Discovery of (2S,4R)-1-(2-aminoacetyl)-4-benzamidopyrrolidine-2-carboxylic acid hydrochloride (GAP-134)13, an orally active small molecule gap-junction modifier for the treatment of atrial fibrillation

Article abstract of DOI:10.1021/jm801558d

Rotigaptide (3) is an antiarrhythmic peptide that improves cardiac conduction by modifying gap-junction communication. Small molecule gap-junction modifiers with improved physical properties were identified from a Zealand Pharma peptide library using phar

Full text of DOI:10.1021/jm801558d

908
J. Med. Chem. 2009, 52, 908–911
Chart 1. Antiarrhythmic Peptides AAP (1), AAP10 (2), and
Rotigaptide (3)
Discovery of (2S,4R)-1-(2-Aminoacetyl)-4-
benzamidopyrrolidine-2-carboxylic Acid
Hydrochloride (GAP-134)13, an Orally Active
Small Molecule Gap-Junction Modifier for the
Treatment of Atrial Fibrillation
John A. Butera,*,† Bjarne Due Larsen,# James K. Hennan,
Edward Kerns,† Li Di,† Asaf Alimardanov,†
Robert E. Swillo, Gwen A. Morgan, Kun Liu,
Qiang Wang, Eric I. Rossman, Rayomand Unwalla,†
Leonard McDonald,† Christine Huselton,‡ and
Jørgen S. Petersen#
Chemical Sciences, Wyeth Research, CN 8000, Princeton, New
Jersey 08543, Chemical Sciences, Wyeth Research, 401 N.
Middletown Road, Pearl RiVer, New York 10965, Chemical
Sciences, Drug Safety and Metabolism, and CardioVascular and
Metabolic Diseases, Wyeth Research, 500 Arcola Road,
CollegeVille, PennsylVania 19426, and Zealand Pharma A/S,
Smedeland 36 DK-2600 Glostrup, Denmark
second generation of enzymatically stable, efficacious gap-
junction modifiers.6 Compound 3 was evaluated clinically for
the iv treatment of VF.7 The goal of this study was to develop
a third generation of small molecule gap-junction modifiers
possessing the appropriate pharmaceutical properties required
for oral administration to treat chronic conditions such as AF.
The molecular target for this class of compounds remains
elusive, although it has been reported that 2 and 3 may modulate
the phosphorylation state of several serine residues near the
C-terminus of connexin-43 (Cx43), the transmembrane spanning
protein comprising the connexon hemichannel that ultimately
forms the pore of the gap-junction.7 Lacking knowledge of the
molecular target precluded the use of a high-throughput screen
to identify small molecule leads for optimization. An established
in vivo mouse arrhythmia model involving gap-junction com-
munication and uncoupling was utilized as the primary assay
for activity.8 In this model increased intracellular calcium
concentration leads to uncoupling of gap-junctions and second
degree cardiac conduction block. Compounds that maintain or
re-establish GJIC produce a delay in the time required to induce
cardiac conduction block.
ReceiVed December 9, 2008
Abstract: Rotigaptide (3) is an antiarrhythmic peptide that improves
cardiac conduction by modifying gap-junction communication. Small
molecule gap-junction modifiers with improved physical properties were
identified from a Zealand Pharma peptide library using pharmaceutical
profiling, established SAR around 3, and a putative pharmacophore
model for rotigaptide. Activity of the compounds was confirmed in a
mouse cardiac conduction block model of arrhythmia. Dipeptide 9f
(GAP-134) was identified as a potent, orally active gap-junction
modifier for clinical development.
The isolation of a peptidic compound from bovine atria that
restored rhythmicity in cultured cardiac myocytes led to the
discovery of AAPa (1, Chart 1).1,2 Subsequent mechanistic
studies on 1 revealed that it lacked direct effects on classical
ion channels, thereby fueling an interest in this new class of
antiarrhythmic agents.3 Dhein and co-workers reported on the
synthesis and characterization of a series of AAP analogues,
the most potent of which was AAP10 (2).4 The underlying
mechanism of action of 2 and its analogues reported in 1997
involved the re-establishment of gap-junction intercellular
communication (GJIC) in adjacent cardiomyocytes,5 thus re-
versing cardiac conduction slowing and electrical heterogeneity
thought to be responsible for arrhythmias. Inherent enzymatic
instability in first generation peptidic gap-junction modifiers 1
and 2 precluded their further development as useful antiarrhyth-
mic agents. The discovery of rotigaptide (3) as a retroinverso
analogue (reverse sequence of amino acids with concomitant
replacement of L-amino acids with unnatural D configuration)
of 2 represents a significant advance in the field and led to a
Using this assay, scientists at Zealand Pharma generated a
pharmacophore model for 3 based on the SAR of a hexapeptide
library generated during the optimization campaign for 3. The
SAR highlights for 3 are illustrated in Figure 1. Truncation and
amino acid substitution exercises on 3 showed that (i) the tandem
* To whom correspondence should be addressed. Address: Wyeth
Research, CN 8000, Princeton, NJ 08543. Phone: 732-274-4289. Fax: 732-
274-4129. E-mail: Buteraj@wyeth.com.
†
Chemical Sciences, Wyeth Research.
Zealand Pharma A/S.
#
Cardiovascular and Metabolic Diseases, Wyeth Research.
‡
Drug Safety and Metabolism, Wyeth Research.
a Abbreviations: AF, atrial fibrillation; VF, ventricular fibrillation; AAP,
antiarrhythmic peptide; AV, atrioventricular; GJIC, gap-junction intercellular
communication; Cx43, connexin-43.
Figure 1. SAR trends derived from a hexapeptide library of rotigaptide
analogues administered iv.
10.1021/jm801558d CCC: $40.75
2009 American Chemical Society
Published on Web 01/28/2009

Letters
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 4 909
Chart 2. Top Tier Hits from Focused Screen: Three Lys-Gly
Analogues (4-6), Two Asn-Tyr Analogues (7, 8), and Singleton
Gly-Pro Analogue (9f)
(CYP3A4, CYP2D6, and CYP2C9) at 3 µM. Clustering of
compounds allowed us to further reduce the screening set by
selecting representatives from each structural class, with a bias
toward smaller di- and tripeptides that adhered to the putative
pharmacophore model.
Figure 2. 1H NMR based conformational analysis of 3 shows a 1:3
ratio of the extended (A) vs horseshoe (B) conformations.
In the initial screening iteration, the 25 representative small
peptides were evaluated for their ability to prolong the time to
cardiac conduction block in a mouse CaCl2-induced arrhythmia
model.8,10 The compounds (in a volume of 100 µL/30 g mice)
were administered iv (at doses of 10-12, 10-11, 10-10, and 10-9
mol/kg) into the tail vein of anesthetized CD-1 mice equipped
with ECG electrodes. Infusion of CaCl2 solution (30 mg/mL,
0.2 (mL/min)/30 g mice) was initiated 3 min after drug or
vehicle administration. The dose-dependent time lag (∆T) to
onset of AV block was measured for each compound and
compared to vehicle effect and to the response seen with
hexapeptide 3 as a positive control. A prolongation of the time
to the onset of AV block was indicative of antiarrhythmic
activity and suggestive of improved GJIC and conduction
velocity.8,10
Six small peptides (Chart 2) whose potency and efficacy in
the mouse CaCl2 model were indistinguishable from hexapeptide
3 were ranked as top tier compounds for initial follow-up studies.
The data for prolongation of the time to conduction block for
4-9f and parent hexapeptide 3 are shown in Table 1. In a dose
dependent manner, small peptides 4-9f had a significant effect
on ∆T values after iv administration, typically prolonging the
time to AV block by ∼150 to over 200% versus control values.
To further study the three scaffolds of hits identified in the initial
screen, we selected 6 as a representative of the Lys-Gly scaffold,
7 as a representative of the Asn-Tyr scaffold, and Gly-Pro
singleton 9f for additional characterization. The graphical
representation of data for 6, 7, 9f, parent hexapeptide 3, and
vehicle control are shown in Figure 4.
Figure 3. Triage of small peptide library to generate initial screening
set for mouse CaCl2 model.
proline-hydroxy-proline motif was essential for activity, (ii)
the glycine-alanine section showed a fair degree of tolerance
for modification as long as proper spacing was maintained, and
(iii) the N-terminal tyrosine and C-terminal glycine were
essential for activity. Cyclization of the tyrosine and glycine
using one of several spacer amino acids gave rise to several
cyclic heptapeptides that were still able to prolong time to
conduction block in the mouse CaCl2 model. On the basis of
these SAR data, a hypothesis was put forward that the terminal
tyrosine and glycine units are required to be relatively close to
each other in three-dimensional space, probably facilitated by
the tandem proline units and one or several intramolecular
hydrogen bonds formed in the hexapeptide backbone.
Further evidence to support this putative pharmacophore
1
model was obtained from H NMR conformational analysis
(Figure 2) of 3, which showed two major conformations, the
“extended” and “horseshoe” conformations, in approximately
a 1:3 ratio. A hydrogen bond between the proline carbonyl and
the alanine NH stabilizes this conformation. On the basis of
the long-range NOE observed between the tyrosine methylene
protons and the alanine methyl group, the energy minimized
computer-generated model of conformer B is shown in the lower
part of Figure 2. This was consistent with a similar bias toward
the horseshoe conformation reported for 2 by Dhein and co-
workers.9
Compounds 6, 7, and 9f were also evaluated in the rat atrial
strip conduction velocity model.11 In this model, metabolic stress
(glucose and O2-free media) induces conduction velocity slowing
(measured with two microelectrodes) in rat atrial strips, presum-
Table 1. Effect of Tier 1 Compounds on Time to Cardiac Conduction
Block in Mice Receiving an Intravenous Infusion of Calcium Chloridea
To generate a focused screening compound library for initial
evaluation in the mouse CaCl2-induced arrhythmia model, the
Zealand Pharma small peptide library (∼340 compounds) was
triaged using pharmaceutical profiling data, calculated properties,
scaffold clustering, and adherence to the putative pharmacophore
model based on the hexapeptide 3. The 340 compounds were
subjected (Figure 3) to pharmaceutical profiling assays including
solubility at pH 7.4, PAMPA permeability at pH 7.4, rat
microsomal stability, rat plasma stability, and CYP450 inhibition
compd 10-12 mol/kg 10-11 mol/kg 10-10 mol/kg 10-9 mol/kg
4
5
6
7
8
9f
3
192 ( 22
124 ( 15
220 ( 25*
188 ( 29*
228 ( 26
162 ( 9*
199 ( 30*
177 ( 12
206 ( 19
133 ( 7
232 ( 24*
157 ( 8
207 ( 21*
207 ( 26*
151 ( 22
197 ( 14
128 ( 11
203 ( 20*
205 ( 15
148 ( 20*
152 ( 10*
158 ( 9
201 ( 30
164 ( 19*
187 ( 29*
136 ( 11
173 ( 11*
156 ( 15*
a Reported as % of control response, mean ( SEM; n ) 7-11.

910 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 4
Letters
Table 2. Rat PK Parameters for 6, 7, and 9f (n ) 3)
6
7
9f
iv (0.5 mg/kg)
Cl ((mL/min)/kg)
Vss (L/kg)
T1/2 (h)
31 ( 12.3
0.4 ( 0.4
0.3 ( 0.1
55.4 ( 7.8
0.9 ( 0.0
0.3 ( 0.0
22 ( 4
0.61 ( 0.28
0.8 ( 0.9
po (5 mg/kg)
rat F (%)
2
0.7
9
T1/2 (h)
Cmax (ng/mL)
Tmax (h)
1.5 (N ) 2)
0.5 (N ) 2)
1.9 ( 0.6
105 ( 26
0.5 ( 0
360 ( 123
28
31.5
0.4
36
0.75
71
AUC (ng ·h/mL)
Figure 4. Mouse CaCl2 data for 6, 7, 9f, and hexapeptide 3 after iv
administration.
It has been shown that 3 attenuates calcein dye uptake and cell
swelling by reducing permeabilization in cultured neonate rat
ventricular myocytes.14 Connexin hemichannels in the cell mem-
brane are thought to regulate dye retention in this model, thus
suggesting that gap-junctions and re-establishment of GJIC may
play a cardioprotective role. Compound 9f was evaluated for its
ability to reduce dye uptake. Figure 6 shows that 9f dose
dependently reduces dye uptake in cultured C6 glioma cells in a
manner similar to that of parent 3.
Figure 7 illustrates an overlay (using MULTIFIT15 routine within
SYBYL) of the minimized structure of 916 with the H NMR
1
derived horseshoe structure of 3.17-19 It is our working hypothesis
that the benzoyl moiety of 9f is mimicking the N-terminal tyrosine
phenyl group of 3 while the glycine portion of 9f might be
providing a similar hydrogen bond donor-acceptor local environ-
ment as the C-terminal glycine carboxamide of 3.
Figure 5. Compounds 6, 7, and 9f prevent metabolic stress-induced
conduction velocity slowing in rat atrial strips. Results are comparable
to those seen with hexapeptide 3.
Finally, 9f was evaluated for in vivo efficacy in the mouse CaCl2
model after oral administration. For doses of 5-20 mg/kg po, 9f
significantly prolonged (Figure 8) the time to conduction block
(∼200% versus vehicle control; n ) 6-12) in mice after the
infusion of CaCl2. Hexapeptide 3, at 10 mg/kg, was inactive after
oral administration, consistent with its poor oral bioavailability.
ably due to the electrical heterogeneity that follows after the
uncoupling of gap-junctions under these conditions. A com-
pound that re-establishes GJIC would prevent stress-induced
conduction velocity slowing in this model. At 10 nM, 6, 7,
and 9f prevented (Figure 5) the conduction velocity slowing
seen in the control tissue after induction of metabolic stress.
At this concentration, the compounds did not affect atrial
contractility. The effects were identical to those observed
with hexapeptide 3.
Scheme 1. Synthesis of 9fa
Since the oral bioavailability of 3 in multiple species was
negligible (unpublished data), a key focus of the follow-on
campaign was to identify equipotent compounds with enhanced
PK properties that could have potential to treat chronic AF.
Preliminary rat PK studies were performed on the three
prototypes 6, 7, and 9f, and the data are shown in Table 2. With
an oral bioavailability of 9% in the rat, 9f was selected for
additional characterization and was found to exhibit 21% oral
bioavailability in the dog (unpublished results; for additional
data suggestive of oral bioavailability of 9f in dog, see ref 12).
Compound 9f did not demonstrate permeability characteristics
in pharmaceutical profiling, thereby suggesting uptake by active
transport mechanisms.
a Reaction conditions: (i) PhCOCl, EtOAc, H2O, NaHCO3 (94%); (ii)
HCl, Et2O (99%); (iii) Boc-Gly-OH, EtNCN(CH2)3NMe2 ·HCl, HOBT
(74%); (iv) NaOH, MeOH (93%); (v) HCl (88%).
Compound 9fwas resynthesized according to the route
depicted in Scheme 1.13 (2S,4R)-1-tert-Butyl 2-methyl 4-ami-
nopyrrolidine-1,2-dicarboxylate (9a) was benzoylated in the
presence of NaHCO3 to give 9b which upon deprotection in
HCl afforded aminoproline derivative 9c. This compound was
reacted with Boc-Gly-OH under standard peptide coupling
protocol to give dipeptide 9d. Saponification of the methyl ester
followed by treatment with HCl to remove the Boc group
afforded 9f as a crystalline HCl salt. The sequence was highly
efficient (overall yield ) 56%), amenable for kilogram-scale,
and was performed with no chromatography.
Figure 6. Dye uptake data for 3 and 9f in C6 glioma cells.

Letters
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 4 911
References
(1) (a) Aonuma, S.; Koama, Y.; Akai, K.; Komiyama, Y.; Nakajima, S.;
Wakabayashi, M.; Makino, T. Studies on heart. XIX. Isolation of an
atrial peptide that improves the rhythmicity of cultured myocardial
cell clusters. Chem. Pharm. Bull. 1980, 28, 3332–3339. (b) Aonuma,
S.; Koama, Y.; Akai, K.; Iwasaki, S. Studies on heart. XX. Further
effects on bovine ventricle protein (BVP) and antiarrhythmic peptide
(AAP) on myocardial cells in culture. Chem. Pharm. Bull. 1980, 28,
3340–3346.
(2) (a) Dikshit, M.; Srivastava, R.; Kundu, B.; Mathur, K. B.; Kar, K.
Antiarrhythmic and antithrombotic effect of antiarrhythmic peptide
and its synthetic analogs. Indian J. Exp. Biol. 1988, 26, 874–876. (b)
Kohama, Y.; Okimoto, N.; Mimura, T.; Fukaya, C.; Watanabe, M.;
Yokoyama, K. A new antiarrhythmic peptide, N-3-(4-hydroxyphenyl)
propionyl Pro-Hyp-Gly-Ala-Gly. Chem. Pharm. Bull. 1987, 35, 3928–
3930.
(3) Argentieri, T.; Cantor, E.; Wiggins, J. R. Antiarrhythmic peptide has
no direct cardiac actions. Experientia 1989, 45, 737–738.
(4) Dhein, S.; Manicone, N.; Muller, A.; Gerwin, R.; Ziskoven, U.;
Irankhahi, A.; Minke, C.; Klaus, W. A new synthetic antiarrhythmic
peptide reduces dispersion of epicardial activation recovery interval
and diminishes alterations of epicardial activation patterns induced
by regional ischemia. A mapping study. Naunyn-Schmiedeberg’s Arch.
Pharmacol. 1994, 350, 174–184.
(5) (a) Mu¨ller, A.; Schaefer, T.; Linke, W.; Tudyka, T.; Gottwald, M.;
Klaus, W.; Dhein, S. Actions of the antiarrhythmic peptide AAP10
on intercellular coupling. Naunyn-Schmiedeberg’s Arch. Pharmacol.
1997, 356, 76–82. (b) Mu¨ller, A.; Gottwald, M.; Tudyka, T.; Linke,
W.; Klaus, W.; Dhein, S. Increase in gap junction conductance by an
antiarrhythmic peptide. Eur. J. Pharmacol. 1997, 327, 65–72.
(6) Haugan, K.; Petersen, J. S. Gap junction-modifying antiarrhythmic
peptides: therapeutic potential in atrial fibrillation. Drugs Future 2007,
32, 245–260.
1
Figure 7. Pharmacophore overlay of 9f (magenta) with the H NMR
derived putative bioactive conformer of 3 (green). Atoms used in
MULTIFIT alignment are numbered: (1) centroids of benzoyl moiety
of 9f and N-tyrosinephenyl group of 3; (2, 3) the glycine portion of 9f
with the C-terminal carboxamide of 3.
(7) Kjølbye, A. L.; Haugan, K.; Hennan, J. K.; Petersen, J. S. Pharma-
cological modulation of gap junction function with the novel compound
rotigaptide: a promising new principle for prevention of arrhythmias.
Basic Clin. Pharmacol. Toxicol. 2007, 101, 215–230.
(8) Lynch, J. J.; Rahwan, R. G.; Witiak, D. T. Effects of 2-substituted
3-dimethylamino-5,6-methylenedioxyindenes on calcium-induced ar-
rhythmias. J. CardioVasc. Pharmacol. 1981, 3, 49–60.
(9) Grover, R.; Dhein, S. Spatial structure determination of antiarrhythmic
peptide using nuclear magnetic resonance spectroscopy. Peptides 1998,
19, 1725–1729.
(10) Kjølbye, A. L.; Knudsen, C. B.; Jepsen, T.; Larsen, B. D.; Petersen,
J. S. Pharmacological characterization of the new stable antiarrhythmic
peptide analog Ac-D-Tyr-D-Pro-D-Hyp-Gly-D-Ala-Gly-NH2 (ZP123):
in vivo and in vitro studies. J. Pharmacol. Exp. Ther. 2003, 306, 1191–
1199.
(11) Haugan, K.; Olsen, K. B.; Hartvig, L.; Petersen, J. S.; Holstein-Rathlou,
N. H.; Hennan, J. K.; Nielsen, M. S. The antiarrhythmic peptide analog
ZP123 prevents atrial conduction slowing during metabolic stress.
J. CardioVasc. Electrophysiol. 2005, 16, 537–545.
(12) Laurent, G.; Leong-Poi, E.; Mangat, I.; Moe, G.; Hu, X.; So, P.P.-S.;
Tarulli, E.; Ramadeen, A.; Rossman, E. I.; Hennan, J. K.; Dorian, P.
Effects of chronic gap junction conduction-enhancing antiarrhythmic
peptide GAP-134 administration on experimental atrial fibrillation in
dogs. Circulation, in press.
(13) Larsen, B. D.; Petersen, J. S.; Haugan, K. J.; Butera, J. A.; Hennan,
J. K.; Kerns, E. H.; Piatnitski, E. L. Preparation of Peptidomimetics,
Especially Modified Lysine-Mimetic Compounds, as Antiarrhythmic
Agents. U.S. Pat. Appl. 0149460-A1, 2007.
(14) Clarke, T. C.; Thomas, D.; Petersen, J. S.; Evans, W.; Martin, H.;
Patricia, E. M. The antiarrhythmic peptide rotigaptide (ZP123)
increases gap junction intercellular communication in cardiac myocytes
and HeLa cells expressing connexin 43. Br. J. Pharmacol. 2006, 147,
486–495.
(15) SYBYL, version 6.7; Tripos Associates: St. Louis, MO, 2000.
(16) MacroModel; Schrodinger, LLC: Portland, OR.
(17) Nilges, M.; Kuszewski, J.; Brunger, A. T. Computational Aspects of
the Study of Biological Macromolecules by NMR; Hoch, J. C., Ed.;
Plenum Press: New York, 1991.
Figure 8. Mouse CaCl2 data for 9f and hexapeptide 3 after oral
administration.
The bell shape dose response observed in this model has been
observed previously with other antiarrhythmic compounds includ-
ing verapamil and amiodarone and is likely model specific.10
In summary, we have reported our efforts to identify the first
orally bioavailable small molecule gap-junction modifier. Screening
of a focused subset of the Zealand Pharma small peptide library
resulted in the identification and characterization of three small
molecule chemotypes that displayed the same potency and efficacy
as parent hexapeptide 3. After demonstrating oral bioavailability
in dog (21%), 9f ((2S,4R)-1-(2-aminoacetyl)-4-benzamidopyrroli-
dine-2-carboxylic acid hydrochloride, GAP-134) was selected for
additional studies and was shown to re-establish GJIC, prevent
conduction velocity slowing in metabolically stressed atrial tissue,
and reduce dye uptake and cell swelling in C6 glioma cells. The
compound significantly prolonged the time to AV block in the
mouse CaCl2 arrhythmia model after oral administration. Com-
pound 9f was found to have a highly desirable in vitro and in vivo
safety profile including (1) lack of direct effects on membrane ion
currents, hERG inhibition, and CYP450 liabilities, (2) negative
AMES result, (3) a benign NovaScreen profile, and (4) no effects
on systemic hemodynamics.20 On the basis of these and other data
to be reported in due course, 9f was selected as a development
compound and is currently in phase 1 clinical trials for the treatment
of AF.
(18) Kuszewski, J.; Nilges, M.; Brunger, A. T. Sampling and efficiency of
metric matrix distance geometry: a novel partial metrization algorithm.
J. Biomol. NMR 1992, 2, 33–56.
(19) InsightII, version 2000.0; Accelrys, Inc.: San Diego, CA.
(20) Rossman, E. I.; Liu, K.; Morgan, G. A.; Swillo, R. E.; Butera, J. A.;
Gruver, M.; Kantrowitz, J.; Feldman, H. S.; Haugan, K.; Petersen,
J. S.; Hennan, J. K. Effects of the gap junction modifier, GAP-134,
on conduction and atrial fibrillation/flutter inducibility and maintenance
in dogs. Circulation 2007, 116 (16, Suppl. S), 392.
Supporting Information Available: Procedures for the synthesis
and characterization of 9f, details of the in vitro and in vivo
1
biological protocols, and details of the H NMR and 13C NMR
conformational analysis and modeling. This material is available
free of charge via the Internet at http://pubs.acs.org.
JM801558D