Catch and Anchor Approach to Combat Both Toxicity and Longevity of Botulinum Toxin A

Article abstract of DOI:10.1021/acs.jmedchem.0c01006

Botulinum neurotoxins have remarkable persistence (～weeks to months in cells), outlasting the small-molecule inhibitors designed to target them. To address this disconnect, inhibitors bearing two pharmacophores - a zinc binding group and a Cys-reactive wa

Full text of DOI:10.1021/acs.jmedchem.0c01006

Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE
Article
A Catch and Anchor Approach to Combat both
Toxicity and Longevity of Botulinum Toxin A
Lucy Lin, Margaret E Olson, Takashi Sugane, Lewis D Turner, Margarita A. Tararina, Alexander L. Nielsen,
Elbek K Kurbanov, Sabine Pellett, Eric A Johnson, Seth M. Cohen, Karen N Allen, and Kim D Janda
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c01006 • Publication Date (Web): 04 Sep 2020
Downloaded from pubs.acs.org on September 6, 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 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A Catch and Anchor Approach to Combat both
Toxicity and Longevity of Botulinum Toxin A
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
Lucy Lin‡, §, Margaret E. Olson†,‡,§, Takashi Sugane†,‡,§, Lewis D. Turner , Margarita A.
§
║
§
†,¤
┴
┴
Tararina , Alexander L. Nielsen , Elbek K. Kurbanov , Sabine Pellett , Eric A. Johnson , Seth
¤
║
*,§
M. Cohen , Karen N. Allen and Kim D. Janda
§Departments of Chemistry and Immunology, The Skaggs Institute for Chemical Biology, Worm
Institute of Research and Medicine (WIRM), Scripps Research, 10550 North Torrey Pines Road,
La Jolla, California 92037, United States.
║Department of Chemistry, Boston University, 590 Commonwealth Avenue, Boston,
Massachusetts 02215, United States.
¤Department of Chemistry and Biochemistry, University of California San Diego, 9500 Gilman
Drive, La Jolla, California 92093, United States.
┴Department of Bacteriology, University of Wisconsin, 1550 Linden Drive, Madison, Wisconsin
5
3706, United States.
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 92
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
ABSTRACT.
Botulinum neurotoxins have remarkable persistence (~weeks to months in cells), outlasting the
small molecule inhibitors designed to target them. To address this disconnect, inhibitors bearing
two pharmacophores — a zinc binding group and a Cys-reactive warhead — were designed to
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
st
leverage both affinity and reactivity. A series of 1 generation bifunctional inhibitors was
achieved through structure-based inhibitor design. Through X-ray crystallography, engagement
of both the catalytic Zn2+ and Cys165 was confirmed. A 2nd generation series improved on
affinity by incorporating known reversible inhibitor pharmacophores; the mechanism was
confirmed by exhaustive dialysis, mass spectrometry, and in vitro evaluation against the C165S
rd
mutant. Finally, a 3 generation inhibitor was shown to have good cellular activity and low
toxicity. In addition to our findings, an alternative method of modeling time-dependent inhibition
that simplifies assay set-up and allows comparison of inhibition models is discussed.
1
. INTRODUCTION. Botulinum neurotoxins (BoNTs), the etiological agents of botulism, are
the most potent known human poisons, with the BoNT/A serotype exhibiting a lethal dose
LD50) of ~1–2 ng/kg of body weight.1-2 Although pharmaceutical preparations of BoNT/A are
(
routinely used at very low doses in medical procedures, including the treatment of facial
wrinkles, migraines and strabismus among others, the implications of BoNT/A misuse, either
through medical malpractice or as a bioterrorism agent, are dire.1-4 Adverse events, while rare,
are associated with extended use and misuse of pharmaceutical BoNTs and include muscle
5
weakness, depressed respiration, dysphagia, and necrotizing fasciitis. Because of its remarkable
toxicity, ease of procurement and absence of effective post-exposure interventions, the Centers
for Disease Control (CDC) classifies BoNT/A as a Tier 1 Category A agent, emphasizing its high
ACS Paragon Plus Environment
2

Page 3 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
risk to public health safety.2, 6-8 In humans, botulism causes long lasting muscle paralysis, which
in severe cases may progress to paralysis of the diaphragm, inhibiting normal respiration,
causing death. Even with extensive and prolonged supportive therapy including nutritional and
digestive support and in severe cases respiratory support, mortality is about 5%. Widespread
dissemination of aerosolized or foodborne BoNT/A would overwhelm the country’s healthcare
infrastructure, particularly the need for mechanical ventilators as life support. Botulism also
occurs naturally, though infrequently. Many U.S. botulism cases in adults result from
unintentional exposure to BoNT produced by Clostridium botulinum, the anaerobic, Gram
positive, BoNT/A-producing bacteria, in mishandled food.9-10 The most common type of
botulism in the US occurs through Clostridium botulinum colonization of infants’ immature
intestines (infant botulism).10 Additionally, botulism can be contracted through wound
infections, and has found an unexpected connection to the opioid epidemic through association
with the use of black-tar heroin.11-13
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
The pathology of BoNT/A toxicity results from intoxication of peripheral neurons,
predominantly motor-neurons. The BoNT/A heavy chain (HC) binds to neuronal cell receptors
and facilitates translocation of the BoNT/A light chain (LC) into the neuron terminal.14-15 The
2
+
LC, a Zn -dependent endopeptidase, cleaves soluble SNARE proteins; BoNT/A specifically
16
cleaves SNAP-25. Cleavage of SNAP-25 impairs the process of vesicular fusion at neuronal
termini, thus inhibiting the release of acetylcholine into neuromuscular junctions.17
Currently, the only available treatment for botulism is an equine-derived heptavalent antitoxin
18
that is severely limited by adverse effects and a restricted window of utility. The antibody-
based antitoxin targets and neutralizes BoNT/A holotoxin (intact HC + LC) but cannot reach the
intracellular LC; thus, effectiveness is lost once BoNT/A LC enters neuronal cells (12–24 h post-
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 92
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
exposure), where it persists for several weeks to months.6, 19-22 Unlike immunotherapies, small
molecules have the ability to gain entry to the neuronal compartment and inhibit BoNT/A after
nerve intoxication. Thus, the discovery of clinically relevant pharmacotherapeutic BoNT/A
inhibitors remains of urgent need to deliver the first post-exposure treatment for botulism.
Extensive efforts over the last two decades have produced several in vitro and in vivo
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
2
3
inhibitors of BoNT/A LC translocation and enzyme activity. The most potent class of BoNT/A
inhibitors remains zinc-chelating molecules, particularly those inclusive of a hydroxamic acid
pharmacophore.24-33 Zinc-chelating inhibitors bind competitively at the BoNT/A active site,
directly preventing the association of the substrate with the enzyme catalytic site. While best-in-
class reversible inhibitors have achieved sub-micromolar potency and modest protection against
BoNT/A challenge in murine models, no drug candidate has reached clinical trials.
An important consideration for the development of small molecule inhibitors of BoNT/A is the
toxin’s inherently long half-life, both in circulation and in nerve cells. In particular, BoNT/A
persists intraneuronally on a timescale of several months to >1 year, remaining fully active.19
Comparatively, the majority of pharmacotherapeutic countermeasures being studied for BoNT/A
inhibition exhibit half-lives that are a fraction of this timescale (hours to days). The design of
reversible BoNT/A inhibitors is limited by this transient mechanism-of-action, even with
perfectly optimized binding affinity and drug metabolism and pharmacokinetic (DMPK)
properties. Irreversible inhibition using an electrophilic small molecule offers notable benefits
1
9
for BoNT/A drug design, particularly as the toxin is non-regenerative. However, irreversible
inhibitors for metalloproteases are rare.24, 34
ACS Paragon Plus Environment
4

Page 5 of 92
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
Figure 1. Summary of past efforts to develop irreversible BoNT/A LC inhibitors, and this work.
(A) Previously reported covalent inhibitors of BoNT/A LC. kinact/K values are presented as
i
reported. (B) Graphical representation of the bifunctional inhibitor strategy, shown with the
fragments used in this manuscript. The green puzzle piece represents a reversible zinc-chelating
active-site inhibitor, which “catches” the enzyme, and is connected to a red anchor representing
the Cys-reactive warhead.
To date, five irreversible BoNT/A inhibitors have been reported (Figure 1A). These covalent
inhibitors putatively target Cys165, a catalytically important residue approximately 7–8 Å from
the catalytic zinc. Mutagenesis of Cys165 to serine reduces BoNT/A catalytic activity 50-fold,
3
5
highlighting the importance of this amino acid. The first class of covalent inhibitors included
rationally-designed benzylidene cyclopentenediones (1), reported by our group in 2010.36
-1 -1
Although impressively potent in vitro (520–580 M s ), we found that the cyclopentenedione
pharmacophore was completely consumed within five minutes when incubated with bovine
serum and glutathione. Despite no possibility for in vivo translation, this report demonstrated that
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 92
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
irreversible inhibition of BoNT/A is possible. The second disclosure of an irreversible BoNT/A
inhibitor leveraged a benzimidazole acrylonitrile pharmacophore, which was identified through
high-throughput screening.37 While the lead inhibitor (2) also exhibited preferable in vitro
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
inhibition kinetics (326–979 M ·s ) and in silico modelling showed a mechanism of covalent
3
8
adduct formation at Cys165, no further development of this class has been reported. Stura et al.
accomplished the first X-ray co-crystal structure of covalent BoNT/A inhibition at Cys165 using
well-characterized,
small
sulfonyl-containing
probes,
namely
(2-aminoethyl)
methanethiosulfonate (MTSEA, 3) and (3-aminopropyl) methanethiosulfonate (MTSPA, 4), that
3
5
specifically react with cysteine residues. As a follow up to the previous report, a library of
electrophilic fragments was screened and benzoquinone was identified as an active warhead for
irreversible inhibition of BoNT/A LC.39 Efforts to elaborate upon the benzoquinone
pharmacophore (5), including an attempt to deliver the first bifunctional (zinc-chelating and
covalent) inhibitor, significantly compromised covalent inhibition. The selenium-containing
compound Ebselen (6), initially investigated in 2010 by Saunders et al. and more recently by the
Bogyo group, was shown to irreversibly inhibit BoNT/A by reaction with Cys165, with a
-
1 -1 40-41
reported kinact/K
i
of 23,000 M s .
Ebselen, however, is a well-known PAINS scaffold with
extensive activity against a number of biological targets, including the recently identified main
protease Mpro of COVID-19.42-43
In this effort, we highlight a new strategy to address toxin persistence; namely, the rational
design of irreversible inhibitors of BoNT/A LC with dual mechanism-of-action (bifunctional)
scaffolds (Figure 1B). To accomplish this strategy, previously optimized BoNT/A inhibitory
scaffolds with non-covalent, zinc-chelating mechanism-of-action were synthetically linked to
electrophilic moieties with irreversible inhibitory potential. An important aspect of this strategy
ACS Paragon Plus Environment
6

Page 7 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
is the use of a competitive inhibitor that will inhibit fully to prevent substrate depletion; as will
be demonstrated in this work, partial covalent inhibition may not be sufficient. This work
provides proof-of-concept that Cys165 and the catalytic zinc of BoNT/A can be simultaneously
engaged by a small molecule, offering a strategy for the future development of irreversible,
bifunctional BoNT/A inhibitors that irrevocably inhibit toxin, decrease intraneuronal half-life
and accelerate nerve cell recovery. Notably, this strategy is somewhat serotype-selective;
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
examination of available X-ray crystal structures of other human-relevant BoNTs B, E and F
show that only BoNT/F LC possesses Cys residues (C165, C166) proximal to the catalytic Zn.44-
4
6
In addition, BoNTs D, G and X, which have not been conclusively linked to botulism
pathology in humans, also lack an active-site proximal Cys.47-49
2
. RESULTS & DISCUSSION.
2.1 1ST GENERATION BIFUNCTIONAL INHIBITORS. The first bifunctional inhibitors of
BoNT/A LC were designed by leveraging privileged pharmacophores; specifically, we
hypothesized that the combination of previously optimized zinc-binding groups (ZBG) with
cysteine reactive electrophiles would afford an irreversible inhibitor with improved potency and
selectivity. As discussed by Bremer et al., this inhibitor design strategy could increase the
effective molarity of an electrophilic moiety for Cys165, in turn enhancing kinact.39 Although
initial attempts to design a bifunctional inhibitor with a benzoquinone electrophile and an amino
acetamide ZBG were mostly unsuccessful, the importance of overcoming BoNT/A LC
persistence dictates a deeper investigation into this design strategy. BoNT/A LC endopeptidase
activity is dependent on zinc-mediated hydrolysis of SNAP-25 between residues Gln197 and
Arg198. The catalytic, tetrahedral zinc is coordinated by a water molecule and a consensus zinc-
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 92
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
binding motif of His-Glu-X-X-His, where X represents any amino acid and the His and Glu
residues coordinate the zinc metal (Figure 2).50 The most successful inhibitor design for
BoNT/A LC, both in vitro and in vivo, has targeted this zinc-binding region and is headlined by
hydroxamates, quinolinols, and recently, steroidal 4-aminoquinolines.24-33, 51-55 Despite their
well-known liabilities, hydroxamates remain the most potent BoNT/A LC inhibitors. As such,
the hydroxamate functional group was selected for our bifunctional inhibitor design. For the
electrophile, methanethiosulfonate (MTS) was selected due to its ligand efficiency, notable
BoNT/A inhibition kinetics (Figure 1), and available crystallographic structural information
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
(
Figure 2B, C).35
ACS Paragon Plus Environment
8

Page 9 of 92
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
Figure 2. Visualizing the bifunctional design strategy with X-ray co-crystal structures. (A)
Rendering of the active site of BoNT/A LC in complex with 2,4-dichlorocinnamic hydroxamic
acid (7) (PDB 2IMA). The inhibition of BoNT/A LC by 7 is driven by hydroxamate-mediated
2+
Zn -chelation and displacement of the active site water. Cys165 is ~6–8 Å from the catalytic
2
+
Zn metal and is separated by a shallow, unrestricted cleft. (B) Structure and kinetic properties
of reversible inhibitor 7, and MTS covalent inhibitors MTSEA (3) and MTSPA (4). (C) X-ray
crystal structure of the active site of BoNT/A LC with 3 covalently bound to Cys165 (PDB
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 92
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
4
ELC). The zinc ion is depicted as a purple sphere. Residues coordinating zinc are colored light
blue. Catalytically important residues are shown in green, and the Cys165 covalent adduct is
colored yellow.
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
Initial proof-of-concept studies investigated the simplest bifunctional scaffold, comprised of a
hydroxamic acid, aliphatic carbon linker and MTS electrophile. Because Cys165 is separated
from the catalytic Zn by ~6–8 Å through a shallow, hydrophobic cleft, carbon linkers of lengths
4–7 were synthesized. A linker length of six carbons between the ZBG and electrophile was
hypothesized to exhibit superior inhibitor activity because the linker length was closest to 7 Å.
st
Synthesis of the 1 generation bifunctional inhibitor series began with bromoaliphatic carboxylic
acids (8–11). The bromo-containing acids were subjected to nucleophilic substitution with
sodium methanethiosulfonate (NaMTS) at elevated temperatures to give 12–15 in good yields.
Subsequent amide coupling with O-tetrahydropyran (THP)-protected hydroxylamine (OTX)
afforded 16–19 in good yields. Finally, these were subjected to mild deprotecting conditions to
give the final hydroxamate-MTS compounds 20–23.
st
Scheme 1. Synthesis of 1 generation bifunctional BoNT/A LC inhibitors. Reagents and
Conditions: (a) NaMTS, DMF, 80 °C, 86–92%; (b) OTX, HATU, DIPEA, DMF, 68–86%; (c)
PPTS, EtOH, 65 °C, 3–49%.
This series of putative, bifunctional inhibitors was evaluated using a time-dependent assay
56
using the FRET substrate SNAPtide. Enzyme and inhibitor were pre-incubated for 0.5–30 min,
followed by 100-fold dilution into substrate solutions to initiate the enzymatic reaction, while
ACS Paragon Plus Environment
1
0

Page 11 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
effectively diluting the inhibitor and halting further covalent inhibition. Quantitative analysis of
5
7
BoNT/A LC activity was determined as previously described. The kinetic parameter kinact/K
i
was determined by plotting the first order rate constant (kobs) versus inhibitor concentration, and
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
fitting to a hyperbolic equation.58 Ultimately, all four of the putative bifunctional inhibitors
-1
-1
demonstrated irreversible inhibition of BoNT/A LC with kinact/K = 1–10 M ·s (Table 1). As
i
predicted, the bifunctional inhibitor with the six-carbon linker (22) was approximately five to
ten-fold more potent than analogues with four (20), five (21) and seven-carbon (23) linkers.
st
Although irreversible inhibition among this series was modest, this 1 generation design
demonstrated that irreversible inhibition with a putative bifunctional scaffold was possible.
st
Table 1. Kinetic Parameters of 1 Generation Bifunctional BoNT/A Inhibitors
k
inact/K
i
-1
Compound
n
K
i
(μM)
kinact (s )
-1 -1
(
M s )
2
2
22
23
0
1
4
5
6
7
66
88
41
0.00007
0.00013
0.00041
0.00191
1.0
1.5
10.1
1.9
998
Furthermore, we obtained X-ray crystal structures of 20, 21, and 22 in complex with BoNT/A
LC at 1.78, 1.90, and 1.92 Å resolution, respectively. Co-crystallization was achieved by
simultaneously mixing solutions of the protein, inhibitor, seed solution and crystallization
reagents as described in the Experimental Section. Phases were calculated by molecular
replacement using the apoprotein (PDB 3BON) as the search model. Overall, there were no large
conformational changes of the protein compared to the unliganded protein (rmsd 0.19 – 0.32 Å)
the structures show formation of the respective Cys165-adduct in the BoNT/A LC active site
ACS Paragon Plus Environment
1
1

Journal of Medicinal Chemistry
Page 12 of 92
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
with binding of the hydroxamate moiety to the Zn2+ cofactor, confirming the bifunctional
mechanism of action. The X-ray crystal structure for 22 is shown in Figure 3, and structures of
complexes with 20 and 21 can be found in Figure S3 of the supplemental text. Notably, the
structure of BoNT/A LC/21 revealed that suboptimal linker length leads to a noticeable
difference in the positioning of Cys165 in that there is electron density corresponding to partial
occupancy of the rotamer typically observed in the unmodified protein (Figure S3B). In this
conformation, the Cys165 side chain is rotated ~180° and no covalent attachment can be formed.
This observation complements kinetics data showing decreased potency of 21 compared with 22.
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
nd
With these results in-hand, a 2 generation series with expanded functionality for molecular
recognition was explored.
Figure 3. X-ray crystal structure of BoNT/A LC in complex with 22 show as green sticks. The
2+
2+
Zn ion is depicted as a purple sphere and residues coordinating Zn are colored light blue. The
flexible nature of the active site accommodates simultaneous interaction with the catalytic zinc
2+
and Cys165. The distance between the cysteine thiol and Zn is shown as red dashes with an
interatomic distance of 7.8 Å.
2
.2 A CONTINUOUS ASSAY FOR COVALENT INHIBITION. Concurrent with our
st
exploration of 1 generation bifunctional inhibitors, we sought to improve the method by which
covalent inhibitors are evaluated. The previously described assay can be considered the
ACS Paragon Plus Environment
1
2

Page 13 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
conventional method of determining covalent kinetic parameters, dependent on back-calculating
k
inact and K from kobs. Although theoretically straightforward, reproducibility issues can arise
i
from the time-sensitive nature of the assay being affected by physical limitations of equipment,
reagent sensitivity, and order of operations. Importantly, the data analysis methods are dependent
on understanding the mode of inhibition; incorrect assumptions about the inhibition mechanism
can lead to the use of incorrect models, and outdated analysis methods can have significant
impact on the accuracy of data.59 Our own difficulties with replicating previously reported
covalent kinetic parameters using both identical and modified assay protocols motivated us to
develop an alternative method.
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
Rather than measuring activity remaining as a function of reaction time after the “removal” of
pre-incubated covalent inhibitors, described vide supra, the reduction in enzyme activity as a
function of time in the presence of inhibitor can be directly observed. Equation 1, which
describes product formation as a function of time, inhibitor concentration, and inhibitor affinity
and rate-constant for inactivation, was obtained by integrating the rate equations for a system
containing a covalent inhibitor and a substrate (Figure S5A). Integration of the rate equations of
an alternative scenario — a system in which the enzyme is covalently modified by an inhibitor
that does not fully prevent substrate binding (partial inhibition, Figure S5B) — gives rise to
Equation 2. The ability to distinguish between these two scenarios is crucial to our bifunctional
strategy, which links a covalent warhead to an active-site directed inhibitor; lack of warhead
reactivity may simply result in a reversible inhibitor, but disruption of the active site inhibitor
portion may be harder to detect.
ACS Paragon Plus Environment
1
3

Journal of Medicinal Chemistry
Page 14 of 92
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
[퐼]
ꢂ
푖푛푎푐푡퐾
퐼
[푆]
ꢀ
ꢁ
ꢁ·ꢁꢄ
푘푐푎푡퐾
[푆] [퐼]
ꢃ+
+
푀
퐾
퐾
푀 퐼
ꢅ
)
(Equation 1)
(Equation 2)
푃=푘푖푛푎푐푡[퐼]퐸푇 1−푒
퐾
퐼
(
ꢂ
ꢊ퐼ꢋ
[푆]
푖푛푎푐푡
푖
·ꢍꢃ+
ꢎ
퐾
퐾
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
퐼푖
푀
ꢅ
ꢄ
[
푆]
ꢊ퐼ꢋ
ꢀ
푖
[푆] ꢊ퐼푖ꢋ ꢊ퐼푖ꢋ[
푆]
푘푐푎푡
ꢇꢈꢉ퐾퐼
ꢌ
퐾
ꢃ+
+
+
ꢅ
ꢑ
푀
퐾
퐾
훼퐾푀퐾퐼푖)
푖
푖
푀 퐼
(
푃=[퐸] ꢆ
ꢏ 1−푒
ꢐ
ꢎ
푇
[퐼]
[푆]
푖
푘푖푛푎푐푡퐾 ·ꢍꢈꢉ퐾
퐼푖
푀
(
)
Using our current FRET substrate and assay conditions, we determined that the linear range of
activity extends up to 50 minutes. This time frame allows us to confidently observe reaction half-
lives (t1/2) from 10–30 minutes, which is the expected range for Cys-reactive warheads.60-62
Fitting fluorescent data obtained in kinetics mode to either Equations 1 or 2, constraining total
enzyme ([E]
T
), inhibitor concentration ([I]), substrate concentration ([S]), and substrate K
M
to
defined values, yield values for kinact and K
i
. To validate these models, comparison of kinact and K
i
values obtained from this method and previously described assay methods is reported, vide infra.
In addition to the facile determination of inhibition kinetics, this assay setup allows qualitative
differentiation of reversible and covalent inhibitors during the assay by observing the linearity of
fluorescence, as demonstrated in Figure 4.
ACS Paragon Plus Environment
1
4

Page 15 of 92
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
Figure 4. Fluorescence output of the SNAPtide FRET assay. Black represents no inhibitor added,
green represents 50 μM of 4, blue represents 0.5 μM of 7. Reversible inhibitor 7 shows a linear
increase in fluorescence, whereas covalent inhibitor 4 shows a non-linear response.
Notably, when MTSPA (4) was tested, the reported K of 7.7 μM could not be replicated via
i
either the continuous method or the conventional method. A comparison of fits was made
between Equations 1 and 2 using the extra sum-of-squares F-test, and it was found that the
irreversible partial inhibitor model was preferred (P < 0.0001) (Figure S6). Furthermore, with
this model, the inhibition kinetic constants of MTSPA (4) were determined to be
-1
k
inact = 0.00114 s and Ki = 52 μM. Examination of the crystal structure of the related compound
MTSEA (3, Figure 2C) is consistent with this model; even with an additional carbon, the small
probe does not occupy the same area as other reported active site inhibitors. Considering the
documented flexibility of the active site, it is possible that substrate-binding in a reduced
capacity may be tolerated even in the presence of this covalent inhibitor.29, 57 Therefore, covalent
modification alone is not sufficient in depleting enzyme activity—full occupancy of the enzyme
active site and covalent modification is required to ensure complete inhibition. This outlines the
need to develop more effective bifunctional inhibitors.
ACS Paragon Plus Environment
1
5

Journal of Medicinal Chemistry
Page 16 of 92
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
2
.3 2ND GENERATION BIFUNCTIONAL INHIBITORS. Despite being among the first
reported small molecule BoNT/A LC inhibitors, 2,4-dichlorocinnamic hydroxamic acid (7)
remains one of the most potent.25, 27, 63 Inhibitor 7 exhibits an IC50 against BoNT/A LC of
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
.41 μM, K
i
= 300 ± 12 nM, and at ~10 mg/kg rescued 16% (5/31) of mice challenged with
~
2.5–5x the intraperitoneal LD50 of BoNT/A in a post-exposure toxicity model.25, 32 In the
surviving animals, no signs of botulism were observed and importantly, there was no small
molecule-associated toxicity upon treatment.64-65 The mechanism of BoNT/A inhibition by 7 was
50
determined using X-ray crystallography of enzyme-inhibitor complexes (Figure 2). Inhibition
of protease activity was driven by hydroxamate chelation of the active site zinc. The
dichlorophenyl ring of the inhibitor binds within a tight, hydrophobic pocket comprised of
Ile161, Phe163, Phe194, and Phe369 in the 370 loop. Notably, the dichlorophenyl ring
completely fills this hydrophobic pocket, explaining previous structure-activity relationship
(SAR) studies which found that bulky substituents at the 3- and 4-positions translated to
25
significantly reduced potency. This hydrophobic pocket is orientated away from Cys165; thus,
no modification to the 2,4-dichlorocinnamic functional group is required. Alternatively, the
hydroxamate amide nitrogen, which faces the shallow hydrophobic cleft towards Cys165, may
nd
be modified with a linked electrophile. The goal of this 2 generation bifunctional series was to
leverage the excellent in vitro and promising in vivo performance of 7 to accomplish a potent and
irreversible BoNT/A inhibitor.
st
Biochemical evaluation of the 1 generation series identified the six-carbon linker (22) as the
optimal spacer between the ZBG and the Cys-reactive warhead. Despite the added cinnamoyl
moiety, the optimal spacing between the pharmacophores was expected to remain the same. As
nd
such, three putative inhibitors were synthesized in the 2 generation series, with linkers of five
ACS Paragon Plus Environment
1
6

Page 17 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
st
to seven carbons. As with the 1 generation series (20–23), the six-carbon linked compound was
expected to have significantly enhanced inhibitory potency over its five- and seven-carbon
counterpart. Synthesis of the 2nd generation series (Scheme 2) first proceeded with S
2
N
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
displacement of bromoalkanes (24–26) using NaMTS to afford 27–29 in moderate yields.
Subsequent substitution of the remaining bromine with OTX was carried out; we found that
THP-protected intermediates were highly reactive and therefore required immediate coupling to
acyl chloride (30). Upon purification, an in situ deprotection was observed affording putative, 2nd
generation bifunctional inhibitors 31–33.
nd
Scheme 2. Synthesis of 2 generation bifunctional BoNT/A inhibitors. Reagents and conditions:
(
a) NaMTS, DMF, 70 °C, 1 h, 62–66%; (b) OTX, DMF, 48 h; (c) 2,4-dichlorocinnamoyl
chloride (30), DCM, 0.1–10% over three steps.
nd
Table 2. Kinetic parameters of 2 generation bifunctional BoNT/A inhibitors.
k
inact/K
i
-1
Compound
n
K
i
(μM)
kinact (s )
-1 -1
(
M s )
33.5
418
3
3
1
2
5
6
7
8.9
2.0
88
0.000297
0.000836
0.00013
33
1.5
ACS Paragon Plus Environment
1
7

Journal of Medicinal Chemistry
Page 18 of 92
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
Using the continuous, time dependent SNAPtide assay described vide supra, 31–33 were
evaluated for irreversible inhibition (Table 2). Compound 32 was also evaluated in the
-1
conventional assay, with observed kinetic parameters of K = 2 μM and kinact = 0.00035 s , in
i
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
agreement with the results of the continuous assay. As expected, compound 32 was not only
more potent than its five-carbon and seven-carbon analogues (31 and 33, respectively), but also
st
-1 -1
the 1 generation bifunctional inhibitors (20–23), with a kinact/K = 418 M s . The improvement
i
st
nd
in inhibition kinetics between the 1 and 2 generation series can be explained by the increased
molecular recognition between the 2,4-dichlorocinnamoyl substituent and the small hydrophobic
pocket adjacent to the active site, defined by the 370 loop. Although the K
is significantly lower when compared to that of 7 (K = 300 ± 12 nM), the irreversible
mechanism of inhibition compensates for any potency losses by catalyzing an irreversible event.
i
value of 32 (2.0 μM)
i
Figure 5. Dialysis of various BoNT inhibitors. Error bars represent standard deviation (n = 3).
BoNT/A LC activity pre-dialysis, post-dialysis, and post-dialysis after addition of ZnCl is
2
represented by black, dark-teal, and light-blue bars, respectively. Incubation concentrations:
BoNT/A LC (10 nM), TPEN (100 μM), 4 (20 μM), 7 (10 μM) and 32 (20 μM).
ACS Paragon Plus Environment
1
8

Page 19 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To demonstrate the irreversible mechanism-of-action, and further validate the need for the
bifunctional inhibition, we incubated BoNT/A LC with 4, 7, and 32. N,N,N′,N′-tetrakis(2-
pyridinylmethyl)-1,2-ethanediamine (TPEN) was included as a control compound which
irreversibly inhibits BoNT/A LC via zinc stripping. Bifunctional inhibitor 32 exhibited a large
reduction in BoNT/A LC activity that remained constant after exhaustive dialysis/re-addition of
zinc to the enzyme, indicating that the loss of activity was due to permanent occupation of the
active site through irreversible covalent modification. By contrast, enzyme incubated with
reversible parent compound 7 showed ~50% inhibition before dialysis, which is a result of the
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
1
0-fold assay dilution not completing diluting out inhibitor, but recovered ~20% activity upon
exhaustive dialysis, indicating a reversible mechanism-of-action. Covalent inhibitor 4 showed a
much smaller reduction in activity when compared to 32 before dialysis, remaining constant after
dialysis — indicating covalent modification. The difference in BoNT/A LC activity between 32
and 4 is consistent with the hypothesis that full occupation of the active site is needed in order to
fully perturb enzyme activity. These results highlight the strength of the bifunctional strategy and
underscore the need for active-site engagement for potency, and covalent modification for
longevity.
To further validate the necessity for bifunctional inhibition, we synthesized additional
compounds masking either the ZBG or the Cys165 reactive warhead (Scheme 3). N-Boc-6-
bromohexylamine (34) was substituted with NaMTS at elevated temperature, followed by
deprotection with TFA to furnish 36. This amine was coupled to 2,4-dichlorocinnamic acid to
give 37, which no longer possesses a ZBG. Free thiol analogue 38 was obtained through
reduction of 32 using NaBH
4
. In addition to the control compounds, a BoNT/A LC variant,
ACS Paragon Plus Environment
1
9

Journal of Medicinal Chemistry
Page 20 of 92
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
which possesses a serine in place of Cys165, was expressed to further validate the mechanism-
of-action of compounds 31–33.
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
Scheme 3. Synthesis of control compounds. Reagents and conditions: a) NaMTS, EtOH, reflux,
1
4
6 h, 40%; b) TFA, DCM, r.t., 1 h, quant.; c) 2,4-dichlorocinnamic acid, HATU, DIPEA, DMF,
6%; d) NaBH , EtOH, 29%.
4
Wild type (WT) BoNT/A LC and the C165S variant were each incubated with 32, 37, and 38,
then analyzed by high-resolution mass spectrometry (HR-MS) to observe changes in mass upon
covalent adduct formation (Figure S7). In addition to Cys165, BoNT/A LC possesses a second
cysteine, Cys134, that is located away (~34 Å) from the active site. As might be expected, a
mass difference corresponding to the addition of two adducts was observed when either 32 or 37
was incubated with WT. The C165S mutant, however, only showed the addition of one adduct.
Notably, when 38 was incubated with either BoNT/A LC WT or C165S, no mass change was
observed. The free thiol lacks the reactivity necessary to form a disulfide bond with BoNT/A,
even when oriented by the hydroxamic acid group, which increases the thiol’s effective molarity
with respect to Cys165.
ACS Paragon Plus Environment
2
0

Page 21 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To further examine the activity of compounds 32 and 38, a cursory screen for inhibition of
BoNT/A LC C165S was performed. At 100 μM, 32 inhibited the enzyme by 49%, and 38
inhibited the enzyme by 65% (Figure S8). As expected, no time-dependent irreversible inhibition
was observed. The difference in inhibitory potency can likely be attributed to the smaller thiol
being more easily accommodated in the active site compared to the much larger MTS group.
Despite significant efforts to obtain an X-ray structure of the complex between 32 and
BoNT/A LC, no structural data was obtained. Instead, we performed covalent docking66 to
visualize the binding mode of 32. As expected from the shape complementarity and potent
binding, the 2,4-dichlorocinnamoyl moiety of 32 in the docked structure overlaid well with that
in the previously determined structure of the complex with 7 (Figure 6A). Thus, occupation of
the hydrophobic pocket by the 2,4-dichlorocinnamoyl moiety not only results in more favorable
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
st
binding, but also allows for better structural guidance than the 1 generation bifunctional
inhibitors. This additional reversible molecular recognition directs the MTS-moiety towards the
shallow, hydrophobic cleft, increasing proximity to Cys165. Notably, upon docking both 31 and
3
3, we observed that the flexibility of the linker allowed similar accommodation in the binding
pocket upon formation of the covalent adduct (Figure S4). Thus, the binding preference of
BoNT/A LC for the six-carbon linker over the five-carbon or seven-carbon analogues likely
stems from enhanced reactivity resulting from the orientation and fit of the MTS group near
Cys165.
ACS Paragon Plus Environment
2
1

Journal of Medicinal Chemistry
Page 22 of 92
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
Figure 6. The predicted binding mode resulting from covalent docking of 32 into the BoNT/A
LC active site (PDB 2IMA). Covalently bound 32 is depicted as beige sticks and noncovalently
bound 7 is depicted in cyan sticks. Residues making interactions with inhibitor are shown as grey
2+
sticks. The catalytic zinc metal is rendered as a purple sphere and Zn -coordinating residues are
colored light blue. The covalent target, Cys165, is colored yellow. The hydrophobic binding
pocket is comprised of Ile161, Phe194, and Phe369 (loop 370).
rd
2
.4 3RD GENERATION BIFUNCTIONAL INHIBITORS. A 3 generation series of bifunctional
inhibitors was pursued with the goal of identifying an agent with improved activity and more
2
7
drug-like properties. The most potent BoNT/A inhibitory hydroxamate is an (R)-hydroxy ethyl
analogue (39) of 7 that was designed to extend the inhibitor within the polar space of the
enzyme’s active site and incorporate chirality. This best-in-class hydroxamate exhibits a
K
i
= 0.16 ± 0.02 μM for BoNT/A LC inhibition. Presuming that this hydroxamate would also
offer the most potent bifunctional inhibitor, a series of putative leads were synthesized
(Scheme 4). For ease, the bifunctional inhibitor was first synthesized as the racemic mixture
ACS Paragon Plus Environment
2
2

Page 23 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
starting from 2,4-dichlorobenzaldehye (48). The aldehyde was condensed with ethyl acetoacetate
(49) in a Knoevenagel condensation, followed by saponification to give dicarboxylic acid 50.
Ring closure of 50 to its corresponding anhydride (51) by reflux in acetic anhydride proceeded
with excellent yield before ring opening with OTX, giving protected hydroxamate 52. Linkers of
three to six carbons in length (4, 36, 46–47) were then installed via standard coupling conditions.
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
rd
Finally, deprotection of the THP protecting group afforded the 3 generation bifunctional
compounds (53–56).
rd
Scheme 4. Synthesis of 3 generation bifunctional inhibitors with structural similarity to 39.
Reagents and conditions: a) NaMTS, DMF, 70 °C, 57–79%; b) TFA, DCM, r.t., 1h, quant.; c)
ACS Paragon Plus Environment
2
3

Journal of Medicinal Chemistry
Page 24 of 92
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
piperidine; KOH (20 M), H
2
O, 78%; d) Ac
2
O, 120°C, 2h, 89%; e) OTX, CHCl , r.t., 77%; (f) 4
3
or 36 or 46–47, HATU, DIPEA, DMF, 5–32%.
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
rd
The series of 3 generation bifunctional inhibitors was evaluated for time dependent inhibition
using the continuous SNAPtide assay. All analogues exhibited inferior inhibition kinetics
compared to 32 (Table 3). In fact, inhibitors of three, five, and six carbon lengths (53, 55–56)
st
exhibited inhibition kinetics reflecting the 1 generation of bifunctional inhibitors (20–23.
-1 -1
Inhibitor 54, with a four-carbon linker, showed a slightly improved kinact/K (19.38 M ˑs ) but
i
nd
remained lesser than the 2 generation inhibitors.
rd
Table 3. Kinetic parameters of 3 generation bifunctional inhibitors.
k
inact/K
i
-1
Compound
n
K
i
(μM)
kinact (s )
-
1 -1
(M s )
5
5
5
3
4
5
3
4
5
6
17
32
48
38
0.000130
0.00062
0.00008
0.00006
7.6
19.38
1.67
56
1.55
We confirmed the mechanism of time-dependent inhibition by incubating 53 with the BoNT/A
LC C165S variant. We found that 53 only inhibited 32% of the enzyme’s activity at 100 μM and
that the inhibition did not appear to be time dependent in nature (Figure S8). This result indicates
that covalent inhibition is dependent on Cys165. HR-MS analysis of the WT and C165S variant
incubated with 53 further supports this conclusion; an increase in mass consistent with two
ACS Paragon Plus Environment
2
4

Page 25 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
adducts was observed in the WT, whereas an increase in mass of only one adduct was observed
with the C165S variant (Figure S7).
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
rd
Scheme 5. Synthesis of 3 generation control compounds 57 and 59. Reagents and conditions:
a) 4, CHCl
2%.
3
, 12%; b) 1-aminopentane, HATU, DMF, r.t., 1 h, 76%; c) PPTS, EtOH, 60°C, 16 h,
4
In order to further explore the requirements for successful covalent inhibition, we synthesized
control compounds 57 and 59, which lack the hydroxamic acid ZBG and the MTS group,
respectively. Compound 57 was obtained by the reaction of MTSPA (4) with anhydride 51,
while compound 59 was obtained by coupling 52 with 1-aminopentane. Both compounds were
tested in the continuous SNAPtide enzyme assay, and no inhibition by 57 was observed up to a
concentration of 500 μM; 59 showed only reversible inhibition (IC50 = 5.7 ± 0.6 μM, Figure S9).
We concluded from this result that the Cys-reactive MTS group itself is not capable of mediating
ACS Paragon Plus Environment
2
5

Journal of Medicinal Chemistry
Page 26 of 92
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
covalent inhibition; in other words, the hydroxamic acid ZBG is likely required to position the
electrophile in proximity to Cys165 for successful adduct formation.
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
X-ray crystal structures of lead 53 and control compound 59 were resolved to 2.5 and 1.67 Å
resolution, respectively (Figures S1 and S2). Inhibitor complexes were co-crystallized as
described for complexes 20-22. Unexpectedly, electron density corresponding to the zinc atom
was not present in the active site of the BoNT/A LC/53 X-ray crystal structure, although zinc
was present in the crystallization milieu. As a result, the inhibitor ZBG is orientated away from
the Zn2+ binding motif (HEXXH). Notably, this structure shows the covalent adduct of 53
reacted with Cys165. This unique orientation in the crystal structure is consistent with the
rd
inhibition kinetics, showing decreased potency of the 3 generation inhibitors compared to the
2nd generation. In addition, the X-ray crystal structure of BoNT/A LC/59 shows that the alkyl
chain of 59 fits into a hydrophobic pocket resulting in a rotation of Phe194; this hydrophobic
pocket may be employed to improve inhibitor affinity if future iterations of this scaffold are
pursued. A more detailed discussion of the X-Ray crystal structures of BoNT/A LC complexed
with 53 or 59 is included in the Supporting Information. Together, these structures demonstrate
the plasticity of the active site and is consistent with the induced-fit model of specificity,29
providing a structural framework for the design of future, highly specific inhibitors of BoNT/A
LC.
Optimization of reversible binding interactions while maintaining an optimal vector/placement
of the Cys-warhead is challenging and current efforts are focused on improving both these
aspects in unison. The barriers to creating optimal binding interactions and Cys-warhead vector
ACS Paragon Plus Environment
2
6

Page 27 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
clearly impact affinity and reactivity, the latter of which is sensitive to warhead placement, as
st
nd
illustrated with the 1 and 2 generation inhibitors.
2.5 CELLULAR EVALUATION OF BIFUNCTIONAL INHIBITORS
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
The promising in vitro data of the 2nd and 3 series prompted evaluation for BoNT/A LC
inhibition in cells. In this assay, human induced pluripotent stem cell (hiPSC) derived matured
neurons were intoxicated with BoNT/A1 before inhibitor was added. After incubation, hiPSC
derived neurons were lysed and analyzed by Western blot using an anti-SNAP-25 antibody.
Absence of the cleaved substrate band demonstrates successful BoNT/A inhibition.
rd
nd
rd
Figure 7. Cellular evaluation of 2 and 3 generation inhibitors. A) Western blot evaluation of
3
1, 32, and 53 against post-exposure BoNT/A1 in hiPSCs. BoNT/A cleavage of the SNAP-25
substrate is measured by quantifying the lower molecular weight band corresponding to the
SNAP-25 breakdown product. Cell rounding as a result of toxicity was observed at higher
ACS Paragon Plus Environment
2
7

Journal of Medicinal Chemistry
Page 28 of 92
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
concentrations, resulting in lower SNAP-25 signal. Full immunoblots are presented in
Figure S12. B) Dose-response curves of 31, 32, 53 (n = 3).
nd
Bifunctional 2 generation compounds 31 and 32 were found to have IC50 values of 59 ± 15 μM
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
rd
and 34 ± 4 μM, respectively, and 3 generation compound 53 was found to have an IC50 of
rd
3
5 ± 6 μM (Figure 7). Interestingly, none of the other analogues in the 3 generation series
demonstrated inhibitory activity in cells (Figure S12). Despite the difference in inhibition
constants, 32 and 53 are equipotent in cells, and nearly two-fold more potent than 31. Although
5
3 may have lower affinity and reactivity against the enzyme (Table 3), its cellular activity
compensates for its shortcomings. Overall, cellular activity of these proof-of-concept inhibitors
is modest; in the past, we have found that hydroxamates may perform better when dosed as pro-
drugs.28
ACS Paragon Plus Environment
2
8

Page 29 of 92
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
nd
rd
Figure 8. Cellular toxicity and MMP specificity of 2 and 3 generation compounds. A) Cellular
viability of HEK293 cells after incubation with 31, 32 (IC50 = 60 ± 5 μM, n = 3), 53, and 59.
Results for compounds 54 and 55 can be found in the supplemental text, Figure S13. B) MMP
panel screen of 32 and 53, (n = 3). Compounds were screened at 25 μM.
Following the observation of signs of toxicity at high concentrations in the hiPSC assay, the
effect of the compounds on cellular proliferation was tested. Compound 59 was included to
probe for toxicity stemming from the MTS group. Notably, the 2nd generation compounds
rd
inhibited HEK293 proliferation more than 3 generation compounds 53 and 59, which did not
exhibit cellular toxicity up to a concentration of 100 μM. Compound 32 inhibited cells with an
IC50 of 60 ± 5 μM; compound 31 did not inhibit cells to the extent that an IC50 value could be
nd
calculated. Moreover, hiPSC derived neurons treated with high concentrations of 2 generation
ACS Paragon Plus Environment
2
9