An aminomethylpyrimidine DPP-IV inhibitor with improved properties

Article abstract of DOI:10.1016/j.bmcl.2004.04.048

A recently identified DPP-IV inhibitor (1) was found to induce phospholipidosis and to inhibit CYP3A4. A small series of less lipophilic and less amphiphilic analogues was synthesized in an effort to overcome these issues. One compound from this series was equipotent to 1, did not induce phospholipidosis and showed a reduced CYP3A4 inhibition.

Full text of DOI:10.1016/j.bmcl.2004.04.048

Bioorganic & Medicinal Chemistry Letters 14 (2004) 3575–3578
An aminomethylpyrimidine DPP-IV inhibitor with
improved properties
Jens-Uwe Peters,* Daniel Hunziker, Holger Fischer, Manfred Kansy, Silja Weber,
ꢀ
€
Stephane Kritter, Aranka Muller, Angelina Wallier, Fabienne Ricklin,
Markus Boehringer, Sonia Maria Poli, Miklos Csato and Bernd-Michael Loeffler
Pharma Division, Preclinical Research, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
Received 27 January 2004; revised 29 March 2004; accepted 8 April 2004
Abstract—A recently identified DPP-IV inhibitor (1) was found to induce phospholipidosis and to inhibit CYP3A4. A small series of
less lipophilic and less amphiphilic analogues was synthesized in an effort to overcome these issues. One compound from this series
was equipotent to 1, did not induce phospholipidosis and showed a reduced CYP3A4 inhibition.
ꢀ 2004 Elsevier Ltd. All rights reserved.
Table 1. Properties of 1
Glucagon-like peptide 1 (GLP-1) has become a promi-
nent target for the treatment of type 2 diabetes.1 GLP-1
is secreted by the gastrointestinal tract in response to
food intake, stimulates insulin secretion, and inhibits
hepatic glucose production.2 Continuous intravenous
NH2 NH2
N
N
infusion of GLP-1 nearly normalizes blood glucose
Cl
Cl
levels in type 2 diabetic patients.3 However, subcutane-
ous bolus injections of GLP-1 have proven ineffective4
because GLP-1 is rapidly metabolized by dipeptidyl
peptidase IV (DPP-IV).5;6 Inhibition of DPP-IV is an
indirect way to increase the levels of circulating GLP-1,7
and DPP-IV inhibitors have been shown to improve
glucose excursion in patients with type 2 diabetes.8
1
IC50(DPP-IV)
logD7.4
IC50 (CYP 3A4)
=
=
=
10nM
3.0
5.4µM
• phospholipidosis induction
In a search for novel DPP-IV inhibitors, we identified
aminomethylpyrimidine 1 (Table 1) as a lead structure.9
Compound 1 has many favorable properties, such as
high solubility,10 high membrane permeability,11 and
high metabolic stability in human liver microsome
preparations.12 However, 1 was also found to be an
inhibitor of cytochrome P450 3A4 (CYP3A4)13 and was
found to induce phospholipidosis in cultured fibroblasts
at all test concentrations (2.5–20 lM) in a concentra-
tion-dependent manner.14 Inhibition of CYP3A4 is a
leading cause of drug–drug interactions,15 and phos-
pholipidosis might have adverse physiological conse-
quences.16 For these reasons, we felt that these latter
issues had to be addressed to improve the quality of our
lead.
Compound 1 is rather lipophilic17 and has a pKa of 7.8.18
As a consequence, 1 should be a cationic, amphiphilic
compound under physiological conditions.19 Lipophi-
licity is a recognition criterion for CYP3A4,20 and
amphiphilicity is predictive for the phospholipidosis
potential of cationic compounds.21 Therefore, we
anticipated that less lipophilic and less amphiphilic
analogues of 1 should be less likely to show CYP3A4
inhibition and phospholipidosis induction. In an at-
tempt to obtain such improved analogues, we replaced
the lipophilic 2-phenyl group of 1 by small motifs like
Me (2, Table 2), MeO (3), and NH2 (4a). However,
these compounds had a reduced inhibitory activity at
DPP-IV.22
* Corresponding author. Tel.: +41-61-6882636; fax: +41-61-6886459;
e-mail: jens-uwe.peters@roche.com
0960-894X/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.04.048

3576
J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3575–3578
Table 2. Exploration of small substituents R
In 4b, the 2-Ph substituent of 1 is replaced by a mor-
pholino substituent. Compound 4b is twofold more ac-
tive than 4a, but is still much less active than 1. The
thiomorpholino compound 4c is structurally closely re-
lated to 4b, but is in contrast to 4b an outstandingly
active DPP-IV inhibitor. The calculated lipophilicity of
4c is, however, rather high. Formal removal of one
methylene unit led to the smaller thiazolidine analogue
4d, which is less lipophilic but also less active than 4c.
Compound 4e and its acyclic analogue 4f have an even
further decreased activity. The introduction of a hy-
droxyl function gave a highly active DPP-IV inhibitor,
4g, with a greatly reduced lipophilicity, but a borderline
DDGAM value. The high absolute DDGAM value for 4g is
an exception to the otherwise generally observed corre-
lation between DDGAM and KOW_ClogP within the
series of 4, and is due to the conformer selection pro-
cedure in CAFCA.24 The corresponding methyl ether,
4h, had the most appealing balance between activity,
lipophilicity, and amphiphilicity, and was further char-
acterized. Compound 4h is comparable to 1 with respect
to solubility, membrane permeability, microsomal sta-
bility, and inhibitory activity at DPP-IV. Compound 4h
has a much reduced measured lipophilicity (as compared
to 1), which correlates well with the predicted value.
Gratifyingly, 4h shows a sufficiently reduced CYP3A4
inhibition, and was found not to induce phospholipi-
dosis in cultured fibroblasts up to the highest test con-
centration (20 lM) (Table 4).
NH2
N
NH2
N
R
Cl
Cl
R ¼
Me
IC50 (DPP-IV) (nM)
log D7:4
2
250
79
1.4
1.6
0.4
3
4a
OMe
NH2
270
To identify similar compounds with restored activity
and low CYP3A4/phospholipidosis potential, we pre-
pared a number of N-alkylated analogues of 4a that
were predicted by computational tools to be less lipo-
philic (KOW_ClogP)23 and less amphiphilic (CAFCA)24
than compound 1 (4b–h, Table 3). More specifically, we
aimed for compounds with a calculated free energy of
amphiphilicity (DDGAM) greater than )6 kJ/mol, be-
cause it was found previously that such compounds have
a low probability of being positive in the fibroblast
phospholipidosis assay, whereas DDGAM values below
)6 kJ/mol were found to indicate compounds with a
high risk of phospholipidosis.21
The in silico results and the measured inhibitory activ-
ities of 4b–h are documented in Table 3.
Aminomethylpyrimidines 1–4 were obtained according
to Scheme 1: benzylamidine (5, R ¼ Ph), O-methyliso-
urea (5, R ¼ OMe), or urea (5, R ¼ NH2) were reacted
with 2,4-dichlorobenzylidenemalononitrile (6) under
basic conditions with a subsequent oxidative step to give
the corresponding 5-cyanopyrimidines 7.25 Reduction of
the nitrile functionality furnished 1–4a. S-Meth-
ylisothiourea (5, R ¼ SMe) and 6 were converted to
pyrimidine 7-SMe. Oxidation provided sulfone 7-
SO2Me. Nucleophilic displacements with a variety of
amines R2NH gave 7b–h,26 which were reduced to
aminomethylpyrimidines 4b–h. For a detailed procedure
for the preparation of 4h, see Ref. 27.
Table 3. Exploration of amine substituents R
NH2 NH2
N
R
N
Cl
Cl
a
R ¼
KOW_ClogP DDGAM IC50 (DPP-IV)
(kJ/mol) (nM)
1
Ph
2.6
1.4
)6.6
10
N
N
4b
)5.3
140
In summary, we addressed the phospholipidosis liability
and the CYP3A4 inhibition in a series of aminometh-
ylpyrimidine DPP-IV inhibitors. We speculated that
compounds with lowered lipophilicity and amphiphilic-
O
4c
4d
2.3
1.8
)5.7
)5.6
0.2
S
S
N
11
24
55
Table 4. Properties of 4h
N
NH2 NH2
4e
4f
2.2
2.3
)5.7
)5.8
N
O
N
N
N
Cl
4h
Cl
HO
O
N
4g
4h
0.9
1.6
)6.0
)5.6
0.5
IC50 (DPP-IV)
=
=
=
9nM
1.6
30µM
N
9
logD7.4
IC50 (CYP 3A4)
a Calculated free energy of amphiphilicity as a measure of phospho-
lipidosis potential.
• no phospholipidosis

J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3575–3578
3577
NH2
NH2 NH2
CN
Ar
CN
Ar
CN
N
N
a
b
NH
+
R
N
R
N
Ar
R
NH2
5
6
7
1-3, 4a
NH2 NH2
NH2
NH2
NH2
N
CN
Ar
CN
Ar
CN
Ar
N
N
N
N
d
c
b
O
R
R
S
N
S
N
R
N
N
R
N
Ar
O
7-SMe
7-SO2Me
7b-h
4b-h
Scheme 1. Ar ¼ 2,4-dichlorophenyl. Reagents and conditions: (a) K2CO3, MeOH, 50 ꢁC, 3 h, then, after evaporation of solvent, KMnO4, acetone, rt,
3 h (e.g., 7-Ph 52%; 7-SMe 88%); (b) LiAlH4, THF, 40 ꢁC, 3 h (e.g., 1 40%; 4h 18%); (c) MCPBA, CH2Cl2, 15 min 0 ꢁC fi rt (quant.); (d) R2NH,
dioxane, 2 h, rt (e.g., 7h 84%).
13. Measurements were performed according to: Stresser, D.
M.; Turner, S. D.; Blanchard, A. P.; Miller, V. P.; Crespi,
C. L. Drug Metab. Dispos. 2002, 30, 845 (BFC was used as
a substrate).
ity might be superior in terms of CYP3A4 inhibition and
phospholipidosis potential. Predictive computational
tools guided us in the synthesis of a series of such
compounds. From this series we identified a low
nanomolar DPP-IV inhibitor, 4h, that did not induce
phospholipidosis and showed a reduced CYP3A4 inhi-
bition. Additionally, sub-nanomolar inhibitors of DPP-
IV (4c,g, Table 3) were found in the course of our work.
€
14. (a) Handrock, K.; Lullmann-Rauch, R.; Vogt, R. D.
Toxicology 1993, 85, 199; (b) Lullmann-Rauch, R.; Pods,
€
R.; von Witzendorff, B. Toxicology 1996, 110, 27; (c)
Mason, R. J.; Walker, S. R.; Shields, B. A.; Henson, J. E.;
Williams, M. C. Am. Rev. Respir. Dis. 1985, 131, 786.
15. Dresser, G. K.; Spence, J. D.; Bailey, D. G. Clin.
Pharmacokinet. 2000, 38, 41.
16. Reasor, M. J.; Kacew, S. Exp. Biol. Med. 2001, 226, 825.
17. Measurements were performed according to an in-house
developed miniaturized method to measure the shake flask
octanol–water partition coefficient at pH 7.4 (log D).
18. Measurements were performed according to: Allen, R. I.;
Box, K. J.; Comer, J.; Peake, C.; Tam, K. Y. J. Pharm.
Biomed. Anal. 1998, 17, 699.
19. An amphiphilic molecule comprises a hydrophobic and a
hydrophilic part. Such molecules have an inherent ten-
dency to orient themselves in a suitable environment, for
example, a phospholipid bilayer.
20. (a) Riley, R. J.; Parker, A. J.; Trigg, S.; Manners, C. N.
Pharm. Res. 2001, 18, 652; (b) Smith, A. D.; van de
Waterbeemd, H.; Walker, D. K. In Pharmacokinetics and
Metabolism in Drug Design; Mannhold, R., Kubinyi, H.,
Timmermann, H., Eds.; Wiley-VCH: Weinheim, 2001.
21. Fischer, H.; Kansy, M.; Potthast, M.; Csato, M. In
Proceedings of the 13th European Symposium on Quanti-
tative Structure–Activity Relationships, Duesseldorf, Ger-
many, Aug 27–Sept 1, 2000; Prous Science: Barcelona,
Spain, 2001.
References and notes
1. For reviews, see: (a) Drucker, D. J. Diabetes Care 2003,
26, 2929; (b) Augustyns, K.; Van der Veken, P.; Senten,
K.; Haerners, A. Exp. Opin. Ther. Patents 2003, 13, 499.
2. For a review, see: Holst, J. J.; Orskov, C. Scand. J. Clin.
Lab. Invest. 2001, 61(Suppl. 234), 75.
3. Rachman, J.; Barrow, B. A.; Levy, J. C.; Turner, R. C.
Diabetologica 1997, 40, 205.
4. Nauck, M. A.; Wollschlager, D.; Werner, J.; Holst, J. J.;
Orskov, C.; Creutzfeldt, W. Diabetologica 1996, 39, 1546.
5. Knudsen, L. B.; Pridal, L. Eur. J. Pharmacol. 1996, 318,
429.
6. For a review, see: Rosenblum, J. S.; Kozarich, J. W. Curr.
Opin. Chem. Biol. 2003, 7, 496.
7. Pauly, R. P.; Demuth, H. U.; Rosche, F.; Schmidt, J.;
White, H. A.; Lynn, F.; McIntosh, C. H.; Pederson, R. A.
Metab.: Clin. Exp. 1999, 48, 385.
8. For a review, see: Drucker, D. J. Exp. Opin. Invest. Drugs
2003, 12, 87.
9. Peters, J.-U.; Weber, S.; Kritter, S.; Weiss, P.; Wallier, A.;
Boehringer, M.; Hennig, M.; Kuhn, B.; Loeffler, B.-M.
Bioorg. Med. Chem. Lett. 2004, 14, 1491.
10. Measurements were performed according to an in-house
developed method to assess solubility from a 10 mM
DMSO stock solution. This method is similar to the
classical thermodynamic shake-flask solubility with the
only difference that DMSO is removed in the beginning by
an additional lyophilization step. Therefore this assay is
called lyophilizated solubility assay (LYSA).
11. Measurements were performed according to: Kansy, M.;
Senner, F.; Gubernator, K. J. Med. Chem. 1998, 41, 1007.
12. Measurements were performed according to: Obach, R. S.;
Baxter, J. G.; Liston, T. E.; Silber, B. M.; Jones, B. C.;
MacIntyre, F.; Rance, D. J.; Wastall, P. J. Pharm. Exp.
Ther. 1997, 283, 46.
22. DPP-IV inhibitors were measured for their ability to
inhibit DPP-IV mediated cleavage of Ala-Pro-7-amido-4-
trifluoromethylcoumarin in a fluorogenic assay. All com-
pounds were measured in triplicate at 5–7 concentrations
in the range of 100 lM to 100 pM. IC50 values were
calculated with a nonlinear best fit regression model. All
assays were calibrated with NVP-DPP728 as internal
standard inhibitor. NVP-DPP728 under the conditions of
the assay showed an IC50 of 15 4 nM (M SD, n ¼ 12) at
50 lM substrate concentration and a Ki of 11 3 nM
determined at substrate concentration range of 10–
600 lM. IC50 values of unknown compounds were
accepted when the IC50 (x) measured for NVP-DPP728
in the assay was 11 < x < 19 nM.
23. Meylan, W. M.; Howard, P. H. J. Pharm. Sci. 1995, 84,
83.

3578
J.-U. Peters et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3575–3578
24. Fischer, H.; Kansy, M.; Bur, D. Chimia 2000, 54, 640.
25. Abd-Elfattah, A. M; Hussain, S. M.; El-Reedy, A. M.;
Yousif, N. M. Tetrahedron 1983, 39, 3197.
26. Masquelin, T.; Sprenger, D.; Baer, R.; Gerber, F.;
Mercadal, Y. Helv. Chim. Acta 1998, 81, 646.
(CDCl3): d 1.55 (2H, br s), 3.33 (3H, s), 7.45 (2H, s), 7.59
(1H, s); (c) 7h: A mixture of N-2-(methoxyethyl)methyl-
amine (62 mg, 0.699 mmol) was added to 7-SO2Me
(200 mg, 0.583 mmol) in dioxane (4 mL). The mixture
was stirred for 2 h at rt during which time it turned into a
clear brown solution. The solvent was evaporated and the
residue was taken up in EtOAc. The mixture was washed
(1 N HCl, brine) and dried (Na2SO4). After evaporation,
7h (173 mg, 84%) was obtained by column chromatogra-
phy (silica gel, AcOEt/hexane ¼ 1:2). 1H NMR (DMSO): d
3.13 (3H, s), 3.21 (3H, s), 3.48–3.50 (2H, m), 3.71–3.76
(2H, m), 7.35 (2H, br s), 7.49–7.51 (1H, m), 7.55–7.57 (1H,
m), 7.79 (1H, s); (d) 4h: At a temperature of 0 ꢁC, a
solution of 7h (19 mg, 0.054 mmol) in THF (2 mL) was
added to a suspension of LiAlH4 (20 mg, 0.53 mmol) in
THF (1 mL). The mixture was stirred for 2 h at rt and then
quenched with H2O (0.5 mL). After filtration and evapo-
ration of the solvent, 4h (3.4 mg, 18%) was isolated from
the reaction mixture by preparative HPLC [Pro 18
column, solvent gradient 5–95% CH3CN in 0.1% TFA(aq)
27. Preparation of 4h: (a) 7-SMe: S-Methylisothiourea sulfate
(5-SMe, 4.49 g, 16.1 mmol) 2,4-dichlorobenzylidenema-
lononitrile (6, 6.00 g, 27.0 mmol), and K2CO3 (6.50 g,
47.0 mmol) were suspended in MeOH (50 mL) and heated
to reflux for 90 min. The solvent was then evaporated and
the residue was distributed in EtOAc/H2O. The organic
layer was separated and dried (Na2SO4). After evapora-
tion of the solvent, the residue was taken up in acetone
and treated with KMnO4 at rt overnight. After filtration
(dicalite) and evaporation, 7-SMe (7.40 g, 88%) was
isolated by column chromatography (silica gel, AcOEt/
1
hexane ¼ 1:2). H NMR (DMSO-d6): d 2.46 (3H, s), 7.57–
7.62 (2H, m), 7.83 (1H, s), 7.80–8.30 (2H, br s); (b) 7-
SO2Me: At 0 ꢁC, MCPBA (865 mg, 5.01 mmol) was added
to a suspension of 7-SMe (1.30 g, 4.18 mmol) in CH2Cl2
(40 mL). The mixture was stirred for 1 h at rt, during
which time the product partly crystallized. CH2Cl2 was
added, the mixture was washed with satd NaHCO3 and
H2O, and dried (Na2SO4). The product, 7-SO2Me (1.49 g,
100%), was used without further purification.1H NMR
1
over 6.0 min, k ¼ 230 nm, flow rate 40 mL/min]. H NMR
(CDCl3): d 1.61 (2H, br s), 3.01 (3H, s), 3.23 (3H, s), 3.31
(2H, s), 3.45–3.48 (2H, m), 3.61–3.65 (2H, m), 6.70
(2H, br s), 7.38–7.40 (1H, m), 7.46–7.48 (1H, m), 7.67
(1H, s).