Bioorganic & Medicinal Chemistry Letters 21 (2011) 3738–3742
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl
Triazoloquinazolines as a novel class of phosphodiesterase 10A (PDE10A) inhibitors Jan Kehler a,⇑, Andreas Ritzen a, Morten Langgård a, Sebastian Leth Petersen a, Mohamed M. Farah b, Christoffer Bundgaard a, Claus Tornby Christoffersen c, Jacob Nielsen d, John Paul Kilburn a a
H. Lundbeck A/S, Department of Discovery Chemistry & DMPK, 9 Ottiliavej, DK-2500 Valby, Denmark Department of Chemistry, Faculty of Science, University of Malaya, 50603 Kuala Lumpur, Malaysia c H. Lundbeck A/S, Department of Molecular Pharmacology, 9 Ottiliavej, DK-2500 Valby, Denmark d H. Lundbeck A/S, Department of Synaptic Transmission 2, 9 Ottiliavej, DK-2500 Valby, Denmark b
a r t i c l e
i n f o
Article history: Received 24 February 2011 Revised 12 April 2011 Accepted 15 April 2011 Available online 30 April 2011 Keywords: PDE10A Phosphodiesterase Psychosis Schizophrenia Structure Triazoloquinazoline
a b s t r a c t Novel triazoloquinazolines have been found as phosphodiesterase 10A (PDE10A) inhibitors. Structure– activity studies improved the initial micromolar potency which was found in the lead compound by a 100-fold identifying 5-(1H-benzoimidazol-2-ylmethylsulfanyl)-2-methyl-[1,2,4]triazolo[1,5-c]quinazoline, 42 (PDE10A IC50 = 12 nM) as the most potent compound from the series. Two X-ray structures revealed novel binding modes to the catalytic site of the PDE10A enzyme. Ó 2011 Elsevier Ltd. All rights reserved.
The cyclic nucleotides cyclic adenosine monophosphate (cAMP) and cyclic guanosine monophosphate (cGMP) are ubiquitous second messengers responsible for mediating the biological response of a variety of extracellular signals, including hormones, light, and neurotransmitters and they influence a vast array of processes, for example, muscle contraction, learning, differentiation, and apoptosis as well as metabolic processes.1 In neurons, this includes the activation of cAMP- and cGMP-dependent kinases and subsequent phosphorylation of proteins involved in acute regulation of synaptic transmission as well as in gene expression and neuronal differentiation and survival.2,3 The phosphodiesterases (PDEs) are a family of bimetallic hydrolase enzymes that inactivate cAMP/cGMP by catalytic hydrolysis of the 30 -ester bond, forming the inactive 50 monophosphate thereby terminating the signaling.1,3 PDE10A is a relatively newly identified phosphodiesterase with a distinct brain localisation of the protein with predominant expression in the medium spiny neurons of the striatal complex.4–6 Based on this unique localisation, research has focused extensively on using PDE10A modulators as a novel therapeutic approach for dysfunction in the basal ganglia circuit including Parkinsons’ disease, Huntingtons disease, schizophrenia, addiction and obsessive
⇑ Corresponding author. E-mail address:
[email protected] (J. Kehler). 0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.bmcl.2011.04.067
compulsive disorder.7 Notably, MP-10 (Fig. 1) is currently being evaluated in clinical trials for the treatment of schizophrenia.7,8 Preclinical evidence suggests that a PDE10A inhibitor could provide anti-psychotic, pro-cognitive and negative symptom efficacy.7,8 In a HTS campaign at H. Lundbeck A/S we identified 2-amino[1,2,4]triazolo[1,5-c]quinazolines 5a–d (Fig. 1) and the (2-methyl-[1,2,4]triazolo[1,5-c]quinazolin-5-ylsulfanyl)-acetic acid ethyl ester 6 (Fig. 1) to be low-micromolar PDE10 inhibitors. We found the triazoloquinazoline scaffold to be an attractive starting point for ligand design, because the triazoloquinazoline makes up a rigid template, which is ideally placed within the CNS drugability space due to its molecular properties (MW 300, TPSA 70 Å2, c log P 2.5).12 However, initial SAR-evaluation of the 2-amino triazoloquinazolines revealed a relatively flat SAR (for a few representative examples see Table 1). The flat SAR of the amino-triazoloquinazolines appeared problematic to us and we therefore turned our attention to explore the SAR around the 2-methyltriazoloquinazoline thioether 6. The compounds were synthesized using the three-step protocol outlined in Scheme 1. Commercially available substituted 2-amino-benzonitriles, 7, were converted to isothiocyanates, 8, by treatment with thiophosgene in methylene dichloride and water, and stirring at room temperature for 4–20 h. The organic phase was extracted and evaporated to yield crystalline solids that were used without
3739
J. Kehler et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3738–3742
N N
N O
N
N
O
O N
O
N
O
N
O
O
1, papaverine
N
N N
3, MP-10
2, PQ10 NH2
N
O
O
N
N
N
N
N
N
N N
N
S O
N
R
OEt
5a-e
4
6
Figure 1. Some published PDE10A inhibitors 1,7d 2,8b 3,9 4,10 and the novel triazoloquinazoline leads 5a–d and 6.
Table 1 IC50 of compounds 1–8 for the PDE10A enzyme Compound
PDE10A IC50a (nM)
1, papaverine 2, PQ10 (Pfizer) 3, MP10 (Pfizer) 4, (Wyeth–Elion) 5a, R = 2-Br 5b, R = 4-Br 5c, R = 3,4-MeO 5d, R = 2-Cl 5e, R = H 6
210 64 2.6 32 1000 31% inh@10,000 4500 41% inh@10,000 27% inh@10,000 3000
a Values are means of three experiments, and a typical standard deviation was ±30%.
a)
R1
C
N
S
NH2
R1
N
Cl
Cl
rt/CH2Cl2
Table 2 IC50 of compounds 10–21 for the PDE10A enzyme
N
7
C
8
S
H2N
H N
R2 O
b)
9
µw 130°C EtOH 2
R1 10
1
9
N
8 7
R2
N N3
N 5 S R3 11-51
purification in the following step. Alkyl acid hydrazide 9 (1.3 equiv) was added, and the materials dissolved in ethanol and heated to reflux and evaporated to give the substituted 6H[1,2,4]triazolo[1,5-c]quinazoline-5-thiones 10. The reactions were followed by LC–MS, and except in one case the evaporated reaction mixture was used as starting material for the following alkylation without preceding purification. The sulphur was selectively alkylated by different alkylating reagents to give final products 11–51. Initial SAR evaluation focused on the in vitro potency for PDE10A measured in a PDE10A enzyme inhibition assay. First we investigated the SAR around the substituent on the thioether and the methyl group in the 2-position (R1 = H) (Table 2). The SAR shown in Table 2 indicate that a substituent on the sulphur is critical for PDE10A affinity, but there seem to be some tolerability to the type of thioether since both methyl, ethyl, acetamide, ethylene-morpholine give rise to equipotent compounds with affinities in the 1–2 lM range. However, the acetonitrile derivative is favored with 10-fold increase in potency compared with the original ester hit 6. A few small variations of the R2 group were explored with the R3 group kept constant as acetonitrile, but no improvement in potency beyond R2 = Me was achieved, see compounds 17–21 in Table 2. In order to better understand the SAR of this compound series we carried out a number of docking studies, which demonstrated that several different binding modes could be possible. To clarify the binding mode more precisely we obtained a crystal of the complex between 14 and the PDE10A enzyme for which the X-ray diffraction data was consistent with the binding mode shown in Figure 2. The crystal structure of 14 revealed a novel binding mode compared with previously published crystal structures of PDE10A inhibitors.8d From a structural point of view, there are two major contributions to binding, one predominantly polar and the other predominantly hydrophobic. The nitrogen in the 1-position of the triazole part of the triazoloquinazoline core accepts a hydrogen bond from the conserved Gln-726 and the aromatic part of the core interacts through hydrophobic and p-stacking aromatic interactions with Phe-726 and Ile-692 in the central part of the binding pocket, which has been termed ‘the clamp’. The thiomethylene acts as a linker reaching backwards in the binding pocket into what has been termed ‘the Q1-pocket’.8 Here, the acetonitrile functions as a hydrogen bond acceptor to the hydroxyl group of Ser-667 with a distance to the to the hydroxyl group of Ser-667 of 1.62 Å. The
Compound
R2
R3
PDE10A IC50a (nM)
10 11 12 13 14
CH3 CH3 CH3 CH3 CH3
H CH3 Ethyl –CH2C(O)NH2 –CH2CN
>10000 1800 2400 1300 310
O 15
CH2
CH3
L
N
R3
c)
N N H
N S
O
16
CH3
17 18 19
H –CH2OH –CH2CH2OH
20
CH N
10 21
Scheme 1. Reagents and conditions: (a) thiophosgene in methylene dichloride and water, and stirring at room temperature for 4–20 h; (b) microwave heating, 130 °C, for 4–20 h; (c) NaH (60%), toluene, microwave heating, 135 °C. L is a leaving group.
1300
O
R2 R1
N
CH N
CH2
OH O
N
2500
–CH2CN –CH2CN –CH2CN
1300 >10,000 57% inhib@10000
–CH2CN
960
–CH2CN
>10,000
a Values are means of three experiments, and a typical standard deviation was ±30%.
3740
J. Kehler et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3738–3742
a
a
b TYR GLN 726 693 O H2N OH
Q2
N N
S N
b
Q1 C N
TYR 693
GLN 726
Q1
O
SER 667 HO
N
H2N
OH
Q2
N N H
N N
S
N
N
Water
Water Zn+2 Mg +2 Figure 2. (a) X-ray structure of compound ligand 14; (b) Cartoon showing binding interactions and pharmacophore of ligand 14 with the catalytic pocket in the PDE10A enzyme.
X-ray therefore nicely is consistent with the relatively high affinity of compound 14 and especially the particular role of the cyanogroup. However, it is very unlikely that the observed (scaffold) binding mode for 14 can be adapted by all structures in Table 2. In particular it is clear, that the bigger morpholine groups of 15 and 16 will force a reorientation of the core scaffold in the binding site similarly to what is observed for compound 42 in Figure 3 (vide infra). We next attempted to optimise the potency by exploring the SAR of different substituents at positions 7–10 of the triazolo-quinazoline scaffold, while keeping the acetonitrile (R3) and 2-methyl (R2) substituents constant. The resulting potency data are shown in Table 3. The SAR described in Table 3, show that the the substituents in the positions 7–10 of the triazoloquinazoline ring are important for PDE10A affinity. It seems that highest potency is obtained with relatively small, lipophilic substituents at the 9-position. Bromine gives the highest potency, approximately 7- to 10fold higher than methyl, fluorine and methoxy, and 3-fold higher than chlorine. This SAR is consistent with the scaffold binding mode shown in Figure 2, which predicts a relatively lipophilic area in the clamp region of the binding pocket (see above) where the triazoloquinazoline binds. Relatively large substituents like phenyl and morpholine were not tolerated at position 9. The most potent compound, the 9-bromo-derivative 33, had an IC50 of 35 nM.
Mg +2
Zn+2
Figure 3. (a) X-ray structure of compound ligand 42; (b) Cartoon showing binding interactions and pharmacophore of ligand 42 with the catalytic pocket in the PDE10A enzyme.
Compound 33 was screened in a panel of seven other PDEs and showed greater than 300-fold selectivity towards PDE1b, PDE2, PDE3A, PDE7b, PDE9. However, compound 33 had only about a 100-fold selectivity over PDE4d6 (IC50 = 2200 nM) and PDE5a (IC50 = 5700 nM). Preliminary screening of a few other compounds in this series indicated that selectivity towards PDE4 pose a problem for some of these compounds, since the PDE4 potency of compound 32 (PDE4/PDE10 IC50 = 3000/1900 nM), compound 14 (PDE4/PDE10 IC50 = 1300/310 nM) and compound 38 (PDE4/ PDE10 IC50 = 640/790 nM) were in the same range as the potency at PDE10A. Due to the limited selectivity towards PDE4 we decided to revisit structural variations of and expand the SAR-study around the thioether moiety using compound 11 as a lead-structure. Starting from compound 10 and using the alkylation reactions shown in Scheme 1, we prepared compounds 39–51 shown in Table 4. The SAR indicates that the PDE10 potency is very sensitive to the type of sulphur substituent, and the methylene acetophenone-derivative 39 completely lost activity at the PDE10 enzyme. However, the introduction of heterocycles like pyridyl (compound 40) and quinolinyl (compound 47) resulted in an order of magnitude higher PDE10 potency compared with the thiomethyl lead 11. The most potent heterocycles turned out to be the benzimidazole (42 and 43) and imidazopyridine (46) which had a 100-fold higher PDE10A potency compared with the lead-compounds. Ring-transformation
3741
J. Kehler et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3738–3742 Table 3 IC50 of compounds 14, and 22–38 for the PDE10A enzyme
Table 4 IC50 of compounds 39-51 for the PDE10A enzyme Compound
R1
10
N
8 7
N
PDE10A IC50, nMa
2
N
9
R3’ Cl
N
Cl
S
C N
39
>10000
*
22-38
O
Compound
R1
PDE10A IC50a (nM)
14 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
H 10-F 9-F 8-F 7-F 10-Me 9-Me 8-Me 10-CF3 10-Br 8-Br 8-CF3 9-Br 9-Cl 9-MeO 9-Ph 9-(1-Morpholinyl) 8,9-DiMeO
310 1600 460 320 1500 170 460 1100 450 1000 2300 1900 35 130 250 >10,000 8900 790
a Values are means of three experiments, and a typical standard deviation was ±30%.
40
H N
41
990
* N H N
42
12
* N
N
43
14
* N O
44
110
* N
N *
45
55 N
N
N
46
*
15
N N
*
from the benzimidazole 43 to phenyl imidazoles (49–51) gave a 5to 10-fold drop in PDE10A potency compared with the fused bicyclic benzimidazoles. When revisiting the X-ray structure of compound 14, it became clear from docking-studies, that the heterocycles in compounds 39–51 would not fit in the same binding mode due to lack of space in the binding pocket as visualised in Figure 2. In order to better understand the SAR and binding mode of this series of compounds, we obtained a high-resolution X-ray structure of the catalytic site of the PDE10A enzyme with the most potent compound (42) in (Fig. 3).11 Notably, structural differences between the two crystal structures of the catalytic site of the PDE10A enzyme are negligible and overlaps consistently with earlier published X-ray structures of PDE10A. Interestingly, this X-ray structure revealed yet another novel bidentate binding mode. Here the 1-nitrogen of the triazoloquinazoline core accepts a hydrogen bond from Gln-726 and the aromatic part of the core participates with hydrophobic and aromatic interactions in the clamp with Phe-729 and Ile-692. The thiomethylene acts as a linker reaching into the (unique) Q2-pocket, which among the PDEs is found only in PDE10A. The benzimidazole fills out this lipophilic Q2-pocket and the benzimidazole nitrogen functions as a hydrogen bond acceptor to the hydroxyl group of Tyr-693. The Xray therefore nicely is consistent with the high potency of compound 42. The 5-(1H-Benzoimidazol-2-ylmethylsulfanyl)-2-methyl[1,2,4]triazolo[1,5-c]quinazoline, 42, was screened in a panel of six other PDEs and showed greater than 1000-fold selectivity towards PDE1b, PDE2, PDE3A, PDE4d6, PDE5a, PDE7b, PDE9 and about 100-fold selectivity towards PDE11. Compound 42, therefore, looked like an attractive candidate for in vivo studies, however, we disappointingly found that compound 42 did not have any brain exposure with a brain/plasma ratio (B/P)
