Discovery and optimization of a novel CNS penetrant series of mGlu4 PAMs based on a 1,4-thiazepane core with in vivo efficacy in a preclinical Parkinsonian model

Article abstract of DOI:10.1016/j.bmcl.2021.127838

A high throughput screen (HTS) identified a novel, but weak (EC₅₀ = 6.2 μM, 97% Glu Max) mGlu₄ PAM chemotype based on a 1,4-thiazepane core, VU0544412. Reaction development and chemical optimization delivered a potent mGlu₄ PAM VU6022296 (EC₅₀ = 32.8 nM, 108% Glu Max) with good CNS penetration (K_p = 0.45, K_p,uu = 0.70) and enantiopreference. Finally, VU6022296 displayed robust, dose-dependent efficacy in reversing Haloperidol-Induced Catalepsy (HIC), a rodent preclinical Parkinson's disease model.

Full text of DOI:10.1016/j.bmcl.2021.127838

Bioorg. Med. Chem. Lett. 37 (2021) 127838
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery and optimization of a novel CNS penetrant series of mGlu4 PAMs
based on a 1,4-thiazepane core with in vivo efficacy in a preclinical
Parkinsonian model
,
b
,
,
b
,
,
,b
Caitlin N. Kenta 1, Mark G. Fultona 1, Kaylee J. Stillwella b, Jonathan W. Dickersona
,
Matthew T. Locha b, Alice L. Rodrigueza b, Anna L. Blobauma b, Olivier Boutauda
,
,
,
,
,b
Jerri L. Rooka b, Colleen M. Niswendera
e, P. Jeffrey Conna
e, Craig W. Lindsleya
,
,
b
,
d
,
,
b
,
d
,
,b,c,e,*
a Warren Center for Neuroscience Drug Discovery, Vanderbilt University, Nashville, TN 37232, USA
b Department of Pharmacology, Vanderbilt University School of Medicine, Nashville, TN 37232, USA
c Department of Chemistry, Vanderbilt University, Nashville, TN 37232, USA
d Vanderbilt Kennedy Center, Vanderbilt University Medical Center, Nashville, TN 37232, USA
e Vanderbilt Institute of Chemical Biology, Vanderbilt University, Nashville, TN 37232, USA
A R T I C L E I N F O
A B S T R A C T
A high throughput screen (HTS) identified a novel, but weak (EC50 = 6.2
Keywords:
μ
M, 97% Glu Max) mGlu4 PAM che-
mGlu4
motype based on a 1,4-thiazepane core, VU0544412. Reaction development and chemical optimization delivered
a potent mGlu4 PAM VU6022296 (EC50 = 32.8 nM, 108% Glu Max) with good CNS penetration (Kp = 0.45, Kp,uu
= 0.70) and enantiopreference. Finally, VU6022296 displayed robust, dose-dependent efficacy in reversing
Haloperidol-Induced Catalepsy (HIC), a rodent preclinical Parkinson’s disease model.
Metabotropic glutamate receptor
Positive allosteric modulator (PAM)
Structure-activity-relationship (SAR)
Introduction
sp3-enriched chemotypes in hopes that more attractive overall profiles
would result. Towards this end, we performed a high-throughput screen
The effective, long-term treatment of Parkinson’s disease (PD) re-
mains a significant healthcare challenge, and due to both non-motor
symptoms that develop over time, as well as dyskinesia issues with
(HTS), initially employing an mGlu2/4 heterodimer HEK293 cell line
with native coupling to G protein-coupled inwardly-rectifying potas-
sium channels, or GIRK (using thallium flux as a read-out), followed by a
homomeric mGlu4 counter-screen.20 This screening exercise identified a
novel mGlu4 PAM chemotype based on an sp3-hybridized 1,4-thiazepane
core, which was
dopamine replacement therapy,
a non-dopaminergic approach is
needed.1–6 Selective activation of the metabotropic glutamate receptor
subtype 4 (mGu4), by virtue of positive allosteric modulators (PAMs),
has been shown to significantly reduce and/or eliminate motor symp-
toms in preclinical models of PD by decreasing output of the indirect
Multiple domains of VU0544412 (7) were explored chemically in
search of the optimal mGlu4 PAM within this unique series (Fig. 2).
Elements of interest included the stereochemistry at the C-3 center (the
HTS hit was 3-(R), derived from L-cysteine), ring contracted variants,
replacement of the sulfur with carbon or oxygen and a diverse amide
scan.
pathway in the basal ganglia, akin to deep brain stimulation (DBS).7–15
A
number of mGlu4 PAM chemotypes 1–5 have demonstrated preclinical
efficacy in PD models (Fig. 1), but only recently has an mGlu4 PAM
(Prexton Therapeutics Foliglurax®, 6) entered clinical devel-
opment.15–19 However, the chemical landscape for mGlu4 PAMs has
been largely dominated by flat, highly sp2-hybridized molecules that
engender poor solubility, formulation challenges or possess undesirable
chemical moieties.7–19 Based on these chemotype-driven, physi-
ochemical challenges, we sought to identify mGlu4 PAMs based on new,
Synthesis of the 1,4-thiazepane core initially was attempted
following the only literature report, a 1971 protocol by Blondeau and co-
workers (Scheme 1).21 Here, methyl acrylate 9 underwent a 1,4-addition
with L-cysteine 10 to deliver thioether 11 in 88% yield. Intramolecular
cyclization to form the 1,4-thiazepane 12 proceeded in the presence of 7
* Corresponding author.
E-mail address: craig.lindsley@vanderbilt.edu (C.W. Lindsley).
1
These authors contributed equally
https://doi.org/10.1016/j.bmcl.2021.127838
Received 4 January 2021; Received in revised form 25 January 2021; Accepted 30 January 2021
Available online 5 February 2021
0960-894X/© 2021 Elsevier Ltd. All rights reserved.

C.N. Kent et al.
Bioorganic & Medicinal Chemistry Letters 37 (2021) 127838
Fig. 1. Structures of prototypical mGlu4 PAMs 1–5 and 6 (Foliglurax), the first mGlu4 PAM to advance into human clinical studies. Optimized into a CNS penetrant
tool compound with in vivo efficacy in the rat haloperidol-induced catalepsy (HIC) model.
Fig. 2. Multi-dimensional optimization plan for the HTS hit mGlu4 PAM 7.
N NH3/MeOH for seven days, resulting in low conversion (13%). With
the acid in hand, HATU-mediated amide coupling with diverse anilines
delivered amide congeners 13 in yields ranging from 5 to 15%. Yields
were modestly better with non-aromatic amines, but these proved
inactive as mGlu4 PAMs. The yields from this route were not sufficient to
drive a hit-to-lead campaign, so the team paused to improve the syn-
thetic route.
couplings employing DIC/HOBt proved optimal with 12, delivering final
analogs 13 in yields ranging from 35 to 70%. While superior to our
initial route and enabled early SAR, this was not suitable for scale-up
chemistry to provide sufficient compound quantities for in vivo studies.
We next revisited the initial strategy, but replaced methyl acrylate 9
with acrylic acid (Scheme 3). Conjugate addition of 10 into acrylic acid
proceeded quickly in one hour giving 11 in 60% yield. Facilitation of the
macrolactamization with HATU provided the methyl ester of 12 in good
yield, followed by a mild saponification procedure (Ba(OH)2, THF:H2O)
to produce 12 in 37% overall yield.
While little literature precedent exists for the preparation of
substituted 1,4-thiazepines as chemical probes, there are several works
that report 1,4-thiazepines as a byproduct of a fluorescent cysteine
detection system; thus, we hijacked and evaluated this application.22,23
Here (Scheme 2), we employed the fluorescent probe 2-(2′-hydroxy-4′-
methoxyphenyl)benzo-thiazole (HMBT) 15, readily prepared from
aldehyde 14 in 91% yield. Treatment of 15 with acroyl chloride
smoothly delivers key substrate 16 in 89% yield. Exposure of 16 with L-
cysteine 10 facilitates a conjugate addition/cyclization cascade to pro-
vide the desired 12 in 62% yield, along with recovered HMBT that could
be re-acylated and recycled to produce more 12. Finally, carbodiimide
With methods in hand for both small and large scale synthesis of
analogs 13, the stage was set for SAR evaluation. As mentioned earlier,
only aniline-based amides displayed mGlu4 PAM activity (aliphatic,
benzyl and heteroaryl amines were all inactive), and the nature and
position of substituents was equally crucial for PAM activity. As shown
in Table 1, the HTS hit, with a 3-Br anilino amide 7 was a weak mGlu4
PAM (EC50 = 6.2
μM, 98% Glu Max); however, substitution with a
chlorine atom, as in 13a, significantly improved PAM activity (EC50
=
2

C.N. Kent et al.
Bioorganic & Medicinal Chemistry Letters 37 (2021) 127838
Scheme 1. Synthesis of analogs 13 of the mGlu4 PAM 7.a. aReagents and conditions: (a) NEt3, EtOH, 5 min, rt, 88%; (b) 7 M NH3/MeOH, 7 days, rt, 13%; (c) HATU,
DIPEA, ArNH2, DMF, rt, 12 h, 5–15%.
Scheme 2. Alternate synthetic strategy to access 12a. aReagents and conditions: (a) 2-Aminobenzenethiol, AgNO3 (1 mol%), DMSO, 1 h, rt, 91%; (b) acroyl chloride,
NEt3, DCM, 12 h, rt, 89%; (c) 10, DMC, 1 h, rt, 1 h, then DIEPA, 62%; (d) DIC, HOBt, ArNH2, DCM, 1–6 h, rt, 35–70%.
Scheme 3. Synthesis of 1,4-thiazepane core 12 of the mGlu4 PAM 7.a. aReagents and conditions: (a) acrylic acid, DIEPA, DCM, 1 h, rt,; (b) HATU, 1 h, rt; (c) Ba(OH)2,
THF, H20, 3 h, rt, 37% over three steps .
148 nM, 120% Glu Max) by > 40-fold. Assuming smaller was better, the
3-F congener was prepared, but inactive, as well as a 3,5-diF derivative,
13b, which was weaker than the HTS hit. Interestingly, the 3-Cl, 5-F
analog 13c, regained PAM potency (EC50 = 278 nM, 112% Glu Max),
and a related 3,5-diCl congener proved equipotent (EC50 = 273 nM, 88%
Glu Max). As a bioisosteric replacement, we surveyed the 3,5-diMe
3

C.N. Kent et al.
Bioorganic & Medicinal Chemistry Letters 37 (2021) 127838
2.28, and a brain-to-plasma partitioning coefficient (Kp) of 0.19 (un-
bound Kp,uu = 0.10), suggesting, that in this series, cLogP is a reasonable
predictor of Kp. This hypothesis was further borne out by 13g, a 3-Br, 5-
Table 1
Cl congener, with a cLogP of 2.41, and an improved Kp of 0.45 (Kp,uu
=
Structure and mGlu4 PAM activities of analogs 13.
0.70). Moreover, 13g (VU6022296) was predicted to be moderately
cleared in both human and rat (13.1 mL/min/kg and 59.9 mL/min/kg,
respectively), with acceptable unbound fraction in both plasma and
brain homogenates (human fu = 0.011, rat fu = 0.018, rat BHB fu
=
Compound
Ar
mGlu4 EC50 (nM)
% Glu Max
0.028). Thus, 13g emerged as a reasonable candidate for an in vivo tool
7
3-Br
6,200
148
8,900
278
351
558
273
33
98
within this series, and it proved to be highly selective for mGlu4 as well
13a
13b
13c
13d
13e
13f
13g
13h
13i
13j
13k
13l
3-Cl
120
102
112
101
102
88
(>10 μM versus mGlu1,2,3,5,7 and no inhibition > 50%@10 μM in a
3,5-diF
3-Cl, 5-F
3-Br, 5-F
3-I, 5-F
3,5-diCl
3-Br, 5-Cl
3,5-diBr
3-I, 5-Br
3,5-diMe
3-Cl, 5-Me
3,5-diCF3
radioligand binding panel of 68 GPCRs, ion channels, transporters, and
nuclear hormones).24 13g also showed enhance solubility (6.2
μM),
versus a prototypical PAM 3, with kinetic solubility of 0.9 μM.
Based on the potency, selectivity and DMPK profile of 13g,25 it was
an advanced into a haloperidol-induced catalepsy (HIC) study, a pre-
clinical Parkinsonian rodent model (Fig. 4).16,18,19 In this study, the
vehicle group treated with haloperidol resulted in catalepsy as deter-
mined by an increase in the mean latency to withdraw from 0 sec to 49.8
sec. Treatment with VU6022296 (13g) or VU0418506 (VU506, 1)
significantly reduced the mean latency to withdraw in these animals
(One-way ANOVA, F4,45 = 7.97p < 0.0001). The administration of the
positive control VU506 (1) significantly reduced cataleptic behavior by
64.0% (mean latency to withdraw 17.9 ± 3.7 s; Fig. 4; ***p < 0.001).
The injection of 3 mg/kg of 13g demonstrated a non-significant reduc-
tion of catalepsy by 26.5% (mean latency to withdraw 36.6 ± 4.7 s; p >
0.05). The administration of 10 and 30 mg/kg of 13g significantly
reduced catalepsy by 42.6% and 55.4% (mean latency to withdraw 28.6
± 5.4 s and 22.2 ± 4.6 s; **p < 0.01, ***p < 0.001). Thus, akin to mGlu4
PAMs in other chemical series and based on other sp2-centric chemo-
types,16,18,19 potentiation of mGlu4 by this novel 1,4-thiazepane-based
PAM series results in robust efficacy in this preclinical model of Par-
kinson’s disease.
108
97
35
115
>10
1,090
950
101
<30
87
102
aThallium mobilization assays with HEK293/GIRK cells co-expressing rat mGlu4
performed in the presence of an EC20 fixed concentration of glutamate; values
represent N = 1 experiment performed in triplicate.
variant 13j, and surprisingly, this compound was devoid of mGlu4 PAM
activity. Whereas the 3-Br HTS hit 7 was weak, unexpectedly, a 3,5-diBr
analog 13h was extremely potent (EC50 = 35 nM, 97% Glu Max) as was a
3-Br, 5-Cl derivative 13g (EC50 = 33 nM, 108% Glu Max). All other
analogs 13 were notably weaker, and replacement of the sulfur atom
with either an oxygen atom (e.g., an oxo-1,4-oxazepane) or a methylene
group (e.g., a 7-oxoazepane) led to inactive compounds as well.
The importance of the (R)-stereochemistry at C3 had yet to be
evaluated, thus, following the route in Scheme 3, but substituting D-
cysteine, representative (S)-analogs 19 were prepared and evaluated
(Fig. 3). The (S)-analog 19a of 7 was inactive, whereas the (S)-analogs of
13g and 13h, 19b and 19c, respectively, were ~ 2-fold less potent
(EC50s of 64 nM and 76 nM, respectively). Therefore, the (R)-enantiomer
was favored, yet viable tools could potentially still be obtained with the
opposite enantiomer. Contraction to 6-membered rings, of either ste-
reochemistry, proved devoid of mGlu4 PAM activity.
In summary, an HTS campaign identified a fundamentally new
mGlu4 PAM chemotype based on a 1,4-thiazepane core with weak PAM
Table 2
In vitro and in vivo DMPK Profiles of Selected Analogs 13.
Property
13a
13g
13h
We then evaluated the representative analogs 13 (13a, 13g and 13h)
in a battery of in vitro and in vivo DMPK assays (Table 2) to determine if
the potent mGlu4 PAMs discovered within this novel chemotype had the
profiles necessary to serve as in vivo tool compounds. PAM 13a, the 3-Cl
analog, was a low molecular weight compound (MW = 284), with a low
cLogP (1.52), high predicted hepatic clearance in rat (CLhep = 63.6 mL/
min/kg), but moderate in human (CLhep = 9.4 mL/min/kg) and with
acceptable free fraction (fus 0.02 – 0.03). However, in a rat plasma and
brain level (PBL) tissue distribution cassette, 13a was not CNS pene-
trant, showing no detectable drug levels in the brain, in agreement with
the low cLogP. In contrast, 13h, a 3,5-diBr analog displayed a cLogP of
MW
284
1.52
58
363
2.41
58
408
2.28
58
cLogP
tPSA
In vitro PK parameters
CLHEP (mL/min/kg), rat
63.6
9.4
59.9
58.1
CLHEP (mL/min/kg), human
13.1
12.1
Rat fuplasma
0.02
0.03
0.01
0.018
0.011
0.028
0.011
0.007
0.006
Human fuplasma
Rat fubrain
In vivo Parameters Rat IV PBL cassette (0.2 mg/kg)
Kp
N/A
N/A
0.45
0.70
0.19
0.10
Kp,uu
Fig. 3. Examples of the activity of the (S)-enantiomers 19.
4

C.N. Kent et al.
Bioorganic & Medicinal Chemistry Letters 37 (2021) 127838
PAM chemotype is underway and will be reported in due course.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgments
We thank the Warren Family and Foundation for establishing the
William K. Warren, Jr. Chair in Medicine (C.W.L.) and endowing the
Warren Center for Neuroscience Drug Discovery.
References
1 Lindsley CW, Hopkins CR. Exp Opin Ther Patent. 2012;22:461–481.
2 Robichaud AJ, Engers DW, Lindsley CW, et al. ACS Chem Neurosci. 2011;2:433–449.
3 Lindsley CW, Emmitte KA, Hopkins CR, et al. Chem Rev. 2016;116:6707–6741.
4 Marino MJ, Hess JF, Liverton N. Curr Top Med Chem. 2005;5:885–895.
5 Battaglia G, Busceti CL, Molinaro G, et al. J Neurosci. 2006;26:7222–7229.
6 Hopkins CR, Lindsley CW, Niswender CM. Future Med Chem. 2009;1:501–513.
7 Maj M, Bruno V, Dragic Z, et al. Neuropharmacology. 2003;45:895–906.
8 Williams R, Zhou Y, Niswender CM, et al. ACS Chem Neurosci. 2010;1:411–419.
9 Kalinichev M, Le Poul E, Bolea C, et al. J Pharmacol Exp Ther. 2014;350:495–505.
10 Bennouar KE, Uberti MA, Melon C, et al. Neuropharmacology. 2013;66:158–169.
11 Engers DW, Niswender CM, Weaver CD, et al. J Med Chem. 2009;52:4115–4118.
12 Engers DW, Blobaum AL, Gogliotti RD, et al. ACS Chem Neurosci. 2016;7:1192–1200.
13 Niswender CM, Jones CK, Lin X, et al. ACS Chem Neurosci. 2016;7:1201–1211.
14 Hong S-P, Liu KG, Ma G, et al. J Med Chem. 2011;54:5070–5081.
15 Engers DW, Bollinger SR, Engers JL, et al. Bioorg Med Chem Lett. 2018;28:2641–2646.
16 Bolinger SR, Engers DW, Panarese JD, et al. J Med Chem. 2019;62:342–358.
17 https://parkinsonsnewstoday.com/2017/07/28/prexton-initiates-new-phase-2-trial-
foliglurax-parkinsons-disease/.
18 Panarese JD, Engers DW, Wu YJ, et al. ACS Med Chem Lett. 2019;10:255–265.
19 Panarese JD, Engers DW, Wu YJ, et al. Bioorg Med Chem Lett. 2019;29:342–346.
20 Fulton MG, Loch MT, Rodriguez AL, et al. Bioorg Med Chem Lett. 2020;30:127212.
21 Blondeau P, Gauthier R, Berse C, et al. Can J Chem. 1971;49:3866–3876.
22 Wang J, Li B, Zhao W, et al. ACS Sens. 2016;1:882–887.
23 Liu H, Wang X, Xiang Y, et al. Anal Methods. 2015;7:5028–5033.
24 See, https://www.eurofinsdiscoveryservices.com/catalogmanagement/viewitem/
LeadProfilingScreen-Panel/PP68.
25 Synthesis of 13g: (R)-N-(3-bromo-5-chlorophenyl)-5-oxo-1,4-thiazepane-3-
carboxamide. (R)-5-oxo-1,4-thiazepane-3-carboxylic acid (12).To a 100 mL flask, 16
(0.3 mmol) and L-cysteine (0.375 mmol, 1.25 eq) were combined in 3 mL of DCM,
and the mixture stirred at rt for 2 h. Then, Et3N (80 μL) was added and the solution
stirred for 40 min. The reaction mixture under reduced pressure and the crude solid
was subjected to column chromatography to afford 75 mg of HMBT 15 and 39 mg of
12 as an off-white solid. 1H NMR (400 MHz, MeOD) δ 4.54 (dd, J = 8.1, 2.0 Hz, 1H),
3.10 (dd, J = 14.5, 2.0 Hz, 1H), 3.01 – 2.83 (m, 3H), 2.78 – 2.69 (m, 2H). 13C NMR
(101 MHz, MeOD) δ 177.15, 170.87, 57.53, 40.65, 33.48, 23.70. LC-MS [m/z+H] =
176.3. Analytical data are in accordance with the literature. To a solution of 3-bro-
mom-5-chloroaniline (1.0 eq) in DCM (0.1 M), was added (R)-5-oxo-1,4-thiazepane-
3-carboxylic acid (1.0 eq), HOBt (1.5 eq), and DIC (1.2 eq) at rt. Reaction progress
was monitored by TLC and upon completion, the reaction mixture was washed with
water, concentrated, and redissolved in DMSO. The urea by product of DIC is poorly
soluble in DMSO and will crash out of solution after 15 min. Crude DMSO solutions
were filtered of the solid DIU byproduct and purified using a Gilson HPLC system (30
x 50 mm column; H2O with 0.1% TFA:acetonitrile). Fractions containing the desired
product were quenched with saturated NaHCO3, extracted with 4:1 CHCl3:IPA, and
concentrated to isolate the pure product as a white solid. 1H NMR (400 MHz, DMSO)
δ 10.50 (s, 1H), 7.81 (t, J = 1.8 Hz, 1H), 7.72 (t, J = 1.9 Hz, 1H), 7.45 (t, J = 1.8 Hz,
1H), 7.19 (d, J = 6.7 Hz, 1H), 4.50 (ddd, J = 8.0, 6.6, 2.5 Hz, 1H), 3.04 (dd, J = 14.5,
2.5 Hz, 1H), 2.97 (dd, J = 14.5, 8.0 Hz, 1H), 2.90 – 2.74 (m, 2H), 2.70 (td, J = 4.9,
2.4 Hz, 2H). 13C NMR (101 MHz, DMSO) δ 174.77, 168.82, 141.41, 126.05, 122.54,
121.02, 118.53, 58.22, 40.79, 34.09, 23.88. LC-MS [m/z+H] = 362.9.
Fig. 4. VU6022296 (13 g) dose dependently reduces haloperidol-induced
catalepsy in rats. A) data shown as percent reversal and B) data shown as la-
tency to withdraw. The administration of VU6022296 (13 g) significantly
reduced mean latency to withdraw at 10 and 30 mg/kg as compared to vehicle-
treated controls. The 30 mg/kg dose reversed haloperidol-induced catalepsy to
a similar level of the positive control VU0418506, VU506 (1). Statistical
analysis was performed using a one way ANOVA with Dunnett’s post hoc. **p
< 0.01, ***p < 0.001 compared to vehicle treated control animals. N = 10 per
dosing group.
potency, significant sp3 character and no detectable CNS penetration.
Chemical optimization improved potency by ~ 200-fold, afforded good
brain penetration in rat (Kp and Kp,uu) and showed dose-dependent ef-
ficacy in haloperidol-induced catalepsy, a preclinical model of Parkin-
son’s disease. Thus, VU6022296 (13g) is a novel in vitro and in vivo
mGlu4 PAM tool compound. Further exploration of this unique mGlu4
5