The identification, and optimisation of hERG selectivity, of a mixed NET/SERT re-uptake inhibitor for the treatment of pain

Article abstract of DOI:10.1016/j.bmcl.2010.11.021

Hit compound 1, a selective noradrenaline re-uptake transporter (NET) inhibitor was optimised to build in potency at the serotonin re-uptake transporter (SERT) whilst maintaining selectivity against the dopamine re-uptake transporter (DAT). During the opt

Full text of DOI:10.1016/j.bmcl.2010.11.021

Bioorganic & Medicinal Chemistry Letters 21 (2011) 271–275
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
The identification, and optimisation of hERG selectivity, of a mixed NET/SERT
re-uptake inhibitor for the treatment of pain
Derek Angus c, Matilda Bingham a, , Dawn Buchanan b, Neil Dunbar a, Linsday Gibson a, Richard Goodwin a,
⇑
Anders Haunsø b, Andrea Houghton c, Margaret Huggett a, Richard Morphy a, Susan Napier a,*, Olaf Nimz a,
Joanna Passmore a, Glenn Walker b
a Department of Medicinal Chemistry, MSD, Newhouse, Motherwell, ML1 5SH, United Kingdom
b Department of Molecular Pharmacology, MSD, Newhouse, Motherwell, ML1 5SH, United Kingdom
c Department of Pharmacology, MSD, Newhouse, Motherwell, ML1 5SH, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
Hit compound 1, a selective noradrenaline re-uptake transporter (NET) inhibitor was optimised to build
in potency at the serotonin re-uptake transporter (SERT) whilst maintaining selectivity against the dopa-
mine re-uptake transporter (DAT). During the optimisation of 1 it became clear that selectivity against
the Kv11.1 potassium ion channel (hERG) was also a parameter for optimisation within the series. Dis-
crete structural changes to the molecule as well as a lowering of global cLogP successfully increased
the hERG selectivity to afford compound 11m, which was efficacious in a mouse model of inflammatory
pain, complete Freund’s adjuvant (CFA) induced thermal hyperalgesia and a rat model of neuropathic
pain, spinal nerve ligation (SNL) induced mechanical allodynia.
Received 17 September 2010
Revised 30 October 2010
Accepted 2 November 2010
Available online 6 November 2010
Keywords:
Noradrenaline re-uptake transporter
inhibitor
Serotonin re-uptake transporter inhibitor
Monoamine re-uptake transporter inhibitor
Ó 2010 Elsevier Ltd. All rights reserved.
A number of antidepressant medications have demonstrated
efficacy in treating pain associated with nerve damage (neuro-
pathic pain). In particular, the tricyclic antidepressants (TCAs) have
proven clinical efficacy; one third of patients with neuropathic
pain who took a TCA such as amitriptyline obtained moderate pain
relief or better.1 Although the TCAs are classical ‘dirty’ compounds
with binding affinity at a range of receptor targets, there is a grow-
ing body of evidence which attributes the efficacy of the TCAs in
neuropathic pain to the inhibition of serotonin and noradrenaline
re-uptake at pre-synaptic transporters. This is further substanti-
ated by evidence that the newer class of antidepressants venlafax-
ine and duloxetine (Fig. 1) have similar effectiveness to the TCAs
with reduced side effects.2 Both venlafaxine and duloxetine inhibit
re-uptake of both of the key neurotransmitters, serotonin and nor-
adrenaline, involved in the descending modulation of neuropathic
pain.3 Venlafaxine and duloxetine are serotonin–noradrenaline
re-uptake inhibitors or SNRIs with an inhibitor profile of SERT >
NET >> DAT, however of the two compounds only duloxetine is
registered for the treatment of neuropathic pain by the FDA.
The most frequently reported adverse events for duloxetine
include nausea, somnolence, dizziness, and fatigue and it is likely
these symptoms are caused by the elevation of serotonin rather
S
N
N
H
H
O
OH
N
H
O
venlafaxine
duloxetine
amitriptyline
Figure 1.
than noradrenaline. Thus, we hypothesised that the optimal mixed
re-uptake inhibitor profile for achieving the desired analgesic
efficacy, without the side effects commonly attributed to serotonin
re-uptake, would be an inhibitor with a potency profile of NET >
SERT >> DAT; an ‘NSRI’ re-uptake inhibitor with reduced potency
at SERT.
An in-house transporter project looking at triple re-uptake
inhibitors had identified a number of interesting leads including
compound 1 which was selected as a start point for our NSRI pro-
gramme (Fig. 2). Compound 1 was a selective NET inhibitor and
thus the challenge was to dial in SERT affinity without compro-
mising the DAT selectivity and NET potency. During the optimisa-
tion of 1 it became clear that hERG selectivity was also a
⇑
Corresponding authors. Tel.: +44 01698736000.
E-mail addresses: matilda.bingham@merck.com, matilda.bingham@virginme-
dia.com (M. Bingham), susan.napier@merck.com (S. Napier).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.11.021

272
D. Angus et al. / Bioorg. Med. Chem. Lett. 21 (2011) 271–275
Table 1
NET, DAT and SERT re-uptake assay results for compounds 10a–10m
R1
1
N
N
SERT IC50 7.6µM
NET IC50 26nM
DAT IC50 >10µM
hERG Ki 1.3 µM
O
O
NH
10a-m
R
Figure 2.
R1
R
NETa (IC50/nM) SERTa (IC50/nM) DATa (IC50/nM)
10a
10b
10c
10d
10e
10f
10g
10h
10i
H
Me
H
H
H
H
H
H
H
H
H
54 32
86
347 203
19
188 142
5655 2235
>10,000
737 198
437 200
2261 94
380 52
1318 127
201 33
804 141
750 141
2427 28
5
R1
p-Chloro
m-Chloro
o-Chloro
m-Bromo
m-Iodo
m-Fluoro
m-
Methyl
m-Nitro
m-CF3
m-OMe
m-CN
43
6
R1
X
7
135 46
881 166
136 62
106 37
1245 136
624 190
R2
R2
13
12
4
3
OH
HO
X
O
67 16
92 28
Y
3
Y
N
N
R
Boc
a)
10j
10k
10l
H
H
H
H
91 15
128 51
222 111
126 56
86 30
99 17
369 49
2010 584
3861 426
2775 771
1404 378
4
X = CH or N
2
Y = CH or N
R1 and R2 = H, F, Cl, Br, CN, CF3
10m
25
4
a
Values are mean SEM of at least three independent experiments carried out in
Scheme 1. Reagents and conditions: (a) PPh3, DEAD, THF, rt, 18 h.
triplicate.
parameter for optimisation within the series. We report herein
the results of those optimisation studies, including the observa-
tion that discrete structure–activity relationships (SAR) at hERG
could be used to dial out hERG affinity without significant
changes to global cLogP.
Scheme 1 illustrates a typical synthetic route which was used to
prepare analogues 4.4 Where phenol or pyridinone derivatives 3
were available, or readily accessible from Suzuki-Miyaura coupling
of commercially available 3-bromophenols or 6-bromopyridones,
the final compounds 4 were prepared via mitsunobu reaction of
compounds 3 with Boc protected endo-8-azabicyclo[3.2.1]octan-
3-ol 2.
In cases where the requisite phenol or pyridone 3 was not avail-
able then the final compounds were prepared from Boc protected
exo-8-azabicyclo[3.2.1]octan-3-ol 6, using an alkylation, borylation
sequence to obtain the intermediate boronic acid as illustrated for
compound 8. The final compound 9 was synthesized via a Suzuki-
Miyaura coupling of 8 with 3-cyano, 2-chloropyridine and subse-
quent Boc deprotection (Scheme 2).
The compounds thus prepared were assayed using whole-cell
re-uptake assays as described by Haunso and Buchanan5 using
MDCK cells overexpressing NET, and HEK293 cells overexpressing
DAT and SERT. Inhibition of re-uptake of 3H-noradrenaline,
3H-dopamine or 3H-serotonin, respectively, was then measured.6
The human Kv11.1 potassium ion channel (hERG) dofetilide
binding assay was carried out as previously described.7
Studies in our laboratories with substituted monoaryl tropane
ethers such as 10, had indicated that the transporter profile could
be modified by appropriate selection of substitution pattern on the
aromatic ring (Table 1).
In particular we had observed that potency at SERT could be in-
creased by substitution at the meta- or para-position of the aryl
ring. In the case of para-substitution as in example 10c this was
F
O
HO
O
O
OH
O
OH
b)
Br
c)
O
N+
O–
N
N
+
N
N
d)
Boc
Boc
Boc
N+
2
–O
5
6
N
N
N
NC
Cl
HCl
N
CN
OH
f), g)
O
e)
O
Br
B
O
OH
N
H
N
N
N
9
8
7
Boc
Boc
Scheme 2. Reagents and conditions: (b) PPh3, DEAD, THF, rt, 18 h, 75%; (c) 4 N NaOH (aq), THF, 72 h, 89%; (d) NaH, DMF, rt then
Pr)3 À78 °C to rt (ii) HCl 2 M aq 90%; (f) Pd2dba3, K3PO4, PCy3, dioxane/water 2.5:1 100 °C, 30 min, 59%; (g) (i) TFA, DCM (ii) SCX (iii) HCl/diethyl ether 2 M 87%.
lw 180 °C, 900 s; (e) (i) n-BuLi À78 °C, B(Oi-

D. Angus et al. / Bioorg. Med. Chem. Lett. 21 (2011) 271–275
273
Table 2
groups in the meta-position were preferred by the SERT with the
nitrile 10m as the optimal substituent. Our strategy for the optimi-
sation of hit 1 was thus to see if this SAR could be translated to the
biaryl series via the synthesis of compounds such as 11a–k. For
synthetic expediency the initial R1 SAR screen was carried out on
the bi-phenyl series (Table 2). We were delighted to observe that
the SAR did appear to translate; again the most pronounced boost
in SERT potency was seen for the chloro and nitrile analogues 11c
and 11e,f, with the nitrile affording a ꢀ37-fold increase in SERT
affinity (11f vs 1).
We also carried out a SAR screen in the second phenyl ring B,
illustrative examples 11g–i are given in Table 2. In this case the po-
tency at SERT was most enhanced by para-substitution 11g,
whereas the NET potency could be significantly boosted by addi-
tion of an ortho-substituent 11i. Importantly this SAR was tolerant
of re-introduction of the A-ring pyridyl nitrogen from hit 1 as in
compound 11j and was additive with the SAR for the A-ring, since
re-introduction of the nitrile R1 again boosted the SERT potency
and afforded a potent NSRI 11k.
A ubiquitous issue in the development of ligands for mono-
amine receptors is hERG selectivity. Mindful of this liability, we
introduced the hERG dofetilide binding assay7 early in our screen-
ing cascade. It quickly became evident that the biaryl series based
around hit 1 has inherent hERG affinity and our optimisation goals
were thus focussed on four assays SERT, NET and the selectivity
assays, DAT and hERG. As is frequently observed,8 we identified a
positive correlation between cLogP and hERG dofetilide binding
within the series and a negative correlation with polar surface area
(Fig. 3).
NET, DAT and SERT re-uptake assay results for compounds
H
N
R2
B
O
Y
A
11a-k
R1
Y
R1
R2
NETa
(IC50/nM)
SERTa
(IC50/nM)
DATa
(IC50/nM)
1
N
H
H
H
H
H
H
H
H
p-CN
m-CN
o-CN
o-CN
o-CN
26 10
891 166
132 63
30 10
85
148 73
90 32
7644 1166
4624 593
2610 1359
452 129
905 360
100 14
208 34
105 29
<10,000
2550 216
8981 1019
>10,000
>10,000
>10,000
11a
11b
11c
11d
11e
11f
11g
11h
11i
11j
11k
CH
CH
CH
CH
CH
N
CH
CH
CH
N
CF3
Br
Cl
F
CN
CN
H
H
H
H
CN
5329 1723
8710 1290
9550 282
>10,000
173 100
481 49
8066 1934
>10,000
>10,000
8
60
31 19
9
5
9
2
1
2
N
13
48
6
a
Values are mean SEM of at least three independent experiments carried out in
triplicate.
Over and above this correlation we also identified some discrete
hERG SAR which could be used to modify the hERG selectivity (Ta-
ble 3). In particular we observed that moving the nitrile to the sec-
ond aryl ring as in 11j reduced the hERG affinity by 60-fold,
without a significant effect on the cLogP. Indeed compound 11j is
an interesting compound in its own right as a very potent NRI.
We also observed that moving the nitrogen from the A-ring to
the B-ring as in compound 11l lead to a reduction in hERG potency,
but this time retaining the SERT potency. The effect can perhaps be
attributed to a change in the pKa of the pyridine conjugate acid.
Combining the above changes into a single compound 11m, affor-
ded a potent NSRI with reduced hERG affinity relative to 11f.
Compound 11m had the best balance of NET/SERT potency and
selectivity in the hERG dofetilide assay and was selected for further
profiling. The compound was screened against our in-house selec-
tivity panel and was selective against a range of receptors (includ-
ing <60% inhibition at 10
l
M for M1, M2, M3, M4, 5HT1a, 5HT2b
,
5HT2c a1A a2a D2 and H1 and 64% at 10
,
lM for 5HT2a). The com-
pound had moderate Caco2 permeability (A–B 178 nm/s and B–A
402 nm/s ER 2.2) and a brain/plasma ratio of 1.5 (1 mpk iv, ICR
male mice) predictive of reasonable brain exposure. In vitro micro-
some and hepatocyte assays predicted low intrinsic clearance for
the majority of analogues within the series and 11m was no excep-
tion (rat Clint <12 lL/min/mg protein, and human Clint <12 lL/
min/mg protein in microsomes) however, this did not translate
into the in vivo clearance which was measured as 103 mL/min/
Kg for compound 11m. Compound 11m was shown to have a
blood/plasma ratio of 3:1 and a high Vss which may in part explain
this discrepancy.
Compound 11m was selected as a tool compound to evaluate
the NSRI profile in our in-house in vivo pain models. The com-
pound was tested in a mouse model of inflammatory pain, com-
plete Freund’s adjuvant (CFA) induced thermal hyperalgesia and
a rat model of neuropathic pain, spinal nerve ligation (SNL)
induced mechanical allodynia. In the mouse CFA model 11m com-
pletely reversed the thermal hyperalgesia induced by CFA after oral
Figure 3. Graph to show the correlation between cLogP and hERG dofetilide
binding, and the inverse relationship between polar surface area and hERG
dofetilide binding. Data points are shaded to reflect polar surface area increasing
from red to green.
usually accompanied by a reduction in NET affinity which ran
counter to our desired NSRI-like profile. However, in the case of
m-substitution as in 10d the SERT potency could be raised without
a loss of affinity at the NET transporter. Electron withdrawing
administration of 100 lmol/kg. Similarly 11m dose dependently

274
D. Angus et al. / Bioorg. Med. Chem. Lett. 21 (2011) 271–275
Table 3
hERG dofetilide binding of analogues 11f, j, l, k and m
H
N
NC
H
N
O
N
H
N
11j
O
N
N
O
H
N
11f
CN
CN
N
11m
CN
O
11l
CN
Compound
hERG dofetilide (Ki/nM)
cLogP
NETa (IC50/nM)
90 33
SERTa
DATa (IC50/nM)
11f
11j
11l
11k
11m
119 14
3.06
2.95
2.39
2.50
2.01
208 34
8981 1019
169 45
94
101 11
>10,000
>10,000
>10,000
>10,000
>10,000
7080 864
1621 834
251 147
9
1
111 88
13
23
4
8
7
3802 1130
a
Values are mean SEM of at least three independent experiments carried out in triplicate.
model,9 however, in our hands, SRIs (e.g., fluoxetine) are not active,
indicating that noradrenaline re-uptake inhibition is key for anal-
gesic efficacy in this model. Addition of the component of seroto-
nin re-uptake inhibition as in compound 11m may confer
additional benefits compared with NRIs hence 11m provides a tool
compound for wider pharmacological profiling with comparator
NRI/SNRI/SSRI compounds in neurochemistry studies and neuro-
pathic pain models (Table 4).
15
12
9
6
InsummarywehaveoptimisedtheNETselectivehit 1toaffordan
NSRI 11m with reduced hERG liability compared with the starting
compound. The key discovery which allowed us to build in the SERT
inhibition, without introducing DAT inhibition was the incorpora-
tionof a nitrile inthe meta-positionof the A-ring. Thechemicalseries
was optimised further to dial out hERG binding affinity, via a dual
strategy of global cLogP reduction and discrete structural changes,
to affordcompound11mwhichhaddemonstrated efficacyinanimal
models of inflammatory and neuropathic pain. Further characterisa-
tion of compound 11m identified that the window between the esti-
mated peak plasma exposure at the MED (minimum effective dose)
and adverse effects in the Langendorf model,10 a model of cardiovas-
cular conductance, was not sufficient and development of this series
was stopped in favour of more promising chemotypes. Compound
11m is however a useful tool for evaluating the NET > SERT >> DAT
profile in neuropathic pain models.
3
0
Pre-Spinal
Nerve Ligation
0
30
Time after oral administration of 11m (min)
10µmol/kg 11m 30µmol/kg 11m 100µmol/kg 11m
60
120
vehicle
Figure 4. Effect of compound 11m in a rat model of neuropathic pain, spinal nerve
ligation (SNL) induced mechanical allodynia.
attenuated the mechanical allodynia induced by SNL. This effect
was significant compared to vehicle treated animals after a dose
of 30 or 100 lmol/kg PO (Fig. 4). Both NRIs and to a more modest
extent SNRIs (e.g., duloxetine) have proven efficacy in the SNL
Table 4
In vivo DMPK data for 11m
1 mg/Kg iva
10 mg/Kg p.oa
Vss (L/Kg)
Cl (mL/min/kg)
103
AUC (ng/mL h)
563
F (%)
Tmax (h)
t1/2 (h)
11m
6.9
34
1.7
3.2
a
Male wistar rat.

D. Angus et al. / Bioorg. Med. Chem. Lett. 21 (2011) 271–275
275
20 nM) for 3, 5 and 10 min for DAT, SERT and NET, respectively, termination of
the assay by cold HBSS wash and addition of MicroScint-20 scintillation
mixture to determine 3H-monoamine uptake was carried out. All incubations
were performed at room temperature except for NET cells which were
incubated at 37 °C. IC50 values were determined form 6-point concentration–
response curves from three independent experiments carried out in triplicates.
7. Finlayson, K.; Turnbull, L.; January, C. T.; Sharkey, J.; Kelly, J. S. Eur. J. Pharmacol.
2001, 430, 147.
8. Jamieson, C.; Moir, E. M.; Rankovic, Z.; Wishart, G. J. Med. Chem. 2006,
49, 5029.
9. Vu, A. T.; Cohn, S. T.; Zhang, P.; Kim, C. Y.; Mahaney, P. E.; Bray, J. A.; Johnston,
G. H.; Koury, E. J.; Cosmi, S. A.; Deecher, D. C.; Smith, V. A.; Harrison, J. E.;
Leventhal, L.; Whiteside, G. T.; Kennedy, J. D.; Trybulsk, E. J. J. Med. Chem. 2010,
53, 2051.
References and notes
1. McQuay, H. J.; Tramér, M.; Nye, B. A.; Carroll, D.; Wiffen, P. J.; Moore, R. A. Pain
1996, 68, 217.
2. Jann, M. W.; Slade, J. H. Pharmacotherapy 2007, 27, 1571; Wernicke, J. F.;
Pritchett, Y. L.; D’Souza, D. N.; Waninger, A.; Tran, P.; Iyengar, S.; Raskin, J. A.
Neurology 2006, 67, 1411.
3. Basbaum, A. I.; Fields, H. L. Annu. Rev. Neurosci. 1984, 7, 309.
4. Napier, S. E.; Bingham, M. J.; Dunbar, N. A. WO 2007063071 A1, 2007.
5. Haunso, A.; Buchanan, D. J. Biomol. Screening 2007, 12, 378.
6. Cells were seeded the day before the experiment at 35,000 cells/well in 96 well
plates. On the day of the experiment, cells were washed and pre-incubated
with assay buffer (HBSS supplemented with pargyline and ascorbic acid)
containing monoamine transport inhibitors for 5 min. Subsequent incubation
with 3H-dopamine, 3H-serotonin or 3H-noradrenaline (final concentration of
10. Hondeghem, L. M.; Carlsson, L.; Duker, G. Circulation 2001, 103, 2004.