Investigating the effects of the core nitrogen atom configuration on the thermodynamic solubility of 6,5-bicyclic heterocycles

Article abstract of DOI:10.1016/j.bmcl.2020.127752

Physicochemical properties, such as solubility, are important when prioritising compounds for progression on a drug discovery project. There is limited literature around the systematic effects of core changes on thermodynamic solubility. This work details

Full text of DOI:10.1016/j.bmcl.2020.127752

Bioorg. Med. Chem. Lett. 33 (2021) 127752
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Investigating the effects of the core nitrogen atom configuration on the
thermodynamic solubility of 6,5-bicyclic heterocycles
a
,
b
,
*
a
a
b
Brett Cosgrove
, Kenneth Down , Sophie Bertrand , Nicholas C.O. Tomkinson ,
a
Michael D. Barker
a
Medicinal Chemistry, GlaxoSmithKline Medicines Research Centre, Stevenage, UK
Department of Pure and Applied Chemistry, University of Strathclyde, Glasgow, UK
b
A R T I C L E I N F O
A B S T R A C T
Keywords:
Physicochemical properties, such as solubility, are important when prioritising compounds for progression on a
drug discovery project. There is limited literature around the systematic effects of core changes on thermody-
namic solubility. This work details the synthesis of nitrogen containing 6,5-bicyclic heterocyclic cores which are
common scaffolds in medicinal chemistry and the analysis of their physicochemical properties, particularly,
thermodynamic solubility. Crystalline solids were obtained where possible to enable a robust comparison of the
ChromlogD7.4
Fasted State Simulated Intestinal fluid (FaSSIF)
General Solubility Equation (GSE)
Kinetic solubility
Physicochemical properties
Simulated lung fluid (SLF)
Thermodynamic solubility
thermodynamic solubility. Other parameters such as pK
a
, melting point and lipophilicity were also measured to
determine the key factors affecting the observed solubility.
Introduction
development.11 It is important to measure solubility during lead opti-
1
2
misation, and to be able to have strategies for improving it. Solubility
is reported as either thermodynamic or kinetic; these refer to two
different aspects of behavior that a molecule possesses when undergoing
dissolution.13 Kinetic solubility refers to any solubility that is associated
Physicochemical and molecular properties such as solubility,
permeability and lipophilicity are important parameters in drug dis-
covery. They can influence the pharmacokinetic, pharmacodynamic and
1
14
safety profile of a potential drug. There is often a balancing act between
with a metastable form of a compound. Thermodynamic solubility is
these various parameters to find candidates with the appropriate profile
the maximum amount of the most stable form of the compound (pref-
erably crystalline) that can remain in solution under equilibrium con-
ditions.15 In the literature, kinetic solubility is widely reported and
accepted due to the fact that quantification can be automated
for the desired target.2
–4
To introduce novelty and overcome physicochemical barriers, the
medicinal chemist often explores alternative ring systems, with an
average of six new rings systems entering the drug market each year.5
16
effectively.
6
,5-Bicyclic heterocycles in particular, are ubiquitous to drug discovery
Within the literature, methods of improving the solubility of a
compound are limited and the data presented frequently does not report
full information with regards to the effect of a change on each property
of the molecule. A typical medicinal chemistry approach to enhancing
the solubility of a compound is to introduce groups such as a basic amine
and in 2019, the FDA approved multiple drugs containing this type of
motif (Fig. 1).6 Benzimidazoles, indazoles and purine type scaffolds
shown in Fig. 1 are well-established substructures, found in both frag-
6
,7
ments and drug molecules. A less common core found in the Bruton’s
tyrosine kinase inhibitor Zanubrutinib was developed through a ring-
merging approach of common kinase analogues. This led to a clinical
candidate with excellent selectivity and a desirable pharmacokinetic
or a carboxylic acid group. These simple modifications can affect other
parameters.1
7,18
For instance, introducing a basic amine can lead to an
increase in the volume of distribution which can lead to accumulation in
8
19
profile. Therefore, understanding the physicochemical properties of
certain tissues of the body, or even introduce other challenges such as
6
,5-bicyclic heterocycles is important to the medicinal chemistry
hERG inhibition.20 Adding a carboxylate group is known to be detri-
9
21
community.
Solubility is a critical parameter in drug discovery.10 For oral dosing,
mental to the permeability of the molecule. Alternative approaches for
3
improving solubility have been reported and include: increasing the sp
2
2
23–24
poorly soluble compounds can have reduced gut absorption and vari-
ability in drug plasma concentration which can lead to attrition during
character, distorting the crystal lattice structure
the aromatic ring count.25 However, these approaches can also
and decreasing
*
Corresponding author.
https://doi.org/10.1016/j.bmcl.2020.127752
Received 23 September 2020; Received in revised form 11 December 2020; Accepted 13 December 2020
Available online 29 December 2020
0
960-894X/© 2020 Elsevier Ltd. All rights reserved.

B. Cosgrove et al.
Bioorganic & Medicinal Chemistry Letters 33 (2021) 127752
determine how the properties were affected.
Cores 1, 4 and 5 were synthesised according to literature methods
see supplementary information for experimental details).2
6–29
Cores 2
(
and 3 required synthetic development (Scheme 1 and Scheme 2).
The first step of the synthesis for compounds 2a-c (Scheme 1)
involved refluxing 2.1 with aqueous sodium hydroxide to give the tri-
carbonyl product (2.2) in good yield (62%). Compound 2.2 was heated
with hydrazinecarboxamide under basic aqueous conditions to form the
bicyclic heterocycle in a single step (2.3, 79%). The pyrimidone species
(
3
2.3) was subsequently chlorinated using POCl then iodinated with NIS
to give the key bis-halogenated intermediate (2.5) in good yield (62%
over two steps). Compound 2.5 was further functionalised through an
S
N
Ar reaction to give the products 2.6–2.8. Finally, a Suzuki cross-
coupling reaction with 3,4-dimethoxyphenyl boronic acid under stan-
dard reaction conditions gave compounds 2a-2c, which were found to
be crystalline after isolation.
The synthesis of compounds 3a-c are shown in Scheme 2. Compound
3.1 and chloropropanone were heated together under forcing reaction
conditions to form the cyclised product (3.2) in good yield (63%). It
proved important to carry out the reaction under solvent free conditions
to prevent the formation of multiple alkylation side products (see Sup-
porting Information). The bicyclic heterocycle 3.2 was iodinated with
Fig. 1. Examples of FDA drug approvals in 2019 containing 6,5-bicyclic
heterocycles.
compromise other properties of the molecule in lead optimisation.
Within this work, an alternative strategy was applied to investigate the
physicochemical properties of a family of bicyclic heterocycles con-
taining a bridgehead nitrogen atom.
◦
NIS at 0 C to give the key bis-halogenated compound 3.3 (79%). The
N
penultimate step of the synthesis involved a series of S Ar reactions to
A range of 6,5-bicyclic heterocycles varying only in their nitrogen
atom configuration were synthesised, with the substituents held con-
stant to determine how the nitrogen atoms within the core bicyclic ring
affected the properties of the molecule. The aim was to identify whether
these changes would influence the solubility without introducing other
liabilities into the molecules. This approach might offer alternative
templates with better physicochemical properties to build a molecule
around and provide a greater chance in meeting candidate quality
criteria. Analysis of the physicochemical properties on crystalline
batches was carried out to give a robust comparison of the thermody-
namic solubility data of the derivatives prepared. Plane polarised light
microscopy was initially used to detect signs of crystallinity for all final
compounds (1a-5c) and X-ray powder diffraction (XRPD) was used to
give compounds (3.4–3.6) in low to moderate yield (24–55%). Finally, a
Suzuki cross-coupling reaction gave the compounds (3a-3c) as gummy
solids. Multiple crystallisation techniques were performed on these
products, however, only amorphous solids were obtained, as confirmed
by plane polarised light microscopy and XRPD analysis.
Compounds (1a-5c) were profiled in three solubility assays
(Table 1). The thermodynamic solubility was measured in two bio-
relevant media, Fasted State Simulated Intestinal Fluid (FaSSIF6.5) and
3
0
Simulated Lung Fluid (SLF6.9). Both biorelevant media mimic physi-
ological conditions of the lungs and stomach. FaSSIF is buffered at pH
6.5 and contains 0.75 mM lecithin which comprises fatty lipids; SLF is
buffered at pH 6.9 and contains both 0.75 mM lecithin and a protein,
bovine serine albumin (BSA).31 The presence of the fatty lipids and
protein influence the solubility of the compounds. For example, the
lipids could potentially form micelles within the buffer and the BSA
could interact with the compounds through non-covalent interactions.
The kinetic solubility was measured from a 10 mM DMSO stock solution
(Kinetic7.4). The lipophilicity was measured using a chromatographic
a
confirm the crystalline material. Additional properties such as pK ,
melting point and chromatographic logD were also measured to un-
derstand how subtle core changes influenced physicochemical
properties.
To enable the direct comparison of the effect of altering the five cores
3
2
(
Fig. 2) on the physicochemical profile of the molecule, the pendant
reverse phase method (ChromlogD7.4) and calculated using clogP.
The pK and melting points of all final compounds (1a-5c) were
measured.
groups were kept constant. A 3,4-dimethoxybenzene substituent was
incorporated in all compounds prepared. The pendant amine R’ groups
selected consist of typical medicinal chemistry functional groups: an
aromatic ring (a), a heteroaromatic ring (b) and a fully saturated ring
a
The individual compound solubility data shown in Table 1 was
averaged for each core (1–5) and plotted in Fig. 3. The effects of the
different amine monomers (a-c) are shown in Fig. 4.
(
c). These groups provided a range of functionality on each core to
All five cores had an average FaSSIF solubility that was greater than
the average SLF solubility. The 0.4 log unit difference in the pH of the
two media equates to a 2.5-fold difference. In the more acidic FaSSIF
media, the equilibrium will be shifted towards the protonated state for a
compound that contains a basic centre. An increased ionisation state will
a
lead to an increase in aqueous solubility. The pK of each compound was
measured to investigate this possibility further. However, it is important
to consider the buffer composition and the effects it could have on the
basicity of the compound. The presence of the lipids and BSA within the
SLF buffer are likely to reduce the pK
a
of the compound and in turn
values could be over-
reduce solubility. Therefore, the measured pK
a
estimated in the biorelevant buffers for measuring the thermodynamic
solubility.
Core 2, a pyrazolopyrimidine motif, demonstrated the lowest
average solubility of all five cores across all three solubility assays. The
greatest average ChromLogD7.4 (5.6) was observed for this core, the high
lipophilicity being associated with a low solubility. Introducing polarity
into a molecule is often used to reduce the lipophilicity.33 Interestingly,
Fig. 2. 6,5-Bicyclic heterocycles investigated. Each core (1–5) was appended
with three groups (a-c).
2

B. Cosgrove et al.
Bioorganic & Medicinal Chemistry Letters 33 (2021) 127752
◦
◦
Scheme 1. Synthetic route for the preparation of Core 2 compounds (2a-c). (a) NaOH, EtOH, 2 h, 100 C, 62% (b) hydrazinecarboxamide, Na
2
CO
3
, H
2
O, 1 h, 100 C,
◦
◦
’
◦
7
9% (c) PhNMe
2
, POCl
3
, 16 h, 110 C, 69% (d) NIS, CH
CO , 1,4-dioxane/H
2
Cl
2
, 1 h, 0 C, 54% (e) R –NH
2
, NEt
3
, 1,4-dioxane, 5 h, 120 C, 35–92% (f) 3,4-dimethoxyphenyl boronic acid,
◦
XPhosPdG2 (10 mol%), Cs
2
3
2
O (4:1), 5 h, 130 C, 22–74%.
◦
◦
Scheme 2. Synthesis of Core 3 analogues (3a-3c). (a) Chloropropanone, 2.5 h, 130 C, 63% (b) NIS, CH
2
Cl
2
, 4 h, 0 C, 79% (c) R-NH
2
, NEt
3
, 1,4-dioxane, 5 h,
1
30 ◦C, 24–55% (d) 3,4-dimethoxyphenyl boronic acid, PdCl
2
(dppf)CH
2
Cl
2
adduct (20 mol%), Cs
2
CO
3
, 1,4-dioxane/H
2
O (4:1), 5 h, 130 ◦C, 33–87%.
shifting the bridgehead nitrogen atom one position lowers the Chrom-
LogD to 4.5 and increases the pK H to 5.3 (Core 3, an imidazopyrazine
template) which significantly improves the solubility of this series
Figure 3). This illustrates the significance of the nitrogen atom
Medicinal chemists frequently use kinetic solubility to drive SAR as this
can be established by a higher throughput assay. The data obtained
within this study demonstrates a considerable drop in the thermody-
namic solubility when compared to the kinetic solubility e.g. see Table 1
kinetic vs SLF solubility. Core 3 was amorphous, and it did not follow
the same trends in solubility as the other four cores. Therefore, it is
important to develop a stable crystalline polymorph of a compound as
early as possible to obtain a realistic measurement of the thermody-
namic solubility.
a
(
configuration in the core and how this subtle transformation alters the
physicochemical properties of the molecule.
Core 1, containing an imidazopyridine bicycle, was the most basic
core of the five examined (pK
FaSSIF solubility of the four crystalline cores (116 µg/mL). Interestingly,
the pK of around 6.6 falls between the pH values of the FaSSIF and SLF
a
H = 6.6). This had the greatest average
a
Core 4, a pyrazolopyrimidine core, and Core 5, a pyrazolotriazine
core, both displayed high kinetic solubilities. A significant drop in the
thermodynamic solubility was observed for both cores in comparison to
the kinetic solubility, emphasising the need to monitor thermodynamic,
in addition to kinetic solubility.
media. Changing from FaSSIF to SLF the solubility drops, probably due
to a decrease in the proportion of ionised compound in the SLF assay
medium.
Another trend observed was the average kinetic solubility (measured
from a 10 mM DMSO solution) was higher than the SLF solubility
Each core had been appended with an aromatic, a heteroaromatic
and a saturated group. This provided a range of functionality to
(
measured from solid) for all of the crystalline compounds examined.
3

B. Cosgrove et al.
Bioorganic & Medicinal Chemistry Letters 33 (2021) 127752
Table 1
Physicochemical profile of compounds (1a-5c).
Cpd
Kinetic
FaSSIF
SLF
pK
a
ChromlogD7.4
Mp
logS
clogP
a
1
29
86
64
6
2
6.6
4.3
4.9
6
203–205
85–88
ꢀ 5.2
ꢀ 4.0
ꢀ 4.4
3.9
3.9
4.2
b
c
c
c
c
c
1
109
233
33
29
6.5/4.7
6.7
c
1
90–94
a
2
19
35
44
5
<6
6
3.8
4.8
5.5
6.4
194–198
101–103
104–107
ꢀ 4.2
ꢀ 3.3
ꢀ 3.6
3
b
2
31
13
2.8/4.9
4.4
3
c
2
3
3.3
a
b
b
3
53
67
99
32
9
5
3.9
4.5
5.2
191-193b
128-131b
ꢀ 4.5
ꢀ 3.8
ꢀ 3.8
3.3
3.3
3.6
b
b
b
3
335
58
5.3/4.0
5.6
c
b
b
b
3
7
10
93-95
a
a
a
4
79
15
1
5.4
3.7
4.3
5.2
238–240
149–151
157–159
ꢀ 4.6
ꢀ 3.8
ꢀ 4.1
3
b
a
a
4
>107
>180
141
39
5.4/4.1
5.7
3
c
a
a
4
25
18
3.3
a
5
19
32
36
29
10
36
18
2.9
3.9
4.5
5.1
167–169
149–152
143–145
ꢀ 3.2
ꢀ 3.1
ꢀ 3.3
2.3
2.3
2.6
b
5
>150
>165
2.5/4.7
3.1
c
5
a
b
c
The solubility experiments were performed in at least duplicate and on crystalline material unless otherwise noted. n = 1, = amorphous material, = pK
a
of meta-
◦
pyridine measured. Melting point (Mp, C), Kinetic solubility7.4 (µg/mL), FaSSIF6.5 solubility (µg/mL), SLF6.9 solubility (µg/mL), and ChromlogD7.4. clogP was pre-
dicted using GSK Biobyte/Daylight tool.
understand how each of the groups affected the solubility. The effect of
each monomer (benzamide (a), meta-pyridine (b) and tetrahydropyran
(
c)) on the average solubility was plotted (Fig. 4).
The cores containing the pyridine monomer (b) show the highest
thermodynamic solubility in both FaSSIF and SLF. This was probably
due to the additional basic centre resulting in a higher proportion of
protonated species, thus leading to a higher observed solubility. This
observation is further supported by the fact that a higher proportion of
compound would be protonated in the lower pH system (FaSSIF) which
was reflected in the improved solubility for compounds containing this
monomer.
Compounds appended with the benzamide monomer (a) displayed
the poorest solubility across all three solubility assays. This was prob-
ably due to the intermolecular hydrogen bonding interactions between
molecules. Also, the additional carboaromatic ring can support
π π
-
stacking, with the amide group planar to the aromatic ring. This led to
an enhanced crystal lattice energy and as shown in Table 1, compounds
appended with the benzamide monomer showed the highest melting
point. Furthermore, as represented in the General Solubility Equation
Fig. 3. Compounds (1a-5c) measured kinetic solubility (pH = 7.4) and ther-
modynamic (FaSSIF, pH = 6.5 & SLF, pH = 6.9) solubility. Data is displayed as
an average for each Core 1–5, in numerical order.** 4b, 4c, 5b and 5c contain a
greater than (>) modifier in the kinetic solubility measurement. * 2a contains a
less than (<) modifier in the SLF solubility measurement.
(
GSE), a higher melting point is associated with a lower solubility,
3
4,35
consistent with the data presented in Table 1.
The GSE, also known as the Yalkowsky equation, is shown in Eq. 1. It
is used to predict the aqueous solubility (logS) of a given compound if
3
4
the melting point (Mp) and clogP of the compound are known. The
equation is a predictive tool; variables such as crystallinity and melting
point can influence the accuracy of the GSE and so it should be viewed
with caution; however, it provides a particularly useful platform from
◦
which to analyse data. The equation predicts that an increase of 100 C
in melting point would reduce the intrinsic solubility by a factor of 10
(
logarithmic scale) which equates to a unit lower in logS. It should be
noted that 85% of drugs have logS values between ꢀ 1 and ꢀ 5, ꢀ 1 being
highly polar molecules that may have low membrane permeability and
3
6
ꢀ 5 being hydrophobic molecules with low aqueous solubility.
logS = 0.5 ꢀ clogP ꢀ 0.01(Mp ꢀ 25)
Eq. 1: Yalkowsky (GSE) equation.
It is important to know the physical state of the compound, as an
amorphous solid does not have a definitive melting point,3 whereas, a
stable crystalline polymorph displays a single sharp melt. Therefore,
predicting the solubility (logS) based on the melting point of a crystal-
line solid should provide a robust and reliable method for the compar-
ison of data across a series.
7
Fig. 4. The average solubility for cores containing each monomer (a-c attached
to the core of the molecule). ** 4b, 4c, 5b and 5c contain a greater than (>)
modifier in the kinetic solubility measurement. * 2a contains a less than (<)
modifier in the SLF solubility measurement.
4

B. Cosgrove et al.
Bioorganic & Medicinal Chemistry Letters 33 (2021) 127752
The data in Table 1 shows a clear correlation between the thermo-
generating the physicochemical data and advice on the assays; and Kate
Dennis for XRPD analysis. We also thank the EPSRC for funding via
Prosperity Partnership EP/S035990/1.
dynamic solubilities of crystalline compounds and their predicted sol-
ubility values (logS). Compound 1a (FaSSIF 6 µg/mL, logS ꢀ 5.2)
displays the lowest solubility for Core 1 analogues and it was ranked
with the lowest solubility by the GSE equation. Compound 1b (FaSSIF
Appendix A. Supplementary data
1
09 µg/mL, logS ꢀ 4.0) is around 18-fold more soluble in FaSSIF than
compound 1a, which is mirrored by a 1.2 log unit increase in logS. This
trend was consistent with the majority of the crystalline compounds
where the logS value was reflective of the compounds in each core. The
thermodynamic solubility values for the three analogues of Core 5 were
very similar and this was reflected in the narrow range of logS for this
series of compounds. The exception to the trend was for Core 3, where
each of the compounds were amorphous, and the predicted solubilities
of these compounds did not follow the measured thermodynamic solu-
bilities. This can be explained by the amorphous form of the compounds
and the uncertainty associated with the melting points determined. This
again, highlights the importance of gathering thermodynamic solubility
data on crystalline compounds.
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.bmcl.2020.127752.
References
[
[
[
1] Nakashima S, Yamamoto K, Arai Y, Ikeda Y. Chem Pharm Bull. 2013;61:1228.
2] Morgan P, Van Der Graaf PH, Arrowsmith J, et al. Drug Discov Today. 2012;17.
3] Saal C, Petereit AC. Eur J Pharm Sci. 2012;47:589.
[4] Volpe DA. Future Med Chem. 2011;3:2063.
[
[
5] Taylor RD, MacCoss M, Lawson AD. J Med Chem. 2014;57:5845.
6] https://www.fda.gov/drugs/new-drugs-fda-cders-new-molecular-entities-and-
new-therapeutic-biological-products/novel-drug-approvals-2019 (Accessed
13.01.2020).
[
[
[
7] Giordanetto F, Jin C, Willmore L, Feher M, Shaw D. E. J Med Chem. 2019;62:3381.
8] Guo Y, Liu Y, Hu N, et al. J Med Chem. 2019;62:7923.
9] Wenlock MC, Austin RP, Barton P, Davis AM, Leeson PD. J Med Chem. 2003;46:
1250.
In summary, five alternative 6,5-bicyclic nitrogen containing het-
erocycles were synthesised and crystalline batches of compounds were
obtained for 4 of the 5 cores. The physicochemical properties of com-
pounds (1a-5c) were profiled and the solubility was found to vary
significantly due to these subtle changes in nitrogen atom position. The
intrinsic properties of the nitrogen atom configuration and substituents
were shown to influence the solubility considerably through modifica-
tions to lipophilicity, basicity and crystallinity. For each of the crystal-
line cores, the analogue with the least solubility was generally the
compound with the highest melting point and lowest logS. The data
gathered showed that the kinetic solubility was often higher compared
to the thermodynamic solubility. The measured thermodynamic solu-
bility was dependent on the biorelevant media (FaSSIF6.5 vs SLF6.9), as
the solubility in SLF, was in all cases, lower than FaSSIF. This focus on
the thermodynamic solubility of the core of crystalline compounds at an
early stage in a drug discovery programme could be a useful strategy to
adopt to ensure physicochemical properties are optimised in parallel
with potency and selectivity for the desired target.
[10] Valko K. Physicochemical and biomimetic properties in drug discovery. John Wiley &
Sons, Inc.; 2014.
[
[
11] Sugano K. Int J Pharm. 2010;405:79.
12] Ahmad NM. Bioorg Med Chem Lett. 2016;26:2975.
[13] Walker MA. Bioorg Med Chem Lett. 2017;27:5100.
[
[
14] Covarrubias CH, Vilchis-Reyes MA, Y ´e pez-Mulia L, et al. Eur J Med Chem. 2012;52:
93.
15] Valko K, Butler J, Eddershaw P. Expert Opin Drug Discov. 2013;8:1225.
1
[16] Hill AP, Young RJ. Drug Discov Today. 2010;15:648.
[
[
17] Cisneros JA, Robertson MJ, Mercado BQ, Jorgensen WL. ACS Med Chem Lett. 2017;
:124.
18] Lassalas P, Gay B, Lasfargeas C, et al. J Med Chem. 2016;59:3183.
8
[19] Charifson PS, Walters WP. J Med Chem. 2014;57:9701.
[
[
[
20] Jamieson C, Moir EM, Rankovic Z, Wishart G. J Med Chem. 2006;49:5029.
21] Thomae AV, Wunderli AH, Kr a¨ mer SD. Biophys J. 1802;2005:89.
22] Lovering F, Bikker J, Humblet C. J Med Chem. 2009;52:6752.
[23] Ishikawa M, Hashimoto Y. J Med Chem. 2011;54:1539.
[
[
[
24] Wang H-L, Katon J, Balan C, et al. J Med Chem. 2007;50:3528.
25] Ritchie TJ, Macdonald SJF. Drug Discov Today. 2009:14.
26] Catalano JG, Gaitonde V, Beesu M, et al. Tet Lett. 2015;56:6077.
[27] Long J, Lee W-S, Chough C, Bae IH, Kim BM. J Org Chem. 2015;80:4716.
[
[
[
[
28] Wavhale RD, Martis EAF, Ambre PK, et al. Bioorg Med Chem. 2017;25:4835.
29] Mejdrova I, Chalupska D, Kogler M, et al. J Med Chem. 2015;58:3767.
30] Hughes JP, Rees S, Kalindjian SB, Philpott KL. Br J Pharmacol. 2011;162:1239.
31] Williams HD, Trevaskis LN, Charman AS, et al. Pharmacol Rev. 2013;65:315.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
[32] Liang C, Qiao JQ, Lian HZ. J Chromatogr A. 2017;1528:25.
[
[
[
[
[
33] Arnott JA, Planey SL. Exp Opin Drug Discov. 2012;10:863.
34] Pinal RJ. Org Biomol Chem. 2004;2:2692.
35] Ran Y, He Y, Yang G, Johnson JLH, Yalkowsky SH. Chemosphere. 2002;48:487.
36] Sanghvi T, Jain N, Yang G, Yalkowsky SH. QSAR Comb Sci. 2003;22:258.
37] Ran Y, Jain N, Yalkowsky SH. J Chem Inf Comput Sci. 2001;41:1208.
Acknowledgments
The authors are thankful to Iain Reid and Shenaz Bunally for
5