Structure-based design of N-substituted 1-hydroxy-4-sulfamoyl-2-naphthoates as selective inhibitors of the Mcl-1 oncoprotein

Article abstract of DOI:10.1016/j.ejmech.2016.02.006

Structure-based drug design was utilized to develop novel, 1-hydroxy-2-naphthoate-based small-molecule inhibitors of Mcl-1. Ligand design was driven by exploiting a salt bridge with R263 and interactions with the p2 pocket of the protein. Significantly, target molecules were accessed in just two synthetic steps, suggesting further optimization will require minimal synthetic effort. Molecular modeling using the Site-Identification by Ligand Competitive Saturation (SILCS) approach was used to qualitatively direct ligand design as well as develop quantitative models for inhibitor binding affinity to Mcl-1 and the Bcl-2 relative Bcl-x_L as well as for the specificity of binding to the two proteins. Results indicated hydrophobic interactions in the p2 pocket dominated affinity of the most favourable binding ligand (3bl: K_i = 31 nM). Compounds were up to 19-fold selective for Mcl-1 over Bcl-x_L. Selectivity of the inhibitors was driven by interactions with the deeper p2 pocket in Mcl-1 versus Bcl-x_L. The SILCS-based SAR of the present compounds represents the foundation for the development of Mcl-1 specific inhibitors with the potential to treat a wide range of solid tumours and hematological cancers, including acute myeloid leukemia.

Full text of DOI:10.1016/j.ejmech.2016.02.006

Accepted Manuscript
Structure-Based Design of N-Substituted 1-Hydroxy-4-sulfamoyl-2-naphthoates as
Selective Inhibitors of the Mcl-1 Oncoprotein
Maryanna E. Lanning, Wenbo Yu, Jeremy L. Yap, Jay Chauhan, Lijia Chen, Ellis
Whiting, Lakshmi S. Pidugu, Tyler Atkinson, Hala Bailey, Willy Li, Braden M. Roth,
Lauren Hynicka, Kirsty Chesko, Eric A. Toth, Paul Shapiro, Alexander D. MacKerell,
Jr., Paul T. Wilder, Steven Fletcher
PII:
S0223-5234(16)30075-7
10.1016/j.ejmech.2016.02.006
EJMECH 8357
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 9 November 2015
Revised Date: 2 February 2016
Accepted Date: 3 February 2016
Please cite this article as: M.E. Lanning, W. Yu, J.L. Yap, J. Chauhan, L. Chen, E. Whiting, L.S. Pidugu,
T. Atkinson, H. Bailey, W. Li, B.M. Roth, L. Hynicka, K. Chesko, E.A. Toth, P. Shapiro, A.D. MacKerell
Jr., P.T. Wilder, S. Fletcher, Structure-Based Design of N-Substituted 1-Hydroxy-4-sulfamoyl-2-
naphthoates as Selective Inhibitors of the Mcl-1 Oncoprotein, European Journal of Medicinal Chemistry
(2016), doi: 10.1016/j.ejmech.2016.02.006.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Structure-Based Design of N-Substituted 1-Hydroxy-4-
sulfamoyl-2-naphthoates as Selective Inhibitors of the
Mcl-1 Oncoprotein
†
‡
†‡
†
†
†
#
Maryanna E. Lanning, Wenbo Yu, Jeremy L. Yap, Jay Chauhan, Lijia Chen, Ellis Whiting,
∆
#
^
#
∆
&
Lakshmi S. Pidugu, Tyler Atkinson, Hala Bailey, Willy Li, Braden M. Roth, Lauren Hynicka,
†
$
∆$
†$
†$
∆$
Kirsty Chesko, Eric A. Toth, Paul Shapiro, Alexander D. MacKerell, Jr. Paul T. Wilder* and
†
$
Steven Fletcher*
†
Department of Pharmaceutical Sciences, University of Maryland School of Pharmacy, 20 N. Pine St.,
#
∆
Baltimore, MD 21201; School of Chemistry, University of Cardiff, CF10 3AT, UK; Department of
Biochemistry and Molecular Biology, Center for Biomolecular Therapeutics, University of Maryland
&
School of Medicine, 22 S. Greene St., Baltimore, MD 21201; Department of Pharmacy Practice and
#
Science, University of Maryland School of Pharmacy, 20 N. Pine St., Baltimore, MD 21201; PharmD
^
Program, University of Maryland School of Pharmacy, 20 N. Pine St., Baltimore, MD 21201; Bryn
$
Mawr School, 109 W. Melrose Ave., Baltimore, MD 21210, USA; University of Maryland
Greenebaum Cancer Center, 22 S. Greene St., Baltimore, MD 21201, USA.
KEYWORDS. Mcl-1, Bcl-x , protein–protein interaction, apoptosis, cancer
L
ABSTRACT. Structure-based drug design was utilized to develop novel, 1-hydroxy-2-naphthoate-based
small-molecule inhibitors of Mcl-1. Ligand design was driven by exploiting a salt bridge with R263 and
1

ACCEPTED MANUSCRIPT
interactions with the p2 and p3 pockets of the protein. Significantly, target molecules were accessed in
just two synthetic steps, suggesting further optimization will require minimal synthetic effort. Molecular
modeling using the Site-Identification by Ligand Competitive Saturation (SILCS) approach was used to
qualitatively direct ligand design as well as develop quantitative models for inhibitor binding affinity to
Mcl-1 and the Bcl-2 relative Bcl-x as well as for the specificity of binding to the two proteins. Results
L
indicated hydrophobic interactions with the p2 pockets dominate the affinity of the most favourable
binding ligand (3bl: K
Selectivity of the inhibitors was driven by interactions with the deeper p2 pocket in Mcl-1 versus Bcl-
. The SILCS-based SAR of the present compounds represents the foundation for the development of
i L
= 31 nM). Compounds were up to 20-fold selective for Mcl-1 over Bcl-x .
x
L
Mcl-1 specific inhibitors with the potential to treat a wide range of solid tumours and hematological
cancers, including acute myeloid leukaemia.
Introduction
The B-cell lymphoma-2 (Bcl-2) family of proteins regulates the intrinsic apoptosis pathway that is
responsible for programmed cell death, or apoptosis. The pathway involves protein–protein interactions
(PPIs) between pro-apoptotic members of the Bcl-2 family, such as Bim, Bak and Bad, and anti-
apoptotic members, such as Bcl-x and myeloid cell leukemia-1 (Mcl-1).[1,2] More specifically,
L
through conserved hydrophobic crevices, the anti-apoptotic Bcl-2 proteins capture the BH3 α-helical
domains of their pro-apoptotic counterparts, effectively “neutralizing” their cell killing functions.
Evasion of apoptosis is a hallmark of cancer, and is also one culprit for the development of resistance to
current chemo- and radiotherapies.[3] More than a decade of research has seen the emergence of potent
inhibitors of Bcl-x , including the prototypical BH3 mimetic ABT-737.[4-6] In particular ABT263, or
L
Navitoclax, which is a dual inhibitor of Bcl-x and Bcl-2, but not of Mcl-1,[7] has performed well in
L
2

ACCEPTED MANUSCRIPT
phase I/II clinical trials for several malignancies, validating the inhibition of the anti-apoptotic Bcl-2
proteins as a strategy to treat cancer.[8-11]
Mcl-1 overexpression and/or amplification of the Mcl-1 gene immortalizes cells, and has been
observed in many human solid tumours, including pancreatic, prostate, cervical, lung and breast cancers,
[
12-19] as well as B-cell lymphomas and hematological cancers, including acute myeloid leukemia
AML).[20,21] While Navitoclax continues to perform well in clinical trials, its low affinity for Mcl-1 is
(
a contributing factor to the observed resistance of several tumour cell lines.[22-26] Moreover, the
upregulation of Mcl-1 has been directly linked to the reduced efficacy of several FDA-approved anti-
cancer chemotherapies. Meanwhile, Zhang and colleagues demonstrated that RNAi-mediated
downregulation of Mcl-1 decreased tumorigenicity of a mouse xenograft model.[27] Taken together,
these findings indicate that the pharmacologic inhibition of Mcl-1 is an attractive, complementary
and/or adjuvant strategy towards the execution of cancer cells by re-activating apoptosis.
In a similar vein to the inhibition of Bcl-x , it is envisaged that the development of synthetic
L
agents capable of disrupting the interaction between Mcl-1 and the BH3 α-helical “death” domains of
pro-apoptotic Bcl-2 proteins will “neutralize” Mcl-1’s cell survival role. Indeed, several groups have
implemented this stratagem and successfully developed effective inhibitors of Mcl-1.[26,28-44] We
herein report a structure-based design approach that has led to the discovery of potent inhibitors of Mcl-
1
based on a novel 1-hydroxy-2-naphthoate scaffold.
Design
Fesik and co-workers recently reported the identification of potent and selective inhibitors of
Mcl-1 through fragment-based drug design (FBDD).[28,42] Their inhibitor design, for example 1
Scheme 1), encompasses a hydrophobic bi-aryl scaffold, projected from which is a carboxylic acid that
(
recognizes R263 (bound by D67 of Bim-BH3), and also a hydrophobic linker and “tail” that probes into
the p2 pocket (bound by L62 of Bim-BH3). Meanwhile, researchers at AbbVie, also employing FBDD,
3

ACCEPTED MANUSCRIPT
discovered 5-substituted salicylates, such as 2, as potent inhibitors of Mcl-1 wherein the carboxylic acid
also binds R263 and the 5-substituent delves into the p2 and/or p1 pockets.[38] Inspired by these
reports, we considered that merging the two scaffolds to afford a 1-hydroxy-2-naphthoic acid core
would provide a novel and alternative platform from which to inhibit Mcl-1, wherein the carboxylic acid
was predicted to bind R263, and the distal phenyl ring towards and/or in the p3 pocket. Further
engineering to gain access to the p2 pocket was driven by the hydroxyl group and the carboxylic acid
whose ortho- and meta-directing effects, respectively, synergize to promote the regioselective 4-
chlorosulfonylation of the scaffold;[45] subsequent amination of this readily introduced functional
group is expected to facilitate occupancy of the p2 pocket. In addition, although not discussed by
Abbvie, we reasoned that the hydroxyl group of their salicylate moiety might promote cell permeability
through an intramolecular hydrogen bond that serves to mask the negative charge of the carboxylic acid.
Whilst this work was in progress, similar small-molecules derived from 2-((1-hydroxynaphthalen-2-
yl)thio)acetic acid were reported to inhibit Mcl-1,[39,46] although the present work differs in that the
bicycle core is unique and the synthetic route to access target molecules is simpler and shorter (Scheme
2
).
4

ACCEPTED MANUSCRIPT
Scheme 1: Structure-based design of novel 1-hydroxy-4-sulfamoyl-2-naphthoates as Mcl-1 inhibitors.
To facilitate ligand design, the Site Identification by Ligand Competitive Saturation (SILCS)
method[47-50] was used. SILCS FragMaps represent the 3D free energy functional group requirements
for different types of functional groups in and around the protein. The FragMaps include contributions
from solute-protein interactions and both solute and protein desolvation in the context of protein
flexibility.[48,49] In addition, to assure that all regions accessible to solute atoms can be occupied by
the studied ligands, the protein surface is defined based on SILCS exclusion maps.[51] Notably, the
information content of SILCS may be used to qualitatively direct ligand design as well as to rapidly
make quantitative estimates of relative ligand affinities.[52]
As motivation for the present ligand design was in part based on the compound reported by
Fesik and coworkers (1), initial analysis of the SILCS maps on Mcl-1 was performed with respect to this
compound. Figure 1 shows the generic FragMaps overlaid on Mcl-1 along with the crystallographic and
docked orientations of 1. The docked orientation closely mimics the crystallographic orientation of the
ligand. The ligand occupies nonpolar FragMaps, including a region in the upper left of panel A that is
occupied by the chlorine of 1. In addition, a negative (NEG) FragMap (orange) is occupied by the acid
moiety of the ligand. While these features largely directed the docked orientation of 1, the inclusion of
protein flexibility in the SILCS method is essential. Shown in Figure 1B is a surface representation of
the crystal structure of Mcl-1 used to initiate the SILCS simulations onto which the crystallographic
orientation of 1 is overlaid. As can be seen, the inhibitor structure overlaps with the protein surface,
such that if the protein surface from the crystal structure was used for docking, it would not be possible
to attain the experimental binding orientation. However, as the SILCS method includes protein
flexibility allowing for solutes in the SILCS simulations to penetrate the initial protein surface the
surface may alternatively be defined based on the region not sampled by the solutes or water during the
simulations. This allows the protein surface to be defined in terms of an exclusion map that, as shown in
5

ACCEPTED MANUSCRIPT
Figure 1C, accounts for the structural changes in the protein required to properly dock 1. A notable
feature of the bound orientation of 1 and the SILCS FragMaps is the pocket into which the 4-chloro-3,5-
dimethylphenyl moiety binds. This is occupied by a nonpolar FragMap, which extends beyond the 4-
chloro-3,5-dimethylphenyl ring, suggesting that larger nonpolar groups could facilitate binding in that
region consistent with the ligand design strategy.
Figure 1. (A) Predicted binding orientation (cyan carbons) of 1 overlaid on its crystal orientation (pink
carbons). FragMaps are shown as wire mesh in the following colours and GFE cutoff for generic
nonpolar (APOLAR, green, −1.2 kcal/mol), neutral donor (HBDON, blue, −1.0 kcal/mol), neutral
acceptor (HBACC, red, −1.0 kcal/mol), and negative acceptor (NEG, orange, -1.3 kcal/mol). (B & C)
The crystallographic orientation of the inhibitor 1 (from PDB ID: 4HW3) overlaid (B) on the crystal
protein surface of the Mcl-1-BH3 peptide complex structure (PDB ID: 4HW4) used to initiate the
SILCS simulations and (C) on the SILCS exclusion map.
Results and Discussion
Molecular modeling and SILCS functional group affinity mapping (FragMaps)[47-50] of the
Mcl-1 binding site indicated that the carboxylic acid of designed molecule 3a (Figure 2) would occupy
an energetically favourable region indicated by a negative FragMap, associated with a salt bridge
interaction with R263, while the distal ring of the naphthyl core is close to the p3 pocket demarcated by
6

ACCEPTED MANUSCRIPT
a favourable non-polar FragMap, consistent with the published studies.[28,38,39] As anticipated, the
aniline was directed into the hydrophobic p2 pocket, which is also demarcated by a nonpolar FragMap.
With the molecular modeling data in hand, compound 3a was then synthesized according to Scheme 2.
Briefly, commercially available 1-hydroxy-2-naphthoic acid (4) was regioselectively 4-
chlorosulfonylated to yield 5, which was isolated by pouring over ice and used without further
purification. Sulfonyl chloride 5 was next reacted with 4-bromoaniline to furnish the target molecule 3a
in excellent overall yield (83%). This convenient and simple 2-step synthesis allowed us to access the
range of ligands presented below, resulting in the discovery of a low-nanomolar binder, and is in stark
contrast to the lengthier synthesis of other families of Mcl-1 inhibitors.[39]
Figure 2. (A) The binding mode of 3a, as predicted by SILCS. Important residues within the p2 binding
pocket are shown in stick representation with the carbon atoms coloured in green. (B) Compound 3a
and 2 are shown in stick representation with the carbon atoms coloured in cyan and pink, respectively.
Colours and cutoffs used to show the FragMaps are the same as in Figure 1. (C) 2D chemical structure
of 3a.
Evaluation of 3a in a fluorescence polarization competition assay (FPCA)[53] indicated that it
disrupted the Mcl-1–Bak-BH3 PPI with an IC50 of 10.9 ꢀM, corresponding to a K
i
of 2.76 ꢀM. Given
7

ACCEPTED MANUSCRIPT
the ability of 3a to inhibit Mcl-1, we embarked on a structure-activity relationship (SAR) campaign, the
results of which are presented in Tables 1-3. The target molecules in Table 1 were all prepared
according to the concise synthetic route depicted in Scheme 2, those in Table 2 were synthesized using
anilines prepared (or purchased) as described in Scheme 3, whilst those in Table 3 were generated by
following the synthetic route in Scheme 4. SAR model development was based on visual inspection of
predicted binding orientations based on the SILCS modeling, with quantitative analysis based on the
SILCS ligand grid free energy (LGFE) scores, as described in more detail below.
1
1 2
3 2 2
Scheme 2. (a) ClSO H, 0 °C, 1 h; (b) ArNH , pyr, acetone, 50 °C, 3 h or R NH /R R NH, DIPEA,
acetone, RT, 5 h.
8

ACCEPTED MANUSCRIPT
Table 1. Experimental and computational Mcl-1 and Bcl-x
L
inhibitory profiles of first generation
inhibitors. Experimental data determined by a fluorescence polarization competition assay.
Mcl-1
Bcl-x
L
Mcl-1 LGFE
Bcl-xL LGFE
1
2
Compound
R
R
Selectivity
19
a
a
-1
-1
K
i
(µM)
K
i
(µM)
(kcal mol )
(kcal mol )
3
3
a
4-Br-C
6
H
4
H
H
H
2.76 ± 1.26
99.0 ± 11.2
56.3 ± 2.1
51.4 ± 31.3
NA
-39
-31
-39
-36
b
H
3
c
Bn
NA
2
CH -(2-Cl-
C H )
6 4
3
d
H
H
H
50.3 ± 11.8
5.03 ± 2.78
21.0 ± 3.9
-
-
-37
-39
-41
CH
2
-(3-Cl-
3
e
-
-
C H )
6 4
CH
2
-(4-Cl-
3
f
338 ± 203
16
11
-39
-41
C
6
4
H )
3
3
g
CH
2
-C
6
H
11
H
7.31 ± 1.42
25.7 ± 4.86
77.6 ± 14.3
NA
-42
-36
h
Me
Me
3
i
-CH
2
CH
2
CH
2
CH
2
CH
2
-
2.95 ± 0.43
-
-43
3
j
-CH
-CH
Ph
2-Br-C
3-Br-C
2
CH
2
N(Ph)CH
2
CH
2
-
4.54 ± 3.86
3.41 ± 0.37
106 ± 26
-
-45
-34
-37
-37
-39
-39
-40
-39
-43
-41
-44
-42
-40
-43
-39
-42
3
k
2
CH
2
N(Bn)CH
2
CH
H
H
H
H
H
H
H
H
H
H
H
H
H
H
2
-
-
3
l
NA
3
m
6
H
5
5
6.84 ± 2.23
5.64 ± 1.76
0.420 ± 0.163
11.2 ± 2.0
-
3
n
o
p
q
6
H
-
3
2,4-di-Br-C
6
H
4
2.31 ± 1.35
110 ± 45
68.5 ± 36.3
-
5.5
10
15
-38
-38
-37
3
3
1-Naphthyl
2-Naphthyl
4.58 ± 1.56
8.50 ± 1.34
1.88 ± 0.62
1.54 ± 0.46
4.05 ± 2.11
1.91 ± 0.26
2.50 ± 0.65
34.3 ± 1.0
3
r
2-Ph-C
3-Ph-C
4-Ph-C
2-CF -C
4-Cl-C
4-CF -C
4-Me-C
4-(iPr)-C
6 4
H
3
s
6 4
H
3
t
6
H
4
-
3
u
v
3
6
H
4
19.3 ± 9.0
4.8
-41
3
6
H
4
-
-
-
-
3w
3
6 4
H
3
3
x
y
6 4
H
6
H
4
3.86 ± 1.53
3
z
4-OMe-C
4-(OiPr)-C
4-CN-C
6
H
4
H
H
H
11.9 ± 2.5
54.0 ± 4.6
22.2 ± 1.6
71.9 ± 21.0
6
-38
-41
-37
-36
-38
3
aa
ab
6
H
4
NA
-
3
6
H
4
3
ac
4-NO
2
-C
6
H
4
H
H
4.49 ± 1.20
26.9 ± 0.6
23.3 ± 2.5
5.2
-39
-36
3
ad
3-CN-C H
6 4
a
K values determined by Nikolovska-Coleska equation from IC values.[54] Data represent the average
i
50
b
of at least two independent assays; errors are standard deviations. Selectivity is defined as the K (Bcl-
i
x ) divided by the K (Mcl-1). NA, no activity.
L
i
9

ACCEPTED MANUSCRIPT
Unsubstituted sulfonamide 3b was the weakest Mcl-1 inhibitor with a K of 99.0 µM, consistent
i
with the less favourable LGFE score as compared to 1 (Table 1, see supporting information).
Replacement of the NH group with benzylic amines and anilines improved Mcl-1 inhibitory activity in
2
every case, which is attributed to the occupancy of the apolar FragMaps with the apolar FragMap in the
p2 pocket yielding more favourable LGFE scores. In addition, the FragMaps extend further towards the
interior of the protein, a region partially occupied by the 4-chloro-3,5-dimethylphenyl of 2, suggesting
that further hydrophobic extension of benzylic amines would improve activity. Indeed, chlorobenzylic
amine derivatives 3d–3f were more potent than unsubstituted benzylic amine derivative 3c by up to 8-
fold. Replacement of the benzyl group in 3c with a cyclohexylmethyl group (3g) also afforded an 8-fold
i
improvement in inhibitory activity with a K of 7.31 µM. Piperidine derivative 3i achieved good
inhibition (K = 2.95 µM), as did the neutral and basic piperazines 3j and 3k, respectively, indicating a
i
positively-charged group here is not detrimental to binding. Substitution of the aniline ring in 3l with
bromine atoms afforded between 10 to 54-fold improvement in activity, with the most profound effect
observed with para-substituted derivative 3a (K = 2.76 µM cf 83.8 µM for 3l). Introduction of an
i
additional bromine atom into the ortho position of 3a to afford 2,4-dibromo derivative 3o led to an even
more potent inhibitor with a K of 420 nM. Concomitantly, however, this change also led to an erosion
i
in selectivity of more than ten-fold for Mcl-1 over Bcl-x . Naphthalene derivatives 3p and 3q were also
L
more active than unsubstituted 3l. Functionalization of the aniline ring in 3l with additional phenyl rings
to afford biphenyls 3r–3t resulted in improved inhibition in each case due to occupancy of the apolar
map adjacent to residues in the p2 pocket, which include V249 and M250, as well as residues in a non-
peptide-binding sub-pocket (A227 and M231) in the vicinity of the p3 pocket. Once again, the para-
substituted derivative 3u proved the most potent of the series (K = 1.54 µM). It is noteworthy that this
i
sub-pocket was recently shown to be important in the recognition of Fesik’s inhibitor 1.[28] For the
remainder of the compounds in Table 1, substitution of the aniline ring in 3l with hydrophobic and polar
1
0

ACCEPTED MANUSCRIPT
groups resulted in enhanced activities in all instances with the greatest improvements observed with
hydrophobic groups in the para position.
Motivated by the improved activity of extended hydrophobic substituents on the sulfonamide
phenyl ring (e.g. 3o and 3s) along with the extended apolar FragMap in the p2 pocket (Figure 2b), we
chose to expand our library of para-substituted anilines by introduction of various aryloxy groups into
the para position. Requisite anilines were synthesized according to Scheme 3, and then coupled to
sulfonyl chloride 5 according to Scheme 2. As shown in Table 2, compounds 3ba–3bi afforded potent
i
inhibition of Mcl-1 with the tightest binder delivering a K of 79 nM. This ~1000-fold enhancement in
inhibitory activity relative to unsubstituted aniline 3l is associated with the increased overlap of the
nonpolar groups with the second apolar FragMaps in the p2 pocket, as quantified by the systematically
more favourable LGFE scores (Table 2). Simultaneously, these modifications also greatly increased
affinity of the compounds to Bcl-x , which is associated with a change in the predicted binding
L
orientation with Bcl-x as discussed below. The sulfonamide NH group in 3bi offers an additional
L
position from which further inhibitory activity might be acquired associated with occupancy of the
apolar FragMap in the vicinity of residue V253, as shown in Figure 2B for 3a. Indeed, alkylation of this
nitrogen atom with cyclopentyl (3bk) and benzyl (3bl) resulted in our most potent inhibitors yet with
K ’s of 33 nM and 31 nM, respectively. Pleasingly, these alkylations also led to a recovery of selectivity
i
for Mcl-1 over Bcl-x of up to eleven-fold.
L
Scheme 3. (a) ArOH, K CO , 100 °C, 16 h; (b) SnCl •2H O, EtOAc, 50 °C, 16 h; (c) isobutyraldehyde,
2
3
2
2
benzaldehyde or cyclopentanone, NaBH(OAc) , 1,2-dichloroethane, RT, 16 h.
3
1
1

ACCEPTED MANUSCRIPT
Table 2. Experimental and computational Mcl-1 and Bcl-x inhibitory profiles of second-generation
L
inhibitors. Experimental data determined by a fluorescence polarization competition assay.
2
3
b
Compound
R
R
Mcl-1
Bcl-x
L
Selectivity
Mcl-1 LGFE
Bcl-x
L
LGFE
a
a
-1
-1
K
i
(µM)
K
i
(µM)
(kcal mol )
(kcal mol )
3
ba
bb
iBu
H
iPr
Ph
0.487 ± 0.035
1.15 ± 0.28
6.08 ± 0.81
ND
12
-
-39
-44
-44
-47
-45
-46
-47
-46
-47
-48
-40
-52
-36
3
3
bc
bd
be
bf
H
4-Me-C
6
H
4
0.335 ± 0.027
0.082 ± 0.012
0.114 ± 0.017
0.284 ± 0.125
0.079 ± 0.010
0.173 ± 0.070
0.117 ± 0.060
0.080 ± 0.019
0.033 ± 0.025
0.031 ± 0.017
0.84 ± 0.05
0.22 ± 0.06
0.30 ± 0.04
0.39 ± 0.10
0.19 ± 0.02
ND
2.5
3.7
3.8
1.4
2.4
-
-43
-48
-44
-46
-46
3
H
1-Naphthyl
3-Br-C
3,5-di-Me-C
2,4-di-Cl-C
4-Cl-C
3
H
6 4
H
3
H
6
H
3
3
bg
bh
H
6 3
H
3
H
6 4
H
3
bi
bj
H
4-Cl-3,5-di-Me-C
4-Cl-3,5-di-Me-C
4-Cl-3,5-di-Me-C
4-Cl-3,5-di-Me-C
6
6
6
6
H
2
0.48 ± 0.05
0.17 ± 0.06
0.29 ± 0.05
0.34 ± 0.13
4.1
2.1
8.8
11
-47
-47
-39
-51
3
iBu
Cp
Bn
H
2
3
bk
bl
H
2
3
H
2
a
i
K values determined by Nikolovska-Coleska equation from IC50 values.[54] Data represent the average
b
of at least two independent assays; errors are standard deviations. Selectivity is defined as the K
x ) divided by the K (Mcl-1). ND, not determined.
i
(Bcl-
L
i
Finally, we wanted to verify the importance of the carboxylic acid and hydroxyl groups of the
inhibitor scaffold. Analysis of Figure 2B showing overlap of the negative FragMaps with the acid
moiety indicate the importance of that group to binding while analysis of the acceptor FragMaps at a
1
2

ACCEPTED MANUSCRIPT
lower contour level of -0.3 kcal/mol (not shown) indicated small, but favourable contributions from the
alcohol moiety. Accordingly, we synthesized the compounds shown in Scheme 4. As seen in Table 3,
the carboxylic acid is critical to activity, which is consistent with its likelihood of forming a salt bridge
with R263. Methylation of the hydroxyl group to deliver compound 12 resulted in around a 40-fold
reduction in activity indicating its importance, too. Although SILCS predicted the methoxy derivative
1
2 (LGFE = -42 kcal/mol) to bind with similar affinity than its hydroxy counterpart 3ca (LGFE = -42
kcal/mol), the observed disparity may be explained by an anticipated greater acidity for the latter
through a six-membered intramolecular hydrogen bond between the hydroxyl and the carboxylate that
affords a stronger salt bridge with R263.
2 2 3
Scheme 4. (a) ArNH , pyr, acetone, 50 °C, 3 h; (b) MeI, K CO , DMF, rt, 16 h; (c) NaOH,
MeOH/THF/H O, 3:1:1, rt, 16 h.
2
Table 3. Investigation into the significance of the inhibitor’s carboxylic acid and hydroxyl group
towards the inhibition of Mcl-1. Experimental data determined by a fluorescence polarization
competition assay.
1
3

ACCEPTED MANUSCRIPT
Compound
X
Y
Mcl-1
Mcl-1 LGFE
a
-1
K
i
(µM)
(kcal mol )
3
ca
CO
2
H
OH
OH
0.566 ± 0.031
>500
-42
-34
-35
-42
1
0
1
2
CO
2
2
Me
Me
1
CO
OMe
OMe
>500
1
CO
2
H
23.4 ± 2.6
a
50
K values determined by Nikolovska-Coleska equation from IC values. Data represent the average
i
50
of at least two independent assays; errors are standard deviations.
1
5
172-327
NMR HSQC experiments using N-labeled Mcl-1
(recombinant human Mcl-1 residues
1
72 to 327) confirmed the direct interaction of this class of compounds with the protein. Specifically,
1
5
172-327
the HSQC spectra of N-labeled Mcl-1
(Figure 3: black = Mcl-1; red = Mcl-1 and inhibitor)
shows the addition of compound 3ba causes perturbations in the chemical shifts or loss of Mcl-1
1
15
backbone H- N correlations due to chemical exchange broadening in amino acids consistent with the
predicted binding mode of 3a shown in Figure 2. When the residues that are perturbed or lost due to
chemical exchange broadening are mapped onto the 3D structure of Mcl-1 (Figure 4) their positions
support the model shown in Figure 2. The majority of the changes occur in residues near the predicted
binding mode of the compound including R263 and N260, and residues in the p2 pocket (V249, M250,
and F270). There are also chemical shift perturbations near the p3 pocket (A227 and M231), although
these may be associated with indirect effects. The residues undergoing chemical shift changes are also
consistent with the X-ray crystal structure of 1 bound to Mcl-1 (PDB ID: 4HW3)[28] with most changes
located on the end of helix 2 (residues 204 to 224) and in most of helices 3 (residues 226 to 236), 4
(residues 239 to 255), and 5 (residues 260 to 280) that form the BH3-binding groove.
1
4

ACCEPTED MANUSCRIPT
Figure 3. The NMR HSQC spectra shows chemical shift perturbations caused by the direct interaction
of compound 3ba with Mcl-1. The spectrum of the control sample (black) is overlaid with the spectrum
of Mcl-1 bound to 3ba (red). Inset is enlargement of boxed region.
15

ACCEPTED MANUSCRIPT
Figure 4: Mcl-1 residues (blue) whose chemical shifts are perturbed or lost in the presence of
compound 3ba.
Comprehensive SILCS SAR analysis of Mcl-1 binding
Additional SILCS based modeling was undertaken to predict the bound orientation of the
active ligands, to quantitatively predict the relative affinities of the ligands as well as the
contributions of the different regions of the ligands to binding to Mcl-1 and Bcl-x and to
L
develop a quantitative model of differential binding of the compounds to Mcl-1 versus Bcl-xL.
This information would be of utility in designing ligands with improved potency and selectivity
for Mcl-1. Thus, subsequent modeling involved docking of each studied compound using the
SILCS-Pharm approach as described in the supporting information (Figure S1).[50,51] The
docked orientation of each compound was then subjected to MC-SILCS sampling to allow the
molecules to conformationally sample the local binding region as defined by the FragMaps and
the exclusion maps, with the LGFE scores calculated from the MC SILCS conformational
sampling. Comparison of the LGFE values for all the tested compounds with the experimental
16

ACCEPTED MANUSCRIPT
affinities converted to free energies (
ꢁG = -RTln
K
i
, where R is the Boltzmann constant and T is
2
the temperature) for Mcl-1 yields a correlation of R = 0.53 and a high predictive index[55] of
0
.70, (Figure S2, supporting information). Thus, the LGFE scores correlate with the Mcl-1
experimental data, indicating the quality of the bound orientations from the SILCS-Pharm-MC-
SILCS protocol.[49-51]
Given the predictive capabilities of the LGFE scores, further analysis was undertaken to
use the SILCS modeling to further interpret the experimental SAR data. This analysis focused on
four compounds: 3b, 3a, 3bi and 3bl. These compounds differ in the number of phenyl rings
attached to the sulfonamide moiety going from 0 to 3 rings, in that order, with the binding
2
affinities varying from 99
ꢀ
M down to 31 nM. For these compounds a correlation of R = 0.99
between the LGFE and experimental G values is obtained.
The predicted conformation of the four compounds is presented in Figure 5 along with
the crystallographic orientation of 1.[28] For all four compounds, the carboxylic acid overlaps
with the negative FragMap (orange arrows), associated with its interaction with R263 of Mcl-1
and the aromatic rings suitably overlap with hydrophobic FragMaps (green arrows), binding
within the hydrophobic groove. For 3b, the sulfonamide group is placed outside of the
hydrophobic binding pocket corresponding to an H-bond donor FragMap (blue arrow in (A))
indicating its possible H-bonding interaction with residue H224. The hydroxyl group on the
naphthalene ring overlaps with an H-bond acceptor FragMap (red arrow in (A)), possibly making
an H-bond to residue T266. For the other three compounds, the naphthalene group flipped over
relative to 3b, maintaining the position of the carboxylic acid while allowing the additional
aromatic groups to interact with the hydrophobic binding pocket, consistent with that observed
17

ACCEPTED MANUSCRIPT
with 1. Notably the sulfonamide oxygen atoms in this orientation overlap with an acceptor
FragMap associated with interactions with residue T266 (red arrows).
Figure 5. The orientations of the four compounds binding to Mcl-1 predicted by SILCS. (A) 3b;
(B) 3a; (C) 3bi; (D) 3bl. Compounds shown in stick format with atom-based colouring. The
compound, 2, from the previously reported complex structure 4HW3 is also shown, with the
carbon atoms coloured in pink. Mcl-1 residue R263 is shown. Same colour and GFE cutoff is
used as in Figure 1.
Additional analysis of the contribution of individual atoms and functional groups on the
compounds to binding was performed by analyzing the atom-based GFE contributions of the
individual atoms to the overall LGFEs. Presented in Figure 6 are the atom-based GFE
18

ACCEPTED MANUSCRIPT
contributions for the most favourable binding conformations of the four compounds. For all
compounds, the naphthyl and acid moieties make favourable contributions to binding. With 3b,
significant favourable contributions occur with both the sulfonamide and hydroxyl moieties. In
the remaining three compounds with the naphthyl moiety flipped; the sulfonamide and hydroxyl
moieties still make favourable contributions to binding though the magnitude is generally less
than that of 3b. For example, while some of the sulfonamide oxygens make more favourable
contributions with the larger compounds, the amide NH makes a very unfavourable contribution.
However, the less favourable interactions of the hydroxyl and sulfonamide moieties are
overcome by favourable contributions from the additional aromatic groups, which lead to the
improved binding of the larger inhibitors. Interestingly, the quite favourable contributions from
2
the second phenyl ring coming off the sulfonamide at the R position was counter-balanced by
1
decreased contributions from the aromatics group at the R position upon going from 3bi to 3bl.
This is consistent with the experimental free energy of binding changing by only -0.8 kcal/mol
versus 3bi while differences going from 3b to 3a and 3a to 3bi are -2.1 and -1.9 kcal/mol,
respectively. This atomic detailed interpretation of the experimental SAR data is anticipated to
facilitate further improvements in the compounds.
19

ACCEPTED MANUSCRIPT
Figure 6. Atom GFE contributions for the most LGFE favourable binding conformation of the
four compounds. (A) 3b; (B) 3a; (C) 3bi; (D) 3bl.
SILCS quantitatively captures the binding specificity of compounds for Mcl-1 over Bcl-x
L
A long-standing challenge in targeting the anti-apoptotic Bcl-2 proteins is achieving
family member specificity, particularly Mcl-1 specificity, although selective ligands are
beginning to emerge.[5,6,28,39,41,42] To investigate the binding specificity of compounds for
Mcl-1 versus Bcl-x , SILCS simulations and GFE FragMap generation followed by LGFE
L
20

ACCEPTED MANUSCRIPT
2
scoring were also done for Bcl-x
.58 and a high predictive index of 0.76 was found between the LGFE and experimental binding
data for Bcl-x (Figure S3 of the supporting information). Difference FragMaps, FragMaps,
FragMaps favour Mcl-1 over
FragMaps may then be used to qualitatively understand the contributions driving
specificity as well as quantitatively obtain difference LGFE scores ( LGFE). Correlation
L
(Table S1 of the supporting information). A correlation of R =
0
ꢁ
L
were then calculated as GFEBcl-xl – GFEMcl-1 such that positive
Bcl-x . The
ꢁ
ꢁ
L
ꢁ
analysis between the
Mcl-1 and Bcl-x
ꢁ
LGFE scores and the experimental differences in the binding affinities for
2
L
yielded an R = 0.57 and predictive index of 0.83 (Figure S4 in supporting
information), indicating the quality of the SILCS
ꢁFragMaps in modeling the relative affinities
for Mcl-1 versus Bcl-x . Additional analysis was therefore undertaken on 3a as this compound
L
showed the largest specificity for Mcl-1 over Bcl-x . Figure 7A shows the
ꢁ
FragMaps between
L
Mcl-1 and Bcl-x , and includes the docked orientation of 3a with the atom contributions to the
L
ꢁ
LGFE between Mcl-1 and Bcl-x mapped onto this orientation. The p-bromophenyl group
L
(
green arrow) that binds deeply in the p2 pocket between helices
α
4 and
FragMap indicating that region to favour binding to Mcl-1. Similarly, the sulfonamide oxygen
. Alternatively, the NH moiety
α5 is in a Mcl-1 positive
ꢁ
(red arrow) contributes to the binding of 3a to Mcl-1 over Bcl-x
L
of the sulfonamide contributes to more favourable binding to Bcl-x over Mcl-1. The acid
L
moiety in the lower right of the figure is also predicted to favour binding to Bcl-x over Mcl-1.
L
Summing over the different ꢁFragMap classes showed the overall ꢁLGFE score of 3.0 kcal/mol
favouring binding to Mcl-1 to have favourable contributions from the hydrophobic and neutral
acceptor terms, 4.4 and 1.6 kcal/mol, respectively, while the neutral donor, negatively charged
acceptor and hydroxyl groups favour Bcl-x
L
binding, with values of -1.8, -0.7 and -0.5 kcal/mol,
respectively. Similar trends are observed for the other compounds for which a selectivity of Mcl-
21

ACCEPTED MANUSCRIPT
1
over Bcl-x
L
was found, as shown in Table S2 in the supporting information. Indeed, given that
all the tested compounds favoured Mcl-1 binding over Bcl-x , these results indicate that the
L
exploitation of the binding pocket inherently favours Mcl-1 binding, as previously discussed by
Fesik.[28] The docked orientation of 3a to the two proteins in Figure 7B shows that this pocket is
not being exploited with Bcl-x . This is consistent with the fact that Mcl-1 shows more opening
L
at the p2 pocket as compared to Bcl-x as illustrated in Figure 7B,[28] disallowing access of 3a
L
L
deep into the hydrophobic pocket in the p2 site of Bcl-x .
However, further extension of the ligands (e.g. 3bl) leads to improved binding affinity,
but not improved specificity. This may be explained by different predicted binding orientations
of the compound to the two proteins. As seen in Figure 7C, 3bl binds with its additional 4-
3
2
chloro-3,5-dimethylphenyl R group deep in the p2 pocket of Mcl-1, and with the R benzyl
group occupying the upper region of the p2 pocket. In contrast, the orientation of 3bl bound to
Bcl-x shows the 4-chloro-3,5-dimethylphenyl still interacting with the p2 site, though not as
L
deep as that occurring with Mcl-1. This leads to a shift in the location of the naphthyl ring as
2
well as the R phenyl group into the p3 site where they can exploit the hydrophobic character of
that sub-pocket as well as interactions with R263, as indicated by the apolar and negative Bcl-x
L
FragMaps in Figure 7D, respectively. Thus, the significant increase in the size of 3bl leads to
increased affinity with respect to both Mcl-1 and Bcl-x , with that larger size predicted to change
L
the binding orientation with Bcl-x such that further increased specificity for Mcl-1 is not gained
L
over 3a.
22

ACCEPTED MANUSCRIPT
Figure 7. (A) Difference FragMaps (ꢁFragMaps) between Mcl-1 and Bcl-xL showing the
favourable binding patterns for Mcl-1. Hydrophobic and neutral donor ꢁFragMaps are shown in
green and blue, respectively. The atom GFE contributions to the binding affinity difference are
23

ACCEPTED MANUSCRIPT
L
also shown. (B) Experimental structures of Mcl-1 (blue, 4HW4) and Bcl-x (red, 1BXL) shows
that the binding pocket between helix α4 (upper right) and α5 (bottom left) is larger for Mcl-1
than Bcl-x . The predicted binding modes of 3a for Mcl-1 (blue) and Bcl-x (red) differs due to
L
L
this p2 pocket difference. The 4-bromophenyl ring of 3a binds deeply to the opened pocket for
Mcl-1 while it binds on the protein surface for Bcl-x . (C) Binding orientations of 3bl to Mcl-1
L
(
blue, 4HW4) and Bcl-x (red, 1BXL) along with the ꢁFragMaps. (D) Binding orientations of
L
3
L L
bl to Bcl-x (1BXL) along with the Bcl-x APOLAR and NEG FragMaps at the contour levels
specified in Figure 1.
Cellular activity of Mcl-1 inhibitors
.
To investigate the cellular activity of our Mcl-1 inhibitors, the most potent compound 3bl
was evaluated for its ability to inhibit the proliferation of A375 and SK-MEL-5 melanoma cells,
both of which express high levels of Mcl-1.[57] Unfortunately, the IC50 for 3bl was 50
µM and
90 µM in A375 and SK-MEL-5 cells, respectively (Table 4). We attributed the poor efficacy in
cells to the charged carboxylic acid of 3bl. To test this hypothesis, we prepared two ester
prodrugs of the potent, but more hydrophilic, inhibitor 3ba, methyl ester 13 and acetoxymethyl
ester 14, according to Scheme 3. As expected, both prodrugs 13 and 14 demonstrated no binding
affinity to Mcl-1 in vitro. (data not shown). Naphthoate derivative 3ba exhibited worse activity
than 3bl in cells, consistent with its weaker affinity to Mcl-1 (Table 3). However, the
acetoxymethyl ester 14 displayed an IC50 of 15
µM in both A375 and SK-MEL-5 cells,
presumably due to improved cell penetration and then intracellular hydrolysis to the active
metabolite 3ba (Table 4). We postulate that the lack of cellular activity for methyl ester 13 might
be due to limited hydrolysis to 3ba, as we have observed elsewhere.[58] These results suggest a
24

ACCEPTED MANUSCRIPT
path forward with respect to rationally improving the cell penetration capabilities of this class of
compounds based on a prodrug strategy.
Scheme 5. (a) SOCl , MeOH, rt, 48 h; (b) BrCH OCOCH , K CO , MeCN, 0 °C, 16 h.
2
2
3
2
3
Table 4. Inhibition of cell proliferation by select compounds, as determined by a cell titer blue
assay.
25

ACCEPTED MANUSCRIPT
Conclusions
Presented is a novel class of Mcl-1 inhibitors that target R263 and the p2 and p3 pockets within
the hydrophobic BH3-binding crevice on the surface of the protein. Inhibitor design was
motivated by compounds reported by Fesik and coworkers and by AbbVie that are known to
exploit the p2 pocket and interactions with R263. SILCS analysis of this region of the BH3
binding site supported this strategy, with the use of the SILCS exclusion maps indicating
significant additional opening of the p2 pocket that could be exploited in ligand design.
Accordingly, a series of compounds were designed, synthesized and subjected to experimental
evaluation using a FP assay with binding of 3ba to targeted pocket of Mcl-1 verified using
NMR. From this strategy, 3bl was identified with a K = 31 nM for Mcl-1 with a specificity of
i
1
1-fold over Bcl-x . Although only moderate cell activity was observed with the 2-
L
hydroxynaphthoate-based Mcl-1 inhibitors, this was enhanced by the preparation of an
acetoxymethyl ester prodrug of the critical carboxylic acid.
The two-step synthesis developed in this study allowed us to rapidly synthesize and test
over 40 analogs, with the selected compounds based, in part, on commercially available starting
materials. These synthetic capabilities allowed us to readily synthesize compunds that would be
able to reach into and complement the hydrophobic p2 pocket. This yielded compounds with
affinities spanning over a 10,000-fold range. Detailed SILCS analysis of all of these compounds
quantified the contributions of the different moieties leading to improved affinity. These results
3
suggest that fine tuning of interactions with the p2 pocket via appropriate modifications of the R
1
ring (Table 2) as well as of the R ring will likely lead to improved affinity. Additional SILCS
L
analysis on the differential binding of selected compounds to Mcl-1 versus Bcl-x indicates that
3
fine tuning of the R ring occupying the p2 pocket may also improve specificity. Further
26

ACCEPTED MANUSCRIPT
modifications that may contribute to improved specificity include alkylation of the NH moiety of
the sulfonamide moiety and additions of both hydrogen bond acceptor (e.g. OMe) or
hydrophobic (e.g. Cl) functional groups to the naphthyl ring as indicated by the differential
Fragmaps in the top, central region of Figure 7A. Future studies will involve the design of new
compounds using the validated SILCS affinity and specificity models for quantitative analysis of
synthetically accessible compounds.
Experimental
Computational
Molecular modeling studies were initiated with the crystal structure of the Mcl-1-BH3 peptide
complex (PDB ID: 4HW4),[28] following removal of the 16mer BH3 peptide, and with the Bcl-
xL-Bak complex structure (PDB ID: 1BXL), following removal of the Bak peptide. For both
proteins, the Reduce software[59] was used to choose optimal Asn, Gln, and His side-chain
orientations and determine the optimal protonation states of His residues. The protein was
immersed in a box of an aqueous solution containing eight small probe molecules at
approximately 0.25 M each with water at ~55 M. The size of the simulation box was chosen so
as to have the protein extrema separated from the edges by a minimum of 8 Å based on non-
hydrogen atoms. The small molecules, or solutes, include benzene, propane, methanol,
formamide, acetaldehyde, methylammonium and acetate, as previously used,[49] along with
imidazole. Ten such protein-small molecule aqueous systems were generated for each protein
with each simulation system differing in the initial positions and orientations of the small
molecules and water. The SILCS[47-50] molecular dynamics (MD) simulations were performed
27

ACCEPTED MANUSCRIPT
using the GROMACS[60] simulation program with the CHARMM36 force field,[61]
CHARMM general force field (CGenFF)[62,63] and TIP3P water model[64] to describe the
protein, small molecules and water, respectively. The simulations were each extended for 40 ns,
yielding a total of 400 ns of simulation time for each protein. Additional MD simulation details
can be found in Reference 49.[49]
From the simulation, 3D probability distributions of selected atoms from the small
molecules were constructed to form the FragMaps, yielding a total of ten different FragMaps.
The final FragMaps are converted to free energies, termed grid free energies (GFE), by
normalizing the distributions with respect to the distributions of the solutes in an aqueous
solution in the absence of the proteins followed by Boltzmann transformation to yield the GFE
values.[49] As the GFE FragMaps are normalized with respect to the fragment probabilities in
solution, they contain both favourable regions as well as unfavourable regions. The unfavourable
regions typically range from 0 to 3 kcal/mol, with the upper limit based on sampling issues. In
addition to the FragMaps, exclusion maps were constructed by calculating the 3D probability
distributions of all non-hydrogen atoms of the water and solutes together and identifying those
voxels with zero occupancies, which defines the exclusion maps. This exclusion map represents
regions forbidden to the small molecules and water and may be considered an alternate to more
traditional representations of the protein surface. For quantitative analysis, these voxels were
assigned a very high energetic penalty (1000 kcal/mol) while the remaining voxels were assigned
energies associated with the specific FragMaps. The availability of the GFE FragMaps and the
exclusion maps allows for quantitative estimates of binding affinities to be made, referred to as
ligand grid free energies (LGFE). These are a simple summation of the GFE energy contribution
of all the atoms in each ligand that are classified with respect to the FragMap types followed by
28

ACCEPTED MANUSCRIPT
normalization of the summed energies by the number of classified ligands atoms and subsequent
multiplication by the total number of non-hydrogen atoms, yielding the final LGFE values. Note
that the LGFE scores are not directly equivalent to experimental free energies due to the additive
approximation of the LGFE scores (i.e. the individual atom-based GFEs are summed to yield the
LGFE), the lack of accounting for the energy cost of connecting the fragments that comprise the
full compounds and issues associated with the standard state in the experimental and
computational conditions.
In the present study the specific FragMaps were used for calculation of GFE and LGFE
scores, while generic FragMaps are used for visualization. The specific and generic FragMap
types that were used include: (1) aromatic, AROM (benzene carbons); (2) aliphatic, ALIP
(propane carbons); (3) dual role hydrogen bonding atom, MEOO (methanol oxygens); (4)
FORN, (formamide nitrogen); (5) FORO, (formamide oxygen); (6) IMIN, (imidazole acceptor
nitrogen); (7) IMIH, (imidazole donor nitrogen); (8) AALO, (acetaldehyde oxygen); (9) positive
donor, POS (methylammonium nitrogen); (10) negative acceptor, NEG (acetate oxygens); (11)
generic nonpolar, APOLAR (benzene and propane carbons); (12) generic neutral hydrogen-bond
donor, HBDON (formamide and imidazole donor nitrogen); and (13) generic neutral hydrogen-
hydrogen acceptor, HBACC (formamide, acetaldehyde oxygens and imidazole acceptor
nitrogen). The convergence of the FragMaps was examined by calculating the overlap coefficient
between two sections of the simulations (trajectories 1–5 and trajectories 6–10), as previously
described.[49] All the generic FragMaps show an overlap coefficient of greater than 0.6,
indicating satisfactory convergence.
To identify the binding modes of the tested compounds with Mcl-1, the SILCS-Pharm
protocol[50,51] was used in which the FragMaps are used to define the pharmacophore features.
29