In Situ Assembly of Choline Acetyltransferase Ligands by a Hydrothiolation Reaction Reveals Key Determinants for Inhibitor Design

Article abstract of DOI:10.1002/anie.202011989

The potential drug target choline acetyltransferase (ChAT) catalyses the production of the neurotransmitter acetylcholine in cholinergic neurons, T-cells, and B-cells. Herein, we show that arylvinylpyridiniums (AVPs), the most widely studied class of ChAT inhibitors, act as substrate in an unusual coenzyme A-dependent hydrothiolation reaction. This in situ synthesis yields an adduct that is the actual enzyme inhibitor. The adduct is deeply buried in the active site tunnel of ChAT and interactions with a hydrophobic pocket near the choline binding site have major implications for the molecular recognition of inhibitors. Our findings clarify the inhibition mechanism of AVPs, establish a drug modality that exploits a target-catalysed reaction between exogenous and endogenous precursors, and provide new directions for the development of ChAT inhibitors with improved potency and bioactivity.

Full text of DOI:10.1002/anie.202011989

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: In situ assembly of choline acetyltransferase ligands by a
hydrothiolation reaction reveals key determinants for inhibitor
design
Authors: Daniel Wiktelius, Anders Allgardsson, Tomas Bergström,
Norman Hoster, Christine Akfur, Nina Forsgren, Christian
Lejon, Mattias Hedenström, Anna Linusson, and Fredrik
Ekström
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202011989
Link to VoR: https://doi.org/10.1002/anie.202011989

10.1002/anie.202011989
Angewandte Chemie International Edition
RESEARCH ARTICLE
In situ assembly of choline acetyltransferase ligands by a
hydrothiolation reaction reveals key determinants for inhibitor
design
Daniel Wiktelius+, Anders Allgardsson+, Tomas Bergström, Norman Hoster, Christine Akfur, Nina
Forsgren, Christian Lejon, Mattias Hedenström, Anna Linusson and Fredrik Ekström*
[*]
Dr. D. Wiktelius+, Dr. A. Allgardsson+, Dr. Tomas Bergström, B.Sc. C. Akfur, Dr. N. Forsgren, MSc. C. Lejon, Dr. F. Ekström
Swedish Defence Research Agency
CBRN Defence and Security
SE-901 82 (Sweden)
E-mail: fredrik.ekstrom@foi.se
M.Sc. N. Hoster, Dr. Mattias Hedenström, Prof. Dr. Anna Linusson
Department of Chemistry, Umeå University
901 87 Umeå (Sweden)
+ These authors contributed equally to this work
Supporting information for this article is given via a link at the end of the document.
and co-workers have identified commercially avalible compounds by
virtual screening that show inhibitory activity in in vitro assays[18,19]
NH2
Abstract: The potential drug target choline acetyltransferase (ChAT)
catalyzes the production of the neurotransmitter acetylcholine in
cholinergic neurons, T-cells, and B-cells. Herein, we show that
arylvinylpyridiniums (AVPs), the most widely studied class of ChAT
inhibitors, act as substrate in an unusual coenzyme A-dependent
hydrothiolation reaction. This in-situ synthesis yields an adduct that is
the actual enzyme inhibitor. The adduct is deeply buried in the active
site tunnel of ChAT and interactions with a hydrophobic pocket near
the choline binding site have major implications for the molecular
recognition of inhibitors. Our findings clarify the inhibition mechanism
of AVPs, establish a drug modality that exploits a target-catalysed
reaction between exogenous and endogenous precursors, and
provide new directions for the development of ChAT inhibitors with
improved potency and bioactivity.
.
N
N
O
O
N
O
P
O
P
O
O
O
S
N
H
N
H
R
N
O
OH
OH OH
O
HO
HO
CoA (R = H)
AcCoA (R = Ac)
OH
P O
I
Cl
NH2
N
N
O
N
1
2
I
I
Introduction
N
Cl
Cl
Br
Choline acetyltransferase (ChAT) is a ubiquitous enzyme in the
animal kingdom and the central upstream actor in acetylcholine
(ACh) metabolism in cholinergic neurons[1], T-cells[2], and B-
cells[3]. Alterations in ChAT expression or activity have been
linked to conditions including Alzheimer’s disease[4],
schizophrenia[5], congenital myasthenic syndromes (CMS)[1], and
chronic viral infections[6]. ChAT is also a potential target for
pharmacotherapy for a range of medical conditions, including
blood pressure disorders[7] and cholinergic overstimulation
caused by organophosphorus nerve agents[8]; however, validation
of ChAT´s utility as a drug target has not been possible due to the
lack of useful inhibitors. Arylvinylpyridiniums (AVPs, e.g. the
prototypical compound 1, Figure 1) are the most widely studied
ChAT inhibitors, with reported half maximal inhibitory
concentrations (IC50) in the low micromolar range[9]. While AVPs
are reasonably potent in vitro, their confounding pharmacological
4
3
5
I
N HCl
N
Cl
Cl
6
Figure 1. Ligands of ChAT. Chemical structures of compounds used in this
study.
ChAT catalyses the synthesis of ACh from choline, using
acetylcoenzyme A (AcCoA, Figure 1) as the acyl donor. Crystal
structures of the apo- and holoenzyme show that ChAT has
binding and catalytic domains within a narrow active site tunnel
that extends through the enzyme (Figure 2)[20]. The choline and
AcCoA binding sites are located on opposite sides of the tunnel.
The choline hydroxyl group is activated by His324 and
subsequently attacks the carbonyl carbon of the AcCoA
thioester[20]. AcCoA and CoA have similar affinity for ChAT and
in the absence of choline, ChAT catalyses the rapid hydrolysis of
profile[10,11]
,
their propensity to photoisomerize[12]
,
their
electrophilic scaffold, and their permanent charge limits their
usefulness in many applications. Furthermore, the mechanism of
ChAT inhibition by AVPs is hitherto unknown[13–15], and the
structure-activity relationships (SAR) of these compounds have
yielded few clues as to how they could be improved[16]. In addition
to AVPs, aryl-3-oxopropanaminium compounds (e.g alfa-NETA)
are known inhibitors of ChAT [17] and a recent work by Darreh-Shori
AcCoA to CoA[20]
.
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202011989
Angewandte Chemie International Edition
RESEARCH ARTICLE
deficient pyridinium ring of 1-CoA forming an edge-to-face
interaction with the arene of Tyr552 and additional contacts to
His324, Tyr436 and Ser438 (Figure 3 A). The naphthyl moiety of
1-CoA is accommodated in a hydrophobic pocket with contacts to
residues Pro98, Cys322, Leu332, Val542, Cys550, Tyr552, and
Cys563 (Figure 3 B). The CoA moiety binds in a pose closely
resembling that observed for the ChAT holoenzyme[20]. While the
use of ChAT-SERM was necessary for these structural studies,
subsequent experiments were performed using the 69 kDa R form
of wild-type ChAT.
To verify that the hydrothiolation reaction observed in the crystal
structure also occurred in solution, we studied the reaction using
Ultra High Performance Liquid Chromatography - High Resolution
Mass Spectrometry (UHPLC-HRMS).
A
reaction mixture
containing ChAT, 1, and CoA was filtered using a 10 kDa cut off
filter to remove the protein fraction, and analysis of the
supernatant showed the presence of 1-CoA in the solution (Figure
4, Supporting Information Figure S1). Importantly, no reaction
product was detected in a control sample without enzyme,
confirming that the reaction is catalysed by ChAT. Enzyme
catalysed hydrothiolation is known to occur in natural product
biosynthesis[30], but to the best of our knowledge, this is the first
observation of such a reaction in vitro.
To confirm the structure of the inhibitor adduct suggested by X-
ray crystallography and UHPLC-HRMS, we prepared 1-CoA as a
mixture of diastereomers by organic synthesis. The reaction
between CoA•3 Li and 1 progressed slowly at room temperature
and required no catalyst or base when performed in
dimethylformamide. A more practical reaction rate was achieved
by using an excess of CoA at 40 °C. The only detectable side
product formed under these conditions was the disulfide-linked
CoA-CoA homodimer. The non-enzymatic synthesis of 1-CoA is
an unusual example of a Michael-type hydrothiolation that
proceeds at mild conditions and involves a highly functionalised
substrate. The HRMS spectrum of the product was identical to
that of 1-CoA produced in the ChAT-catalysed reaction, and the
structure was further confirmed by a full Nuclear Magnetic
Resonance (NMR) characterization (Supporting Information
Materials and Methods).
Figure 2. Molecular architecture of ChAT. Molecular architecture of ChAT
with choline (pink) and CoA (orange) shown using stick representations. The
active site His324 and the choline-binding Tyr552 are shown in yellow. The
catalytic domain is shown in cyan and the binding domain is shown in grey. The
figure was generated by a superposition of pdb entry codes 2FY3 and 2FY4.
Michael adducts that are formed reversibly may undergo retro-
reaction, resulting in an equilibrium between reactants and
product. By using NMR to assess the stability of 1-CoA in
phosphate buffered saline (PBS) at pH 7.4, we found that an
equilibrium molar ratio of 9:1:1 between 1-CoA, 1, and CoA was
rapidly reached upon dissolution (Supporting Information Figure
S2 A, B). The stability of the adduct in aqueous buffer was further
corroborated by UHPLC-HRMS, which showed that the
concentration of 1-CoA in ammonium acetate buffer (pH 6.8) was
constant during 7 hours at 10 °C (Supporting Information Figure
S2 C). Although the specific experimental conditions may
influence the Michael - retro - Michael equilibrium and thereby
determine the effective concentrations of 1-CoA, 1 and CoA,
these experiments show that the inhibitor adduct was sufficiently
stable in aqueous buffer for further experiments.
To determine whether the in situ formation of CoA-adducts is a
common inhibitory mechanism for all AVPs, we reviewed the
chemical structures of previously reported compounds in this
class[9,22–27], and found that all active compounds possess a
Michael-acceptor moiety (i.e. a vinyl linker and an electron-
withdrawing heteroarene) and could therefore all be
hydrothiolation substrates. To test this hypothesis, we studied the
reaction in solution using ChAT, CoA and a representative subset
of AVPs (2-5, Figure 1). In all cases, the expected hydrothiolation
product (i.e. 2-5-CoA) was observed by UHPLC-HRMS
(Supporting Information Table S2). Conversely, no reaction
product was observed for the negative control 6 (Figure 1), which
lacks the Michael acceptor motif. A Principal Component
Analysis[31] (PCA, Supporting Information Figure S3) of
parameters describing molecular size, electronic properties and
lipophilicity shows that the chosen subset is representative of
Results and Discussion
To understand the molecular recognition of AVPs by ChAT, and
the binding mode of this compound class, we determined the
crystal structure of Homo sapiens ChAT modified with surface
entropy reducing mutations (ChAT-SERM)[21] in complex with 1
and CoA (Supporting Information Table S1). The resulting 2.3 Å
resolution X-ray crystal structure revealed the presence of a
covalent bond between the thiol of CoA and the b-carbon of the
vinyl linker of 1, suggesting that a hydrothiolation reaction had
occurred in close proximity to the catalytic His324 of ChAT,
forming the adduct 1-CoA (Figure 3 A, B). The pyridinium and
naphthyl rings are not coplanar, indicating that 1-CoA has a
saturated linker (Figure 3A), consistent with a Michael-type
addition facilitated by the electron-withdrawing pyridinium ring.
The conformation of the His324 sidechain differs from that in the
apo- and holoenzyme[20], and the structure suggests that the
imidazole activates the thiol of CoA towards nucleophilic attack
on the alkene moiety of 1. Accordingly, the site of His324-assisted
activation in the enzymatic synthesis of 1-CoA differs from that in
the natural reaction of ChAT, where the imidazole instead
activates choline. The in situ assembled ligand exhibits excellent
shape complementarity towards the interfacial active-site tunnel
formed between the binding and catalytic domains of ChAT. Over
160 different AVPs and some analogous compounds with
inhibitory activity against ChAT have been reported[9,22–27] , but no
structural information about their interaction with ChAT is
available. Our crystal structure shows that 1-CoA spans the
choline and AcCoA binding sites of ChAT, with the electron-
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202011989
Angewandte Chemie International Edition
RESEARCH ARTICLE
oxopropanaminium compound a-NETA[17]. In our hands, this
compound was not sufficiently stable to allow structural or
biochemical characterisation of its interaction with the enzyme.
active AVPs (i.e. reported apparent half maximal inhibition
concentration IC50 lower than 15 µM). We therefore suggest that
in-situ assembly of CoA-adducts is
a common inhibition
mechanism for all AVP inhibitors. We also synthesised the aryl-3-
Figure 3. ChAT catalyses a hydrothiolation reaction between 1 and CoA. A) X-ray crystal structure of the hydrothiolation reaction product 1-CoA (green) in
complex with ChAT-SERM with the catalytic and binding domains shown in cyan and grey respectively. The Polder omit map is shown at a contour level of 4 sigma
(blue). B) The naphthyl moiety of 1-CoA is accommodated in a hydrophobic pocket.
reaction. Furthermore, the naphthyl containing inhibitors 1,2-CoA
provide a higher degree of stabilization than adducts that have a
substituted phenyl moiety (3-5-CoA). Interestingly, we see only
minor differences between the charged N-methylpyridinium 3-
CoA and the pyridine derivative 5-CoA, indicating that interaction
with Tyr552 only marginally contribute to their stabilization of
ChAT.
Assays used for measuring the inhibitory potency of AVPs
include CoA or AcCoA[28,33,34], which is rapidly hydrolysed to
CoA[20], and do not account for the fact that AVPs and CoA are
co-substrates in the enzyme-catalysed hydrothiolation reaction.
Figure 4. Confirmation of the chemical structure of 1-CoA by HRMS. The
The propensity for inhibitor adduct decomposition by retro-
fragmentation of the product formed in the reaction between 1 (40 µM) and CoA
Michael reaction for a given compound may also give confounding
(80 µM) catalysed by ChAT (4.6 µM) in HRMS analysis. Measured (bold) and
results when determining the inhibitory potency. It is therefore
unsurprising that the SAR of AVPs remains poorly understood,
and that that there has been no development of improved
inhibitors based on the AVP scaffold. The chemically prepared 1-
CoA offers the opportunity to circumvent these problems by
studying a pre-assembled ChAT inhibitor.
theoretical (bold italic) high resolution masses of fragments are indicated.
Ligand-induced shifts in the midpoint of a protein’s melting
temperature (Tm) correlate with the protein’s affinity for the ligands
in question[32]. We therefore used Circular Dichroism (CD)
spectroscopy to investigate the individual contributions of the
reactants and the in situ formed CoA adducts to the temperature
stability of ChAT. Compounds 1-6 had no effect on the Tm of ChAT,
but binding of CoA increased the Tm by approximately 3 °C (Table
1, Supporting Information Figure S4). When both 1-5 and CoA
were present, the Tm increased even further by up to 12 °C. For
comparative purposes, compound 6 (which can presumably
occupy the same binding site as 1-5 but lacks the Michael
acceptor motif) did not shift the Tm, even in the presence of CoA.
These results show that 1-5-CoA stabilize the folded state of the
enzyme considerably and that this effect is due to the formed
adducts and not to the binding of the individual components of the
The association kinetics was studied by following the time-
dependent decline in the enzymatic activity at various
concentrations of chemically prepared 1-CoA. Under the assay
conditions, where choline, CoA and 1-CoA compete for the same
binding sites, we found that the inhibition is slow and occurs on a
time scale of minutes, even at inhibitor concentrations as high as
100 µM (Figure 5). For comparative purposes, the Michaelis
constants (KM) are 11.9 and 8.8 µM for AcCoA and CoA,
respectively[35]
.
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202011989
Angewandte Chemie International Edition
RESEARCH ARTICLE
Table 1. Thermal stabilization of ChAT by inhibitor adducts and their target residence time.
Compounda
koff (min-1)c
Tm (°C)b
t (min)c
- CoA
+ CoA
+ CoA
-
49.2 (48.0 - 50.4)
48.3 (45.8 - 50.8)
49.4 (48.6 - 50.3)
50.0 (49.2 - 50.9)
49.2 (46.9 - 51.5)
48.9 (47.4 - 50.4)
49.6 (48.5 - 50.7)
48.8 (46.6 – 51.2)
52.4 (51.9 - 52.9)
63.8 (62.2 - 65.4)
64.5 (62.0 – 66.9)
59.0 (57.0 - 61.0)
57.5 (54.9 - 60.0)
57.3 (55.0 - 59.6)
52.5 (49.7 - 55.3)
52.7 (51.0 – 53.9)
-
1
0.080 (0.077 - 0.083)
0.097 (0.093 - 0.102)
0.342 (0.317 - 0.371)
0.135 (0.129 - 0.141)
0.156 (0.148 - 0.164)
-
12.5
10.3
2.9
2
3
4
7.4
5
6
6.4
Choline
-
aAll samples contained ChAT
bDetermined by temperature scanning CD experiments, mean, n=3-4, (95% CI)
cDetermined by jump dilution experiments, mean, n=3 independent experiments, each measured in quadruplicates, (95% CI)
Figure 5. The inhibition kinetics of 1-CoA. The inhibition progress curve at
various concentrations of chemically prepared 1-CoA. The enzymatic activity
was determined using a fluorescence assay and the reaction mixture also
contained CoA (9.3 µM) and acetylthiocholine (9.5 mM). The data (n=3, mean
and SD plotted) was baseline corrected and the solid line represents the fit of
the experimental data of the control sample to a linear regression line
(R2=0.984).
The thermodynamics of binding was characterized by
isothermal calorimetry (ITC). The Gibbs free energy of binding
(DG°) of 1-CoA was mainly governed by entropic contributions (-
TDS°), while the binding of CoA was dominated by enthalpic
terms (DH°) (Figure 6 A, Supporting Information Figure S4). Both
ligands have dissociation constants in the high nanomolar range
Kd (1-CoA) = 8.9 ´ 10-7 M (95% CI 1.3 ´ 10-7 – 1.8 ´ 10-6) and Kd
(CoA) = 8.5 ´ 10-7 M (95% CI 8.1 ´ 10-7 – 8.9 ´ 10-7). A
comparison of the crystal structure of ChAT-SERM in complex
with CoA (pdb entry code 2FY4) and the complex between 1-CoA
and ChAT-SERM reveals that several water molecules are
displaced from a hydrophobic pocket by the naphthyl moiety of 1-
CoA, likely contributing to the difference in thermodynamic profile
(Figure 6 B).
Figure 6. The thermodynamics of 1-CoA. (A) The thermodynamic
parameters of 1-CoA and CoA (n=3-4, mean and 95% CI are plotted, the
isotherms are shown in Extended Data Figure 5). B) Water displacement upon
1-CoA binding to ChAT. The binding of 1-CoA displaces water molecules
from the arene-binding hydrophobic pocket and the choline binding site.
The figure was generated by a superpositioning of the crystal structure of
the complex between ChAT-SERM and CoA (pdb entry code 2FY4) and
the crystal structure of the complex between 1-CoA and ChAT-SERM.
Water molecules that are sterically incompatible with the AVP moiety of 1-
CoA are shown as red spheres. Potential hydrogen bonds are shown as
dashed lines, 1-CoA is shown in green and the active site tunnel is shown
as a surface representation.
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202011989
Angewandte Chemie International Edition
RESEARCH ARTICLE
We next investigated the pre-steady-state kinetics of the in situ
assembly of 1-CoA by pre-incubating ChAT with CoA to generate
CoA•ChAT. The reaction was initiated by adding 1, and following
termination, UHPLC-HRMS was used to quantify the production
of 1-CoA. The process was too fast to be resolved at 25 °C by
discontinuous sampling, so the experiments were conducted at
0 °C. We found that the in-situ hydrothiolation reaction exhibited
a distinct burst phase during the first 20 minutes (Figure 7). A
second burst was observed upon adding a second equivalent of
ChAT apo enzyme after 25 min. Both burst phases could be fitted
to the same pseudo-first order association kinetics model. We
determined the rate constant for the pre-steady-state phase (kpss
)
to be 0.05 min-1 (95% CI 0.034 - 0.080) and the doubling time of
1-CoA synthesis to be 12.8 min (95% CI 8.68 - 20.12). The rate
constant during the steady state (kss, i.e. 55-105 min) was
reduced tenfold, to 0.005 min-1 (95% CI 0.0018x - 0.0091). Taken
together, these results show that the hydrothiolation reaction is
fast and that it is the association of 1 to ChAT that limits the rate
of production of 1-CoA during the pre-steady-state phase. The
rate of 1-CoA assembly is reduced due to product inhibition once
a steady state is established.
Figure 8. The dissociation kinetics. A) Results from jump-dilution
experiments in which ChAT (2.7 µM), CoA (80 µM) and 1, 2, 3, 4 or 5 (100
µM) were mixed, incubated, and then diluted 4200-fold. Enzymatic activity
was determined using a fluorescence assay (n=3 independent reactions
each measured in quadruplicates, mean and SD plotted). The solid lines
represent fits of experimental data to the integrated rate equation (1) from
which the dissociation constants (koff) are determined. (B) The narrow
binding site of 1-CoA. The X-ray crystal structure shows that 1-CoA binds
through a constriction of the active site tunnel formed by two b-sheets and
one a-helix.
Figure 7. The kinetics of in situ hydrothiolation. Kinetic analysis of 1-CoA
formation. The reaction was initiated by adding 1 (40 µM) to ChAT (0.9 µM)
preloaded with CoA (80 µM) at 0 °C. A second equivalent of ChAT apo enzyme
was added at 25 min (n = 3, mean and SD plotted, data fitted to a pseudo-first
order association kinetics model, R2 = 0.884 and R2 = 0.877 for the 0-25 min,
and 25-105 min periods, respectively).
Conclusion
We have shown that 1 and CoA are substrates of a ChAT-
catalysed hydrothiolation reaction, and that the inhibition of ChAT
attributed to 1 is actually due to the in situ synthesis of the adduct
1-CoA. This mechanism is common to a representative subset of
AVPs (1-5), suggesting that it is the main mechanism of ChAT
inhibition for this compound class and that thia-Michael acceptor
ability is a crucial component of the SAR of AVPs. Our results
show that the bioactive forms of AVPs are their adducts with CoA,
which thus constitute a new class of ChAT inhibitors. Given these
new insights into the mechanism of ChAT inhibition by AVPs and
the reversibility of adduct formation, we advocate a reassessment
of previous studies on AVPs. Our kinetic data show that the
association of chemically prepared 1-CoA as well as the
dissociation of in situ assembled 1-5-CoA is slow and may involve
a structural reorganisation of ChAT. The slow trafficking may
provide an opportunity to develop non-reactive, tunnel spanning
inhibitors with prolonged target residence time. The X-ray crystal
structure of 1-CoA in complex with ChAT-SERM establishes the
binding mode of a tunnel-spanning inhibitor and provides a
template for such development efforts. Analysis of the
To investigate the dissociation kinetics of in situ assembled
inhibitor adducts, we performed jump-dilution experiments in
which the reactants were mixed, incubated, and then diluted
4200-fold. The dissociation of the adduct was followed by
continuously monitoring the enzymatic activity. We found that the
dilution restored the catalytic activity of ChAT in a time-dependent
manner (Figure 8 A). Furthermore, a non-linear phase was
observed for samples containing in situ assembled 1-5-CoA due
to a slow dissociation of the adducts from ChAT. We determined
the first order rate constant for dissociation (koff) and found that
the target residence time (t) varies between 2.9 and 12.5 minutes
(Table 1). As a reference, the catalytic turnover (kcat) of the natural
reaction is 99.6 s-1 [1], showing that the exit of 1-5-CoA is several
orders of magnitude slower than the exit of CoA. We also note
that inhibitors that have a naphthyl moiety (1,2-CoA) have a
slower dissociation and a longer target residence time than
adducts with a substituted phenyl moiety (3-5-CoA, Table 1).
Entry of chemically prepared 1-CoA into the active site tunnel
of ChAT shown by the inhibition progress curves (Figure 5), as
well as the reversibility of binding of in situ assembled inhibitor
adducts shown by the jump dilute experiments (Figure 8 A) is
unexpected since the X-ray crystal structure shows that the tunnel
is constricted by an a-helix (residue 329-342) and two b-sheets
(residue 435-440 and 537-542, Figure 8 B) that form a clamp
around the propanamido linker of CoA. This suggests that
dynamics, likely involving a significant structural rearrangement
or domain movement of ChAT may be involved in the trafficking
of 1-CoA.
dissociation
kinetics,
shifts
in
thermal
stabilization,
thermodynamics and the X-ray crystal structure show that
interactions between the arene of the AVPs and a hydrophobic
pocket near the choline binding Tyr552 play an important role in
the molecular recognition of AVP-CoA adducts and is a significant
determinant of inhibitor potency. In conclusion, our results
establish the mechanism of ChAT inhibition by AVPs, provide
directions for the development of non-reactive ChAT inhibitors
and illustrate a novel drug modality concept based on in situ
inhibitor assembly from an exogenous precursor and an
endogenously produced co-substrate.
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202011989
Angewandte Chemie International Edition
RESEARCH ARTICLE
Enzymol. 1970, 206, 343–358.
B. R. Baker, R. E. Gibson, J. Med. Chem. 1972, 15,
639–642.
A. Y. Chweh, J. F. DeBernardis, J. F. Siuda, N. G.
Rondan, J. E. Abola, D. J. Abraham, J. Med. Chem.
1984, 27, 825–830.
V. Chandrasekaran, G. B. McGaughey, C. J. Cavallito,
J. P. Bowen, J. Mol. Graph. Model. 2004, 23, 69–76.
B. V. R. Sastry, N. Jaiswal, L. K. Owens, V. E. Janson,
R. D. Moore, J. Pharmacol. Exp. Ther. 1988.
R. Kumar, A. Kumar, B. Långström, T. Darreh-Shori,
Sci. Rep. 2017, DOI 10.1038/s41598-017-16033-w.
R. Kumar, A. Kumar, A. Nordberg, B. Långström, T.
Darreh-Shori, Alzheimer’s Dement. 2020, DOI
10.1002/alz.12113.
A. R. Kim, R. J. Rylett, B. H. Shilton, Biochemistry 2006,
49, 14621–14631.
A. R. Kim, T. Dobransky, R. J. Rylett, B. H. Shilton, Acta
Crystallogr. Sect. D Biol. Crystallogr. 2005, 61, 1306–
1310.
Acknowledgements
[14]
[15]
We thank Dr. Brian Shilton for helpful comments and discussions
during this work. We are also grateful to the staff at ESRF
beamline ID23-2, David C. Andersson for help with the CD
measurements, and Kristoffer Brännström at the Biochemical
Imaging Centre at Umeå University for assistance with the ITC
measurements.
Funding We thank Swedish Ministry of Defence, the Knut and
Alice Wallenberg foundation program NMR for life, SciLifeLab.
NH is funded by the Industrial Doctoral School for Research and
Innovation at Umeå University and by the Swedish Defence
Research Agency (FOI).
[16]
[17]
[18]
[19]
Competing interests The authors declare no competing
interests.
[20]
[21]
Data and material availability: The coordinate file is available at
rcsb.org with accession number 7AMD.
[22]
[23]
[24]
[25]
[26]
C. J. Cavallito, H. S. Yun, T. Kaplan, J. C. Smith, F. F.
Foldes, J. Med. Chem. 1970, 13, 221–224.
B. R. Baker, R. E. Gibson, J. Med. Chem. 1971, 14,
315–322.
R. L. Ryan, W. O. McClure, Neurochem. Res. 1981, 6,
163–173.
N. B. Mehta, D. L. Musso, H. L. White, Eur. J. Med.
Chem. 1985, 20, 443–446.
J. F. DeBernardis, P. Gifford, M. Rizk, R. Ertel, D. J.
Abraham, J. F. Siuda, J. Med. Chem. 1988, 31, 117–
121.
E. Alcalde, T. Roca, A. Barat, G. Ramirez, P. Goya, A.
Martinez, Bioorganic Med. Chem. Lett. 1992, 12, 1493–
1496.
C. J. Cavallito, H. S. Yun, J. Crispin Smith, F. F. Foldes,
J. Med. Chem. 1969, 12, 134–138.
C. J. Cavallito, H. S. Yun, M. L. Edwards, C. J.
Cavallito, H. S. Yun, M. L. Edwards, F. F. Foldes, J.
Med. Chem. 1971, 14, 130–133.
K. L. Dunbar, D. H. Scharf, A. Litomska, C. Hertweck,
Chem. Rev. 2017, DOI 10.1021/acs.chemrev.6b00697.
H. Abdi, L. J. Williams, Wiley Interdiscip. Rev. Comput.
Stat. 2010, DOI 10.1002/wics.101.
Keywords: Choline acetyltransferase; in situ assembly;
hydrothiolation, Coenzyme A, Drug discovery
[1]
[2]
K. Ohno, A. Tsujino, J. M. Brengman, C. M. Harper, Z.
Bajzer, B. Udd, R. Beyring, S. Robb, F. J. Kirkham, A.
G. Engel, Proc. Natl. Acad. Sci. U. S. A. 2001, 98,
2017–2022.
M. Rosas-Ballina, P. S. Olofsson, M. Ochani, S. I.
Valdés-Ferrer, Y. A. Levine, C. Reardon, M. W. Tusche,
V. A. Pavlov, U. Andersson, S. Chavan, T. W. Mak, K.
J. Tracey, Science (80-. ). 2011, DOI
[27]
[28]
[29]
10.1126/science.1209985.
[3]
C. Reardon, G. S. Duncan, A. Brüstle, D. Brenner, M.
W. Tusche, P. Olofsson, M. Rosas-Ballina, K. J. Tracey,
T. W. Mak, Proc. Natl. Acad. Sci. U. S. A. 2013, DOI
10.1073/pnas.1221655110.
[30]
[31]
[32]
[4]
[5]
[6]
T. D. Bird, S. Stranahan, S. M. Sumi, M. Raskind, Ann.
Neurol. 1983, 14, 284–293.
C. N. Karson, M. F. Casanova, J. E. Kleinman, W. S. T.
Griffin, Am. J. Psychiatry 1993, 150, 454–459.
M. A. Cox, G. S. Duncan, G. H. Y. Lin, B. E. Steinberg,
L. X. Yu, D. Brenner, L. N. Buckler, A. J. Elia, A. C.
Wakeham, B. Nieman, C. Dominguez-Brauer, A. R.
Elford, K. T. Gill, S. P. Kubli, J. Haight, T. Berger, P. S.
Ohashi, K. J. Tracey, P. S. Olofsson, T. W. Mak,
Science (80-. ). 2019, 363, 639–644.
M. W. Pantoliano, E. C. Petrella, J. D. Kwasnoski, V. S.
Lobanov, J. Myslik, E. Graf, T. Carver, E. Asel, B. A.
Springer, P. Lane, F. R. Salemme, J. Biomol. Screen.
2001, 6, 429–440.
L. T. Potter, V. A. Glover, J. K. Saelens, J. Biol. Chem.
1968, 243, 3864–3870.
R. Kumar, B. Långström, T. Darreh-Shori, Sci. Rep.
2016, 6, 31247.
L. B. Hersh, B. Coe, L. Casey, J. Neurochem. 1978, 30,
[33]
[34]
[35]
[7]
[8]
P. S. Olofsson, B. E. Steinberg, R. Sobbi, M. A. Cox, M.
N. Ahmed, M. Oswald, F. Szekeres, W. M. Hanes, A.
Introini, S. F. Liu, N. E. Holodick, T. L. Rothstein, C.
Lövdahl, S. S. Chavan, H. Yang, V. A. Pavlov, K.
Broliden, U. Andersson, B. Diamond, E. J. Miller, A.
Arner, P. K. Gregersen, P. H. Backx, T. W. Mak, K. J.
Tracey, Nat. Biotechnol. 2016, 34, 1066–1071.
A. P. Gray, R. D. Platz, T. C. P. Chang, T. R.
Henderson, K. Takahashi, K. L. Dretchen, J. Med.
Chem. 1988, 31, 807–814.
1077–1085.
[9]
J. C. Smith, C. J. Cavallito, F. F. Foldes, Biochem.
Pharmacol. 1967, 16, 2438–2441.
[10]
S. -M Aquilonius, L. Frankenberg, K. -E Stensiö, B.
Winbladh, Acta Pharmacol. Toxicol. (Copenh). 1971,
30, 129–140.
[11]
[12]
[13]
B. A. Hemsworth, F. F. Foldes, Eur. J. Pharmacol.
1970, 11, 187–194.
H. L. White, C. J. Cavallito, BBA - Enzymol. 1970, 206,
242–251.
H. L. White, C. J. Cavallito, Biochim. Biophys. Acta -
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202011989
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
We show that arylvinylpyridiniums (AVPs), the most widely studied class of inhibitors of the potential drug target choline
acetyltransferase (ChAT), act as substrate in an unusual coenzyme A-dependent hydrothiolation reaction. This in-situ synthesis
yields an adduct that is the actual enzyme inhibitor. Our findings clarify the inhibition mechanism of AVPs and establish a drug
modality that exploits a target-catalyzed reaction between exogenous and endogenous precursors.
7
This article is protected by copyright. All rights reserved.