Discovery of ONO-8590580: A novel, potent and selective GABAA α5 negative allosteric modulator for the treatment of cognitive disorders

Article abstract of DOI:10.1016/j.bmcl.2020.127536

The identification and SAR development of a series of negative allosteric modulators of the GABA_A α5 receptor is described. This novel series of compounds was optimised to provide analogues with high GABA_A α5 binding affinity, high α5 negative allosteric modulatory activity, good functional subtype selectivity and low microsomal turnover, culminating in identification of ONO-8590580.

Full text of DOI:10.1016/j.bmcl.2020.127536

Bioorganic&MedicinalChemistryLetters30(2020)127536
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of ONO-8590580: A novel, potent and selective GABAA α5
negative allosteric modulator for the treatment of cognitive disorders
A. Lewisa, A. Beresforda, M.S. Chambersa, G. Clarka, D.C. Hartleya, K.L. Hirsta, M. Higashinob,
S. Kawahadarab, M. Nakanishib, T. Saitob, A. Imagawab, H. Habashitab, S. Maidmenta,
⁎
A.M. Macleoda, A.P. Owensa, A. Raea, , C. Rousea, G. Wisharta
a Charles River Discovery Research Services, Chesterford Research Park, Little Chesterford, Essex CB10 1XL, United Kingdom
b Ono Pharmaceutical Co. Ltd., Minase Research Institute, Shimamoto-cho, Mishima-gun, Osaka 618-8585, Japan
A R T I C L E I N F O
A B S T R A C T
Keywords:
The identification and SAR development of a series of negative allosteric modulators of the GABAA α5 receptor is
described. This novel series of compounds was optimised to provide analogues with high GABAA α5 binding
affinity, high α5 negative allosteric modulatory activity, good functional subtype selectivity and low microsomal
turnover, culminating in identification of ONO-8590580.
GABAA α5
Negative allosteric modulator
Cognition enhancement
ONO-8590580
GABAA receptors are pentameric membrane proteins that gate
chloride ions in response to the binding of γ-aminobutyric acid (GABA).
GABAA ion channels containing the α5-subunit (GABAA α5) are highly
expressed in the hippocampus, a key area of the brain associated with
cognitive processing,1 and it has previously been shown that GABAA α5
negative allosteric modulators (NAMs) are efficacious in models
thought to reflect pro-cognitive properties in rodents.2,3,4 Furthermore,
mice with a disrupted α5 gene (α5 −/− mice) show improved per-
formance in the spatial learning water maze test compared with wild-
type mice.5 Taken together, these data support the hypothesis that
GABAA α5 NAMs may offer therapeutic potential for the treatment of
cognitive disorders, such as Down’s syndrome and Alzheimer’s disease.
However, in adopting this therapeutic approach it is essential that the
GABAA α5 NAM should demonstrate binding or functional selectivity
for the GABAA α5-subunit, as negative modulation of other GABAA
subunits expressed in the CNS can lead to unwanted pro-convulsant
(GABAA α1) or anxiogenic (GABAA α2 or GABAA α3) effects.6,7
progression of α5IA was terminated due to renal toxicity in pre-clinical
species.12 MRK-016 entered Phase I clinical evaluation but was subse-
quently terminated due to poor tolerability in elderly patients and
variable pharmacokinetics.2
In order to identify starting points for a GABAA α5 NAM drug dis-
covery program, approximately 124,000 compounds were screened
from the Charles River Laboratories screening collection, at a con-
centration of 10 μM, in both receptor binding and FLIPR functional
assays against human GABAA α5β3γ2 expressed in HEK293 cells. In the
high throughput screen (HTS), receptor binding was assessed by mea-
suring the inhibition of binding of commercially available [3H]-Ro15-
1788 ([3H]-flumazenil), and in the FLIPR assays an EC20 concentration
of GABA was co-applied with the test compound, in order to determine
the functional response.13 Following the HTS, 525 compounds
showed > 40% inhibition at 10 μM in both the binding and FLIPR as-
says, which were progressed to determination of binding Ki and con-
centration-response functional activity against GABAA α5β3γ2. In the
FLIPR assay, 26 of the 525 compounds demonstrated an IC50 value
against GABAA α5 of < 1 μM, which were progressed to evaluation of
functional selectivity against the GABAA α1, α2 and α3 subtypes. A
promising hit from this screening exercise was imidazopyridine 1,
which demonstrated moderate binding affinity, significant negative
modulation of the GABA response at the GABAA α5 receptor, and
minimal effect on the GABA response at the α1, α2 and α3 subtypes
(Table 1). However, 1 was rapidly metabolised in both rat and human
liver microsomes. Consequently, a program of work was initiated
This approach to cognition enhancement has attracted considerable
attention in recent years, with several GABAA α5 NAMs, including
α5IA, MRK-016 and basmisanil reaching clinical evaluation
(Fig. 1).2,3,4,8,9 In Down’s Syndrome subjects, basmisanil did not de-
monstrate efficacy in the cognition primary end-points10 and has sub-
sequently been withdrawn from Roche’s pipeline. Merck’s α5IA was
evaluated for cognitive performance in ethanol-impaired volunteers,
and those subjects on drug demonstrated a significant improvement in a
list recall experiment compared to those on placebo.11 However,
⁎ Corresponding author.
E-mail address: alastair.rae@crl.com (A. Rae).
https://doi.org/10.1016/j.bmcl.2020.127536
Received 11 May 2020; Received in revised form 28 August 2020; Accepted 31 August 2020
Availableonline06September2020
0960-894X/©2020ElsevierLtd.Allrightsreserved.

A. Lewis, et al.
Bioorganic&MedicinalChemistryLetters30(2020)127536
Fig. 1. α5IA, MRK-016 and Basmisanil.
and, importantly, 4-fold higher GABAA α5 binding affinity and sig-
nificantly improved rat and human microsomal stability compared to 1.
Consequently, 8 became the focus of further optimisation.
Table 1
Binding affinity, functional efficacy and microsomal stability data for imida-
zopyridine 1.
A variety of structural modifications to the benzoxazinone sub-
stituent were investigated and key analogues are shown in Table 3. The
first set of compounds 9–14 were designed to probe the structure–-
activity relationship around the oxazinone ring. The secondary amide 9
retained high stability following incubation with human liver micro-
somes, and also provided improved rat microsomal stability, possibly as
a result of eliminating N-de-methylation as a route of metabolism.
However, this change also resulted in an undesirable elevation of the
levels of negative modulation at α1 and α2. The isomeric benzox-
azinone 10 maintained a similar profile to 8, albeit with 4-fold reduced
binding affinity, whereas the ring contraction analogue – benzox-
azolone 11 – suffered from an unacceptable functional selectivity pro-
file, and inferior metabolic stability compared to 8 or 9. Lactam 12
retained good binding affinity but was a positive modulator at α1. In-
troduction of a methyl group adjacent to the carbonyl group of the
oxazinone ring to afford compound 13 was well tolerated for GABAA
α5 affinity and enhanced rat microsomal stability compared to the
parent compound 8. The difluorinated benzoxazinone 14 afforded an
excellent functional selectivity profile with good affinity against GABAA
α5, although microsomal stability was sub-optimal. To assist with
identification of the main routes of metabolism in this series and inform
the design of more stable analogues, benzoxazinone 14 was subjected to
a metabolite ID study in human liver microsomes. This study indicated
de-methylation of the amide and oxidation of either the aryl ring or
imidazopyridine scaffold as primary metabolic pathways. A key
strategy towards investigating the potential for oxidative aryl metabo-
lism of the benzoxazinone group was to reduce the electron density of
the phenyl ring by replacing phenyl with pyridyl to afford the aza-
benzoxazinones 15–17. This approach proved to be successful, with all
three isomers retaining high stability in the presence of human liver
microsomes, and enhanced stability in rat liver microsomes compared
to their phenyl counterpart 8. Of these three isomers it was aza-
benzoxazinone 17 that showed the best overall profile, with high sta-
bility in both human and rat liver microsomes, high GABAA α5 negative
allosteric modulation and no significant functional response against the
other key GABAA subtypes. Consequently, this optimised substituent
was retained in a further program of work around the imidazopyridine
scaffold, with the objective of increasing binding affinity to maximise
the likelihood of achieving high receptor occupancy in vivo.
GABAA α5
Ki (nM)2
76
Efficacy1 (%)
Half-life in liver microsomes (min)
α5
α1
α2
α3
Rat
Human
14
−42
−1
−7
−6
< 5
1Effect of 1 μM compound on the current induced by the EC20 concentration of
GABA on human recombinant GABAA α5, α1, α2, or α3 subunits expressed in
HEK293 cells. Current recordings were performed using PatchXpress
(Molecular Devices, LLC). A detailed protocol for the efficacy assays is de-
scribed in the Supplementary Information.
2In vitro binding affinity at the human GABAA α5β3γ2 receptor, measured using
a scintillation proximity assay (radioligand [3H]-Ro15-1788). A detailed pro-
tocol for this binding assay is described in the Supplementary Information.
around 1, targeting increased GABAA α5 receptor affinity and improved
microsomal stability, with retention of GABAA α5 functional selectivity.
Importantly, analogues from the imidazopyridine series exemplified by
1 did not show any binding selectivity for α5 versus the other main
GABAA subtypes, and consequently the program strategy was to invoke
subtype selectivity in a functional manner.
The electron-rich dimethoxyphenyl substituent in 1 was surmised to
be a potential metabolic liability, with susceptibility to O-demethyla-
tion and oxidation of the aryl ring. Therefore, the SAR around the aryl
group was investigated as a priority, with emphasis on reduction of
electron density in the ring and/or lowering of the molecule’s lipo-
philicity, targeting analogues with retained or improved binding affi-
nity and improved microsomal stability. Some specific examples syn-
thesised for the dimethoxy replacement program are shown in Table 2.
Removal of one or both of the methoxy substituents to lower electron
density on the ring (compounds 2 and 3) resulted in a significant loss of
binding affinity. Aliphatic substituents such as cyclopropylmethyl
(compound 4) were not tolerated. An alternative electron-donating
substituent that afforded a reduction in lipophilicity was the morpho-
line of analogue 5, which had similar binding affinity to 1 and some-
what improved metabolic stability, although this benefit came with a
cost of reduced GABAA α5 negative allosteric modulation. Introducing
the more polar carboxamide group (6) retained the acceptable GABAA
α5 inverse agonism of 1 but had 4-fold weaker GABAA α5 affinity.
However, by incorporating the hydrogen-bond acceptor group in a
fused bicycle to afford the benzoxazole 7 or benzoxazinone 8 provided
analogues with very encouraging in vitro profiles. Most notable was the
benzoxazinone 8, which had a reasonable functional selectivity profile
Earlier investigation of SAR in the 1H-imidazo[4,5-c]pyridine series
had shown that modifications to the imidazole ring such as nitrogen
deletion or substitution at C2 were not tolerated, and that alternative
substituents to cyclopropyl at N1 did not confer improvements to the
overall profile. Hence, subsequent analogue synthesis was aimed to-
wards retaining the attractive microsomal stability and GABAA α5
functional selectivity profile of compound 17 and improving the α5
affinity by evaluating modifications to the six-membered ring of the
imidazopyridine scaffold (Table 4).
Methylation (18) at the C4 position of the scaffold maintained
2

A. Lewis, et al.
Bioorganic&MedicinalChemistryLetters30(2020)127536
Table 2
In vitro data for selected C6-substituted imidazopyridines.
R
ID
cLogP
4.0
GABAA α5 Ki (nM)
76
Efficacy (%)
α5
Half-life in liver microsomes (min)
α1
α2
α3
Rat
Human
14
1
−42
−1
−7
−6
< 5
2
3
3.8
3.5
459
972
−43
−26
−20
–
−20
–
3
–
–
–
–
–
4
5
3.1
3.1
> 15,000
119
–
–
–
–
–
–
–
–
–
−26
36
23
6
2.9
319
−44
−11
−11
−11
–
–
7
8
3.2
2.8
85
17
−42
−46
−16
−8
−10
−2
−6
10
19
38
−17
> 100
binding affinity relative to 17, but the functional selectivity profile for
this 4-alkylated 1H-imidazo[4,5-c]pyridines was inferior. The isomeric
3H-imidazo[4,5-b]pyridine core exemplified in analogues 19–21 also
provided enhanced GABAA α5 binding affinity (Ki 2–19 nM) compared
to 17, but their functional or microsomal stability profiles were less
attractive.
25 had high GABAA α5 inverse agonism and essentially negligible
functional activity at either α1, α2 or α3. However, compound 25 ex-
hibited a relatively poor microsomal stability profile. As had been ob-
served when comparing the purines 22 and 23, C4-methylation also
enhanced metabolic stability in the benzimidazole series, with com-
pound 26 demonstrating improved rat microsomal stability compared
to 25. Indeed, 26 had an encouraging in vitro profile, with a GABAA α5
Ki of 7 nM, high α5 inverse agonism (−49%) and good functional se-
lectivity over the other GABAA subtypes.
Introducing further polarity into the central core to afford the 9H-
purine scaffolds provided compounds 22 and 23 which had favourable
α5 affinity compared to 17, but sub-optimal microsomal stability and
GABAA α5 functional selectivity profiles.
As the results in Table 4 demonstrate, the substituted benzimida-
zoles (24–26) exhibited high GABAA α5 binding affinity and GABAA α5
NAM activity, and in the case of 25 and 26, a good functional selectivity
profile. Consequently, in an effort to understand the major routes of
metabolism in this class, the 5-fluorobenzimidazole 25 was subjected to
a metabolite ID study. As previously observed for the 1H-imidazo[4,5-c]
pyridine 14, the major in vitro metabolites for benzimidazole 25 were
derived from N-dealkylation or oxidation of the azabenzoxazinone unit.
Hence, further efforts were instigated to find replacements for the C6
substituent, focussing on the 4-methyl-5-fluorobenzimidazolyl (26),
To complete SAR development around the scaffold, an array of
benzimidazoles was also synthesised. Within this sub-series, the 5-tri-
fluoromethoxy- (24) 5-fluoro- (25) and 4-methyl-5-fluoro analogues
demonstrated improved GABAA α5 binding affinity compared to 17, the
most notable of which was 24, which exhibited sub-nanomolar GABAA
α5 affinity (Ki 0.5 nM). Unfortunately, compound 24 demonstrated
significant negative modulation against the GABAA α2 (−16%) and
GABAA α3 (−31%) subtype, thus rendering it liable to the undesired
side-effects demonstrated by non-selective GABAA NAMs.6,7 In contrast,
3

A. Lewis, et al.
Bioorganic&MedicinalChemistryLetters30(2020)127536
Table 3
In vitro data for selected analogues incorporating modifications to the C6 benzoxazinone substituent.
R
ID
GABAA α5 Ki (nM)
17
Efficacy (%)
α5
Half-life in liver microsomes (min)
α1
α2
α3
Rat
19
Human
> 100
8
−46
−8
−2
−17
9
42
65
−48
−41
−18
0
–32
−18
−13
59
25
> 100
77
10
−10
11
12
13
14
62
16
17
11
−50
−38
−40
−55
−31
+32
+19
+1
–
−21
+13
+60
+12
8
73
–
25
55
22
70
+54
−1
> 100
52
15
16
17
45
11
38
−46
−45
−52
−22
−45
0
−39
−27
−8
−27
−26
−8
39
93
45
> 100
> 100
> 100
and 4-methyl 3H-imidazo[4,5-b]pyridyl (20) cores, both of which had
previously demonstrated high α5 binding affinity, and strong negative
modulation against α5.
variety of pendent heterocycles, in a manner that allowed retention of a
hydrogen-bond acceptor atom in a similar region of space to that of the
oxazinone carbonyl. As shown in Table 5, this strategy provided mul-
tiple analogues that retained high GABAA α5 binding affinity and high
GABAA α5 NAM activity. Within this series of compounds, it was ONO-
8590580 that emerged as the compound with the best overall in vitro
As bicyclic systems to replace the benzoxazinone had been in-
vestigated earlier in the campaign, an alternative approach to C6 sub-
stituent design was taken in which this substituent was replaced with a
4

A. Lewis, et al.
Bioorganic&MedicinalChemistryLetters30(2020)127536
Table 4
In vitro data for different scaffolds.
ID
GABAA α5 Ki (nM)
Efficacy (%)
α5
Half-life in liver microsomes (min)
α1
0
α2
α3
Rat
Human
17
18
38
58
−52
−50
−8
−8
> 100
> 100
−16
−15
+5
> 100
> 100
19
20
19
4
−33
−57
+17
+21
−2
+2
nd
7
14
29
+2
22
21
2
−43
+36
+5
+24
67
63
22
23
17
14
−33
−43
+9
+7
−9
+3
18
55
55
+15
+20
> 100
24
0.5
−55
−8
−16
−31
33
10
25
26
12
7
−47
−49
1
−7
0
17
55
32
39
+3
−10
−7
5

A. Lewis, et al.
Bioorganic&MedicinalChemistryLetters30(2020)127536
Table 5
Data for selected analogues incorporating pendent heterocycles.
R
ID
α5 Ki (nM)
Efficacy (%)
Half-life in liver microsomes (min)
α5
α1
α2
α3
Rat
Human
27
7
−55
−5
−11
−11
24
46
28
29
4
2
−42
−52
−6
−19
−8
−18
−10
74
> 100
> 100
−21
> 100
30
31
32
33
1
−50
−50
−42
−54
−11
−8
−10
−4
−5
−3
6
28
20
27
44
41
15
2
40
−10
−24
−1
> 99
> 100
2
−10
−24
34
9
8
−50
−50
−11
−2
−17
+5
−15
−6
18
> 100
60
ONO-8590580
> 100
6

A. Lewis, et al.
Bioorganic&MedicinalChemistryLetters30(2020)127536
Table 6
In vitro and in vivo data for ONO-8590580.
ID
α5 Ki
α5 efficacy
MDCK-MDR1 Papp (x 10-6
cms−1)/efflux ratio
Plasma protein binding, Brain homogenate binding,
In vivo receptor occupancy of GABAA
(nM)
(%)
rat (% bound)
93
rat (% bound)
96
α5 receptors in rat brain (%)1
ONO-8590580
8
−50
47 / 0.9
71
1Occupancy of the binding site of rat brain GABAA α5 was measured using an in vivo [3H]-Ro15-4513 binding assay (n = 3–4 in each group). [3H]-R015-4513 is a
GABAA α5-specific binding radioligand. ONO-8590580 was orally administered at 10 mg/kg 1 h before sacrifice. [3H]-Ro15-4513 was intravenously administered
10 min before sacrificing the rats. The amount by which ONO-8590580 reduced the specific binding of [3H]-Ro15-4513 relative to the binding in non-treated rats
was defined as the occupancy.14
Table 7
In vivo rat PK parameters for ONO-8590580.
Route1
Dose (mg/kg)
Vss (L/kg)
Cmax (ng/mL)
t1/2 (min)
AUCinf (μg/mL.min)
Cl (mL/min/kg)
%F2
i.v.
0.44
3
1.7
–
278
520
72
24
18
–
–
p.o.
131
107
65
1Vehicle for i.v.: DMSO:Cremophor:water (10:10:80). Vehicle for po: 0.5% w/v methylcellulose in water.
2Bioavailability.
Scheme 1. Synthetic route to ONO-8590580. (i) Benzyltrimethylammonium tribromide, CH2Cl2/MeOH (97%); (ii) Trimethylorthoformate, p-TsOH, reflux; (iii)
cyclopropylmethylamine, CuI, DBU, DMSO, 120 °C (62% for 2 steps); (iv) Pd(PPh3)4, Na2CO3, DMF/water, 100 °C (55%); (v) 7, Pd2dba3, xantphos, Cs2CO3, dioxane,
90 °C (27%).
profile. ONO-8590580 has very high negative allosteric modulatory
activity against GABAA α5 and high α5 binding affinity. Moreover, it
demonstrates an excellent functionally selectivity profile, with essen-
tially no functional activity against the other subtypes, high rat mi-
crosomal stability and good human microsomal stability. Consequently,
ONO-8590580 was taken forward for further ADME evaluation and in
vivo profiling.
benzimidazole ring (3 → 4).15
In summary, optimisation of the hit imidazopyridine 1 focussing on
modification of physicochemical properties and utilising metabolite ID
studies to drive improvements in metabolic stability has led to identi-
fication of the benzimidazole ONO-8590580. This compound displayed
good binding affinity, high negative modulation of the GABAA α5 re-
ceptor subtype, an excellent functional selectivity profile and, as pre-
viously reported,14 is efficacious in two in vivo models of cognition
enhancement. Consequently, it is our belief that ONO-8590580 is a
valuable tool compound to pre-clinically evaluate the pharmacological
effects of selectively negatively modulating the functional response of
the GABAA α5 receptor subtype.
As shown in Table 6, ONO-8590580 exhibited good cell perme-
ability and only a modest efflux ratio in the MDCK-MDR1 cell line,
consistent with potential for brain penetration. Rat plasma protein
binding and rat brain homogenate binding were 93% and 96%, re-
spectively, providing confidence that good oral exposure would achieve
a satisfactory level of target engagement in the brain.
As shown in Table 7, in rats ONO-8590580 had modest plasma
clearance (18 mL/min/kg) and 65% oral bioavailability, following i.v.
and oral dosing at 0.44 mg/kg and 3 mg/kg, respectively. Moreover,
after dosing at 10 mg/kg p.o. ONO-8590580 afforded 71% occupancy
of GABAA α5 receptors in the rat hippocampus (Table 6).14 ONO-
8590580 was also assessed for functional response against the hERG
ion channel, for which it demonstrated an IC50 of 11 µM.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Appendix A. Supplementary data
As reported previously by Kawaharada et al.,14 ONO-8590580
dose-dependently reversed MK-801-induced cognitive deficit in the rat
passive avoidance and 8-arm radial maze tests, and did not elicit an-
xiogenic or pro-convulsant side-effects when administered at doses that
provide high GABAA receptor occupancy in the brain.
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.bmcl.2020.127536.
References
The synthesis of ONO-8590580 is shown in Scheme 1, and was
accomplished using a convergent strategy, via a Buchwald-Hartwig
coupling of the 6-bromobenzimidazole 4 and the substituted 2-amino-
pyridine 7. A key step was the copper-mediated formation of the
1. Quirk K, Blurton P, Fletcher S, et al. Neuropharmacology. 1996;35:1331–1335.
2. Atack JR, Maubach KA, Wafford KA, et al. J Pharmacol Exp Ther. 2009;331(2):470–484.
3. Dawson GR, Maubach KA, Collinson N, et al. J Pharmacol Exp Ther. 2006;316:1335–1345.
4. Ballard TM, Knoflach F, Prinssen E, et al. Psychopharmacology. 2009;202:207–223.
7

A. Lewis, et al.
Bioorganic&MedicinalChemistryLetters30(2020)127536
5. Collinson N, Kuenzi FM, Jarolimek W, et al. J Neurosci. 2002;22(13):5572–5580.
6. Vergnes M, Boehrer A, He X, et al. Epilepsy Res. 2001;47:43–53.
2007;53(7):810–820.
12. Atack JR. Pharmacol Ther. 2010;125(1):11–26.
7. Horowski R, Dorrow R. Psychopharmacology. 2002;162:223–224.
13. Protocols for the α5 SPA binding and FLIPR assay used in the high throughput screen are
described in the Supplementary Information.
8. Sternfeld F, Carling RW, Jelley RA, et al. J Med Chem. 2004;47(9):2176–2179.
9. https://clinicaltrials.gov/ct2/show/NCT02024789, Accessed 26 Nov 2019.
10. http://www.edsa.eu/roche-clematis-trial-discontinued/, Accessed 25 Aug 2020.
11. Nutt DJ, Besson M, Wilson SJ, Dawson GR, Lingford-Hughes AR. Neuropharmacology.
14. Kawaharada S, Nakanishi M, Nakanishi N, et al. J Pharmacol Exp Ther. 2018;366:58–65.
15. Hirano K, Biju AT, Glorius F. J Org Chem. 2009;74:9570–9572.
8