Development of acetophenone ligands as potential neuroimaging agents for cholinesterases

Article abstract of DOI:10.1016/j.bmc.2016.08.048

Association of cholinesterase with β-amyloid plaques and tau neurofibrillary tangles in Alzheimer's disease offers an opportunity to detect disease pathology during life. Achieving this requires development of radiolabelled cholinesterase ligands with high enzyme affinity. Various fluorinated acetophenone derivatives bind to acetylcholinesterase with high affinity, including 2,2,2-trifluoro-1-(3-dimethylaminophenyl)ethanone (1) and 1-(3-tert-butylphenyl)-2,2,2-trifluoroethanone (2). Such compounds also offer potential for incorporation of radioactive fluorine (¹⁸F) for Positron Emission Tomography (PET) imaging of cholinesterases in association with Alzheimer's disease pathology in the living brain. Here we describe the synthesis of two meta-substituted chlorodifluoroacetophenones using a Weinreb amide strategy and their rapid conversion to the corresponding trifluoro derivatives through nucleophilic substitution by fluoride ion, in a reaction amenable to incorporating¹⁸F for PET imaging. In vitro kinetic analysis indicates tight binding of the trifluoro derivatives to cholinesterases. Compound 1 has a K_ivalue of 7?nM for acetylcholinesterase and 1300?nM for butyrylcholinesterase while for compound 2 these values are 0.4?nM and 26?nM, respectively. Tight binding of these compounds to cholinesterase encourages their development for PET imaging detection of cholinesterase associated with Alzheimer's disease pathology.

Full text of DOI:10.1016/j.bmc.2016.08.048

Accepted Manuscript
Development of acetophenone ligands as potential neuroimaging agents for
cholinesterases
Courtney T. Jollymore-Hughes, Ian R. Pottie, Earl Martin, Terrone L.
Rosenberry, Sultan Darvesh
PII:
S0968-0896(16)30656-3
DOI:
http://dx.doi.org/10.1016/j.bmc.2016.08.048
BMC 13231
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
18 July 2016
22 August 2016
26 August 2016
Please cite this article as: Jollymore-Hughes, C.T., Pottie, I.R., Martin, E., Rosenberry, T.L., Darvesh, S.,
Development of acetophenone ligands as potential neuroimaging agents for cholinesterases, Bioorganic &
Medicinal Chemistry (2016), doi: http://dx.doi.org/10.1016/j.bmc.2016.08.048
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Development of acetophenone ligands as potential neuroimaging agents for cholinesterases
Courtney T. Jollymore-Hughesa, Ian R. Pottieb,c, Earl Martinb, Terrone L. Rosenberryd, Sultan
Darvesha,b, e,*
aDepartment of Medical Neuroscience, Dalhousie University, Halifax, Nova Scotia, Canada
bDepartment of Chemistry, Mount Saint Vincent University, Halifax, Nova Scotia, Canada
cDepartment of Chemistry, Saint Mary’s University, Halifax, Nova Scotia, Canada
dDepartment of Neuroscience, Mayo Clinic, Jacksonville, Florida, USA
eDepartment of Medicine (Neurology and Geriatric Medicine), Dalhousie University, Halifax,
Nova Scotia, Canada
Corresponding Author*
Sultan Darvesh
Mailing address: Room 1308, Camp Hill Veterans’ Memorial Building 1
5955 Veterans’ Memorial Lane, Halifax, Nova Scotia, Canada, B3H 2E1.
Phone: 902-473-2490.
Fax: 902-473-7133.
E-mail: sultan.darvesh@dal.ca.
1

ABSTRACT
Association of cholinesterase with β-amyloid plaques and tau neurofibrillary tangles in
Alzheimer’s disease offers an opportunity to detect disease pathology during life. Achieving this
requires development of radiolabelled cholinesterase ligands with high enzyme affinity. Various
fluorinated acetophenone derivatives bind to acetylcholinesterase with high affinity, including
2,2,2-trifluoro-1-(3-dimethylaminophenyl)ethanone (1) and 1-(3-tert-butylphenyl)-2,2,2-
trifluoroethanone (2). Such compounds also offer potential for incorporation of radioactive
fluorine (18F) for positron emission tomography (PET) imaging of cholinesterases in association
with Alzheimer’s disease pathology in the living brain. Here we describe the synthesis of two
meta-substituted chlorodifluoroacetophenones using a Weinreb amide strategy and their rapid
conversion to the corresponding trifluoro derivatives through nucleophilic substitution by
fluoride ion, in a reaction amenable to incorporating 18F for PET imaging. In vitro kinetic
analysis indicates tight binding of the trifluoro derivatives to cholinesterases. Compound 1 has a
Ki value of 7 nM for acetylcholinesterase and 1300 nM for butyrylcholinesterase while for
compound 2 these values are 0.4 nM and 26 nM, respectively. Tight binding of these
compounds to cholinesterase encourages their development for PET imaging detection of
cholinesterase associated with Alzheimer’s disease pathology.
2

Graphical Abstract
3

Highlights
1. Trifluoroacetophenones have high affinities for cholinesterases
2. Chlorodifluoroacetophenones are suitable precursors for trifluoro derivatives
3. Precursors are rapidly converted to trifluoro analogues via F- halogen exchange
4. 18F- incorporation can provide imaging agents for cholinesterases in Alzheimer pathology
Abbreviations
Aβ: β-amyloid, AD: Alzheimer’s disease, AChE: acetylcholinesterase, BuChE:
butyrylcholinesterase, BuLi: butyl lithium, DTNB: 5, 5'-dithiobis 2-nitrobenzoic acid, HPLC:
high performance liquid chromatography, HRMS: high resolution mass spectrum; LRMS: low
resolution mass spectrum, NFTs: neurofibrillary tangles, PET: Positron Emission Tomography,
TMTFA: m-(N,N,N-trimethylammonium)trifluoroacetophenone.
Keywords
Cholinesterases, Positron Imaging Tomography, Alzheimer’s Disease, Acetylcholinesterase,
Butyrylcholinesterase
4

1. Introduction
Alzheimer’s disease (AD) is the most common neurodegenerative disorder that causes
dementia.1 Neuropathological hallmarks of this disease include β-amyloid (Aβ) plaques and tau
neurofibrillary tangles (NFTs)2 which are required indicators for definitive diagnosis of AD,
post-mortem. Since AD can exhibit many clinical symptoms in common with other forms of
dementia,3 it is imperative to develop methods to aid in early definitive diagnosis of AD during
life4 to effect timely treatment of the disease.
Much effort has been devoted to developing imaging agents to detect Aβ in plaques or tau in
NFTs.5-10 Three Positron Emission Tomography (PET) imaging agents were recently approved
for use in imaging Aβ deposits but were not approved as specific agents for diagnosing AD11
because AD pathology can also occur in the brain tissue of up to 30% of cognitively normal
older individuals.12, 13 Therefore, the use of specific Aβ or tau imaging agents could generate
many false positive indications of AD11 and are insufficient for a definitive AD diagnosis during
life.14
The cholinesterases, acetylcholinesterase (AChE) and butyrylcholinesterase (BuChE), are co-
regulators of acetylcholine. These enzymes have been found to co-localize with Aβ plaques that
accumulate in various regions of the AD brains, especially in the cerebral cortex.15-18 In the
normal human brain BuChE activity is associated with certain populations of neurons distinct
from those with AChE activity.19-21 In AD, there is an association of cholinesterase activity,
particularly that of BuChE, with Aβ plaques.15-18 Imaging of cholinesterase activity in Aβ
plaques22 has the potential to permit differentiation between such plaques in the brains of
cognitively normal older adults from those in AD brains.16
5

Investigations into the distribution of cholinesterases in the brain have been conducted with
radiolabelled ligands.23 A number of 11C cholinesterase-targeting radioligand substrates have
been prepared and tested for the ability to image brain cholinesterase activity. These include N-
11C-methylpiperidyl acetate and N-11C-methylpiperidyl propionate24, 25 for detection of AChE
and N-11C-methylpiperidyl butyrate24, 26-28 for detection of BuChE. However, these agents have
yielded only partial recapitulation of known cholinesterase distribution in human brains.23, 26-28
These substrate-radiotracers, however, have provided proof-of-concept that a sufficient amount
survives hydrolysis by cholinesterases in the periphery, cross the blood-brain-barrier, bind to and
are hydrolyzed by, cholinesterases to provide PET images through metabolic trapping.23
Several cholinesterase inhibitors have also been radiolabelled with 11C as potential
cholinesterase imaging agents. These include 11C-donepezil, 11C-methyltacrine, and 11C-
physostigmine, which also have had limited success in demonstrating the known histochemically
defined cholinesterase distribution in the brain.23 Thus, there is a need to develop more effective
radioligands that can detect, in the living brain, the distribution of cholinesterases characteristic
of AD pathology.
A number of trifluorinated acetophenone derivatives have been shown to be very potent, time-
dependent inhibitors of AChE.29, 30 Foremost among these is N,N,N-trimethyl ammonium
trifluoroacetophenone (TMTFA), which has very high affinity for AChE.31 However, this
derivative is a quaternary ammonium cation, a property that may represent a possible
impediment to crossing the blood-brain barrier for PET imaging cholinesterase activity in the
brain. Somewhat less potent neutral trifluoroacetophenones have also been shown to bind to and
inhibit AChE29 and, being uncharged, should cross the blood-brain barrier more readily with
estimated LogP values between 2.7 and 3.5. Such 18F-radiolabelled compounds could potentially
6

be used for PET imaging of cholinesterase-associated pathology in the living brain.
Radiolabelling of acetophenones has been reported through electrophilic addition using 18F-19F
gas.32 The focus of the present work is to develop a generally applicable methodology involving
nucleophilic substitution of 18F- for chlorine in chlorodifluoro acetophenone derivatives, thereby
avoiding the inherent difficulties of handling 18F-19F gas. Herein we report the synthesis of m-
(N,N-dimethylamino)trifluoroacetophenone (1) and m-(tert-butyl)trifluoroacetophenone (2) and a
comparable synthesis of the chlorodifluoro analogues 3 and 4 as precursors for conversion to 1
and 2 by displacement of a chlorine leaving group by F-,33 a reaction amenable to incorporation
of 18F-. Each of the synthesized acetophenones (1-4, Figure 1) was tested for its ability to inhibit
the enzyme activities of human AChE and BuChE, and all exhibited affinity for both
cholinesterases. The reactions of the chlorodifluoromethylketones with AChE have not been
reported or quantified previously, and reactions of none of the four compounds with BChE have
been previously described. The Weinreb amide strategy for the synthesis of the acetophenone
derivatives is consistent with expectations from previous literature reports and represent a
valuable practical addition for the preparation of PET tracers. Additional in vivo work is beyond
the scope of this report but will be conducted in the future.
7

Figure 1. Chemical structures of 1-[3-(dimethylamino) phenyl]-2, 2, 2-trifluoroethanone (1), 1-
(3-tert-butylphenyl)-2,2,2-trifluoroethanone (2), 1-[3-(dimethylamino) phenyl]-2-chloro-2,2-
difluoroethanone (3) and 1-(3-tert-butylphenyl)-2-chloro-2,2-difluoroethanone (4).
The ability of the trifluoroacetophenones to bind strongly to cholinesterases suggests their
potential as PET imaging agents for detecting the association of cholinesterases with AD
pathology.
2. Results
2.1. Synthesis of acetophenone derivatives from bromobenzenes. Four meta-substituted
acetophenones (Figure 1), 2,2,2-trifluoro-1-(3-dimethylaminophenyl)ethanone (1), 1-(3-tert-
butylphenyl)-2,2,2-trifluoroethanone (2), 2-chloro-2,2-difluoro-1-(3-
dimethylaminophenyl)ethanone (3) and 2-chloro-2,2-difluoro-1-(3-tert-butylphenyl)ethanone
(4), were synthesized, as outlined in Scheme 1, from a reaction between an organolithium
reagent and a Weinreb amide, which was synthesized according to Scheme 2, to generate the
desired acetophenone derivatives with non-optimized yields of 42-79%.
8

Scheme 1. Synthesis of meta-substituted acetophenones. Reagents and conditions (a) THF,
-90°C, 1.6M BuLi in hexanes, 15 min (b) For X = F: N-methoxy-N-methyl-2,2,2-
trifluoroacetamide, for X = Cl: N-methoxy-N-methyl-2- chloro-2,2-difluoroacetamide, 30 min at
-78°C then -78°C to room temperature.
Scheme 2. The synthetic strategy to produce N-methoxy-N-methyl-2-chloro-2,2-
difluoroacetamide.
In this method, a meta-substituted bromobenzene, having either m-dimethylamino or m-tert-butyl
group, was mixed with butyl lithium at −90°C to generate the corresponding lithium salt through
a halogen-metal exchange reaction. The resulting organolithium reagent was then mixed with
either commercially available N-methoxy-N-methyl-2,2,2-trifluoroacetamide or freshly prepared
N-methoxy-N-methyl-2-chloro-2,2-difluoroacetamide (Scheme 2) to directly give the
corresponding products (1-4; Figure 1). The chlorodifluoroacetophenones 3 and 4 were found to
9

have good stability (several weeks) when stored at 5°C in the dark under an atmosphere of
anhydrous argon.
2.2. Conversion of chlorodifluoroacetophenone precursors 3 or 4 to trifluoroacetophenone
products 1 or 2. We examined the facility with which the chlorodifluoroacetophenones 3 and 4
were converted to the corresponding trifluoroacetophenone products 1 and 2 with a method
(Scheme 3) that could be adapted for radiolabelling with nucleophilic 18F-.
Scheme 3. Conversion of a meta-substituted chlorodifluoroacetophenone to a meta-substituted
trifluoroacetophenone using a fluorination procedure. Reagents and conditions: (c) KF,
Kryptofix®, K2CO3, CH3CN, 110°C, 30 min.
The chlorine atom in compounds 3 and 4 was displaced with fluoride anion to produce a chloride
anion and the desired trifluoro compounds by a modification of earlier fluorination reaction
procedures.33, 34 To simulate the process that will be used with radioactive 18F-, identical steps
were carried out with non-radioactive fluoride ion isotope (19F-) as the nucleophilic agent.
Production of 18F-, generated from bombardment of H2 O in a cyclotron, is subject to
18
10

contamination by metal ions. These contaminants are readily removed by chromatographic
separation using the Sep-Pak® Plus QMA system. In addition, since F- is subject to being
surrounded by an aqueous solvent shell, which reduces its nucleophilicity, a crown ether,
4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo[8.8.8]hexacosane (Kryptofix®), was employed.33, 34
The Kryptofix®/fluoride mixture was then dried by azeotropic distillation with acetonitrile. The
dried fluoride reagent was mixed with the cholorodifluoro precursor (3 or 4) in anhydrous
acetonitrile and the reaction container was sealed with a cap under an atmosphere of argon and
heated to 110°C for 30 min for compound 3 and 85°C for 30 min for compound 4. The product
in the cooled reaction mixture was first purified by HPLC, using a C18 reverse phase column,
with acetonitrile as the mobile phase (see Materials and Methods). The total time to obtain the
purified trifluoro product in this simulated radiolabelling was approximately 60 min. The
trifluoro products, obtained from the nucleophilic displacement of the chlorine atom with
fluoride ion, using Scheme 3, corresponded to the same compounds obtained from the method
outlined in Scheme 1. That is, mass spectral data (Supplementary Figure S25 and S26) for the
trifluoro compounds 1 and 2 obtained via Scheme 3 were identical to those for 1 and 2
synthesized via Scheme 1.
2.3. Inhibition of AChE by acetophenones 1 - 4. All synthesized acetophenone derivatives
were evaluated as potential ligands for human AChE. Although not a focus in this study as
potential imaging agents, the precursor chlorodifluoroacetophenones (3 and 4) were also
examined in order to note any significant differences in structure-activity relationships resulting
from the halogen exchange. Previously, a number of fluoroacetophenones have been observed to
react with AChE on a time scale of s to min.29 Therefore, the derivatives prepared here (1 - 4,
11

Scheme 1) were initially examined for any time-dependent reaction with human AChE using a
modification37 of the Ellman method.38 Three of the four compounds clearly exhibited this type
of inhibition of AChE (Table 1).
Table 1. Kinetic constants for the interaction of acetophenone-based ligands with
acetylcholinesterasea
3
1
2
4
ka
0.018 0.021
0.024 0.004b
n/a
0.036 0.013
(nM-1min-1)
kd
0.36 0.39
7 2
0.014 0.006b
0.4 0.1
n/a
0.08 0.08
2.9 0.6
(min-1)
Ki (nM)
experimental
Ki (nM)
calculated as kd/ka
42 7
n/a
20 31
0.56 0.27b
2.1 0.9
aTrifluoro ketones are known to be equilibrium mixtures of the free ketone and the ketone
hydrate, and values of ka and Ki can be corrected for hydration.29 However, no corrections were
made here either for hydration or for competition between substrate and inhibitor for the enzyme
active site.29
bValues of kobs were fitted to Equation 3, and ka and kd were calculated from Equations 4 – 6.29
Enzyme and acetophenone were incubated in 0.1M phosphate buffer (pH 7.4) containing
Ellman’s reagent; aliquots were removed at set time points up to 70 min, and assays were
12

initiated with the addition of acetylthiocholine. Assay rate values (v in ∆A/min) were plotted
against time for several acetophenone concentrations, as illustrated in the Figure 2A inset for
compound 2.
A
0.2
[compound 2], nM
0.4
1.7
3.3
-1
k, min
v = ∆A/min
17
33
0.1
0.2
0.0
0
30
60
0.0
time, min
0
20
40
compound 2, nM
13

[compound 2], nM
B
0
0.6
24
12
0
0.17
0.33
v-1 min/∆A
slopes
0.3
0.0
0
15
30
0.0
0.3
[AcSCh]-1 mM-1
[compound 2], nM
Figure 2. Inhibition of AChE by tert-butyl trifluoroacetophenone (2). A) Time-dependent
inhibition is shown by values of kobs determined by non-linear regression analysis (Equation 2) of
the change in enzyme activity (v in ∆A/dt) over time for inhibitor concentrations 1.7, 3.3, 17, and
33 nM (inset). Plotting kobs vs. the concentration of 2 according to Equations 3 – 6 gave the
fitted values of ka and kd shown in Table 1. B) Lineweaver-Burk plots obtained from AChE
activities (v) at various concentrations of acetylthiocholine substrate ([AcSCh]) and the indicated
concentrations of compound 2. Enzyme and inhibitor were incubated for 30 min before initiating
the assays by addition of substrate. Plots of 1/v vs. 1/[AcSCh] were generated (left panel), and
replots of the slopes of these lines against [compound 2] (right panel) gave Ki (see Table 1 for
corrected value).
14

These plots were fitted by nonlinear regression according to Equation 1 to give the pseudo first-
order rate constant kobs.
v = vf + (v0 - vf)exp(-kobst)
(1)
Previous reports have indicated that replots of kobs versus inhibitor concentration are often linear,
corresponding to Scheme 4 and Equation 2.29,31
Scheme 4. Enzyme (E) and Inhibitor (I) interaction showing one equilibrium
kobs = kon[I] + koff
(2)
However, the replots of kobs versus inhibitor concentration in this report were all found to be
nonlinear, and they could be fitted by Equation 3.
kobs = k4 + k3[I]/(K21 + [I])
where
(3)
K21 = k2/k1
(4)
(5)
(6)
(7)
ka = k3[I]/K21
kd = k3 k4/(k3 + k4)
Ki = kd/ka
15

Equation 3 corresponds to a limiting case of the two-step equilibrium in Scheme 5 in which a
rapid initial inhibitor binding is followed by a slower re-equilibration of the complex, as was
proposed by Nair et al,.29 to fit nonlinear replots of kobs versus acetophenone concentration.
Scheme 5. Enzyme (E) and Inhibitor (I) interaction showing two successive equilibrias.
The replot of kobs versus the concentration of 2 is shown in Figure 2A, and the nonlinear fit to
Equation 3 allowed determination of an overall association rate constant ka = 0.024 nM-1min-1
and an overall dissociation rate constant kd = 0.014 min-1 from Equations 4 – 6 (see Table 1).
Compounds 1 and 4 also showed time dependent inhibition of AChE, and values of ka and kd for
the reaction of these compounds with AChE were obtained with Equations 3 – 6 and are shown
in Table 1. This value of kd (0.014 min-1) corresponds to a half-life of approximately an hour for
the complex of compound 2 with AChE, long enough to allow good wash-out of unreacted
compound 2 but short enough to avoid toxicity resulting from too much AChE inactivation.
Apparent equilibrium constants Ki for inhibitor binding to AChE were obtained from steady-
state measurements following pre-incubation of enzyme and inhibitor. The analysis is illustrated
for compound 2 in Figure 2B. Lineweaver-Burk double reciprocal plots were obtained in the
presence and absence of inhibitor. A replot of the slopes of the lines versus inhibitor
concentration gave Ki as the intercept on the x-axis (Table 1, Ki experimental). Since Ki is a
16

measure of the equilibrium dissociation constant of inhibitor and enzyme, a smaller Ki value
indicates a higher affinity for the enzyme. Values of Ki obtained for the other acetophenones 1,
3, and 4 with AChE also are listed in Table 1. The ratio of kd to ka also should equal Ki
(Equation 7 and see Reference 29), and comparison of this calculated Ki with that obtained from
the steady-state measurements provides an additional check on the consistency of the kinetic
data.
Ki = kd/ka
(7)
Calculated Ki values are also shown in Table 1, and their agreement with the experimental Ki
values is within the experimental error.
2.4. Inhibition of BuChE by acetophenones 1 - 4. Only one of the four acetophenones,
compound 2, showed time-dependent inhibition of BuChE (Figure 3A).
17

2
1
0
A
-1
k, min
0.4
v = ∆A/min
[compound 2], nM
0.2
3.3
6.7
33
0.0
0
6
12
time, min
0
20
40
compound 2, nM
[compound 2], nM
B
18
0
13
67
0.3
v-1 min/∆A
slopes
9
0.0
0
0
15
30
0
60
[BuSCh]-1 mM-1
[compound 2], nM
18

Figure 3. Inhibition of BuChE by tert-butyl trifluoroacetophenone (2). Time-dependent (A) and
steady-state inhibition (B) were measured with BuChE and butyrylthiocholine (BuSCh) as in
Figure 2 at the indicated concentrations of compound 2 and 15 min preincubation of enzyme and
inhibitor. Parameter values are shown in Table 2.
Analytical procedures for the time-dependent and the steady-state assays were as described for
AChE except that the substrate butyrylthiocholine was used in place of acetylthiocholine.
Results are summarized in Table 2.
19

Table 2. Kinetic constants for the interaction of acetophenone-based ligands with
butyrylcholinesterasea
3
1
2
4
ka
n/a
0.13 0.07b
n/a
n/a
(nM-1min-1)
kd
n/a
3.4 2.1c
26 8
n/a
n/a
n/a
(min-1)
Ki (nM)
experimental
Ki (nM)
1300 200
n/a
1900 300
n/a
32 7
n/a
calculated as
kd/ka
aTrifluoro ketones are known to be equilibrium mixtures of the free ketone and the ketone
hydrate, and values of ka and Ki can be corrected for hydration.29 However, no corrections were
made here either for hydration or for competition between substrate and inhibitor for the enzyme
active site.29
bValues of kobs were fitted to Equation 3, and ka and kd were calculated from Equations 4 – 6.29
cThe value of k4 fitted from Equation 3 was too close to zero to allow an estimate of kd. The kd
shown was calculated from kd = kaKi, where Ki was the experimental Ki.
20

The ka for 2 with BuChE was about five-fold higher than that with AChE, while the calculated kd
for 2 with BuChE was more than 100-fold larger than that with AChE (Tables 1 and 2). The
reversible inhibition constants (Ki) obtained from the BChE steady-state data for 2 (Figure 3B)
and Ki values for the other three acetophenones are shown in Table 2. The tert-butyl derivatives,
2 and 4, were the most potent reversible inhibitors of BuChE but the respective Ki values for all
four acetophenones with BuChE were higher than the corresponding Ki values with AChE,
indicating a lower affinity for BuChE. The Ki for BuChE was 180-fold higher for 1, 70-fold
higher for 2, 45-fold higher for 3, and 11-fold higher for 4, compared to the same values for
AChE (Tables 1 and 2).
3. Discussion
Current attempts to obtain a definitive diagnosis of AD through specific amyloid neuroimaging
fall short since β-amyloid is frequently found in brains of cognitively normal older adults.12, 13 In
contrast, using available neuroimaging techniques to detect the characteristic association of
cholinesterases with AD pathology may provide a more definitive diagnosis during life. The
development of ligands that bind tightly to cholinesterases and that are also amenable to efficient
radiolabelling, with isotopes like 18F for PET imaging, may offer new opportunities to reveal AD
pathology in the living brain and enable more timely diagnosis and monitoring of treatment for
the disease. Trifluoroacetophenones are good candidates for such ligands.
Certain acetophenones have been shown29 to inhibit AChE through covalent interaction
between the carbonyl functionality of the ketone inhibitor and the active site serine of the
enzyme (Scheme 5).
21

Scheme 6. Hydration and hemiketal equilibria for acetophenone compounds.
This interaction is facilitated by enhancing the electrophilicity of the ketone carbonyl carbon
with adjacent electron-withdrawing groups such as a trifluoromethyl group (e.g., Compounds 1
and 2; Figure 1). Replacement of hydrogen atoms with small electronegative fluorine atoms
produces a bioisostere40 with comparable structure but enhanced ability to form a hemiketal
intermediate that blocks substrate access to the enzyme. Trifluoromethyl ketones are known to be
potent inhibitors of AChE.41-43 The nucleophile reacting with the ketone is the activated
alkoxide of the catalytic serine residue of the enzyme (Ser203 in human AChE), that results in
the formation of the hemiketal-enzyme complex29 (Scheme 5).44,45
The present study sought to determine if chlorodifluoroacetophenones, such as 3 and 4, were
amenable to a radiolabelling methodology to generate potential 18F imaging agents with high
cholinesterase affinity. A number of procedures have been reported40 for the synthesis of
trifluoromethyl ketones, including oxidation of α-trifluoromethyl alcohols, electrophilic addition
22

of trifluoroacetic anhydride, alkylation and decarboxylation of fluorinated β-ketoesters or, as
described here, through the use of organometallic intermediates. The latter approach (Scheme 1)
was chosen as it offered a rapid direct method to obtain the desired m-substituted
chlorodifluoroacetophenones to be precursors for fluoride exchange reactions. Since the
chlorodifluoro derivatives were not commercially available, these were synthesized (Scheme 2)
using an organolithium/Weinreb amide strategy. This approach also enabled the direct synthesis
of trifluoroacetophenones for proof of identity with the halogen exchange reaction products and
for the assessment of cholinesterase affinities (Tables 1 and 2). Due to the relatively short half-
life of 18F (t1/2 ~ 110 min), the fluorination reaction to produce the radiolabelled
trifluoroacetophenone ligand and all subsequent purification steps must be performed rapidly to
maximize the specific activity of the radiolabelled product. It is also important to rapidly
separate the desired product from remaining precursor, especially given the structural similarities
of the chlorodifluoro precursors and trifluoro products. Purification by C18 reverse phase HPLC
allowed for rapid separation of product from the precursor. The halogen exchange reaction with
19F isotope, and purification, can be performed within one hour and with reasonable yields.
Considering time to synthesize and yields, the preparation of trifluoroacetophenone compounds
from the chlorodifluoroacetophenone precursors represents a viable method for incorporation of
18F radioisotope for PET imaging of cholinesterase activity associated with AD pathology. The
ability to detect this association by PET imaging is dependent on the affinity of the radioligand
for cholinesterase and the length of time it remains in the enzyme-inhibitor hemiketal complex.
Previous studies have mostly focused on the interaction of trifluoromethyl ketones with
AChE.29, 30, 45, 46 Here, BuChE is also considered since it represents a major enzyme component
associated with AD pathology.15-18 Two trifluoroacetophenones, compounds 1 and 2, were
23

examined as representative potential radioligands. For comparison, the immediate chlorodifluoro
precursors (3 and 4, respectively) were also evaluated kinetically for any significant effects on
enzyme inhibition due to exchanging halogen species. The ability of fluorinated acetophenones
to interact with the catalytic serine of cholinesterases and form a hemiketal complex is indicated
by measurable time-dependent inhibition of the enzyme and determined primarily by the
magnitude of the first order dissociation rate constant kd for the enzyme-inhibitor complex.
Lower kd values are correlated with higher affinity for enzyme as measured by the experimental
Ki (Table 1) and slower attainment of steady state equilibrium (Scheme 4). tert-Butyl
trifluoroacetophenone (2) showed the highest affinity for AChE, with a Ki of 0.36 nM (Table 1).
Earlier reported Ki values for 2 with eel and torpedo AChE (1.9 and 3.6 pM, respectively),29 and
mouse AChE (3.7-3.8 pM).30,47 appear to indicate much higher affinities than that reported here
for 2 with human AChE (Table 1). However, typical of most ketones and aldehydes,
acetophenones are subject to a hydration equilibrium in aqueous solution (Scheme 4), and these
previously reported values were adjusted to take into account the hydration equilibrium constant
Khyd to reflect only the presence of the active ketone form of the molecule. Khyd has been
reported as ~500 for 2.30 Before adjustment for Khyd, the earlier Ki values for 2 become 0.9 and
1.8 nM with eel and torpedo AChE , respectively, and 1.9 nM for mouse AChE. These values
are only slightly higher than the 0.36 nM value in Table 1. Compound 1 also has been reported
to show time-dependent inhibition of eel and torpedo AChE with respective Ki values of 4 nM
and 28 nM29 and of mouse AChE with a Ki of 15 nM,30 all prior to correction for hydration, in
reasonable agreement with the Ki value of 7 nM for human AChE in Table 1. The Ki values
reported in Tables 1 and 2 were not adjusted for hydration to better reflect the actual activity of
24

these molecules in living systems where both the hydrated and nonhydrated forms will be present
during imaging studies.
In contrast to AChE, BuChE (Table 2) exhibited a measurable time-dependent inactivation
only with compound 2 among the four acetophenones in this study. Values of Ki obtained from
steady-state inhibition of BuChE by all four acetophenones were, as noted in the Results, 10- to
200-fold greater (indicating lower affinity) than those for the same compounds with AChE
(Tables 1 and 2).
The three acetophenones that showed time-dependent inhibition of AChE gave association rate
constants (ka) in excess of 107 M-1min-1 (Table 1). The ka values for 1 and 2 are within the range
of those reported for eel, torpedo, and mouse AChEs.29,30,47 Only compound 2 provided a
measurable ka value with BuChE; compounds 1, 3 and 4 attained steady state equilibrium with
BuChE so rapidly that no reaction rate constants could be measured here. The ka value of 13 x
107 M-1min-1 for 2 with BuChE (Table 2) was five-fold more than the ka for 2 with AChE (Table
1), consistent with the larger and more accessible active site gorge in BuChE.48,49 The
observation of a nonlinear dependence of kobs on the inhibitor concentration for all of the time-
dependent reactions in Tables 1 and 2 indicated that the successive equilibria for acetophenone
binding in Scheme 5 and Equations 3 – 6 were more appropriate here than the single equilibrium
in Scheme 4 and Equation 2. Scheme 5 and Equations 3 - 6 were invoked previously to account
for the binding of compound 1 to eel and torpedo AChE,29 but it is noteworthy that in that report
Scheme 5 and Equation 2 were sufficient to account for the binding of compound 2 to eel or
torpedo AChE.29
The hemiketal complex formed by acetophenones with cholinesterases is analogous to the
tetrahedral transition state intermediate formed during the hydrolysis of ester substrates like
25

acetylcholine, and enzyme stabilization of this complex promotes catalytic hydrolysis. Since all
the acetophenone compounds described here are halogenated methyl ketones, the formation of
the hemiketal is likely to be qualitatively similar for all, and this mechanism alone cannot
account for the differences in the observed affinities of the derivatives with AChE and BuChE.
However, there is a strong correlation between the meta-substituent on a trifluoroacetophenone
and the potency of its inhibition of AChE. The inhibitor m-(N,N,N-
trimethylammonium)trifluoroacetophenone (TMTFA) has the highest reported affinity of any
trifluoroacetophone with AChE.29,30 Its Ki with mouse AChE is 5 fM after correction for
hydration and 0.3 nM before correction, values that are nearly 1000-fold higher after correction
but only 6-fold higher before correction than those for compound 2 with mouse AChE.29,47 The
high in vitro affinity of TMTFA can be attributed largely to a stabilizing cation-π interaction
between the positively charged nitrogen of the trimethylammonio group with the indole ring of a
tryptophan residue (Trp86 in mouse AChE) and, to a lesser extent, to a more electrophilic
carbonyl carbon resulting from this electron-withdrawing group.29 This electrophilic effect could
account for the 6- to 8-fold higher affinities of the trifluoroacetophenones 1 and 2 for AChE
relative to the corresponding chlorodifluoroacetophenones 3 and 4 , which have less electrophilic
carbonyl carbons. In contrast, there are no differences in the affinities of 1 relative to 3 and 2
relative to 4 for BuChE, where hemiketal formation occurs to a lesser extent. Alternatively,
these differences may reflect a smaller acyl pocket in the active site of AChE relative to that of
BuChE.48,49
Tables 1 and 2 indicate that the m-tert-butylacetophenones 2 and 4 are 15- to 50-fold more
potent reversible inhibitors of both AChE and BuChE than the corresponding m-
dimethylaminoacetophenones 1 and 3. It is possible that the lower affinity of the m-
26

dimethylamino derivatives may result from repulsion between the lone pair on the neutral
nitrogen atom in the m-dimethylamino group of the compound and the indole nitrogen of the
corresponding tryptophan residue of human cholinesterases. The absence of such repulsion for
compounds 2 and 4 and the larger size of the tert-butyl moiety may favour inhibitor stabilization
within the active site gorge29 and help to account for their higher affinities relative to the m-
dimethylamino acetophenone compounds (1 or 3) in our studies.
The absence of an observable time dependence of kobs on inhibitor concentration for compound
3 with AChE and compounds 1, 3 and 4 with BuChE parallels a similar absence of time
dependence for some related fluoro ketones with AChE. Nair et al.29 found that trifluoromethyl
acetophenones with small meta-substituents such as methyl, ethyl, amino or nitro groups did not
inhibit AChE in a time-dependent manner but appeared to be reversible inhibitors. They
suggested that the larger meta-substituents in compounds 1 and 2 can undergo London dispersion
interactions in the cholinesterase active site that increase the affinity of trifluoroacetophenones
for the enzyme. Such an increase in affinity adds to their appeal as potential imaging agents.
Selective labeling of the target macromolecule is an important goal in neuroimaging. Because
of their unique character as tetrahedral transition state analogs of cholinesterases,
trifluoroacetophenones should have very high selectivity for AChE and BuChE in the central
nervous system. This high selectivity is based on the high affinity of these compounds for the
AChE and BuChE active sites, and it is enhanced by the slow dissociation rate constants that
several trifluoroacetophenones exhibit with these enzymes. Slow dissociation implies that these
compounds will persist in complex with their target molecules during the washout typical of PET
imaging analysis. These desirable properties are especially evident in m-tert-
butyltrifluoroacetophenone (2), which exhibits both high affinities (low Kis) for the
27

cholinesterase active sites and a low dissociation rate constant kd. These features promise to
ensure more selective binding of 18F labeled acetophenone with cholinesterases associated with
AD pathology for PET detection.
4. Conclusion
Two m-substituted trifluoroacetophenones, 1 and 2, have been prepared in a reaction with
potential to generate 18F radioligands for the neuroimaging of cholinesterases associated with AD
neuropathology. The precursor molecules of the possible radioligands, m-substituted
chlorodifluoroacetophenones 3 and 4, are readily prepared and sufficiently stable to be made in
advance and stored for extended periods without chemical degradation. Kinetic evaluations of
the trifluoroacetophenones 1 and 2 indicate they bind with high affinity to both AChE and
BuChE. The ability to rapidly introduce 19F by converting 3 to 1 and 4 to 2 indicates that these
molecules have potential for development as radioligands for the neuroimaging of
cholinesterases and the localization of these enzymes with AD pathology. The present work
provides the basis for future studies directed towards incorporation of 18F into these molecules
for PET imaging of cholinesterases in the brain.
5. Materials and Methods
5.1. Materials and Reagents. Tetrahydrofuran (HPLC grade, Caledon Labs, CAN) was dried
immediately before use in a solvent purification system from Innovative Technology, Newbury
Port, MA, USA. Diethyl ether (anhydrous, ACS grade) was used as supplied from Caledon Labs,
28

CAN. When needed, acetonitrile (HPLC Grade, Caledon Labs, CAN) was dried by refluxing
and distilling the solvent over CaH2 and immediately used. All other solvents were purchased as
HPLC grade and used as supplied from Caledon Labs, CAN. All glassware was dried in an oven
at 110°C for 24 h then cooled to room temperature inside a desiccator before use. 1-Bromo-3-
tert-butylbenzene was purchased from Oakwood Products, West Columbia, SC, USA. Butyl
lithium (1.6M) in hexanes, 3-bromo-N,N-dimethylaniline, N-methoxy-N-methyl-2,2,2-
trifluoroacetamide, N,O-dimethylhydroxylamine hydrochloride, chlorodifluoroacetic anhydride,
sodium sulfate, potassium fluoride, 4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo[8.8.8]hexacosane
(Kryptofix®), potassium carbonate, acetylthiocholine iodide, S-butyrylthiocholine iodide, 5,5'-
dithiobis(2-nitrobenzoic acid) and purified recombinant human acetylcholinesterase were
purchased from Sigma-Aldrich, USA. Purified human plasma butyrylcholinesterase was a gift
from Dr. Oksana Lockridge (University of Nebraska Medical Center). Chemical reactions were
performed under ultra-high purity argon or nitrogen atmosphere (Air Liquide, CAN). Sep-Pak®
Plus QMA (# WAT020545) were purchased from Waters Corp, USA and used as supplied.
5.2. Characterization of synthesized compounds. Nuclear magnetic resonance spectra were
recorded at the Green Chemistry NMR facility at Saint Mary’s University, Halifax, Nova Scotia
on a Bruker Avance 300, operating at 300.1 MHz for 1H or 75.4 MHz for 13C experiments; or at
the Nuclear Magnetic Resonance Research Resource (NMR-3), Dalhousie University, Halifax,
Nova Scotia, Canada on a Bruker AVANCE 500, operating at 500.1 MHz for 1H or 125.8 MHz
for 13C. Chemical shifts are reported in parts per million relative to Me4Si in CDCl3. For 1H
NMR experiments, the coupling constants are reported in Hertz and the multiplicities are
apparent. Infrared spectra were recorded neat between sodium chloride plates on a Nicolet
29