Discovery of quinuclidine modulators of cellular progranulin

Article abstract of DOI:10.1016/j.bmcl.2021.128209

Phenotypic screening of an annotated small molecule library identified the quinuclidine tetrahydroisoquinoline solifenacin (1) as a robust enhancer of progranulin secretion with single digit micromolar potency in a murine microglial (BV-2) cell line. Subsequent SAR development led to the identification of 29 with a 38-fold decrease in muscarinic receptor antagonist activity and a 10-fold improvement in BV-2 potency.

Full text of DOI:10.1016/j.bmcl.2021.128209

Bioorg. Med. Chem. Lett. 47 (2021) 128209
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of quinuclidine modulators of cellular progranulin
James C. Lanter*, Angela Y.-P. Chen, Toni Williamson, Gerhard Koenig, Jean-François Blain,
Duane A. Burnett
Arkuda Therapeutics, 200 Arsenal Yards Blvd Suite 220, Watertown, MA 02472, USA
A R T I C L E I N F O
A B S T R A C T
Keywords:
Phenotypic screening of an annotated small molecule library identified the quinuclidine tetrahydroisoquinoline
solifenacin (1) as a robust enhancer of progranulin secretion with single digit micromolar potency in a murine
microglial (BV-2) cell line. Subsequent SAR development led to the identification of 29 with a 38-fold decrease in
muscarinic receptor antagonist activity and a 10-fold improvement in BV-2 potency.
Progranulin
Frontotemporal dementia
Quinuclidine
Neurodegeneration
CNS
Frontotemporal dementia (FTD) is the second most common form of
dementia after Alzheimer’s disease (AD), but, compared to AD, FTD is
considered a rare disease with an incidence estimated to be 1.6–4.1/
100,000 annually.1–3 This disorder is distinguished by its age of onset,
rate of progression and localization in the frontal and temporal lobes,
impacting personality, behavior, language, memory and movement.
FTD of the progranulin subtype (FTD-GRN) represents about 25% of
cases,4 and is caused by autosomal dominant mutations5 in the gene
encoding the secreted glycoprotein progranulin (PGRN). All known
mutations cause haploinsufficiency leading to reduced levels of PGRN,
suggesting that its restoration to normal levels will be therapeutically
beneficial to such patients. The current absence of an effective treatment
makes at-risk subjects often choose not to be genotyped possibly also
contributing to an underestimation of the patient population. This point
highlights the urgent need for therapies that will not only affect the
course of the disease, but also prevent, or at least delay, the onset of
behavioral symptoms and cognitive decline.
(Bafilomycin A and chloroquine),8 histone deacetylase inhibitors as well
as mTOR inhibitors and the autophagy activator trehalose9 have been
reported to induce PGRN secretion. Several molecules representing
these mechanisms were part of the library and served as positive con-
trols in the initial screen. Among the molecules screened, one stood out
from the others: the muscarinic receptor antagonist drug solifenacin10
(Fig. 1, 1). Two features of this hit raised its profile: muscarinic receptor
antagonism was not reported in the literature to affect progranulin levels
and evaluation of the scaffold using a multiparameter optimization al-
gorithm11 for CNS drug like properties indicated that it had good po-
tential for brain distribution relative to the periphery.
It was encouraging to see that our first modification, reversal of the
quinuclidine configuration (Fig. 1, 2), maintained the BV-2 potency
while reducing the annotated (M3 antagonist) activity more than 30-
fold, suggesting that muscarinic receptor engagement was not
involved in progranulin release. With this information, we began our
SAR development by exploring substitution on the phenyl substituent
pendant to the tetrahydroisoquinoline (THIQ) ring. To enable this and
subsequent SAR work we employed the synthetic route depicted in
Scheme 1. Acylation of phenethylamine followed by Bischler-Napier-
alski12 cyclization delivered the right half architecture. Reduction of the
cyclic imine was accomplished via hydrogenation with concomitant
creation of a stereocenter at the 1-position of the bicycle. For SAR
exploration of the pendant aryl ring, resolution of the antipodes was
effected by chiral chromatography but for exploring the SAR of analogs
with a fixed aryl moiety, iridium catalyzed asymmetric hydrogenation13
was employed to provide optically pure material. Coupling of this ma-
terial to the quinuclidine was typically done by activating the latter as
even carbamoyl chlorides of the THIQ were of generally low reactivity.
It is in this context that we initiated a program to identify small
molecules effective in raising PGRN levels. To this end we employed a
phenotypic screening approach utilizing an immortalized murine
microglial line (BV-2) to evaluate a moderately sized chemogenomic6
library of about 3500 molecules.7 The members of this library were
chosen with annotated activity across a range of molecular targets
spanning all major protein classes: G-protein coupled and nuclear re-
ceptors, enzymes, transporters, and ion channels. To date, small mole-
cule enhancers of PGRN release have only demonstrated effectiveness in
preclinical models. Conversely an anti-sortilin monoclonal antibody was
shown raise CSF PGRN in the clinic, albeit by blocking a major avenue of
PGRN cellular entry. Of the small molecules, lysosome alkalizing agents
* Corresponding author.
https://doi.org/10.1016/j.bmcl.2021.128209
Received 1 April 2021; Received in revised form 10 June 2021; Accepted 14 June 2021
Available online 18 June 2021
0960-894X/© 2021 Elsevier Ltd. All rights reserved.

J.C. Lanter et al.
Bioorganic & Medicinal Chemistry Letters 47 (2021) 128209
Fig. 1. Screening hit and initial analog.
Scheme 1. Synthesis of Quinuclidine THIQ.molecules.
Table 1
Table 2
Pendant aryl ring SAR.
Linker SAR.
a
b
a
b
Compound
X
PGRN EC50
M3 IC50
MPO Score
Compound
R1
R2
R3
PGRN EC50
M3 IC50
3
–OCO-
257 ± 138
265 ± 133
17,000
4.19
4.92
4.34
4.17
4.92
5.28
5.35
5.17
5.24
4.50
1.96
1
1300
24
13
14
15
16
17
18
19
20
21
22
–NHCO-
10,000
2
H
H
H
H
H
H
H
H
F
H
H
H
H
H
H
H
H
H
F
1830
733
-NMeCO-
–CH2CO-
1030 ± 253
222 ± 38
811
–
3
F
257 ± 138
339 ± 33
2870
17,000
30,000
4
Cl
1480
–OCH2CO-
–CH(OH)CO–
-CH2SO2-
-NHSO2-
–
–
–
–
–
–
–
5
OMe
OCF3
OcycPr
SO2Me
H
–
–
–
–
895
6
2600
1700
7
>10,000
>10,000
3550
>10000
1610
8
–NHCOCO-
–OCOCO-
–CH2CH(CF3)-
9
>10000
3070
10
11
12
F
F
2030
10,000
9400
F
H
F
247 ± 138
1110
H
F
a
in BV-2 cells (nM, ±SD; Values are the average of duplicate runs. Where
a
in BV-2 cells (nM, ±SD; Values are the average of duplicate runs. Where
reported, standard deviation values are calculated from multiple assays).
b
reported, standard deviation values are calculated from multiple assays).
(nM).
b
(nM).
in muscarinic receptor activity as well as a substantial increase in BV-2
potency.
While most quinuclidine derivatives were commercially available, in
select cases (e.g. 15) quinuclidine-3-one was a useful precursor.
Representative results of a comprehensive SAR survey of the pendant
aryl ring revealed a narrow tolerance for substitution (Table 1). Addition
of a fluorine atom to the 4-position (3) led to both a profound reduction
Surprisingly, while installing a chlorine (4) at the same position
maintained the BV-2 potency, it also increased the M3 activity 10-fold.
Further exploration of 4-position substituents with either electron
donating (5, 7) or withdrawing (6, 8) substituents led to a substantial
2

J.C. Lanter et al.
Bioorganic & Medicinal Chemistry Letters 47 (2021) 128209
Fig. 2. Further SAR exploration; asterisks indicate unchanged portions of structure 2(PGRN EC50 in BV-2 Cellsa). a in BV-2 cells (nM, ±SD; Values are the average of
duplicate runs. Where reported, standard deviation values are calculated from multiple assays).
decrease in potency. These results prompted us to look in greater detail
at fluorine substitution. Moving the atom from the 4- to the 3-position
(9) caused about a 10-fold drop in potency as did adding a fluorine
atom to the 3-position (10) of 3. While the 2,4-difluoro analog (11)
maintained BV-2 potency on par with 3, the 2,3-difluoro molecule (12)
was about 4-fold less potent. A wide variety of polar substituents, het-
eroatom substitutions and fused heterocycles on the pendant ring, as
well as a smaller group of conservative modifications on the fused aro-
matic ring were also explored (data not shown). None of these analogs
were more active than those reported in Table 1, indicating 3 repre-
sented an optimized right hemisphere of the pharmacophore.
investigation. Bioisosteric replacement of the amide linker (22) with
trifluoroethyl amine,14 in addition to losing more than ten-fold BV-2
potency (as a 1:1 diastereomeric mixture), also suffered from a reduced
MPO score and was not further pursued.
Having established the quinuclidine stereochemistry needed to
mitigate muscarinic receptor engagement, optimized pendant aryl ring
substitution and explored a range of linkers to connect the two halves of
the molecule, we turned our attention to both the tetrahydroisoquino-
line and quinuclidine moieties. Although the SAR shown thus far in-
corporates (S) stereochemistry off the THIQ ring, as a consequence of
our synthetic approach the (R) isomers of most compounds were also
isolated and tested. Without exception, they were less potent than the
(S) configured molecules as exemplified by compound 23 vs 2 (Fig. 2,).
Moving the carbonyl from the linker (15) to the THIQ ring (24 vs 15)
was also deleterious to the activity as were other changes to the scaffold
such as installation of a methyl group next to the pendant fluorophenyl
ring (25). Steric shielding of the quinuclidine basic nitrogen (26), qui-
nuclidine fluorination (27) or homologation of the linker (28) all
reduced potency to varying degrees. Interestingly, moving the
We next turned our attention to the linker between the right hemi-
sphere and the quinuclidine, selecting a conservative group of alterna-
tives with good, predicted properties (Table 2). Compared to the
carbamate (3), both the trisubstituted urea (13) and the amide (15)
maintained not only the high level of cellular potency but also a good
margin with M3 receptor engagement. None of the other linkers assessed
(16–21) including the N-methyl substituted urea (14) were able to
maintain enough of the cellular potency to warrant further
3

J.C. Lanter et al.
Bioorganic & Medicinal Chemistry Letters 47 (2021) 128209
aggregate, both molecules have pharmacokinetic characteristics suit-
able for the exploration of progranulin changes in vivo; such work is
ongoing and will be reported in due course.
Table 3
In vitro ADME properties for selected molecules.
Compound
MDCK-
MDR1a
PappA→B
[ER]
Aqueous
Protein
Bindingc,
Fu
Brain
Tissue
Bindingc,
Fu
Hepatocytes
Solubilityb
t1/2d, CLint
e
In summary, cellular phenotypic screening of an annotated small
molecule library delivered a tractable scaffold for further investigation.
Systematic modification of this hit greatly reduced the known musca-
rinic receptor antagonist activity while substantially improving the po-
tency of the desired progranulin-releasing effects. Two molecules were
identified with ADME-PK properties warranting further in vivo
investigation.
3
7.59
1990
2.0 (m)
8.6 (r)
6.0 (h)
0.8 (m)
0.8 (r)
29, 23.0(r)
10, 69.8 (c)
>240, 1.9
(h)
[0.94]
13
15
29
0.95
5965
–
–
–
–
–
–
[39.46]
2.68
–
[18.90]
3.01
Declaration of Competing Interest
2670
4.8 (m)
9.6 (r)
0.8 (m)
0.7 (r)
17, 39.7 (r)
22, 31.8(c)
198, 3.5(h)
[3.62]
The authors declare the following financial interests/personal re-
lationships which may be considered as potential competing interests:
The authors are currently, or were at the time the work was performed,
employees of Arkuda Therapeutics. A central goal of Arkuda Thera-
peutics is the discovery of small molecule modulators of progranulin for
the treatment of Frontotemporal dementia.
17.6 (c)
8.6 (h)
a
(*10ꢀ 6 cm/s).
b
c
μ
M, pH 7.4 (PBS).
%; m: mouse, r: rat, d: dog; c: cynolmologous monkey, h: human.
minutes.
d
e
μ
L/min/kg.
Acknowledgements
The authors would like acknowledge several partners in the reported
work. Andrew Cook designed the chemogenomic library from which
solifenacin’s progranulin releasing activity was discovered. Mark Tebbe
served as a chemistry consultant in the early days of the company and
William Greenlee has been a valuable medicinal chemistry resource
throughout our journey. All the compounds reported were synthesized
by partners at Symeres and Shanghai ChemPartner. The latter organi-
zation also conducted both in vitro ADME and in vivo DMPK
characterizations.
Table 4
Pharmacokinetic parameters for 3 and 29.
CLa
Vdss
b
Compound
(Species)
Dose
PO
t1/2
(h)
IV
Cmax,
Kpuu
@ 8
h
F
IV/PO
t1/2
(h)
(%)
u
(nM),
3 (mouse)
3 (rat)
1/10
1/10
1/10
1/10
1/3
4.1
2.1
2.9
5.2
6.8
2.1
1.9
2.2
2.2
6.4
28
37
30
17
32
1.08
0.26
0.09
0.40
0.33
64
16
41
16
42
2.65
11.9
3.11
6.09
1.24
4.91
4.65
7.23
16.2
9.60
29 (mouse)
29 (rat)
29 (cyno)
a
L/h/kg.
Appendix A. Supplementary data
b
L/kg.
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.bmcl.2021.128209.
attachment point from the 3- to the 4-position of the bicyclic amine was
well tolerated (29) though the homolog (30) was not. All attempts to
directly connect the 4-quinuclidinol to the THIQ as a carbamate failed,
likely due to an unfavorable steric environment during coupling.
With several interesting molecules in hand, we proceeded to assess
their in vitro ADME properties (Table 3). It became immediately
apparent from their poor permeability and efflux ratio (ER) that neither
the urea (13) nor amide (15) linkers merited further investigation. Of
the two remaining molecules, the 3-substituted quinuclidine (3) had
superior permeability and efflux potential compared to the 4-substitued
molecule (29) though they were similar across measures of protein
binding, nonspecific brain tissue binding and intrinsic clearance.
We next evaluated the pharmacokinetic properties of both molecules
in mice and rats (Table 4). While 3 outperformed 29 in mice with respect
to its unbound partition coefficient (Kpuu), other PK parameters were
similar. In contrast, rat PK data indicated that 29 regained substantial
ground on the Kpuu front, albeit at some expense to oral bioavailability
and clearance. The cynomologus monkey PK parameters of 29 echoed its
mouse data with the notable exceptions of Kpuu and half-life. In
References
1 Woollacott IOC. Toher JD Neurochemistry. 2016;138:6.
2 Coyle-Gilchrist ITS, Dick KM, Patterson K, et al. Neurology. 2016;86(18):1736.
3 Knopman DS, Roberts RO. J Mol Neurosci. 2011;54(3):330.
4 Mackenzie IRA. Acta Neuropathol. 2007;114(1):49.
5 van Swieten JC, Heutink P. Lancet Neurol. 2008;10(965).
6 Jones LH, Bunnage ME. Nat Rev Drug Discov. 2017;16:285.
7 (a)Larson K, Cook, A, Soper, J et al. A phenotypic screen of a mouse microglial cell
line reveals novel mechanisms to modulate levels of progranulin. in 9th ICFTD),
Vancouver, Canada, 2014; Abstract 137.
(b)Cook A, Larson, KC, Soper, JH, et al. Building a compound phenotypic screening
collection and the discovery of mechanisms to modulate levels of progranulin to treat
FTLD in BSOBC, Boston, MA, 2014.
8 Capell A, Liebscher S, Fellerer K, et al. J Neurosci. 2011;31(5):1885.
9 Cenik B, Sephton CF, Dewey CM, et al. J Biol Chem. 2011;18:16101.
10 Basra R, Kelleher C. Ther Clin Risk Manag. 2008;4(1):117.
11 Wager TT, Hou X, Verhoest PR, et al. ACS Chem Neurosci. 2016;7(6):767.
12 Heravi MM, Khaghaninejad S, Nazari N. Adv Het Chem. 2014;112:183.
13 Kim AN, Ngamnithiporn A, Welin ER, et al. ACS Catal. 2020;10(5):3241.
14 Black WC, Bayly CI, Davis DE, et al. Bioorg Med Chem Lett. 2005;15(21):4741.
4