Discovery of Selective Inhibitors Targeting Acetylcholinesterase 1 from Disease-Transmitting Mosquitoes

Article abstract of DOI:10.1021/acs.jmedchem.6b00967

Vector control of disease-transmitting mosquitoes is increasingly important due to the re-emergence and spread of infections such as malaria and dengue. We have conducted a high throughput screen (HTS) of 17,500 compounds for inhibition of the essential AChE1 enzymes from the mosquitoes Anopheles gambiae and Aedes aegypti. In a differential HTS analysis including the human AChE, several structurally diverse, potent, and selective noncovalent AChE1 inhibitors were discovered. For example, a phenoxyacetamide-based inhibitor was identified with a 100-fold selectivity for the mosquito over the human enzyme. The compound also inhibited a resistance conferring mutant of AChE1. Structure-selectivity relationships could be proposed based on the enzymes' 3D structures; the hits' selectivity profiles appear to be linked to differences in two loops that affect the structure of the entire active site. Noncovalent inhibitors of AChE1, such as the ones presented here, provide valuable starting points toward insecticides and are complementary to existing and new covalent inhibitors.

Full text of DOI:10.1021/acs.jmedchem.6b00967

Subscriber access provided by CORNELL UNIVERSITY LIBRARY
Article
Discovery of selective inhibitors targeting
acetylcholinesterase 1 from disease-transmitting mosquitoes
Cecilia Engdahl, Sofie Knutsson, Fredrik J. Ekström, and Anna Linusson
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b00967 • Publication Date (Web): 06 Sep 2016
Downloaded from http://pubs.acs.org on September 7, 2016
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 free 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 accessible to all
readers and 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.
Journal of Medicinal Chemistry 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 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of selective inhibitors targeting
acetylcholinesterase 1 from diseaseꢀtransmitting
mosquitoes
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
†
,$
†,$
‡*
†*
Cecilia Engdahl, Sofie Knutsson, Fredrik Ekström, and Anna Linusson
†
Department of Chemistry, Umeå University, SEꢀ90187 Umeå, Sweden
‡Swedish Defense Research Agency, CBRN Defense and Security, SEꢀ906 21 Umeå, Sweden
Corresponding author
*
$
These authors contributed equally
Eꢀmail address: anna.linusson@umu.se; fredrik.ekstrom@foi.se
ABSTRACT
Vector control of diseaseꢀtransmitting mosquitoes is increasingly important due to the reꢀ
emergence and spread of infections such as malaria and dengue. We have conducted a high
throughput screen (HTS) of 17,500 compounds for inhibition of the essential AChE1 enzymes
from the mosquitoes Anopheles gambiae and Aedes aegypti. In a differential HTS analysis
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 55
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
including the human AChE, several structurally diverse, potent, and selective nonꢀcovalent
AChE1 inhibitors were discovered. For example, a phenoxyacetamideꢀbased inhibitor was
identified with a 100ꢀfold selectivity for the mosquitoꢀ over the human enzyme. The compound
also inhibited a resistance conferring mutant of AChE1. Structureꢀselectivity relationships could
be proposed based on the enzymes’ 3D structures; the hits’ selectivity profiles appear to be linked
to differences in two loops that affect the structure of the entire active site. Nonꢀcovalent
inhibitors of AChE1, such as the ones presented here, provide valuable starting points towards
insecticides and are complementarity to existing and new covalent inhibitors.
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
INTRODUCTION
Mosquitoes act as vectors for parasitic and viral infectious diseases such as malaria, dengue,
chikungunya, Zika infection, and lymphatic filariasis that affect the health of hundreds of
millions of people every year. Many previously controlled vectorꢀborne diseases are reꢀemerging
1
ꢀ2
or spreading to new parts of the world, partly because of globalization and climate change. For
example, dengue is one of the fastest spreading vectorꢀborne diseases and is now considered a
2ꢀ4
serious public health threat with 40% of the world’s population at risk. The incidence of the
deadliest vectorꢀborne disease, malaria, has declined following a massive global effort over the
5
ꢀ6
last fifteen years, but 1.2 billion people remain at high risk of infection. International initiatives
and campaigns aiming to increase the coverage of current mosquitoꢀcontrol interventions, which
are primarily based on insecticideꢀtreated bed nets (ITN) and indoor residual spraying (IRS),
5
ꢀ6
have helped to significantly reduce the risk of malaria in SubꢀSaharan Africa. These methods
currently rely on four chemical classes of insecticides that are recommended by the World Health
Organization (WHO) for mosquito vector control purposes: chlorinated hydrocarbons,
ACS Paragon Plus Environment

Page 3 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
organophosphates, carbamates, and pyrethroids (Figure 1). All of them are recommended for IRS
while only the pyrethroids are recommended for treatment of bed nets. The insecticidal effects on
the targeted organisms are similar for all chemicals; interference with the nervous system which
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
7
causes paralysis and death of the mosquitoes.
While the largeꢀscale production and widespread use of insecticides has had beneficial effects
in terms of disease control and crop protection it has also caused the insecticides to accumulate in
7ꢀ9
ecosystems, with harmful effects on nonꢀtarget species including humans. Moreover, it has
10ꢀ13
promoted the development and spread of insecticideꢀresistant mosquito populations.
Consequently, there is a need for new vector control strategies, ideally based on combinations of
different approaches, such as biopesticides (i.e. microorganisms, viruses, or natural products that
1
4ꢀ17
target insect vectors) and insecticides with novel mode of actions.
Figure 1. Some representative insecticides currently used for vector control. From the left: the
chlorinated hydrocarbon DDT, the organophosphate fenitrothion, the carbamate propoxur, and
the pyrethroid deltamethrin.
The organophosphate and carbamate insecticides target the essential enzyme
acetylcholinesterase (AChE, EC 3.1.1.7) through a covalent modification of the catalytic serine
8
, 18ꢀ19
residue.
AChE terminates cholinergic transmission through a rapid hydrolysis of the
neurotransmitter acetylcholine. Inhibition of AChE causes an accumulation of acetylcholine,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 55
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
leading to an overstimulation of the nervous system. While current insecticides are very efficient
at controlling diseaseꢀtransmitting mosquitoes, the compounds are nonꢀspecific and inhibit AChE
1
9ꢀ20
enzymes from many different species including humans.
In addition, insecticide insensitive
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
mosquito strains have evolved in several species. Various mutations of mosquitoꢀAChE1 that
confer resistance to existing insecticides have been reported in literature. The most abundant and
wellꢀstudied mutations are the G122S substitution (human AChE (hAChE) numberings are used
throughout the text; G122S corresponds to G119S in Torpedo californica AChE (TcAChE))
2
1
found in Anopheles gambiae and Culex pipiens among others, the F338W (F331W in TcAChE)
2
2
23
in Culex triaeniorhynchus, and the F297V (F290V in TcAChE) in Culex pipiens.
Mosquitoes have two genes encoding AChE enzymes: ace-1 and ace-2, expressing the AChE1
2
4ꢀ26
and AChE2 enzymes, respectively,
while vertebrates have only one gene. In mosquitoes,
AChE1 is the isoform responsible for AChEꢀmediated insecticide resistance and also have higher
21, 27
catalytic activity than AChE2.
The role of AChE2 is still not fully understood. In this work
we are focusing on AChE1 from Anopheles gambiae (AgAChE1; vector of the malaria parasite)
and Aedes aegypti (AaAChE1; vector of the dengue, yellow fever, Zika, and chikungunya
viruses). Both the AChE1 and AChE2 enzymes from Anopheles gambiae and Aedes aegypti
2
8ꢀ33
share many functional and structural characteristics with vertebrate AChEs.
For example,
sequence comparisons, crystallographic studies of enzymes from various species, and homology
modelling of AgAChE1 suggest that the overall three dimensional structures of vertebrate and
3
0, 34ꢀ36
mosquito enzymes are similar.
Despite the similarities between vertebrate and mosquito AChEs, promising covalent inhibitors
of mosquito AChE1 have been obtained using two different design strategies; modification of an
36ꢀ40
free cysteine unique for insect AChEs
and modification of the catalytic serine residue using
Both strategies have yielded covalent inhibitors displaying
2
0, 30, 41ꢀ46
reversible covalent inhibitors.
ACS Paragon Plus Environment

Page 5 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
38, 40ꢀ41
20, 46
selectivity for mosquito AChE1 over hAChE
and other nonꢀtarget organisms.
In a
recent functional comparison of AChEs from vertebrates (hAChE and Mus musculus AChE
(mAChE)) and mosquitoes (AgAChE1 and AaAChE1) we have showed small but significant
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
33
differences in ligand binding properties for nonꢀcovalent inhibitors as well. Also the resistance
conferring G122S mutant of AgAChE1 (G122SꢀAgAChE1) has been targeted by both carbamateꢀ
4
2ꢀ45
47
based covalent inhibitors
and nonꢀcovalent inhibitors.
To explore nonꢀcovalent inhibitors of mosquito AChE1 on a larger scale, we here present a
high throughput screening (HTS) of a library comprising 17,500 compounds. The hits’ chemical
diversity and potential selectivity were investigated in a differential HTS study in which the
screening results for the mosquito AChE1s were compared with previously reported data from
4
8
the human enzyme. We also present the synthesis and biochemical evaluation, including
G122SꢀAgAChE1, of two hits with different selectivity profiles and present structureꢀselectivity
analyses for both compounds based on three dimensional proteinꢀligand complex structures.
RESULTS
Discovery of AgAChE1 and AaAChE1 inhibitors
To identify inhibitors of AgAChE1 and AaAChE1, we adapted the activityꢀbased colorimetric
49
Ellman assay to a format suitable for screening of 17,500 compounds at a single concentration.
The performance of the screening campaign was monitored by repeated runs of a reference plate
containing eight compounds with different inhibition capacities, for example the previously
3
3, 48
characterized compound 1 (Table 1).
The developed assay proved to be robust; it exhibited a
good ability to distinguish between active and inactive compounds, having satisfying average Z’ꢀ
50
factors of 0.65 and 0.73 for AgAChE1 and AaAChE1, respectively (see Supporting Information,
Figures S1ꢀS7, Tables 1ꢀ3). On the basis of the reference plate analyses, the proportions of false
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 55
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
positives for the HTS campaigns were estimated to be 6% and 11% for AgAChE1 and AaAChE1,
respectively; the proportion of false negatives was estimated to be 0.5% for both enzymes.
The AChE1 screens identified 338 compounds that significantly reduced the activity of
AgAChE1, AaAChE1, or both proteins (henceforth, AChE1 is used to refer to either or both of
AgAChE1 and AaAChE1) at a concentration of 50 ꢀM. The identified hits inhibited the enzymes’
ability to hydrolyse the substrate by 33ꢀ98% in the case of AgAChE1 and 31ꢀ96% for AaAChE1
at the tested concentration; the overall hit rates for the two enzymes were 1.3% and 1.6%,
respectively. The inhibition profiles of AgAChE1 and AaAChE1 derived from our screening
studies indicated that the two mosquito enzymes behaved very similar: only 35 of the 338
screening hits exhibited appreciable differences in their capacity to inhibit AgAChE1 and
AaAChE1.
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
We next confirmed the hits’ inhibitory activity by determining their half maximal inhibitory
concentrations (IC ) for AgAChE1 and AaAChE1 (Table 2). IC values based on freshly
5
0
50
prepared stock solutions were determined for 61 of the screened compounds that had different
physicochemical properties and represented three different activity ranges of AChE1 (set A: 70ꢀ
100% inhibition; set B: 31ꢀ69% inhibition; set C: ≤ 30% inhibition; see Supporting Information,
Tables S4 and S5 and Figures S8ꢀS17). In addition, a fourth set of 15 hits was reꢀtested. This last
set was selected from the group of compounds showing different inhibitory activity towards
AgAChE1 and AaAChE1 (set D ꢀD : ≥ 30% difference in measured inhibition). The chemical
1
2
structures of nine of the studied compounds and the corresponding inhibition data are presented
in Table 1 (the complete inhibition results for all the reꢀtested compounds can be found in the
Supporting Information, Table S6). The results of this full doseꢀresponse analysis generally
agreed well with the screening data. A good correlation between the inhibition at a single
ACS Paragon Plus Environment

Page 7 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
concentration and the corresponding IC value was also reported in the previous HTS targeting
5
0
4
8
hAChE.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. Reference compound 1 and selected compounds from sets AꢀD with their HTS
inhibition and doseꢀresponse data.
HTS (%
inhibition)
a
IC (µM)
50
b
C
Set
Structure
Ag
Aa
Ag
Aa
h
S.R
0.030
0.26
0.44
(0.028ꢀ
1
2
R
A
93
92
91
76
(0.13ꢀ0.52)
(0.17ꢀ1.1)
0.034)
0.07
141
0.21
0.22
31
(0.12ꢀ0.37) (0.12ꢀ0.38)
(29ꢀ34)
3
4
A
A
86
94
90
93
>100
.3
>100
>200
5.1
n.a.
8
8.8
(6.8ꢀ10)
(7.1ꢀ11)
(4.4ꢀ6.0) 0.6
5
A
76
82
>1000
>1000
>500
>1000
n.a.
6
7
B
B
23
32
38
33
>500
>500
>200
n.a.
>12
17
12
(12ꢀ22)
(9.6ꢀ14)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 55
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
1
19
43
8
9
C
24
18
(19ꢀ23)
(17ꢀ20)
(41ꢀ46)
2
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
D1
D2
63
41
1
>1000
>1000
2.9
>1000
>200
n.a.
>67
2.9
10
80
(1.6ꢀ5.2)
(2.0ꢀ4.2)
a
IC50 values were determined using 2 replicates. The 95% confidence interval is given in
parentheses. Selectivity ratios were computed by taking the compound’s IC50 value against
hAChE and dividing by the higher of its IC values against AgAChE1 and AaAChE1. n.a. refers
b
5
0
to not applicable.
Of the 76 compounds tested in the follow up assays, more than half inhibited AChE1 with IC50
values ranging from 0.2 to 128 µM (Table 2). The remainder had IC values above 200 µM, or
5
0
showed a poor solubility, preventing a full doseꢀresponse analysis. All but three of the
compounds in set A inhibited AChE1 in a doseꢀdependent manner. This group included the
inhibitors that showed the highest potency, such as compound 2 which had IC values of 0.21
5
0
µM and 0.22 µM for inhibition of AgAChE1 and AaAChE1, respectively (Table 1, Supporting
Information, Table S6). Several of the hits in set B were also confirmed to be inhibitors, with IC50
values between 2.4 and 22 µM. However, the false positive rate for this group appeared to be
higher than that for set A. Four of the compounds from set C proved to be false negatives and
showed doseꢀdependent inhibition of AChE1. These four inhibitors had similar structures to some
compounds from sets A and B but were generally less potent than their analogues, having IC50
values ranging from 11ꢀ128 µM (compared to 0.21ꢀ22 µM for compounds from the first two
sets). The IC data for the compounds in set D showed that none of them were significantly more
5
0
active against one mosquito enzyme than the other. In fact, most of these compounds were false
ACS Paragon Plus Environment

Page 9 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
positives: only two were active inhibitors, and both of them were equally potent against
AgAChE1 and AaAChE1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 2. HTS and doseꢀresponse data for selected compounds from the screening campaigns
targeting AgAChE1 and AaAChE1.
a,b
a,c
a,d
a,e
a,f
Set A
Set B
Set C
Set D1
9
Set D2
No. of tested compounds
24
23
14
6
HTS inhibition AgAChE1 (%)
HTS inhibition AaAChE1 (%)
61 ꢀ 96
72 ꢀ 96
13 ꢀ 66
25 ꢀ 69
ꢀ21 ꢀ 24
ꢀ11 ꢀ 25
37 ꢀ 63
ꢀ5 ꢀ 2
ꢀ40 ꢀ 41
47ꢀ 80
Confirmed inhibitors of AChE1 21
IC AgAChE1 (µM)
0.21 ꢀ 86
13
4
0
ꢀ
2
2 ꢀ 22
2 ꢀ 19
6
11 ꢀ 73
3 ꢀ 7
2 ꢀ 6
2
5
0
IC AaAChE1 (µM)
0.22 ꢀ 66
12 ꢀ 128
ꢀ
5
0
IC AChE1 (> 200µM)
3
0
8
2
9
0
5
0
g
nd
4
2
a
If a hit fell into one activity group for one enzyme and a different activity group for the other,
b
it was assigned to the group corresponding to its highest level of inhibition. Hits showing ≥70%
inhibition in the HTS. Hits showing 31ꢀ69% inhibition in the HTS. Hits showing ≤ 30%
inhibition in the HTS. Hits showing ≥ 30% higher inhibition of AgAChE1 than AaAChE1. Hits
showing ≥ 30% higher inhibition of AaAChE1 than AgAChE1. Could not be determined due to
c
d
e
f
g
poor solubility.
Chemical diversity and selectivity of AChE1 inhibitors
The discovery of potent AChE1 inhibitors prompted us to further analyse the data from the two
4
8
screens against AChE1 and the previously reported HTS targeting hAChE (i.e. differential HTS
analysis). The assay conditions for the three screens were comparable (e.g. plate format, buffer
conditions, substrateꢀ and compound concentrations) and each screening was evaluated based on
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 55
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
its own statistics. Combining the HTS hits for AChE1 and hAChE yielded 425 unique hits. An
investigation of the AChE1ꢀ and hAChE hits revealed that both sets consisted of chemically
diverse compounds that were similar with respect to their overall molecular size, hydrophobicity,
and flexibility: the MW values for the AChE1 and hAChE hits ranged from 199ꢀ629 and 234ꢀ
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
596, respectively; their logP values were between 0.6ꢀ8.7 and ꢀ1.2ꢀ8.1, respectively; and their
numbers of rotatable bonds were between 0ꢀ10 and 0ꢀ12, respectively. Many of the compounds
contained a basic amine capable of forming activated CH⋅⋅⋅arene interactions, which are
5
1
commonly observed in AChEꢀinhibitor complexes. However, the sets of AChE1 and hAChE
hits also included some neutral and anionic inhibitors.
Interestingly, comparisons of the HTS data showed that only 10% of the AChE1 hits were also
scored as hits against hAChE. Mosquito over human selectivity was further investigated by
determining the IC values of the compounds from sets AꢀD for inhibition of hAChE (Table 1,
5
0
Supporting information, Table S6). Indeed, the full doseꢀresponse analysis agreed with the data
from the HTS and many of the compounds were confirmed to be selective for AChE1 over
hAChE; 26 compounds were identified with IC values for AChE1 that were at least a factor ten
5
0
lower than the corresponding value for hAChE. This finding shows that the mosquito and human
enzymes have different preferences for different chemotypes. In addition, this analysis also
demonstrated that it was possible to identify potentially AChE1 selective hits on the basis of their
inhibition (%) in the HTS (Table 1, Supporting Information, Table S6). Hits with more than a
fivefold higher inhibition potency in the HTS also had IC ꢀbased selectivity ratios (S.R.
5
0
IC50
values) of ten or more.
Analysis of the HTS data from the three screens showed that the combined set of AChE1ꢀ and
hAChE hits contained 163 AChE1 selective hits, 74 hAChE selective hits and 37 nonꢀselective
ACS Paragon Plus Environment

Page 11 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
hits (see Experimental Section for selectivity criteria). The remaining hits were classified as
miscellaneous. A multivariate regression discriminant analysis of the selective AChE1ꢀ and
hAChE hits based on selected molecular descriptors indicated chemical differences between the
two groups of compounds. The analysis showed that the AChE1 selective hits were smaller, more
flexible, and had a larger fraction of saturated hydrocarbons than the hAChE selective hits (see
Supporting Information, Tables S7 ꢀS8 and Figure S19).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. Chemical structural tree of the 425 AChE1 and hAChE hits. Three major classes (IꢀIII)
of hits are shown, each against a grey background. Based on the HTSꢀdata, hits showing potential
selectivity for AChE1 are shown as orange dots, nonꢀselective hits as green squares, and hits
showing potential selectivity for hAChE as grey triangles, while nonꢀmarked hits were
miscellaneous. Compounds 1, 2, 4, 7, and 10 are labeled and their symbols are outlined in black.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 55
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
The hits’ chemical diversity and selectivity were investigated further by performing a cluster
analysis of their structural fingerprints and selectivity profiles, yielding the chemical structural
tree shown in Figure 2. The tree revealed three major groups of hit structures (IꢀIII); the first
contained more flexible compounds, the second had high contents of heteroatoms (N, O, and S),
and the last contained hits with higher contents of aromatic and/or rigid substructures. Mapping
the compounds’ selectivity profiles onto the tree revealed clusters of similar molecules that had
distinct effects on the studied enzymes – one type comprising compounds with high potential
selectivity for AChE1, another comprising compounds with high potential selectivity for hAChE,
and a third comprising compounds with mixed selectivity profiles (Figure 2). For example,
compounds 2, 7, and 10 belonged to clusters in which most of the compounds showed potential
selectivity for AChE1 while compounds 1 and 4 belonged to clusters in which the hits were
potentially nonꢀselective or exhibited mixed selectivity. Compounds 2 and 7 had high selectivity
ratios (S.R.IC50 > 12) indicating preferential inhibition of AChE1, while compound 1, which
shares a number of structural features with 2 and 7, was potent against all three enzymes. To
better understand the selectivity properties of AChE1 and hAChE, we decided to further
investigate the highly potent and AChE1 selective compound 2, and the chemically similar but
nonꢀselective 1.
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
Synthesis and biochemical evaluation of 2 and 1
Compound 2 was prepared in two steps as outlined in Scheme 1 by performing a Williamson
ether synthesis with 4ꢀphenylphenol (11) and chloroacetic acid to give 12 followed by an amide
coupling with 1ꢀethylꢀ4ꢀpipiridinꢀ4ꢀyl piperazine, using TBTU as the coupling reagent. The
synthesis of 1 was accomplished by nucleophilic substitution of 4ꢀnitrobenzyl bromide (13) with
piperidine followed by reduction to yield the corresponding aniline 15. Coupling of this
ACS Paragon Plus Environment

Page 13 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
intermediate with chloroacetic acid chloride provided the desired amide 16, allowing the final
compound 1 to be obtained by ether formation with 2,4ꢀdichlorophenol (Scheme 2).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a
Scheme 1
a
Reagents and conditions: (a) chloroacetic acid (5 eq), NaOH (s) (7 eq), MeOH, reflux 24h,
yield 29% (b) 1ꢀethylꢀ4ꢀpiperidinꢀ4ꢀylꢀpiperazine (1.2 eq), TBTU (1.2 eq), TEA (3 eq), DMF, rt
5 days (c) Half sat. HCl in CH Cl , 0 °C to rt, 20h, yield 47% over two steps.
2
2
a
Scheme 2
a
Reagents and conditions: (a) piperidine (3.1 eq), THF, rt, 1h, yield 93% (b) SnCl *2 H O (5
eq), EtOAc, rt, overnight, yield 92% (c) chloroacetyl chloride (20 eq), AcOH, NaOAc, 3h, rt,
yield 92% (d) 2, 4ꢀdichlorophenol (1.5 eq), K CO (2 eq), KI (0.05 eq) DMF, rt, 16 h, yield 59%.
2
2
2
3
The potencies of the synthesized 2 and 1 were investigated by determining their IC values for
50
inhibition of AgAChE1, AaAChE1, hAChE, as well as mAChE and the resistance conferring
mutant G122SꢀAgAChE1 (Figure 3). Compound 2 was confirmed to be selective for the
mosquito enzymes, having S.R.IC50 values of 141 and 82 for AChE1 over hAChE and mAChE,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 55
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
respectively. The IC50 values of compound 2 was 0.21 µM and 0.22 µM for inhibition of
AaAChE1 and AgAChE, respectively. The corresponding constants for the vertebrate enzymes
were 31 µM (hAChE) and 18 µM (mAChE). The IC value of 2 for inhibition of the resistance
5
0
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
conferring mutant G122SꢀAgAChE1 was 1.3 µM, corresponding to a more than ten times
selectivity over the vertebrate enzymes (Figure 3A). The nonꢀselective binding of compound 1 to
AChE1 was confirmed (Figure 3B). In fact, compound 1 had a slightly higher potency for the
vertebrate enzymes (0.030 µM and 0.026 µM for hAChE and mAChE, respectively) than the
mosquito enzymes (0.26 µM and 0.44 µM for AgAChE1 and AaAChE1, respectively). Previous
measurements of the vertebrate enzymes resulted in IC values for 1 that were approximately
5
0
3
3, 48
seven to twelve times higher than reported here.
The experimental conditions in this work
(e.g. pH, compoundꢀ and enzyme batch) were slightly different from those previously used.
Importantly, all IC50 determinations included in the present study were determined using the
same conditions and are thus directly comparable. The nonꢀselective 1 had an IC value of 1.2
50
3
3
µM for inhibition of G122SꢀAgAChE1, which is in agreement with previous determinations.
Figure 3. Doseꢀresponse curves for compounds 2 (A) and 1 (B). The plotted results are means ±
standard deviations based on 2ꢀ3 measurements. IC ꢀvalues against Agꢀ, AaAChE1, and hAChE
5
0
ACS Paragon Plus Environment

Page 15 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
are presented in Table 2; for compounds 2 and 1 the IC ꢀvalues for mAChE were 18 (8.7ꢀ38) µM
5
0
and 0.026 (0.018ꢀ0.039), respectively, and for G122SꢀAgAChE1 1.3 (0.98ꢀ1.7) µM and 1.2 (0.77ꢀ
1.9) µM, respectively.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Structure – selectivity relationships of 2 and 1
To investigate the structural basis for the marked differences in selectivity of 2 and 1 we
studied their binding modes using a combination of Xꢀray crystallography and homology
modeling. The 2.5 Å resolution crystal structure of 2 in complex with mAChE (2•mAChE, pdb
code: 5FUM, see Supporting Information, Table S10, authors will release the atomic coordinates
and structure factors upon article publication) illustrates the general, extended binding mode of
the ligands (Figure 4). The compounds span the entire 20 Å deep active site gorge of AChE and
form interactions with both Trp286 and Trp86, residues located at the rim and the base of the
gorge, respectively. The binding poses in 2•mAChE and the previously reported 1•mAChE (pdb
5
1
code: 5FOQ ) also involves a similar set of contacts and interactions; arene···arene interactions
with indole of Trp286 are formed by the 2,4ꢀdichlorophenyl or biphenyl arenes while their
tertiary amines are located at the base of the gorge. Ligands featuring a positively charged moiety
interacting with a residue deep in the catalytic site of AChE has been observed in the crystal
51ꢀ54
structures of potent inhibitors.
For example, the piperidine moiety of 1 has been shown to
5
1
form a strong activated CH···arene hydrogen bond with Trp86. A superposition of 2•mAChE
and 1•mAChE reveals that the piperidine fragment in 1 is located approximately 2.5 Å deeper
down in the gorge than the ethylpiperazine of 2 (Figure 5AꢀB). This will influence the
CH···arene hydrogen bond and may contribute to the lower affinity of 2 than 1. Furthermore, it
appears that steric effects could also contribute to the reduced potency of 2, since the bulky
piperidine linker is positioned in the narrow bottleneck formed by Tyr124 and Tyr337.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 55
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. Structures of enzymeꢀinhibitor complexes. (A) Crystal structure of 2•mAChE (pdb
code: 5FUM) with compound 2 in pink and its electron density shown in blue (simulated
annealing F ꢀF omit map at 3 σ). (B) Homology model of AgAChE1 shown in cyan
o
c
superimposed to 1•mAChE (pdb code: 5FOQ) in white. The two loops at the entrance of the
gorge responsible for the main differences between AgAChE1 and mAChE are highlighted in
magenta.
To investigate the binding of 2 and 1 to the mosquito enzymes, the corresponding structures
were generated by homology modelling. The models of 1•AgAChE1 and 1•AaAChE1 were
constructed using a crystal structure of Drosophila melanogaster AChE (DmAChE) as the main
template (38% and 39% sequence identity for AgAChE1 and AaAChE1, respectively). Note that
no AChE1 crystal structure was available and that DmAChE is encoded by the ace-2 gene. The
sequence identity of AgAChE1 and AaAChE1 was 93%, and the RMSD value based on all heavy
atoms in the two homology models was low (1.2 Å). In general, the structures of mAChE,
ACS Paragon Plus Environment

Page 17 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
AgAChE1, and AaAChE1 are very similar (Figure 4B); the structure of the active site is
conserved and most of the amino acids in the binding site are identical. As previously reported,
the most apparent differences in sequences between mosquito and vertebrate enzymes’ occur in
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
30, 36
two loops located at the entrance of the gorge (Figure 4B).
The first loop (loop 1), involving
residues that define the acyl pocket, is adjacent to Trp286 and extends from Leu289 to Phe297 in
the mouse and human enzymes; in the mosquito enzymes this loop is truncated by three amino
acids. In addition, the residue Phe295, which in vertebrates has frequently been observed to
5
1ꢀ52, 54ꢀ56
interact with ligands,
is replaced by a cysteine in the mosquito enzymes. The presence of
a free cysteine in the active site gorge has previously been used to develop selective covalent
3
7ꢀ40
AChE1 inhibitors.
The second loop (loop 2) extends from Gly342 to Lys348 in
mAChE/hAChE; in the mosquito enzymes, loop 2 contains one additional amino acid and the
proline that directs the loop’s structure in mAChE/hAChE (Pro344) is missing in AChE1. While
this loop may not be directly involved in ligand binding, it could affect the position of the alpha
helix that includes residues such as Tyr337, Phe338, and Tyr341, all of which line the active site
gorge.
The homology models showed that most of the interactions between 1 and
AgAChE1/AaAChE1 were similar to those found in 1•mAChE; the key interactions to Trp86 and
Trp286 were preserved and no potentially new interactions between 1 and AChE1 were identified
(see Supporting Information, Figure S22 and S23). To investigate the higher potency of 2 for
AgAChE1 and AaAChE1 compared to mAChE, we rigidly modelled 2 into the homology models
based on the binding pose of 1 in 1•mAChE (see Experimental section for details). The modelled
conformation of 2 was readily accommodated in the binding sites of AgAChE1 and AaAChE1 (2
in complex with AgAChE1 and AaAChE are shown in Figure 5C and Supporting Information,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure S24A, respectively), showing that 2 can adopt a slightly different binding pose in the
mosquito enzymes compared to the one observed in 2•mAChE. By binding further down in the
active site gorge, the piperazine moiety of 2 can thus form CH···arene hydrogen bonds with the
indole of Trp286, in contrast to the suboptimal distances and geometries seen in 2•mAChE. In
addition, the bulky piperidine linker of 2 is well accommodated in the bottleneck region of the
mosquito enzymes.
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 homology model of AgAChE1 including the mutation of Gly122 to serine showed that the
mutation appears not to significantly affect the binding modes of compounds 1 and 2, which is in
agreement with the observed minor reduction in potency. The G122S mutation is located in the
3
5
oxyanion hole near the catalytic Ser203 and is thereby not making contact with either of the
ligands; the closest distance is approximately 5 Å between 2 and the OH of the mutated serine
(see Supporting Information, Figure S24B).
Figure 5. Binding poses of inhibitors in the active site gorge of mAChE (AꢀB; pdb codes:
5FOQ and 5FUM) and AgAChE1 (C, homology model). The structures of mAChE and AgAChE1
are shown in white and cyan, respectively; compounds 1 and 2 are shown in grey and pink,
respectively.
ACS Paragon Plus Environment

Page 19 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
DISCUSSION
We have experimentally screened a library of 17,500 compounds for their inhibition of AChE1
from the mosquitoes Anopheles gambiae and Aedes aegypti, and the resulting hits were analyzed
together with corresponding data from the previously presented HTS of the human enzyme. The
screening allowed us to explore a larger chemical space of potential inhibitors compared to
efforts aiming for covalent inhibitors. Covalent inhibitors target a specific amino acid in AChE1,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
36ꢀ40
20, 30, 41ꢀ46
for example Cys295
or Ser203.
In contrast, nonꢀcovalent inhibitors could in
principle bind to any region of the protein as long as the catalytic activity is inhibited. The HTS
resulted in 338 inhibitors of AChE1 that had different chemical topologies and physicochemical
properties suggesting that they exploit different interaction patterns and binding poses with
AChE1 compared to known covalent inhibitors, including currently used insecticides.
The AChE1ꢀ and hAChE hits showed similar global chemical features (e.g., molecular
weight and logP), reflecting the conserved structure and function of AChEs from different
species. Nevertheless, multivariate modelling based on selective hits for AChE1 and hAChE,
respectively, indicated chemical differences between the two groups of compounds. This finding
has important implications for the development of selective nonꢀcovalent inhibitors as it indicates
that, in addition to the overlapping molecular properties of the two groups of hits, there also exist
distinct chemical spaces that can be explored in an optimization process. From a structural point
of view, the homology modeling of mosquito AChE1 suggested that the observed selectivity for
the mosquito enzymes is related to structural differences of two loops at the entrance of the active
site gorge. Interactions with residues in these loops have previously been proposed to confer
30
hAChE over AgAChE1 selectivity for nonꢀcovalent inhibitors. Interestingly, the mosquito vs.
vertebrate differences in the loops at the entrance of the gorge appear to influence the structure of
the entire active site. This may explain the large number of selective hits despite the fact that
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 55
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
AChE1, hAChE, and mAChE have identical residues in the actual catalytic sites. In fact, the
combination of Xꢀray crystallography and homology modeling of 2, suggested that the preference
for AChE1, at least partly, was due to specific interactions with Trp86, a residue more than 15 Å
away from the loop regions at the entrance of the gorge. The hypothesis of differences in the
catalytic sites of AChE1 and hAChE are further supported by previous work on covalent
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
0, 41
inhibitors targeting Ser203 where bulky inhibitors led to selectivity for AgAChE1
and our
biochemical characterization of AChEs, showing that AChE1 has a higher tolerance for larger
3
3
substrates compared to vertebrates enzymes.
The resistance conferring mutation in G122SꢀAgAChE1, located in the oxyanion hole of
3
5
AgAChE1, results in a profound effect on the enzyme’s ligand binding properties. So far it has
proven challenging to develop covalent AChE1 inhibitors that are selective and active on the
42ꢀ45
G122S enzyme.
Nonꢀcovalent inhibitors targeting G122SꢀAgAChE1 have been reported,
4
7
however the mosquito over human selectivity of the compounds was not presented in the study.
Encouragingly, both compounds 1 and 2 inhibited G122SꢀAgAChE1. Thus, 2 combines a
selectivity for the mosquito enzymes with a potency for G122SꢀAgAChE1. This demonstrates a
possibility to develop selective nonꢀcovalent inhibitors that target both the wildꢀtype and the
mutated form of AChE1, although further investigations are needed to determine the general
applicability.
Our long term goal is to develop new insecticides for vector control that selectively target
mosquitoes and overcome problems arising from insecticide resistance. The differential HTS
generated a substantial number of hits that provide valuable chemical starting points for the
insecticide development process. However, in order to be effective in vivo the inhibitors must not
only target a relevant protein with sufficient potency, they also have to be able to penetrate the
exoskeleton and survive the metabolism of mosquitoes. New AChE1 inhibitors have shown
ACS Paragon Plus Environment

Page 21 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
41ꢀ43, 45, 47
insecticidal activity on mosquitoes,
although a generally much lower in vivo compared
to in vitro potency of these inhibitors highlights the importance of the pharmacokinetic as well as
the pharmacodynamic properties. Furthermore, factors such as human safety, cost of goods, and
environmental sustainability also need to be adequately addressed. During such a multiobjective
optimization process (cf. drug discovery process) many compounds will likely fail, making
multiple compound classes highly desirable in order to increase the prospect of success.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CONCLUSIONS
To discover new inhibitors of AgAChE1 and AaAChE1, we successfully adapted the Ellman
assay to a format where we could screen 17,500 compounds. The assay proved to be robust and
338 hits of AgAChE1 and AaAChE1 were identified. The doseꢀresponse analysis generally
agreed well with the screening data, confirming the quality of the screen. Importantly, several
inhibitors were identified with subꢀmicromolar to low micromolar potency for the mosquito
enzymes.
Considering the evolutionary conserved structure and function of AChE, a surprisingly large
fraction of the hits was selective for AChE1 over the human enzyme. Chemical differences could
be identified between the two sets of hits, although their overall global properties were similar.
The chemical cluster tree revealed that structurally related hits were prone to have similar
selectivity profiles, although inhibitors with subtle differences in their chemical structures could
still have different selectivity. The subꢀmicromolar inhibitors 2 and 1 were synthesized and their
different selectivity profiles were confirmed. They were also shown to be potent inhibitors of the
resistance conferring mutant G122SꢀAgAChE1. The structureꢀselectivity relationship of 2 and 1
for the mosquito and mouse enzymes could be established by Xꢀray crystallography and
homology modelling. Two loops at the entrance of the active site gorge that differ between the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 55
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
enzymes appear to cause differences in the entire active site explaining the selectivity profiles of
2
and 1. The structural analysis can be extended to the complete set of hits as the proposed
differences in the active site gorge may explain the large number of selective inhibitors.
Multiple compound classes will be needed in the development process of new insecticides for
vector control to combat mosquitoꢀborne diseases. Nonꢀcovalent inhibitors of AChE1, such as the
ones presented here, provide valuable starting points, and are complementary to existing and new
covalent inhibitors since they are chemically different and have different interaction
characteristics with the enzymes.
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
EXPERIMENTAL SECTION
High throughput screening
The collection of 17,500 compounds used in this work is maintained as part of the screening
5
7
platform of the Laboratories for Chemical Biology, Umeå (LCBU). Recombinant AChE1
enzymes from the mosquitoes Anopheles gambiae and Aegypti aegypti were expressed in insect
33
Sf-9 cells using a baculovirus expression system, as described previously. The screen was
performed using secreted nonꢀpurified proteins in growth medium, and enzymatic activity was
4
9
measured using the Ellman assay adapted to a 96ꢀwell format. Assays were performed in a final
volume of 200 µL of 0.1 M phosphate buffer (pH 7.4) containing 0.2 mM 5,5′ꢀdithiobis(2ꢀ
nitrobenzoic acid) and 1 mM acetylthiocholine iodide. Stock solutions of the compounds in
P
DMSO (5 mM) were transferred to the assay plates using a Biomek NX Laboratory Automation
Workstation (Beckman Coulter, Inc.) to give a final compound concentration of 50 µM. Each
plate contained 80 compounds in DMSO, eight positive controls (without compound) and eight
negative controls (without compound and substrate). The enzymatic reaction was measured by
monitoring changes in the absorbance of individual wells at 412 nM over 60 s in an Infinite
ACS Paragon Plus Environment

Page 23 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
M200 Microplate reader (Tecan Group Ltd., Männedorf, Switzerland). The average slope
(
typically 0.40±0.03 and 0.53±0.03 dA/min for AgAChE1 and AaAChE1, respectively)
determined for the eight positive controls on each plate was taken to represent 100% activity and
the activity observed in the sample wells were quantified in relation to this value. Compounds
that reduced the measured activity by at least three times the standard deviation of the mean for
the 17,500 compounds were scored as hits.
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
Assay stability and robustness
th
A reference plate was analysed after every 10ꢀ20 plate of samples to monitor the stability of
the screening assay; in total, 27 reference plates were run while screening with AgAChE1 and 12
with AaAChE1. Each plate contained eight compounds, with eight evenly spaced replicates of
each compound. Five of the compounds had IC values ranging from 0.05 to 100 µM for both
5
0
enzymes; the other three compounds were inactive up to 1000 µM and used as positive controls
5
0
(
see Supporting Information, Figure S4 and Table S2). The Z’ꢀfactor was calculated for each
reference plate using the following expression: Z’=1 – ((3SDc+ + 3SD )/|µ ꢀ µ |). In these
cꢀ
c+
cꢀ
calculations, data for eserine and AL045 were used as negative and positive controls, respectively
(see Supporting Information, Figures S4ꢀS7 and Table S2).
Selection of hits for IC determinations
50
The hits were divided into groups based on their activity in the HTS (if a hit was classified into
different groups for the two AChE1 enzymes, it was assigned to a group based on its highest
observed inhibitory potency). The first group consisted of hits that reduced the enzymatic activity
of AChE1 by ≥ 70%; a representative subset of compounds with diverse structural and
physicochemical features was manually selected from this group and designated set A. Similarly,
another set of compounds was selected from the group of compounds that achieved 31ꢀ69%
inhibition in the HTS (set B), and a third set (set C) was selected from the group of compounds
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 55
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
that were not scored as hits but had similar structural and physicochemical properties to the
compounds in sets A and B. Finally, a fourth set of compounds were manually selected from the
hits whose inhibitory potency towards Agꢀ and AaAChE1 differed by ≥ 30%. For more details,
see the Supporting Information, Tables S4 and S5 and Figures S8ꢀS17.
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
IC determinations
50
4
9
Doseꢀresponse experiments based on the Ellman assay were performed on hit compounds that
were selected for follow up evaluation. The enzymes used in these experiments were the
3
3
58
59
recombinant variants of AChE1, hAChE, and mAChE. The compounds were purchased
5
7
from ChemꢀBridge (San Diego, CA) or accessed via the LCBU, except 2 and 1 which were
1
synthesized (see Synthesis section below). The compounds’ structures were confirmed by H
1
NMR and H COSY, and their purity was confirmed to be ≥ 95% by LCꢀMS (detection at 254
nm; see the Synthesis section below and Supporting Information, pages S32ꢀS36). Stock
solutions of the compounds were prepared from solid material in DMSO at a concentration of
100 mM and working dilutions thereof were prepared in either 0.1 M sodium phosphate buffer
(pH 7.4) or MilliQ water. Compound solutions of at least eight different concentrations up to a
maximum of 1 mM were used. The assay was performed under the same conditions as described
above for the HTS, with the difference that it was handled manually, only eight wells were
monitored at a time, and the assay was run at 30 °C. The average enzymatic activity determined
for the eight wells of unꢀinhibited enzyme on the plate was defined as 100% activity and the
activity observed in the inhibited wells was quantified in relation to this value. IC values were
5
0
6
0
calculated using nonꢀlinear regression (curve fitting) in GraphPad Prism and the log [inhibitor]
vs. response variable slope equation was fitted using four parameters. In certain cases, threeꢀ
parameter analyses were used instead. The IC values presented in Table 1 were determined
5
0
independently twice using 1ꢀ2 replicates on each occasion for all compounds other than 9, which
ACS Paragon Plus Environment

Page 25 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
exhibited poor solubility. The complete HTS and follow up data of 1ꢀ10 and the remaining
compounds in set AꢀD, whose activity was assessed once using 1ꢀ2 replicates, is presented in
Supporting Information, Table S6.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Selectivity profiles
Based on the observations from the doseꢀresponse evaluations, the following criteria were
established in order to delineate selectivity profiles for the hits based on their HTS data.
Potentially AChE1 selective hits were required to i) be identified as hits against at least one
mosquito enzyme, ii) not be identified as a hit against hAChE, and iii) inhibit both AgAChE1 and
AaAChE1 at least five times more effectively than hAChE in the HTSs. Potentially hAChE
selective compounds were identified as hits that i) were identified as hAChE hits, ii) were not
identified as AgAChE1 or AaAChE1 hits, and iii) inhibited hAChE at least five times more
effectively than both AgAChE1 and AaAChE1 in the HTSs. Nonꢀselective hits were required to
be identified as hits of AChE1 and hAChE. All remaining hits were classified as miscellaneous.
Discriminant analysis of AChE1ꢀ and hAChE selective hits
Prior to calculation of physicochemical 2Dꢀdescriptors and MACCS fingerprints, the neutral
chemical structures of the hit compounds were prepared by removing small fragments (i.e. metals
or groups in salts), adding explicit hydrogens, and scaling of bonds using the ‘wash’ option in
61
MOE. A regression discriminant analysis was made based on all the AChE1 selective
compounds that were scored as hits against both AgAChE1 and AaAChE1 (83 hits) and the
hAChE selective compounds (74 hits) according to the criteria in the section above. The
physicochemical properties of the hits were described by 85 2Dꢀdescriptors that were calculated
6
1
62
using the MOE software. Orthogonal partial least squaresꢀdiscriminant analysis (OPLSꢀDA),
a supervised multivariate classification method, was used to identify physicochemical properties
that could discriminate between the AChE1ꢀ and hAChE selective hits. The data were meanꢀ
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 55
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
centered and scaled to unit variance before an OPLSꢀDA model was calculated using the
6
3
SIMCAꢀP+ software. The model was iteratively refined by excluding 2D descriptors with small
influence on the model (i.e. weighted sum of squares of the w* weights less than 0.5). The
number of significant model components was determined by crossꢀvalidation. The results are
presented in the Supporting Information, Tables S7 and S8, and Figure S18. The significance of
model variables (2D descriptors) for discriminating between AChE1ꢀ and hAChE hits was
confirmed by Student’s Tꢀtest (α = 0.05) (see Supporting Information, Table S8 and Figure S18).
Chemical structural tree
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 chemical structures of the AChE1ꢀ and hAChE hits were described in terms of 166
6
4
61
65
MACCS fingerprints using the MOE software. A matrix of Soergel distances between the
hits was calculated, where the Soergel distance between two hits A and B was calculated as DAB
=
(a+b ꢀ 2*c) / (a + b ꢀc). Here, a is the number of bits set to “1” in the MACCS fingerprint
vector of hit A, b is the number of bits set to “1” in the fingerprint vector of hit B, and c is the
number of bits set to “1” in the fingerprint vectors of both hit A and hit B. The hits were clustered
66
on the basis of their Soergel distances using the NeighbourꢀJoining method in the PHYLIP
6
7
software after randomizing their order. The clustering was visualized in the form of an unrooted
tree with the angle of the arc set to 360 degrees; the tree’s branch lengths correspond to the
Soergel distances between the hits. A short distance between two hits means that they are
chemically similar. The selectivity profile of each hit (which could be potentially AChE1
selective, nonꢀselective, or potentially hAChE selective) was mapped onto the tree.
Synthesis
General. All reactions were carried out under inert atmosphere (N ) unless otherwise stated.
2
THF and DMF were dried in a solvent drying system and freshly collected prior to reaction (THF
was passed through neutral alumina, DMF was passed thorough activated molecular sieves
ACS Paragon Plus Environment

Page 27 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
followed by an isocyanate scrubber). All microwave reactions were carried out in a monomode
reactor using Smith process vials sealed with a Teflon septum and an aluminum crimp top. The
temperature was measured with an IR sensor and reaction times refer to the irradiation time at the
target temperature. Reactions were monitored using TLC (silica gel matrix, layer thickness 200
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
µm, particle size 25 µm) with UVꢀdetection (254 nm) or developed using KMnO solution. Flash
4
column chromatography (eluents given in brackets) was performed on normal phase silica gel
1
13
(
Merck, 60 Å, 40ꢀ63 µm). H and C NMR spectra were recorded on a Bruker DRXꢀ400
instrument at 298 K in CDCl using residual CHCl (δ = 7.26 ppm) or CDCl (δ = 77.0 ppm) as
3
3
H
3
C
an internal standard, (CD ) SO using residual (CD )(CD H)SO (δ = 2.50 ppm) or (CD ) SO (δ
C
3
2
3
2
H
3 2
=
40.0 ppm) as an internal standard, or CD OD using residual CD HOD (δ = 3.31 ppm) or
3 2 H
CD OD (δ = 49.0 ppm) as an internal standard, or D O using traces of CH OH (δ = 3.34 ppm)
3
C
2
3
H
or CH OH (δ = 49.50 ppm) as an internal standard (Supporting Information, pages S32ꢀS42).
3
C
LCꢀMS analyses were performed on a Waters LC system using a Xterra MS C18 18.5 µm 4.6x50
mm column and an acetonitrile:water eluent system containing 0.2% formic acid. Eluting
compounds were detected by monitoring the eluent’s absorption at 254 nm and mass
spectrometry was performed in positive ion mode using a Waters micromass ZG 2000
electrospray instrument. Analytical HPLC was performed using a Nexera UHPLC system
(Shimadzu, US) connected to a diode array detector (SPP M20A). Samples were analyzed using a
Nucleodur C18 HTec column (EC 150 × 4.6, 5 ꢁm, MachereyꢀNagel) with a flow rate of 1
ml/min. Aliquots of 2 ꢁl of each sample were injected, and detection was performed at 254 nm.
The mobile phase was composed of solvent A (H O with 0.1% TFA) and solvent B (acetonitrile
2
with 0.1% TFA). The binary gradient profile was as follows with solvent B as the reference: 0–
1
.5 min, 10% B; 1.5–11.5 min, 10–100% B; 11.5–26.5 min, 100% B; 26.5–27 min, 100–10% B;
7–30.5 min, 10% B. Before first injection, the column was equilibrated at 10% B for 1.5 min.
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 55
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
The synthesized compounds used in the biological evaluations exhibited ≥ 95% purity by the
HPLC analysis.
2ꢀ(2,4ꢀDichlorophenoxy)ꢀNꢀ[4ꢀ(piperidinꢀ1ꢀylmethyl)phenyl]acetamide (1). K CO (142
2
3
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
mg, 1.03 mmol) and KI (4 mg, 0.03 mmol) were added to 16 (137 mg, 0.51 mmol) and 2,4ꢀ
dichlorophenol (126 mg, 0.77 mmol) in dry DMF (4 ml) at rt while stirring. Stirring was
continued at rt for 16 h before the reaction mixture was poured into EtOAc and washed three
times with NaHCO (sat), followed by water and brine. The organic layer was dried over Na SO ,
3
2
4
filtered, and concentrated under reduced pressure to give an offꢀwhite solid. Recrystallization
1
from EtOH:heptane gave 1 (120 mg, 59% yield) as white needles. H NMR (400 MHz, CDCl ) δ
3
8.52 (br s, 1H), 7.52 (d, J = 8.5 Hz, 2H), 7.41 (d, J = 2.5 Hz, 1H), 7.29 (d, J = 8.5 Hz, 2H), 7.22
(dd, J = 2.5 Hz, J = 8.8 Hz, 1H), 6.86 (d, J = 8.8 Hz, 1H), 4.58 (s, 2H), 3.43 (s, 2H), 2.38ꢀ2.30
1
2
13
(
m, 4H), 1.58ꢀ1.52 (m, 4H), 1.44ꢀ1.38 (m, 2H). C NMR (100 MHz, CDCl ) δ 164.8, 151.3,
3
1
35.4, 135.4, 130.1, 129.8 (2C), 128.0, 127.6, 123.7, 119.6 (2C). 114.8, 68.3, 63.2, 54.3 (2C),
25.9 (2C), 24.3.
ꢀ([1,1'ꢀBiphenyl]ꢀ4ꢀyloxy)ꢀ1ꢀ[4ꢀ(4ꢀethylꢀ1ꢀpiperazinyl)ꢀ1ꢀpiperidinyl]ꢀethanone
hydrochloride (2). TBTU (42 mg, 0.13 mmol) and 1ꢀethylꢀ4ꢀpiperidinꢀ4ꢀylꢀpiperazine (26 mg,
.13 mmol) were added to 12 (25 mg, 0.11 mmol) in dry DMF (1 ml), at rt while stirring. TEA
46 µl, 0.33 mmol) was added dropwise followed by an additional 1 ml of dry DMF. The cloudy
2
0
(
reaction mixture was stirred at rt for five days. Upon addition of NaHCO (sat.) to the reaction
3
mixture a white precipitate was formed that dissolved upon addition of H O. The resulting
2
aqueous layer was extracted three times with EtOAc, after which the combined organic layers
were washed three times with brine, dried over Na SO , filtered, and concentrated. The resulting
2
4
crude material was cooled on ice, treated with halfꢀsaturated HCl in CH Cl (5 ml) and stirred for
2
2
20 h while slowly being allowed to attain rt. The solvent was then removed under reduced
ACS Paragon Plus Environment

Page 29 of 55
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pressure and the resulting crude product was recrystallized from MeOH:EtOAc to give a white
solid that was filtered off and washed with a small amount of EtOAc. The mother liquor was
concentrated and the resulting solid triturated with EtOAc before the recrystallization was
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
repeated once to give 24 mg (47% yield) of 2 as a white solid. H NMR (400 MHz, D O) δ 7.62ꢀ
2
7
.58 (m, 4H), 7.48ꢀ7.44 (m, 2H), 7.38ꢀ7.34 (m, 1H), 6.99 (d, J = 8.7 Hz, 2H), 4.82 (d, J = 15.1
Hz, 1H), 4.71 (d, J = 15.1 Hz, 1H), 4.51 (d, J = 13.6 Hz, 1H), 3.95 (d, J = 13.6 Hz, 1H), 3.75ꢀ
.44 (m, 9H), 3.32 (q, J = 7.3 Hz, J = 14.7 Hz, 2H), 3.12 (t, J = 12.8 Hz, 1H), 2.70 (t, J = 12.8
3
1
2
1
3
Hz, 1H), 2.19 (m, d, J = 11.5 Hz, 2H), 1.69ꢀ1.49 (m, 2H), 1.34 (t, J = 7.3 Hz, 3H). C NMR (100
MHz, D O) δ 168.9, 157.6, 140.3, 134.6, 129.8 (2C), 128.7 (2C), 128.0, 127.1 (2C), 115.8 (2C),
2
6
6.1, 63.5, 53.0, 49.2 (2C), 46.6 (2C), 43.5, 41.2, 27.1, 26.5, 9.2.
(Biphenylꢀ4ꢀyloxy)ꢀacetic acid (12). Chloroacetic acid (278 mg, 2.94 mmol) followed by
NaOH (s) (165 mg, 4.11mmol) were added to 4ꢀphenylphenol 11 (100 mg, 1.76 mmol) in MeOH
6 ml) at rt while stirring. The reaction mixture was then heated and stirred at reflux for 24 h,
(
allowed to cool to rt, and acidified by adding 1M HCl. The resulting solution was concentrated
under reduced pressure to remove MeOH and the formed white precipitate was filtered off and
1
washed with small amounts of water to yield 12 (39 mg, 29% yield) as a white solid. H NMR
(
400 MHz, CD OD) δ 7.56 (m, 4H), 7.41ꢀ7.37 (m, 2H), 7.27 (t, J = 7.4 Hz, 1H), 7.01 (d, J = 8.7
3
13
Hz, 2H), 4.69 (s, 2H). C NMR (100 MHz, CD OD) δ 172.7, 159.0, 142.0, 135.8, 129.8 (2C),
3
129.0 (2C), 127.8, 127.6 (2C), 116.0 (2C), 65.9.
1
ꢀ(4ꢀNitrobenzyl)piperidine (14). Piperidine (750 µl, 7.58 mmol) followed by dry THF (3 ml)
were added to 4ꢀnitrobenzyl bromide 13 (523 mg, 2.42 mmol) in dry THF (3 ml) at rt while
stirring. A precipitate was immediately formed and the reaction mixture was allowed to stir at rt
for 1 h before being diluted with Et O. The resulting mixture was washed three times with 1 M
2
NaOH and then extracted three times with 1 M HCl. The combined aqueous layers were made
ACS Paragon Plus Environment