Bioorganic & Medicinal Chemistry Letters 24 (2014) 1657–1660
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl
Pyrazoles as non-classical bioisosteres in prolylcarboxypeptidase (PrCP) inhibitors Thomas H. Graham a,⇑, Min Shu a, Andreas Verras a, Qing Chen c, Margarita Garcia-Calvo b, Xiaohua Li c, JeanMarie Lisnock d, Xinchun Tong c, Elaine C. Tung c, Judyann Wiltsie d, Jeffrey J. Hale a, Shirly Pinto b, Dong-Ming Shen a a
Department of Medicinal Chemistry, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA Department of Metabolic Disorders, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA Department of Drug Metabolism, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA d Department of In Vitro Sciences, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA b c
a r t i c l e
i n f o
Article history: Received 12 January 2014 Revised 22 February 2014 Accepted 25 February 2014 Available online 6 March 2014 Keywords: Non-classical bioisostere Pyrazole Amide Enzyme inhibitor Prolylcarboxypeptidase (PrCP) Angiotensinase C
a b s t r a c t Bioisosteres are integral components of modern pharmaceutical research that allow structural optimization to maximize in vivo efficacy and minimize adverse effects by selectively modifying pharmacodynamic, pharmacokinetic and physicochemical properties. A recent medicinal chemistry campaign focused on identifying small molecule inhibitors of prolylcarboxypeptidase (PrCP) initiated an investigation into the use of pyrazoles as bioisosteres for amides. The results indicate that pyrazoles are suitable bioisosteric replacements of amide functional groups. The study is an example of managing bioisosteric replacement by incorporating subsequent structural modifications to maintain potency against the selected target. A heuristic model for an embedded pharmacophore is also described. Ó 2014 Elsevier Ltd. All rights reserved.
Classical bioisosteres encompass the structural changes to a molecule that include substituting atoms and the associated valences or exchanging of ring equivalents. In contrast, non-classical bioisosteres include the replacement of entire functional groups, exchanging cyclic and non-cyclic forms and the inversion of functional groups (retroisosteres).1 Modern pharmaceutical research relies on bioisosteres to maximize in vivo efficacy and minimize undesired effects by selectively modifying pharmacodynamic, pharmacokinetic and physicochemical properties.2 A recent medicinal chemistry campaign that focused on discovering novel inhibitors of prolylcarboxypeptidase (PrCP) investigated the use of a pyrazole as a bioisostere for an amide. The study revealed that the pyrazole initially attenuated the in vitro potency; however, additional structural modifications restored in vitro potency. Prolylcarboxypeptidase (PrCP) (angiotensinase C) is a single chain serine peptidase that cleaves the amide bond between a C-terminal amino acid and a proline residue (i.e. peptide-Pro-XxxOH).3 Several hypotheses concerning the biological role of PrCP are based on the ability of PrCP to degrade bioactive peptides4,5
⇑ Corresponding author. Tel.: +1 732 594 2918. E-mail address: [email protected] (T.H. Graham). http://dx.doi.org/10.1016/j.bmcl.2014.02.070 0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.
Me
Me
O N
N O
O N
Ph N H
Ph 1
PrCP IC50 (h,m): 3.5, 5.0 nM
Cl
Ph Ph
2
PrCP IC50 (h,m): 2.5, 1.4 nM
Figure 1. Recently discovered structures of potent PrCP inhibitors.
as well as the extensive tissue distribution of PrCP.6,7 Studies have suggested that PrCP may be involved in cardiovascular control,8 metabolism regulation,9,10 inflammation induction,11 angiogenesis12 and chemotherapy resistance.13 The possibility for numerous therapeutic indications has resulted in the rapid development of small molecule inhibitors of PrCP.14 A program initiated to identify small molecule inhibitors of PrCP has produced several classes of potent inhibitors.15 The aminocyclohexanes, represented by 1 and 2 (Fig. 1) were recently discovered as potent inhibitors of PrCP.16,17 During lead-optimization studies, a bioisosteric replacement of the amide functional group was explored. Many heterocycles have been used as
1658
T. H. Graham et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1657–1660
Table 1 Initial attempts to incorporate a pyrazole bioisostere into the lead molecules
Table 3 Optimization of the lead with the pyrazole bioisostere F
PrCP IC50 h,ma (nM)
Compd
HN N
HN N
Me b
Cl
860,—
3
R
N
Compd
O
PrCP IC50 h,ma (nM)
R=
8b
HN N
Me
N
b
Cl
4
74, 20 Cl
940,—
3.5, 2.2 92,— 114,— 40, 24
9 4-epi-9 ent-9 4-epi-(ent-9)
N O
Br
OH
10
HN N
5
13, 5.9
270, 220
Cl
O
69, 30
11
N O Br
12b HN N
epi-5
13b
580
Cl
14b
N O
15b
a
Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse. b Compound was tested as the racemate.
16b
Cl
150,— F F
F F
40, 56 F
F
Me Me
1500,— Me
66, 180
Me
150,— Me
17b
bioisosteres of amides including oxazoles, thiazoles, triazoles, tetrazoles and oxadiazoles; however, the use of pyrazoles as amide bioisosteres, is less common.18 The present study was designed
Table 2 Data supporting the heuristic model of a linear pharmacophore Compd
PrCP IC50 h,ma (nM) F
O
6
10.4, 4.7 N
N H
Ph Ph
Cl
F
O
7
N N
990,— Ph Ph
Cl F
O
850,—
ent-7 N N
Ph Ph
Cl a Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse.
H
1000,—
a
Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse. b Compound was tested as the racemate.
to explore the possibility that a pyrazole can replace an amide functional group in a series of PrCP inhibitors.19 Replacement of the amide in 1 with a pyrazole afforded 3; however, the potency was significantly attenuated by the structural modification.20 Attempts to regain the potency with additional structural modifications afforded compounds 4 and 5 that also proved to have attenuated inhibitory activity. In addition, compounds 5 and epi-5 indicated that the relative stereochemistry of the morpholine substituent has minimal effect on the in vitro inhibition of PrCP.21 The data in Table 1 indicated that the incorporation of the pyrazole as an amide bioisostere resulted in a substantial loss in potency. An additional set of compounds suggested a possible reason for the decrease in activity (Table 2). Compound 6, the fluorophenyl analog of 2, displayed similar potency to 2; however, 7 and ent-7, as structural isomers of 6 with the nitrogen shifted by one position, demonstrated a substantial loss of potency despite the retained amide functional group. The results in Table 2 suggested that the distance between the basic nitrogen and the amide hydrogen bond acceptor (HBA) may be a crucial parameter for the inhibition of the PrCP enzyme. The hypothesis of a linear pharmacophore embedded within 1 and 2 suggested that the compounds in Table 1 lost potency because of an expansion of the distance between the basic nitrogen and the HBA (e.g. 1 vs 4), while the compounds in Table 2 lost potency because of a contraction of the basic nitrogen-HBA distance (e.g. 6 vs 7). Combining the pyrazole bioisostere shown in Table 1
1659
T. H. Graham et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1657–1660 Table 4 The results of shifting the aryl substituent to remove the stereogenic centers Compd
PrCP IC50 h,ma (nM) X
18 (X = Cl) 19 (X = F)
105,— 30, 23
HN N
N
F
20
7.0, 3.6 N N Me
N
F
300,—
21 N N
N
Me
F
N
3.3, 0.7
22 N
N N Me
a Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse.
and the piperidine core shown in Table 2 would balance the linear distance between the basic nitrogen and the HBA. The changes would appear as a one bond frameshift of the original pharmacophore that is in 1 and 2. Importantly, envisioning this model requires selecting the proper tautomeric form of the pyrazole. To test the heuristic model, the racemic pyrazole hybrid 8 was prepared and demonstrated a greater than 10-fold improvement in potency (Table 3). Replacing the p-chlorophenyl with a phenyl restored the potency to the level exhibited by the initial leads 1 and 2. Thus, moving the nitrogen to an endocyclic position and exchanging the chlorine atom for a hydrogen atom afforded potent inhibitors of PrCP that incorporated a pyrazole as a replacement of the amide in the lead structures. In addition, the results for the stereoisomers of 9 suggested that the exact positioning of the p-fluorophenyl aromatic relative to the core structure is less influential on the PrCP inhibition than the stereogenicity at the pyrazole center.22,23 In general, additional geometric or electronic perturbations could also explain differences in binding affinities, but the
simplicity of the heuristic model guided the rapid design of new molecular entities during the lead optimization process. Having restored the potency with the pyrazole bioisostere in place, a brief SAR survey of the nitrogen substituent was conducted. Substitution on the b-carbon with a hydroxyl (10) lost 2to 4-fold potency relative to 9 and substitution with a ketone (11) resulted in a more substantial decrease in inhibitory activity. Removal of a methylene from 8 afforded analog 12 with a 2-fold loss in potency. The trifluoropropyl analog 13 lost 11- and 25-fold relative to 9. Trifluoroethyl analog 14 was substantially less potent. The progressive loss in potency for 13 and 14 is likely attributable to inductive effects reducing the basicity of the nitrogen rather than steric effects because the tert-butyl (15) and isopropyl (16) analogs were more potent.24 The results for compounds 13–16 gave additional credence to the importance of a basic nitrogen in the pharmacophore. Finally, the secondary amine 17 had an IC50 value of 1 lM. As mentioned earlier, the position of the p-fluorophenyl aromatic of 9 relative to the pharmacophore appears to be less restricted to variations. To test this hypothesis, a series of analogs, with the aromatic shifted by one carbon atom, was prepared (Table 4). Compound 18 lost potency relative to the closest analog 9; however, substitution of the chloro- with a fluoro-group afforded a modest improvement with compound 19. Methylation of the pyrazole afforded 20 and 21, which demonstrated a greater than 40-fold difference in potency against human PrCP and may indicate the preferred pyrazole tautomer of 19. Finally, replacement of the phenyl in 20 with a pyridyl afforded 22, which is the most potent compound in the series. The effect of the bioisosteric replacement on the mouse pharmacokinetics was also evaluated (Table 5). Both amides 1 and 2 demonstrated high plasma protein binding (PPB) with the morpholine in 1 imparting a measurable free-fraction. The pharmacokinetics of 1 and 2 reveal the differences in PPB because 1 has higher total clearance and a consequently shorter half-life compared to 2. Compounds 9 and 12, like compound 2, were highly plasma protein bound and the pharmacokinetics parameters are generally comparable to 2, with long half-life and improved oral bioavailability. Similarly, compounds 15 and 16 also had improved oral bioavailabilities relative to 1 and 2. Finally, compounds 18–22, with the aromatic group shifted by one carbon atom thus removing the stereogenic carbons, demonstrated high total clearances and modest oral bioavailabilities in the mouse. The results in Table 5 demonstrated that the pyrazole bioisostere can have mouse pharmacokinetic parameters that are comparable to those of the original amides 1 and 2.
Table 5 Mouse pharmacokinetic parameters for selected compounds PPBb Compd 1d 2 9e 12 15 16 18 19 20 22 a
a
Mouse pharmacokinetics (C57Bl/6)c
h
m
Cl (mL/min/kg)
Vd (L/kg)
t½ (h)
AUCNpo (lM h kg/mg)
%F
2
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl
Pyrazoles as non-classical bioisosteres in prolylcarboxypeptidase (PrCP) inhibitors Thomas H. Graham a,⇑, Min Shu a, Andreas Verras a, Qing Chen c, Margarita Garcia-Calvo b, Xiaohua Li c, JeanMarie Lisnock d, Xinchun Tong c, Elaine C. Tung c, Judyann Wiltsie d, Jeffrey J. Hale a, Shirly Pinto b, Dong-Ming Shen a a
Department of Medicinal Chemistry, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA Department of Metabolic Disorders, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA Department of Drug Metabolism, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA d Department of In Vitro Sciences, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065-0900, USA b c
a r t i c l e
i n f o
Article history: Received 12 January 2014 Revised 22 February 2014 Accepted 25 February 2014 Available online 6 March 2014 Keywords: Non-classical bioisostere Pyrazole Amide Enzyme inhibitor Prolylcarboxypeptidase (PrCP) Angiotensinase C
a b s t r a c t Bioisosteres are integral components of modern pharmaceutical research that allow structural optimization to maximize in vivo efficacy and minimize adverse effects by selectively modifying pharmacodynamic, pharmacokinetic and physicochemical properties. A recent medicinal chemistry campaign focused on identifying small molecule inhibitors of prolylcarboxypeptidase (PrCP) initiated an investigation into the use of pyrazoles as bioisosteres for amides. The results indicate that pyrazoles are suitable bioisosteric replacements of amide functional groups. The study is an example of managing bioisosteric replacement by incorporating subsequent structural modifications to maintain potency against the selected target. A heuristic model for an embedded pharmacophore is also described. Ó 2014 Elsevier Ltd. All rights reserved.
Classical bioisosteres encompass the structural changes to a molecule that include substituting atoms and the associated valences or exchanging of ring equivalents. In contrast, non-classical bioisosteres include the replacement of entire functional groups, exchanging cyclic and non-cyclic forms and the inversion of functional groups (retroisosteres).1 Modern pharmaceutical research relies on bioisosteres to maximize in vivo efficacy and minimize undesired effects by selectively modifying pharmacodynamic, pharmacokinetic and physicochemical properties.2 A recent medicinal chemistry campaign that focused on discovering novel inhibitors of prolylcarboxypeptidase (PrCP) investigated the use of a pyrazole as a bioisostere for an amide. The study revealed that the pyrazole initially attenuated the in vitro potency; however, additional structural modifications restored in vitro potency. Prolylcarboxypeptidase (PrCP) (angiotensinase C) is a single chain serine peptidase that cleaves the amide bond between a C-terminal amino acid and a proline residue (i.e. peptide-Pro-XxxOH).3 Several hypotheses concerning the biological role of PrCP are based on the ability of PrCP to degrade bioactive peptides4,5
⇑ Corresponding author. Tel.: +1 732 594 2918. E-mail address: [email protected] (T.H. Graham). http://dx.doi.org/10.1016/j.bmcl.2014.02.070 0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.
Me
Me
O N
N O
O N
Ph N H
Ph 1
PrCP IC50 (h,m): 3.5, 5.0 nM
Cl
Ph Ph
2
PrCP IC50 (h,m): 2.5, 1.4 nM
Figure 1. Recently discovered structures of potent PrCP inhibitors.
as well as the extensive tissue distribution of PrCP.6,7 Studies have suggested that PrCP may be involved in cardiovascular control,8 metabolism regulation,9,10 inflammation induction,11 angiogenesis12 and chemotherapy resistance.13 The possibility for numerous therapeutic indications has resulted in the rapid development of small molecule inhibitors of PrCP.14 A program initiated to identify small molecule inhibitors of PrCP has produced several classes of potent inhibitors.15 The aminocyclohexanes, represented by 1 and 2 (Fig. 1) were recently discovered as potent inhibitors of PrCP.16,17 During lead-optimization studies, a bioisosteric replacement of the amide functional group was explored. Many heterocycles have been used as
1658
T. H. Graham et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1657–1660
Table 1 Initial attempts to incorporate a pyrazole bioisostere into the lead molecules
Table 3 Optimization of the lead with the pyrazole bioisostere F
PrCP IC50 h,ma (nM)
Compd
HN N
HN N
Me b
Cl
860,—
3
R
N
Compd
O
PrCP IC50 h,ma (nM)
R=
8b
HN N
Me
N
b
Cl
4
74, 20 Cl
940,—
3.5, 2.2 92,— 114,— 40, 24
9 4-epi-9 ent-9 4-epi-(ent-9)
N O
Br
OH
10
HN N
5
13, 5.9
270, 220
Cl
O
69, 30
11
N O Br
12b HN N
epi-5
13b
580
Cl
14b
N O
15b
a
Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse. b Compound was tested as the racemate.
16b
Cl
150,— F F
F F
40, 56 F
F
Me Me
1500,— Me
66, 180
Me
150,— Me
17b
bioisosteres of amides including oxazoles, thiazoles, triazoles, tetrazoles and oxadiazoles; however, the use of pyrazoles as amide bioisosteres, is less common.18 The present study was designed
Table 2 Data supporting the heuristic model of a linear pharmacophore Compd
PrCP IC50 h,ma (nM) F
O
6
10.4, 4.7 N
N H
Ph Ph
Cl
F
O
7
N N
990,— Ph Ph
Cl F
O
850,—
ent-7 N N
Ph Ph
Cl a Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse.
H
1000,—
a
Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse. b Compound was tested as the racemate.
to explore the possibility that a pyrazole can replace an amide functional group in a series of PrCP inhibitors.19 Replacement of the amide in 1 with a pyrazole afforded 3; however, the potency was significantly attenuated by the structural modification.20 Attempts to regain the potency with additional structural modifications afforded compounds 4 and 5 that also proved to have attenuated inhibitory activity. In addition, compounds 5 and epi-5 indicated that the relative stereochemistry of the morpholine substituent has minimal effect on the in vitro inhibition of PrCP.21 The data in Table 1 indicated that the incorporation of the pyrazole as an amide bioisostere resulted in a substantial loss in potency. An additional set of compounds suggested a possible reason for the decrease in activity (Table 2). Compound 6, the fluorophenyl analog of 2, displayed similar potency to 2; however, 7 and ent-7, as structural isomers of 6 with the nitrogen shifted by one position, demonstrated a substantial loss of potency despite the retained amide functional group. The results in Table 2 suggested that the distance between the basic nitrogen and the amide hydrogen bond acceptor (HBA) may be a crucial parameter for the inhibition of the PrCP enzyme. The hypothesis of a linear pharmacophore embedded within 1 and 2 suggested that the compounds in Table 1 lost potency because of an expansion of the distance between the basic nitrogen and the HBA (e.g. 1 vs 4), while the compounds in Table 2 lost potency because of a contraction of the basic nitrogen-HBA distance (e.g. 6 vs 7). Combining the pyrazole bioisostere shown in Table 1
1659
T. H. Graham et al. / Bioorg. Med. Chem. Lett. 24 (2014) 1657–1660 Table 4 The results of shifting the aryl substituent to remove the stereogenic centers Compd
PrCP IC50 h,ma (nM) X
18 (X = Cl) 19 (X = F)
105,— 30, 23
HN N
N
F
20
7.0, 3.6 N N Me
N
F
300,—
21 N N
N
Me
F
N
3.3, 0.7
22 N
N N Me
a Compounds were tested as the trifluoroacetate salt; values are based on one or two experiments, each in triplicate; h = human, m = mouse.
and the piperidine core shown in Table 2 would balance the linear distance between the basic nitrogen and the HBA. The changes would appear as a one bond frameshift of the original pharmacophore that is in 1 and 2. Importantly, envisioning this model requires selecting the proper tautomeric form of the pyrazole. To test the heuristic model, the racemic pyrazole hybrid 8 was prepared and demonstrated a greater than 10-fold improvement in potency (Table 3). Replacing the p-chlorophenyl with a phenyl restored the potency to the level exhibited by the initial leads 1 and 2. Thus, moving the nitrogen to an endocyclic position and exchanging the chlorine atom for a hydrogen atom afforded potent inhibitors of PrCP that incorporated a pyrazole as a replacement of the amide in the lead structures. In addition, the results for the stereoisomers of 9 suggested that the exact positioning of the p-fluorophenyl aromatic relative to the core structure is less influential on the PrCP inhibition than the stereogenicity at the pyrazole center.22,23 In general, additional geometric or electronic perturbations could also explain differences in binding affinities, but the
simplicity of the heuristic model guided the rapid design of new molecular entities during the lead optimization process. Having restored the potency with the pyrazole bioisostere in place, a brief SAR survey of the nitrogen substituent was conducted. Substitution on the b-carbon with a hydroxyl (10) lost 2to 4-fold potency relative to 9 and substitution with a ketone (11) resulted in a more substantial decrease in inhibitory activity. Removal of a methylene from 8 afforded analog 12 with a 2-fold loss in potency. The trifluoropropyl analog 13 lost 11- and 25-fold relative to 9. Trifluoroethyl analog 14 was substantially less potent. The progressive loss in potency for 13 and 14 is likely attributable to inductive effects reducing the basicity of the nitrogen rather than steric effects because the tert-butyl (15) and isopropyl (16) analogs were more potent.24 The results for compounds 13–16 gave additional credence to the importance of a basic nitrogen in the pharmacophore. Finally, the secondary amine 17 had an IC50 value of 1 lM. As mentioned earlier, the position of the p-fluorophenyl aromatic of 9 relative to the pharmacophore appears to be less restricted to variations. To test this hypothesis, a series of analogs, with the aromatic shifted by one carbon atom, was prepared (Table 4). Compound 18 lost potency relative to the closest analog 9; however, substitution of the chloro- with a fluoro-group afforded a modest improvement with compound 19. Methylation of the pyrazole afforded 20 and 21, which demonstrated a greater than 40-fold difference in potency against human PrCP and may indicate the preferred pyrazole tautomer of 19. Finally, replacement of the phenyl in 20 with a pyridyl afforded 22, which is the most potent compound in the series. The effect of the bioisosteric replacement on the mouse pharmacokinetics was also evaluated (Table 5). Both amides 1 and 2 demonstrated high plasma protein binding (PPB) with the morpholine in 1 imparting a measurable free-fraction. The pharmacokinetics of 1 and 2 reveal the differences in PPB because 1 has higher total clearance and a consequently shorter half-life compared to 2. Compounds 9 and 12, like compound 2, were highly plasma protein bound and the pharmacokinetics parameters are generally comparable to 2, with long half-life and improved oral bioavailability. Similarly, compounds 15 and 16 also had improved oral bioavailabilities relative to 1 and 2. Finally, compounds 18–22, with the aromatic group shifted by one carbon atom thus removing the stereogenic carbons, demonstrated high total clearances and modest oral bioavailabilities in the mouse. The results in Table 5 demonstrated that the pyrazole bioisostere can have mouse pharmacokinetic parameters that are comparable to those of the original amides 1 and 2.
Table 5 Mouse pharmacokinetic parameters for selected compounds PPBb Compd 1d 2 9e 12 15 16 18 19 20 22 a
a
Mouse pharmacokinetics (C57Bl/6)c
h
m
Cl (mL/min/kg)
Vd (L/kg)
t½ (h)
AUCNpo (lM h kg/mg)
%F
2
