Quinazolin-4-piperidin-4-methyl sulfamide PC-1 inhibitors: Alleviating hERG interactions through structure based design

Article abstract of DOI:10.1016/j.bmcl.2009.04.006

PC-1 (NPP-1) inhibitors may be useful as therapeutics for the treatment of CDDP (calcium pyrophosphate dehydrate) deposition disease and osteoarthritis. We have identified a series of potent quinazolin-4-piperidin-4-ethyl sulfamide PC-1 inhibitors. The se

Full text of DOI:10.1016/j.bmcl.2009.04.006

Bioorganic & Medicinal Chemistry Letters 19 (2009) 3339–3343
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Quinazolin-4-piperidin-4-methyl sulfamide PC-1 inhibitors: Alleviating
hERG interactions through structure based design
à
§
*
Snahel D. Patel , Wendy M. Habeski, Alan C. Cheng , Elisa de la Cruz, Christine Loh , Natasha M. Kablaoui
Pfizer Global Research and Development, Cambridge Laboratories, Cambridge, MA 02139, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
PC-1 (NPP-1) inhibitors may be useful as therapeutics for the treatment of CDDP (calcium pyrophosphate
dehydrate) deposition disease and osteoarthritis. We have identified a series of potent quinazolin-4-
piperidin-4-ethyl sulfamide PC-1 inhibitors. The series, however, suffers from high affinity binding to
hERG potassium channels, which can cause drug-induced QT prolongation. We used a hERG homology
model to identify potential key interactions between our compounds and hERG, and the information
gained was used to design and prepare a series of quinazolin-4-piperidin-4-methyl sulfamides that retain
PC-1 activity but lack binding affinity for hERG.
Received 12 February 2009
Revised 31 March 2009
Accepted 3 April 2009
Available online 9 April 2009
Keywords:
PC-1 inhibitor
hERG
Ó 2009 Elsevier Ltd. All rights reserved.
Homology model
PC-1 (Plasma Cell Membrane Protein-1), also known as NPP-1,
belongs to a family of enzymes with nucleoside triphosphate pyro-
phophohydrolase (NTPPPH) activity; PC-1 catalyzes the hydrolysis
of ATP to ePPi (extracellular inorganic pyrophosphate) and AMP.1
PC-1 is a class II transmembrane glycoprotein that is highly ex-
pressed in articular chondrocytes, and is the major source of the
elevated PPi levels in chondrocytes and fibroblasts of patients with
familial CPPD (calcium pyrophosphate dehydrate) deposition dis-
ease.2,3 CPPD crystals may be a hallmark of the pathology of osteo-
arthritis.4 This suggests that inhibitors of PC-1 may block PPi
production by chondrocytes which may result in the reduction of
CPPD crystals, thereby having utility in the treatment of chondro-
calcinosis and osteoarthritis. Further, PC-1 interferes with insulin
receptor signaling, and epidemiological data suggests a link be-
tween mutations in PC-1 and insulin resistance.5–8 PC-1 inhibitors
may therefore also have use in treating diabetic patients. Since
both osteoarthritis and diabetes require chronic treatment, a PC-
1 inhibitor drug requires an excellent safety profile. Our PC-1
inhibitor lead compound 1 is a potent ligand for the human
‘ether-a-go-go’ (hERG) channel, a strong cardiovascular safety lia-
bility, so understanding and alleviating this liability is a major goal
in the project.
tion in man; in particular they can prolong the QT interval in the
electrocardiogram.10 This QT prolongation can, in rare cases, lead
to drug-induced ventricular fibrillation and sudden death, Torsades
de Pointes (TdP).11 Affinity for the hERG potassium channel is com-
monly tested using a whole cell voltage clamp assay, which is a low
throughput and technically challenging technique.10 We instead
employed a predictive, high-throughput competitive binding assay
that measures the ability of a test compound to displace [3H]-dof-
etilide from the hERG channel stably expressed in HEK-293 cells.12
Dofetilide is an antiarrhythmic agent, the structure of which is
shown in Figure 1. While a binding assay such as this does not pro-
vide functional information on drug effects, the tritiated dofetilide
binding assay is often used in early drug discovery projects as a
high throughput surrogate for the patch clamp assay.
Our lead compound 1 is a potent inhibitor of PC-1 (36 nM, Table
1), as measured by an in vitro Kinase-Glo assay (Promega) moni-
toring the consumption of ATP by the PC-1 enzyme. However,
the compound is also a potent ligand for the hERG channel, mea-
suring 601 nM in the tritiated dofetilide binding assay. Therefore
we sought out the structural requirements for hERG binding, hop-
ing to find divergent SAR between hERG binding and PC-1
inhibition.
Compounds with a high affinity for the hERG potassium chan-
nel, which plays a key role in regulating cardiac rhythm,9 have
the propensity to cause undesirable effects on cardiac repolariza-
Me
S
O
O
O
N
O
O
* Corresponding author. Tel.: +1 617 551 3473; fax: +1 860 686 8090.
E-mail address: Natasha.m.kablaoui@pfizer.com (N.M. Kablaoui).
Present address: Genentech Inc., South San Francisco, CA 94080, USA.
S
Me
Me
N
H
à
Present address: Amgen Inc., Cambridge, MA 02139, USA.
Present address: Sirtris Pharmaceuticals, Cambridge, MA 02139 USA.
§
Figure 1. Structure of dofetilide.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.04.006

3340
S. D. Patel et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3339–3343
Table 1
channels, respectively. The homology model employed contains
Modifications to the quinazoline region of compound 1
four subunits spanning Gly546 to Ile662, which comprise the
highly symmetrical homo-tetrameric pore domain. Docking of
compound 1 into both conformations of this model (Fig. 2) reveals
two areas of important interactions: stacking of the quinazoline
with the Phe656 residues (Fig. 3), and extensive hydrogen bonding
of the sulfamide to Ser624 residues in a tight pocket (Fig. 4). The
nitrogen in the 3-position of the quinazoline may also be in a posi-
N
O
S
HN
O
tion to participate in cation–p interactions with a Tyr652 residue
NH2
(Fig. 3); this nitrogen has a predicted pKa of 9.7 and would there-
fore be the first site of protonation, based on calculations using
ACD/pKa DB software (v. 8.07, Advanced Chemistry Development,
Inc., Toronto, Canada). All of these interactions are in keeping with
the known key residues necessary for the binding of many potent
hERG channel blockers.19,22
Structure
PC-1 (
lM)
Dofetilide (
0.601
lM)
MeO
N
N
1
2
3
4
5
6
0.0362
0.187
>10
N
N
MeO
MeO
Modifications to the quinazoline region of compound 1 are out-
lined in Table 1. As suggested by observation of the docking model,
small modifications to the quinazoline ring do not have a large ef-
fect on the compounds’ ability to displace tritiated dofetilide, as
illustrated by 7-OMe quinazoline compound 2, which is equipotent
in PC-1 and dofetilide assays. However, an electron withdrawing
group on the quinazoline, such as 7-Cl or 6-Cl quinazoline deriva-
tives 3 and 4, results in a reduction in dofetilide potency; unfortu-
nately it also significantly interferes with PC-1 potency as well.
Substitution of the quinazoline in the 2-position enhances activity
in the dofetilide assay, as evidenced by 2-Et compound 5, which
has dofetilide activity of 17 nM. Compound 6, also substituted in
the 2-position of the quinazoline, also displays submicromolar
activity in the dofetilide assay. Both 2-substituted examples main-
tain PC-1 potency. Two examples are shown that break up the qui-
nazoline, and in both cases the dofetilide binding efficiency is
0.1202
Cl
N
N
3.4
N
5.98
5.42
N
Cl
reduced: 6-methylimidizole substituted pyrimidine
detectable dofetilide binding activity, but retains submicromolar
PC-1 activity, and the 2-methylimidizole substituted pyrimidine
7 has no
MeO
MeO
N
Et
0.214
0.0353
0.0174
0.276
N
N
8 shows a dofetilide binding potency of 6.8
lM, but has PC-1 inhib-
itory activity of only 2.5 M. While replacement of the quinazoline
l
N
with a substituted pyrimidine initially seemed promising as a path
forward for separating the SAR between dofetilide binding and
PC-1 activity, we were unable to identify a compound with PC-1
activity better than the 600 nM potency of compound 7 using this
strategy. Compounds were synthesized according to Scheme 1.
A second strategy suggested by the model is to shorten the dis-
tance between the quinazoline, which is locked into place through
MeO
MeO
N
Me
N
N
N
the
p-stacking with Phe656 residues and a cation–p interaction
7
8
0.6
>10
with a Tyr652 residue, and the sulfamide, which is also in a tight
pocket with extensive hydrogen bonding interactions to Ser624
residues. Figure 5 shows the docking of lead compound 1 to the
model compared to the docking of compound 9, in which the linker
from the piperidine to the sulfamide is shortened from two car-
bons to one. The model suggests that compound 9 does not form
the network of hydrogen bonds to Ser624 side chains that occurs
in 1 and would therefore bind with lower affinity.
N
N
N
N
2.45
6.867
Me
N
Experimentally, when the linker from the piperidine to the sulf-
amide is shortened from two carbons to one, we find that in most
cases tested, compounds have no detectable dofetilide binding
activity. More importantly, we also find that PC-1 activity is pre-
served in the quinazoline analogs. Table 2 shows four comparisons
of quinazoline (or pyrimidine) head groups with differing linker
lengths, along with PC-1 enzymatic data, dofetilide binding data,
and the ratio of the two assay readouts. The dofetilide/PC-1 ratio
can be interpreted as a safety window.
Analogs with the 6,7-dimethoxyquinazoline head group are
represented by compounds 1, 9, and 10, which have an ethylene,
methylene, and no spacer, respectively. PC-1 inhibitory activity is
maintained for compounds 1 and 9, but there is no detectable
PC-1 inhibition noted for compound 10. However, compounds 9
There have been several ligand-based quantitative structure–
activity relationship (QSAR) and pharmacophore models built
using structurally diverse compounds known to bind to and block
hERG channels.13–16 Similar models built with large amounts of di-
verse internal compound data were poor predictors of hERG bind-
ing liability for this series. Therefore we explored the use of a
structure-based hERG homology model for guidance in determin-
ing the most important functional group interactions facilitating
the hERG binding affinity of our lead compound. Similarly to sev-
eral reported hERG homology models,15–19 the models for both
the open and closed conformations of the hERG channel were gen-
erated using crystal structures of the MthK20 and KcsA21 potassium

S. D. Patel et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3339–3343
3341
Figure 2. Compound 1 (purple) modeled binding mode in homology models of the open (a) and closed (b) conformations of the hERG channel, with key residues shown in
stick format. One monomer of the tetrameric channel is not shown to improve clarity.
Figure 3. Detail of compound 1 modeled into the open conformation of the hERG
channel; the quinazoline t-stacks with at least one of the Phe656 residues of the
Figure 4. Detail of compound 1 modeled into the open conformation of the hERG
channel; the sulfamide is in a tight binding pocket with the Ser624 residues from
tetrameric channel, as indicated by the orange bar. Quinazoline N3 may also
the four symmetrical subunits. Potential hydrogen bonds are indicated with black
participate in a cation–pi interaction with Tyr652.
dashed lines.
and 10 are both >10 lM in the dofetilide binding assay, as sug-
gested by the dockings shown in Figure 5. Thus compound 9, with
the methylene spacer, has the optimal ratio of PC-1 inhibition to
dofetilide binding, at >160-fold. This trend is mirrored in the 7-
OMe headgroup series, compounds 2 and 11, in that the methylene
spacer 11 is equipotent in PC-1 activity to the ethylene spacer com-
pound 2, but has no detectable dofetilide binding activity, resulting
in a much greater safety window (>60-fold for 11 vs 1-fold for 2).
The 2-substituted quinazoline example is slightly different, in that
the methylene spacer compound 12 still has dofetilide binding
activity, although an order of magnitude less potent than ethylene
R1
R2
N
N
H
Bn
N
N
N
n
d
a, b, c
.HCl
n
HN
S
n
H2N
O
NH2
O
HN
S
O
NH2
R1
N
O
S
spacer compound 6 (3.3
lM and 0.28
l
M, respectively); however
Me
O
O
N
N
the trend is in the same direction and the safety window is en-
hanced by having the shorter linker (52-fold for 12 vs 9-fold for
6). Interestingly, for the non-quinazoline example shown, both
the ethyl and methyl spacer compounds (7 and 13) lack detectable
dofetilide binding activity, but only ethylene spacer compound 7
has PC-1 inhibitory activity.
N
R2
Me
BocHN
Cl-
Cl
A
B
Scheme 1. Synthetic route to analogs in Tables 1 and 2. Reagents and conditions:
(a) DIPEA, DCM, A,26 16 h (79–91%); (b) 15%Pd/C, EtOH; (c) HCl (g), DCM (89–97%,
two steps); (d) TEA, DCE, 50 °C, B, 16 h (68–97%).

3342
S. D. Patel et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3339–3343
Figure 5. Modeling suggests that 9 does not form the network of hydrogen bonds to Ser624 side chain and backbone that occurs with 1.
Table 2
Comparison of compounds with differing linker lengths between the sulfamide and the piperidine
Me
N
N
MeO
MeO
N
N
MeO
N
N
MeO
MeO
N
N
N
N
N
N
n
N
n
N
n
N
n
HN
S
HN
S
O
O
HN
S
O
O
HN
S
H2N
O
O
H2N
O
H2N
H2N
O
n = 2
1
2
6
7
PC-1 (
l
M)
0.036
0.601
17
0.187
0.12
1
0.035
0.276
8
0.6
>10
>16
dof (
lM)
Ratio dof/PC-1
n = 1
9
11
12
13
PC-1 (
l
M)
0.061
>10
>160
0.166
>10
>60
0.063
3.296
52
>10
>10
—
dof ( M)
l
Ratio dof/PC-1
n = 0
10
PC-1 (
l
M)
>10
>10
NA
NA
NA
NA
NA
NA
dof ( M)
l
In conclusion, the hERG homology model enabled the PC-1
program to identify the cause of the unwanted hERG activity
in the lead compound and to rapidly design potent PC-1 inhibi-
tors with high selectivity over binding to the hERG ion channel.
It is interesting to note that the usual pharmacophore for hERG
References and notes
1. Goding, J. W.; Grobben, B.; Slegers, H. Biochim. Biophys. Acta 2003, 1638, 1.
2. Terkeltaub, R.; Rosenbach, M.; Fong, F.; Goding, J. Arthritis Rheum. 1994, 37, 934.
3. Johnson, K.; Pritzker, K.; Goding, J.; Terkeltaub, R. J. Rheumatol. 2001, 28, 2681.
4. Misra, R. Cell. Mol. Life Sci. 2000, 57, 421.
5. Maddux, B.; Sbraccia, P.; Kumakura, S.; Sasson, S.; Youngren, J.; Fisher, A.;
Spencer, S.; Grupe, A.; Henzel, W.; Stewart, T.; Reaven, G.; Goldfine, I. Nature
1995, 373, 448.
blockage is thought to be through
p-stacking interactions with
the Phe656 and cation– interactions with Tyr652; however, in
p
this series, we found that the polar group interaction with the
Ser624, often thought to be of lesser importance,13 seems to
be the determining factor in eliminating hERG binding. Consis-
tent with our observations, published mutagenesis work sug-
gests that Ser624 is a significant contributor to hERG inhibitor
binding of droperidol, risperidone, haloperidol,23 cisapride,24
and dofetilide.25 Understanding the hERG liability and designing
away from it at an early stage in the project is imperative, and
utilizing structure-based drug design through a hERG homology
model is a potent approach.
6. Dong, H.; Maddux, B.; Altomonte, J.; Meseck, M.; Accili, D.; Terkeltaub, R.;
Johnson, K.; Roungren, J.; Goldfine, I. Diabetes 2005, 54, 367.
7. Kumadura, S.; Maddux, B.; Sung, C. J. Cell. Biochem. 1998, 68, 366.
8. Meyre, D.; Bouatia-Naji, N.; Tounian, A.; Samson, C.; Lecoeur, C.; Vatin, V.;
Ghoussaini, M.; Wachter, C.; Hercberg, S.; Charpentier, G.; Patsch, W.; Pattou,
F.; Charles, M.; Tounian, P.; Clement, K.; Jouret, B.; Weill, J.; Maddux, B.;
Goldfine, I.; Walley, A.; Boutin, P.; Dina, C.; Froguel, P. Nat. Genet. 2005, 37, 863.
9. Keating, M.; Sanguinetti, M. Cell 2001, 104, 569.
10. Fermini, B.; Fossa, A. Nat. Rev. Drug Disc. 2003, 2, 439.
11. De-Ponti, F.; Poluzzi, E.; Montanaro, N. Eur. J. Clin. Pharmacol. 2000, 56, 1.
12. Greengrass, P.; Stewart, M.; Wood, C.; Pfizer, I.; Eds.; Patent Report WO
03021271 A2. 2003, p. 33.
13. Aronov, A. Curr. Opin. Drug Disc. Dev. 2008, 11, 128.

S. D. Patel et al. / Bioorg. Med. Chem. Lett. 19 (2009) 3339–3343
3343
14. Cavalli, A.; Poluzzi, E.; Ponti, F. D.; Racanatini, M. J. Med. Chem. 2002, 45, 3844.
15. Farid, R.; Day, T.; Friesner, R. A.; Pearlstein, R. A. Bioorg. Med. Chem. Lett. 2006,
14, 3160.
16. Pearlstein, R. A.; Vaz, R. J.; Kang, J.; Chen, X. L.; Preobrazhenskaya, M.;
Shchekotikhin, A. E.; Korolev, A. M.; Lysenkova, L. N.; Miroshnikova, O. V.;
Hendrix, J.; Rampe, D. Bioorg. Med. Chem. Lett. 2003, 13, 1829.
17. Rajamani, R.; Tounge, B. A.; Li, J.; Reynolds, C. H. Bioorg. Med. Chem. Lett. 2005,
15, 1737.
21. Doyle, D.; Cabral, J.; Pfuetzner, R.; Kuo, A.; Gulbis, J.; Cohen, S.; Chait, B.;
MacKinnon, R. Science 1998, 280, 69.
22. Mitcheson, J. S.; Perry, M. D. Curr. Opin. Drug Disc. Dev. 2003, 6,
667.
23. Nowak, M. W.; Zacharias, N. M.; Kulkami, A. A.; Nicholas, J. B.; Sahba, S. B.;
Lally, B. S.; Lesso, H. P. S.; Reyes, J.; Mackey, E. D.; Shiva, N. W.; Bennett, P. B.
In National Meeting of the Americal Chemical Society; San Diego, CA,
2005.
18. Osterberg, F.; Aqvist, J. FEBS Lett. 2005, 579, 2939.
19. Mitcheson, J. S.; Chen, J.; Lin, M.; Culberson, C.; Sanguinetti, M. C. PNAS 2000,
97, 12329.
24. Imai, Y. N.; Ryu, S.; Oiki, S. J. Med. Chem. 2009, 52, 1630.
25. Kamiya, K.; Niwa, R.; Mitcheson, S. S.; Sanguinetti, M. C. Mol. Pharmacol. 2006,
69, 1709.
20. Jiang, U.; Lee, A.; Chen, J.; Cadene, M.; Chalt, B.; MacKinnon, R. Nature 2002,
417, 515.
26. Winum, J.; Toupet, L.; Barragan, V.; Dewynter, G.; Montero, J. Org. Lett. 2001, 3,
2241.