Neladenoson Bialanate Hydrochloride: A Prodrug of a Partial Adenosine A1 Receptor Agonist for the Chronic Treatment of Heart Diseases

Article abstract of DOI:10.1002/cmdc.201700151

Adenosine is known to be released under a variety of physiological and pathophysiological conditions to facilitate the protection and regeneration of injured ischemic tissues. The activation of myocardial adenosine A₁ receptors (A₁Rs) has been shown to inhibit myocardial pathologies associated with ischemia and reperfusion injury, suggesting several options for new cardiovascular therapies. When full A₁R agonists are used, the desired protective and regenerative cardiovascular effects are usually overshadowed by unintended pharmacological effects such as induction of bradycardia, atrioventricular (AV) blocks, and sedation. These unwanted effects can be overcome by using partial A₁R agonists. Starting from previously reported capadenoson we evaluated options to tailor A₁R agonists to a specific partiality range, thereby optimizing the therapeutic window. This led to the identification of the potent and selective agonist neladenoson, which shows the desired partial response on the A₁R, resulting in cardioprotection without sedative effects or cardiac AV blocks. To circumvent solubility and formulation issues for neladenoson, a prodrug approach was pursued. The dipeptide ester neladenoson bialanate hydrochloride showed significantly improved solubility and exposure after oral administration. Neladenoson bialanate hydrochloride is currently being evaluated in clinical trials for the treatment of heart failure.

Full text of DOI:10.1002/cmdc.201700151

DOI: 10.1002/cmdc.201700151
Full Papers
Very Important Paper
Neladenoson Bialanate Hydrochloride: A Prodrug of
a Partial Adenosine A1 Receptor Agonist for the Chronic
Treatment of Heart Diseases
Daniel Meibom,*[a] Barbara Albrecht-Kꢀpper,*[b] Nicole Diedrichs+,[a] Walter Hꢀbsch,[a]
Raimund Kast,[b] Thomas Krꢁmer,[a] Ursula Krenz,[a] Hans-Georg Lerchen,[a]
Joachim Mittendorf,[a] Peter G. Nell+,[a] Frank Sꢀssmeier,[a] Alexandros Vakalopoulos,[a] and
Katja Zimmermann[c]
In memory of Jçrg Keldenich
Adenosine is known to be released under a variety of physio-
logical and pathophysiological conditions to facilitate the pro-
tection and regeneration of injured ischemic tissues. The acti-
vation of myocardial adenosine A1 receptors (A1Rs) has been
shown to inhibit myocardial pathologies associated with ische-
mia and reperfusion injury, suggesting several options for new
cardiovascular therapies. When full A1R agonists are used, the
desired protective and regenerative cardiovascular effects are
usually overshadowed by unintended pharmacological effects
such as induction of bradycardia, atrioventricular (AV) blocks,
and sedation. These unwanted effects can be overcome by
using partial A1R agonists. Starting from previously reported
capadenoson we evaluated options to tailor A1R agonists to
a specific partiality range, thereby optimizing the therapeutic
window. This led to the identification of the potent and selec-
tive agonist neladenoson, which shows the desired partial re-
sponse on the A1R, resulting in cardioprotection without seda-
tive effects or cardiac AV blocks. To circumvent solubility and
formulation issues for neladenoson, a prodrug approach was
pursued. The dipeptide ester neladenoson bialanate hydro-
chloride showed significantly improved solubility and exposure
after oral administration. Neladenoson bialanate hydrochloride
is currently being evaluated in clinical trials for the treatment
of heart failure.
Introduction
Cardiovascular diseases are still the main cause of mortality
worldwide, and they will be an increasing challenge in a grow-
ing elderly population. Although a high standard of care has
been established for several cardiac indications including coro-
nary artery disease and heart failure (HF), with a decreased
ejection fraction,[1] there is still a high unmet medical need for
new drugs that are capable of improving long-term prognosis
and quality of life. Furthermore, despite medical advances in
the treatment of HF over the last two decades, there are no
approved treatment options for HF patients with preserved
ejection fraction. Direct and selective activation of adenosi-
ne A1 receptors (A1Rs) is a new mode of action that is not cur-
rently addressed by other cardiovascular drugs. Activation of
myocardial A1Rs has been shown to inhibit a variety of myocar-
dial pathologies associated with ischemia and reperfusion
injury, including myocardial stunning, arrhythmogenesis, ab-
normalities of coronary vasomotor tone, ventricular dysfunc-
tion, acute myocardial infarction, apoptosis, and chronic HF,[2–4]
suggesting a broad spectrum of therapeutic possibilities for
the treatment of cardiac diseases.
Adenosine, a purine nucleoside, is present in all cells at
tightly regulated concentrations. It is released under a variety
of physiological and pathophysiological conditions to facilitate
protection and regeneration of injured tissues. In the heart, ad-
enosine-induced protective and regenerative effects are medi-
ated predominantly by A1R activation.[5] A1Rs are expressed in
cardiomyocytes, smooth muscle cells,[6] supraventricular tissue
(atria), and ventricles.[7] They belong to the family of seven-
transmembrane-spanning purinergic P1 receptors and couple
to the inhibitory Gi protein, which decreases the concentration
of the secondary messenger cyclic adenosine monophosphate
(cAMP).[8,9] Furthermore, A1Rs couple to phospholipase C,
thereby influencing inositol triphosphate and Ca2+ release
from internal stores. In addition, activation of A1Rs is cardiopro-
tective by modulating potassium channels such as mitochon-
drial KATP channels which improves and protects mitochondrial
[a] Dr. D. Meibom, Dr. N. Diedrichs,+ Dr. W. Hꢀbsch, Dr. T. Krꢁmer, U. Krenz,
Dr. H.-G. Lerchen, Prof. Dr. J. Mittendorf, Dr. P. G. Nell,+ F. Sꢀssmeier,
Dr. A. Vakalopoulos
Medicinal Chemistry Wuppertal, Bayer AG, 42113 Wuppertal (Germany)
E-mail: daniel.meibom@bayer.com
[b] Dr. B. Albrecht-Kꢀpper, Dr. R. Kast
Department of Cardiology Research Wuppertal, Bayer AG, 42113 Wuppertal
(Germany)
E-mail: barbara.albrecht@bayer.com
[c] Dr. K. Zimmermann
Lead Discovery Wuppertal, Bayer AG, 42113 Wuppertal (Germany)
[+] Present address: Casebia Therapeutics, San Francisco, CA, 94158 (USA)
Supporting information for this article can be found under:
https://doi.org/10.1002/cmdc.201700151.
ChemMedChem 2017, 12, 1 – 11
1
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
function in cardiomyocytes under hypoxia.[10] Effects on L-type
calcium channels[2] are also known to be mediated by A1Rs.
The main limitations of using full A1R agonists in cardiovas-
cular indications like HF are undesired cardiac effects, such as
negative chronotropic and dromotropic effects leading to bra-
dycardia and potentially to a higher degree, atrioventricular
(AV) block as shown with the full A1R agonists tecadenoson (1)
and selodenoson (2) (Figure 1).[11]
in patients with stable angina pectoris.[17] No induction of AV
blocks was reported in this study. Furthermore, 4 showed ben-
eficial effects, for example, on LV ejection fraction in a canine
HF model without any effects on blood pressure or heart
rate.[4] In a rodent model of acute myocardial infarction in-
duced by temporary occlusion of the left anterior descending
artery (LAD), treatment with capadenoson (4) decreased infarct
size, presumably through pharmacological preconditioning.[2]
The degree of agonism of 4, however, turned out to be too
high still, as central effects such as vertigo and dizziness have
been observed in the above-mentioned clinical study.[17] Fur-
thermore, the low solubility of capadenoson (4) hampered
tablet development.[18] Taken together, safety and formulation
aspects prompted us to start a follow-up project.
Results and Discussion
Addressing safety margins
CNS effects of adenosine including vertigo and sedation are
mediated by the activation of A1Rs in the brain. Full A1R ago-
nists such as tecadenoson (1) are capable of inducing pro-
longed sedation.[19] The central effects observed with the par-
tial A1R agonist capadenoson (4) were mild in nature and
became visible in some cases at high plasma concentrations.[17]
This led to the hypothesis that with a compound more partial
than 4, it might be possible to avoid unwanted A1R-mediated
pharmacological effects. More specifically: an A1R agonist with
the right degree of partiality could be capable of eliciting the
desired pharmacological responses in the heart (e.g., cardio-
protection) while being no longer agonistic enough to induce
signaling in brain tissues. Limiting blood–brain barrier (BBB)
permeation was not regarded as a valid strategy, as the BBB is
often not intact in elderly heart disease patients.
Figure 1. Full A1R agonists: Tecadenoson (1) reached phase III clinical trials
for the treatment of paroxysmal supraventricular tachycardia. Selodenoson
(2) was evaluated in patients with atrial fibrillation. Partial A1R agonists: GS-
9667 (3) has been developed for type 2 diabetes, but development was ter-
minated after phase I clinical trials. Capadenoson (4) was evaluated in pha-
se II in patients with stable angina pectoris.
The separation of desired from undesired pharmacological
effects might be rationalized by different receptor reserves for
A1R-induced signaling cascades, as has been described for A1Rs
in guinea pig atrial myocytes.[20] Whereas a full A1R agonist has
high efficacy on A1Rs and thereby elicits the full spectrum of
physiological effects independent of the receptor reserve in
different tissues, a partial A1R agonist has lower efficacy and
therefore only activates certain physiological responses, mostly
with high receptor reserve. The decreased efficacy of a partial
A1R agonist is postulated to be due to its binding only to cer-
tain receptor activity states with compromised coupling to the
G protein. This results in a selected activation of signaling cas-
cades, as has been shown for the activation of A2a receptors by
partial A2a receptor agonists.[21] The activation process can thus
be viewed from the perspective of populations of key func-
tional states, and the action of ligands on these conformational
states through conformational selection.[21] The lower efficacy
resulting in decreased signaling is therefore dependent on the
conformational state addressed by the partial agonist and the
specific receptor reserve involved in the induction of different
physiological effects.
Furthermore, non-cardiac effects such as decreased neuro-
transmitter release in the central nervous system (CNS), result-
ing in sedation,[12,13] for example, are known to be triggered by
full A1R agonists. Therefore, partial A1R agonists might be ben-
eficial as a new therapeutic option in treating HF.[14] We previ-
ously reported that partial A1R agonists modulate and trigger
primarily favorable pharmacological responses for HF therapy,
such as improvement of left ventricular (LV) function and cardi-
oprotection without evoking the adverse effects of full A1R ag-
onists.[4] To date, the two partial A1R agonists GS-9667 (3) and
capadenoson (4) have reached clinical testing. GS-9667 (3)
showed free fatty acid lowering effects in obese subjects while
being devoid of hemodynamic properties.[15] However, the de-
velopment of 3 was terminated due to unfavorable human
pharmacokinetics data.[16] Capadenoson (4) belongs to the di-
cyanopyridine class of A1R agonists, which was found in
a high-throughput screen of our compound library. Hit optimi-
zation led to the identification of 4, a potent and selective par-
tial A1R agonist with favorable DMPK properties.[2,3] In phase II
clinical studies, capadenoson (4) improved total exercise time
To quantify the partial agonistic activity on the A1R, a GTP
shift assay was used (see Supporting Information for details): Ki
&
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
2
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 2. A GTP shift assay (Supporting Information) was used to categorize A1R modulators as weak agonists, partial agonists, or full agonists. Rat brain
membranes expressing the A1R were incubated with [3H]DPCPX as radioligand and A1R modulator at various concentrations in the presence and absence of
GTP. Binding was determined after filtration and scintillation measurement. DPCPX (GTP shift 1.0) and tecadenoson (1, GTP shift 15.0) were used as controls.
measurements for A1R modulators in the presence
and absence of GTP were performed. The assay uses
the fact that A1R modulators have different affinities
for the G-protein-coupled receptor (without GTP ad-
dition) or the uncoupled receptor (with GTP addition)
depending on their partiality. Full A1R agonists have
high affinity for the G-protein-coupled receptor and
low affinity for the uncoupled receptor, whereas
weak agonists show high affinity in both cases. The
factor between the Ki values from both measure-
ments (KiGTP/Ki) is the GTP shift, which is high for full
agonists and intermediate for partial A1R agonists
Figure 3. Effects on locomotion were assessed in a mouse running wheel model (Sup-
porting Information). Compounds were dosed orally once daily by gavage, and running
(Figure 2). The degree of partiality of the agonists
translates into different efficacies for signal induction
by A1R activation. Full agonists have high efficacy,
whereas the efficacy is decreased with increasing par-
tiality of the agonists.
distances were analyzed over 8 h during the night activity phase. Potent partial A1R ago-
nists with similar PK properties (data not shown) and a GTP shift of 4.0 or lower did not
decrease running distance significantly. Potency was measured by light emission of Chi-
nese hamster ovary (CHO) cells expressing the human A1R and a cAMP/luciferase report-
er gene (Supporting Information). GTP shift: rat brain membranes expressing the A1R
were incubated with [3H]DPCPX as radioligand and A1R modulator at various concentra-
tions in the presence and absence of GTP (Supporting Information).
Dicyanopyridines were found to cover the full GTP
shift range. To identify the optimal GTP shift range to
achieve increased safety margins toward sedation on
the one hand and cardioprotection on the other, we tested
a variety of dicyanopyridines with a GTP shift lower than that
of capadenoson (4, GTP shift 5.0) for effects on locomotion
and infarct size reduction. Effects on locomotion were assessed
in a mouse running wheel model (see Supporting Information
for details), in which nighttime running distances of mice were
used as a surrogate parameter for central sedative properties.
As capadenoson decreased the running distance in this model
and showed some hints for CNS effects in humans,[17] we con-
cluded that results obtained in mice might translate to
humans. Compounds with a GTP shift of 4.0 and lower did not
decrease running distance significantly, as exemplified by the
dicyanopyridine 7 (Figure 3).
Cardioprotective effects were analyzed in a rat model of
acute myocardial infarction induced by occlusion of the LAD
(see Supporting Information for details). Dicyanopyridines of
similar potency were administered intravenously (i.v.) as bolus
prior to occlusion of the LAD. After reperfusion for two hours,
the infarcted area was quantified. It was found that com-
pounds with a GTP shift of 3.0 or above are able to decrease
infarct size, presumably through pharmacological precondi-
tioning (Figure 4).
In an extension of our previous work[2,3] we started with
a modification of the northern hydroxyethoxyphenyl substitu-
ent (Table 1) to identify compounds with suitable partiality
(GTP shift 3.0–4.0). The pyridine derivative 11 was found to be
a very potent and selective partial A1R agonist, but with a GTP
shift falling outside the desired range. Five-membered hetero-
cycles, as represented by the hydroxyethyl-substituted pyra-
zole 12, exhibited GTP shifts that were too low. Modification of
the hydroxyethoxy side chain provided a variety of potent par-
tial A1R agonists (examples 9, 13–17). However, only the tri-
fluoromethyl-substituted compound 9 and the dihydroxy de-
rivative 13 displayed favorable GTP shifts between 3.0 and 4.0.
Both compounds were deprioritized due to unfavorable PK
characteristics based on a first assessment in rats (oral adminis-
tration, data not shown; 9) or insufficient selectivity versus the
A2b receptor (compound 13).
We then turned our attention to the southeastern part of ca-
padenoson (4, Table 2). Replacement of the thiazole moiety of
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
3
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Figure 4. Cardioprotective properties of partial A1R agonists were assessed in an acute myocardial infarction model in rats (Supporting Information). Com-
pounds were dosed intravenously before ischemia by bolus injection, and infarct sizes were measured after reperfusion. Administration of potent dicyanopyri-
dines with a GTP shift of 3.0 or above led to infarct size reduction (ISR). Potency was measured by light emission of CHO cells expressing the human A1R and
a cAMP/luciferase reporter gene (Supporting Information). GTP shift: rat brain membranes expressing the A1R were incubated with [3H]DPCPX as radioligand
and A1R modulator at various concentrations in the presence and absence of GTP (Supporting Information).
4 by an oxazole yielded compound 18 with a markedly de-
creased GTP shift of 1.6. Introduction of a 5-methyl group pro-
vided no improvement (19, Table 2). Variations on the phenyl
ring, on the other hand, had a greater impact on the GTP shift,
as represented by compounds 20 and 21. However, partial ag-
onism with a GTP shift in the desired range could not be ach-
Table 1. Exploration of northern SAR: potency, selectivity, and GTP shift.
Table 1. (Continued)
Compd
R
EC50 [nm][a]
EC50 ratio[a]
A2a/A1 A2b/A1 shift[b]
GTP
A1
Compd
R
EC50 [nm][a]
EC50 ratio[a]
A2a/A1 A2b/A1 shift[b]
GTP
4[c]
0.1
1800
900
5.0
A1
15 (ent1,[d] trans)
0.3
3900
3600
870
1500
2.0
5.0
6.2
11
0.04
1.9
30000 10000
5.2
2.5
3.5
12
1000
7200
730
260
16 (ent2,[d] trans)
0.7
850
9 (ent)
0.8
17
0.05
1500
13
0.05
680
85
3.0
1.5
[a] Potency was measured with CHO cells expressing either the human
A1, A2a, or A2b receptor and a cAMP/luciferase reporter gene (Supporting
Information). A1 cells were pre-stimulated with 1 mm forskolin. Cells were
incubated with increasing doses of test compounds, and luciferase ex-
pression was measured by light emission. [b] GTP shift: rat brain mem-
branes expressing the A1R were incubated with [3H]DPCPX as radioligand
and A1R modulator at various concentrations in the presence and ab-
sence of GTP (Supporting Information). [c] Capadenoson. [d] Enantiomer 1
was separated from enantiomer 2 by chiral chromatography; absolute
stereochemistry was not elucidated.
14 (rac)
0.3
4800
2400
&
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
4
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 2. Exploration of southeastern SAR: potency, selectivity, and GTP
Table 2. (Continued)
shift.
Compd
R
EC50 [nm][a]
EC50 ratio[a]
A2a/A1 A2b/A1
GTP
shift[b]
Compd
R
EC50 [nm][a]
EC50 ratio[a]
A2a/A1 A2b/A1
GTP
shift[b]
A1
A1
25
0.5
710
140
2.8
18
0.1
0.3
410
180
1.6
2.0
[a] Potency was measured with CHO cells expressing either the human
A1, A2a, or A2b receptor and a cAMP/luciferase reporter gene (Supporting
Information). A1 cells were pre-stimulated with 1 mm forskolin. Cells were
incubated with increasing doses of test compounds, and luciferase ex-
pression was measured by light emission. [b] GTP shift: rat brain mem-
branes expressing the A1R were incubated with [3H]DPCPX as radioligand
and A1R modulator at various concentrations in the presence and ab-
sence of GTP (Supporting Information).
19
2100
3200
20
21
0.3
2200
2600
4.8
24, and triazole 25 also provided no improvement with respect
to potency and degree of partial agonism.
In contrast, evaluation of the southwestern region turned
out to be more successful. As represented by the examples in
Table 3, introduction of various substituents on the primary
amino group of capadenoson (4) resulted in a decreased GTP
shift while maintaining favorable potency and selectivity. Par-
tial A1R agonism within the desired range could be achieved
with the glycine amide side chain (27) and especially with
cyclic amino substituents (7, 28–31), which was unexpected
and is still not fully understood.1 Amide 27 and piperidines 7
and 31 showed unfavorable PK properties in rats after oral ad-
ministration (data not shown) and were not further pursued.
The azetidine derivatives 28 and 29 had to be deprioritized
due to instability of the four-membered rings under acidic con-
ditions. The pyrrolidine 30 (neladenoson) showed the best
overall profile in terms of potency, selectivity versus other ade-
nosine receptors (including A3, EC50 >3000 nm), and partiality
on the A1R. Furthermore, no significant interference with 67
off-targets (MDS Pharma, Supporting Information), no CYP or
hERG inhibition, and no in vitro mutagenicity were detected.
The in vivo PK profile of neladenoson (30) is characterized
by low clearance and moderate to long half-life in rats, dogs,
and monkeys (Table 4) after i.v. administration. Compared with
capadenoson (4, Table 5), it showed an improved clearance
and half-life, particularly in dogs. However, oral bioavailability
values of neladenoson (30) were only in the low-to-moderate
range due to low aqueous solubility (<0.1 mgLÀ1, crystalline
0.02
610
140
10.0
22
23
35
10
15
3.6
4.2
1.0
870
800
24
0.4
2700
300
4.3
ieved. Shifting the chlorophenyl substituent from the 2-posi-
tion as in compound 18 to the 5-position of the oxazole result-
ed in significantly decreased A1R agonistic potency (22, EC50 =
35 nm). Replacement of the thiazole of capadenoson 4 by ni-
trogen-containing heterocycles such as pyrazole 23, imidazole
1
The molecular understanding of receptor activation by partial agonists is
a topic of current research. A recent publication by Ye et al.[21] explains the
mechanism of partial agonism on the A2a receptor.
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
5
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 3. Exploration of southwestern SAR: potency, selectivity, and GTP
shift.
Table 5. Pharmacokinetic profile of capadenoson (4) in selected
species.[a]
[b]
[c]
Species
CLb [LhÀ1 kgÀ1
]
VSS [LkgÀ1
]
t1/2 [h]
F [%][d]
Rat
Dog
Monkey
1.4
1.2
0.21
10
2.3
2.7
8.4
2.4
9.8
57
21
60
[a] Values were derived by i.v. and oral (gavage) administration. [b] Blood
clearance. [c] Volume of distribution at steady state. [d] Oral bioavailabil-
ity.
Compd
R
EC50 [nm][a]
EC50 ratio[a]
GTP
shift[b]
A1
A2a/A1
A2b/A1
use of organic solvents like liquid-filled capsules and solid dis-
persions. However, these approaches were quickly dismissed,
as the expected human dose was not soluble in acceptable
capsule or tablet sizes. Several drugs are known, in which the
introduction of prodrug moieties led to improved solubility
and bioavailability.[22–25] We therefore turned our attention to
investigating whether it would be beneficial to explore pro-
drug forms of neladenoson (30). A suitable candidate needed
to have improved solubility and bioavailability from suspension
and it had to be sufficiently stable for tablet development and
to enable rapid liberation of the drug in vivo. The hydroxy
group of 30 was selected as point of attachment for prodrug
moieties.
26
27
0.1
0.9
460
790
360
4.3
3.6
900
500
28
29
0.3
0.3
1000
2000
3.6
3.5
8600
30[c]
7
0.1
0.6
6700
5000
800
3.9
3.9
5000
The first part of our prodrug screening cascade consisted of
stability testing in aqueous solution at various pH values to
identify prodrugs with good stability, particularly at pHꢁ4, to
ensure that parent drug liberation in the stomach at low pH
does not take place. Esters of natural and unnatural amino
acids,[25,26] which are protonated on the terminal amino group
under acidic conditions, fulfilled this requirement (Figure 5).
The next filter in our prodrug screening cascade was the
availability of crystalline prodrugs by stirring in various sol-
vents. Crystalline material with defined hydrochloride stoichi-
ometry was obtained from isopropanol as exemplified by the
isoleucine derivative 34 (Figure 6).
31
0.6
510
1200
3.2
[a] Potency was measured with CHO cells expressing either the human
A1, A2a, or A2b receptor and a cAMP/luciferase reporter gene (Supporting
Information). A1 cells were pre-stimulated with 1 mm forskolin. Cells were
incubated with increasing doses of test compounds, and luciferase ex-
pression was measured by light emission. [b] GTP shift: rat brain mem-
branes expressing the A1R were incubated with [3H]DPCPX as radioligand
and A1R modulator at various concentrations in the presence and ab-
sence of GTP (Supporting Information). [c] Neladenoson.
Table 4. Pharmacokinetic profile of neladenoson (30) in selected
Further criteria were bioavailability of 30 in rats after oral ad-
ministration of suspensions of crystalline prodrugs and the
seven-day stability of the solid prodrugs at 908C. Among dif-
ferent dipeptide esters, the bis-l-alanyl prodrug 35 (neladeno-
species.[a]
[b]
[c]
Species
CLb [LhÀ1 kgÀ1
]
VSS [LkgÀ1
]
t1/2 [h]
F [%][d]
Rat
Dog
Monkey
1.1
0.2
0.5
8.0
1.9
0.9
6.3
9.6
4.3
37
51
39
[a] Values were derived by i.v. (rat: bolus, dose 0.5 mgkgÀ1; dog
monkey: 0.25 h infusion, dose 0.3 mgkgÀ1
and oral (gavage, dose
&
)
1.0 mgkgÀ1) administration of solutions in EtOH/PEG400/H2O vehicles.
[b] Blood clearance. [c] Volume of distribution at steady state. [d] Oral bio-
availability.
material). The exposure was further markedly decreased when
administered as a suspension of crystalline material in 0.5%
Tylose (Fꢀ1%).
Addressing solubility and bioavailability
Figure 5. Stability of prodrugs in aqueous solution was monitored for 24 h.
No ester cleavage and liberation of neladenoson (30) was detected at pH 4.
At pH 7.4 slow conversion to 30 took place.
The low solubility and bioavailability of 30 in aqueous media
prompted us to evaluate formulation techniques that make
&
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
6
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
thioacetamide with two equivalents of the Knoeve-
nagel product 39 and benzylic chloride 37. The se-
quence likely starts by reacting one equivalent of 39
with 2-cyanothioacetamide, leading to a dihydropyri-
dine intermediate which is oxidized in situ[29] to a pyr-
idine by another equivalent of 39 in the second
step. Finally, 37 alkylates the sulfur-substituted pyri-
dine obtained in the previous reaction to give 4. The
pyrrolidine was introduced via Sandmeyer reaction
and substitution of the chlorine atom. The dipeptide
43 was obtained via sequential coupling of two tert-
butoxycarbonyl (Boc)-protected alanine moieties. Re-
moval of the Boc group with 1n HCl in the final
step afforded the prodrug neladenoson bialanate
hydrochloride (35).
Figure 6. Crystallization experiments were performed with suspensions of prodrugs
under a variety of conditions including a solvent screen. Stirring in isopropanol for seven
days at room temperature gave good results.
son bialanate hydrochloride) showed the best overall profile
In vivo pharmacodynamics studies
with good solubility, rapid liberation of neladenoson (30)
in vivo, improved oral exposure of 30, and no degradation in
the solid state. Compound 35 was deemed stable enough for
tablet development and was selected as the development can-
didate (Table 6).
Capadenoson (4) and neladenoson (30) are both selective and
potent A1R agonists with EC50 values of 0.1 nm on human A1Rs
and a selectivity factor of 1800 and 6700 versus A2a, 900 and
800 versus A2b, respectively, and no significant activity on A3 re-
ceptors.[2] The partial agonistic character of compounds 30 and
4 was analyzed relative to the full A1R agonist 2-chloro-N6-cy-
clopentyladenosine (CCPA) on human A1Rs by a [35S]GTP bind-
ing assay (Supporting Information). Neladenoson (30) and ca-
padenoson (4) reached ~67 and 79% of the efficacy of CCPA,
respectively (Figure 7).
Table 6. Profile of neladenoson bialanate hydrochloride (35).
Parameter
Result
Stability in aqueous solution
Solubility in aqueous medium
Prodrug concentrations in rat plasma
Bioavailability of 30 from suspension of 35
Stability as solid
stable 24 h at pHꢁ5[a]
6 gLÀ1 at pH 4.5[b]
negligible[c]
52%[d]
stable 7 days at 908C[a]
[a] No degradation (ester cleavage) to neladenson (30) was detectable.
[b] Measured from crystalline material. Solubility of neladenoson (30) at
pH 4: <0.1 mgLÀ1. [c] After oral administration of 35 to rats, 35 was only
sporadically detected in the systemic circulation at single time points and
at very low concentrations. [d] Calculation based on study with oral
(gavage) administration of crystalline material of 35 as suspension (0.5%
Tylose) to rats. Exposure of neladenoson (30) exceeded the one achieva-
ble by administration of a solution of 30, whereas bioavailability of nela-
denson (30) from suspension of crystalline material of 30 accounted to
only ꢀ1%.
^
Figure 7. Degree of partiality of neladenoson (30, ) on A1Rs compared with
!
*
capadenoson (4, ) and CCPA ( ). Partial character of A1R agonists was as-
sessed on human A1Rs from frontal lobe in a [35S]GTP binding assay (Sup-
porting Information). Data are the meanÆSEM (n=8).
Although the difference in the level of partiality between
both compounds is only 12%, there were significant differen-
ces in their pharmacological effects. Capadenoson (4) showed
a significant decrease of locomotion activity, as surrogate
marker for sedative CNS effects, in a mouse running wheel
model (Figure 8). The compound decreased the overnight run-
ning distance of mice in a dose-dependent manner starting
from 0.3 mgkgÀ1 onward. In contrast, neladenoson (30) had no
significant effect on the running distance up to doses of
3 mgkgÀ1.
Synthesis
Neladenoson bialanate hydrochloride (35) was synthesized in
an eight-step sequence[27,28] starting from commercially avail-
able thioamide 36 and aldehyde 38 (Scheme 1). The first key
step is the Hantzsch pyridine synthesis to afford capadenoson
(4) in a one-pot/three-reaction sequence by reacting 2-cyano-
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
7
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Scheme 1. Synthesis of neladenoson bialanate hydrochloride (35). Reagents and conditions: a) 1,3-dichloroacetone, 2-propanol, 558C, 7.5 h, 91%; b) malononi-
trile, piperidine, 2-propanol, 808C, 16 h, 95%; c) 2-cyanothioacetamide, tributylamine, MeOH, RT, 20 h, 68%; d) isopentyl nitrite, CuCl2, MeCN, 708C, 6 h, 62%;
e) pyrrolidine, THF, RT, 30 min, 81%; f) Boc-l-alanine, EDC·HCl, HOBt·H2O, N,N-diisopropylethylamine, DMF, RT, 1day, 98%; g) TFA, CH2Cl2, RT, 1day, 98%; h) Boc-
l-alanine, EDC·HCl, HOBt·H2O, N,N-diisopropylethylamine, DMF, RT, 3 h, 91%; i) 1n HCl in Et2O, CH2Cl2, RT, 1day, 91%. EDC=N-(3-dimethylaminopropyl)-N’-eth-
ylcarbodiimide; HOBt=hydroxybenzotriazole.
Determination of free brain concentrations of capadenoson
(4) and neladenoson (30) showed equivalent exposures (data
not shown). Therefore, the decreased effects on locomotion by
30 cannot be attributed to differences in the penetration of
the BBB. It is more likely that differences on the A1R reserve
are responsible for the differentiated pharmacological effects
depending on the level of partiality of the agonists. Based on
our data, we assume that there is a low receptor reserve for
the induction of sedative effects measured as decreased loco-
motor activity. The higher partiality of neladenoson (30) does
not allow the induction of these effects. The findings are sup-
ported by clinical studies with neladenoson bialanate hydro-
chloride (35) in humans (HF patients and healthy subjects), in
which no safety issue with respect to sedation was observed
(manuscript in preparation).
such as CCPA.[30] Because capadenoson (4) showed beneficial
effects in a canine model of HF with reduced ejection frac-
tion,[4] benefits of neladenoson (30) on cardiac dysfunction
were evaluated in a rodent HF model; chronic effects of 30 on
heart function were measured in rats with permanent LAD oc-
clusion. One week after LAD occlusion, treatment with
0.3 mgkgÀ1 30 once daily was started over eight weeks fol-
lowed by an invasive measurement of cardiac function by left
ventricular catheterization (Supporting Information). Placebo
rats showed a significant decrease in contractility and relaxa-
tion parameters and thereby systolic and diastolic function
versus sham rats (dp/dtmax 5575Æ479 versus 7501Æ
1031 mmHgsÀ1
;
dp/dtmin
5399Æ748
versus
7694Æ
1011 mmHgsÀ1; tau 17.3Æ3.1 versus 11.3Æ1.5 ms) (Figure 10).
Treatment with neladenoson (30) at 0.3 mgkgÀ1 p.o. once daily
resulted in a significant improvement of contraction and relax-
Although neladenoson (30) had decreased effects on motor
activity (sedation), it showed slightly higher cardioprotection
measured as a decrease of infarct size (IS) relative to capadeno-
son (4), whereas both compounds are similar in their unbound
exposure. When administered before induction of ischemia by
left anterior descending artery (LAD) occlusion, 30 decreased
IS from 28.7Æ1.5% to 19.8Æ3.3% at 0.03 mgkgÀ1 i.v. signifi-
cantly in a rat model of acute myocardial infarction (Figure 9).
In contrast, 4 decreased IS to 22.0Æ3.4% only at the higher
dose of 0.1 mgkgÀ1. The level of infarct size reduction by nela-
denoson (30) is similar to published data for full A1R agonists
ation parameters (dp/dtmax 7503Æ765 mmHgsÀ1
; dp/dtmin
7385Æ1298 mmHgsÀ1; tau 14.4Æ2.1 ms).
Conclusions
Starting from capadenoson (4) we evaluated options to identi-
fy partial A1R agonists within a specific partiality range (GTP
shift 3.0–4.0) to improve safety margins regarding CNS effects.
The introduction of cyclic amines in the southwest region of
the dicyanopyridine core provided compounds with suitable
&
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
8
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 8. Effects of A) neladenoson (30) and B) capadenoson (4) on over-
night running distance of mice in a running wheel. Running distance was
automatically measured (Supporting Information). Both compounds were
applied orally. Data are the meanÆSEM (n=8–10); *p<0.05, **p<0.01 vs.
placebo.
Figure 10. Effects of neladenoson (30) on heart function in rats treated with
sham (control), placebo, and 30 (0.3 mgkgÀ1 p.o.) in a rat HF model of per-
manent LAD occlusion (Supporting Information). Improvement in contractili-
ty [A) dp/dtmax] and relaxation [B) dp/dtmin; C) tau] were observed for nelade-
noson (30) after eight weeks treatment. Data are the meanÆSEM (n=9–
14); *p<0.05, **p<0.01, ***p<0.001 vs. placebo.
PK profile after i.v. administration. However, low solubility led
to low oral exposure and potential tablet development issues.
We therefore embarked on a prodrug approach to identify
a candidate with good solubility and bioavailability. Stability
was tailored to allow efficient liberation of the drug in vivo
while showing no degradation in the solid state. The hydro-
chloride form of the dipeptide ester neladenoson bialanate
(35) fulfilled all desired criteria (Table 7).
Neladenoson (30) preserved the high potency and cardio-
protective effects observed with capadenoson (4), while
having significantly reduced central effects. In phase I and pha-
se IIa studies the neladenoson prodrug 35 was well tolerated
and showed no safety issues with respect to sedation or AV
block induction (manuscript in preparation), which makes the
compound suitable for chronic treatment of patients with car-
diac diseases such as heart failure (HF). Neladenoson bialanate
Figure 9. Cardioprotection of partial A1R agonists in rodents. Cardioprotec-
tive properties of neladenoson (30) and capadenoson (4) by reduction of in-
farct size given as percentage of area at risk (AAR) were assessed in an acute
myocardial infarction model in rats (Supporting Information). Compound 30
and 4 were dosed intravenously by bolus injection (0.03 and 0.1 mgkgÀ1
before LAD occlusion and infarct sizes were measured after reperfusion.
Data are the meanÆSEM (n=9–20); *p<0.05, **p<0.01 vs. placebo.
)
GTP shift values. The potent and selective A1R agonist nelade-
noson (30) had the preferred level of partiality and a favorable
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
9
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
[6] T. Hussain, S. J. Mustafa, Circ. Res. 1995, 77, 194–198.
[7] B. Musser, M. E. Morgan, M. Leid, T. F. Murray, J. Linden, R. E. Vestal, Eur.
J. Pharmacol. 1993, 246, 105–111.
Table 7. Comparison of capadenoson (4), neladenoson (30), and nelade-
noson bialanate hydrochloride (35).
[8] D. van Calker, M. Mꢀller, B. Hamprecht, Nature 1978, 276, 839–841.
[9] C. Londos, D. M. F. Cooper, J. Wolff, Proc. Natl. Acad. Sci. USA 1980, 77,
2551–2554.
[10] F. Xiang, Y. Huang, D. Zhang, Z. Chu, J. Zhang, Q. Zhang, Clin. Exp. Phar-
macol. Physiol. 2010, 37, 343–349.
Compd
A1 EC50 [nm][a]
GTP shift[b]
Aq. sol. [mgLÀ1
]
F [%][c]
4
30
35
0.1
0.1
–
5.0
3.9
–
<0.1[d]
<0.1[d]
6000[e]
ꢀ1
ꢀ1
52
[11] W. F. Kiesman, E. Elzein, J. Zablocki, Handb. Exp. Pharmacol. 2009, 193,
25–58.
[12] B. Johansson, L. Halldner, T. V. Dunwiddie, S. A. Masino, W. Poelchen, L.
Gimenez-Llort, R. M. Escorihuela, A. Fernandez-Teruel, Z. Wiesenfeld-
Hallin, X.-J. Xu, A. Hardemark, C. Betsholtz, E. Herlenius, B. B. Fredholm,
Proc. Natl. Acad. Sci. USA 2001, 98, 9407–9412.
[13] R. Basheer, L. Halldner, L. Alanko, R. W. McCarley, B. B. Fredholm, T.
Porkka-Heiskanen, Neuroreport 2001, 12, 1577–1580.
[14] S. J. Greene, H. N. Sabbah, J. Butler, A. A. Voors, B. E. Albrecht-Kꢀpper,
H.-D. Dꢀngen, W. Dinh, M. Gheorghiade, Heart Failure Rev. 2016, 21,
95–102.
[a] Potency was measured with CHO cells expressing the human A1R and
a cAMP/luciferase reporter gene (Supporting Information). A1 cells were
pre-stimulated with 1 mm forskolin. Cells were incubated with increasing
doses of test compounds, and luciferase expression was measured by
light emission. [b] GTP shift: rat brain membranes expressing the A1R
were incubated with [3H]DPCPX as radioligand and A1R modulator at vari-
ous concentrations in the presence and absence of GTP (Supporting In-
formation). [c] Oral bioavailability in rats from suspension in 0.5% Tylose.
[d] Measured at pH 4. [e] Measured at pH 4.5.
[15] P. M. Staehr, A. K. Dhalla, J. Zack, X. Wang, Y. L. Ho, J. Bingham, L. Belar-
dinelli, J. Clin. Pharmacol. 2013, 53, 385–392.
[16] Gilead Sciences Inc., 10-K Annual Report 2011, 63.
[17] M. Tendera, E. Gaszewska-Zurek, Z. Parma, P. Ponikowski, E. Jankowska,
K. Kawecka-Jaszcz, D. Czarnecka, M. Krzeminska-Pakula, Z. Bednarkie-
wicz, M. Sosnowski, M. O. Kilama, R. Agrawal, Clin. Res. Cardiol. 2012,
101, 585–591.
[18] H.-G. Lerchen, U. Krenz, J. Keldenich, N. Diedrichs, T. Krahn, C. Hirth-Die-
trich, B. Albrecht-Kꢀpper (Bayer Schering Pharma AG), Int. PCT Pub. No.
WO2009015811 A1, 2009.
hydrochloride (35) is currently under evaluation in clinical trials
for the treatment of HF. Phase IIb studies are planned for 2017.
Experimental Section
All experimental data are shown in the Supporting Information.
[19] L. A. Sorbera, J. Castaner, L. Martin, M. Bayes, Drugs Future 2002, 27,
846–849.
[20] M. Srinivas, J. C. Shryock, D. M. Dennis, S. P. Baker, L. Belardinelli, Mol.
Pharmacol. 1997, 52, 683–691.
Acknowledgements
[21] L. Ye, N. V. Eps, M. Zimmer, O. P. Ernst, R. S. Prosser, Nature 2016, 533,
265–268.
[22] P. Ettmayer, G. L. Amidon, B. Clement, B. Testa, J. Med. Chem. 2004, 47,
2393–3404.
We thank Heike Gielen-Haertwig as well as Stephan Vettel for
continuous support of this project and Stefanie Vogt for valuable
technical assistance with the manuscript.
[23] Design of Prodrugs: Bioreversible Derivatives for Various Functional
Groups and Chemical Entities (Ed.: H. Bundgaard), Elsevier, Amsterdam,
1985.
Conflict of interest
[24] K. Beaumont, R. Webster, I. Gardner, K. Dack, Curr. Drug Metab. 2003, 4,
461–485.
[25] B. S. Anand, Y. E. Nashed, A. K. Mitra, Curr. Eye Res. 2003, 26, 151–163.
[26] M. Sugawara, W. Huang, Y.-J. Fei, F. H. Leibach, V. Ganapathy, M. E. Gana-
pathy, J. Pharm. Sci. 2000, 89, 781–789.
The authors declare no conflict of interest.
Keywords: adenosine A1 receptor
· biological activity ·
[27] U. Rosentreter, T. Krꢁmer, M. Shimada, W. Hꢀbsch, N. Diedrichs, T. Krahn,
K. Henninger, J.-P. Stasch, R. Wischnat (Bayer Healthcare AG), Int. PCT
Pub. No. WO2003053441 A1, 2003.
dicyanopyridines · medicinal chemistry · partial agonists ·
prodrugs
[28] A. Vakalopoulos, D. Meibom, B. Albrecht-Kꢀpper, K. Zimmermann, J. Kel-
denich, H.-G. Lerchen, P. Nell, F. Sꢀssmeier, U. Krenz (Bayer Schering
Pharma AG), Int. PCT Pub. No. WO2010086101 A1, 2010.
[29] N. M. Evdokimov, A. S. Kireev, A. A. Yakovenko, M. Y. Antipin, I. V. Mage-
dov, A. Kornienko, J. Org. Chem. 2007, 72, 3443–3453.
[30] G. J. Gross, J. N. Peart, Am. J. Physiol. Heart Circ. Physiol. 2003, 285,
H921–H930.
[1] P. F. Kantor, J. Lougheed, A. Dancea, M. McGillion, N. Barbosa, C. Chan,
R. Dillenburg, J. Atallah, H. Buchholz, C. Chant-Gambacort, J. Conway, L.
Gardin, K. George, S. Greenway, D. G. Human, A. Jeewa, J. F. Price, R. D.
Ross, S. L. Roche, L. Ryerson, R. Soni, J. Wilson, K. Wong, Can. J. Cardiol.
2013, 29, 1535–1552.
[2] B. Albrecht-Kꢀpper, K. Leineweber, P. G. Nell, Purinergic Signalling 2012,
8, 91–99.
[3] P. G. Nell, B. Albrecht-Kꢀpper, Prog. Med. Chem. 2009, 47, 163–201.
[4] H. N. Sabbah, R. C. Gupta, S. Kohli, M. Wang, S. Rastogi, K. Zhang, K.
Zimmermann, N. Diedrichs, B. E. Albrecht-Kꢀpper, Circ. Heart Fail. 2013,
6, 563–571.
Manuscript received: March 7, 2017
Revised: April 12, 2017
[5] S. J. Mustafa, R. R. Morrison, B. Teng, A. Pelleg, Handb. Exp. Pharmacol.
2009, 193, 161–188.
Accepted Article published: April 12, 2017
Final Article published: && &&, 0000
&
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
10
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

FULL PAPERS
Good gets even better: Partial adeno-
sine A1 receptor (A1R) activation is asso-
ciated with a positive impact on heart
failure without the detriments of full
A1R agonism. Starting from capadeno-
son, we describe the identification of
neladenoson bialanate hydrochloride,
which has a good pharmacokinetic and
safety profile. Neladenoson bialanate
hydrochloride is currently being as-
sessed in clinical studies for the treat-
ment of heart failure.
D. Meibom,* B. Albrecht-Kꢀpper,*
N. Diedrichs, W. Hꢀbsch, R. Kast,
T. Krꢁmer, U. Krenz, H.-G. Lerchen,
J. Mittendorf, P. G. Nell, F. Sꢀssmeier,
A. Vakalopoulos, K. Zimmermann
&& – &&
Neladenoson Bialanate Hydrochloride:
A Prodrug of a Partial Adenosine A1
Receptor Agonist for the Chronic
Treatment of Heart Diseases
ChemMedChem 2017, 12, 1 – 11
www.chemmedchem.org
11
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ