Imaging Mutant Huntingtin Aggregates: Development of a Potential PET Ligand

Article abstract of DOI:10.1021/acs.jmedchem.0c00955

Mutant huntingtin (mHTT) protein carrying the elongated N-Terminal polyglutamine (polyQ) tract misfolds and forms protein aggregates characteristic of Huntington's disease (HD) pathology. A high-Affinity ligand specific for mHTT aggregates could serve as a positron emission tomography (PET) imaging biomarker for HD therapeutic development and disease progression. To identify such compounds with binding affinity for polyQ aggregates, we embarked on systematic structural activity studies; lead optimization of aggregate-binding affinity, unbound fractions in brain, permeability, and low efflux culminated in the discovery of compound 1, which exhibited target engagement in autoradiography (ARG) studies in brain slices from HD mouse models and postmortem human HD samples. PET imaging studies with 11C-labeled 1 in both HD mice and WT nonhuman primates (NHPs) demonstrated that the right-hand-side labeled ligand [11C]-1R (CHDI-180R) is a suitable PET tracer for imaging of mHTT aggregates. [11C]-1R is now being advanced to human trials as a first-in-class HD PET radiotracer.

Full text of DOI:10.1021/acs.jmedchem.0c00955

Subscriber access provided by the University of Exeter
Article
Imaging Mutant Huntingtin Aggregates: Development of a Potential PET ligand
Longbin Liu, Michael E Prime, Matt R Lee, Vinod Khetarpal, Christopher J Brown, Peter D Johnson,
Patricia Azpiazu, Ashley Chen, Daniel Clark-Frew, Samuel Coe, Randall Davis, Anthony Dickie, Andreas
Ebneth, SIMONE ESPOSITO, Elise Gadouleau, Xinjie Gai, Sebastien Galan, Samantha Green, Catherine
Greenaway, Paul Giles, Christer Halldin, Sarah Hayes, Todd Herbst, Frank Herrmann, Manuela Hessmann,
Zhisheng Jia, Alexander Kiselyov, Adrian Kotey, Thomas Krulle, John E. Mangette, Richard W Marston,
Sergio Menta, Matthew R Mills, Edith Monteagudo, Sangram Nag, Martina Nibbio, Laura Orsatti, Sabine
Schaertl, Christoph Scheich, Joanne Sproston, Vladimir Stepanov, Marie Svedberg, Akihiro Takano,
Malcolm Taylor, Wayne Thomas, Miklós Toth, Darshan Vaidya, Katarina Varnas, Derek Weddell, Ian
Wigginton, John Wityak, Ladislav Mrzljak, Ignacio Muñoz-Sanjuan, Jonathan A Bard, and Celia Dominguez
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c00955 • Publication Date (Web): 14 Jul 2020
Downloaded from pubs.acs.org on July 17, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Takano, Akihiro; Karolinska Institutet
Taylor, Malcolm; Evotec UK Ltd
Thomas, Wayne; Evotec UK Ltd
Toth, Miklós; Karolinska Institutet
Vaidya, Darshan; Evotec UK Ltd
Varnas, Katarina; Karolinska Institutet,
Weddell, Derek ; Evotec UK Ltd
Wigginton, Ian; Evotec UK Ltd
Wityak, John; CHDI Foundation Inc Los Angeles
Mrzljak, Ladislav; Takeda Pharmaceutical Co Ltd, Neuroscience
Muñoz-Sanjuan, Ignacio; CHDI Foundation Inc Los Angeles
Bard, Jonathan; CHDI Management, Inc.,
Dominguez, Celia; CHDI Foundation Inc Los Angeles, Advisors to CHDI
Foundation Inc.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Imaging Mutant Huntingtin Aggregates: Development of a Potential PET ligand
§*
†*
§
§
Longbin Liu, Michael E. Prime, Matt R. Lee, Vinod Khetarpal, Christopher J.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
†
†
∆
¶
†
Brown, Peter D. Johnson, Patricia Azpiazu, Ashley Chen, Daniel Clark-Frew,
†
¶
†
†
ϕ
Samuel Coe, Randall Davis, Anthony Dickie, Andreas Ebneth, Simone Esposito,
†
†
†
†
Elise Gadouleau, Xinjie Gai, Sebastien Galan, Samantha Green, Catherine
†
†
∆
†
§
Greenaway, Paul Giles, Christer Halldin, Sarah Hayes, Todd Herbst, Frank
‡
‡
∆
§
†
Herrmann, Manuela Heßmann, Zhisheng Jia, Alexander Kiselyov, Adrian Kotey,
†
¶
†
ϕ
Thomas Krulle, John E. Mangette, Richard W. Marston, Sergio Menta, Matthew R.
†
ϕ
∆
ϕ
ϕ
Mills, Edith Monteagudo, Sangram Nag, Martina Nibbio, Laura Orsatti, Sabine
‡
‡
†
∆
∆
Schaertl, Christoph Scheich, Joanne Sproston, Vladimir Stepanov, Marie Svedberg,
∆
†
†
∆
†
Akihiro Takano, Malcolm Taylor, Wayne Thomas, Miklós Toth, Darshan Vaidya,
∆
†
†
§
§
Katarina Vanräs, Derek Weddell, Ian Wigginton, John Wityak, Ladislav Mrzljak,
§
§
§*
Ignacio Munoz-Sanjuan, Jonathan A Bard, and Celia Dominguez
†Evotec (U.K.) Ltd, 114 Innovation Drive, Milton Park, Abingdon, OX14 4RZ, U.K.
‡Evotec AG, Manfred Eigen Campus, Essener Bogen 7, 22419 Hamburg, Germany.
1
ACS Paragon Plus Environment

Page 3 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
§CHDI Management/CHDI Foundation, 6080 Center Drive, Suite 700, Los Angeles,
California 90405, United States.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
¶Albany Molecular Research, Inc., 1001 Main St., Buffalo, NY 14203, United States.
∆Karolinska Institutet, Department of Clinical Neuroscience, Centre for Psychiatric
Research, Karolinska Hospital, S-17176 Stockholm, Sweden.
ϕIRBM, IRBM Science Park S.p.A., Via Pontina Km 30,600 - 00071 Pomezia (RM) Italy.
KEYWORDS PET imaging ligand, Huntington’s disease, HTT, polyglutamine (polyQ)
aggregate, CAG expansion
ABSTRACT Mutant huntingtin (mHTT) protein carrying the elongated N-terminal
polyglutamine (polyQ) tract misfolds and forms protein aggregates characteristic of HD
pathology. A high-affinity ligand specific for mHTT aggregates could serve as a positron
emission tomography (PET) imaging biomarker for HD therapeutic development and disease
progression. To identify such compounds with binding affinity for polyQ aggregates, we
embarked on systematic structural activity studies; lead optimization of aggregate-binding
affinity, unbound fractions in brain, permeability and low efflux culminated in the discovery of
compound 1, which exhibited target engagement in autoradiography (ARG) studies in brain
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
slices from HD mouse models and post-mortem human HD samples. PET imaging studies with
11C-labelled 1 in both HD mice and WT nonhuman primates (NHPs) demonstrated that the right-
1
1
hand side labelled ligand [ C]-1R (CHDI-180R) is a suitable PET tracer for imaging of mHTT
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
11
aggregates. [ C]-1R is now being advanced to human trials as a first-in-class HD PET
radiotracer.
INTRODUCTION
Huntington’s disease (HD) is a fatal, neurodegenerative disorder characterized by progressive
deterioration of a patient’s motor, cognitive, and psychiatric function. In the United States,
approximately 30,000 people suffer from HD, with symptoms typically manifesting around 40
1
years of age and a median survival of ~20 years from motor onset. Unlike more prevalent
neurological disorders—such as Alzheimer’s (AD) and Parkinson’s (PD) diseases, whose genetic
contribution and lineage can be complex — HD is solely caused by a mutation in a single
autosomal dominant gene. The mutant huntingtin gene (mHTT) harbors an expanded CAG
trinucleotide repeat (>36 CAG triplets) that encodes an elongated polyQ tract in the N-terminus
of the mutant huntingtin protein (mHTT). Normal HTT is important for both embryonic
development and cellular homeostasis; it interacts with proteins involved in transcription,
2
cellular transport, and synaptic and cell signaling. Although HTT is ubiquitously expressed
throughout the body, the expanded gene and its protein appear to selectively target only brain
cells leading to degeneration of medium spiny neurons of the striatum and atrophy of the cortex.3
Accrued evidence indicates that elongated polyQ repeats (especially ≥40) in mHTT induce
3
ACS Paragon Plus Environment

Page 5 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
conformational flux and structural modifications that trigger proteolysis and aggregation of the
4
huntingtin protein during disease progression. Notably, neurites and neuronal intranuclear
inclusions containing N-terminal mHTT fragments have been observed in neuronal and glial
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
6
7
cells, both in animal models of HD, and in the cortex and striatum of HD patient brains. The
age of onset of the disease is proportional to the length of the expanded polyQ repeats which in
8
turn correlate with the propensity for misfolding and aggregate formation. Consequently, one
major area of HD research has been to develop therapies that reduce mHTT in the brain.9
Preclinically, several approaches targeting the mHTT gene at the DNA/RNA levels have been
evaluated, including (1) blocking mHTT transcription by CRISPR-Cas9 and engineered zinc-
finger proteins (ZFPs); (2) repressing RNA translation via RNA interference (siRNAs, miRNAs)
and antisense oligonucleotides (ASOs); and (3) nonsense-mediated decay by small molecule-
mediated preRNA mis-splicing.10,11,12,13,14,15 Recent clinical trials of ASOs have demonstrated
HTT reduction in HD patients’ cerebrospinal fluid (CSF).16 At the protein level, application of
proteolysis targeting chimeras (PROTACs) to the degradation of mHTT may offer an attractive
therapeutic approach.17 Nevertheless, development of any such therapy would likely require
and undoubtedly benefit from a biomarker that can reliably monitor the disease pathology with
sufficient sensitivity. Currently, the most commonly used HTT-detection methods rely on
antibody-based luminescent readouts in biofluid samples, an assay platform that has been used
successfully to quantify mHTT levels in the CSF from HD patients and then adapted for the use
in the on-going clinical trials with the ASO drug, IONIS-HTTRX /RG6042 (now called
tominersen).18,19 However, directly measuring mHTT changes in living human brains, especially
in the striatum and the cerebral cortex where neuropathology is primarily observed, is not yet
possible. Given that misfolded mHTT oligomers and aggregates are a hallmark of HD
4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
pathogenesis and that reduction of mHTT aggregates can ameliorate of mHTT-related
pathophysiology in HD animal models, we hypothesized that direct in vivo quantitation of
mHTT aggregates could serve as an HD biomarker.20 With a suitable radiotracer, we envisioned
that positron emission tomography (PET) imaging would be the ideal platform to fill this critical
gap in HD research.18 PET radiotracers targeting the central nervous system are generally small
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1
18
molecules containing a single positron-emitting isotope (typically C or F). Molecular
imaging with PET radiotracers offers a highly sensitive and relatively non-invasive quantitation
of receptor occupancy or target expression within a living human21 In addition, PET ligands for
aggregated proteins are in clinical use, specifically those targeting amyloid plaques and
neurofibrillary tangles in AD are currently undergoing clinical evaluation.22-24 Here we describe
11
our discovery of CHDI-180R ([ C]-1R, Figure 1) as a mHTT aggregate-specific PET agent with
physicochemical properties suitable for live brain imaging.
O
O
N
N N
R2
O
R1
N
O
R = CH , R = CH : 1
1
3
2
3
11
11
R = CH , R = CH : [ C]-1R (CHDI-180R)
1
3
2
3
11
11
R = CH , R = CH : [ C]-1L
1
3
2
3
11
Figure 1. Structure of CHDI-180, [ C]-1.
RESULTS AND DISCUSSION
Aggregated PolyQ As Surrogates For mHTT-Aggregates
5
ACS Paragon Plus Environment

Page 7 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To facilitate ligand design targeting the mHTT aggregates in HD brains, we needed a
relevant in vitro reporter assay that would allow for facile compound screening. HTT is a very
large protein of 3144 amino acids and difficult to handle; the N-terminal exon-1 fragment is
comprised of the first 17 amino acid-residues (the N17 region) followed by the polyQ repeats
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(polyQ domain) which in turn is flanked by a proline-rich domain (PRD). The polyQ expansion
in mHTT causes the formation of aggregated huntingtin species associated with the altered
neuronal functions in HD. Expression of the mHTTexon-1 fragment alone can recapitulate both
cellular toxicity and HD pathogenesis in mouse models,25,26 and this pathology mirrors the
fragment’s propensity towards aggregation as higher-order species, including spherical
oligomers, proto-fibrils, mature amyloid fibrils, and large fibril clusters.7,27-29 Solid-state NMR
and Föster resonance energy transfer studies of mHTT exon-1 aggregation revealed a complex
3
0
dynamic process involving initial nucleation of the N17-domain, random coil to β-hairpin
transition of the polyQ tract, and oligomerization of β-sheets through interdigitating extended
side chains leading to fibril formation.31,32 At present, the significance of intact full-length
mHTT in relation to the inclusion bodies observed in HD brains remains unclear. A recent cryo-
electron microscopy study of the first full-length human wild-type (WT) HTT in complex with
HTT-associated protein 40 (HAP40) provided little information on how the exon-1 domain
interacts with the rest of the molecule.33 We reasoned that the readily aggregating mHTT exon-1
fragment (polyQ expansion > 40) could serve as a proxy for the pathological mHTT and
facilitate screening to identify high-affinity HTT small molecule binders. However, when
we began this project it was challenging to produce sufficient quantities of aggregated mHTT
exon-1 for compound screening. This was ameliorated by the fact that polyQ peptides show
propensity toward aggregate formation similar to that of mHTT exon-1 fragments, and that
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
exogenously aggregated polyQ species as well as polyQ-induced intracellular aggregates elicit
toxicity in cells.34-36 We therefore opted to use the aggregates derived from a 46-glutamine
peptide (Q46 aggregate) as a surrogate for physiologically relevant aggregated mHTT species.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In the radioligand binding assay (RBA), compounds were incubated with Q46-aggregate in
the presence of a suitable radioligand and the concentration response of inhibition of the
radioligand binding to such aggregates was used to derive the binding potency (IC ) which was
5
0
used for SAR analysis and compound selection. Additional binding assays, including a mHTT
exon1-polyQ46 binding assay, helped validate the Q46-based RBA and these are described in
detail elsewhere.37 Here, we present the progression of compound selection based on the
screening cascade shown in Chart 1. Briefly, compounds were tested for their binding affinity to
the Q46-aggregates. Potent binders were triaged based on desired physicochemical properties
suitable for reaching the CNS with minimal non-specific binding as well as acceptable plasma,
microsomal and hepatocyte stability. Promising compounds were then labelled with tritium
isotope and evaluated for target binding in brain sections from mouse HD models and human
post-mortem samples. Brain uptake was evaluated in micro-PK (µ-PK) where both exposure and
the brain-to-plasma ratio was measured to derive the unbound brain-to-plasma partition
coefficient (Kp,uu,brain). Compounds with good BBB penetration and significant target binding
compared to either non-specific binding or non-target binding (i.e. wild-type mouse and non-HD
human) were labelled with a positron-emitting nuclide and subject to proof of concept in vivo
PET studies in mouse HD models. Finally, a clinical candidate was selected based on non-
human primate (NHP) PET imaging profiles that support favorable brain uptake and washout
kinetics in a non-target expressing large animal model (i.e. lacks mHTT aggregates).
7
ACS Paragon Plus Environment

Page 9 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Chart 1. Screening cascade for mHTT PET ligand candidate selection
PET Ligand Design Principle
A major hurdle to targeting mHTT-aggregate is that the size and density of aggregates in
the brains of HD patients are substantially lower relative to those observed in other
neurodegenerative diseases such as Aβ plaque in AD brains (based on immunofluorescent
8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
stains).37 In addition, mHTT aggregate pathology is found exclusively intracellularly,
including the nucleus and neuropil. For these reasons, radioligands with high molar
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
activity (MA) of tritium labelling (≥3 Tritium-labels per molecule) must be used in binding
studies with HD human brain slices in the autoradiography (ARG) studies (see below). Given
the relatively low abundance of binding sites (Bmax) available to a mHTT-directed PET ligand, it
is critical to identify ligands with both low non-specific binding and high target-binding affinity
(
K approximated from the RBA IC ) to ensure acceptable binding potential (B /K ) for PET
d, 50 max d
imaging studies.38 At the initiation of this program, no ligands were known to bind
specifically to mHTT aggregates. We commenced with surveying known PET ligands
from the Alzheimer’s field, since amyloid plaques are known to have -sheet structures
that are common features of aggregated proteins.39,40 We hypothesized that by
optimizing binding to the polyQ-based aggregates, it should be possible to weaken or
eradicate affinities for those targets from which the initial hits were selected. During the
early phase of our investigations, a close analog of the amyloid PET ligand FDDNP
(
2),41-43 3, was used as the initial radioligand for the screening of a 1,600-compound
collection mostly consisting of structural motifs found in amyloid binders (Figure 2).
44
From that effort, MK-3328 (4) was identified as being moderately potent in the RBA
9
ACS Paragon Plus Environment

Page 11 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3
assay with an IC of 60 nM using its tritium analog [ H]-4 as the new radioligand. In
50
45
comparison, several structurally related PET ligands such as AZD2148 (5), AZD4694
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
46
47
(
6), and PiB (7) were significantly less potent (see IC values in Figure 2).
50
Interestingly, the structurally distinct compounds 2 and 3 showed modest binding to the
Q46 aggregate with IC values of 109 and 66 nM, respectively. We were intrigued by
50
the presence of the nitrile groups in the latter series and hypothesized that the
ethylidene malononitrile moiety might behave like an aryl nitrile, as shown in compound
8. The slight loss of binding potency in 8 (IC = 236 nM) relative to 2 was encouraging
50
and indicative that the nicotinonitrile pharmacophore could be further exploited for the
design of novel chemical series with specificity for poly-Q aggregates.
N
N
HO
F
N
S
N
Me
NH
O
N
N
R
N
N
Me
R
4
(R = CH , MK-3328)
5 (AZD2184)
3
6
38 nM
2
R: 
F
3
60 nM
R:
N
(
FDDNP)
09 nM
66 nM 
Me
[3
H]-4 (R = CT3)
O
1
Me
N
N
O
Me
NH
HO
S
N
Me
NH
N
HO
Me
F
O
6
(AZD4694)
83 nM
7 (PiB)
888 nM
8
2
236 nM
1
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 2. Structures of known amyloid PET ligands. The IC values quoted are
50
3
measured in the Q46 RBA assay using compound [ H]-4 as radioligand and are the
geometric mean of triplicates.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
To further improve upon compound 8, we turned our focus to the modification of
the naphthalene core. Given the frequent presence of [6,5]-bicyclic scaffold in existing
structure templates from the CNS PET literature, we were keen to pursue a
hybridization approach, by replacing the naphthalene ring with a [6,5]-heterocyclic
moiety that was common to the known ligands described earlier (4-7, Figure 2). As
shown in Table 1, when a thiazole core was installed, the resulting compound 9 not only
retained potency with an IC of 104 nM but also showed reduced lipophilicity compared
50
to parent 8, as indicated by its low experimental LogD measured at pH 7.4 (eLogD (< 3))
7.4
using the shake-flask method.48 Incorporation of a benzofuran core resulted in compound
10 which showed enhanced binding potency (IC = 33 nM) albeit with higher
50
lipophilicity (eLogD = 3.7). On the other hand, N-methyl benzimidazole analog 11
showed a dramatic reduction in potency (IC > 1000 nM); presumably the presence of
50
the N-methyl group was not compatible with the planar biaryl conformation necessary
11
ACS Paragon Plus Environment

Page 13 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
for binding to -sheet structures common to misfolded proteins. This detrimental effect

was partially reversed in the unsubstituted benzimidazole 12 (IC = 644 nM) where,
50
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
although unhindered, the presence of a hydrogen-bond donor group (NH) seemed not
well tolerated. Indeed, additional potency-rescue was realized by eliminating the
hydrogen-bond donor NH group as shown in indazole 13 which showed similar binding
potency (IC = 147 nM) to benzothiazole 9. Interestingly, the partially saturated
50
derivative isoindoline 14 was no less potent (IC = 76 nM) relative to the parent
50
compound 9. These results demonstrated that, notwithstanding the subtle steric and
electronic contributions of the [6,5]-bicyclic cores, the benzofuran-based scaffold in 10
was preferentially recognized by the poly-Q46 aggregates.
Table 1. Core replacement of naphthalene in 9 with [6,5]-systems.
N
R
N
RBA IC50
(nM)*
Compound R
9
eLogD
S
N

104.4 ± 31.3
2.7
Me
O
1
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
O

1
0
1
32.8 ± 1.4
>1000
3.7
Me
Me
O
O
N
N


1
1.4
1.8
Me
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
12
644.0 ± 651
Me
N
H
O
N 
N 
1
3
4
147.4 ± 110.5
76.0 ± 42.3
2.1
3.7
Me
Me
N
O
O
1
3
*
Q46 RBA IC values (nM) determined using compound [ H]-4 as radioligand and are
5
0
the geometric mean of 3 triplicates).
The fact that a longer alkyl chain in compound 2 was tolerated in relation to the
methoxy analogs 3 and 8 prompted us to examine the effect of left-hand side
substituents as the next logical step toward further optimization of 10 (Table 2). Indeed,
extension of the methoxy group in 10 to simple (O)-alkyl chains, as shown by fluoroethyl
(
15), methoxymethyl (16), and methoxyethyl (17) derivatives had no detrimental effect
on the potency. However, introduction of (O)-benzylic systems resulted in significant
improvement of potency into the single-digit nanomolar range, as demonstrated by 3-
pyridyl (18) and 5-methoxy-2-pyridyl (19) analogs. It is worth noting that direct aryl
13
ACS Paragon Plus Environment

Page 15 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ether substitution, as exemplified by compound 20, was less tolerated than alkyl ethers.
The loss of potency could be attributed to the unique non-planar conformational
preference of the pyridine ring relative to the benzofuran scaffold. At this juncture, a
separate SAR campaign with the benzofuran core linked to a different RHS moiety and
employing a different radioligand for the RBA binding assay, examined a diverse set of
left-hand side structural inputs (data not shown). From that study, we concluded that
the 5-methoxy-2-pyridyl methyl ether shown in 19 represented the most privileged
structural motif for maintaining balanced potency and lipophilicity in this region. As a
result, this particular left-hand moiety became the new foothold for future SAR studies.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 2. Left-hand side SAR expansion around compound 10
N
O
N
R1
RBA IC50
Compound R1
eLogD
3.7
(nM)*
3
2.8 ±
.4
10
CH O-
3
1
28.7 ±
.9
15
FCH CH O-
3.1
2
2
1
14
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
1
3
1.6 ±
1
1
1
1
2
6
7
8
9
0
CH OCH O-
3.2
2.8
3.8
2.8
3.5
3
2
.5
3.5 ±
CH OCH CH O-
3
2
2
1.4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0

O
6.0 ± 1.6
N
N

O
2.4 ± 1.6
Me
O
94.7 ±
N

O
1.2
*
3
Q46 RBA IC values (nM) determined using compound [ H]-4 as radioligand and are
5
0
the geometric mean of triplicates.
Lead Optimization
With compound 19 poised as the promising lead candidate for additional
evaluation, we made a particularly important finding. We realized early on that the
nitrile group in 10 was important for the observed binding: removal of it led to over 20x
loss of potency (from 33 nM to 740 nM). Now, with the fully optimized left-hand side
ether in place, the nitrile group in 19 could be removed without significant loss of
potency. Thus the 4-pyridine analog 21 still showed good potency with an IC of 7 nM,
50
although its lipophilicity was over one-log unit higher (eLogD ~ 4.1, Table 3).
15
ACS Paragon Plus Environment

Page 17 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Additionally, we found that inserting a nitrogen atom to the furan ring and shifting the
pyridine ring junction from the 4- to 3-position (22) was sufficient to offset the lipophilic
gain (due to the absence of the CN group) with no loss of the potency.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 3. Key modifications on the right-hand side of compound 19
O
R2
N
X
O
Me
O
RBA IC50
nM)*
Compound X
R2
eLogD
(
N
19
21
CH
2.4 ± 1.6
2.8


N
N
CH
N
7.0 ± 1.5
8.2 ± 1.6
4.1
3.1

22
N
*
3
Q46 RBA IC values (nM) determined using compound [ H]-4 as radioligand and are
50
the geometric mean of triplicates.
Identifying nanomolar binders for poly-Q46 aggregates was notable since it
validated our design strategy of transforming known CNS PET ligands targeting Aβ
1
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
plaques into potential tool compounds for HD research. Despite a majority of the
identified high potency polyQ aggregate binders being lipophilic (eLogD > 3), analogs
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
19 and 22 possessed lipophilicity within the desired range (eLogD ≤ 3). They also both
3
11
possessed a convenient handle (methoxy group) for [ H]- and [ C]-radiolabeling.
Subsequently, these compounds were profiled in several in vitro DMPK assays (Table
4). To ensure adequate partition across the blood-brain barrier (BBB), passive permeability,
along with P-gp and BCRP-mediated efflux, were measured in MDCK cells. Both
compounds exhibited moderate permeability with no apparent basal to apical efflux by
either transporter. In mouse liver microsomes, the nicotinonitrile moiety in 19 seemed
to have contributed to higher metabolic clearance than the pyridine analog 22 which
showed moderate stability. Since our intended target organ is the brain, the two lead
compounds were evaluated for their brain uptake in mice. Thus, compounds 19 and 22
were dosed intravenously (iv) to C57BL/6 mice (2.0 mg/kg) and the plasma and brain
exposure (area under the curve from 5 to 60 min, AUC5-60) were measured over 1 h
(Figure S1). The unbound brain-to-plasma partition coefficient (Kp,uu,brain) values,
calculated based on plasma and brain AUCs and fraction unbound in plasma (f ) and
u,p
17
ACS Paragon Plus Environment

Page 19 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
brain homogenate (f ), of 0.6 (19) and 1.1 (22), were indicative that the unbound
u,b
fraction of both compounds achieved acceptable brain uptake (Kp,uu,brain ≥ 0.5).49
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 4. Comparison of lead compounds identified from initial SAR exploration.
N
O
O
N
N
Compound
O
O
N
Me
N
Me
N
O
19
O
22
RBA IC (nM)
2.4 ± 1.6
2.8
8.2 ± 1.6
3.1
50
eLogD
MDCK_WT Papp
nm/s)
1
14
.7
0.3
68
0.6
0.7
(
0
MDR1_ER
BCRP_ER
Mouse Microsomal
243
102
Cl (mL/min/kg)
int
fu,p
0.001
0.004
0.6
0.010
0.014
1.1
fu,b
Kp,uu,brain
ARG_Hippocampus*
ARG_Cortex*
ARG_Striatum*
1541.6 | 1251.8 | 127.2
1651.5 | 1269.4 | 121.5
1140.1 | 1115.1 | 136.5
165.2 | 50.8 | 23.3
138.5 | 44.2 | 20.9
133.7 | 92.5 | 61.9
3
*
ARG specific binding of [ H]-radioligands (2 nM) in mice (HOM_zQ175 | WT | NSB,
fmol/mg); non-specific binding (NSB) values were derived from the WT tissue.
18
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
We evaluated the in vitro mHTT-aggregate binding of 19 and 22 using10-month-
old homozygous zQ175 knock-in mice that carry an expanded CAG repeat (~188) in the
mouse exon-1 gene and harbor both diffuse mHTT and aggregated inclusion bodies in
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
brain tissues.50,51 Each compound was labeled at the methoxy position as CT ether in
3
high molar activity (Scheme S1) and was incubated at 2 nM with mouse brain slices (for
37
comprehensive assay development, see reference ). As shown in the autoradiograms
at the level of hippocampus (Figure 3A), the benzofuran-based nicotinonitrile 19
exhibited very high binding density (>1100 fmol/mg, Table 4) in the hippocampus,
cortex, and striatum regions of both zQ175 and WT mice, indicating considerable off-
target binding. In contrast, benzoxazole-based pyridine 22 displayed significantly higher
binding density in the zQ175 brain slices (134-165 fmol/mg) compared to the
corresponding WT (44-92 fmol/mg), as shown in Figure 3B. Treatment of radioligand-
incubated WT brain slices with the corresponding cold compounds (10 M, to displace
the radioligand) indicated that there was a lower non-specific binding for 22 (23-62
fmol/mg) than for 19 (121-136 fmol/mg).
19
ACS Paragon Plus Environment

Page 21 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
3
Figure 3. Autoradiography images of [ H]-19 (A) and [ H]-22 (B) using WT (top row) and
homozygous zQ175 (bottom row) brain slices. Coronal sections were cut at the level of
hippocampus. Left side of each panel shows total binding (2 nM of radioligand),
whereas the right side shows non-specific binding when co-incubated with cold
radioligands (10 µM).
Given the similar profiles exhibited by compounds 19 and 22 (Table 4), the
observed higher non-specific binding from 19 in the ARG study was likely due to its 3-
fold lower unbound fraction (f ) in the brain homogenate assay (0.004). To determine
u,b
whether in general unbound brain fraction correlates with non-specific binding, a
selected set of compounds across different chemical series with fu,b values of between
0.001 and 0.04 were tritium-labeled and tested for their bindings in WT mouse brain
tissues. The observed non-specific bindings were plotted against unbound brain
20
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
fraction and a clear inverse relationship was revealed (Figure 4, Table S2). This finding
was in line with a recent report by the Pfizer group on the multi-parameter optimization (MPO)
of PET ligands, where 67% of clinically validated CNS PET ligands displayed f values
u,b
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
of greater than 0.05.52-54 Based on both CHDI and Pfizer’s results, we adopted a
criterion for fu,b to be ≥ 0.05 in our screening cascade to minimize the risk of non-
specific binding. In addition, due to the unusually high non-target related binding signal
associated with 19 in the WT mice, the nicotinonitrile series was deprioritized, and we
focused on optimizing the pyridine moiety in 22.
500
50
5
0
0.01
0.02
0.03
0.04
0.05
0.06
Fraction unbound in brain (fu,b)
21
ACS Paragon Plus Environment

Page 23 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 4. Correlation between fraction unbound in brain and non-specific binding in WT
mouse (9-month-old) brain slices using tritium-labeled compounds (ligand
concentration: 1.5 − 1.8 nM). Non-specific binding values (from the cortex region) were
obtained by co-incubating with cold ligands (10 µM).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
At this juncture, the RBA binding assay was modified by replacing [ H]-4 with the
3
more potent radioligand [ H]-22. As a result, the IC values for both 4 and 22 were
50
right-shifted by 2-fold (Table 5). We began the right-hand side optimization by
incorporating a second nitrogen into the pyridine ring, anticipating reduced eLogD and
increased fu,b from the resulting analogs.55 Indeed, diazines (23 − 26) uniformly
returned lower eLogD, high f (≥ 0.05) and, except for pyridazine 25 (IC ~39 nM),
u,b
50
comparable potency to 22 (IC : ~16 nM). On the contrary, adding a methyl or a
50
methoxy group on the pyridine ring enhanced binding potency (IC : 3 ─ 5 nM), and, not
50
surprisingly, elevated the lipophilicity (eLogD: 3.7 ─ 4.1, see 27, 28, 29). Polar groups
containing a hydrogen-bond donor such as MeNH- (30) and carboxamide (31) also
afforded potent binders with high lipophilicity and unacceptable unbound brain fractions
(
<0.01). Interestingly, pyridone 32 seemed to be more potent (IC : 8.0 nM) than the
50
2
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
parent pyridine 22 without compromising either eLogD or f . In addition, N-methylation
u,b
of 32 to 33 (IC : 6.4 nM) appeared to be beneficial as the unbound brain fraction of the
50
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
later (0.035) was only slightly higher than the parent (0.023). This finding prompted us
to consider further modulating the lipophilicity of 33 by incorporating an additional
nitrogen atom. Thus, pyridazinone 1 maintained similar potency (IC : 5.6 nM) to 32 but
50
showed higher unbound brain fraction (fu,b ~0.058), while pyrazinone 34 (IC : 3.9 nM)
50
showed an even more dramatic increase in fu,b (0.27). Replacing the 6-membered
pyridine ring in 22 with alternative heterocycles was also examined. For example,
pyrazole 35 was more potent (IC : 8.6 nM) whereas imidazole 36 was less so (IC : 25
50
50
3
nM). Fused bicyclic amines with sp -centers such as isoindoline 37 (IC : 3.9 nM) and
50
tetrahydroisoquinoline 38 (IC : 8.4 nM) were potent but also highly bound (f ~ 0.01).
50
u,b
From the results summarized in Table 5, it became evident that the binding potency for
polyQ aggregates and the lipophilicity profile for a given compound can be influenced
independently by subtle structural changes on the right-hand side of the lead compound
22. Through the combined efforts from core selection, left-hand side and right-hand
side optimizations, six candidates were selected with unbound brain fractions higher
23
ACS Paragon Plus Environment

Page 25 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
than the targeted level of 0.05. Among these, compounds 23, 26, 1 and 34 also had the
highest binding potency with IC ≤ 10 nM. Consequently, all four compounds were
50
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
labeled as their corresponding CT ethers and were evaluated for target binding in mice.
3
Table 5. Right-hand side SAR exploration.
O
R
N
N
O
Me
O
Compoun
RBA IC50
(nM)*
eLog
D
R
fu,b
d
4
─
60.1± 2.5
2.9
0.005
0.013

22
15.6 ± 1.8
13.8 ± 1.3
3.1
2.4
N
N
N
N

N
23
24
25
26
0.083
0.053
0.110
0.090


17.8 ± 1.3
38.9 ± 2.1
12.8 ± 2.0
2.5
2.3
2.6
N N
N


N
Me
Me
27
28
29
30
5.1 ± 2.2
3.2 ± 2.4
4.0 ± 1.1
3.9 ± 1.4
4.0
4.1
3.7
2.7
0.005
0.006
<0.01
<0.001
N

N
Me


O
N
Me
NH
N
24
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
O
3
1



4.6 ± 1.5
8.0 ± 1.7
3.6
2.7
<0.007
0.023
N
NH
N
HN Me
O
O
32
33
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6.4 ± 1.9
5.6 ± 1.6
2.8
2.5
0.035
0.058
Me

O
1
N N
N
Me

O
34
3.9 ± 1.5
2.0
0.270
N
Me
N

35
36
37
8.6 ± 2.6
24.9 ± 1.4
3.9 ± 1.3
3.0
2.6
2.7
0.026
0.077
0.011
N
Me
N
 N
 N
N
N

N
38
8.4 ± 0.1
1.7
0.014
3
*
Q46 RBA IC values (nM) were determined using compound [ H]-22 as radioligand
50
and reported as geomean ± sd (n >3).
Here the mouse ARG binding studies were conducted in the brains of another
HD model: 12-week old transgenic R6/2 mice that express the human exon-1 gene with
37 25
1
20-CAG repeats (see reference ). In Table 6, the binding results for 23, 26, 34 and
1
were indicated as the specific binding in the hippocampus from both the transgenic
and the WT cohorts. Comparable levels of mHTT-specific binding (13-17 fmol/mg) were
25
ACS Paragon Plus Environment

Page 27 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
displayed by all four compounds in the transgenic mouse brains. However, notable off-
target binding from 23 and 26 was observed in the WT brains with specific binding of
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5.7 and 9.5 fmol/mg, respectively. The latter findings indicated that compounds 23 and
26 are less specific binders for mHTT-aggregates in the mouse model. Additionally,
significantly high liver microsomal clearances were observed with 23 in NHP and 26 in
mouse, respectively, further limiting their utility for preclinical in vivo imaging studies. By
contrast, negligible off-target binding was detected for compounds 34 and 1 in the WT
mouse brains, suggesting that both compounds were highly specific binders to mHTT
exon-1 aggregates. Microsomal turnover was low for 34 across all three species tested
while moderate clearance was seen for 1 in NHP and human preparations. Both 34 and
1
exhibited modest permeability in parental MDCK cells with no Pgp-mediated efflux as
shown by the MDR1 over-expressing cells. However, compound 34 appeared to be a
substrate for the BCRP transporter with an effective efflux ratio of 5.2, indicating its CNS
exposure may potentially be limited for NHP and human studies.56 Based on these
considerations, we elected to advance compound 1 to additional in vitro evaluations.
26
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 121
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 6. Comparison of in vitro and in vivo properties of selected compounds.
MDCK wt
Cmp HPC_SB* Cl (µL/min/mg)**
MDR1 BCRP Kp,uu,brai
int
B-A / A-B
nm/s)
─ | 23
d
R6/2 | WT M | NHP | hu
EER
EER
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
n
(
2
3
6
15.9 | 5.7
16.9 | 9.5
13.4 | 0.1
16.8 | 0
65 | 290 |108
─
─
─
2
239 | 126 | 44
<44 | 31 | 22
<70 | 81 | 103
75 | 52
65 | 41
87 | 59
0.9
0.8
0.9
1.1
5.2
1.3
2.1
1.5
5.8
34
1
*R6/2: binding was based on ligand concentration at 0.3 nM (fmol/mg)
**microsomal intrinsic clearance
Permeability measurement was based on compound at 0.5 µM
Liver microsomal data was based on compound at 0.25 µM (mouse, human) and 0.5
µM (NHP)
Among various HD mouse models, the zQ175 mHTT knock-in model is more
suitable for longitudinal studies of HD therapeutic and biomarker development.50,51 For
this reason, we chose to evaluate 1 in this model for the proof-of-concept studies. The
3
in vitro target binding studies were carried out using [ H]-1 (see Scheme 7 for synthesis)
in 9-month-old heterozygous mice brains. As shown by the autoradiogram and binding
quantification in Figure 5, compound 1 displayed excellent specific binding at all
27
ACS Paragon Plus Environment

Page 29 of 121
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
concentrations tested (cortex: 13.0 fmol/mg at 0.1 nM, 18.3 fmol/mg at 0.3 nM, and 25.9
fmol/mg at 1 nM; Fig 5B) with minimal binding signals detectable in the WT mice 0.01,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0.78, and 1.3 fmol/mg, respectively). It is worth noting that these findings also resonate
3
with the binding results of [ H]-1 from the R6/2 mice studies (Table 6), reaffirming the
favorable target affinity and specificity of 1.
28
ACS Paragon Plus Environment