FragLites - Minimal, Halogenated Fragments Displaying Pharmacophore Doublets. An Efficient Approach to Druggability Assessment and Hit Generation

Article abstract of DOI:10.1021/acs.jmedchem.9b00304

Identifying ligand binding sites on proteins is a critical step in target-based drug discovery. Current approaches to this require resource-intensive screening of large libraries of lead-like or fragment molecules. Here, we describe an efficient and effective experimental approach to mapping interaction sites using a set of halogenated compounds expressing paired hydrogen-bonding motifs, termed FragLites. The FragLites identify productive drug-like interactions, which are identified sensitively and unambiguously by X-ray crystallography, exploiting the anomalous scattering of the halogen substituent. This mapping of protein interaction surfaces provides an assessment of druggability and can identify efficient start points for the de novo design of hit molecules incorporating the interacting motifs. The approach is illustrated by mapping cyclin-dependent kinase 2, which successfully identifies orthosteric and allosteric sites. The hits were rapidly elaborated to develop efficient lead-like molecules. Hence, the approach provides a new method of identifying ligand sites, assessing tractability and discovering new leads.

Full text of DOI:10.1021/acs.jmedchem.9b00304

Subscriber access provided by ECU Libraries
Article
FragLites - minimal, halogenated fragments displaying pharmacophore
doublets. An efficient approach to druggability assessment and hit generation.
Daniel Wood, José Daniel Lopez-Fernandez, Leanne Knight, Islam Al-
Khawaldeh, Conghao Gai, Shengyin Lin, Mathew P. Martin, Duncan Miller,
Celine Cano, Jane Endicott, Ian Hardcastle, Martin Noble, and Michael J. Waring
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00304 • Publication Date (Web): 12 Mar 2019
Downloaded from http://pubs.acs.org on March 13, 2019
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 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 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.
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 13
Journal of Medicinal Chemistry
1
2
3
4
5
FragLites - minimal, halogenated fragments displaying pharmacophore
doublets. An efficient approach to druggability assessment and hit generation.
6
7
8
9
Daniel J. Wood1, J. Daniel Lopez-Fernandez2, Leanne E. Knight2, Islam Al-Khawaldeh2, Conghao Gai2,
Shengyin Lin2, Mathew P. Martin1, Duncan C. Miller2, Céline Cano2, Jane A. Endicott1, Ian R.
Hardcastle2, Martin E. M. Noble1* and Michael J. Waring2*.
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
1Northern Institute for Cancer Research, Paul O’Gorman Building, Medical School, Framlington Place, Newcastle
University, Newcastle upon Tyne, NE2 4HH, U.K.; 2Northern Institute for Cancer Research, Chemistry, School of Natural
and Environmental Sciences, Bedson Building, Newcastle University, Newcastle upon Tyne, NE1 7RU, U.K.
prior to commencing investigations, a significant amount of effort
is wasted in attempting to identify leads against intractable targets.
ABSTRACT: Identifying ligand binding sites on proteins is
a critical step in target-based drug discovery. Current
approaches to this require resource intensive screening of
large libraries of lead-like or fragment molecules. Here we
describe an efficient and effective experimental approach to
mapping interaction sites using a set of halogenated
compounds expressing paired hydrogen-bonding motifs,
termed FragLites. The FragLites identify productive drug-like
interactions, which are identified sensitively and
unambiguously by X-ray crystallography, exploiting the
anomalous scattering of the halogen substituent. This
mapping of protein interaction surfaces provides an
assessment of druggability and can identify efficient start
points for the de novo design of hit molecules incorporating
the interacting motifs. The approach is illustrated by mapping
cyclin-dependent kinase 2, which successfully identifies
orthosteric and allosteric sites. The hits were rapidly
elaborated to develop efficient lead-like molecules. Hence,
the approach provides a new method of identifying ligand
sites, assessing tractability and discovering new leads.
Considerable attempts to address this issue have been made
using both computational and experimental methods.
Computational methods can provide useful indications of the
presence of binding sites10 but have, in general, not been accurate
enough to be used with confidence as a tractability filter.9,11
Experimental methods are generally based on screening
compounds. In such a setting, the application of screening by
NMR12,13 or X-ray14,15,16 based methods is attractive, since these
techniques can detect the binding of ligands that have relatively low
affinity for their target (Kd values up to 10 mM).17 X-ray
crystallography, in particular, provides an information rich
assessment of binding sites, revealing not only the location of the
binding site, but also the specific protein-ligand interactions.
Although crystal structures of compounds with very small ligands
are known, compounds usually need to have a threshold level of
affinity to achieve the degree of occupancy to be detected with
confidence. This typically requires compounds of a certain size
(molecular mass greater than 200 Da) as well as complementarity
of shape and pharmacophoric interactions.18,19 Smaller compounds
are likely to bind with lower affinity, hence with lower occupancy,
and be hard to distinguish from ordered solvent in electron density
maps. The number of drug-like molecules within the higher mass
range is enormous,20,21 consequently libraries that sufficiently
cover the corresponding chemical space are both large (containing
thousands of compounds22) and hard to design to populate chemical
space systematically. It would be possible to cover chemical space
far more comprehensively with smaller compounds, but such
compounds may lack the affinity and shape recognisability
required for their binding to be unambiguously confirmed by X-ray
crystallography.
Introduction
The identification of ligand binding sites is a critical initial step
in assessing whether a protein target is amenable to modulation by
a chemical probe and, ultimately, a drug molecule.1 Binding of a
small molecule to a specific site preferentially occurs where the
protein offers a poorly solvated, hydrophobic pocket.2 However,
for a small molecule to be specific in its interactions (as is required
for a chemical probe3) as well as possess the necessary physical
properties to be functional in a biological environment (as is
required for a drug), it must also contain polar groups that may
contribute to solubility, modulate pharmacokinetics, and form
geometrically constrained, and hence selective, interactions with
the protein.4,5,6 This combination of requirements makes it difficult
to predict the number and location of drug-like compound binding
sites on a protein from the 3D structure of the protein alone. The
active sites of proteins, which are refined by evolution to bind to
other macromolecules, co-factors or substrates, can generally be
recognised from their shape and physicochemical properties.7
Consequently, such sites can, in some cases, be targeted in drug or
probe discovery campaigns. Allosteric sites or sites of regulatory
protein-protein interaction, by contrast, are harder to identify and
harder to exploit.8,9 Since the existence of potential binding sites,
and their propensity to bind drug-like molecules, cannot be known
A further complication that arises when experimentally assessing
target tractability is that the observation of molecules bound to a
protein pocket does not necessarily establish that pocket to be
amenable to binding a small molecule that has properties
compatible with development into a drug or chemical probe. For
this to be the case, some of the molecule’s binding affinity must
derive from polar interactions, in particular through the formation
of hydrogen bonds that are energetically more favourable than
those it might otherwise make with solvent.23 The penalty of
desolvation upon protein binding for functional groups capable to
forming such interactions can be such that they contribute little or
nothing to the overall binding energy.2 This penalty can be
minimised if more than one polar interaction can be made by
groups on a ligand that are in close proximity to one another, such
that the combined solvation sphere of the groups is reduced relative
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 13
1
2
3
4
5
to that of the two groups in isolation. Protein – ligand pairs that are
capable of making such cooperative interactions are likely to have
the combination of high affinity and relatively low lipophilicity
required for a drug molecule or chemical probe.22
6
7
Results
We hypothesised that such potential interacting sites could be
found by designing a library of molecules of minimal size and
complexity, that comprises compounds displaying a combination
of two functionalities capable of forming hydrogen bonding
interactions in proximity but with different spatial orientations,
termed pharmacophore doublets (Fig. 1). In addition to providing
potential for polar protein-ligand interactions, the doublet of polar
groups would be expected to confer a degree of aqueous solubility
to library members, allowing them to be used at high concentrations
in crystallographic and other assay conditions - a requirement
because molecules of this size are expected to have low affinity for
their targets. Presenting these doublets on an aromatic scaffold
would provide potential for lipophilic interactions and to conform
to the typical architecture of drug-like molecules. Incorporating a
heavy halogen atom (bromine or iodine) into each compound
would further allow unambiguous identification of the position and
orientation of the fragments by detection of a signal in X-ray
crystallography that arises from the anomalous dispersion of the
halogen atom.24,25,26 Confidence in this concept was provided by a
recent paper in which bromo- and iodopyrazole were shown to bind
to numerous sites on three proteins, with ligand binding occurring
at known active sites and at previously undescribed allosteric
sites.27
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
Accordingly, we designed a set of compounds that individually
display all the combinations of a hydrogen bond donor-acceptor or
acceptor-acceptor doublet with 1 to 5 bond connectivity between
them. We also incorporated some redundancy into the set,
populating the likely more productive combinations such as 1,2
donor-acceptors and 1,2 acceptor-acceptors multiple times. This
allows assessment of reproducibility and incorporates variety in the
orientation of the pharmacophore doublet with respect to the
halogen. This resulted in the selection of a set of 31 compounds
that populate all combinations of pharmacophore doublets (Fig. 1).
Where not commercially available, these molecules were
synthesised using standard chemistry (see Supporting
Information). We term these molecules “FragLites” [defined as
small (≤13 heavy atoms) compounds bearing a pharmacophore
doublet and a heavy halogen atom] due to their minimal size,
maximal simplicity and high visibility in X-ray crystallography.
Figure 1 – Design of the FragLite set and their matching to
pharmacophore doublets of donors (red) and acceptors (blue) along
with their halogen tag (green), grouped by the connectivity
(number refers to the number of bonds separating the doublet.
To test our hypothesis, each FragLite was soaked into an
exemplar protein. The well diffracting system of cyclin-dependent
kinase 2 (CDK2) was selected for this initial assessment due to its
precedent for binding small molecules at the orthosteric site,28
isolated reports of allosteric small molecule binding sites,16,29 the
well characterised allosteric binding of cyclin modulators and the
potential biological significance of the discovery of new allosteric
sites. Of the 31 FragLites that were soaked, 9 were observed to
bind to CDK2 in 6 fully distinct sites (compounds 1, 2, 4, 6, 7, 13,
14, 16 and 31, Fig. 2). These include the orthosteric site, a known
allosteric site29 and 4 previously unidentified allosteric sites.
FragLites 21, 22, 24 and 30 compromised the crystal and so no
diffraction could be collected. The remaining compounds gave rise
to no identifiable signal in normal or anomalous scattering electron
density maps (Fig.2bc). This represents a significant overall hit
rate, supporting the hypothesis that the FragLites have general
utility as universal fragments.
Unambiguous detection of the low affinity binding events was
achieved by looking for a peak more than 5 standard deviations
above the mean value in an anomalous Fourier – a Fourier map
calculated from phased X-ray diffraction data that presents features
at sites occupied by anomalously scattering atoms. Whereas all
atoms display a small amount of anomalous scattering, the
magnitude of the scattering broadly increases with atomic number,
being detectable for atoms with an atomic number greater than 8,
depending on the wavelength of X-rays used and the signal to noise
radio of the data collected.
ACS Paragon Plus Environment

Page 3 of 13
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To evaluate the enhancement in sensitivity provided by the
6Q4C, 31 6Q4D, b) The anomalous signal generated resulting from
the binding of 1 allows for unambiguous fragment hit identification
and aids orientation of the two binding modes (b & c). Anomalous
LLG map (orange surface contoured at 3σ), Weighted map from
refinement (blue mesh contoured at 1σ). d) Evaluation of the dataset
through PanDDA, shows native PanDDA event map (green
contoured at 1σ) allowing identification of both binding modes of 1.
e) Six additional binding events identified through anomalous signal,
not identified by PanDDA analysis, 6,14 and four copies of 31.
anomalous signal, we compared the hit rate achieved using
anomalous scattering with that achieved using only normal
scattering. For this purpose, we applied the Pan-Dataset Density
Analysis (PanDDA) algorithm, a state-of-the art statistical
approach for detecting weak binding events in electron density
maps.30 The sensitivity of PanDDA derives from building a model
for the variance of pixels in “apo” electron density maps so that
significant deviations can be identified. Subsequently, the
PanDDA algorithm exploits these variances to optimally subtract a
proportion of the confounding ground state, thereby revealing a
deconvoluted image of the ligand-bound state.31 PanDDA analysis
was performed on 31 ligand bound crystal datasets, bootstrapped
from a variance model made from 43 apo-CDK2 crystal datasets.
Through this analysis, PanDDA was able to identify 10 of the 16
binding events identified through anomalous scattering. Binding
events caught by anomalous scattering but missed in the PanDDA
analysis involved the unique sites observed for 6 and 14 and four
of the six binding events found for 31. In addition to highlighting
sites that were not picked up by PanDDA, the anomalous signal
from FragLite molecules contributed to the unambiguous
determination of binding pose for several fragments, including
compound 1 (Fig. 2b). While the deconvoluted PanDDA map was
suggestive of a bound molecule, additional consideration of the
associated anomalous peaks allowed a clear identification of two
alternate binding modes for the fragment.
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 orthosteric (ATP binding) pocket bound seven different
FragLites, with six (1, 2, 7, 13, 14 and 16) making one or more
hydrogen bonding contacts with the “hinge” region and 2-methoxy-
4-bromophenyl)acetic acid 31 binding in a distal pocket. It is well
established that targeting the ATP-competitive site of CDK2 with
drug-like small molecules is feasible, with most such molecules
forming hydrogen bonding interactions with the hinge region.
Hence, the observation that this region binds the FragLite
molecules with the highest frequency establishes that the mapping
exercise can successfully identify the most tractable site, and
therefore contribute to ranking and/or prioritising sites for further
investigation.
3-Bromopyrazole 1 bound in two subtly different modes, the first
with the pyrazole 2-nitrogen accepting a hydrogen bond from the
backbone of Leu83 and the other showed the 2-nitrogen accepting
a H-bond from Leu83 and the NH forming a H-bond with the
backbone carbonyl of Glu81 (Fig 3a); 3-iodopyrazole 2 showed the
same interactions (Supplementary Fig. 1). 4-Iodopyridone 7 also
formed the donor acceptor paired interaction with the NH and
carbonyl of Leu83 (Fig. 3b) and both 5-bromopyrimdine 13 and 5-
iodopyrmidine 14 formed a hydrogen bond to the NH and a CH
hydrogen bond from the pyrimidine 6-hydrogen to the backbone
carbonyl of Glu81 (Fig. 3c and Supplementary Fig. 2).
Bromonaphthyridine 16 bound to the hinge via a halogen mediated
interaction.
In addition to the direct interactions with the hinge, (2-methoxy-4-
bromophenyl)acetic acid 31 bound in the ATP-pocket, slightly
offset from the hinge, with the carboxylate interacting with the
backbone NH and sidechain hydroxyl of Thr14 and the amino
group of Lys129, the benzene ring also formed a π-cation
interaction with Lys33 (Fig. 3d).
Figure 2 – Binding sites of CDK2 revealed by FragLites. a)
Overview of binding events to CDK2 (white ribbon), depicting the 6
distinct sites identified by the FragLite library (yellow), PDB codes:
1 6Q3C, 2 6Q3B, 4 6Q3F, 6 6Q49, 7 6Q48, b 6Q4B, 14 6Q4A, 16
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 13
1
2
Br
a)
Br
Br
a)
b)
3
4
N
N
H
N
NH
2
N
H
6
O
1
4
5
6
7
8
9
c)
b)
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
c)
d)
Br
13
7
O
O
O
31
e)
d)
e)
f)
16
31
Figure 3 – Interactions of FragLites in the ATP pocket. a) 3-
bromopyrazole 1, b) 4-iodopyridone 7, PDB code 6Q48, c) 5-
bromopyrimidine 13, PDB code 6Q4B, d) (2-methoxy-4-
bromophenyl)acetic acid 31, PDB code 6Q4D, e) 4-
bromonaphthyridine 16, PDB code 6Q4C. Protein atoms involved
in direct interactions are highlighted in bold. The surface colour
indicates regions of hydrophobicity (green), polarity (red) and
solvent exposure (blue).
Figure 4 – Interactions of the FragLites in the identified allosteric
sites. a) 2-amino-4-bromopyridine 4, PDB code 6Q3F, b) 4-
bromopyridone 6, PDB code 6Q49, c-f) (2-methoxy-4-
bromophenyl)acetic acid 31, PDB code 6Q4D. Protein atoms
involved in direct interactions are highlighted in bold. The surface
colour indicates regions of hydrophobicity (green), polarity (red)
and solvent exposure (blue).
Of the five allosteric sites, 2-amino-4-bromopyridine 4 bound in
the lower region of the C-terminal lobe on the C-helix side (Fig. 2).
Binding interactions were made by the donation of hydrogen bonds
from both hydrogens of the amino group to the backbone amide of
Lys178 and Tyr180, along with π-cation interactions between the
pyridine ring and Arg126 and Arg150 (Fig. 4a). A second allosteric
site at the very bottom of the C-lobe was identified by 4-
bromopyridone 6, which formed a hydrogen bond between the
carbonyl and the backbone NH of Tyr269 as well as a π-cation
interaction with Arg245 (Fig. 4b).
Together, the mapping of the ATP pocket identifies several
productive hydrogen bonding interactions that reinforce the
conclusion that this location is the most druggable site. Moreover,
their proximity suggests that it may be possible to use the results of
these experiments to design active leads by combining the FragLite
signatures into merged fragments. For example, three hydrogen
bonding contacts at the hinge were identified by a combination of
3-bromopyrazole and 4-iodopyridone, overlaying these structures
suggested molecules that may exploit all three of these contacts
simultaneously via a contiguous array of hydrogen bond donor -
acceptor - donor, each separated by two bonds (Fig. 5a).
Accordingly, we conceived of fragments such as 2,6-
diaminopyridine 32, and 6-aminopurine 33 that would satisfy this
combined pharmacophore.
The remaining allosteric sites were all identified by (2-methoxy-4-
bromophenyl)acetic acid 31, which bound in three different
orientations at a region in the palm of N-terminal lobe, a single
location below the hinge region and a peripheral site close to the
cyclin binding region (Fig. 2). Each of these binding events were
dominated by interactions of the carboxylate group. In the N-lobe
palm, one copy formed interactions with the amino groups of
Lys34, Lys75 and Tyr77 (Fig. 4c), a second copy formed an
interaction with the amino group of Lys6 (Fig. 4d) and the third
copy formed hydrogen bonds to the backbone NH groups of Glu2,
Asn3 and Phe4 (Fig. 4e). At the bottom left of the C-lobe, making
a single hydrogen bond contact to the backbone NH of Leu101 in a
relatively solvent exposed region (Fig. 4f).
Upon soaking into CDK2, both 32 and 33 were observed to make
all three of the postulated hydrogen bond contacts in a manner that
directly overlaid with that suggested by the FragLite map (Fig. 5b
and c). Moreover, compound 33 showed increased affinity relative
to the original hit; 33 showed a Kd of 350 μM (ligand efficiency
0.44) by isothermal calorimetry (ITC). In contrast, original hits 1
and 8 showed no detectable binding by ITC at 5 mM, consistent
with the premise that such small fragments have low affinity
(Supplementary Fig. 3). The affinity of 32 could not be determined
due to poor solubility. These results show that the FragLite map
can be used to effectively design highly ligand efficient fragment-
like leads in a rational manner without extensive screening; the
ACS Paragon Plus Environment

Page 5 of 13
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ligand efficiency (LE)32 value for 36 is 0.44. Moreover, the lead
(Scheme 1b).Compound 36 was synthesised from 6-chloro-9H-
structures are reminiscent of substructures of known CDK2
purin-2-amine by bis-dihydropyran protection followed by
stepwise SNAr substitutions of the chloro-substituent with
DABCO, followed by methyl 2-(3-hydroxyphenyl)acetate and
subsequent ester hydrolysis (Scheme 1c). Compound 37 was also
prepared from bis-tetrahydropyranyl-protected 6-chloro-9H-purin-
2-amine by Suzuki-Myaura coupling with ethyl 2-(3-
(tetramethyldioxaborolanyl)phenyl)acetate followed by acidic
deprotection (Scheme 1d).
ligands, which have been optimised into potent compounds.33
a)
Leu83
O
H
N
N
H
D
H
D
H
NH
O
HN
O
Glu81
A
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
O
a)
N
N
i
N
Merge
Acceptor
Donor
N
HN
Br
CO2H
34
H N N
NH
b)
H
b)
N
O
OMe
N
O
N
OMe
ii
i
32
O
Br
CO2Me
CO2H
35
THP
HN N
c)
THP
N
H
H
N
H
N
H2N
N
H2N
N
O
N
H2N
N
i
ii
iii
N
N
N
N
N
N
N
N
+
N
N
Cl
Cl
CO2H
36
H
N
THP
HN
d)
THP
H2N
N
THP
H
N
THP
N
N
H2N
N
N
HN
iii
N
ii
N
i
N
N
H
N
N
N
N
N NH
N
I
N
N
N
c)
Cl
Cl
CO2H
CO2H
37
33
Scheme 1 – Synthesis of merged FragLites. a) 34, reagents and
conditions: i) (3-aminophenyl)acetic acid, DMF, 80 °C; b) 35,
reagents and conditions: i) methyl (4-hydroxyphenyl)acetate (1.5
eq.), 3,4,7,8-tetramethylphenanthroline (cat.), CuI (cat.), Cs2CO3,
toluene, 110 °C, 24h; ii) pyridine hydrochloride (5 eq.), 180 °C; c)
36, reagents and conditions: i) dihydropyran (10 eq.), aq. 2M HCl
(cat.), DCM, r.t., 16h; ii) DMSO, DABCO (2 eq.), r.t., 3h; iii) 1)
methyl 2-(3-hydroxyphenyl)acetate (1.33 eq.), DBU (1.1 eq.),
DMF, 65°C, 72h; 2) aq. 2M HCl, THF, r.t., 4h; d) 37, reagents and
conditions: i) dihydropyran (10 eq.), HCl (2M, cat.), DCM, r.t.,
16h; ii) ethyl 2-(3-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-
yl)phenyl)acetate (2.5 eq.), K2CO3 (2.5 eq.), Pd(PPh3)4 (0.05 eq.),
DMF:H2O (1:1), 140 °C, 20', MW hv; iii) aq. 2M HCl, THF, r.t.,
2h.
Kd= 350M
LE = 0.44
Figure 5 – Fragment merging from FragLites to produce fragment
leads. a) 3-point pharmacophore derived from two FragLites
binding to the hinge, b) 2,6-diaminopyridine 32 fulfilling the 3-
point pharmacophore and making the predicted interactions, PDB
code 6Q4F, c) 4-iodo-6-aminopurine 33 making the predicted
interactions, PDB code 6Q4E, with a measurable a Kd by ITC.
Crystallisation of expanded structures pyrimidine 34 and pyridone
35 revealed their binding mode to be as predicted by the FragLite
mapping. The pyrimidine of 34 and pyridone of 35 engaged the
hinge region with the same interactions as 13 and 7 respectively
and carboxylate 31 binding to Lys129 with good overlay with the
original hits (Fig. 6b and c). Aminopurines 36 and 37 also bound
in the manner predicted by the FragLite map, with the carboxylate
and the hinge motifs of the combined fragments making the same
interactions with the same orientations as the merged fragment 33
and original hit 31 (Figure 6d and e). Compound 36 did not show
The observation of (2-methoxy-4-bromophenyl)acetic acid 31
binding in the ATP pocket adjacent to the hinge provided further
opportunities for increasing potency by fragment growing (Fig. 6a).
The analysis of FragLite binding suggested it was the carboxylate
of 31 that made the critical interactions and its orientation
suggested that it could be linked to hinge binding Fragments via a
direct bond or a single atom linker. Accordingly, compounds 34
and 35, combining this template with the hinge binders 7 and 13 as
well as biarylether 36 and biaryl 37 incorporating the merged
aminopurine motif revealed in 31, were designed.
a
binding isotherm by ITC, however, 37 demonstrated
concentration dependent binding in ITC with a Kd of 50 μM,
equivalent to a ligand efficiency of 0.30 (Fig. 6e). These results
demonstrate that the FragLite mapping can be used to efficiently
inform fragment growing strategies to derive lead molecules with
promising binding affinity in short order without the synthesis of
large numbers of analogues.
Pyrimidine 34 was prepared in a single step SNAr displacement of
4-bromopyrimidine with (3-aminophenyl)acetic acid (Scheme 1a).
Pyridone 35 was prepared by copper catalyzed etherification of 4-
bromo-2-methoxypyridine with methyl (4-hydroxyphenyl)acetate
followed by global demethylation with pyridine hydrochloride
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 13
1
2
3
4
5
6
7
observation suggests that the FragLite mapping identifies tractable
allosteric sites for which optimisable leads can be identified. It also
suggests that the lipophilic scaffolds employed in the design of the
FragLites, in addition to the polar hydrogen bonding pairs, are
important for binding, and that the combination of the two is
essential for identifying tractable sites.
N
a)
b)
N
HN
34
COH
2
8
9
O
OH
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
c)
O
NH
Cl
35
38
d)
36
Figure 7 – Cinnamic acid fragment 38 binding to the allosteric
pocket at the bottom of the C-lobe, PDB code 6Q4K.
Discussion
The key principle of fragment screening is to screen smaller
compounds to cover more chemical space with fewer compounds
than traditional lead-like libraries. The smaller the compounds, the
fewer will be required to fully cover chemical space. FragLites
provide the smallest possible constructs that can display hydrogen-
bonding doublet on aromatic scaffolds. The hit rate for binding to
CDK2 (8/31, 26%) is considerably higher than could be expected
for the hit rate in a screen of fragments, which is typically around
5%.19 This hit rate far exceeds that observed previously for
fragment screening against CDK2 with regard to hit rate34 and the
number of identifiable sites.29 This provides evidence that
FragLites provide much more extensive coverage of chemical
space with fewer compounds than is possible with compound
libraries of larger molecules.
e)
37
Kd= 50M
LE = 0.30
Figure 6 – Linking hinge binding compounds with carboxylate
containing FragLite 31. a) Overlay of 33 (grey carbon atoms),
indicating orientation of the hinge motif, and 31 (cyan carbon
atoms) as they bind to the CDK2 ATP pocket, b) Linked
aminopyrimidine 34 binding to CDK2 and forming interactions
with the hinge and Thr14, PDB code 6Q4J, overlay shows the
alignment of 34 (grey) with the original hits 13 and 31 (cyan), c)
Linked pyridone 35 binding to CDK2, PDB code 6Q4I, with 34
(grey) overlaid with 7 and 31 (cyan), d) Linked pyridone
biarylether 36 (grey) binding with the acid making comparable
interactions to 31 (cyan), PDB code 6Q4H, e) Linked biarylether
37 (grey) binding to CDK2making comparable interactions to 31
(cyan), PDB code 6Q4G, and demonstrating concentration
dependent binding by ITC.
Our results demonstrate that FragLite mapping of a protein
provides accurate experimental identification of productive ligand
binding sites. The observation that the frequency of FragLites
binding to the region established to be most druggable (the ATP
pocket binds 6/31 compounds) and that the majority of these (5/31)
make the well characterised hydrogen bonding interactions with the
hinge gives confidence that this exercise can be used as an
experimental druggability assessment. The identification of
additional pockets, one of which we have shown, with very limited
follow up to be amendable to binding traditional fragments,
suggests that the analysis can also be used to identify novel
allosteric pockets.
In addition to the hinge interactions, to illustrate that follow up hit
development for the allosteric pockets was also possible, we
studied further analogues of (2-methoxy-4-bromophenyl)acetic
acid 31 to investigate their propensity to bind to the same allosteric
pockets. This investigation revealed that amide substituted
cinnamic acid 38 was able to bind to the site identified by 31 at the
bottom of the C-lobe (Fig. 4f). In this case, the carboxylate does
not make the same interactions in the analogue but the distal p-
chlorophenyl ring overlays with the aromatic ring of 31, in the
lipophilic region of this relatively polar pocket (Fig. 7). This
The points of interaction identified by the FragLites could be used
prospectively to define a pharmacophore for virtual screening or to
focus future library chemistry. For example, in our CDK example,
further molecules could be selected satisfying the three-point
hydrogen bonding pharmacophore identified at the CDK2 hinge
coupled with the interaction made by the carboxylate. This is
potential advantage of this approach in that it can identify
productive interactions within a tractable pocket (and presumably
ACS Paragon Plus Environment

Page 7 of 13
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
not unproductive ones) to focus future hit finding. Other
approaches, such as computational druggability calculations are
unable to do this with confidence.
Despite these results, we believe that the definition of the optimal
set of FragLites that strike the best balance between minimising the
numbers of compounds to screen and maximising the coverage of
chemical space will require further optimisation across a wide
range of proteins and larger number of compounds. Our results
suggest that some redundancy in the screening set is required.
There are differences between the binding of analogous brominated
and iodinated FragLites and not all the hits expressing the same
pharmacophore doublet bind in the same manner. Perhaps most
strikingly, it is surprising that 4-bromo-2-aminopyridine 4 was not
detected binding to the hinge region and 4-iodo-2-aminopyridine 5
did not bind at all. These compounds express the same 1,3-donor-
acceptor doublet as 4-iodopyridone 7 which made the critical pair
of H-bonds in the hinge region. It is possible that the geometry of
the bound pose required for 4 and 5 to bind to the hinge is
incompatible with accommodation of the heavy halogen atom.
Consistent with this theory, 4-chloropyridine did bind at the hinge
in the manner that was predicted by the binding mode of 7
(Supplementary Fig. 4). This suggests that the ideal set should
cover a range of differing geometries between the pharmacophore
doublets and the halogen atom. Future work will explore these
observations further to define the optimal FragLite set.
Furthermore, the results of fragment merging leading to hit-like
levels of affinity and high ligand efficiency demonstrate that the
FragLite map can be used in hit generation directly, without
recourse to further hit generation activities. We would not contend
that this approach renders further screening unnecessary but could
be applied if further screening approaches are unavailable or
resource limited. The short order in which we were able to produce
potent merged fragments demonstrates that this can be a very
efficient way to find hit 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
A further extension of this work would be to investigate why
certain residues within proteins are more able to form productive
polar interactions such as hydrogen bonds when others are not.
Why, for example, are the hydrogen bonding interactions with the
peptide backbone in the hinge region identified when other
backbone interactions are not? This particular case is well
established with almost all kinase inhibitors requiring hydrogen
bonding to the hinge region to achieve potency.35
The incorporation of the halogen atom is essential to this approach.
Without the analysis of anomalous scattering, it would not be
possible to determine the location and orientation of the hits.
Another potential advantage of the incorporation of an anomalous
scatter is their use in weaker diffracting crystal systems, to help
resolve ambiguity of binding events. When reviewing the
crystallographic binding event data, consideration was given to the
generation of halide ions from the compound upon exposure to
synchrotron radiation.36,37,38 In some instances, we observed
anomalous signal for a liberated halide ion but this was of no
consequence for the analysis. The degree of radiation damage can
be minimised by controlling the radiation energy.
Conclusion
The definition of the set of 31 FragLites and mapping of CDK2
protein to identify the most tractable, orthosteric site and
exploitable allosteric sites suggest that FragLites offer a new
approach to assessment of druggability that provides a facile and
informative hit-finding activity. Development of potent hit
compounds directly from the FragLite map shows that the
information can also be used for hit-finding in isolation. We
believe that this has the potential to be a nuanced approach to hit
finding that will supplement and inform screening of more
traditional fragment and lead-like libraries.
Future work will explore the application of this approach to a wider
range of proteins and with a wider range of designed FragLites.
This will establish the general applicability of the approach and
optimise the compound set for prospective screening. The results
of these studies will be reported in due course.
The PanDDA analysis method for identification of low occupancy
ligands identified the majority of those detected using the
anomalous signal but failed to identify the binding events of two
FragLites and several of the binding sites of FragLite 31.
Increasing ligand occupancy would require screening at much
higher concentrations at which many of the desired compounds
would be insoluble. Similarly, use of an anomalous signal may
allow for application to weakly diffracting crystal systems, which
are typically not suitable for a fragment crystallography campaign.
Experimental Section
Protein Purification
Full length human CDK2 for crystallography was cloned into a
pGEX6P1 vector (GE Healthcare) and grown in BL21(DE3)STAR
Escherichia coli (E. coli) cells.39 For ITC, full length human CDK2
was cloned into a modified pGEX3C plasmid40 containing
Saccharomyces cerevisiae Civ1 for activation loop T160
phosphorylation and grown in BL21(DE3)pLysS cells.
The FragLite mapping can be carried out very rapidly, due to the
small number of compounds required. Our selected set can be
soaked into protein crystals in a single experiment and X-ray
diffraction data on the entire set collected within a day. Critical to
the further development of this approach is the definition of the
optimum set of compounds that cover chemical space most
effectively with the minimal number of compounds. Our work
shows that additional information can be obtained, at least for
CDK2, by screening a larger compound set than that explored by
Arnold.27 It is noteworthy, however, that bromo- and iodopyrazole,
the most promiscuous compounds in Arnold’s experiment, have
also proven to be the most useful here. We contend that this is
likely due to that fact that the 1,2-donor-acceptor motif can form
highly productive complementary hydrogen bonds with a protein
target. This observation could be useful for future ligand design.
This work also suggests that the promiscuous carboxylate 31 may
offer an alternative “universal fragment” that could be used for
SAD phasing as established by Arnold for bromopyrazole.
All E. coli were then grown at 37 ⁰C until an optical density of 0.4-
0.6 before decreasing the temperature to 18 ⁰C with 0.2 mM IPTG
(Isopropyl β-D-1-thiogalactopyranoside; Sigma) induction for
BL21(DE3)STAR
and
Rosetta(DE3)pLysS
cells.
BL21(DE3)pLysS cells were grown in auto-induction media
(Formedium) and so didn’t require induction with IPTG. All other
constructs were grown in Luria Broth (LB; Invitrogen). All cells
were incubated for a further 16-20 hours at 18 C, before harvesting
via centrifugation (4,000 xg, 30 mins, 4 ⁰C) and resuspension in 50
mM HEPES, 150 mM NaCl, 2 mM DTT (dithiothreitol), pH 7.4
and EDTA (ethylenediaminetetraacetic acid)-free protease
inhibitor tablets (Roche). For lysis, cell suspensions were then
supplemented with 2 μg/mL DNase I, 10 μg/mL RNase A, 25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 13
1
2
3
4
5
6
7
8
9
μg/mL lysozyme and 5 mM MgCl2 before pulsed sonication for 5
All ITC experiments were carried out at 30 C using a Microcal
mins (20 s on, 40 s off) on ice. Cell lysates were then recovered via
centrifugation (40, 000 xg, 60 mins, 4 ⁰C) before 0.45 μm filtering
and batch binding to glutathione sepharose 4B resin (GE
Healthcare) for 60 mins, 4 ⁰C with rolling. Batch-bound lysate was
then poured down a gravity flow column, with GST-CDK2 (all
constructs are GST-fusions) eluted from the resin using 50 mM
HEPES, 150 mM NaCl, 2 mM DTT, 10 mM reduced glutathione,
pH 7.4. GST was then cleaved from CDK2 using 1:100 w/w 3C
prescission protease overnight at 4 ⁰C with rolling. Protein was then
separated from GST using gel filtration (26-60 HiLoad Superdex
75 preparative grade (GE Healthcare)) before subtractive GST
using a 5 mL GSTrap fast flow column (GE Healthcare).
PEAQ ITC instrument (Malvern) with 1× injection of 0.4 L for
0.8s followed by 19× 2 L, 4s injection with 120 s spacing between
injections, using a reference power of 6 cal/s. Data were then
analysed using Malvern Microcal PEAQ-ITC analysis software
using a one-set-of-sites model. Data analysis was also conducted at
low-c51,52 so, in accordance with the crystallographic evidence,
stoichiometry was set at n=1 set of sites, during fitting. Data was
analysed using Microcal PEAQ Evaluation software (Malvern)
using the one set of sites model to calculate the dissociation
constant and the error of the fit. Experiments were carried out twice
with two separate biological preparations of CDK2.
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
For the phosphorylated CDK2 construct (used for ITC
experiments), protein was then concentrated appropriately using a
Millipore spin concentrator and flash frozen using liquid nitrogen
for storage at -80 ⁰C
All compounds had purity ≥95% as determined by HPLC (UV
detection) and 1H NMR analysis.
Compounds 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, 30, 31, 32 are commercially
available.
For crystallisation, CDK2 was then buffer exchanged into 100 mM
4:1(Dibasic:monobasic) K2:Na phosphate, 2 mM DTT buffer and
concentrated to 10 mg/mL prior to crystalisation.41
4-Bromo-1-(2-hydroxyethyl)pyridin-2(1H)-one (28)
Crystallisation and structure determination
2-Bromoethanol (326 µL, 4.6 mmol) was added dropwise to a
mixture of potassium carbonate (1.59 g, 11.5 mmol) and 4-bromo-
2(1H)-pyridinone (0.20 g, 1.1 mmol) in anhydrous THF (10 mL)
under nitrogen. The mixture was stirred at 80 ºC for 16 h, allowed
to cool to room temperature, the precipitate filtered off and the
filtrate evaporated in vacuo. The crude residue was purified by
flash column chromatography (EtOAc:MeOH 9:1) to afford 28 (94
mg, 38%) as a white solid.
Crystals of CDK2 protein were prepared using the hanging drop
vapour diffiusion method. Protein was mixed 1:1 with precipitant
solution (6% PEG3350, 100 mM HEPES pH 7.4) over a reservoir
of 50 mM HEPES pH 7.4, 25 mM K2HPO4, 25 mM NaH2PO4.
Crystals were then soaked for 48 hours in 50 mM compound
prepared to 10 % DMSO in CDK2 precipitant solution (6%
PEG3350, 100 mM HEPES pH 7.4). Crystals were then harvested
in 30 % ethylene glycol prepared in CDK2 precipitant solution and
flash-cooled in liquid nitrogen.
Rf 0.33 (100% EtOAc); mp. 134-135 ºC; IR: νmax 3293, 3039, 2943,
1629 cm-1; 1H NMR (500 MHz, CDCl3) δ 7.23 (d, J = 7.2 Hz, 1H),
6.87 (d, J = 2.2 Hz, 1H), 6.38 (dd, J = 7.2, 2.2 Hz, 1H), 4.14-4.08
(m, 2H), 3.96 (q, J = 5.0 Hz, 2H); 13C NMR (125 MHz, CDCl3) δ
162.2, 138.4, 136.0, 123.0, 110.5, 61.5, 52.9; HRMS-ESI (m/z): [M
X-ray diffraction data were then recorded either at Diamond Light
Source, Oxford, UK or in house using a rotating copper anode X-
ray source. Structural data were determined using the CCP4 suite
of programs.42 Data were firstly processed using DIALS and XDS
+
+ H]+ C7H9BrNO2 , 217.9811; found 217.9806.
43
within Xia2.44 The structures were solved using molecular
replacement using PHASER45 and protein databank (PDB) code
1HCK46 as a search model. Refinement was carried out using
REFMAC5,47 with model building performed and compound
descriptions generated using COOT48 and AceDRG,49 respectively.
Figures were prepared using the CCP4 molecular graphics
programme CCP4mg50 and MOE (Molecular Operating
Environment, 2013.08; Chemical Computing Group ULC, 1010
Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7,
2018). PanDDA (Pan-Dataset Density Analysis)30 was undertaken
using 31 FragLite datasets against a variance model derived from
42 apo CDK2 crystal datasets. Derived PanDDA maps were
analysed using the panda.inspect command and results tabulated
appropriately.
4-Bromo-1-(2-methoxyethyl)pyridin-2(1H)-one (29)
2-Bromoethyl methyl ether (216 µL, 2.3 mmol) was added
dropwise to a mixture of potassium carbonate (0.95 g, 6.9 mmol)
and 4-bromo-2(1H)-pyridinone 6 (0.20 g, 1.1 mmol) in anhydrous
MeCN (10 mL) under nitrogen. The mixture was stirred at 80 ºC
for 16 h, allowed to cool to room temperature, the precipitate
filtered off and the filtrate evaporated in vacuo. The crude residue
was purified by flash column chromatography (EtOAc 100%) to
afford 29 (119 mg, 45%) as a brown oil.
Rf 0.25 (1:1 Petrol Ether:EtOAc); IR νmax 2891, 1625, 1111 cm-1;
1H NMR (500 MHz, CDCl3) δ 7.23 (d, J = 7.2 Hz, 1H), 6.84 (d, J
= 2.2 Hz, 1H), 6.32 (dd, J = 7.2, 2.2 Hz, 1H), 4.12-4.12 (m, 2H),
3.64-3.68 (m, 2H), 3.30 (s, 3H); 13C NMR (125 MHz, CDCl3) δ
161.3, 138.8, 135.7, 122.7, 109.7, 70.1, 59.0, 49.4; HRMS-ESI
Isothermal Titration Calorimetry (ITC)
CDK2 was buffer exchanged using a 5 mL HiTrap desalting
column (5 mL) (GE Healthcare) into ITC buffer (50 mM HEPES,
150 mM NaCl, 0.25 mM TCEP, pH 7.4) and protein concentration
was then determined using a Nanodrop 2000 at an absorbance of
280 nm. with sequence derived extinction coefficients
(http://web.expasy.org/protparam/). Protein was prepared in ITC
buffer, with 1 % DMSO in the cell at 20 M with compounds
prepared at 5 mM or 1 mM in ITC buffer at 1 % DMSO final, in
the syringe.
+
(m/z): [M + H]+ C8H11BrNO2 , 231.9968; found 231.9966.
6-Iodo-9H-purin-2-amine (33)
Hydroiodic acid (5 mL, 47% aqueous) was added to 2-amino-6-
chloropurine (0.42 g, 2.49 mmol) at 0 °C. The mixture was stirred
at 0 °C for 8 hours, treated with 5 mL of deionised water at 5 °C
for 1 hour. The precipitate was filtered and washed with water three
ACS Paragon Plus Environment

Page 9 of 13
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
times. The filter cake was transferred to a 50 mL round bottom flask
Rf 0.61 (DCM:MeOH 9:1); mp. 260-261 ºC; IR: νmax 3443, 1613,
and of 6 M NaOH aqueous solution was added to the solid. The
solution obtained was added slowly to a solution of acetic acid in
water (2:25, 27 mL). The resulting suspension was stirred at room
temperature for 2 h. The solid was collected by filtration, washed
with water (2 × 5 mL), followed by petrol (5 mL). The solid was
dried in vacuo to afford 33 (501 mg, 77%) as a yellow powder.
1216, 1182 cm-1; 1H NMR (500 MHz, (CD3)2SO) δ 11.35 (s, 1H),
7.36 (d, J = 7.3 Hz, 1H), 7.34 (d, J = 8.4 Hz, 2H), 7.07 (d, J = 8.4
Hz, 2H), 5.97 (dd, J = 7.3, 2.5 Hz, 1H), 5.31 (d, J = 2.5 Hz, 1H),
3.53 (s, 2H); 13C NMR (125 MHz, (CD3)2SO) δ 172.7, 167.7,
163.6, 151.6, 136.7, 132.8, 131.2 (×2), 120.7 (×2), 100.5, 98.6 (Cq
+
absent); HRMS-ESI (m/z): [M+H]+ C13H11NO4 , 246.0688; found
246.1010.
Rf 0.21 (DCM:MeOH 9:1); mp. 230-231 ºC; IR: νmax 3488, 3293,
3170 cm-1; 1H NMR (500 MHz, (CD3)2SO) δ 12.81 (br s, 1H), 8.05
(s, 1H), 6.67 (s, 2H); 13C NMR (125 MHz, (CD3)2SO) δ 160.1,
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
6-chloro-N,9-bis(tetrahydro-2H-pyran-2-yl)-9H-purin-2-
amine
+
151.8, 141.2, 130.5, 122.4; HRMS-ESI (m/z): [M + H]+ C5H5IN5 ,
3,4-Dihydropyran (2.7 mL, 29.5 mmol) was added to 6-chloro-2-
aminopurine (1 g, 2.95 mmol) in anhydrous DMF (50 mL)
followed by the addition of aqueous hydrochloric acid (50 μL, 5.7
M). The mixture was stirred at 60 ºC for 2 h. The mixture was
evaporated in vacuo and the crude product was purified by flash
column chromatography (silica, EtOAc:Petrol Ether 2:3) to afford
the desired product (1.55 g, 78%) as a pale yellow oil that
crystallised upon freezing.
261.9584; found 261.9582.
2-(3-(Pyrimidin-4-ylamino)phenyl)acetic acid (34)
4-Chloropyrimidine (100 mg, 0.87 mmol), 2-(3-aminophenyl)
acetic acid (100 mg, 0.66 mmol) and DMF (4 mL) were stirred at
80 ºC for 3 h. The mixture was evaporated in vacuo and the crude
product was purified by flash column chromatography (silica,
DCM:MeOH:AcOH, 8:2:0.05) to afford 34 (15 mg, 10%) as a
beige solid.
Rf 0.44 (100% EtOAc); mp. 171-174 ºC; IR: νmax 3269, 2940, 2855,
1
1716, 1614 cm-1; H NMR (500 MHz, (CD3)2SO) δ 8.42 (s, 1H),
Rf 0.47 (DCM:MeOH:AcOH 9:1:0.05); mp. 186-190 ºC; IR: νmax
8.14 (br s, 1H), 5.53 (d, J = 10.2 Hz, 1H), 5.20-5.12 (m, 1H), 4.02
(d, J = 11.1, 1H), 3.84 (d, J = 11.1 Hz, 1H), 3.46 (td, J = 2.8, 11.1
Hz, 1 H), 2.21-2.35 (m, 1H), 2.03-1.34 (m, 11H); 13C NMR (125
MHz, (CD3)2SO) δ 158.1, 153.7, 150.0, 142.3, 124.7, 81.6, 80.4,
68.2, 66.0, 30.7, 30.1, 30.0, 25.4, 24.9, 22.9; HRMS-ESI (m/z): [M
+ H]+ C15H21ClN4O2+ 338.1378; found 338.1378.
1
3045, 2827, 1649, 1580, 1516, 1474 cm-1; H NMR (500 MHz,
(CD3)2SO) δ 12.40 (br s, 1H), 9.89 (s, 1H), 8.65 (s, 1H), 8.28 (s,
1H), 7.63 (d, J = 7.8 Hz, 1H), 7.53 (s, 1H), 7.28 (t, J = 7.8 Hz, 1H),
6.95 (d, J = 7.8 Hz, 1H), 6.85 (d, J = 5.7 Hz, 1H), 3.56 (s, 2H); 13
C
NMR (125 MHz, (CD3)2SO) δ 159.9, 158.0, 155.1, 139.6, 128.6,
123.5, 120.6, 118.2, 107.2, 41.0 (Cq absent –CO2H); HRMS-ESI
(m/z): [M + H]+ C12H11N3O2+ , 230.0924; found 230.0925.
Methyl
2-(3-((9-(tetrahydro-2H-pyran-2-yl)-2-((tetrahydro-
2H-pyran-2-yl)amino)-9H-purin-6-yl)oxy)phenyl)acetate
Methyl 2-(4-((2-methoxypyridin-4-yl)oxy)phenyl)acetate
DABCO (88 mg, 0.59 mmol) was added to a mixture of 6-chloro-
N,9-bis(tetrahydro-2H-pyran-2-yl)-9H-purin-2-amine (100 mg,
0.30 mmol) and anhydrous DMSO (3 mL). The mixture was stirred
at room temperature for 3 h. The mixture was added to a stirred
suspension of DBU (49 µL, 0.33 mmol) and methyl 2-(3-
hydroxyphenyl)acetate (98 mg, 0.40 mmol) in anhydrous DMF (2
mL). The mixture was stirred at 65 ºC for 72 h. The mixture was
treated with aqueous saturated sodium bicarbonate (8 mL) and was
extracted three times with EtOAc and the combined extracts
washed three times with brine, dried over sodium sulfate, filtered,
and concentrated in vacuo to give the crude product (145 mg)
which was used as such in the next step without further purification.
4-Bromo-2-methoxypyridine (1.2 g, 6.4 mmol), 3,4,7,8-
tetramethylphenantroline (139 mg, 0.59 moles), copper (I) iodide
(54 mg, 0.28 mmol), caesium carbonate (2.9 g, 8.9 mmol) and
methyl(4-hydroxyphenyl)acetate (1.5 g, 9.0 mmol) in toluene (12
mL) were stirred in a sealed tube under nitrogen atmosphere at 120
ºC for 3 days. The mixture was allowed to cool to room
temperature, treated with water (12 mL) and extracted three times
with EtOAc. The combined organic extracts were washed with
brine, dried over sodium sulfate, filtered, and concentrated in
vacuo. The crude residue was purified by flash column
chromatography (silica; EtOAc: Petrol Ether 1:9 to 10:0) to afford
the desired product (351 mg, 10%) as a pale yellow oil.
2-(3-((2-Amino-9H-purin-6-yl)oxy)phenyl)acetic acid (36)
1
Rf 0.72 (EtOAc:Petrol Ether 9:1); H NMR (500 MHz, CDCl3) δ
8.03 (d, J = 5.9 Hz, 1H), 7.32 (d, J = 8.5 Hz, 2H), 7.04 (d, J = 8.5
Hz, 2H), 6.54 (dd, J = 5.9, 2.1 Hz, 1H), 6.18 (d, J = 2.1 Hz, 1H),
3.91 (s, 3H), 3.72 (s, 3H), 3.64 (s, 2H); 13C NMR (125 MHz,
CDCl3) δ 171.8, 167.0, 165.9, 153.3, 147.8, 131.0, 120.9, 107.4,
97.5, 53.8, 52.1, 40.5 (Cq absent).
The
crude
methyl
2-(3-((9-(tetrahydro-2H-pyran-2-yl)-2-
((tetrahydro-2H-pyran-2-yl)amino)-9H-purin-6-
yl)oxy)phenyl)acetate (142 mg, 0.3 mmol) was dissolved in THF
(5 mL) and 2M aqueous HCl (2.5 mL) and stirred at room
temperature for 4 h. The mixture was evaporated in vacuo and the
crude product was purified by reverse phase chromatography (C18
silica, water:CH3CN:NH4OH 70:30:0.1) to afford 36 (8 mg, 10%)
as a white powder.
2-(4-((2-Oxo-1,2-dihydropyridin-4-yl)oxy)phenyl)acetic acid
(35)
Pyridine hydrochloride (146 mg, 1.2 mmol) and methyl 2-(4-((2-
methoxypyridin-4-yl)oxy)phenyl)acetate (57 mg, 0.2 mmol) were
combined and the mixture was stirred at 120 ºC until full
conversion was observed by TLC. The mixture was allowed to cool
to room temperature and purified by flash column chromatography
(C18 silica, CH3CN:H2O 1:9 to 2:3) to give 35 (19 mg, 37%) as a
beige solid.
Rf 0.18 (DCM:MeOH 9:1); mp. 284-285 ºC; IR: νmax 1233, 1265,
1376, 1568 cm-1; 1H NMR (500 MHz, CD3OD) δ 7.97 (s, 1H), 7.38
(t, J = 8.1 Hz, 1H), 7.27-7.21 (m, 2H), 7.11 (dd, J = 8.1, 1.1 Hz,
1H), 3.60 (s, 2H); 13C NMR (125 MHz, (CD3)2SO) δ 173.2, 160.22,
160.18, 153.0, 138.2, 129.6, 126.5, 122.8, 120.2, 41.9 (2 Cq and 1
+
Ct absent)*; HRMS-ESI (m/z): [M + H]+ C15H14N3O3 286.0935;
found 286.0931.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 13
1
2
3
4
5
6
7
8
9
*Weak signals at 139.8 ppm and 159.7 ppm were assigned by
HMBC but are not sufficiently resolved.
were added, and the mixture was heated at 140 ºC for 18 h. After
cooling to room temperature, water (20 mL/mmol) was added and
the mixture passed through a pad of Celite. The organic layer was
extracted with Et2O (3 × 10 mL), separated and washed with water
(10 mL) and brine (2 × 10 mL) and dried over magnesium sulfate.
The solvent was concentrated in vacuo to give the crude product.
Purification by flash chromatography (silica; 0-100% DCM in
petrol ether) afforded the title compound as a white solid (25 mg,
29%).
2-(3-(2-Amino-9H-purin-6-yl)phenyl)acetic acid (37)
A
mixture of 6-chloro-N,9-bis(tetrahydro-2H-pyran-2-yl)-9H-
purin-2-amine (100 mg, 0.3 mmol), potassium carbonate (102 mg,
0.74 mmol), ethyl 2-(3-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-
yl)phenyl)acetate
(86
mg,
0.3
mmol),
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
tetrakis(triphenylphosphine)palladium(0) (17 mg, 0.015 mmol),
DMF (1.5 mL) and deionised water (1.5 mL) was stirred at 140 ºC
for 30 minutes under microwave irradiation under nitrogen. The
mixture was treated with deionised water (5 mL), extracted three
times with EtOAc and the combined extracts washed with brine,
dried over sodium sulfate, filtered, and concentrated in vacuo. The
crude residue was dissolved in THF (5 mL) and 2M aqueous HCl
(2.5 mL) and stirred at room temperature for 4 h. The mixture was
neutralised with 2M aqueous NaOH, extracted three times with
EtOAc, and the combined extracts washed with brine, dried over
sodium sulfate, filtered, and concentrated in vacuo to give a yellow
oil that was triturated with anhydrous methanol (5 mL) to afford 37
(5 mg, 6%) as a white powder.
Rf 0.14 (100% DCM); mp. 148.7-155.0 ºC; UV λmax (EtOH) 274
nm; IR νmax 3312, 2981, 2928, 2109, 1703, 1650, 1595, 1535, 1490
cm-1; 1H NMR (500 MHz, CDCl3) δ 8.35 (br s, 1H), 7.94 (s, 1H),
7.81 (d, J = 8.0 Hz, 1H), 7.64-7.59 (m, 4H), 7.44 (t, J = 8.0 Hz, 1H),
7.31-7.28 (m, 2H), 6.43 (d, J = 16.0 Hz, 1H), 4.23 (q, J = 7.0 Hz,
2H), 1.31 (t, J = 7.0 Hz, 3H); 13C NMR (125 MHz, CDCl3) δ 166.7,
165.4, 143.2, 136.4, 135.4, 135.1,131.2, 129.8, 129.4, 129.1, 128.6,
126.6, 121.7, 60.8, 14.3; HRMS-ESI (m/z): [M + H]+ C18H17O3NCl
330.0888; found 330.0891.
(E)-3-(3-((4-Chlorophenyl)carbamoyl)phenyl)acrylic acid (38)
A
mixture
of
ethyl
(E)-3-(3-((4-
Rf 0.2 (DCM:MeOH 9:1); mp. 329-330 ºC; IR: νmax 3380, 3163,
3041, 2113, 1911, 638 cm-1; 1H NMR (500 MHz, (CD3)2SO) δ 8.58
(d, J = 7.4 Hz, 1H), 8.50 (br s, 1H), 8.10 (s, 1H), 7.48 (t, J = 7.4 Hz,
1H), 7.40 (d, J = 7.4 Hz, 1H), 3.69 (s, 2H); 13C NMR (125 MHz,
(CD3)2SO) δ 174.8, 156.8, 148.6, 137.8, 135.5, 131.7, 130.6,
129.5, 41.5 (4 Cq absent)*; HRMS-ESI (m/z): [M + NH4]+
C13H12N5O2 + NH4 287.3025; found 287.1240.
chlorophenyl)carbamoyl)phenyl)acrylate (50 mg, 0.15 mmol),
LiOH.H2O, and THF:MeOH:H2O (4:2:1, 1 mL) was stirred at 0 ºC
for 2 h. The mixture was treated with 2M aqueous HCl (2 mL) and
extracted three times with EtOAc. The combined extracts were
dried over sodium sulfate, filtered, and concentrated in vacuo to
give
a white solid that was purified by flash column
chromatography (silica, MeOH: DCM 0:10 to 2:8) to afford 38
(14.4 mg, 31%) as a white powder.
*Weak signals at 123.5 ppm and 159.7 ppm were assigned by
HMBC but are not sufficiently resolved.
Rf 0.10 (DCM:MeOH 9:1); m.p. 259-260 ºC; IR νmax 3334, 2919,
1
2849, 2633, 2485, 2115, 1719, 1648, 1593, 1522 cm-1; H NMR
3-Bromo-N-(4-chlorophenyl)benzamide
(500 MHz, (CD3)2SO) δ 10.44 (1H, s), 8.23 (1H, s), 7.95 (1H, d, J
= 7.8 Hz), 7.89 (1H, d, J = 7.8 Hz), 7.81 (2H, d, J = 8.8 Hz), 7.66
(1H, d, J = 16.0 Hz), 7.57 (1H, t, J = 7.8 Hz), 7.42 (2H, d, J = 8.8
Hz), 6.67 (1H, d, J = 16.0 Hz); 13C NMR (125 MHz, (CD3)2SO) δ
167.6, 165.2, 143.9, 138.0, 135.3, 134.6, 131.4, 129.4, 129.2 128.6,
127.5, 126.9, 122.1, 120.9; HRMS-ESI (m/z): [M + H]+
C16H13O3N35Cl 302.0579, found 302.0578.
3-Bromobenzoic acid (400 mg, 2.00 mmol) was added to DMAP
(242 mg, 1.83 mmol) in anhydrous DCM (20 mL), followed by the
addition of EDC.HCl (460 mg, 2.40 mmol). After 15 min, a
solution of 4-chloroaniline (280 mg, 2.20 mmol) in anhydrous
DCM (5 mL) was added, and the mixture was stirred at rt. for 18 h.
Water (10 mL) was added to quench the reaction, the organic layer
separated and washed with 1.0M aqueous HCl (3 × 10 mL). The
combined organic layers were dried over magnesium sulfate and
the solvent concentrated in vacuo. Purification by flash
chromatography (silica; 0-40% ethyl acetate in petrol ether)
afforded the title compound as a white solid (270 mg, 87%).
Ancillary Information
The Supporting Information is available free of charge on the ACS
Publications website. Supplementary figures for additional crystal
structures of 2, 14 and 2-amino-4-trifluoromethylpyridine; ITC
traces of 1 and 8; crystallographic data table and NMR spectra of
synthesized compounds.
Rf 0.78 (33% EtOAc:Petrol); mp. 136.0-138.7 ºC; UV λmax (EtOH)
274 nm; IR νmax 3361, 2112, 1652, 1595, 1517, 1397 cm-1; 1H NMR
(500 MHz, CDCl3) δ 7.34-7.31 (m, 2H), 7.36 (d, J = 8.0 Hz, 1H),
7.59-7.56 (m, 2H), 7.67 (ddd, J = 1.0 Hz, 2.0 Hz, 8.0 Hz, 1H), 7.76
(dd, J = 1.0 Hz, 8.0 Hz, 1H), 7.83 (br s, 1H), 7.98 (dd, J = 2.0 Hz,
1H); 13C NMR (125 MHz, CDCl3) δ 164.3, 136.6, 136.1, 135.0,
130.4, 130.2, 130.0, 129.2, 125.6, 123.0, 121.6; HRMS-ESI (m/z):
[M + H]+ C13H10ONClBr 309.9627; found 309.9629.
All crystallographic results were deposited to and are accessible
from the Protein Data Bank (PDB) https://www.rcsb.org/. PDB
accession codes for all structures derived from this study are as
follows: 1 6Q3C, 2 6Q3B, 4 6Q3F, 6 6Q49, 7 6Q48, 13 6Q4B, 14
6Q4A, 16 6Q4C, 31 6Q4D, 32 6Q4F, 33 6Q4E, 34 6Q4J, 35 6Q4I,
36 6Q4H, 37 6Q4G and 38 6Q4K.
Corresponding Author
Ethyl (E)-3-(3-((4-chlorophenyl)carbamoyl)phenyl)acrylate
Michael J. Waring, email: mike.waring@ncl.ac.uk, Tel. +44 (0)
191 208 8591.
Palladium(II)
acetate
(6.8
mg,
0.026
mmol)
and
triphenylphosphine (7.9 mg, 0.026 mmol) were added to a solution
of 3-bromo-N-(4-chlorophenyl)benzamide (80 mg, 0.259 mmol) in
DMF (1.0 mL) at room temperature under nitrogen. Ethyl acrylate
(0.04 mL, 0.259 mmol) and sodium acetate (27 mg, 0.325 mmol)
Notes
The authors declare no competing financial interests.
ACKNOWLEDGMENT
ACS Paragon Plus Environment

Page 11 of 13
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
of endothiapepsin. J. Med. Chem. 2016, 59 (16), 7561–7575.
(16) Betzi, S.; Alam, R.; Martin, M.; Lubbers, D. J.; Han, H.; Jakkaraj, S.
R.; Georg, G. I.; Schönbrunn, E. Discovery of
We gratefully acknowledge financial support from Cancer
Research UK (grant reference C2115/A21421), the Medical
Research Council (grant reference MR/N009738/1), Astex
Pharmaceuticals, the JGW Paterson Foundation (studentship award
to DJW), Newcastle University (studentship award to JDL-F) and
Jordan University of Science & Technology (studentship award to
IA-K). We thank beamline staff at Diamond Light Source (Oxford)
for help in refinement of CDK2 structures.
a potential allosteric ligand binding site in CDK2. ACS Chem. Biol.
2011, 6 (5), 492–501.
(17) Hubbard, R. E.; Murray, J. B. Experiences in fragment-
based lead discovery. Methods Enzymol. 2011, 493, 509–531.
(18) Hann, M. M.; Leach, A. R.; Harper, G. Molecular complexity
and its impact on the probability of finding leads for drug discovery.
J. Chem. Inf. Comput. Sci. 2001, 41 (3), 856–864.
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
REFERENCES
(19) Leach, A. R.; Hann, M. M. Molecular complexity and fragment-
based drug discovery: ten years on. Curr. Opin. Chem. Biol. 2011,
15 (4), 489–496.
REFERENCES
(1) Hopkins, A. L.; Groom, C. R. The druggable genome. Nat. Rev. Drug
Discov. 2002, 1 (9), 727–730.
(20) Fink,
T.;
Reymond,
J.-L.
Virtual exploration
of
the chemical universe up to 11 atoms of C, N, O, F: assembly of
26.4 million structures (110.9 million stereoisomers) and analysis
for new ring systems, stereochemistry, physicochemical properties,
compound classes, and drug discovery. J. Chem. Inf. Model. 2007,
47 (2), 342–353.
(2) Homans, S. W. Water, water everywhere--except where it
matters? Drug Discov. Today 2007, 12 (13–14), 534–539.
(3) Arrowsmith, C. H.; Audia, J. E.; Austin, C.; Baell, J.; Bennett, J.;
Blagg, J.; Bountra, C.; Brennan, P. E.; Brown, P. J.; Bunnage, M. E.;
Buser-Doepner, C.; Campbell, R. M.; Carter, A. J.; Cohen, P.;
Copeland, R. A.; Cravatt, B.; Dahlin, J. L.; Dhanak, D.; Edwards, A.
M.; Frederiksen, M.; Frye, S. V; Gray, N.; Grimshaw, C. E.;
Hepworth, D.; Howe, T.; Huber, K. V. M.; Jin, J.; Knapp, S.; Kotz,
J. D.; Kruger, R. G.; Lowe, D.; Mader, M. M.; Marsden, B.; Mueller-
Fahrnow, A.; Müller, S.; O’Hagan, R. C.; Overington, J. P.; Owen,
D. R.; Rosenberg, S. H.; Roth, B.; Roth, B.; Ross, R.; Schapira, M.;
Schreiber, S. L.; Shoichet, B.; Sundström, M.; Superti-Furga, G.;
Taunton, J.; Toledo-Sherman, L.; Walpole, C.; Walters, M. A.;
Willson, T. M.; Workman, P.; Young, R. N.; Zuercher, W. J. The
promise and peril of chemical probes. Nat. Chem. Biol. 2015, 11 (8),
536–541.
(21) Blum, L. C.; Reymond, J.-L. 970 Million druglike small molecules
for virtual screening in the chemical universe database GDB-13. J.
Am. Chem. Soc. 2009, 131 (25), 8732–8733.
(22) Keserű, G. M.; Erlanson, D. A.; Ferenczy, G. G.; Hann, M. M.;
Murray,
C.
W.;
Pickett,
S.
D.
the value
Design principles
of learnings
for fragment libraries: maximizing
from pharma fragment-based drug discovery
(FBDD) programs
for use in academia. J. Med. Chem. 2016, 59 (18), 8189–8206.
(23) Hunter, C. A. Quantifying intermolecular interactions: guidelines for
the molecular recognition toolbox. Angew. Chemie Int. Ed. 2004, 43
(40), 5310–5324.
(4) Waring, M. J. Lipophilicity in drug discovery. Expert Opin. Drug
Discov. 2010, 5 (3), 235–248.
(24) Tiefenbrunn, T.; Forli, S.; Happer, M.; Gonzalez, A.; Tsai, Y.; Soltis,
M.; Elder, J. H.; Olson, A. J.; Stout, C. D. Crystallographic fragment-
based drug discovery: use of a brominated fragment library targeting
HIV protease. Chem. Biol. Drug Des. 2014, 83 (2), 141–148.
(5) Young, R. J.; Green, D. V. S.; Luscombe, C. N.; Hill, A. P. Getting
physical in drug discovery II: the impact of chromatographic
hydrophobicity measurements and aromaticity. Drug Discov. Today
2011, 16 (17–18), 822–830.
(25) Wilcken, R.; Liu, X.; Zimmermann, M. O.; Rutherford, T. J.; Fersht,
A. R.; Joerger, A. C.; Boeckler, F. M. Halogen-enriched fragment
libraries as leads for drug rescue of mutant p53. J. Am. Chem. Soc.
2012, 134 (15), 6810–6818.
(6) Kawasaki, Y.; Freire, E. Finding a better path to drug selectivity.
Drug Discov. Today 2011, 16 (21–22), 985–990.
(7) Sankararaman, S.; Sha, F.; Kirsch, J. F.; Jordan, M. I.; Sjölander, K.
Active site prediction using evolutionary and structural information.
Bioinformatics 2010, 26 (5), 617–624.
(26) Blaney, J.; Nienaber, V.; Burley, S. K. Fragment-
based Lead Discovery
and Optimization Using
X-
ray Crystallography, Computational Chemistry,
and High-
throughput Organic Synthesis. In Fragment‐based Approaches in
Drug Discovery (eds. R. Mannhold, H. Kubinyi, G. Folkers, W.
Jahnke and D. A. Erlanson) 2006; pp. 215–248, Wiley-VCH Verlag,
Weinheim, Germany.
(8) Lu, S.; Huang, W.; Zhang, J. Recent computational advances in the
identification of allosteric sites in proteins. Drug Discov. Today
2014, 19 (10), 1595–1600.
(9) Rathi, P. C.; Ludlow, R. F.; Hall, R. J.; Murray, C. W.; Mortenson,
P. N.; Verdonk, M. L. Predicting “hot” and “warm” spots
for fragment binding. J. Med. Chem. 2017, 60 (9), 4036–4046.
(27) Bauman, J. D.; Harrison, J. J. E. K.; Arnold, E. Rapid experimental
SAD phasing and hot-spot identification with halogenated
fragments. IUCrJ. 2016, 3, 51–60.
(10) Halgren, T. A. Identifying and characterizing binding sites
and assessing druggability. J. Chem. Inf. Model. 2009, 49 (2), 377–
389.
(28) Sharma, P. S.; Sharma, R.; Tyagi, R. Inhibitors of cyclin dependent
kinases: useful targets for cancer treatment. Curr. Cancer Drug
Targets 2008, 8 (1), 53–75.
(11) Ward, R. A. Using protein-ligand docking to assess the chemical
tractability of inhibiting a protein target. J. Mol. Model. 2010, 16
(12), 1833–1843.
(29) Ludlow, R. F.; Verdonk, M. L.; Saini, H. K.; Tickle, I. J.; Jhoti, H.
Detection of secondary binding sites in proteins using fragment
screening. Proc. Natl. Acad. Sci. 2015, 112 (52), 15910–15915.
(12) Gee, C. T.; Arntson, K. E.; Urick, A. K.; Mishra, N. K.; Hawk, L. M.
L.; Wisniewski, A. J.; Pomerantz, W. C. K. Protein-observed 19F-
NMR for fragment screening, affinity quantification and druggability
assessment. Nat. Protoc. 2016, 11 (8), 1414–1427.
(30) Pearce, N. M.; Krojer, T.; Bradley, A. R.; Collins, P.; Nowak, R. P.;
Talon, R.; Marsden, B. D.; Kelm, S.; Shi, J.; Deane, C. M.; von Delft,
F. A multi-crystal method for extracting obscured crystallographic
states from conventionally uninterpretable electron density. Nat.
Commun. 2017, 8, 15123.
(13) Hajduk, P. J.; Huth, J. R.; Fesik, S. W. Druggability indices
for protein targets derived from NMR-based screening data. J. Med.
Chem. 2005, 48 (7), 2518–2525.
(31) Antolín, A. A.; Mestres, J. Distant polypharmacology among
MLP chemical probes. ACS Chem. Biol. 2015, 10 (2), 395–400.
(14) Davies, T. G.; Tickle, I. J. Fragment screening using X-ray
(32) Hopkins, A. L.; Groom, C. R.; Alex, A. Ligand efficiency: a useful
crystallography. Top. Curr. Chem. 2012, 317, 33–59.
metric for lead selection. Drug Discov. Today 2004, 9 (10), 430–431.
(15) Radeva, N.; Krimmer, S. G.; Stieler, M.; Fu, K.; Wang, X.; Ehrmann,
F. R.; Metz, A.; Huschmann, F. U.; Weiss, M. S.; Mueller, U.;
Schiebel, J.; Heine, A.; Klebe, G. Experimental active-site mapping
(33) Arris, C. E.; Boyle, F. T.; Calvert, A. H.; Curtin, N. J.; Endicott, J.
A.; Garman, E. F.; Gibson, A. E.; Golding, B. T.; Grant, S.; Griffin,
R. J.; Jewsbury, P.; Johnson, L. N.; Lawrie, A. M.; Newell, D. R.;
by fragments: hot spots remote
from
the catalytic center
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 13
1
2
3
4
5
6
7
8
9
Noble, M. E.; Sausville, E. A.; Schultz, R.; Yu, W. Identification of
4 i 2: the new graphical user interface to the CCP 4 program suite.
Acta Crystallogr. Sect. D Struct. Biol. 2018, 74 (2), 68–84.
novel purine and pyrimidine cyclin-dependent kinase inhibitors with
distinct molecular interactions and tumor cell growth inhibition
profiles. J. Med. Chem. 2000, 43 (15), 2797–2804.
(43) Waterman, D. G.; Winter, G.; Gildea, R. J.; Parkhurst, J. M.;
Brewster, A. S.; Sauter, N. K.; Evans, G. Diffraction-geometry
refinement in the DIALS framework. Acta Crystallogr. Sect. D
Struct. Biol. 2016, 72 (4), 558–575.
(34) Wyatt, P. G.; Woodhead, A. J.; Berdini, V.; Boulstridge, J. A.; Carr,
M. G.; Cross, D. M.; Davis, D. J.; Devine, L. A.; Early, T. R.; Feltell,
R. E.; Lewis, E. J.; McMenamin, R. L.; Navarro, E. F.; O’Brien, M.
A.; O’Reilly, M.; Reule, M.; Saxty, G.; Seavers, L. C. A.; Smith, D.-
M.; Squires, M. S.; Trewartha, G.; Walker, M. T.; Woolford, A. J.-
(44) Winter, G. Xia2: an expert system for macromolecular
crystallography data reduction. J. Appl. Crystallogr. 2010, 43 (1),
186–190.
A.
Identification
of
N
-(4-piperidinyl)-4-(2,6-
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
(45) McCoy, A. J.; Grosse-Kunstleve, R. W.; Adams, P. D.; Winn, M. D.;
Storoni, L. C.; Read, R. J. Phaser crystallographic software. J. Appl.
Crystallogr. 2007, 40 (Pt 4), 658–674.
dichlorobenzoylamino)-1 H -pyrazole-3-carboxamide (AT7519),
a novel cyclin dependent kinase inhibitor using fragment-based X-
ray crystallography and structure based drug design. J. Med. Chem.
2008, 51 (16), 4986–4999.
(46) De Bondt, H. L.; Rosenblatt, J.; Jancarik, J.; Jones, H. D.; Morgan,
D. O.; Kim, S. H. Crystal structure of cyclin-dependent kinase 2.
Nature 1993, 363 (6430), 595–602.
(35) Xing, L.; Klug-Mcleod, J.; Rai, B.; Lunney, E. A. Kinase hinge
binding scaffolds and their hydrogen bond patterns. Bioorg. Med.
Chem. 2015, 23 (19), 6520–6527.
(47) Murshudov, G. N.; Skubák, P.; Lebedev, A. A.; Pannu, N. S.; Steiner,
R. A.; Nicholls, R. A.; Winn, M. D.; Long, F.; Vagin, A. A.
REFMAC5 for the refinement of macromolecular crystal structures.
Acta Crystallogr. D. Biol. Crystallogr. 2011, 67 (Pt 4), 355–367.
(36) Garman, E. F. Radiation damage in macromolecular crystallography:
what is it and why should we care? Acta Crystallogr. Sect. D Biol.
Crystallogr. 2010, 66 (4), 339–351.
(48) Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K. Features and
development of Coot. Acta Crystallogr. D. Biol. Crystallogr. 2010,
66 (Pt 4), 486–501.
(37) Blanksby, S. J.; Ellison, G. B. Bond dissociation energies of organic
molecules. Acc. Chem. Res. 2003, 36 (4), 255–263.
(38) Ramagopal, U. A.; Dauter, Z.; Thirumuruhan, R.; Fedorov, E.; Almo,
S. C. Radiation-induced site-specific damage of mercury derivatives:
phasing and implications. Acta Crystallogr. D. Biol. Crystallogr.
2005, 61 (Pt 9), 1289–1298.
(49) Long, F.; Nicholls, R. A.; Emsley, P.; Graǽulis, S.; Merkys, A.;
Vaitkus, A.; Murshudov, G. N. AceDRG:
a stereochemical
description generator for ligands. Acta Crystallogr. Sect. D, Struct.
Biol. 2017, 73 (Pt 2), 112–122.
(39) Martin, M. P.; Alam, R.; Betzi, S.; Ingles, D. J.; Zhu, J.-Y.;
Schönbrunn, E. A novel approach to the discovery of small-molecule
ligands of CDK2. ChemBioChem 2012, 13 (14), 2128–2136.
(50) McNicholas, S.; Potterton, E.; Wilson, K. S.; Noble, M. E. M.
AceDRG: a stereochemical description generator for ligands. Acta
Crystallogr. D. Biol. Crystallogr. 2011, 67 (Pt 4), 386–394.
(40) Brown, N. R.; Noble, M. E. M.; Lawrie, A. M.; Morris, M. C.;
Tunnah, P.; Divita, G.; Johnson, L. N.; Endicott, J. A. Effects of
phosphorylation of threonine 160 on cyclin-dependent kinase 2
structure and activity. J. Biol. Chem. 1999, 274 (13), 8746–8756.
(51) Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L. N. Rapid
measurement of binding constants and heats of binding using a new
titration calorimeter. Anal. Biochem. 1989, 179 (1), 131–137.
(52) Turnbull, W. B.; Daranas, A. H. On the value of c: can low affinity
systems be studied by isothermal titration calorimetry? J. Am. Chem.
Soc. 2003, 125 (48), 14859–14866.
(41) Betzi, S.; Alam, R.; Martin, M.; Lubbers, D. J.; Han, H.; Jakkaraj, S.
R.; Georg, G. I.; Schönbrunn, E. Discovery of a potential allosteric
ligand binding site in CDK2. ACS Chem. Biol. 2011, 6 (5), 492–501.
(42) Potterton, L.; Agirre, J.; Ballard, C.; Cowtan, K.; Dodson, E.; Evans,
P. R.; Jenkins, H. T.; Keegan, R.; Krissinel, E.; Stevenson, K.;
Lebedev, A.; McNicholas, S. J.; Nicholls, R. A.; Noble, M.; Pannu,
N. S.; Roth, C.; Sheldrick, G.; Skubak, P.; Turkenburg, J.; Uski, V.;
von Delft, F.; Waterman, D.; Wilson, K.; Winn, M.; Wojdyr, M. CCP
TOC graphic
ACS Paragon Plus Environment

Page 13 of 13
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
Insert Table of Contents artwork here
13
ACS Paragon Plus Environment