Piperidinyl-2-phenethylamino inhibitors of DPP-IV for the treatment of Type 2 diabetes

Article abstract of DOI:10.1016/j.bmcl.2009.02.099

A highly ligand efficient lead molecule was rapidly developed into a DPP-IV selective candidate series using focused small library synthesis. A significant hurdle for series advancement was genetic safety since some agents in this series impaired chromoso

Full text of DOI:10.1016/j.bmcl.2009.02.099

Bioorganic & Medicinal Chemistry Letters 19 (2009) 2220–2223
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Piperidinyl-2-phenethylamino inhibitors of DPP-IV for the treatment of Type 2
diabetes
*
John W. Benbow , Kim A. Andrews, Jiri Aubrecht, David Beebe, David Boyer, Shawn Doran,
Michael Homiski, Yu Hui, Kirk McPherson, Janice C. Parker, Judith Treadway,
Maria VanVolkenberg, William J. Zembrowski
Pfizer Global Research and Development, Groton Laboratories, 558 Eastern Point Road, Groton, CT 06340, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
A highly ligand efficient lead molecule was rapidly developed into a DPP-IV selective candidate series
using focused small library synthesis. A significant hurdle for series advancement was genetic safety
since some agents in this series impaired chromosome division that was detected using the in vitro
micronucleus assay. A recently developed high-throughput imaging-based in vitro micronucleus assay
enabled the identification of chemical space with a low probability of micronucleus activity. Advanced
profiling of a subset within this space identified a compound with a clean safety profile, an acceptable
human DPP-IV inhibition profile based on the rat PK/PD model and a projected human dose that was suit-
able for clinical development.
Received 21 January 2009
Revised 24 February 2009
Accepted 25 February 2009
Available online 28 February 2009
Keywords:
Dipeptidyl peptidase-IV
DPP-IV
Ó 2009 Elsevier Ltd. All rights reserved.
Phenethylamine
An increasing incidence of obesity has, in part, fueled the rise in
new Type 2 Diabetes Mellitus (TTDM) diagnoses and continues to
make the treatment of TTDM a critical global health care issue.1
Present pharmaceutical therapy fails to provide adequate glycemic
control and many patients are unable to achieve their targeted
plasma glucose levels through available regimens. Inhibition of
the serine protease dipeptidyl peptidase-IV (DPP-IV) has recently
emerged as an effective biological target for the treatment of
TTDM.2 Two of the targets for DPP-IV are the incretin hormones
glucagon-like peptide-1 (GLP1) and gastrointestinal inhibitory
peptide (GIP), bioactive molecules that are secreted by the gut in
response to food intake.3 Binding of GLP1 to the pancreas and pan-
creatic b-cells leads to insulin secretion and results in enhanced
glucose disposal. GLP1 has also been reported to provide benefits
to the pancreas through improved beta cell function and possible
regeneration.4 From a safety standpoint, DPP-IV inhibition offers
a significantly reduced risk of hypoglycemia because the endoge-
nous insulin secretagogue (GLP1) is generated only in response
to a glucose stimulus. A first in class DPP-IV inhibitor JanuviaÒ
(MK-0431, 1) has been approved and other agents are in review
at the FDA [GalvusTM (LAF-237, 2); Alogliptin (SYR-322, 3)]
(Fig. 1). We were interested in finding additional novel small
molecular weight inhibitors of DPP-IV that lacked the 2-cyanopyrr-
olidide as a back-up to our clinical candidate, PF-00734200 (4).
This account describes the discovery of a novel piperidinyl-2-phen-
ethylamino series of DPP-IV inhibitors and the tactics used to
manage pre-clinical safety risks to identify promising clinical
candidates.
The laboratory objective of the back-up effort was a once-a-day
oral therapy that would provide P80% inhibition of DPP-IV for at
least 8 h in humans. Compounds would have demonstrated
>100-fold selectivity against DPP2 and DPP8 and have rat PK prop-
erties that would translate into acceptable human PK parameters.
The successful candidate must also have no QTc prolongation risk,
as judged by a 300x therapeutic index against hERG (hERG IC50
/
DPP-IV Ki P 300) (Ki = IC50/1.66) and clean profile in the
a
in vitro genetic toxicity assays (Bioluminescent Salmonella Assay
and in vitro micronucleus assay).5
N
N
F3C
N
F
F
N
N
N
H
CN
HO
O
O
NH2
F
1
2
O
N
N
N
N
H2N
N
O
N
N
N
O
N
H
F
F
NC
4
3
* Corresponding author.
E-mail address: orgodoc@comcast.net (J.W. Benbow).
Figure 1. Advanced DPP-IV inhibitor clinical candidates.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.02.099

J. W. Benbow et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2220–2223
2221
Table 1
DPP-IV inhibition of C1-substituted 1-amino-2-(2,4,5-trifluoro)phenylethanes, 5
F
F
R1
R2 NH
F
5
Compounds
R1
R2
DPP-IV inhibition IC50
,
l
Ma
Ligand efficiency (LE)b
5a
5b
5c
5d
5e
5f
5g
5h
5i
H
i-Pr
H
H
H
H
H
>30
1.93
3.58
>30
25.9
>30
3.25
1.42
>30
1.97
0.272
0.27
0.52
0.41
<0.32
0.39
<0.39
0.34
0.36
<0.28
0.43
0.50
0.36
Cyclohexyl
Cyclohexyl-CH2–
t-Bu
–(CH2)4–
4-CH3O2CPh–
1-(R)-4-CH3O2CPh–
1-(S)-4-CH3O2CPh–
1-(R)-4-Pyridyl
1-(R)-4-Piperidinyl
N-Boc-piperidin-4-yl
H
H
H
H
H
H
5j
5k
5l
a
Values are means of three experiments.
b
Ligand efficiency is measured from the following equation: LE = À1.4 log Ki/# of heavy atoms. Ki was determined from the IC50 values using the Cheng–Prussof equation:
Ki = IC50/(1 + [S]/Km) where [S] = 50 M and Km = 57 M.
l
l
The 2,4,5-trifluorophenethylamino substructure was an estab-
lished proline amide replacement in the S1 pocket and structural
information suggested that the S2 pocket was sufficiently large
and flexible enough to accommodate a variety of drug-like motifs.6
As part of a lead-hopping effort around the pharmacophore sub-
structure, a series of simple phenethylamine derivatives were made
and screened. While the un-substituted parent structure 5a was an
inefficient inhibitor of DPP-IV action, simple alkyl (5b and c) pro-
duced highly ligand efficient molecules (Table 1). A synthetic strat-
egy using accessible chemical space was employed to probe the
area directly outside of the S1 pocket and enabled the discovery of
a viable, novel lead series. Substitution near the primary amine dis-
rupted binding to the critical glutamate residues on the floor of S1
pocket: quaternary substitution alpha to the phenethylamine chain
(5e), secondarynitrogenintheformof asaturatedheterocycle (5f)or
the extension of the alkyl substituent from the main backbone (5d)
was not tolerated. However, non-polar, racemic aryl substitution
5g on the ethyl chain was tolerated and asymmetric HPLC separation
of the 4-methylbenzoate analogs established that the preferred ami-
no stereochemistry was (R). The S2 pocket also tolerated basic polar-
ity (4-pyridyl (5j) and 4-piperidinyl (5k)) and the N-Boc analog (5l)
provided an opportunity to effectively explore this region of the
binding site using a variety of functional groups that could modulate
the physical chemical properties of the core.
with (S)-hydroxypinanone 6 to form the corresponding imine
and diastereoselective alkylation with 2,4,5-trifluorobenzyl chlo-
ride proceeded in good yield to afford 7 with excellent diastereose-
lectivity (>98% de). Removal of the imine through trans-oximation
provided the primary amine intermediate that was readily con-
verted into the N-Boc derivative 8. Hydrogenation of the pyridine
heterocycle (Pt2O, AcOH) afforded the key piperidine intermediate
9 that could be recrystallized from EtOH to afford the N-Boc piper-
idine in good yield.
Having identified a suitable template, medium speed synthesis
methods were used to produce a diverse set of analogs to probe the
DPP selectivity SAR, generate benchmark data for in silico ADME
modeling and to survey the behavior in the secondary safety
screens. The 2-(2,4,5-trifluoro)phenethylamino moiety effectively
delivered the required selectivity over the DPP8 isoform
(IC50 = 5–30 lM) regardless of the substitution elsewhere in the
molecule. Selectivity over the DPP2 isoform could be achieved by
modulating the potency against DPP-IV and DPP2 in tandem:
piperidine-1-yl substituents with rotational flexibility such as
amides, sulfonamides and alkyl chains improved DPP-IV inhibition
and generally decreased the activity against DPP2. The DPP-IV IC50
values in conjunction with the in vitro ADME screening data could
be used to filter compounds for rat PK/PD experiments. Criteria
used for selection were: DPP-IV IC50 <100 nM; clearance in rat
and human microsomes of <19 or <5 mg/mL/min, respectively,
and Apical?Basal (AB) apparent permeability >10 cmÀ1/s in MDCK
cells with negligible efflux (BA/AB <3); single point dofetilide inhi-
An efficient route to the library-enabled template 9 was devised
using the asymmetric alkylation method of Shioiri (Scheme 1).7
The commercially available 4-aminomethylpyridine was reacted
bition <20% at 10 lM compound. In addition, safety data was gen-
erated on a subset of analogs that indicated an unacceptable level
of hERG inhibition and an increase incidence of micronuclei (aneu-
genic activity) (3/4) in the in vitro micronucleus assay.8 At this
stage, an assessment of this series suggested that compounds with
an acceptable selectivity profile and appropriate PK properties
could be identified (Table 2).
N
O
N
a, b
F
F
OH
OH
F
6
7
Because not all of the compounds in this series were positive in
the micronucleus assay, the activity was not believed to be specific
to the pharmacophore. However, the usual low-throughput assay
had insufficient capacity to handle the volume of samples needed
to de-risk this chemical series in a timely manner; the microscopic
scoring of the micronuclei is a laborious and time-consuming pro-
cess. A recently developed image analysis algorithm for micronuclei
scoring and implementation of automated microplate technologies
enabled the testing up to forty compounds at a time.9 The ability
N
c, d
H
N
e
HN
Boc
F
F
Boc
F
NH
F
F
F
9
8
Scheme 1. Asymmetric route to (R)-1-(piperidin-4-yl)-1-amino-2-(2,4,5-tri-
fluoro)phenylethane. Reagents and conditions: (a) 4-aminomethylpyridine,
BF3ÁEt2O, PhCH3; (b) n-BuLi, THF, <À20 °C then 2,4,5-F3BnCl; (c) NH2OH, EtOH,
heat; (d) Boc2O, 1,4-dioxane; (e) H2, Pt2O, EtOH, AcOH.

2222
J. W. Benbow et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2220–2223
Table 2
Table 4
DPP-IV inhibition, DPP2 selectivity, in vitro Human microsomal clearance and MDCK
DPP-IV inhibition, DPP2 selectivity, safety and rat PK/PD of C1-substituted 1-amino-2-
permeability of C1-substituted 1-amino-2-(2,4,5-trifluoro)phenylethane amides, 10
(2,4,5-trifluoro)phenylethanes
O
R1
N
F
F
R1
N
F
F
NH2
10
NH2
F
F
10
Compd R1
DPP-IV IC50
(Ki) (nM)
DPP2 IC50
M)
hERG IC50
(lM)
Rat PK/
PDc (h)
a
b
Compd R1
DPP-IV
IC50 (nM)
DPP2
Human CLb
M) (mL/min/kg)
MDCK ABc
(10À6 s)
a
(
l
IC50
(
l
N
O
F
10d
10e
10f
18
22
27
>30
>30
10.8
85
55
59
8–24
12–16
>24
10a
35
11,100
10,200
9110
<5.3
10.1
7.4
N
F
N
O
N
N
10b
26
19
<5.3
7.2
N
N
N
O
N
10c
17.5
N
NH2
N
N
10g
4.0 (2.1)
16.8
18
>20
a
N
Values are means of three experiments.
Values represent the disappearance of compound from cultured human liver
b
a
Values are means of three experiments.
For a description of the hERG patchclamp assay, please see: Fermini, B. and
Fossa, A. A. Nat. Rev. 2003, 2, 439.
b
microsomes.
c
Values represent the apical to basal migration of compound in a Madin–Darby
c
canine kidney cell line.
Animals were dosed at 5 mg/kg, p.o. with methyl-cellulose solutions through
oral gavage. Data reported shows the time that 50% inhibition of DPP-IV in plasma
was achieved. Ranges represent intra animal variability in the DPP-IV inhibition
profile.
to screen a larger number of compounds using this technology
enabled the identification of chemical sub-series with a predicted
low-probability for chromosomal aberration (<20% positive). The
first round of screening data demonstrated that various piperidine
head-groups could be differentiated. Heteroatom-substituted
indene amides (10f and g) attached through either the five- or six-
membered ring were largely inactive (Table 3) as were aryl or alkyl
sulfonamides, but the sulfonamides were not profiled further due
to poorer DPP-IV inhibition and sub-par ADME properties. Amino-
linked heteroaromatics, benzamide and benzylic substituents deliv-
ered positive results more frequently (P50% positive) and work on
these types of analogs was discontinued. A representative set of het-
eronaphthamide analogs (e.g., 10b) demonstrated a 50% probability
for chromosomal aberrations and work on these analogs was discon-
tinued. A smaller set of heteroarylamide analogs (e.g., 10c) gave
similar results and the set was expanded in the second submission.
These data supported further investment in heteroatom-substituted
indenamides and identified the six-membered heteroaromatic
amide series as viable chemical space. The high-throughput micro-
nucleus assay had enabled the identification of two chemical sub-
series with a low probability of genetic toxicity within six weeks.
The generation of this same data for 60 analogs using the manual
micronucleus assay would have taken 4–5 months, a considerable
savings in opportunity costs.
The SAR of the heteroatom-substituted indenamidoyl and
hetero-arylamidoyl sub-series was expanded through close-in ana-
log synthesis. Optimal affinity for DPP-IV and the desired selectiv-
ity over the DPP2 isoform could be achieved by introducing small
substituents ortho- to the amide carbonyl. Acceptable physical
property space had been maintained throughout this effort and a
small set of pre-candidates emerged that represented both chemi-
cal sub-series for profiling in the rat PK/PD model and the defini-
tive safety assays (Table 4). These agents demonstrated an
acceptable hERG window and all tested negative in the in vitro
micronucleus assay.
The rat PK/PD profile for 10g was sufficiently promising to
advance this compound for a dog PK evaluation. The ADME data
from these two species (rat and dog) allowed the projected human
PK parameters to be calculated (Table 5). The human dose projec-
tion10 from modeling suggested that a 15 mg dose of 10g would
provide 80% inhibition of DPP-IV for 8 h and >50% inhibition for
16 h in humans.10,11
In conclusion, the highly ligand efficient lead molecule 5k was
quickly developed into a candidate series using small focused
library synthesis and a recently developed higher-throughput
image analysis-based in vitro micronucleus screening assay. Ad-
Table 3
In vitro CHO micronucleus screening results for different sub-series of phenethylaminopiperidinyl DPP-IV inhibitors
Compound headgroup
Manuala INMN assay
1st imaging screen
2nd imaging screen
Total % Neg. (n)
+
À
+
À
+
À
Heteroaromatic
Benzamidoyl
1
1
2
3
2
4
1
2
3
0
0
10
5
3
0 (3)
1
0
0
4
3
1
50 (8)
88 (16)
55 (11)
75 (4)
86 (14)
25 (4)
Heteroatom-substituted indenamidoyl
Hetero-napthamidoyl
Sulfonamidoyl
Hetero-arylamidoyl
Benzylic
1
1
3
1
0
9
a
Results are reported as positive (+) or negative (À) based on a reading of the amount of aberrant chromosomes after 24 h treatment of CHO cells with compound at
varying concentrations.

J. W. Benbow et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2220–2223
2223
3. Flatt, P. R. et al Front. Biosci. 2008, 13, 3648.
4. Drucker, D. J. Endocrinology 2003, 144, 5145.
Table 5
Pharmacokinetic data of 10g from rat and dog and the projected human parameters
5. Bioluminescent Salmonella assay: (a) Aubrecht, J. et al Mutagenesis 2007, 22,
335; The in vitro micronucleus assay: (b) Sobol, Z. et al Mutation Res. 2007, 633,
80.
6. Over the course of the DPP-IV project, over 30 crystal structure solutions were
prepared, some of which have been published: (a) Wright, S. W. et al Bioorg.
Med. Chem. Lett. 2007, 17, 5638; (b) Corbett, J. W. et al Bioorg. Med. Chem. Lett.
2007, 17, 6707.
c
d
e
f
Species
CLp (ml/min/kg)
Vdss (L/kg)
t1/2 (h)
F (%)
fu
Rata
18.0
12.9
3.4
2.2
2.2
1.0
5.09
3.71
3.5
70
52
70
49
42
30
Dogb
Human (Projected)
a
Rats (n = 2) were dosed with 1 mg/kg (i.v.) and 5 mg/kg (p.o.) in saline and 5%
methyl cellulose, respectively.
Dogs (n = 2) were dosed with 1 mg/kg (i.v.) and 1 mg/kg (p.o.) in saline and 5%
methyl cellulose, respectively.
7. Irako, N. et al Tetrahedron 1995, 51, 12731.
8. The in vitro micronucleus assay detects the presence of extra-nuclear
b
chromatin (micronuclei) in cytoplasm as
a
consequence of either
chromosome loss (aneugenicity) or chromosome breakage (clastogenicity).
These effects can be differentiated via evaluating the kinetochore complex in
micronuclei: Schuler, M. et al Mutation Res. 1997, 392, 81.
c
Clearance of compound from plasma.
Volume of distribution calculated from the i.v. portion of the experiment.
Half-life was calculated from the i.v. portion of the experiment.
The unbound fraction of drug was measured using fresh plasma.
d
e
9. The higher throughput in vitro micronucleus assay utilizes standard microplate
technology, automated imaging using the Cellomics ArrayScan VTI (Thermo
Fisher Scientific, Rockford, IL) and an image analysis scoring algorithm
developed from an Image-Pro software platform (MediaCybernetics,
Bethesda, MD). The results generated on set of test compounds using this
assay compared favorably with those generated via the standard microscopic
analysis.
10. Human PK parameters were used to generate predicted plasma concentration
versus time profiles at varying doses of 10g (WinNonlin version 5.2, Pharsight
Corp.). The projected human plasma concentration data was then used to
predict the associated DPP-IV inhibition using the equation [i = 1/(1 + Ki/I)],
where ‘i’ is DPP-IV inhibition, ‘Ki’ is the value calculated from the in vitro EC50
value, and ‘I’ is the unbound plasma concentration.
f
vanced profiling in the rat PK/PD model and subsequent human
projections identified 10g as having an acceptable human DPP-IV
inhibition profile with a projected 15 mg/q.d. that was suitable
for a drug development candidate.11
References and notes
11. A low dose (<50 mg q.d.) was desired to lower the risk of idiosyncratic toxicity
in the clinic: Kalgutkar, A. S. et al Exp. Rev. Clin. Pharmacol. 2008, 1, 515.
1. (a) Flier, J. S. Cell 2004, 116, 337; (b) Hogan, P. et al Diabetes Care 2003, 26, 917.
2. Pei, Z. Curr. Opin. Drug Disc. Dev. 2008, 11, 512.