Design and synthesis of potent amido- and benzyl-substituted cis-3-amino-4-(2-cyanopyrrolidide)pyrrolidinyl DPP-IV inhibitors

Article abstract of DOI:10.1016/j.bmcl.2007.10.063

The cis-3-amino-4-(2-cyanopyrrolidide)-pyrrolidine template has been shown to afford low nanomolar inhibitors of human DPP-IV that exhibit a robust PK/PD profile. An X-ray co-crystal structure of 5 confirmed the proposed mode of binding. The potent single

Full text of DOI:10.1016/j.bmcl.2007.10.063

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 17 (2007) 6707–6713
Design and synthesis of potent amido- and benzyl-substituted
cis-3-amino-4-(2-cyanopyrrolidide)pyrrolidinyl DPP-IV inhibitors
J. W. Corbett,* K. Dirico, W. Song, B. P. Boscoe, S. D. Doran, D. Boyer, X. Qiu,
M. Ammirati, M. A. VanVolkenburg, R. K. McPherson, J. C. Parker and E. D. Cox
Pfizer Global Research and Development, Eastern Point Road, Groton, CT 06340, USA
Received 21 August 2007; revised 12 October 2007; accepted 16 October 2007
Available online 22 October 2007
Abstract—The cis-3-amino-4-(2-cyanopyrrolidide)-pyrrolidine template has been shown to afford low nanomolar inhibitors of
human DPP-IV that exhibit a robust PK/PD profile. An X-ray co-crystal structure of 5 confirmed the proposed mode of binding.
The potent single digit DPP-IV inhibitor 53 exhibited a preferred PK/PD profile in preclinical animal models and was selected for
additional profiling.
Ó 2007 Published by Elsevier Ltd.
Type 2 diabetes mellitus (T2DM) is becoming increas-
ingly prevalent and globally has reached epidemic pro-
portions. It is currently estimated that more than 180
million people worldwide have the disease with an esti-
mated 1.1 million deaths in 2005.1Even more disturbing
are the projections that the number of affected people
will double by 2030. Type 2 diabetes is a chronic disease
that is exacerbated by a sedentary lifestyle and obesity,
and is characterized by abnormal glucose regulation,
with diabetics at increased risk of cardiovascular, eye,
kidney, and nerve damage. Type 2 diabetics are unable
to maintain adequate control of blood sugar through
the action of endogenously generated insulin, even
though they generally have functioning pancreatic islet
b-cells. The incretin hormone glucagon-like peptide-1
(GLP-1) stimulates the release of insulin from b-cells.
GLP-1 administration has been shown to have a benefi-
cial effect on islet b-cell function and on maintenance of
proper blood glucose levels without the induction of
hypoglycemia.2 However, GLP-1 is rapidly degraded
and inactivated by the serine protease dipeptidyl pepti-
dase IV (DPP-IV). Therefore, the discovery of safe,
tolerated DPP-IV inhibitors is predicted to provide
effective medicaments for the management of T2DM.
Figure 1. DPP-IV inhibitors currently in market or Phase III clinical
trials.
Several DPP-IV inhibitors have been tested in the clinic,
including multiple compounds that are now in Phase III
clinical trials (Fig. 1).3–5 Sitagliptin (1a, Januvia) has
been approved for use in Mexico and has received ap-
proval from the U.S. Food and Drug Administration.5
Structure-based drug design was used, in conjunction
with the published crystal structure of N-L-valylpyrroli-
dine, to propose the DPP-IV inhibitor scaffold
cis-3-amino-4-pyrrolididepyrrolidine 2 (Fig. 2). The
appropriately substituted pyrrolidine ring was predicted
to project the amide-backbone NH and amine interac-
tions in the same orientation as found in N-L-valylpyrro-
Keywords: Dipeptidyl peptidase IV; DPP-IV; Diabetes.
*
Corresponding author. Tel.: +1 860 686 0252; e-mail: jeffrey.w.
corbett@pfizer.com
0960-894X/$ - see front matter Ó 2007 Published by Elsevier Ltd.
doi:10.1016/j.bmcl.2007.10.063

6708
J. W. Corbett et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6707–6713
shown in Figure 4, where N-L-valylpyrrolidine is over-
layed in blue onto molecule 5 (in yellow).10 The amine
and pyrrolidide groups have the anticipated orientation
as was predicted by modeling. The pyrazine ring of the
quinoxalinyl-group forms favorable aromatic stacking
interactions with Phe357, but part of the benzene ring
binds polar atoms of Glu205, Glu206, Ser209, and
Arg358, suggesting further opportunities for exploring
enzyme–ligand interactions.
Initial SAR work then focused on preparing amide
derivatives using chiral template 2. Several of the more
potent human DPP-IV inhibitors are shown in Table 1.
Good selectivity was observed against a select panel of
other DPP enzymes with the most potent pyrrolidinyl-
amide analog, compound 7, having an IC50 against hu-
man DPP-IV of 30 nM.11
Figure 2. Synthesis of chiral cis-3-amino-4-(pyrrolidide)pyrrolidinyl
single enantiomers.
lidine.6,7 Modest DPP-IV inhibition (<200 nM) was
achieved with the achiral cis-aminopyrrolidide template
(data not shown), so an exhaustive SAR effort was
mounted. An asymmetric synthesis of the desired tem-
plate was effected using a published procedure and both
enantiomers of the template were prepared (Fig. 2).8 It
was found that 3R,4R-pyrrolidinyl template (2)
exhibited DPP-IV activity, as evidenced by the in vitro
activity of enantiomers 5 and 6 (Fig. 3).9 Diastereomeric
pairs of compounds 5 and 6 were not prepared.
The pharmacokinetics of compound 5 was evaluated in
Sprague–Dawley rats and had t1/2 of 1.7 0.5 h and
%F = 90.9 The relatively favorable PK profile of 5
provided further impetus to continue evaluating the
series. Hence, pyrrolidinyl-benzyl derivatives were
evaluated and compounds with human DPP-IV
IC50’s < 100 nM are shown in Table 2. Cursory analysis
of the benzyl derivatives indicated decreased DPP-IV
inhibitory potency and increased displacement of dofeti-
lide in a dofetilide binding assay12 (e.g., 32% and 9%
dofetilide binding inhibition for benzyl and amide ana-
logs 16 and 12, respectively). However, improved cell
permeability of the benzyl series relative to the amides
led to their continued investigation. The dofetilide bind-
ing assay was found to be a useful indicator of hERG
ion channel activity and we demonstrated a correlation
between the dofetilide displacement assay and the
hERG patch clamp assay for a subset of com-
pounds.13,14 The hERG/dofetilide correlation suggested
that <20% inhibition of dofetilide binding was desired in
order to minimize hERG activity.
A representation of the single crystal X-ray structure of
˚
5 in human DPP-IV in a 6 A binding site surface is
Analogs containing a 2-cyanopyrrolidide were prepared
using the synthetic scheme shown in Figure 5 in an effort
to obtain inhibitors with increased potency. The benzyl
bromides used to synthesize final products 25, 26, 28, 29,
and 31–34 were prepared following the method of Tana-
ka et al.15
Figure 3. Structure of chiral DPP-IV inhibitors 5 and 6.
Several of the resultant analogs exhibited a significantly
improved inhibitory profile (Table 3). Analysis of the
emerging SAR indicated that it was possible to obtain
potent DPP-IV inhibitors with >100-fold selectivity
relative to DPP-2 and -8. However, there was a trend to-
ward increased inhibition of dofetilide binding and
DPP-8 inhibition. Therefore, dofetilide binding was
identified as a property that required removal from a
target molecule. Compound 28 was selected for PK/
PD determination since it exhibited excellent DPP-IV
activity and low potential for hERG activity.13 Unfortu-
nately, maximal DPP-IV inhibition in Sprague–Dawley
rats was only 51% between 4 and 6 h post-dose (5 mg/
kg po). This pharmacodynamic effect was considered
suboptimal and may be attributed to modest oral bio-
availability (36 15%). Therefore, a careful analysis of
Figure 4. N-L-Valylpyrrolidine (blue) overlayed with compound 5
(yellow).

J. W. Corbett et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6707–6713
6709
Table 1. Biological activity of (3R,4R)-cis-3-amino-4-(pyrrolidide)pyrrolidinyl amide derivatives with human DPP-IV IC50 < 100 nM
Compound
Ar
Human DPP-4
IC50 (nM)
Human DPP-8
IC50 (lM)
Percent dofetilide binding
inhibition (10 lM)
7
8
9
30
38
40
27.4
26.4
15
Quinolin-6-yl
19
14
>30
10
11
12
13
14
Benzothiazol-6-yl
Isoquinolin-3-yl
Quinolin-3-yl
42
47
49
62
79
23.1
>30
>30
0.9
0.5
9
Benzothiaphen-2-yl
Naphthalen-2-yl
10.2
24.1
5
4
15
86
4.25
14
Table 2. Biological activity of (3R,4R)-cis-3-amino-4-pyrrolidinyl benzyl derivatives with human DPP-IV IC50 < 100 nM
Compound
Ar
R
Human DPP-4
IC50 (nM)
Human DPP-8
IC50 (lM)
Percent dofetilide binding
inhibitiona (10 lM)
16
Quinolin-3-yl
H
27
5.1
32
17
18
H
H
50
1.2
43
69
4.7
ND
19
20
21
22
3-Cyano-4-fluorophenyl
3,4-Difluorophenyl
p-Cyanophenyl
H
78
80
81
83
3.1
2.5
1.9
1.8
9
H
19
ND
13
H
CH3
p-Nitrophenyl
23
H
91
1.0
84
a ND, no data.
des-methyl analog (data not shown) had 27% inhibition
of dofetilide binding); (2) heterocyclic substitution of
p-heteroarylbenzyl compounds (e.g., triazolyl deriva-
tives 25 and 28); and (3) m-substituted benzyl analogs
(e.g., m-cyano-derivatives 19 and 27).
A series of compounds were subsequently synthesized to
test these hypotheses in an attempt to identify low single
digit nanomolar DPP-IV inhibitors possessing low inhi-
bition of dofetilide binding. The requisite benzyl bro-
mides were either commercially available or were
prepared as shown in Figure 6. Of note, the preparation
of 4-cyano-3-methylbenzyl bromide 37 required the
selective reduction of methyl ester 36, which was initially
Figure 5. Synthesis of 2-cyano-pyrrolidide analogs.
extant dofetilide inhibition binding data was undertaken
and multiple SAR trends were identified that correlated
with decreased displacement of dofetilide binding: (1)
a-methylbenzyl substitution (compound 22 had 13%
inhibition of dofetilide binding, while the corresponding

6710
J. W. Corbett et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6707–6713
Table 3. Biological activity of (3R,4R)-cis-3-amino-4-cyanopyrrolidide benzyl derivatives with human DPP-IV IC50 ꢁ 10 nM
Compound
Ar
R
H
Human DPP-4
IC50 (nM)
Human DPP-8
IC50 (nM)
Percent dofetilide binding
inhibition (10 lM)
24
p-Cyanophenyl
2.4
560
47
40
25
H
3.2
461
26
27
H
H
3.3
3.6
148
922
36
10
3-Cyano-4-fluoro
28
H
5.3
858
4
29
30
H
H
5.4
6.2
247
49
36
Quinolin-3-yl
3290
31
32
33
34
H
H
H
H
6.7
8.5
10
326
704
363
190
65
4
71
91
10
accomplished using LiAlH4/SiO2 following a published
procedure. 16 Considerable caution needs to be exercised
when using this protocol beyond low millimolar scale
since explosive generation of hydrogen gas occurs upon
mixing LiAlH4 with silica. A different reductive proce-
dure was subsequently used to prepare the desired ben-
zyl alcohols from methyl esters 36 and 42a,b.17
p-cyano derivative 24. Interestingly, diethyl analog 55
had decreased inhibition of dofetilide binding but exhib-
ited a >3-fold loss in DPP-IV potency relative to com-
pound 53.18 In addition, 57 had ꢀ4-fold increased
inhibition of dofetilide binding compared to 56 (62%
vs 17% inhibition of dofetilide binding). Other examples
of minor structural changes that decreased dofetilide
binding are found by comparing 49 (88%) with 51
(18%) and 57 (62%) with 25 (40%).
Benzyl-substituted compounds, shown in Table 4, affor-
ded potent DPP-IV inhibitors and exhibited substan-
tially improved dofetilide binding as was evident by
comparing 48 and 51 (7% and 18% inhibition of dofeti-
lide binding, respectively) with the corresponding des-
methyl analogs 25 and 24 (40% and 47% inhibition of
dofetilide binding, respectively). The 2-cyano-5-fluoro-
pyrrolidide of 32 did not produce a substantial increase
in DPP-IV potency, and actually had an adverse effect
on inhibition of dofetilide binding (see compound 52
in Table 4). m-Substituted benzyl derivatives 53–56 were
prepared and analogs 53 and 54 had excellent DPP-IV
inhibitory potency, with 53 exhibiting a ꢀ3.4-fold de-
creased dofetilide binding relative to the unsubstituted
Compounds 48 and 53 were selected for further evalua-
tion based upon their in vitro profiles: rat PK data are
summarized in Table 5. Triazolyl analog 48 had a longer
t1/2 but a lower %F than 53. Maximal DPP-IV inhibition
in rats (5 mg/kg po dose) was more robust for 53 than
for 48, with 53 inhibiting rat DPP-IV >90% beyond
8 h post-dose while maximal DPP-IV inhibition with
48 began to diminish after ꢀ4 h.
The decision was made to progress compound 53 owing
to improved synthetic accessibility and since compound
48 had a more pronounced positive effect in an in vitro

J. W. Corbett et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6707–6713
Table 4. Biological activity of benzyl-substituted analogs
6711
Compound
R1
R2
Human DPP-4
IC50 (nM)
Human DPP-8
a
IC50 (nM)
Percent dofetilide binding
inhibition (10 lM)
48
H
3
586
7
49
50
H
H
3.4
3.5
258
88
7
1150
51
52
H
F
5.5
5.3
1040
332
18
17
53
54
H
H
1.3
1.1
460
346
14
21
55
56
H
4.7
417
3
H
H
46
ND
203
17
62
57
3.8
a ND, no data.
micronucleus assay. Compound 53 subsequently tested
negative in a human lymphocyte aberration (HLA)
assay implying less potential for cytotoxicity. Com-
pound instability in human and dog plasma, but not
in rat, became apparent during further analysis of 53
(Table 6). Compound 53 was also unstable in fresh
rat, dog, and human blood. The reasons for instability
were not apparent since 2-cyanopyrrolidides are not
inherently prone to human plasma instability. For
example, vildagliptin (1b) is currently in Phase III clini-
cal trials. The chemical reasons behind plasma instabil-
ity are unknown and a decision was made to not pursue
this series of DPP-IV inhibitors.
In summary, cis-3-amino-4-(2-cyanopyrrolidide)-pyrrol-
idines were shown to be a unique scaffold from which

6712
J. W. Corbett et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6707–6713
CO2R
Br
tion of compound 53 having an improved cytotoxicity
profile. Unexpected plasma instability led to the cessation
of work with this series of DPP-IV inhibitors.
CH2Br
CH2OH
PPh3, CBr4
CH2Cl2
1. CuCN
DMF, Δ
2.
H3B- N
Li+
CN
38
CN
37
35: R = H
MeOH
H2SO4
Acknowledgment
36: R = CH3
Br
The authors thank Bernard Fermini and Shuya Wang
for hERG patch clamp studies.
1) CbzCl
NaHCO3
THF/H2O
CO2Me a) NaNO2
HBr, 0 °C
b) Cu, reflux
2) (PPh3)2PdCl2
TEA, CO
MeOH/DMF
R
R
R
R
NH2
NH2
40a,b
References and notes
39a: R = Me
39b: R = Et
3) TMSI, CH3CN
1. Fact sheet No. 312. World Health Organization, Geneva,
2006.
CuCN
CO2Me
CO2Me 1)
CH2Br
H3B- N
Li+
DMF
Δ
2. Knudsen, L. B. J. Med. Chem. 2004, 47, 4128.
3. Idris, I.; Donnelly, R. Diabetes Obes. Metab. 2007, 9, 153.
4. Feng, J.; Zhang, Z.; Wallace, M. B.; Stafford, J. A.;
Kaldor, S. W.; Kassel, D. B.; Navre, M.; Shi, L.; Skene, R.
J.; Asakawa, T.; Takeuchi, K.; Xu, R.; Webb, D. R.;
Gwaltney, S. L., II J. Med. Chem. 2007, 50, 2297.
5. Gallwitz, B. Drugs Today 2007, 43, 13.
6. Rasmussen, H. B.; Branner, S.; Wilberg, F. C.; Wagt-
mann, N. Nat. Struct. Biol. 2003, 10, 19.
7. Initial scaffold design was conducted by E. Cox.
8. Hanselmann, R.; Zhou, J.; Ma, P.; Confalone, P. N. J.
Org. Chem. 2003, 68, 8739.
9. The dipeptidyl peptidase assays and rat pharmaco kinetic
and pharmacodynamic assays are described in: Wright, S.
W.; Ammirati, M. J.; Andrews, K. M.; Brodeur, A. M.;
Danley, D. E.; Doran, S. D.; Lillquist, J. S.; McClure, L. D.;
McPherson, R. K.; Orena, S. J.; Parker, J. C.; Polivkova, J.;
Qiu, X.; Soeller, W. C.; Soglia, C. B.; Treadway, J. L.;
VanVolkenberg, M. A.; Wang, H.; Wilder, D. C.; Olson, T.
V. J. Med. Chem. 2006, 49, 3068. DPP-IV assay results
varied by 15% between individual runs.
THF
2) PPh3
R
R
R
R
R
R
CBr4
CH2Cl2
Br
CN
42a,b
CN
43a,b
41a,b
CH2Br
R1
CO2Me
R1
CO2R
R1
N
HN
1) LiAlH4
THF
2) CBr4
Polymer-PPh3
CH2Cl2
N
R2
R2
DMF, K2CO3
R2
110 °C
N
N
F
N
N
44a,b: R = H
MeOH
H2SO4
N
47a,b
N
46a,b
45a,b: R = CH3
a: R1 = CH3, R2 = H
b: R1 = H, R2 = CH3
Figure 6. Synthesis of benzyl halides.
Table 5. Rat PK and PD (5 mg/kg po) for 48 and 53
Parameter
Compound
10. The atomic coordinates for the crystal structure of
compound 5 were deposited with the RCSB Protein Data
Bank with file name 2RIP.
48
53
Cmax (ng/mL)a
AUCinf (h ng/mL)a
CL (mL/min/kg)
t1/2 (h)a
1400 ( 264) 1785 ( 827)
7703 ( 1452) 3130 ( 1160)
11. All compounds tested in Tables 1–4 had DPP-2
IC50 > 10 lM (data not shown).
2.2 ( 2.6)
4.3 ( 2.6)
30.6 ( 5.7) 81 ( 19)
21.8 ( 3.0)
2.3 ( 0.4)
12. A filter binding assay with [3H]-dofetilide was used as
described in: Greengrass, P. M.; Stewart, M.; Wood, C.;
M. WO2003021271, 2003.
%Fa
Rat plasma protein binding (% free) 23 ( 0.3)
Maximal DPP-IV inhibition 86%
34 ( 0.3)
95%
13. Dofetilide binding assays have been shown to be predictive
for ranking compounds for their potential to block the
hERG ion channel, as described in Diaz, G. J.; Daniell,
K.; Leitza, S. T.; Martin, R. L.; Su, Z.; McDermott, J. S.;
Cox, B. F.; Gintant, G. A. J. Pharm. Toxicol. Methods
2004, 50, 187.
a Values are means of two experiments, standard deviation is given in
parentheses.
Table 6. Stability of compound 53 in rat, dog, and human plasma
14. A plot of log(hERG IC50) versus percent dofetilide activity
provides a correlation between dofetilide binding and
hERG IC50.
% Drug remaining
Time (h):
0
2
4
6
Species
Rat
Dog
Compound Percent dofetilide binding hERG patch
inhibition (10 lM)
100
100
100
100
93
85
49
99
98
84
27
117
83
clamp IC50 (nM)
Human
Control
13
127
5
32
53
51
54
25
0.241
4
>300,000
165,000
19,600
10,000
28,900
1800
108
14
18
potent DPP-IV inhibitors could be obtained. Careful
analysis of the SAR enabled the design of compounds that
exhibited low displacement of dofetilide in a binding assay
while human DPP-IV inhibition was improved. Concom-
itant risk management of micronucleus positive data, via
selection of a weakly positive micronucleus compound
and subsequent testing in a HLA assay, enabled the selec-
21
40
15. Tanaka, A.; Terasawa, T.; Hagihara, H.; Sakuma, Y.;
Ishibe, N.; Sawada, M.; Takasugi, H.; Tanaka, H. J. Med.
Chem. 1998, 41, 2390.

J. W. Corbett et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6707–6713
6713
16. Kamitori, Y.; Hojo, M.; Masuda, R. Tetrahedron Lett.
1983, 24, 2575.
Harrison, J.; Fuller, J. C.; Goralski, C. T.; Singaram, B.
Tetrahedron Lett. 1992, 33, 4533.
17. The modified reduction procedure utilized a lithium
pyrrolidinoborohydride, as described in: Fisher, G. B.;
18. Compound 53 had hERG IC50 ꢀ20 lM in a hERG patch
clamp assay.