Synthesis and evaluation of diphenylphosphinic amides and diphenylphosphine oxides as inhibitors of Kv1.5

Article abstract of DOI:10.1016/j.bmcl.2012.11.098

Diphenylphosphinic amides and diphenylphosphine oxides have been synthesized and tested as inhibitors of the Kv1.5 potassium ion channel as a possible treatment for atrial fibrillation. In vitro structure-activity relationships are discussed and several compounds with Kv1.5 IC₅₀ values of <0.5 μM were discovered. Selectivity over the ventricular IKs current was monitored and selective compounds were found. Results from a rabbit PD-model are included.

Full text of DOI:10.1016/j.bmcl.2012.11.098

Bioorganic & Medicinal Chemistry Letters 23 (2013) 706–710
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and evaluation of diphenylphosphinic amides and
diphenylphosphine oxides as inhibitors of Kv1.5
b
b
b
b
b
Roine I. Olsson a, , Ingemar Jacobson , Jonas Boström , Tomas Fex , Annika Björe , Christina Olsson ,
Johan Sundell b, Ulrik Gran b, Anna Öhrn b, Andreas Nordin b, Jonna Gyll b, Maria Thorstensson b,
Ahlke Hayen b, Karolina Aplander b, Olle Hidestål b, Fanyi Jiang a, Gunilla Linhardt b, Elin Forsström b,
Teresa Collins b, Monika Sundqvist b, Emma Lindhardt b, Annika Åstrand a, Boel Löfberg b
⇑
a Respiratory & Inflammation iMed, AstraZeneca R&D Mölndal, SE-431 83 Mölndal, Sweden
b Cardiovascular & Gastrointestinal iMed, AstraZeneca R&D Mölndal, SE-431 83 Mölndal, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
Diphenylphosphinic amides and diphenylphosphine oxides have been synthesized and tested as inhibi-
tors of the Kv1.5 potassium ion channel as a possible treatment for atrial fibrillation. In vitro structure–
activity relationships are discussed and several compounds with Kv1.5 IC50 values of <0.5 lM were dis-
covered. Selectivity over the ventricular IKs current was monitored and selective compounds were found.
Results from a rabbit PD-model are included.
Received 22 October 2012
Revised 23 November 2012
Accepted 25 November 2012
Available online 3 December 2012
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
Atrial fibrillation
Kv1.5
IKs
Diphenylphosphinic amides
Diphenylphosphine oxides
Atrial fibrillation (AF) is the most common cardiac arrhythmia
in clinical practice, often giving symptoms that reduce health re-
lated quality of life, causing a considerable burden in hospitaliza-
tion and increasing the risk of stroke and other cardiovascular
morbidities and mortality.1 AF is frequently associated with other
cardiovascular disease such as hypertension, coronary artery dis-
ease, valvular heart disease and congestive heart failure. With con-
tinuous growth of the elderly population, AF can be expected to
have an increasing impact on managed care and public health.
However, so far no atrial specific drug treatment exists and cur-
rently used antiarrhythmic drugs target ion channels that are pres-
ent both in cardiac atria and ventricles.2
The potassium ion channel Kv1.5, encoded by the gene KCNA5,
has been in focus as a target for pharmacological treatment of AF
during the past decade and numerous reports of the design and
synthesis of compounds targeting Kv1.5 have emerged.3–7 Kv1.5
channels conduct the ultra rapidly activating delayed rectifier K+
current, IKur, a current predominantly expressed in human cardiac
atria and block of these channels has the potential to selectively af-
fect the atrial myocytes.
In the present paper we describe the discovery, synthesis and
evaluation of novel phosphorous compounds as inhibitors of
Kv1.5. There were several underlying motives for exploiting phos-
phorous atoms in compound design against this target. In the first
instance, phosphorous is underused compared to carbon despite
the fact that the phosphorous atom has similar properties as the
carbon atom.8 Secondly, a series of diphenylphosphine oxides, like
DPO-1 (Fig. 1), has been identified as Kv1.5 blockers9 and the series
has also been used as a starting point for lead identification.10 Fi-
nally, AstraZeneca has previous experience of phosphorous com-
pounds in clinical development.11
A substructure search of the AstraZeneca corporate collection
provided a few hundred phosphorus containing compounds. From
this list of compounds, diphenylphosphinic amide A (Fig. 1) over-
laid well with the previously identified potent Kv1.5 blocker B12
using shape-matching technology13 (Fig. 2).
The ion channel-blocking potency on the Kv1.5 channel was as-
sessed in a CHO cell line stably expressing the human cardiac ion
channel Kv1.5. Similarly, the potency at blocking hERG and
KvLQT1/minK channels, was assessed. hERG underlies the cardiac
current IKr and KvLQT1/minK the cardiac current IKs, both affect-
ing the cardiac ventricular repolarization. Electrophysiological
studies were performed using a high throughput planar patch
clamp assay and standard extra- and intracellular solutions.14,15
⇑
Corresponding author.
E-mail address: Roine.i.olsson@astrazeneca.com (R.I. Olsson).
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.11.098

R. I. Olsson et al. / Bioorg. Med. Chem. Lett. 23 (2013) 706–710
707
Table 1
SAR and yields of phosphinic amides A, 1 and 2
O
P
N
O
P
N
No.
R1
R2
Kv1.5a
0.12
IKsa
14
Yieldc
N
A
29
51
50
19
N
N
N
N
N
DPO-1
A
B
n.tb
n.tb
10
Figure 1. Structures of DPO-1, A and B.
1a
2a
2b
H
>30
1.6
N
N
0.53
2c
2d
2e
0.62
1.9
>33
31
64
16
25
O
N
N
CN
N
N
1.3
>33
2f
1.0
>33
18
91
42
N
N
N
Figure 2. Overlay of compounds A and B.
2g
0.59
O
When tested, compound A showed an IC50 value of 0.12
lM in
the Kv1.5 assay and an IC50 value of 14 M in the IKs assay (Ta-
l
1b
2h
H
11
n.tb
2.7
60
42
ble 1). Moreover, compound A was found to have a chromato-
graphic LogD value of 2.6 at pH 7.4 compared to 6.0 for DPO-1,
indicating a potential for improved DMPK properties.16 A synthetic
effort to explore the structure–activity relationship (SAR) around
compound A, generated the structurally related phosphinic amides
1 and 2 shown in Figure 3.
The syntheses of phosphinic amides 1 and 2 are shown in
Scheme 1. Diphenylphosphinic chloride was treated with the
appropriate primary or secondary amine under basic conditions.
Secondary phosphinic amides were treated with the appropriate
alkyl halides R2X under phase transfer conditions to give the ter-
tiary phosphinic amide 2.
CN
CN
S
0.20
2i
0.93
>33
n.tb
36
12
O
O
O
SAR of Kv1.5 and IKs, as well as yields, of phosphinic amides 1
and 2 are shown in Table 1. hERG blocking potency of all tested
compounds was generally low, often with IC50 values above
1c
H
17
N
N
33
l
M. IKs blocking potency was more closely monitored since
M.
N
the Kv1.5 lead compound B showed an IKs IC50 value of 1.7
l
1d
1e
H
H
4.1
2.3
7.8
19
72
A mono 3-pyridyl substituent (1a) was found to be inactive
against Kv1.5. A di-3-pyridyl substituent (2a) was active but was
found to be chemically unstable. Replacing one of the 3-pyridyls
with a pyridine-2-ylmethyl (2b) gave a highly active compound
with a moderate activity on IKs. Introducing polarity in the propyl
chain gave lower IKs potency compared to compound A, but also
compromised Kv1.5 activity (2c–2d). Di-pyridine-2-ylmethyl sub-
stitution (2e) gave no IKs activity but also gave a drop in Kv1.5 po-
tency. Replacing 3-pyridyl in compound A with other heterocycles
was tolerated (2f–2g). A mono 3-cyanophenyl group (1b) was
found to give low Kv1.5 activity and further substitution with a
propyl group gave a highly active Kv1.5 compound (2h) but with
a high IKs activity. Replacing cyano with a methylsulphone (2i)
F
N
>33
a
b
c
IC50 values in lM.
The compound was not tested in this assay.
In percent.
gave no activity on IKs with only a minor drop in Kv1.5 potency.
Several biaromatics (1c–1e) were also investigated but were not
found to give any overall improvement.

708
R. I. Olsson et al. / Bioorg. Med. Chem. Lett. 23 (2013) 706–710
O
O
O
P
N
S
O
O
HO
P
a
NH
1
R4
R3
R4
R3
R2
R1
R1
5
2
4
Figure 3. General structures of phosphinic amides 1 and 2.
b
O
O
P
O
b
O
P
NH
P
Cl
d
c
H
R4
R4
R3
R3
R1
R4
R3
1
6
3
Scheme 2. Reagents and conditions: (a) MsCl, NEt3, DCM, 0 °C, 30 min, quantita-
tive; (b) (i) diethyl phosphonate, phenyl magnesium bromide, diethyl ether, 25 °C,
10 min, (ii) 5, 25 °C, 20 h, 5%; (c) (i) toluene, 100 °C, 1.5 h, (ii) diphenylphosphine
oxide, toluene, 110 °C, 18 h, 38–48%; (d) (i) n-BuLi, THF, À78 °C, 30 min, (ii)
chlorodiphenylphosphine, À78 to 25 °C, 20 h, (iii) H2O, O2, 2%.
c
a
O
O
P
N
P
N
R2
R1
R2
R1
Table 2
SAR and yields of phosphine oxide 3
2
2
No.
R3
R4
Kv1.5a
0.18
IKsa
1.0
Yieldc
5
Scheme 1. Reagents and conditions: (a) R1NHR2, NEt3, DCM, 25 °C, 18 h, 50%; (b)
R1NH2, NEt3, DCM, 0 °C, 2 h, 12–72%; (c) R2X, aqueous NaOH, tetrabutylammonium
hydrogen sulfate, DCM, 6–18 h, 19–91%.
3a
N
N
N
N
In some cases it was noted that the phosphinic amides were
sensitive to chemical hydrolysis. Many compounds also contained
anilinic or heteroanilinic fragments that may pose a mutagenic
risk.17 This risk could be avoided by replacing the nitrogen with
a carbon giving phosphine oxide 3 (Fig. 4).
Synthetic approaches to a first set of phosphine oxides 3 are de-
scribed in Scheme 2. Diheteroaryl alcohol 4 was activated by a me-
syl group to give 5 and further treated with premixed diethyl
phosphonate and phenylmagnesium bromide to give 3. Alterna-
tively, 3 was made by treating diphenylphosphine oxide with an
iminium ion containing fragments R3 and R4. Structure 3 was also
made by deprotonation of diaromatic alkane 6 and treating the an-
ion with chlorodiphenylphosphine, followed by oxidation with air
oxygen.
3bb
3cb
3db
0.10
0.14
0.75
0.50
2
48
38
N
1.8
NH
14
N
a
b
c
IC50 values in
l
M.
Tested as a racemate.
In percent.
The activities and yields of compounds 3a–3d on Kv1.5 and IKs
are shown in Table 2. Compounds 3b–3d were tested as racemates.
Di-3-pyridyl substitution (3a) gave a highly potent Kv1.5 blocking
compound but with high activity also on IKs. Replacing one 3-pyr-
idyl to a phenyl group gave a more potent Kv1.5 compound (3b)
but with no better selectivity towards IKs. Replacing the phenyl
group with a piperidine (3c) gave a slightly reduced IKs potency,
but still at an unsatisfactory level. A propylamino group (3d) gave
lower IKs activity with Kv1.5 activity mostly retained.
O
P
R5
N
7
High IKs activity has previously been observed in symmetrical
Kv1.5 blockers and has been addressed by desymmetrization of
the chemical structure.18,19 Using the same strategy, we continued
to explore the achiral, less symmetrical phosphine oxide 7 shown
in Figure 5.
Figure 5. General structure of phosphine oxide 7.
For the synthesis of 7 we started with (2-chloropyridin-3-
yl)methanol which was activated by forming the corresponding
mesylate 8 and then subjected to an Arbuzov reaction20 (Scheme 3).
The intermediate chloropyridine 9 was transformed into 7 by a pal-
ladium catalyzed Suzuki reaction.21
The blocking activities and yields of compounds 7a–7d are
shown in Table 3. m-Chlorophenyl substitution (7a) gave a high
Kv1.5 blocking activity but also a high IKs activity. A m-cyano-
phenyl group (7b) was highly active but with no improvement of
IKs selectivity. 3,5-Dichlorophenyl substitution (7c) gave a very
high Kv1.5 blocking activity and also possessed a good IKs
O
P
R4
R3
3
Figure 4. General structure of phosphine oxide 3.

R. I. Olsson et al. / Bioorg. Med. Chem. Lett. 23 (2013) 706–710
709
O
O
the QT-interval (percent of baseline) during the drug exposure per-
iod was also modeled to give a measure of the effects on ventricu-
lar repolarization. The two compounds exemplified in Table 4 both
caused an increase in the ERP at unbound plasma concentrations
S
O
HO
a
below 1 lM, that is, at the same order of magnitude as the IC50 val-
N
Cl
Cl
N
ues from the cellular assay. Compound A, however, also caused a
marked increase in the QT-interval at plasma concentrations in
the same range. Although the explanation for this increase was un-
clear, a block of the ventricular IKs channel might be part of the
mechanism. Reducing the affinity for block of IKs also appeared
to reduce the effect on the QT-interval since compound 2f had
no significant QT-effect. None of the compounds had any measur-
able effect on the hERG channel.
8
b
O
P
O
c
P
The most potent Kv1.5 compound in vitro, 7c, was both hERG
and IKs active and was judged unsuitable for in vivo testing but
compounds 2b, 2e, 2i and 7b were tested in the same PD rabbit
model. However, CNS-like side effects were observed, leading to
incomplete data and indicating that this class of compounds has
a low margin to CNS-like side effects in the rabbit.24
In summary we have discovered a number of diphenylphosphi-
nic amides and diphenylphosphine oxides to be potent inhibitors
of the Kv1.5 potassium channel. In vitro data showed several com-
R5
Cl
N
N
7
9
Scheme 3. Reagents and conditions: (a) MsCl, NEt3, DCM, 0 °C, 30 min, quantita-
tive; (b) ethoxydiphenylphosphine, toluene, 80 °C, 20 h, 42%; (c) R5B(OH)2, K2CO3,
Pd(dppf)Cl2, MeCN/H2O, 130 °C microwave, 15 min, 26–86%.
pounds with Kv1.5 IC50 values of <0.5 lM and with a selectivity
Table 3
over IKs. Two of the compounds that were tested in a rabbit PD
model produced a marked prolongation of ERP in the atrium at
submicromolar concentrations. A low margin to CNS-like side ef-
fects in the rabbit was observed for several members of this class
of compounds.
SAR and yields of phosphine oxide 7
No.
R5
Kv1.5a
0.33
IKsa
4.9
Yieldb
26
7a
Cl
Acknowledgments
We thank the Separation Science Group at AstraZeneca Mölndal
for help with compound purification. We also thank Mark Furber
for useful comments on this manuscript.
7b
0.54
7.7
75
CN
7c
0.08
2.7
13
33
68
86
References and notes
Cl
Cl
1. Khoo, C. W.; Lip, G. Y. Chest 2009, 135, 849.
2. Lip, G. Y.; Tse, H. F. Lancet 2007, 370, 604.
3. Tamargo, J.; Caballero, R.; Gómez, R.; Delpón, E. Expert Opin. Investig. Drugs
2009, 18, 399.
4. Yang, Q.; Wang, X.; Du, L.; Li, M.; You, Q. Expert Opin. Ther. Pat. 2007, 17,
1443.
5. Bilodeau, M. T.; Trotter, B. W. Curr. Top. Med. Chem. 2009, 9, 436.
6. Wulff, H.; Castle, N. A.; Pardo, L. A. Nat. Rev. Drug Disc. 2009, 8, 982.
7. Milnes, J. T.; Madge, D. J.; Ford, J. W. Drug Discovery Today 2012, 17, 654.
8. Dillon, K. B.; Mathey, F.; Nixon, J. F. Phosphorous: The Carbon Copy; Wiley:
Chichester, 1998.
9. Lagrutta, A.; Wang, J.; Fermini, B.; Salata, J. J. J. Pharmacol. Exp. Ther. 2006, 317,
1054.
10. Beshore, D. C.; Liverton, N. J.; McIntyre, C. J.; Claiborne, C. F.; Libby, B.;
Culberson, J. C.; Salata, J. J.; Regan, C. P.; Lynch, J. J.; Kiss, L.; Spencer, R. H.; Kane,
S. A.; White, R. B.; Yeh, S.; Hartman, G. D.; Dinsmore, C. J. Bioorg. Med. Chem.
Lett. 2010, 20, 2493.
11. Lehmann, A.; Antonsson, M.; Holmberg, A. A.; Blackshaw, L.; Bränden, L.;
Bräuner-Osborne, H.; Christiansen, B.; Dent, J.; Elebring, T.; Jacobson, B-M.;
Jensen, J.; Mattsson, J. P.; Nilsson, K.; Oja, S. S.; Page, A. J.; Saransaari, P.; von
Unge, S. J. Pharmacol. Exp. Ther. 2009, 331, 504.
12. Boström, J. ACS Med. Chem. Lett. 2012, 3, 769.
13. ROCS, OpenEye Scientic Software, Santa Fe, NM, USA. http://www.eyesopen.
com.
14. Schroeder, K.; Neagle, B.; Trezise, D. J.; Worley, J. J. Biomol. Screen. 2003, 8,
50.
15. Kutchinsky, J.; Friis, S.; Asmild, M. A.; Taboryski, R.; Pedersen, S.; Vestergaard,
R. K.; Jacobsen, R. B.; Kryzywkowski, K.; Schröder, R. L.; Ljungström, T.; Hélix,
N.; Sörensen, C. B.; Bech, M.; Willumsen, N. J. Assay Drug Dev. Technol. 2003, 1,
685.
7d
N
a
IC50 values in lM.
In percent.
b
selectivity. m-Pyridine substitution (7d) was found to reduce Kv1.5
potency.
The pharmacodynamic effects of a few selected compounds
were tested in a modified rabbit PD-model22 which measured the
effective refractory period (ERP) in the atrium as well as the QT-
interval. The animals (male, New Zealand white rabbits) were
administered with
15 min infusions, followed by a wash-out period. The doses ranged
3 increasing doses given as consecutive
between 0.1 and 1 lmol/kg/min. PK/PD modeling using link mod-
els23 on effect-plasma concentration data was applied to estimate
the unbound steady-state plasma concentration (Ceu20) giving a
20% change in the atrium ERP. The maximum observed change in
Table 4
In vivo results
16. Additional data for compound A: Clint (HLM) = 158
lL/min  mg protein,
solubility (pH 7.4) = 91
lM, hERG IC50 >33
l
M, Caco-2-Papp = 30 Â 10–6 cm/s,
Rat CL = 11 mL/min  kg, Rat oral bioavailability = 55%.
Compound
Ceu20
(
l
M)
QTmax (% change)
17. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 99;
International Agency for Research on Cancer: Lyon, France, 2010; p 1.
18. Trotter, W. B.; Nanda, K. K.; Wolkenberg, S.; Nolt, B. M.; Manley, P.; Kett, N. R.;
Bilodeau, M. T. PCT Int. Appl. WO 2007/050348.
A
2f
0.8
0.6
23
<10

710
R. I. Olsson et al. / Bioorg. Med. Chem. Lett. 23 (2013) 706–710
19. Trotter, W. B. Abstract of Papers, 234th ACS National Meeting, Boston, MA,
United States, 2007, MEDI-449.
20. Bhattacharya, A. K.; Thyagarajan, G. Chem. Rev. 1981, 81, 415.
21. Miyaura, N.; Suzuki, A. Chem. Rev. 1995, 95, 2457.
22. Sun, H.; Lloyd, J.; Shi, H.; Li, D.; Levesque, P. Heart Rhythm 2008, 5, 213.
23. Ollerstam, A.; Visser, S. A. G.; Persson, A. H.; Eklund, G.; Nilsson, L. B.; Forsberg,
T.; Wiklund, S. J.; Gabrielsson, J.; Duker, G.; Al-Saffar, A. J. Pharmacol. Toxicol.
Methods 2006, 53, 174.
24. The CNS-like side effects ranged from tremors and salivation to occasional
seizures. No clear correlations between side effects and pharmacokinetics or
exposure were observed. Compounds A and 2f showed mild side effects at
higher plasma concentrations (3–4 times Ceu20), but without interrupting the
experiments. Brain penetration was not investigated for this series, but has
been investigated for other Kv1.5 blockers when CNS-like side effects were
observed, see Ref. 10.