Characterization of sphingosine-1-
phosphate lyase activity by electrospray ionization−liquid chromatog-
raphy/tandem mass spectrometry quantitation of (2E)-hexadecenal.
Anal. Biochem. 2011, 408, 12−18.
(18) Miller-Moslin, K.; Peukert, S.; Jain, R. K.; McEwan, M. A.; Karki,
R.; Llamas, L.; Yusuff, N.; He, F.; Li, Y.; Sun, Y.; Dai, M.; Perez, L.;
Michael, W.; Sheng, T.; Lei, H.; Zhang, R.; Williams, J.; Bourret, A.;
Ramamurthy, A.; Yuan, J.; Guo, R.; Matsumoto, M.; Vattay, A.; Maniara,
K
dx.doi.org/10.1021/jm500338n | J. Med. Chem. XXXX, XXX, XXX−XXX