Phenylthiomethyl Ketone-Based Fragments Show Selective and Irreversible Inhibition of Enteroviral 3C Proteases

Article abstract of DOI:10.1021/acs.jmedchem.7b01440

Lead structure discovery mainly focuses on the identification of noncovalently binding ligands. Covalent linkage, however, is an essential binding mechanism for a multitude of successfully marketed drugs, although discovered by serendipity in most cases. We present a concept for the design of fragments covalently binding to proteases. Covalent linkage enables fragment binding unrelated to affinity to shallow protein binding sites and at the same time allows differentiated targeted hit verification and binding location verification through mass spectrometry. We describe a systematic and rational computational approach for the identification of covalently binding fragments from compound collections inhibiting enteroviral 3C protease, a target with high therapeutic potential. By implementing reactive groups potentially forming covalent bonds as a chemical feature in our 3D pharmacophore methodology, covalent binders were discovered by high-throughput virtual screening. We present careful experimental validation of the virtual hits using enzymatic assays and mass spectrometry unraveling a novel, previously unknown irreversible inhibition of the 3C protease by phenylthiomethyl ketone-based fragments. Subsequent synthetic optimization through fragment growing and reactivity analysis against catalytic and noncatalytic cysteines revealed specific irreversible 3C protease inhibition.

Full text of DOI:10.1021/acs.jmedchem.7b01440

Subscriber access provided by Gothenburg University Library
Article
Phenylthiomethyl ketone-based fragments show selective
and irreversible inhibition of enteroviral 3C proteases
Robert Schulz, Amira Atef, Daniel Becker, Franziska Gottschalk, Carolin Tauber, Stefan Wagner,
Christoph Arkona, Atef A Abdel-Hafez, Hassan H Farag, Jörg Rademann, and Gerhard Wolber
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01440 • Publication Date (Web): 12 Jan 2018
Downloaded from http://pubs.acs.org on January 12, 2018
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 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Phenylthiomethyl ketoneꢀbased fragments show
selective and irreversible inhibition of enteroviral 3C
proteases
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
Robert Schulz, † Amira Atef,‡ Daniel Becker,† Franziska Gottschalk,† Carolin Tauber,† Stefan
Wagner,† Christoph Arkona,† Atef A. AbdelꢀHafez,‡ Hassan H. Farag,‡ Jörg Rademann,†
Gerhard Wolber*†
† Pharmaceutical and Medicinal Chemistry, Institute of Pharmacy, Freie Universität Berlin,
KöniginꢀLuise Straße 2+4, 14195
‡ Medicinal Chemistry Department, Faculty of Pharmacy, Assiut University, Assiut 71526,
Egypt
Abstract
Lead structure discovery mainly focuses on the identification of noncovalently binding ligands.
Covalent linkage, however, is an essential binding mechanism for a multitude of successfully
marketed drugs although discovered by serendipity in most cases. We present a concept for the
design of fragments covalently binding to proteases. Covalent linkage enables fragment binding
unrelated to affinity to shallow protein binding sites and at the same time allows differentiated
targeted hit verification and binding location verification through mass spectrometry. We
describe a systematic and rational computational approach for the identification of covalently
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 50
1
2
3
4
5
6
7
8
9
binding fragments from compound collections inhibiting enteroviral 3C protease, a target with
high therapeutic potential. By implementing reactive groups potentially forming covalent bonds
as chemical feature in our 3D pharmacophore methodology, covalent binders were discovered by
highꢀthroughput virtual screening. We present careful experimental validation of the virtual hits
using enzymatic assays and mass spectrometry unraveling a novel, previously unknown
irreversible inhibition of the 3C protease by phenylthiomethyl ketoneꢀbased fragments.
Subsequent synthetic optimization through fragment growing and reactivity analysis against
catalytic and nonꢀcatalytic cysteines revealed specific irreversible 3C protease inhibition.
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
Introduction
The group of covalently binding drugs comprises several prominent and relevant members
such as acetylsalicylic acid, βꢀlactam antibiotics, protonꢀpump inhibitors like omeprazole or the
antiplatelet agent clopidogrel, one of the bestꢀselling drugs.1ꢀ4 Covalent binding, however, is
rarely considered in drugꢀdesign as a general reluctance resides in the pharmaceutical industry
due to the putative risk of adverse drug reactions.5 This bias changed recently as described by the
title of a review by Singh et al. entitled "The resurgence of covalent drugs".6 Covalent linkage is
not only associated with risks, but also with advantageous target binding characteristics: (i)
prolonged target interaction, (ii) increased binding affinity, (iii) pharmacokinetic and
pharmacodynamic decoupling and (iv) improved selectivity.7ꢀ12 In combination with fragmentꢀ
based approaches, covalent binding has been utilized (i) for the identification of starting points
for inhibitor design, (ii) exploration of novel binding sites on proteins or (iii) assessing
ligandability of proteins.13ꢀ16 Covalent bond formation enables fragment binding unrelated to
affinity at shallow protein sites and hit identification and verification through mass spectrometry.
ACS Paragon Plus Environment
2

Page 3 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Existing approaches utilize covalent fragments (i) from accessible compound collection
through substructure filters, (ii) chemical synthesis of probes or entire molecule collections or
(iii) virtual libraries build from commercially available synthetic precursors with subsequent
synthesis of virtual screening hits.12, 14, 15, 17. Recently, we have applied proteinꢀtemplated
fragment ligation as an approach for the discovery of irreversible protease inhibitors. 15 18
Irreversible protein binding consists of the initial formation of a reversible nonꢀcovalent
complex EI, which subsequently undergoes transition to the final complex EꢀI through covalent
bond formation. The potency of such an inhibitor is given by the second order rate constant kinact
/KI [Mꢀ1sꢀ1]. Thereby, kinact denotes the rate of covalent bond formation, whereas KI accounts for
the stability of the initial complex due to the formation of favorable nonꢀcovalent interactions as
represented by the following scheme:
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
.
In order to identify covalent fragments, whose binding is driven by KI rather than kinact
,
Kathmann et al. proposed to only use covalent fragments with identical warheads, where the
reactivity is merely depending on substitution effects. This way, the best hits result from the
formation of favorable nonꢀcovalent interactions.17 Thereby, the pointless investigation of the
mostꢀactive hits that are only driven by the reactivity of the warhead is circumvented. However,
this approach can only be applied when working in a limited chemical space: Warhead types and
the addressed target as initial knowledge about applicable covalent interactions are a
prerequisite. Furthermore, synthesis efforts prior to the fragment screening are necessary.
Here we report a rational, computerꢀdriven approach for the selection of covalently binding
fragments from tangible as well as virtual molecule sources using 3D pharmacophores. Through
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 50
1
2
3
4
5
6
7
8
9
the introduction of a novel covalent bond feature type, a 3D pharmacophore captures the
formation of the essential nonꢀcovalent interaction as well as the necessary spatial orientation of
the warhead moiety for the subsequent covalent binding. This was utilized for identification of
covalently binding fragments for 3C protease (3Cpro), which is essential for the replication of
clinically relevant and emerging enteroviruses such as the coxsackievirus B3 (CVB3) being
associated with (i) aseptic meningitis, (ii) cardiomyopathy, (iii) typeꢀ1 diabetes and (iv) most
recently with dementia.19ꢀ22 Therefore, 3Cpro has been indicated as a promising target for the
design of enterovirus inhibitors for which the shallow polyprotein binding 3Cpro active site poses
a significant challenge for small molecule inhibition.23 Peptidomimetic analogues of the natural
substrate are the most potent inhibitors developed so far.24 These covalently inactivate the 3Cpro
through the binding of a Michael acceptor to the catalytic cysteine. Yet, their inherent poor
pharmacokinetic properties due to their peptidic nature limit the clinical utility and render the
development of nonꢀpeptidic inhibitors a key objective.25
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
Results & Discussion
Covalent pharmacophore model and fragment screening
In order to enable the 3D pharmacophoreꢀbased identification of covalent fragments, a novel
chemical feature type was introduced to our pharmacophore modeling software LigandScout.26
First, covalently binding substructures were assembled from literature.27ꢀ29 Despite the huge
number of reported covalently binding ligands and substructures, the reaction mechanisms of the
majority can be assigned to one of the following categories: (i) Michael addition, (ii) acylation,
(iii) acetal/ ketal formation, (iv) alkylation, (v) boronic ester formation, (vi) sulfonylation.
Thereby, covalently binding substructures, which are not applicable for rational design were
discarded. This includes (i) overly reactive moieties, which are prone for unspecific binding such
ACS Paragon Plus Environment
4

Page 5 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
as alkyl halides, (ii) covalent interactions that depend on prior transformation through the
enzyme, (iii) substructures, which undergo complex transformations and (iv) high reactivity
variability due to leaving groups. For the selected warheads, SMARTS patterns were developed
and implemented in LigandScout. The covalent bond feature is represented by an orange sphere,
which is placed on the ligand atom covalently bound or binding to the protein (figure 1A).
From crystal structures of peptidic ligands covalently bound to the CVB3 3Cpro (see
experimental section), a 3D pharmacophore model was generated, which comprises the essential
nonꢀcovalent interactions and in addition reflects the necessary spatial positioning of the warhead
moiety for covalent interaction with the catalytic Cys147 of the 3Cpro. The model was
subsequently statistically validated by using nonꢀpeptidic ligands reported in the literature and
has a high specificity of 0.87 and sensitivity of 0.99 (see supporting information).30ꢀ32 This shows
that the model, which was derived from coꢀcrystallized peptidic ligands, is able to retrieve
bioactive, nonꢀpeptidic 3Cpro inhibitors (figure 1C). The interactions comprised in the final
model are (i) covalent bond with the catalytic Cys147, (ii) hydrogen bonds to His161 and
Thr142, which mimic the interaction of the P1ꢀglutamine in the S1 pocket and (iii) lipophilic
contacts in the S2 pocket (figure 1B).
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
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 50
1
2
3
4
5
6
7
8
9
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 1. Covalent 3D pharmacophore model captures 3Cpro binding characteristics. (A)
Common covalently binding substructures matched by the novel covalent bond feature type
implemented in LigandScout, depicted in 2D and 3D. From top to bottom: Michael acceptor, αꢀ
ketoamide, αꢀhalomethyl ketone, (B) Validated 3D Pharmacophore model for covalent 3Cpro
binding. Orange sphere indicates covalent interaction, yellow spheres lipophilic contacts, red
arrows hydrogen bond acceptors and green arrows hydrogen bond donors, respectively. (C)
Active set molecule aligned to 3Cpro 3D pharmacophore model.
The generated 3D pharmacophore model was subsequently used to screen an inꢀhouse small
molecule collection, which was pruned prior to the virtual screening to ensure fragmentꢀlike
33 34
characteristics (relaxed rule of 3) and to remove PAINSꢀcontaining compounds.
This
resulted in approximately 3000 fragmentꢀlike molecules, which were included in the 3D
pharmacophoreꢀbased virtual screening. Forty seven virtual hits were retrieved, which were
further analyzed through covalent and nonꢀcovalent docking into the 3Cpro active site. For 19
fragments binding conformations of the initial and the covalent complex were retrieved, in which
the interaction of the 3D pharmacophore model are formed. These 19 fragments represent 12
different chemotypes. From each of these, one fragment was selected for biological evaluation in
an enzyme kinetic assay for 3Cpro inhibition (supporting information, figure S1A). Six fragments
ACS Paragon Plus Environment
6

Page 7 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
inhibited the 3Cpro by more than 30 % at 650 µM (supporting information, figure S1B). One
fragment (F1) showed a concentrationꢀ and timeꢀdependent activity with an apparent IC50* of
355 ± 30 ꢁM. The time dependency is indicated by the flattening of the progress curve, which is
a key characteristic of covalent binding (figure 2C). The covalent interaction results from the
ketone moiety of F1, which can form a reversible thiohemiketal with Cys147 of 3Cpro (figure
2A). This is facilitated through the occupation of the S1 pocket by the primary amide moiety of
F1 forming the crucial hydrogen bonds to His161 and Thr142. Further investigations of the timeꢀ
dependency of the 3Cpro inhibition by F1 were performed through incubation experiments: The
enzyme was incubated with the fragment for different time periods prior to starting the
enzymatic reaction. For F1, each longer incubation periods resulted in a greater reduction of
3Cpro activity in the subsequent enzymatic assay. However, during the progress of the enzymatic
reaction, 3Cpro activity increased for samples with longer incubation periods (figure 2D).
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
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 50
1
2
3
4
5
6
7
8
9
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. Fragment hit F1 shows time-dependent 3Cpro inhibition through formation of a
thiohemiketal. (A) Proposed covalent binding conformation of F1 in the active site of 3Cpro with
interactions depicted as 3D pharmacophore features. (B) 2D depiction of the interactions in the
reversible covalent complex of F1. Theoretical and experimentally observed m/z values for F1
and instability reaction product F1* depicted with its putative structure. (C) Progress curves for
the enzyme reaction in the presence of different concentrations of F1. (D) Relative enzymatic
activity of 3Cpro in the progress of the enzymatic reaction after incubation with F1 at 500 ꢁM for
60 min (diamond), 40 min (square), 20 min (triangle) or 0 min (circle). (E) Superimposed HPLC
chromatograms of F1 after incubation in assay buffer for 0 h, 1 h, 4 h and overnight. Peaks for
F1 and the instability reaction product F1* are indicated.
The timeꢀdependent behavior of F1 was further investigated through stability analysis using
LC/MS. Results indicate that overnight incubation of F1 in aqueous assay buffer resulted in
degradation into its carboxylic acid derivative F1* (figure 2B, E). High resolution MS show a
difference in the m/z values of ~1 Da between F1 and F1* as well as an increased ion count in
ACS Paragon Plus Environment
8

Page 9 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the negative mode. This structure fits the detected m/z values in the positive and negative mode,
respectively. Since deamidation reactions do not occur easily, we surmise the reaction to occur
via a hydrolytic opening of the labile lactone moiety and recyclization under deamination
(supporting information, figure S2). This mechanism is in accordance with a reported synthesis
scheme for chromeneꢀ3ꢀcarboxamides such as F1.35 Furthermore, the conversion of the amide to
an acid would explain the observed reduction in activity after incubation and thus confirm the
proposed binding conformation. SAR data for S1 pocket binding suggests a 50ꢀfold reduction in
activity for the exchange of the primary amide group to an acid.31, 36 In addition, the observed
timeꢀdependent activity of F1, the initial increase as well as the decrease alongside its decay, are
in accordance to the proposed formation of a reversible covalent thiohemiketal. However, this
prevents further confirmation of the covalent binding through mass spectrometry and the
chemical instability of F1 limits the attractiveness of this scaffold for further investigations.
Scaffold hopping
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
In order to identify bioꢀisosteric fragments to F1, a search for analogues was performed
through substructureꢀbased scaffold hopping.37 The essential 3Cpro binding features of F1, the
covalently binding ketone moiety and the primary amide group for S1 pocket hydrogen bonds,
were transferred into a SMARTS pattern and used to filter commercial fragment collections
(figure 3A). This led to the identification of a fiveꢀmembered heterocycleꢀbased scaffold. The
essential primary amide and ketone moieties are arranged in a metaꢀsubstitution pattern, which
enables the respective fragments to match the 3D pharmacophore model and to adopt the binding
conformation of F1 in the 3Cpro active site. From the initially identified fragments of this
scaffold, five were commercially available from a single vendor and thus purchased for
biological evaluation (figure 3B). Single concentration measurements revealed an apparent
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 50
1
2
3
4
5
6
7
8
9
activity cliff.38 Fragment C5 almost completely inactivates the 3Cpro at 650 µM, whereas none of
the other fragments showed more than 20 % inhibition. Doseꢀresponse measurements for C5
showed reasonable concentrationꢀdependent activity with an apparent IC50 in the low threeꢀdigit
micromolar range (figure 3D) and incubation experiments confirmed the timeꢀdependent mode
of action (figure 3E). Plots of the observed inactivation rate against the concentration of C5
proceed through the origin indicating an irreversible interaction with the 3Cpro (figure 3F). This
result provides an explanation for the apparent difference in biological activity within the set of
fragments. However, it also causes suspicion for binding specificity as C5 does not contain an
apparent irreversible warhead and is highly similar to the other analyzed fragments, especially
C4 and C6.
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
ACS Paragon Plus Environment
10

Page 11 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
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 3. F1 analogue search through scaffold hopping enabled identification of irreversible
3Cpro inhibitor C5. (A) SMARTS substructure pattern utilized for scaffold hopping matched for
F1 and fiveꢀmembered heterocycle. Matched substructure is colored in blue. 3D pharmacophoreꢀ
based superimposition of F1 and a pyrroleꢀbased fragment. (B) Selected fragments and (C)
relative 3Cpro inhibition at 650 µM. Inhibition for C7 was not determined (nd) due to solubility
issues. (D) Doseꢀresponse curve for C5. (E) Relative enzymatic activity of the 3Cpro in the
progress of the enzymatic reaction after incubation with C5 at 100 µM for 60 min (diamonds),
40 min (square), 20 min (triangle) or 0 min (circle). (F) Plot of kobs against the concentration of
C5 and inactivation rate (kinact/KI) calculated from the slope of the linear part of the curve.
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 50
1
2
3
4
5
6
7
8
9
Protein mass spectrometry was performed to investigate the binding of C5. Deconvolution of
the obtained mass spectra revealed three components (figure 4A): One with the mass of the apo
3Cpro (21456 Da) and two components with greater masses, 21620 Da and 21562 Da,
respectively. The increase in mass equals the molecular weight of C5 for none of the
compounds, which nourished suspicion concerning the specificity of inhibition. However, we
were able to propose a putative mode of action through detailed analysis of the structural
differences of the fragments and the increase in mass for the different adducts detected. The
major difference of C5 to the other fragments is the αꢀmethyl ketone substituent, which is a
thiophenol group in C5. Its molecular weight of 109 Da equals the increase in mass for the minor
adduct. Furthermore, the mass shift for the major mass peak is 165 Da, which equals the
fragment C5 mass (275 Da) except the 109 Da of the thiophenol moiety. Based on this
observation, we reasoned that C5 reacts similarly to other activated methyl ketones (figure 4B).
The attack of the cysteine thiolate leads to the formation of an irreversible thioether at the αꢀ
carbon of the ketone moiety of C5 through displacement of the thiophenolate. This irreversible
reaction product shows the same mass shift as the major peak (+165 Da). In addition, the
displaced thiophenolate can react with the cysteine through the formation of a disulfide, which
equals the mass shift detected for the minor product (+109 Da). High resolution MALDI/TOF
mass spectrometry further confirmed the exact masses and mass shifts for the 3CproꢀC5 reaction
adducts (figure 4C).
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
ACS Paragon Plus Environment
12

Page 13 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
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 4. Mass spectrometry reveals irreversible 3Cpro binding by C5 through the
displacement of thiophenol. (A) ESI mass spectrum of the 3Cpro with C5 after deconvolution.
(B) Putative reaction mechanism for the formation of the 3CproꢀC5 adducts. (C) Deconvoluted
highꢀresolution MALDIꢀTOF spectra of the 3Cpro after incubation with C5. The main peak of
21634.8 Da corresponds to the thioether adduct formed after reaction of C5 with the 3Cpro.
The MS data and the proposed reaction products provide a rationale for the determined
biological activity of C5 as an irreversible 3Cpro inhibitor as well as for the difference to the
other selected fragments. To further rationalize the binding mode of C5 the initially proposed
binding hypothesis, based on the 3D pharmacophore model for covalent 3Cpro binding, was
revisited. The initial hypothesis suggested the formation of a reversible thiohemiketal from a
nonꢀcovalent complex in which the primary amide moiety forms the crucial hydrogen bonds to
His161 and Thr142 in the S1 pocket and the ketone moiety is positioned in close proximity to the
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 50
1
2
3
4
5
6
7
8
9
catalytic Cys147 (figure 5A, B). In addition, the attack of the cysteine thiolate can also occur at
the αꢀcarbon of the ketone moiety of C5 in the initial nonꢀcovalent complex. This leads to the
irreversible thioether product detected in the mass spectrometry through a SN2 reaction via the
displacement of thiophenolate. Covalent docking shows that the amide moiety of thioether
product is able to form the crucial hydrogen bonds in the S1 pocket and thus indicates a
conserved binding mode confirming the 3D pharmacophoreꢀbased binding hypothesis (figure
5C).
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 5. Proposed covalent binding mode of C5. 3D (top) and 2D (bottom) depictions of the
complexes and interactions formed between C5 and the 3Cpro. Interactions are described as
pharmacophore features. Yellow spheres indicate lipophilic contacts, orange sphere indicates
covalent bond formation, green and red arrows indicate hydrogen bond donor or acceptor
functionalities.
The biological activity and the observed activity cliff result from the formation of an
irreversible thioether by C5 through its αꢀphenylthiomethyl ketone moiety, which, to our
knowledge, has not been reported to date.
Fragment optimization & warhead analysis
ACS Paragon Plus Environment
14

Page 15 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To further investigate the binding mechanism of C5 and to increase 3Cpro inhibition potency,
synthetic fragment optimization was performed. Based on the proposed binding model, growing
vectors for C5 were defined to occupy additional pockets in the 3Cpro active site, which include
the adjacent S2 and S4 pocket, respectively (figure 6A). The pyrrole nitrogen of C5 represents a
putative synthetic handle for the incorporation of additional functionalities to occupy the
aforementioned pockets as the methyl substituent does not contribute to 3Cpro binding and thus
can be substituted. A de novo design workflow was developed for the rational design of pyrrole
substituted C5 analogues occupying the adjacent pockets while maintaining the original binding
mode. Hereby, synthesisꢀtailored transformation rules, encoded as SMIRKS patterns, are applied
in combination with tangible building blocks to generate virtual libraries of synthetically feasible
compounds. For the design of C5 analogues, its norꢀderivative was used as starting point for the
de novo design workflow, in which it was transformed via nucleophilic substitution at the pyrrole
nitrogen with alkyl halide building blocks (figure 6B). The obtained virtual library of
approximately 2000 Nꢀalkylated C5 analogues was virtually screened using the initial 3D
pharmacophore model, which comprises the essential hydrophobic contacts to Leu127 for S2
pocket binding through a lipophilic feature (figure 1B). 3D pharmacophore matching hits were
subsequently docked in all three putative binding states to confirm formation of the crucial
hydrogen bonds in the S1 pocket (figure 6C). The final selection of C5 analogues occupy the S2
pocket through lipophilic pyrrole substituents, which range from a flexible phenyl alkyl ether to
rigid biphenyl methyl moieties. The corresponding commercially available alkyl halide building
blocks are depicted in figure 6D.
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
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 50
1
2
3
4
5
6
7
8
9
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 6. C5 fragment growing in 3Cpro active site driven by de novo design. (A) Nonꢀ
covalent binding conformation of C5 and growing vectors toward S2 and S4 pocket,
respectively. (B) C5 fragment growing scheme starting from putative chemical synthesis
precursor, input structure for de novo designꢀbased fragment growing and generalized structure
of virtual C5 analogues with variable pyrrole substituent. (C) Docking poses of a representative
fragment growing analogue of C5 occupying the S2 pocket of the 3Cpro. All three binding states
are shown: Initial complex (white stick coloring), thiohemiketal intermediate (light blue stick
coloring) and final thioether (orange coloring). Hydrogen bonds are indicated by red dashed
lines. (D) Chemical structures of commercially available building blocks utilized in the de novo
design of virtual screening hits for the S2 pocket.
ACS Paragon Plus Environment
16

Page 17 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The synthesis of the de novo designed C5 analogues addressing the S2 pocket was conducted
according to scheme 1. From the commercially available methyl 2ꢀpyrroleꢀcarboxylate 1 the
ketone 2 was generated through FriedelꢀCrafts acylation. NꢀAlkylation with the respective alkyl
halide led to the Nꢀsubstituted ester 3, which was converted into the corresponding primary
amide 5 via the free acid 4. The phenylthiomethyl ketone 7 was obtained from the reaction of the
bromomethyl ketone 6 with thiophenol under basic conditions. For the synthesis the alkyl
bromides and among those the benzylic and bromomethyl ketones were prioritized due their
higher reactivity for the pyrrole Nꢀalkylation. The second ketone moiety proved to challenging
for the selective bromination step. By using 1ꢀbromoꢀ3ꢀphenylꢀpropene, a first C5 analogue 7a
was synthesized. In addition, the double substituted 2,2ꢀbisꢀ(phenylthiol) methyl ketone 8a was
also obtained using an excess of thiophenol. This unintentionally synthesized analogue provided
the opportunity to assess the specificity of the phenylthiomethyl ketone warhead, which was
extended by including the synthesis intermediates 5a and 6a.
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
Scheme 1: Scheme for the synthesis of the C5 analogues.
Reagents and conditions: (a) AlCl3, CH3COCl, DCE, rt; (b) K2CO3, alkylhalide, DMF; (c)
KOH, MeOH, THF; (d) (COCl)2, NH3, THF; (e) PTTB, THF; (f) K2CO3, DMF, C6H5SH.
The biological evaluation showed activity against the CVB3 3Cpro for all four synthesized
compounds with a timeꢀdependent mode of action. 5a and 8a showed reversible 3Cpro inhibition
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 50
1
2
3
4
5
6
7
8
9
with Ki values of 197±8 ꢁM and 272±8 ꢁM, respectively. In contrast, 6a and 7a were irreversible
3Cpro inhibitors with kinact/KI values of 1,707±33 Mꢀ1sꢀ1 and 54.4±3.8 Mꢀ1sꢀ1, respectively (Table
1). The irreversible binding of the phenylthiomethyl ketone 7a provides an initial confirmation of
the proposed binding mechanism of C5. In contrast, the 2,2ꢀbisꢀ(phenylthiol) methyl ketone 8a is
a reversible 3Cpro inhibitor, in an analog way to perꢀhalomethyl ketones. These are reversible
inhibitors in contrast to their monoꢀhalogenated analogues, such as the bromomethyl ketone 6a,
which binds irreversibly.28 Furthermore, the bulkiness of the thiophenol substituents of 8a
potentially hampers the covalent binding as 8a is only as active as the significantly smaller
ketone 5a. This is less active than the fragment hit C5, which is due to the missing thiophenol
leaving group. Thus, the effect of the introduction of the phenylpropenyl substituent is only
possible by comparison to the other pyrroleꢀbased fragments devoid of a leaving group. These
were inactive at 650 ꢁM showing a significant boost in activity through occupation of the S2
pocket by 5a. However, the increase of the kinact/KI value from C5 to 7a is only threefold, which
raises questions about the contribution of the phenylpropene substituent for 7a.
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. C5 and synthetic analogue 3Cpro inhibition values and binding kinetics (nd: not
determined).
ID Ki [ꢁM]
kinact/KI [Mꢀ1sꢀ1]
Binding kinetics
CVB3
EV D68
ꢀ
CVB3
EV D68
ꢀ
16.8 ± 5
21.3 ± 8
ꢀ
Irreversible
Reversible
C5
5a
197 ± 8
422 ± 8
ꢀ
ꢀ
ꢀ
1,707 ± 33
54.4 ± 4
ꢀ
1,214 ± 12
160 ± 26
ꢀ
Irreversible
Irreversible
Reversible
6a
7a
8a
ꢀ
ꢀ
272 ± 8
nd
ACS Paragon Plus Environment
18

Page 19 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The activity of the synthesized analogues was additionally investigated using a second 3C
protease from the enterovirus D68 (EV D68). This 3Cpro has an overall sequence similarity of 67
% and all residues relevant for ligand binding in the active site are conserved so that a highly
similar binding mode can be assumed (Supporting information, figure S4). The crucial
distinction to the CVB3 3Cpro is the presence of a nonꢀcatalytic cysteine residue (Cys60)
allowing for additional exploration of the reaction specificity of the irreversible inhibitors.
Evaluation of the EV D68 3Cpro inhibition revealed comparable inhibition values (Table 1). The
phenylthiomethyl ketones C5 and 7a show slightly increased kinact/KI values of 21.3±8 Mꢀ1sꢀ1 and
160.4±26 Mꢀ1sꢀ1, respectively. In contrast, the kinact/KI of 6a is reduced by one third to 1,214±12
Mꢀ1sꢀ1 and it shows a steep curvature of the doseꢀresponse plot with a Hill slope of 1.8 indicating
unspecific or multiple binding events (Supporting information, figure S5A). Mass spectrometry
was performed for EV D68 3Cpro after incubation with the irreversible inhibitors to investigate
this hypothesis further (figure 7). For the apo EV D68 3Cpro a single peak of 21261.26 Da was
found. After reaction with 6a, two main peaks occur at 21527.6 Da and 21793.67 Da,
respectively. These correspond to mass shifts of 266 Da and 532 Da, respectively, which is twice
as much as the former. The mass shift of 266 Da equals the mass of 6a except for the bromine.
Thus, the peaks reflect single and double binding of 6a through unspecific reaction with either
cysteine residue. In contrast, for 7a only one peak with 21527.05 Da was found in addition to the
apo mass peak. The mass shift of 266 Da corresponds to the mass of 7a except the thiophenol
moiety and thus a single reaction of 7a. This putatively exclusively occurs via the catalytic
Cys147 as this is more reactive than Cys60 and secondly due to the binding kinetics of 6a with
the high Hill slope indicating that Cys60 binding does not affect the 3Cpro activity. Only a single
adduct peak of 21426.13 Da was detected for C5 with a mass shift of 164 Da, the same as
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
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 50
1
2
3
4
5
6
7
8
9
determined for the CVB3 3Cpro, which confirms the proposed binding mechanism and reaction
specificity of the phenylthiomethyl ketone warhead for catalytic cysteine residues.
The kinact/KI value increase from C5 to 7a is greater for the EV D68, yet the small abundance in
the mass spectrum of 7a indicates a smaller rate of irreversible binding in comparison to C5.
Thus, the increase in activity can be attributed to better nonꢀcovalent interactions due to the
phenylpropene substituent and a better KI.
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 7. EV D68 3Cpro binding of C5 and synthesized analogues elucidated by mass
spectrometry. Deconvoluted ESI/QTOF mass spectra of EV D68 3Cpro apo structure and after
incubation with C5, 7a and 6a (from left to right). Main peaks are labeled with corresponding
m/z value. Schematic depictions of EV D68 3Cpro or 3Cproꢀinhibitor adducts are shown to
indicate binding to catalytic Cys147 or nonꢀcatalytic Cys60, respectively. Note that the single
adduct peak of the 6a MS results from binding to either cysteine residue.
The reaction specificity of the phenylthiomethyl ketone moiety was further investigated
through its reaction with the nucleophile glutathione (GSH) (figure 8B). This biologically
ACS Paragon Plus Environment
20

Page 21 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ubiquitous tripeptide contains a cysteine thiol moiety and can thus react in analogy to the
cysteines of the 3C proteases. In addition to the active ligands C5 and 7a, available
phenylthiomethyl ketone bearing fragments were included in the analysis (Figure 8A). C8 was
taken from our inꢀhouse small molecule collection and C9 was prepared through a singleꢀstep
synthesis from 2ꢀbromoacetophenone and thiophenol. Both compounds do not show activity
against the CVB3 3Cpro in the micromolar range (Supporting information, figure S5A).
Furthermore, the synthesized 2,2ꢀbisꢀ(phenylthiol) methyl ketone 8a and the bromomethyl
ketone 6a were analyzed. Each compound was incubated with GSH (2.5 mM) at an equimolar
concentration. The reaction was monitored via LC/MS for 24 h (Figure 8C). The formation of
new entities or a reduction of the initial compound could only be detected for 6a within the
analyzed time period (figure 8E). For 6a, a second peak with a smaller retention time appears,
whereas the peak of the original compound vanished. This indicates the formation of reaction
product between 6a and GSH. This is supported by the m/z of the ions in the MS spectra of the
new peak, which fit the calculated m/z of a reaction product formed through the nucleophilic
attack of the thiol moiety at the bromomethyl ketone group and displacement of the bromine
(figure 8D). This reactivity of 6a is known and was used for the synthesis of the desired
compound 7a. It also served as positive control for this analysis as the same GSHꢀadduct with
the same m/z and retention time would have been formed by 7a.
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
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 50
1
2
3
4
5
6
7
8
9
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 8. Phenylthiomethyl ketones are stable in glutathione assay checking warhead
reaction specificity. (A) Assembled set of activated αꢀsubstituted methyl ketones, warhead
substructure is highlighted by blue frame. (B) Chemical structure of the nucleophile glutathione
(GSH). (C) Superimposed LC chromatograms of C5 (top) and 7a (bottom) without (gray) and
after incubation with GSH for 2 h (dark gray) and 24 h (black). (D) Chemical structure of the
putative reaction product 6aꢀGSH. (E) Mass spectra of the reaction product 6a-GSH in the
positive (right) and negative mode (left) with relevant ions and m/z. Superimposed LC
chromatograms of 6a without (gray) and after incubation with GSH for 2 h (dark gray) and 24 h
(black).
ACS Paragon Plus Environment
22

Page 23 of 50
Journal of Medicinal Chemistry
1
2
3
4
Binding mechanism
5
6
7
8
The reaction of the phenylthiomethyl ketones with the 3Cpro can lead either to the formation of
a reversible thiohemiketal or an irreversible thioether through SN2 displacement of the
thiophenol moiety. The differences in the biological activity and irreversible binding between C5
and 7a can be explained by a higher affinity of 7a in the thiohemiketal state and slower
displacement of the thiophenol moiety in comparison to C5. The introduction of the
phenylpropenyl substituent in 7a apparently stabilizes the thiohemiketal and thus reduces the
formation of the irreversible thioether product detected by mass spectrometry. This preference of
7a for the thiohemiketal is supported by docking studies, which show that the positioning of the
phenylpropenyl moiety may vary for the covalent binding products (figure 9AꢀC). This is due to
the difference of the ligand atom covalently bound to the catalytic cysteine. The necessary
formation of the hydrogen bonds in the S1 pocket requires a slight perturbation of the
phenylpropenyl substituent occupying the S2 pocket in the thioether product (figure 9C). This
does not apply for the binding of C5 due to the insignificant contribution of its Nꢀmethyl
substituent. Therefore, the formation of the thioether through displacement of the thiophenol
moiety is slower for 7a, which reduces the benefit of the warhead in comparison to C5.
9
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
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 50
1
2
3
4
5
6
7
8
9
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 9: Perturbation of S2 occupation for the different covalent products of 7a. 3D (top)
and 2D (bottom) depiction of the three complexes formed during the 3Cpro binding by 7a.
Dashed arrow indicates the repositioning of the phenylpropenyl substituent during the formation
of the thioether. Hydrogen bonds are indicated as red dashed lines in the 3D depiction. Yellow
lines indicate lipophilic contacts.
The glutathione reaction analysis as well as the mass spectrometry data show specificity of the
phenylthiomethyl ketone reactivity for catalytic cysteines. This implies that the covalent binding
results from the proper spatial positioning of the warhead in the initial complex of the bioactive
thiophenolmethyl ketones C5 and 7a comprising the essential chemical features, especially the
primary amide moiety for S1 pocket binding. In addition, the thiophenol moiety seems to be
favorable for occupying the S1’ pocket as the pyrroleꢀbased fragments C4 and C6, which only
differ in the αꢀsubstituent, are virtually inactive. Especially since the SN2 displacement reaction
would rather suggest the Nꢀmethyltetrazolethiol substituent of C6 to be a better leaving group
ACS Paragon Plus Environment
24

Page 25 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
due to its higher acidity in comparison to the thiophenol moiety. However, the influence of the
catalytic mechanism for the unusual reaction and specificity of the phenylthiomethyl ketone with
the 3Cpro is not clear especially as it has also been reported to be an unreactive warhead for other
targets.39 This would require further analyses to assess its impact for the displacement of the
thiophenol and to verify the proposed binding mechanism. The evaluation of a broader range of
leaving groups such as the related dimethylsulfonium methyl ketones would be insightful to
explore the factors of the irreversible 3C protease inactivation by activated methyl ketones.
Conclusion
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
The presented study reports the identification of covalently binding fragments for enterovirus
3C protease through the use of 3D pharmacophores with novel chemical features representing
potential covalent linkage. The combination of features for covalent and nonꢀcovalent
interactions reflects the general consideration for the relevance of the initial, nonꢀcovalent
complex in the design of specific, KIꢀdriven covalent inhibitors. Fragment hit validation and
confirmation of the binding hypothesis through mass spectrometric analysis allowed for a
rational, structureꢀbased optimization in the absence of crystallographic data on the binding
conformation. The synthesized analogue shows inactivation rates in the threeꢀdigit range against
3C proteases with a nonꢀpeptidic scaffold making it a promising starting point for further
optimization.
Mass spectrometric analysis of selected fragments led to the identification of the
phenylthiomethyl ketone moiety as a covalently binding substructure for irreversible 3C
proteases inactivation. The binding kinetics of the discovered fragments and structural modelling
suggest key relevance of the nonꢀcovalent interactions for the proper initial positioning of the
phenylthiomethyl ketone warhead to enable the subsequent covalent binding through
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 50
1
2
3
4
5
6
7
8
9
displacement of the thiophenolate. This provides novel insights for the reaction of activated
methyl ketones and extends the available range of leaving groups. The conducted experiments
show considerable reaction specificity of the phenylthiomethyl ketone warhead, which suggests
an application for the selective covalent inactivation of target proteins or for activityꢀbased
protein profiling.
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
Experimental procedures
Reagent sources and quality assessment: Commercially available fragments were purchased
from Enamine (Ukraine). The inꢀhouse small molecule collection was assembled by using a
maximum common substructure concept based on bioactive molecules listed in the World Drug
Index as reported by Lisurek et al.40, 41 The compounds were purchased from ChemDiv Inc. (San
Diego, USA). Purity >95 % and identity of the compounds was determined by LC/MS using an
Agilent (Santa Clara, USA) 1100 LC/MS system equipped with a LunaꢀC18 RPꢀcolumn (3 ꢁm)
by Phenomenex (Torrance, USA). Samples were prepared by mixing 1 ꢁL of the inhibitor
DMSO stock solution with 99 ꢁL of a 1:1 mixture of water and acetonitrile. As solvents water
and acetonitrile were used containing each 0.1 % (v/v) of formic acid. The ratio of water to
acetonitrile was changed from 95:5 to 1:99 during the chromatographic run with a constant
flowrate of 1 mL/min.
The small molecule collection was filtered by using a KNIME workflow with the following
physicochemical property cut offs, calculated by RDKit KNIME nodes: Molecular weight ≤325
43
Da, heavy atom count ≤ 20 and cLogP ≤ 3.42,
Compounds were filtered for PAINS
substructures by using inꢀhouse SMARTS patterns in a KNIME workflow based on reported
substructuresꢀ and filters. 34 44
ACS Paragon Plus Environment
26

Page 27 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3Cpro inhibition assay and evaluation: 3Cpro was expressed and purified as described
previously.15 The biochemical evaluation for CVB3 and EV D68 3Cpro inhibition was done by
using the FRET substrate NꢀDabcylꢀKTLEALFQGPPVYEꢀ(Edans)ꢀNH2 purchased from
Biosyntan (Berlin, Germany). In 384 well plates (Corning, USA) 10 μL buffer consisting of 100
mM HEPESꢀNaOH, 1 mM EDTA and 0.01 % TritonX (v/v) at a pH of 7.5 were mixed with 10
μL of 3Cpro (40 μM), 10 μL of substrate, diluted in the assay buffer to a concentration of 20 μM,
and either 1 μL of the inhibitor diluted in DMSO or only DMSO for the positive control. The
fluorescence signal was measured at a wavelength of 538 nm after irradiation at 355 nm with a
Safire II Fluorimeter (Tecan, Austria) over 30 min.
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
Each concentration was tested in triplicates and the obtained data was analyzed using
GraphPad Prism. Apparent IC50 values were calculated by nonꢀlinear regression by using
equation vi/v0=1/(1+[I]/ IC50)h, where vi is the initial enzyme velocity, v0 is the velocity in the
absence of the inhibitor and [I] is the inhibitor concentration. Observed firstꢀorder rate constant
kobs values were calculated for covalent reversible inhibition by using equation [P]=vst+((viꢀ
v0)/kobs)(1ꢀe^(ꢀkobst)), where [P] is the product concentration. Ki values were obtained from the
ratio koff/kon according to equation Ki= koff/kon. kobs values for irreversible inhibition were
determined through equation [P]=vi/kobs (1ꢀe^(ꢀkobst)). The second orderꢀrate constants kinact/KI
were derived from the slopes of the linear part of the graph of kobs as a function of the inhibitor
concentration based on equation kobs/[I] = kinact/KI, where kinact is the rate of maximum enzyme
inactivation at an infinite inhibitor concentration and KI refers to the inhibitor concentration
displaying halfꢀmaximum enzyme inactivation.
3Cpro incubation assay. For the incubation experiments the same reagents and quantities were
used as described above. The protein solution was mixed with the inhibitor or DMSO for the
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 50
1
2
3
4
5
6
7
8
9
control in the assay buffer and covered with a lid to prevent evaporation. After 60, 40, 20 or zero
minutes the enzymatic reaction was initiated by addition of the substrate and monitored for 60
min. The incubation was done in triplicated for each time frame of incubation and the respective
control. The obtained data was analyzed using GraphPad Prism. The relative enzyme activity for
every fiveꢀminute time step of the enzymatic reaction was calculated for every distinct time
frame of incubation. The obtained relative enzyme activity was plotted as a function of the
enzyme reaction time.
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
Glutathione reaction assay. At 2.5 mM the inhibitor, dissolved in DMSO, was mixed with an
equimolar solution of glutathione dissolved in the assay buffer to a volume of 50 μL. The
mixture was incubated at room temperature for 2 h and 24 h, respectively. After each incubation
period an aliquot was analyzed via LC/MS as described above. Controls for each inhibitor were
prepared and treated analogously, except the omission of glutathione.
Protein mass spectrometry. For protein mass spectrometry the same reagents were used as
described above. For CVB3 3Cpro MS 10 ꢁL of the stock solution were mixed with 20 ꢁL buffer
and 1 ꢁL of the inhibitor dissolved in DMSO, which was added at a concentration, leading to 50
% reduction of 3Cpro activity after 30 min in the kinetic assay. The samples were incubated for
60 min and then subjected to LC/MS analysis using the following procedure. For
chromatographic separation, a Phenomenex Jupiter Cꢀ18 RPꢀcolumn (5 ꢁm) was used. As
solvents water and acetonitrile were used containing each 0.1 % (v/v) of formic acid. The ratio of
water to acetonitrile was changed from 95:5 within the first 35 min of the run to 1:99, which was
kept constant for 5 min and then turn back to the initial ratio within the last 4 min of the
chromatographic run. The flow rate was set to 0.5 mL/min. For ionization the following
parameters were used: Capillary voltage 3.3 kV, sampling cone 40 V, source offset 60 V, source
ACS Paragon Plus Environment
28

Page 29 of 50
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
temperature 100 °C, desolvation temperature 250 °C, desolvation gas flow 600 L/h. MaxEnt1
method of the MassLynx software (Waters Co., Milford, UK) was used for deconvolution of the
obtained protein mass spectra.
9
For high resolution masses and the mass spectrometry of the EV D68 3Cpro the following
procedures were used. Small molecule samples were prepared as described above. For protein
MS, 200 ꢁL of EV D68 3Cpro stock solution were mixed with 50 ꢁL buffer and 10 ꢁL of the
inhibitor dissolved in DMSO, which was added at a concentration, leading to 50 % reduction of
3Cpro activity after 30 min in the kinetic assay. The samples were incubated for 60 min and
subsequently concentrated and desalted through ultrafiltration using Roti®Spin MINIꢀ10 devices
from Carl Roth GmbH & Co. KG. For sample purification 1 mL of purified water was used. The
sample was concentrated to a volume of 20ꢀ30 ꢁL.
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
Sample measurements were performed using an Agilent HPLC/QTOFꢀMS instrument. The
Infinity 1290 UHPLC equipped with a Phenomenex Jupiter Cꢀ18 RPꢀcolumn (5 ꢁm) was used
for chromatographic separation. 10 ꢁL of injected samples were eluted with a flow rate of 0.5
mL/min using a gradient of water and acetonitrile containing each 0.1 % (v/v) of formic acid,
which was changed from 95:5 to 1:99 within 3 min of the chromatographic run. This was held
constant for the next 6 min and then changed back within the next minute. The 6550 iFunnel
QTOF was used for mass analysis with the following parameters of the electrospray ionizationꢀ
source: Gas temperature 200 °C, gas flow 14 l/min, nebulizer gas at 35 psig, sheath gas
temperature 350 °C, sheath gas flow 11 L/min, capillary voltage 5 kV, nozzle voltage 1 kV,
fragmentor voltage 420 V. The analysis was run in the positive mode with the reference masses
of 121.050873 m/z and 922.009798 m/z and in the negative mode with the reference mass of
ACS Paragon Plus Environment
29