Identification of a New Zinc Binding Chemotype by Fragment Screening

Article abstract of DOI:10.1021/acs.jmedchem.7b00606

The discovery of a new zinc binding chemotype from screening a nonbiased fragment library is reported. Using the orthogonal fragment screening methods of native state mass spectrometry and surface plasmon resonance a 3-unsubstituted 2,4-oxazolidinedione fragment was found to have low micromolar binding affinity to the zinc metalloenzyme carbonic anhydrase II (CA II). This affinity approached that of fragment sized primary benzenesulfonamides, the classical zinc binding group found in most CA II inhibitors. Protein X-ray crystallography established that 3-unsubstituted 2,4-oxazolidinediones bound to CA II via an interaction of the acidic ring nitrogen with the CA II active site zinc, as well as two hydrogen bonds between the oxazolidinedione ring oxygen and the CA II protein backbone. Furthermore, 3-unsubstituted 2,4-oxazolidinediones appear to be a viable starting point for the development of an alternative class of CA inhibitor, wherein the medicinal chemistry pedigree of primary sulfonamides has dominated for several decades.

Full text of DOI:10.1021/acs.jmedchem.7b00606

Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES
Article
Identification of a New Zinc Binding Chemotype by Fragment Screening
Panagiotis K. Chrysanthopoulos, Prashant Mujumdar, Lucy A. Woods,
Olan Dolezal, Bin Ren, Thomas S. Peat, and Sally-Ann Poulsen
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00606 • Publication Date (Web): 17 Aug 2017
Downloaded from http://pubs.acs.org on August 18, 2017
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 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Identification of a New Zinc Binding Chemotype by Fragment Screening
1,#
1,#
1
2
Panagiotis K. Chrysanthopoulos , Prashant Mujumdar , Lucy A. Woods , Olan Dolezal ,
2
2,*
1,*
Bin Ren , Thomas S. Peat and SallyꢀAnn Poulsen
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
Griffith University, Griffith Institute for Drug Discovery, Nathan, Brisbane, Queensland
4
111, Australia
2
CSIRO, Biomedical Manufacturing Program, 343 Royal Parade, Parkville, Melbourne,
Victoria 3052, Australia
#
Equal first authors
Email s.poulsen@griffith.edu.au, tom.peat@csiro.au
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Abstract
The discovery of a new zinc binding chemotype from screening a nonꢀbiased fragment
library is reported. Using the orthogonal fragment screening methods of native state mass
spectrometry and surface plasmon resonance a 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinedione
fragment was found to have low micromolar binding affinity to the zinc metalloenzyme
carbonic anhydrase II (CA II). This affinity approached that of fragment sized primary
benzene sulfonamides, the classical zinc binding group found in most CA II inhibitors.
Protein Xꢀray crystallography established that 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinediones bound
to CA II via an interaction of the acidic ring nitrogen with the CA II active site zinc, as well
as a hydrogen bond between the oxazolidinedione ring oxygen and the CA II protein
backbone. Furthermore, 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinediones appear to be a viable starting
point for the development of an alternative class of CA inhibitor, wherein the medicinal
chemistry pedigree of primary sulfonamides has dominated for several decades.
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
ACS Paragon Plus Environment

Page 3 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Introduction
Proteins that require a metal cofactor for function have gained significant traction as
1, 2
therapeutic targets, with the mechanism of action of a number of clinically approved drugs
3
ꢀ6
and investigational compounds attributed to metalloenzyme inhibition. Zinc is one of the
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
, 8
most abundant metal cofactors in the human proteome
and small molecule zinc
9
metalloenzyme inhibitors usually comprise a zinc binding pharmacophore that interacts
10ꢀ12
directly with the metal ion to block endogenous activity.
The importance of zinc proteins
is exemplified in recent studies that show a correlation between zinc recruitment and the
1
3
complexity of eukaryotic genomes. There are a number of zinc binding groups (ZBGs) that
have been explored in exquisite structural detail for binding to zinc metalloenzymes, these
include the carboxylate, hydroxamate, sulfonamide, thiol and phosphonate functional groups.
These functional groups are well represented in the ligands of zinc metalloproteinꢀligand
structures that have been deposited with the protein data bank (PDB) and in approved drugs
1
4
targeting metalloenzymes. With the advent of fragment based drug discovery (FBDD) has
come a new opportunity to discover alternate and novel metal binding groups by fragment
15
screening. Fragment screening represents a dramatic change in the approach to drug
discovery that makes use of low molecular weight small molecules (i.e. fragments) for
unbiased identification of novel structural motifs that bind to proteins with high ligand
efficiency. A hit fragment may be structurally elaborated to a new chemical entity with drugꢀ
like properties and functionality to improve strength and specificity of binding interactions
with the target protein. Since 2005, fragment screening has resulted in more than 30 drug
candidates that have entered clinical trials, with two U.S. Food and Drug Administration
16
(FDA) approved drugs and several more compounds in advanced trials. The relatively short
timeframe from fragment to U.S. FDA approved drugs has led to adoption of fragment
16
screening approaches in academia, biotech and pharma. A combination of FBDD with
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 62
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
metalloprotein targets has been the focus of several successful FBDD campaigns by Cohen
and colleagues utilising fragment libraries assembled from pharmacophores that have a
12, 17, 18
known predilection toward metal binding.
Additionally, Klebe and colleagues have
characterised the structure of a number of alternate ZBGs for carbonic anhydrase II (CA II,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
9
EC 4.2.1.1).
Our group is interested in modulating carbonic anhydrase (CA) activity for therapeutic
purposes and has previously targeted inhibition of this zinc metalloenzyme using both
2
0ꢀ24
standard and novel medicinal chemistry strategies.
hCAs (h = human) regulate pH
homeostasis by catalysing the reversible hydration of carbon dioxide to bicarbonate and a
ꢀ
+
proton: CO
2
+ H
2
O ꢀ HCO
3
+ H and evidence is mounting that modulating tumour cell pH
24ꢀ26
homeostasis may prove an effective antiꢀtumour strategy.
The CA active site zinc
O molecule
coordinates to the imidazole sidechain of three histidine residues and to the H
2
involved in the CO hydration reaction, hence the implied target of small molecule CA
2
inhibitors is the CA active site zinc. A key focus of our novel CA targeting strategies has
been the discovery of new chemical entities as these are critical for the drug discovery
2
7
pipeline to deliver new CAꢀbased therapeutics. We recently reported two novel sulfonamide
2
8
22
compounds, the natural product Psammaplin C (1) and the saccharin glycoconjugate (2),
as potent, isozyme selective inhibitors of CAs, Figure 1A.
More recently we reported the findings of an unbiased fragment screening campaign with a
29
7
20ꢀmember fragment library against CA II. Seven hits (3-9) were identified, including
29ꢀ33
known ZBGs such as a primary sulfonamide (3) and carboxylates (4-6), Figure 1B.
More
interesting however was the finding of two 5ꢀsubstituted tetrazoles (7 and 8) and a 3ꢀ
substitutedꢀ1,2,4ꢀtriazole (9) as hits, Figure 1B. Compounds 7-9 are not classic acidic
4
ACS Paragon Plus Environment

Page 5 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
functional groups like carboxylates and sulfonamides, but instead may be considered as
acidic heterocycles. Additionally these heterocycles were not previously known as CA II
binding chemotypes, ZBGs or metal binding groups generally. The strong binding affinity of
the primary sulfonamide functional group for CA II compared to other known ZBGs is
attributed to two key hydrogen bonds with the active site threonine residue, that are in
addition to the zincꢀsulfonamide interaction, Figure 2. The specificity of the two key
hydrogen bonds ensures that the sulfonamide ZBG has very high selectivity for CA with
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
34
minimal binding to other zinc metalloenzymes.
In this manuscript we disclose the discovery of an additional fragment hit (compound 10)
against CA II that was detected from the same 720ꢀmember unbiased fragment library,
bringing the total number of hits to eight, a hit rate of 1.1%. Compound 10 is a 3ꢀ
unsubstitutedꢀ2,4ꢀoxazolidinedione. This chemotype was not previously known for zinc
binding, however compound 10 exhibited particularly strong binding affinity for CA II (KD
=
3.5 ꢀM). This affinity approaches that of the classical primary sulfonamide ZBG of CA
inhibitors such as 3 (K = 1.4 ꢀM), and far exceeds the CA II affinity for compounds 4-9
D
(K s 631 ꢀ 1280 ꢀM). To define the structural basis and boundaries for the observed strong
D
binding to CA, we investigated the detailed structureꢀactivity relationship (SAR) for the
oxazolidinedione chemotype. Our approach to generate SAR employs a combination of
chemical design and synthesis to prepare fragment analogues, followed by biophysical
characterisation of the new fragment analogues with native state mass spectrometry (MS),
surface plasmon resonance (SPR) and protein Xꢀray crystallography (XRC). Furthermore, we
have utilised MS for qualitative, semiꢀquantitative and quantitative fragment screening to
3
5
evaluate fragment SAR, known as “SAR by MS”. The discovery of the 3ꢀunsubstitutedꢀ2,4ꢀ
oxazolidinedione chemotype as a new ZBG contributes to the growing importance of small
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 62
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
molecule CA inhibitors that comprise nonꢀclassical features, as well as alternate and novel
ZBGs for drug development generally.
A.
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
B.
O
O
O
O
OH
OH
OH
O
Cl
4
5
6
KD = 905 ꢀM
KD = 1280 ꢀM
KD = 1080 ꢀM
N
N
NH2
S O
O
H
NH
N
N
O
N
3
N
H
N
O
S
N
KD = 1.4 ꢀM
N
O
N
N
N
Cl
Cl
7
8
9
KD = 631 ꢀM
K
D
= 709 ꢀM
K = 194 ꢀM
D
2
8
Figure 1. A) Unusual carbonic anhydrase (CA) inhibitors Psammaplin C (1) and saccharin
22
glycoconjugate (2) . B) Fragment hits (3-9) identified from an unbiased fragment screening
2
9
campaign of the CSIRO fragment library targeting CA II. The known or probable zinc
binding group of each fragment is circled.
6
ACS Paragon Plus Environment

Page 7 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Thr199
O
N
H
H
O
O
Glu106
O
O
R
H
S
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
N
O
Zn2+
His119
His96
His94
Figure 2. Common binding pose observed for >200 primary sulfonamide:CA II complexes
36
reported in the Protein Data Bank.
Results and Discussion
1ꢀPhenylꢀ2,4ꢀoxazolidinedione 10 was identified and validated as a binder for hCA II by
fragment screening of a nonꢀbiased synthetic fragment library using orthogonal fragment
screening techniques: SPR and native state mass spectrometry, Figure 3A. The binding
affinity (K ) of 10 for hCA II is 3.5 ꢁM (ligand efficiency = 0.57), as determined by SPR.
D
D
This affinity is similar to that of the sulfonamide fragment 3 (K = 1.4 ꢁM, ligand efficiency
= 0.73) but 2ꢀ3 orders of magnitude stronger than that for the nonꢀsulfonamide fragments 4-9
29, 36a
(
K
D
range 631 ꢀ 1280 ꢀM).
We were keen to investigate the structural basis of the strong
binding and ligand efficiency of 10 to hCA II and to establish if the 3ꢀunsubstitutedꢀ2,4ꢀ
oxazolidinedione heterocycle is a viable starting point for the development of a new class of
CA inhibitor to rival the classic primary sulfonamide chemotype, wherein the medicinal
chemistry pedigree has already spanned several decades. Computed physicochemical
descriptors, absorption, distribution, metabolism and excretion parameters, pharmacokinetic
properties, drugꢀlike nature and medicinal chemistry friendliness of benzene sulfonamide and
compound 10 indicate that 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinedione is a developable fragment
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 62
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
3
6b
for drug discovery (supporting information, Figure S20 and S21).
We determined the
protein Xꢀray crystal structure of fragment 10 bound to hCA II, Figure 3B. The fragment
bound via an interaction of the acidic ring nitrogen with the hCA II active site zinc. The pK
value for the ꢀNHꢀ of 10 (pK of 5.5) together with the pK values of a series of phenyl and
a
a
a
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
benzyl substituted five membered heterocycles predicted to be acidic bioisosteres were
measured more than 25 years ago to acquire data for the calculated Log P (cLog P) database,
37
Pomona College MedChem CLOGP. The hCA II:10 structure is consistent with the
37
prediction that 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinedione is an acidic bioisostere. In addition to
the zinc binding interaction there are two hydrogen bonds between the fragment heteroatoms
and hCA II. A hydrogen bond (3.0 Å) is formed between the ring –O– of 10 with the Thr199
backbone nitrogen, while a second hydrogen bond is formed with the ring –N– of 10 and
Thr199 side chain hydroxyl (3.2 Å). These interactions are akin to the primary sulfonamide
binding pose, and likely account for the comparable binding affinity to sulfonamides and the
increased binding affinity over fragments 4-9, Figure 4. Compound 10 is a very efficient
fragment as each of its four heteroatoms contribute to hCA II binding – either by direct
interactions or by imparting increased acidity on the imide nitrogen. A related compound, 1ꢀ
(
2′,4′ꢀdimethoxyphenyl)ꢀ2,4ꢀoxazolidinedione 11, was obtained from the CSIRO compound
collection. Compound 11 has a comparable K (6.1 ꢁM) to 10 and bound to hCA II similarly
D
via interaction with zinc, and two Hꢀbonds to Thr199, Figure 3B.
We next searched the PDB against the 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinedione fragment and
identified only one protein:ligand crystal structure where the ligand comprises this fragment,
the
ligand
is
(5R)ꢀ5ꢀ(3ꢀ{[3ꢀ(5ꢀmethoxybenzisoxazolꢀ3ꢀyl)benzimidazolꢀ1ꢀyl]
methyl}phenyl)ꢀ5ꢀmethyloxazolidinedione (12, Figure 3A) in complex with the peroxisome
38
proliferatorꢀactivated receptor γ subtype (γPPAR). γPPAR is involved in regulating glucose
8
ACS Paragon Plus Environment

Page 9 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
metabolism and insulin sensitivity, and γPPAR modulation has potential for development of
38
treatment of type 2 diabetes mellitus. The oxazolidinedione moiety of 12 was incorporated
as a replacement for a phenoxycarboxylic acid group of a lead compound series. The
structure of the γPPAR:12 complex (PDB ID 3TY0) indicates there are hydrogen bond
interactions of the oxazolidinedione to the Ser342 backbone nitrogen and the His266 side
chain nitrogen of γPPAR. Notably γPPAR is not a metalloprotein hence there is no
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
38
opportunity for a metal binding interaction in this protein. The finding that 10 and 11 bind
to the hCA II active site zinc, coupled with the lack of exemplar 3ꢀunsubstitutedꢀ2,4ꢀ
oxazolidinediones as a ZBG in the PDB, establishes the 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinedione
as a new zinc binding chemotype.
A.
B.
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 62
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 3. (A) 3ꢀUnsubstitutedꢀ2,4ꢀoxazolidinediones 10, 11 (this study) and γPPAR inhibitor
3
8
1
2. (B) Crystal structures of 10 and 11 with carbonic anhydrase II (CA II). In panel A, 10 is
shown in the crystal structure with the zinc atom represented by a grey sphere and active site
residues labelled (His94, His96, Leu198, Thr199, Thr200; His119 is underneath the zinc
atom). Distances from the protein to the compound are shown in Angstroms. In panel B, 11 is
shown in approximately the same orientation as 10, with the same active site residues
labelled as in panel A.
A.
B.
H
N
NH
OH
O
N
O
O
O
S
O
O
NH
OH
O
Thr199
Thr199
Zn2+
Zn2+
His
His
119
119
His94
His94
His9
His
96
6
Figure 4. Binding interactions of the classic primary sulfonamide chemotype (A) with CA II
are emulated by the oxazolidindione chemotype (B), discovered by fragment screening.
10
ACS Paragon Plus Environment

Page 11 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Compound design to establish SAR
To ascertain the structureꢀactivity relationships (SAR) for binding of the oxazolidinedione
fragment to CAs, a series of structural modifications were made to the heteroatoms and to the
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
5
ꢀsubstituent with the acidic cyclic imide NH group retained in all compounds. The SAR
focussed library comprised lead compounds 10 and 11 (Figure 3) and new compounds, 13 –
9, Figure 5. Specifically, the replacement of the ring heteroatom from –O–
2
(
oxazolidinediones, 10 and 11) to either a –S– (thiazolidinediones, 13, 16 and 17) or–NH–
hydantoins, 15, 18, 19, 20 and 21) was examined to determine the effect of both hydrogen
(
bond acceptor strength (–O– versus –S–) and modification of a hydrogen bond acceptor to a
hydrogen bond donor (–O– and –S– versus –NH–), noting the key role of the ring oxygen in
the 10:CA II and 11:CA II crystal structures as a hydrogen bond acceptor (2.9ꢀ3.0 Å to
Thr199N). We also modified both the ring heteroatom from –O– to –NH– (hydrogen bond
acceptor to hydrogen bond donor) together with one of the carbonyls to a thiocarbonyl
(thiohydantoins, 26 and 27). Depending on ease of synthesis, these heteroatom changes were
combined with modification of the 5ꢀsubstituent (phenyl, benzyl or benzylidene derivatives)
with or without added methoxy electron withdrawing groups. Additionally, the replacement
of both the –O– heteroatom and the 4ꢀcarbonyl oxygen with sulfur atoms denotes the 2ꢀ
thioxoꢀ1,3ꢀoxazolidinꢀ4ꢀone (14) and rhodanine heterocycle (rhodanines, 22, 23, 24 and 25),
which have further altered hydrogen bonding capacity compared to 10 and 11. To the best of
our knowledge none of these chemotypes have been tested for binding to or inhibition of
CAs. Finally, we included two FDA approved fragmentꢀsized oxazolidinediones, the
anticonvulsant trimethadione 28 and its demethylated metabolite dimethadione 29, Figure 5,
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 62
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
in this focussed library. The final assembled library comprised 19 compounds (10, 11, 13 –
29) of molecular weight <300 Da.
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 5. Focussed fragment library to establish SAR around the oxazolidinedione ZBG
chemotype (10 and 11) for CA affinity.
Compound Screening
12
ACS Paragon Plus Environment

Page 13 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Screening of the focussed fragment library 13–29 and lead hit fragments 10 and 11, was
carried out first using native state mass spectrometry, followed by SPR for validation. Xꢀray
crystallography was employed to provide a detailed analysis of proteinꢀfragment interactions
for validated hits
.
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
Nanoelectrospray ionisation (nanoESI) mass spectra were acquired with test samples
comprising CA II with one equivalent of added compound (10, 11, 13 – 29). Nine compounds
were observed bound to CA II ꢀ 10, 11, 13, 14, 17, 22, 24, 25 and 27. The observation of
[
protein + fragment] complex in the nanoESI mass spectrum enables qualitative classification
of a fragment as a hit (or “binder”) or nonꢀhit (or “nonꢀbinder”), with arbitrary thresholds
possible to further delineate fragment binding as strong, medium or weak. Native state mass
spectrometry has been applied to the study of noncovalent proteinꢀligand interactions for the
39, 40
measurement of K
D
values,
there are however few examples of mass spectrometry for
29, 41ꢀ45
fragment screening,
and mass spectrometry is not yet fully evaluated as a primary
fragment screening method. To incorporate quantification to the nanoESI method we
introduce here the concept of mass spectrometry fragment binding (FBMS), specifically we
assess the potential of the FBMS metric to quickly and accurately quantify the relative binding
strength from mass spectrometry data. FBMS is expressed as a percentage and is determined
from the fraction of the intensity of the [protein + fragment] peaks relative to total protein
peaks, Equation 1, where I[ꢀ:ꢁ] and I[ꢂ] are the mass spectrometric peak intensities of the
[
protein + fragment] complexes and free/acetate bound protein, respectively, in the ESI mass
spectrum at a specified protein and fragment concentration. FBMS values for CA II and
fragment, when equimolar amounts of protein and fragment (7.5
in Table 1 (columns 3 and 4).
ꢀM each) are analysed, are
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 62
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
ꢅ[ꢆ:ꢇ]
FBꢃꢄ = ꢅ
ꢊ× ꢊ100%ꢊ
Equation 1.
[ꢆ:ꢇ]
ꢈꢅ[ꢉ]
Next K values were determined for eight of the hit fragments (insufficient sample of 14
D
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
available at time of experiment). Experiments were performed with CA II (14.5 ꢁM) titrated
against at least five different fragment concentrations ([F] range 0.5 ꢁM ꢀ 120 ꢁM) until
either full complexation was observed (i.e. I[••] = 0) or the fragment concentration reached
1
20 ꢁM. The FBMS was calculated for each fragment concentration and values plotted against
fragment concentration [F]. The K was calculated upon curve fitting this plot based on Hill
D
46
slope analysis, Figure 6.
observed charge states (+9, +10 and +11) were utilised for both FBMS and K
Table 1, columns 3 and 5). Additionally for comparison, FBMS and K calculations were
performed using only the peak intensities for the +10 charge state, (Table 1, columns 4 and
). Representative mass spectra showing the +10 charge state acquired for the CA II + 24
titration (K = 16.7 ꢁM) are presented in Figure7. The single charge state (+10 only)
calculated values are in excellent agreement with calculated values using all charge states
+9, +10 and +11), indicating that the single charge state calculation of FBMS and K will be
K
D
s ranged from 2.3ꢀ50.7 ꢁM, (Table 1, columns 5 and 6). All
D
calculations
(
D
6
D
(
D
sufficient for future fragment screening campaigns using nanoESIꢀMS. FBMS values provide a
useful rank order of binding strength compared to K calculations determined from more
D
laborious and higher sample consuming titration experiments, the overall trend is consistent.
14
ACS Paragon Plus Environment

Page 15 of 62
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 6. Saturation curves of the nanoESIꢀMS titration of eight hit fragments with CA II
(14.5 ꢁM). FBMS, as calculated at each fragment concentration using all observed charge
states (+9, +10 and +11), is plotted against total fragment concentration [F] with curve fitting
4
6
based on Hill slope analysis used to determine KD.
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 62
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
1
0+
[
CA II]
1
0+
[CA II+24]
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 7. Representative positive ion mode nanoESI mass spectra (+10 charge state) of 14.4
M hCA II (pH 7.4 10 mM NH OAc) titrated with fragment 24 to determine fragment K by
ꢁ
4
D
nanoESIꢀMS: (a) 1.875 ꢁM of 24; (b) 7.5 ꢁM of 24; (c) 15 ꢁM of 24; and (d) 30 ꢁM of 24.
SPR was used as a secondary orthogonal fragment screening method for compounds 10, 11
and 13–29, Table 1 (column 2). Binding sensorgrams for compounds interacting with the
immobilised CA II are provided in Supporting Information. SPR analysis identified the same
nine compounds that bound to CA II as identified by nanoESIꢀMS, with K
3.4ꢀ65.5 ꢁM (excluding 22), Table 1 (column 2). The K values calculated by the nanoESIꢀ
MS experiments are in good agreement (K range 2.3ꢀ50.7 ꢁM) with the values measured in
values of fragment 22 (nanoESIꢀMS K = 35.8 ꢁM,
D
s that ranged from
D
D
solution by SPR with two exceptions. K
D
D
16
ACS Paragon Plus Environment

Page 17 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
D D
SPR K = 355 ꢁM) differ by a factor of ten, however the K for 22 could not be reliably
determined by SPR due to solubility issues of this compound in the SPR buffer and this issue
is likely responsible for the discrepancy observed. Compound 23 was not detected by
nanoESIꢀMS however was detected by SPR. Notably, 23 had the weakest binding of the SPR
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
hits (K ~ 1000 ꢁM), much weaker than all other hit fragments. Nonꢀbinding compounds
D
were fully consistent using the orthogonal methods. The rank order of K
methods followed a similar trend: nanoESIꢀMS K values, low to high: 25, 10, 11, 17, 24, 13,
2, 27/27 ; SPR K values, low to high: 25, 10, 11, 24, 17, 14, 13, 27/27 , 22, 23 (for details
on structure of 27/27 see Table 1 footnote). This is consistent with our earlier findings where
D
s by the two
D
2
′
D
′
′
we also found very good overlap with hits identified using SPR (as the primary screen) and
29
nanoESI MS. The finding that nanoESIꢀMS and SPR provide agreement of both the
D
magnitude and trend of K values suggests that the straightforward metric FBMS, with
equimolar protein and fragment concentrations, offers the opportunity for rapid, semiꢀ
quantitative assessment of fragment binding strengths. If nanoESIꢀMS is selected as a
primary screen we propose that MS has at least equal potential to other screening methods in
use to accelerate and inform the initial steps of FBDD. As discussed by us earlier, limited
overlap among different fragment screening approaches has been reported by Klebe and
2
9, 46a
colleagues.
However in their study SPR was not included. Their study did assess native
29, 46a
state ESIꢀMS, however using very different conditions to those employed herein.
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 62
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
Table 1. Screening results for 10, 11 and 13–29 using native state mass spectrometry, SPR
and protein Xꢀray crystallography.
Compound
No.
SPR
ESI-MS
ESI-MS
ESI-MS
ESI-MS
Protein
b
c
b
c
d
K (
D
ꢀ
M) FBMS (%)
FBMS (%)
K
D
(
ꢀM)
K
D
(
ꢀM)
XRC
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1
1
1
1
1
1
1
1
0
1
3
4
5
6
7
8
9
3.5
6.1
87.7
88.1
91.5
51.1
SB
79.8
90.2
93.7
53.1
NSB
NSB
67.3
NSB
NSB
NSB
NSB
41.5
NSB
85.5
74.5
NSB
23.1
NSB
NSB
2.6
3.6
3.0
3.5
Y
Y
32.9
28.6
17.4
18.4
n.d.
n.d.
n.d.
16.0
n.d.
n.d.
n.d.
n.d.
36.2
n.d.
19.0
2.7
Y
e
n.d.
Y
a
NSB
NSB
26.2
NSB
NSB
NSB
NSB
n.d.
n.d.
13.9
n.d.
n.d.
n.d.
n.d.
35.8
n.d.
16.7
2.3
n.d.
n.d.
Y
NSB
63.9
NSB
NSB
NSB
NSB
46.3
NSB
84.4
68.7
NSB
14.6
NSB
NSB
n.d.
n.d.
n.d.
n.d.
N
20
2
2
1
2
f
355
23
1000
7.7
N
2
2
4
5
Y
3.4
Y
26
NSB
65.5
NSB
n.d.
50.7
n.d.
n.d.
n.d.
43.6
n.d.
n.d.
n.d.
h
g
2
7/27
′
Y
2
2
8
9
n.d.
n.d.
a
NSB
18
ACS Paragon Plus Environment

Page 19 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
b
NSB = No significant binding (K
D
>2000 µM); equimolar amounts of protein and fragment (7.5 ꢀM each)
c
using all observed charge states (+9, +10 and +11); equimolar amounts of protein and fragment (7.5 ꢀM each)
d
e
f
D
using only +10 charge state; XRC – Xꢀray crystallography. n.d. – not determined; K (SPR) could not be
g
reliably determined due to solubility issues of this compound in the SPR buffer. The electron density of the
fragment observed did not correlate with the structure of 27 but instead to an impurity in the sample, later
identified as 27′.
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
Protein X-ray Crystallography
The ten hit fragments identified from the fragment screening were subjected to Xꢀray
29
crystallography following soaking with CA II crystals as described earlier (Table 1). Eight
fragments led to CA II:fragment structures observed by Xꢀray crystallography, with all of
these fragments bound via interaction of the acidic imide nitrogen with the hCA II active site
zinc (1.9ꢀ2.0 Å) and with the Thr199 side chain hydroxyl (3.1ꢀ3.3 Å) (Figure 8, Table 2 and
Supporting Information). The electron density in the structure for 27 showed that there was
an additional atom attached to Cꢀ5 of the ring with electron density consistent with either an
OH or NH
2
substituent at Cꢀ5 of 27 (Figures 8H and 9). We reꢀexamined a sample of 27 by
as
LCMS and HRMS and this showed that the sample comprised the hydrated compound 27
′
an impurity (~5%) (Figure 9). Coꢀformation of Cꢀ5 benzylidine thiohydantoins with hydrated
Cꢀ5 hydroxy/Cꢀ5 benzyl thiohydantoins has been reported, the addition of water to 26 instead
47
of reduction could lead to the formation of 27
′
similarly. As 27
′
makes two additional
potential hydrogen bonds (2.6 and 2.9 Å) with CA II compared to 27, it is plausible that the
additional binding energy enables 27 to easily outcompete 27 for the CA II binding site.
′
Structures for 10 and 11, Figure 8A and 8B, were described in the introduction. For 13 and 17
5ꢀ10ꢀfold weaker binders than 10 and 11), unlike the hydrogen bond formed with the ring –
(
O– of 10 and 11 and the Thr199 backbone nitrogen, there is no corresponding hydrogen bond
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 62
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
formed between the ring –S– of 13 and 17 and the Thr199 backbone nitrogen, Figure 8C and
8E. For 14, Figure 8D, the thiocarbonyl is too far away from Thr199 to make any reasonable
hydrogen bonds, however the ring –O– of 14 is 3.1 Å from the backbone nitrogen of Thr199,
similar to the distance seen in compounds 10 and 11 (~3.0 Å). Compounds 25 (nanoESIꢀMS
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
D D D D
K = 2.7 ꢁM, SPR K = 3.4 ꢁM) and 24 (nanoESIꢀMS K = 19 ꢁM, SPR K = 7.7 ꢁM) are
relatively strong binders, but neither the ring –S– nor the thiocarbonyl of 24 or 25 make any
hydrogen bond interactions with CA II, Figure 8F and 8G.
Figure 8. Difference density maps of fragments in the CA II binding site. (A) 10,
oxazolidinedione chemotype. (B) 11, oxazolidinedione chemotype. (C) 13, thiazolidinedione
chemotype. (D) 14 2ꢀthioxoꢀ1,3ꢀoxazolidinꢀ4ꢀone chemotype. (E) 17, thiazolidinedione
chemotype. (F) 24, rhodanine chemotype. (G) 25, rhodanine chemotype. (H) 27,
thiohydantoin chemotype. Difference density (mFoꢀDFc) for all maps is shown at a 3σ
contour level where the fragments were omitted from the model. The active site zinc atom is
represented as a grey sphere. Thr199, Thr200 and the three catalytic His residues (His94,
His96 and His119) are shown. There is a water molecule (red sphere) within hydrogen
20
ACS Paragon Plus Environment

Page 21 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
bonding distance of a methoxy substituent for compounds 11 and 14 (B and D). Additionally,
for compounds 10, 11 and 14 the ring oxygen is within hydrogen bonding distance of the
backbone nitrogen of Thr199 and this distance is shown (A, B and D). Also note 24 (F) is
modelled in two orientations for the dimethoxyphenyl group whereas the other compounds
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
with this group had one preferred orientation (11, 14, 25 and 27′).
Table 2. Interactions of compounds with CA II identified by protein Xꢀray crystallography.
Protein-fragment interaction (Å)
Compoun
PDB ID
T199OHꢀ
T199Nꢀ
T200NHꢀ
T200OHꢀ
d
ZnꢀNH
NH
O1 (S1)
OH
ꢀ
OH
ꢀ
5
TXY and
TY8
10
1.9
3.2
3.0
5
11
5TY9
5TYA
5U0D
5U0G
5U0E
5U0F
5VGY
2.0
2.0
1.9
2.0
2.0
2.0
1.9
3.1
3.1
3.3
3.1
3.2
3.2
3.1
2.9
(3.7)
3.1
ꢀ
ꢀ
ꢀ
ꢀ
13
14
17
ꢀ
ꢀ
(3.6)
(3.7)
(3.7)
ꢀ
ꢀ
ꢀ
24
ꢀ
ꢀ
25
ꢀ
ꢀ
27
′
2.9
2.6
A.
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 62
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
OMe
OMe
MeO
MeO
H
H
N
S
N
S
HO
NH
NH
O
O
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
7
27'
B.
Figure 9. The electron density observed for compound 27 with CA II is consistent with an
OH substituent at Cꢀ5 (27′). Compound 27′ was subsequently identified as a ~5% impurity in
2
7. This oxygen makes two potential hydrogen bonds with the Thr200 residue: one at 2.6 Å
to the sidechain hydroxyl and another at 2.9 Å to the backbone nitrogen (shown in figure).
The two compounds that failed to provide electron density in the CA II protein Xꢀray crystal
structures were 22 and 23. As discussed above, 23 was the weakest binding of the ten SPR
D
hits (K ~ 1000 ꢁM) and was not detected by nanoESIꢀMS. Compound 22 provided the only
significant discrepancy in the K values determined by nanoESIꢀMS and SPR, however this
D
compound had solubility issues at the higher concentration required for SPR analysis and this
may have also affected its ability to be seen in the crystallographic studies.
Structure-Activity Relationships
22
ACS Paragon Plus Environment

Page 23 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The strong binding properties of 10 and 11 shown previously were confirmed in the present
D D D D
study (SPR K = 3.5 ꢁM, nanoESIꢀMS K = 2.6 ꢁM and SPR K = 6.1 ꢁM, MS K = 3.6 ꢁM,
respectively). Compounds 10 and 11 have similar affinity, indicating that the two methoxy
groups of the dimethoxyphenyl moiety of 11 do not significantly impact on binding even
though they introduce steric bulk on to the phenyl ring (cf. 10) and are electron withdrawing.
Replacement of the ring –O– of 10 with –S– gave the corresponding thiazolidinedione 13,
this change resulted in a ~10ꢀfold reduction in CA II affinity. The larger size of sulfur relative
to oxygen alters the orientation of the ring in the active site; the ring –S– is positioned 3.7 Å
from the backbone nitrogen of Thr199, whereas in compound 10 the corresponding distance
from the ring –O– is 3.0 Å, Table 2. Replacement of the exocyclic carbonyl of 11 with a
thiocarbonyl gave 14, this change resulted in a ~10ꢀfold reduction on CA II affinity. While
the ring oxygen of 14 is 3.1 Å from the backbone nitrogen of Thr199, similar to the
corresponding interaction of compounds 10 and 11, the thiocarbonyl of 14 is too far away
from Thr199 to make any reasonable hydrogen bonds. The dimethoxybenzyl group of 17 was
tolerated with one of the methoxy groups making a potential hydrogen bond with the
sidechain hydroxyl of Thr200 (3.0 Å), however the more rigid dimethoxybenzylidine group
in 16 resulted in loss of binding to CA II. Similarly 26 was a nonbinder, while 27′ was
observed (Figure 9). The introduction of the –NH– in the ring of hydantoins (15, 18 ꢀ 21)
abolished binding completely. Thus, thiohydantoin and hydantoin binding was lost
irrespective of most of the 5ꢀsubstituents, the exception being 27′ with the added Cꢀ5
hydroxyl. This SAR suggests that a hydrogen bond acceptor in the ring 1ꢀposition (–O–)
provides the additional binding affinity observed for 10 and 11. Compounds 24 (5ꢀbenzyl
substituent) and 25 (5ꢀdimethoxybenzyl substituent) were strong binders, although the
electron density shows that the –S– in the ring is too far away from Thr199 to make a
hydrogen bond (3.7 Å) compared to the –O– in the ring of 10 and 11. Compounds 22 (5ꢀ
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
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 62
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
benzylidine substituent) and 23 (5ꢀdimethoxybenzylidine substituent) failed to give a crystal
structure while the reduced binding observed by nanoESIꢀMS and SPR, suggests that the
2
flexibility of the 5ꢀsubstituent may be important as connection through the sp hybridised
carbon of the benzylidine substituents hinders hCA II binding. Rhodanine is identified as a
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
48
‘PAIN’ as this chemotype leads to panꢀassay interference. Our findings indicate that CA II
is a possible offꢀtarget enzyme of 5ꢀsubstituted rhodanine compounds, but that any binding is
likely dependent on the nature of the 5ꢀsubstituent. To the best of our knowledge, CA II as an
offꢀtarget of rhodanines was not previously known. Given that CA II is a ubiquitous enzyme
this finding may provide further caution on the decision to advance rhodanine compounds in
medicinal chemistry campaigns. Benzylidine substitution at C5 of 16 and 26 was also poorly
tolerated, while the corresponding 5ꢀphenyl or 5ꢀbenzyl group was tolerated. This is
consistent with the SAR observed for the Cꢀ5 substituent of rhodanines 22 and 23. Finally,
trimethadione 28 and dimethadione 29 have no binding to hCA II.
The SAR of the tightest binding compounds can be explained primarily by the hydrogen
bonding potential of the compounds in the active site. For 10 and 11 (and to some extent, 14),
the Thr199ꢀNH interaction with heterocycle O is a typical hydrogen bond and is the strongest
interaction of the compound with the protein, other than the acidic heterocyclic nitrogen with
the zinc ion (see Figure 3). The hydrogen from the backbone nitrogen of Thr199 is in the
plane of the peptide bond and this is directly in line with the heterocycle oxygen (i.e. makes a
‘typical’ hydrogen bond interaction at this distance and angle). The heterocyclic oxygen is
closer (2.9 to 3.0 Å for 11 and 10 respectively) to the Thr199 backbone nitrogen than any
other part of the compound and the nitrogen hydrogen is offꢀcentre from the middle of the
heterocycle, suggesting that this could not form a piꢀstacking interaction. The Thr199ꢀOH
interaction with heterocycle N presents a very different case: it is 3.1 to 3.2 Å from the acidic
24
ACS Paragon Plus Environment

Page 25 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
nitrogen of the heterocycle and 3.9 to 4.0 Å from the zinc ion. Although the zinc ion is too far
away to consider it a ‘bond’, it likely orients a lone pair of electrons from the hydroxyl
oxygen, leaving the hydrogen to orient towards the acidic nitrogen, and this is within the
typical hydrogen bonding distance. The angle for this interaction is not as good as for the
Thr199ꢀNH interaction with the heterocycle –O–, but it is still within reason. As a potential
secondary interaction, one that would compete with the acidic nitrogen, the Thr199 hydroxyl
sits 3.25ꢀ3.35 Å from the closest ketone/carbonyl in the hetercycle, which is long for a
hydrogen bond, although the angle is not as acute, so the hydrogen could be better situated to
make that interaction over the interaction to the acidic nitrogen of the heterocycle. As it is
very rare to actually ‘see’ hydrogens in Xꢀray crystallography, we cannot definitively say
whether the distance or angle predominates in this case (whether the hydrogen prefers to
orient towards the acidic nitrogen or to the carbonyl of the heterocycle), but this will be a
weak interaction in either case.
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
Given that the 3ꢀunsubstitutedꢀ2,4ꢀoxazolidinedione fragment is not known as a ZBG in the
PDB, we next searched the PDB for protein:ligand crystal structures comprising the other hit
fragments (as 3ꢀunsubstituted heterocycles) to see if they are known as metal binding groups.
The 2ꢀthioxoꢀ1,3ꢀoxazolidinꢀ4ꢀone of 14 is not in the PDB. One protein:ligand crystal
structure where the ligand comprises a 3ꢀunsubstituted thiohydantoin as in 27′ is known, PDB
49
ID 1HLF, the protein is glycogen phosphorylase B, not a metalloprotein. . Three
protein:rhodanine structures, with a 3ꢀunsubstituted rhodanine, as in 24 and 25, are known,
all with the bacterial enzyme MurD ligase, also not a metalloprotein (PDB IDs 2WJP, 2Y68
5
0ꢀ52
and 2Y1O).
There were more than 20 protein:ligand structures in the PDB for the 3ꢀ
unsubstituted 1,3ꢀthiazolidineꢀ2,4ꢀdiones, none of which have the fragment bound to a metal
53ꢀ61
ion.
–
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 62
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
Interestingly, the hydantoin chemotype, although displaying no CA II binding in this study, is
known as a ZBG. Xꢀray crystal structures of 5ꢀsubstituted hydantoins with the zinc
metalloenzyme, have the acidic imide nitrogen of the hydantoin coordinated to a catalytic
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
62ꢀ64
65
zinc.
Further structures with matrix metalloproteins (MMP) are reported showing a zinc
interaction with the ligand. The zinc in these enzymes is coordinated to three protein histidine
residues, the same coordination as found in hCA II. The significance of this observation is
that a zinc binding group does not correspond to polypharmacology towards zinc
metalloenzymes and ligand specificity is achievable when targeting metalloenzymes.
Collectively the SAR generated in this study confirms the importance of deprotonation of the
acidic cyclic imide NH group for the oxazolidinedione and other heterocycles to act as a
ZBG. The SAR is however more informative, it demonstrates that the zinc interaction alone
is not sufficient for good CA II binding but rather that a interdependent combination of the
acidic imide, a ring heteroatom (with O > S > NH, where O makes a hydrogen bond, S is
neutral and the hydrogen of the N sterically interferes) and 2ꢀcarbonyl/thiocarbonyl group in
parallel with the introduction of a flexible 5ꢀsubstituent contribute to the strength and
specificity of CA II binding. We have identified 3ꢀunsubstituted 2,4ꢀoxazolidinedione, 2ꢀ
thioxoꢀ1,3ꢀoxazolidinꢀ4ꢀone, thiohydantoin and 1,3ꢀthiazolidineꢀ2,4ꢀdione fragments as
ZBGs which were not previously known in the PDB. All have potential for CA targeting
strategies with new chemical entities. In contrast, the hydantoin chemotype was previously
known as a ZBG yet displayed no CA II binding even though the zinc of the hydantoin
binding enzymes is coordinated to three protein histidine residues, the same coordination as
found in hCA II. Together the SAR of this study with the findings in the PDB indicate that
strong and selective targeting of zinc metalloenzymes such as CA is possible. Notably, it is
26
ACS Paragon Plus Environment

Page 27 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
not sufficient to have only the interaction with zinc but a combination of zinc binding with
additional interactions to the protein residues is required, which then gives the ability to build
in specificity.
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
Chemical Synthesis
The synthetic route to 2,4ꢀoxazolidinediones 10 and 11, 2,4ꢀthiazolidinediones 13 and 2ꢀ
thioxoꢀ1,3ꢀoxazolidinꢀ4ꢀone 14 is presented in Scheme 1. Palladiumꢀcatalysed Suzukiꢀ
Miyaura coupling reaction of the phenyl boronic acids 1a-b with ethyl glyoxylate generated
66
the precursor mandelic acid ethyl esters 2a-b in good yield. Condensation of 2a-b with urea
or thiourea in the presence of NaOEt (21 wt.% in ethanol), followed by acidification with
67
HCl (1.0 N), provided the target compounds.
Scheme 1. Synthesis of oxazolidindione (10 and 11), thiazolidinedione (13) and 2ꢀthioxoꢀ1,3ꢀ
oxazolidinꢀ4ꢀone (14).
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 62
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
Reagents and conditions: (i) ethyl glyoxylate, Pd
2
(dba)
3
.CHCl
3
,
2ꢀ(diꢀtertꢀ
2 3
butylphosphino)biphenyl, Cs CO , toluene 80 °C, 4ꢀ5 h; (ii) urea, 21 wt. % NaOEt in EtOH,
dry EtOH, 0 °Cꢀrt, 30 min, then reflux 3.5 h; (iii) thiourea, 21 wt. % NaOEt in EtOH, dry
EtOH, 0 °Cꢀrt 30 min, then reflux 3.5 h.
Compounds 18 and 19 were synthesised by acid mediated coupling of hydantoin with
benzaldehydes 3a-b followed by hydrolysis, Scheme 2. Reduction of the benzylidine double
bond proceeded with Pdꢀcatalysed hydrogenation to provide 20 and 21 in good yield, Scheme
68
2
. Compound 26 was prepared similarly to 19 from 3b and thiohydantoin, Scheme 2.
Attempts to reduce the benzylidine double bond of 26 with Pd/C or Pd(OH)
hydrogenation, transfer hydrogenation or 2.0 M LiBH in THF resulted in no reaction. The
2
catalysed
68
4
28
ACS Paragon Plus Environment

Page 29 of 62
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
reduction of 26 was however achieved with ZnꢀAcOH, affording benzyl thiohydantoin 27 in
69
65% yield (Scheme 2).
Scheme 2. Synthesis of hydantoin (18-21) and thiohydantoin (26 and 27) fragments.
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
Reagents and conditions: (i) (a) hydantoin, anhyd. NaOAc, AcOH, 170 °C, 3 h; (b) H O, rt,
2
overnight; (ii) 10% Pd/C, EtOH, H
2
, rt, 5 h; (iii) (a) 2ꢀthiohydantoin, anhyd. NaOAc, AcOH,
1
70 °C, 3 h; (b) H O, rt, overnight; (iv) Zn, AcOH, reflux, overnight.
2
Benzaldehydes 3a and 3b were also the core reagent for the synthesis of 16-17 and 22-25.
Knoevenagel type condensation of 3a and 3b with rhodanine afforded benzylidine rhodanines
22 and 23, respectively, while reaction of 3b and 2,4ꢀthiazolidinedione afforded benzylidene
70
thiazolidinedione 16, Scheme 3. Regiospecific reduction of 22, 23 and 16 with 2.0 M LiBH
4
in THF provided the corresponding benzyl compounds 24, 25 and 17, respectively (Scheme
71
3).
Scheme 3: Synthesis of 2,4ꢀthiazolidinedione (16-17) and rhodanine (22-25) compounds.
29
ACS Paragon Plus Environment