Bioorganic & Medicinal Chemistry Letters 25 (2015) 1864–1868
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl
Evolution and synthesis of novel orally bioavailable inhibitors of PDE10A q Douglas F. Burdi ⇑, John E. Campbell , Jun Wang à, Sufang Zhao à, Hua Zhong §, Jianfeng Wei §, Una Campbell, Liming Shao –, Lee Herman, Patrick Koch, Philip G. Jones, Michael C. Hewitt k Sunovion Pharmaceuticals Inc., 84 Waterford Drive, Marlborough, MA 01752, United States
a r t i c l e
i n f o
Article history: Received 17 November 2014 Revised 17 March 2015 Accepted 18 March 2015 Available online 26 March 2015 Keywords: PDE10A PDE10A inhibitor Phosphodiesterase inhibitor PCP-induced hyperlocomotion Schizophrenia
a b s t r a c t The design and synthesis of highly potent, selective orally bioavailable inhibitors of PDE10A is reported. Starting with an active compound of modest potency from a small focused screen, we were able to evolve this series to a lead molecule with high potency and selectivity versus other PDEs using structure-based design. A systematic refinement of ADME properties during lead optimization led to a lead compound with good half-life that was brain penetrant. Compound 39 was highly potent versus PDE10A (IC50 = 1.0 nM), demonstrated high selectivity (>1000-fold) against other PDEs and was efficacious when dosed orally in a rat model of psychosis, PCP-induced hyperlocomotion with an EC50 of 1 mg/kg. Ó 2015 Elsevier Ltd. All rights reserved.
The cyclic nucleotide phosphodiesterases are a class of enzymes that regulate signal transduction by hydrolyzing the 30 -50 monophosphate bond of the second messengers 30 ,50 -adenosine monophosphate (cAMP) and 30 ,50 -guanosine monophosphate (cGMP). There are 11 distinct families of phosphodiesterases, designated PDE1 through PDE11, which vary in their sequence, localization and substrate specificities for cAMP and cGMP.1 Some PDE family members hydrolyze cAMP or cGMP selectively, while others, so called dual-substrate PDE’s, hydrolyze both.2 PDE10A falls into the latter category, and hydrolyzes cAMP with a higher affinity (Km = 0.05 lM) than cGMP (Km = 3 lM).3
q Atomic coordinates of the PDE10A crystal structure with compound 14 (accession number 47QH) and compound 39 (accession number 4YS7) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, New Jersey. ⇑ Corresponding author. E-mail address: [email protected] (D.F. Burdi). Present address: Epizyme Inc., 400 Technology Square, 4th Floor, Cambridge, MA 02139, United States. à Shanghai ChemPartner, Ltd, 965 Halei Rd., Zhangjiang Hi-Tech Park, Pudong New Area, Shanghai 201203, China. § PharmAdvance, 159 Middle Chengjiang Rd., Hi-Tech Park, D501, Jiangyin, Jiangsu 214431, China. – Present address: Fudan University, Zhangjiang Institute, Center for Drug Discovery & Development, 826 Zhangheng Rd., Pudong, Shanghai 201203, China. k Present address: Constellation Pharmaceuticals, 215 First Street, Suite 200, Cambridge, MA 02142, United States.
http://dx.doi.org/10.1016/j.bmcl.2015.03.050 0960-894X/Ó 2015 Elsevier Ltd. All rights reserved.
The high expression level of PDE10A in the striatum has made it a target of great interest in diseases associated with basal ganglia dysfunction, such as schizophrenia and Huntington’s Disease.4 Although PDE10A inhibitors are active in the rodent phencyclidine (PCP) and amphetamine (AMPH)-induced hyperlocomotion models of antipsychotic activity, recent clinical trials in schizophrenia have been a disappointment.5 With results showing a neuroprotective role for chronic PDE10A inhibition in a mouse model of Huntington’s Disease,6,7 attention lately has turned to this disorder as a potential indication. Our search for PDE10A inhibitors began with a focused screen8 of a proprietary set of 5040 compounds, and yielded triazole 1 (Fig. 1) as an initial hit. Compound 1 displayed modest potency against PDE10A (IC50 = 5 lM) and showed little or no inhibition of other phosphodiesterases. However, a broad survey of 87 purchased thioether analogs offered no improvement in potency (data not shown), so our efforts turned to modification of the two-atom linker. To this end, imidazole 2 was synthesized and identified as a promising hit, with an improved IC50 of 65 nM, and good selectivity against other phosphodiesterases. The synthesis of 2 is shown in Scheme 1. Acylation of 5-phenyl1H-imidazole with dimethylsulfamoyl chloride afforded 3, which was formylated at the 2-position using LDA/DMF to afford 4. Coupling of 4 to 2-methylquinazolin-4(3H)-one was achieved using zinc chloride in acetic acid, and afforded 5. Reduction of 5 using catalytic hydrogenation afforded 2.
1865
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868 O
Table 1 Representative bicycles on the left-hand side of the phenyl-imidazole series
O NH
H N
S
N
NH
H N
N
N N
N
1
R2 N
R1
2
R3
N
Figure 1. Evolution of the lead series. R1
O H N
O OHC
N
H
2
0.065
73.9
NH
H
H
6
0.320
28.0
NH
H
H
7
0.150
39.8
H
H
8
>10
nt
H
H
9
>10
226
H
H
10
0.480
207
H
H
11
0.225
529
H
H
12
>10
nt
H
H
13
0.060
242
H
H
14
0.009
229
H
Me
15
0.110
178
Me
H
16
0.090
55
O N N
NH d
H
N
4 O
H N
HLM (lL min1 mg1)
O
3
NH
IC50 (lM)
N
N
N
ID
NH c
N
O
R3
O
N S O N
b
a
N
N S O N
R2
H N
N
O O
NH
N
N
O 5
N
2
Scheme 1. Reagents: (a) dimethylsulfamoyl chloride, K2CO3, DMF; (b) LDA, DMF, THF; (c) 2-methylquinazolin-4(3H)-one, ZnCl2, HOAc; (d) H2, Pd/C, MeOH.
We began our SAR survey of this series by examining a series of azole isosteres, and we quickly realized that additional heterocycles beyond imidazole were not tolerated at this position (see Supplemental information).9 We then turned our attention to the bicyclic ring on the left-hand side of the molecule. Table 1 shows a representative set of SAR for a variety of bicyclic substituents on the left-hand side of our lead series. In order to gain a better understanding of how ADME correlates with other physicochemical properties, we also gathered data on human microsomal clearance. Several SAR trends are apparent. Carbocyclic and monocyclic rings, as in 9 and 12, respectively, were not tolerated, but incremental improvements in the potency of the bicyclics were realized by the introduction of one nitrogen, as in 10 and 11, and then two nitrogens, as seen in 13–16. Compounds possessing the oxa-quinazoline ring, such as 2, showed better microsomal stability than their quinazoline counterparts such as 13. Heteroatom substituents on the oxa-quinazoline ring, as seen in 6 and 7, were tolerated, but exocyclic substituents at these positions, as in 8, were not. Finally, methylation of the imidazole ring at either nitrogen or carbon, as seen in 15 and 16, respectively, led to some loss in potency, but N-methylation in particular led to improvements in in vitro microsomal intrinsic clearance. While we were pleased to see achievement of single-digit nanomolar potency with 14, we were still troubled by its poor microsomal stability. The combination of good potency and microsomal stability, at this juncture, remained elusive (Table 2). Based on metabolite ID studies which identified the aryl substituent of the imidazole as a probable site of metabolism, we turned our attention to aryl substitutions of the imidazole ring with the hope of attenuating in vitro intrinsic clearance. Table 2 shows a series of aryl- and heteroaryl-substituted imidazoles, where some trends in the SAR are readily apparent. N-methylation of the imidazole ring usually improved microsomal stability, as seen in the comparison between phenyl analogs 14 and 17 and furans 18 and 19. Addition of a 3-fluoro substituent to the phenyl ring, as in 20, resulted in a slight loss of potency relative to 14, and failed to reduce in vitro microsomal clearance. Replacement of the
N
N
N N N N N N N N N
phenyl substituent with a 4-pyridyl heterocycle, as in 21, voided activity, while the 3-pyridyl substituent seen in 22 and 23 was tolerated. Although we were encouraged by the favorable microsomal stability of 23, its potency was insufficient for lead advancement, so we began to explore other scaffold opportunities for improving affinity. During the course of hit refinement, we had an opportunity to co-crystallize several of our actives with human PDE10A. Figure 2 shows a co-crystal structure of 14 with human PDE10A. The phenyl imidazole resides in the selectivity pocket10 with the imidazole nitrogen forming a hydrogen bond with the side-chain hydroxyl group of Y693. Interestingly, the phenyl ring is nearly coplanar with the imidazole ring, and this observation would become important for our scaffold designs. The quinazoline ring resides deeper in the active site where it is sandwiched in the hydrophobic clamp defined by F696 and F729; the ring nitrogen forms a hydrogen bond with Q726. The co-crystal structure proved useful in our efforts at structure-based drug design. Based on the aforementioned co-planarity of the phenyl and imidazole rings in the co-crystal structure of 14, we reasoned, as shown in Figure 3, that fusion of the pyridine
1866
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868
Table 2 Variations of the aryl imidazole substituent of the lead series R1 N
N N
R1
R2
N
R2
ID
IC50 (lM)
HLM (lL min1 mg1)
H
14
0.009
229
Me
17
0.095
55.2
18
0.021
171
19
0.018
30.5
20
0.058
184
21
>10
188
O
H
O
Me
F
H
Me N
H
N
22
0.11
46.4
Me
N
23
0.23
16.2
Figure 2. Co-crystal structure of 14 with PDE10A.
N
N
N N
N 23
N
N N
N
N
24
Figure 3. Fusion of the pyridine substituent of 23 with its imidazole ring to form 24.
substituent of 23 with the imidazole ring to form 24 would confer an entropic advantage by increasing the conformational constraint between the pyridine and imidazole rings. The synthesis of 24 and closely related analogues was accomplished in straightforward fashion and is shown in Scheme 2.11 Benzene-1,2-diamine 25 was condensed with pyruvic aldehyde to afford quinoxaline 26. Oxidation to aldehyde 27 was achieved using selenium dioxide. Horner–Wadsworth–Emmons olefination using methyl 2-(dimethylphosphono)-acetate afforded 28 as a
mixture of cis and trans isomers, which was taken forward as a mixture and hydrolyzed to carboxylic acid 29. Reduction of the olefin to form 30 was achieved via hydrogenation. Diamine 32, prepared from commercially available 31 via reduction with Raney nickel in hydrazine hydrate, was coupled with 30 using EDCI to form the intermediate amide which was cyclized in situ using acetic acid to form 24. We were pleased to see that the scaffold transition of 23 to 24 gave the desired improvement in potency while in large part maintaining microsomal stability, as shown in Table 3. Armed with the co-crystal structure of 14, we decided to return to optimization of the bicyclic ring in order to make the final push toward single digit nanomolar potency. As shown in Table 3, we designed bicyclic rings that maintained a heteroatom at the position ortho to the ethylene linker in order to maintain hydrogen-bonding contact with Q726. Removal of one nitrogen from the quinazoline ring resulted in 33, which gave a boost in potency, but also led to higher clearance. Addition of a nitrogen atom to the carbocyclic portion of quinoline 33, as seen in 34 and 35, attenuated the microsomal clearance of 33 somewhat, but also led to a loss of potency. Contraction of the pyrazine ring of 24 accompanied by addition of a nitrogen to the fused benzene ring, led to a dramatic loss of potency as seen in 36, while benzoxazole 37 showed only modest potency. During the course of lead optimization, a report was published12 that described a series of ethylene-linked phenyl-imidazoles whose structures bore a close resemblance to compounds 14 and 17–23. Since these molecules were likely to bind in similar fashion to the active site of PDE10A, we decided to incorporate key aspects of these molecules into our lead series. To this end, compounds 38–41 were synthesized and screened, and the results were very informative. Triazolopyridine 38 gave a considerable increase in potency relative to its benzoxazole isostere 37, while also improving in vitro microsomal clearance. The addition of 1,4-dimethyl substituents to the triazolopyridine ring, as in 40, resulted in an additional increase in potency to 1 nM, although this was accompanied by an unwelcomed increase in microsomal clearance. Addition of a nitrogen to 40 resulted in triazolopyrazine 39 where we were pleased to see good potency and microsomal stability. Removal of a nitrogen from the triazole ring gave pyrazolopyrazine 41, which was equipotent to 39, but considerably less stable in human microsomes. Finally, incorporation of the 1,4dimethyl motif into 24 resulted in quinazoline 42, which led to an order of magnitude increase in potency, but a dramatic loss of microsomal stability. Among the 1,4-dimethyl substituted bicycles, 39 thus appears to reside on an ‘island’ of microsomal stability for reasons that remain unclear. In order to gain a clearer understanding of the factors contributing to the potency increase seen in 1,4-dimethyl substitution, we co-crystallized 39 with human PDE10A. As seen in Figure 4a, the tricycle binds as expected in the selectivity pocket with the imidazole nitrogen making hydrogen-bonding contact with Y693. The pyrazolopyrazine ring resides in the hydrophobic clamp near the active site where the triazole nitrogen forms a hydrogen bond with Q726. A close examination of Figure 4b reveals the methyl groups residing in a small hydrophobic sub-site defined by I692, F729 and V678, and this may account for the enhancement in affinity of the 1,4-dimethyl analogs. Based on the favorable properties seen with 39, we decided to profile this compound more thoroughly.13 Selectivity was >10,000-fold for a panel of phosphodiesterases 2–11, and a Cerep panel screen of 80 targets revealed >1000-fold selectivity versus off-targets. No significant hERG or Cyp inhibition was noted, and a Caco-2 assay showed good permeability (apical–basal transport of 59 106 cm s1) with no asymmetry in rates of apical and basal transport. Rat plasma protein binding was 88.5%.
1867
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868 NH2
N
a
NH2
N
b
N
25
N 27
26
N
N
d
N
CO2 Et
c CHO
e
N
COOH 29
28
NH 2 N N
N NH 2
+
g
H N
N
COOH
N
32
30
24
N N
f
NO 2 NH 2 N 31
Scheme 2. Reagents: (a) pyruvic aldehyde, EtOH; (b) SeO2, 1,4-dioxane; (c) (MeO)2POCH2CO2Me, NaOH, CH2Cl2, H2O; (d) NaOH, H2O; (e) H2, Pd/C, EtOH; (f) Raney Ni, N2H4H2O; (g) EDCI/HOBt/DCM, then AcOH.
Table 3 Variation of the bicyclic ring of 24 R
N N
R
N
ID
IC50 (lM)
HLM (lL min1 mg1)
24
0.03
27.8
33
0.006
131
34
0.062
24.1
35
0.016
38.5
36
5.7
nt
37
0.26
77.5
38
0.013
11.0
39
0.001
8.9
40
0.001
97.3
41
0.001
64.9
42
0.003
510
N N
N N
N
N N N
N N H N O N N N
N
N N N
N N N
N N N
Figure 4. Co-crystal structure of 39 with PDE10A showing (a) the selectivity pocket, and (b) hydrophobic interactions near the metal binding site.
N N
Evaluation of compound stability in rat liver microsomes revealed a moderate in vitro clearance of 13.8 lL/min/mg, so we decided to evaluate the rat pharmacokinetics of 39.
1868
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868
Table 4 Summary of rat pharmacokinetics for 39 Route
CL (L/h/kg)
Vdss (L/kg)
T1/2 (h)
3 10
IV PO
1.62
8.9
3.9 4.1
Total Distance Traveled (cm)
Dose (mg/kg)
8000 6000
*
4000
0
Brain/plasma 0.4
27
active when dosed orally in a rodent model of psychosis. More extensive in vivo characterization of compound 39 will be described in a subsequent Letter.
10000
2000
%F
* *
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bmcl.2015.03. 050.
* Veh Veh
1 3 0 0.3 SEP-39 (mg/kg,p.o.)
Supplementary data
Risp PCP
+ PCP Figure 5. Effects of 39 on PCP-induced hyperlocomotion in rat (risperidone used as positive control).
As shown in Table 4, 39 showed moderate clearance and volume of distribution in rat with an oral bioavailability of 27%. This favorable pharmacokinetic profile has enabled us to evaluate this molecule in a number of in vivo rodent assays. Inhibition of stimulant-induced hyperlocomotion has been used to assess antipsychotic-like activity of candidate compounds and Figure 5 demonstrates that 39, when dosed orally at 1 and 3 mg/kg, significantly attenuated PCP-induced hyperlocomotion. Further in vivo evaluation of 39 will be discussed in detail in a subsequent publication.14 In summary, we have used structure-based design to evolve a series of imidazoquinoline PDE10A inhibitors with low nanomolar potency, high PDE selectivity and good bioavailability that are
References and notes 1. 2. 3. 4.
5. 6. 7. 8. 9. 10.
11. 12. 13. 14.
Manallack, D. T.; Hughes, R. A.; Thompson, P. E. J. Med. Chem. 2005, 48, 3449. Rotella, D. Nat. Rev. Drug Disc. 2002, 1, 674. Soderling, S.; Beavo, J. A. Curr. Opin. Cell Biol. 1999, 12, 174. Seeger, T. F.; Bartlett, B.; Coskran, T. M.; Culp, J. S.; James, L. C.; Krull, D. L.; Lanfear, J.; Ryan, A. M.; Schmidt, C. J.; Strick, C. A.; Varghese, A. H.; Williams, R. D.; Wylie, P. G.; Menniti, F. S. Brain Res. 2003, 985, 113. DeMartinis, N. A. Biol. Psychiatry 2012, 71, 62S. Giampa, C.; Patassini, S.; Borreca, A.; Laurenti, D.; Marullo, F.; Bernardi, G.; Menniti, F. S.; Fusco, F. R. Neurobiol. Dis. 2009, 34, 450. Giampa, C.; Laurenti, D.; Anzilotti, S.; Bernardi, G.; Menniti, F. S.; Fusco, F. R. PLoS One 2012, 5, e13417. The details of the in vitro assay for PDE10A are provided in Supplemental section. The experimental procedures for the synthesis of 2 and its azole isosteres, along with analogs 21 and 23 are provided in Supplemental section. Verhoest, P. R.; Chapin, D. S.; Corman, M.; Fonseca, K.; Harms, J. F.; Hou, X.; Marr, E. S.; Menniti, F. S.; Nelson, F.; O’Connor, R.; Pandit, J.; Proulx-LaFrance, C.; Schmidt, A. W.; Schmidt, C. J.; Suiciak, J. A.; Liras, S. J. Med. Chem. 2009, 52, 5188. Campbell, J. E.; Hewitt, M. C.; Jones, P.; Xie, L. WO 150156, 2011. Ritzen, A.; Kehler, J.; Langgard, M.; Nielsen, J.; Kilburn, J. P.; Farah, M. M. WO 152825, 2009. A summary of the in vitro properties of 39 is given in Supplemental section. Jones, P. et al. Manuscript accepted for publication.
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl
Evolution and synthesis of novel orally bioavailable inhibitors of PDE10A q Douglas F. Burdi ⇑, John E. Campbell , Jun Wang à, Sufang Zhao à, Hua Zhong §, Jianfeng Wei §, Una Campbell, Liming Shao –, Lee Herman, Patrick Koch, Philip G. Jones, Michael C. Hewitt k Sunovion Pharmaceuticals Inc., 84 Waterford Drive, Marlborough, MA 01752, United States
a r t i c l e
i n f o
Article history: Received 17 November 2014 Revised 17 March 2015 Accepted 18 March 2015 Available online 26 March 2015 Keywords: PDE10A PDE10A inhibitor Phosphodiesterase inhibitor PCP-induced hyperlocomotion Schizophrenia
a b s t r a c t The design and synthesis of highly potent, selective orally bioavailable inhibitors of PDE10A is reported. Starting with an active compound of modest potency from a small focused screen, we were able to evolve this series to a lead molecule with high potency and selectivity versus other PDEs using structure-based design. A systematic refinement of ADME properties during lead optimization led to a lead compound with good half-life that was brain penetrant. Compound 39 was highly potent versus PDE10A (IC50 = 1.0 nM), demonstrated high selectivity (>1000-fold) against other PDEs and was efficacious when dosed orally in a rat model of psychosis, PCP-induced hyperlocomotion with an EC50 of 1 mg/kg. Ó 2015 Elsevier Ltd. All rights reserved.
The cyclic nucleotide phosphodiesterases are a class of enzymes that regulate signal transduction by hydrolyzing the 30 -50 monophosphate bond of the second messengers 30 ,50 -adenosine monophosphate (cAMP) and 30 ,50 -guanosine monophosphate (cGMP). There are 11 distinct families of phosphodiesterases, designated PDE1 through PDE11, which vary in their sequence, localization and substrate specificities for cAMP and cGMP.1 Some PDE family members hydrolyze cAMP or cGMP selectively, while others, so called dual-substrate PDE’s, hydrolyze both.2 PDE10A falls into the latter category, and hydrolyzes cAMP with a higher affinity (Km = 0.05 lM) than cGMP (Km = 3 lM).3
q Atomic coordinates of the PDE10A crystal structure with compound 14 (accession number 47QH) and compound 39 (accession number 4YS7) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, New Jersey. ⇑ Corresponding author. E-mail address: [email protected] (D.F. Burdi). Present address: Epizyme Inc., 400 Technology Square, 4th Floor, Cambridge, MA 02139, United States. à Shanghai ChemPartner, Ltd, 965 Halei Rd., Zhangjiang Hi-Tech Park, Pudong New Area, Shanghai 201203, China. § PharmAdvance, 159 Middle Chengjiang Rd., Hi-Tech Park, D501, Jiangyin, Jiangsu 214431, China. – Present address: Fudan University, Zhangjiang Institute, Center for Drug Discovery & Development, 826 Zhangheng Rd., Pudong, Shanghai 201203, China. k Present address: Constellation Pharmaceuticals, 215 First Street, Suite 200, Cambridge, MA 02142, United States.
http://dx.doi.org/10.1016/j.bmcl.2015.03.050 0960-894X/Ó 2015 Elsevier Ltd. All rights reserved.
The high expression level of PDE10A in the striatum has made it a target of great interest in diseases associated with basal ganglia dysfunction, such as schizophrenia and Huntington’s Disease.4 Although PDE10A inhibitors are active in the rodent phencyclidine (PCP) and amphetamine (AMPH)-induced hyperlocomotion models of antipsychotic activity, recent clinical trials in schizophrenia have been a disappointment.5 With results showing a neuroprotective role for chronic PDE10A inhibition in a mouse model of Huntington’s Disease,6,7 attention lately has turned to this disorder as a potential indication. Our search for PDE10A inhibitors began with a focused screen8 of a proprietary set of 5040 compounds, and yielded triazole 1 (Fig. 1) as an initial hit. Compound 1 displayed modest potency against PDE10A (IC50 = 5 lM) and showed little or no inhibition of other phosphodiesterases. However, a broad survey of 87 purchased thioether analogs offered no improvement in potency (data not shown), so our efforts turned to modification of the two-atom linker. To this end, imidazole 2 was synthesized and identified as a promising hit, with an improved IC50 of 65 nM, and good selectivity against other phosphodiesterases. The synthesis of 2 is shown in Scheme 1. Acylation of 5-phenyl1H-imidazole with dimethylsulfamoyl chloride afforded 3, which was formylated at the 2-position using LDA/DMF to afford 4. Coupling of 4 to 2-methylquinazolin-4(3H)-one was achieved using zinc chloride in acetic acid, and afforded 5. Reduction of 5 using catalytic hydrogenation afforded 2.
1865
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868 O
Table 1 Representative bicycles on the left-hand side of the phenyl-imidazole series
O NH
H N
S
N
NH
H N
N
N N
N
1
R2 N
R1
2
R3
N
Figure 1. Evolution of the lead series. R1
O H N
O OHC
N
H
2
0.065
73.9
NH
H
H
6
0.320
28.0
NH
H
H
7
0.150
39.8
H
H
8
>10
nt
H
H
9
>10
226
H
H
10
0.480
207
H
H
11
0.225
529
H
H
12
>10
nt
H
H
13
0.060
242
H
H
14
0.009
229
H
Me
15
0.110
178
Me
H
16
0.090
55
O N N
NH d
H
N
4 O
H N
HLM (lL min1 mg1)
O
3
NH
IC50 (lM)
N
N
N
ID
NH c
N
O
R3
O
N S O N
b
a
N
N S O N
R2
H N
N
O O
NH
N
N
O 5
N
2
Scheme 1. Reagents: (a) dimethylsulfamoyl chloride, K2CO3, DMF; (b) LDA, DMF, THF; (c) 2-methylquinazolin-4(3H)-one, ZnCl2, HOAc; (d) H2, Pd/C, MeOH.
We began our SAR survey of this series by examining a series of azole isosteres, and we quickly realized that additional heterocycles beyond imidazole were not tolerated at this position (see Supplemental information).9 We then turned our attention to the bicyclic ring on the left-hand side of the molecule. Table 1 shows a representative set of SAR for a variety of bicyclic substituents on the left-hand side of our lead series. In order to gain a better understanding of how ADME correlates with other physicochemical properties, we also gathered data on human microsomal clearance. Several SAR trends are apparent. Carbocyclic and monocyclic rings, as in 9 and 12, respectively, were not tolerated, but incremental improvements in the potency of the bicyclics were realized by the introduction of one nitrogen, as in 10 and 11, and then two nitrogens, as seen in 13–16. Compounds possessing the oxa-quinazoline ring, such as 2, showed better microsomal stability than their quinazoline counterparts such as 13. Heteroatom substituents on the oxa-quinazoline ring, as seen in 6 and 7, were tolerated, but exocyclic substituents at these positions, as in 8, were not. Finally, methylation of the imidazole ring at either nitrogen or carbon, as seen in 15 and 16, respectively, led to some loss in potency, but N-methylation in particular led to improvements in in vitro microsomal intrinsic clearance. While we were pleased to see achievement of single-digit nanomolar potency with 14, we were still troubled by its poor microsomal stability. The combination of good potency and microsomal stability, at this juncture, remained elusive (Table 2). Based on metabolite ID studies which identified the aryl substituent of the imidazole as a probable site of metabolism, we turned our attention to aryl substitutions of the imidazole ring with the hope of attenuating in vitro intrinsic clearance. Table 2 shows a series of aryl- and heteroaryl-substituted imidazoles, where some trends in the SAR are readily apparent. N-methylation of the imidazole ring usually improved microsomal stability, as seen in the comparison between phenyl analogs 14 and 17 and furans 18 and 19. Addition of a 3-fluoro substituent to the phenyl ring, as in 20, resulted in a slight loss of potency relative to 14, and failed to reduce in vitro microsomal clearance. Replacement of the
N
N
N N N N N N N N N
phenyl substituent with a 4-pyridyl heterocycle, as in 21, voided activity, while the 3-pyridyl substituent seen in 22 and 23 was tolerated. Although we were encouraged by the favorable microsomal stability of 23, its potency was insufficient for lead advancement, so we began to explore other scaffold opportunities for improving affinity. During the course of hit refinement, we had an opportunity to co-crystallize several of our actives with human PDE10A. Figure 2 shows a co-crystal structure of 14 with human PDE10A. The phenyl imidazole resides in the selectivity pocket10 with the imidazole nitrogen forming a hydrogen bond with the side-chain hydroxyl group of Y693. Interestingly, the phenyl ring is nearly coplanar with the imidazole ring, and this observation would become important for our scaffold designs. The quinazoline ring resides deeper in the active site where it is sandwiched in the hydrophobic clamp defined by F696 and F729; the ring nitrogen forms a hydrogen bond with Q726. The co-crystal structure proved useful in our efforts at structure-based drug design. Based on the aforementioned co-planarity of the phenyl and imidazole rings in the co-crystal structure of 14, we reasoned, as shown in Figure 3, that fusion of the pyridine
1866
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868
Table 2 Variations of the aryl imidazole substituent of the lead series R1 N
N N
R1
R2
N
R2
ID
IC50 (lM)
HLM (lL min1 mg1)
H
14
0.009
229
Me
17
0.095
55.2
18
0.021
171
19
0.018
30.5
20
0.058
184
21
>10
188
O
H
O
Me
F
H
Me N
H
N
22
0.11
46.4
Me
N
23
0.23
16.2
Figure 2. Co-crystal structure of 14 with PDE10A.
N
N
N N
N 23
N
N N
N
N
24
Figure 3. Fusion of the pyridine substituent of 23 with its imidazole ring to form 24.
substituent of 23 with the imidazole ring to form 24 would confer an entropic advantage by increasing the conformational constraint between the pyridine and imidazole rings. The synthesis of 24 and closely related analogues was accomplished in straightforward fashion and is shown in Scheme 2.11 Benzene-1,2-diamine 25 was condensed with pyruvic aldehyde to afford quinoxaline 26. Oxidation to aldehyde 27 was achieved using selenium dioxide. Horner–Wadsworth–Emmons olefination using methyl 2-(dimethylphosphono)-acetate afforded 28 as a
mixture of cis and trans isomers, which was taken forward as a mixture and hydrolyzed to carboxylic acid 29. Reduction of the olefin to form 30 was achieved via hydrogenation. Diamine 32, prepared from commercially available 31 via reduction with Raney nickel in hydrazine hydrate, was coupled with 30 using EDCI to form the intermediate amide which was cyclized in situ using acetic acid to form 24. We were pleased to see that the scaffold transition of 23 to 24 gave the desired improvement in potency while in large part maintaining microsomal stability, as shown in Table 3. Armed with the co-crystal structure of 14, we decided to return to optimization of the bicyclic ring in order to make the final push toward single digit nanomolar potency. As shown in Table 3, we designed bicyclic rings that maintained a heteroatom at the position ortho to the ethylene linker in order to maintain hydrogen-bonding contact with Q726. Removal of one nitrogen from the quinazoline ring resulted in 33, which gave a boost in potency, but also led to higher clearance. Addition of a nitrogen atom to the carbocyclic portion of quinoline 33, as seen in 34 and 35, attenuated the microsomal clearance of 33 somewhat, but also led to a loss of potency. Contraction of the pyrazine ring of 24 accompanied by addition of a nitrogen to the fused benzene ring, led to a dramatic loss of potency as seen in 36, while benzoxazole 37 showed only modest potency. During the course of lead optimization, a report was published12 that described a series of ethylene-linked phenyl-imidazoles whose structures bore a close resemblance to compounds 14 and 17–23. Since these molecules were likely to bind in similar fashion to the active site of PDE10A, we decided to incorporate key aspects of these molecules into our lead series. To this end, compounds 38–41 were synthesized and screened, and the results were very informative. Triazolopyridine 38 gave a considerable increase in potency relative to its benzoxazole isostere 37, while also improving in vitro microsomal clearance. The addition of 1,4-dimethyl substituents to the triazolopyridine ring, as in 40, resulted in an additional increase in potency to 1 nM, although this was accompanied by an unwelcomed increase in microsomal clearance. Addition of a nitrogen to 40 resulted in triazolopyrazine 39 where we were pleased to see good potency and microsomal stability. Removal of a nitrogen from the triazole ring gave pyrazolopyrazine 41, which was equipotent to 39, but considerably less stable in human microsomes. Finally, incorporation of the 1,4dimethyl motif into 24 resulted in quinazoline 42, which led to an order of magnitude increase in potency, but a dramatic loss of microsomal stability. Among the 1,4-dimethyl substituted bicycles, 39 thus appears to reside on an ‘island’ of microsomal stability for reasons that remain unclear. In order to gain a clearer understanding of the factors contributing to the potency increase seen in 1,4-dimethyl substitution, we co-crystallized 39 with human PDE10A. As seen in Figure 4a, the tricycle binds as expected in the selectivity pocket with the imidazole nitrogen making hydrogen-bonding contact with Y693. The pyrazolopyrazine ring resides in the hydrophobic clamp near the active site where the triazole nitrogen forms a hydrogen bond with Q726. A close examination of Figure 4b reveals the methyl groups residing in a small hydrophobic sub-site defined by I692, F729 and V678, and this may account for the enhancement in affinity of the 1,4-dimethyl analogs. Based on the favorable properties seen with 39, we decided to profile this compound more thoroughly.13 Selectivity was >10,000-fold for a panel of phosphodiesterases 2–11, and a Cerep panel screen of 80 targets revealed >1000-fold selectivity versus off-targets. No significant hERG or Cyp inhibition was noted, and a Caco-2 assay showed good permeability (apical–basal transport of 59 106 cm s1) with no asymmetry in rates of apical and basal transport. Rat plasma protein binding was 88.5%.
1867
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868 NH2
N
a
NH2
N
b
N
25
N 27
26
N
N
d
N
CO2 Et
c CHO
e
N
COOH 29
28
NH 2 N N
N NH 2
+
g
H N
N
COOH
N
32
30
24
N N
f
NO 2 NH 2 N 31
Scheme 2. Reagents: (a) pyruvic aldehyde, EtOH; (b) SeO2, 1,4-dioxane; (c) (MeO)2POCH2CO2Me, NaOH, CH2Cl2, H2O; (d) NaOH, H2O; (e) H2, Pd/C, EtOH; (f) Raney Ni, N2H4H2O; (g) EDCI/HOBt/DCM, then AcOH.
Table 3 Variation of the bicyclic ring of 24 R
N N
R
N
ID
IC50 (lM)
HLM (lL min1 mg1)
24
0.03
27.8
33
0.006
131
34
0.062
24.1
35
0.016
38.5
36
5.7
nt
37
0.26
77.5
38
0.013
11.0
39
0.001
8.9
40
0.001
97.3
41
0.001
64.9
42
0.003
510
N N
N N
N
N N N
N N H N O N N N
N
N N N
N N N
N N N
Figure 4. Co-crystal structure of 39 with PDE10A showing (a) the selectivity pocket, and (b) hydrophobic interactions near the metal binding site.
N N
Evaluation of compound stability in rat liver microsomes revealed a moderate in vitro clearance of 13.8 lL/min/mg, so we decided to evaluate the rat pharmacokinetics of 39.
1868
D. F. Burdi et al. / Bioorg. Med. Chem. Lett. 25 (2015) 1864–1868
Table 4 Summary of rat pharmacokinetics for 39 Route
CL (L/h/kg)
Vdss (L/kg)
T1/2 (h)
3 10
IV PO
1.62
8.9
3.9 4.1
Total Distance Traveled (cm)
Dose (mg/kg)
8000 6000
*
4000
0
Brain/plasma 0.4
27
active when dosed orally in a rodent model of psychosis. More extensive in vivo characterization of compound 39 will be described in a subsequent Letter.
10000
2000
%F
* *
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bmcl.2015.03. 050.
* Veh Veh
1 3 0 0.3 SEP-39 (mg/kg,p.o.)
Supplementary data
Risp PCP
+ PCP Figure 5. Effects of 39 on PCP-induced hyperlocomotion in rat (risperidone used as positive control).
As shown in Table 4, 39 showed moderate clearance and volume of distribution in rat with an oral bioavailability of 27%. This favorable pharmacokinetic profile has enabled us to evaluate this molecule in a number of in vivo rodent assays. Inhibition of stimulant-induced hyperlocomotion has been used to assess antipsychotic-like activity of candidate compounds and Figure 5 demonstrates that 39, when dosed orally at 1 and 3 mg/kg, significantly attenuated PCP-induced hyperlocomotion. Further in vivo evaluation of 39 will be discussed in detail in a subsequent publication.14 In summary, we have used structure-based design to evolve a series of imidazoquinoline PDE10A inhibitors with low nanomolar potency, high PDE selectivity and good bioavailability that are
References and notes 1. 2. 3. 4.
5. 6. 7. 8. 9. 10.
11. 12. 13. 14.
Manallack, D. T.; Hughes, R. A.; Thompson, P. E. J. Med. Chem. 2005, 48, 3449. Rotella, D. Nat. Rev. Drug Disc. 2002, 1, 674. Soderling, S.; Beavo, J. A. Curr. Opin. Cell Biol. 1999, 12, 174. Seeger, T. F.; Bartlett, B.; Coskran, T. M.; Culp, J. S.; James, L. C.; Krull, D. L.; Lanfear, J.; Ryan, A. M.; Schmidt, C. J.; Strick, C. A.; Varghese, A. H.; Williams, R. D.; Wylie, P. G.; Menniti, F. S. Brain Res. 2003, 985, 113. DeMartinis, N. A. Biol. Psychiatry 2012, 71, 62S. Giampa, C.; Patassini, S.; Borreca, A.; Laurenti, D.; Marullo, F.; Bernardi, G.; Menniti, F. S.; Fusco, F. R. Neurobiol. Dis. 2009, 34, 450. Giampa, C.; Laurenti, D.; Anzilotti, S.; Bernardi, G.; Menniti, F. S.; Fusco, F. R. PLoS One 2012, 5, e13417. The details of the in vitro assay for PDE10A are provided in Supplemental section. The experimental procedures for the synthesis of 2 and its azole isosteres, along with analogs 21 and 23 are provided in Supplemental section. Verhoest, P. R.; Chapin, D. S.; Corman, M.; Fonseca, K.; Harms, J. F.; Hou, X.; Marr, E. S.; Menniti, F. S.; Nelson, F.; O’Connor, R.; Pandit, J.; Proulx-LaFrance, C.; Schmidt, A. W.; Schmidt, C. J.; Suiciak, J. A.; Liras, S. J. Med. Chem. 2009, 52, 5188. Campbell, J. E.; Hewitt, M. C.; Jones, P.; Xie, L. WO 150156, 2011. Ritzen, A.; Kehler, J.; Langgard, M.; Nielsen, J.; Kilburn, J. P.; Farah, M. M. WO 152825, 2009. A summary of the in vitro properties of 39 is given in Supplemental section. Jones, P. et al. Manuscript accepted for publication.
