Synthesis and Biopharmaceutical Evaluation of Imatinib Analogues Featuring Unusual Structural Motifs

Article abstract of DOI:10.1002/cmdc.201500510

A convenient synthesis of imatinib, a potent inhibitor of ABL1 kinase and widely prescribed drug for the treatment of a variety of leukemias, was devised and applied to the construction of a series of novel imatinib analogues featuring a number of non-aromatic structural motifs in place of the parent molecule's phenyl moiety. These analogues were subsequently evaluated for their biopharmaceutical properties (e.g., ABL1 kinase inhibitory activity, cytotoxicity). The bicyclo[1.1.1]pentane- and cubane-containing analogues were found to possess higher themodynamic solubility, whereas cubane- and cyclohexyl-containing analogues exhibited the highest inhibitory activity against ABL1 kinase and the most potent cytotoxicity values against cancer cell lines K562 and SUP-B15. Molecular modeling was employed to rationalize the weak activity of the compounds against ABL1 kinase, and it is likely that the observed cytotoxicity of these agents arises through off-target effects.

Full text of DOI:10.1002/cmdc.201500510

DOI: 10.1002/cmdc.201500510
Communications
Synthesis and Biopharmaceutical Evaluation of Imatinib
Analogues Featuring Unusual Structural Motifs
[
a]
[a, b]
[a]
Kyriacos C. Nicolaou,* Dionisios Vourloumis,*
Sotirios Totokotsopoulos,
[b]
[c]
[c]
[c]
Athanasios Papakyriakou, Holger Karsunky, Hanan Fernando, Julia Gavrilyuk,
Damien Webb, and Antonia F. Stepan*
[d]
[d]
A convenient synthesis of imatinib, a potent inhibitor of ABL1
kinase and widely prescribed drug for the treatment of a variety
of leukemias, was devised and applied to the construction of
a series of novel imatinib analogues featuring a number of
non-aromatic structural motifs in place of the parent mole-
cule’s phenyl moiety. These analogues were subsequently eval-
uated for their biopharmaceutical properties (e.g., ABL1 kinase
inhibitory activity, cytotoxicity). The bicyclo[1.1.1]pentane- and
cubane-containing analogues were found to possess higher
themodynamic solubility, whereas cubane- and cyclohexyl-con-
taining analogues exhibited the highest inhibitory activity
against ABL1 kinase and the most potent cytotoxicity values
against cancer cell lines K562 and SUP-B15. Molecular model-
ing was employed to rationalize the weak activity of the com-
pounds against ABL1 kinase, and it is likely that the observed
cytotoxicity of these agents arises through off-target effects.
Marketed as Gleevec in its mesylate salt form, imatinib (Im-1,
Figure 1) is a widely used anticancer drug. It is effective against
a variety of leukemias (e.g., chronic myeloid leukemia (CML),
acute lymphoblastic leukemia (ALL), chronic myelomonocytic
leukemia, chronic eosinophilic leukemia), as well as gastrointes-
[
1]
tinal stromal cancer. Being the first tyrosine kinase inhibitor
TKI) to be approved for clinical use, this agent signaled a new
paradigm for targeted chemotherapy that now encompasses
(
1
3 clinically approved TKIs covering 13 of the 90 protein tyro-
[2]
sine kinases in the human kinome.
[
a] Prof. K. C. Nicolaou, Dr. D. Vourloumis, Dr. S. Totokotsopoulos
Department of Chemistry, Rice University
100 Main Street, Houston, TX 77005 (USA)
E-mail: kcn@rice.edu
dionysios.vourloumis@rice.edu
6
[
b] Dr. D. Vourloumis, Dr. A. Papakyriakou
Laboratory of Chemical Biology of Natural Products & Designed Molecules
National Centre of Scientific Research (NCSR) “Demokritos”
Agia Paraskevi, Athens 15310 (Greece)
E-mail: d.vourloumis@inn.demokritos.gr
[
c] Dr. H. Karsunky, H. Fernando, Dr. J. Gavrilyuk
Stemcentrx, Inc., 450 East Jamie Court, San Francisco, CA 94080 (USA)
[
d] Dr. D. Webb, Dr. A. F. Stepan
Pfizer Worldwide Research & Development
Figure 1. Molecular structures of imatinib (Im-1) and designed analogues
Im-2–Im-11.
6
10 Main Street, Cambridge, MA 02139 (USA)
E-mail: Antonia.Stepan@pfizer.com
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500510.
ChemMedChem 2016, 11, 31 – 37
31
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Imatinib (Im-1) has a high density of aromatic rings as mea-
3
sured by its low fraction of sp -hybridized carbon atoms
[
3]
(
0.24) and relatively high calculated lipophilicity (clogP=
[
4]
4
.53). This combination of properties is reflected in its very
À1
low aqueous solubility (0.01 mgmL ), making the free base
[5]
nearly insoluble in neutral aqueous media. Although the me-
sylate salt of Im-1 (i.e., Gleevec) is soluble in water at pH<5.5,
we considered imatinib as the free base (Im-1) to be a good
model system for evaluating the biopharmaceutical properties
(
e.g., thermodynamic solubility, metabolism, ABL1 kinase inhib-
itory potency) of imatinib analogues containing a number of
3
sp three-dimensional rigid or flexible structural motifs as bioi-
sosteres of the 1,4-disubstituted phenyl ring of the parent mol-
ecule (Im-1). The aim of our study was to determine the feasi-
bility of improving the pharmacological properties of drugs
and drug candidates by substituting excess aromatic systems
3
[6]
within the molecule with sp structural motifs. Herein, we de-
scribe the design, synthesis and biopharmaceutical evaluation
of a series of imatinib analogues (Im-2–Im-11, Figure 1).
The molecular design of these analogues was inspired from
3
Figure 2. Calculated property (clogP and fraction sp ) space of imatinib (Im-
1) and designed analogues (Im-2–Im-11).
[7]
a previous report by one of us (AFS) on the beneficial effect
of the replacement of the central fluorophenyl ring of the g-
secretase inhibitor BMS-708,163 with the bicyclo[1.1.1]pentane
moiety on the thermodynamic solubility (>30-fold improve-
[
10,11]
2!Im-1, Scheme 1B).
complicated the situation even fur-
ther. A brief investigation in order to circumvent these prob-
lems led to the finding that reductive amination of methyl 4-
formylbenzoate (1a) with 1-methylpiperazine (2), induced by
sodium triacetoxyborohydride (step f, Scheme 1C), furnished
piperazine derivative 1b in 98% yield, which compares favora-
bly to the yield (70%) obtained under the conditions em-
[
5]
ment), metabolic stability and passive permeability. Ana-
logues Im-2–Im-11 (Figure 1) were thus designed for the pres-
ent study in order to probe the generality and scope of this
type of substitution as a means to improve the biopharma-
ceutical properties of potential drug candidates.
[
12]
ployed in an earlier report for the synthesis of 1b (Pt/C, H
(5 bar), 808C, 20 h; Scheme 1A). Furthermore, the final cou-
pling of methyl ester piperazine derivative 1b with aniline de-
rivative 7 (prepared in four steps from 6; Scheme 1B)
ceeded smoothly under the influence of trimethylaluminum
2
[
9]
These analogues include, in addition to the bicyclo[1.1.1]-
pentane system (Im-4), cyclopropane (Im-7 and Im-8), cyclobu-
tane (Im-2 and Im-3), cyclopentane (Im-5), cyclohexane (Im-9),
[
11,13]
pro-
[
8]
n-butane (Im-10), n-pentane (Im-11) and cubane (Im-6) struc-
tural motifs to allow for either increased three-dimensionality
and rigidity or conformational flexibility, in addition to an over-
[
9]
(step g, Scheme 1C; see also 1b+3!4, Scheme 1A ) to
afford easily purified imatinib (Im-1) in 74% yield (unopti-
mized). This sequence proved overall more convenient and ef-
3
3
3
all increased fraction sp (fsp ), defined as the number of sp
carbons/total number of carbons. The calculated fraction sp
3
[9–11,13]
ficient for our purposes than those previously reported.
and logP (clogP) parameters for these designed analogues are
shown in Figure 2, clearly indicating their expected shift, in
comparison with the parent molecule imatinib (Im-1), into
a more polar space (Im-2–Im-11: clogP=1.46–3.30 vs Im-1:
For the synthesis of the designed imatinib analogues, key
building blocks 8, 9 and 15–19 were prepared from starting
[
14]
[7,15–18]
materials 8a and 10a–14a
via intermediates 10–14,
respectively, as summarized in Scheme 2. Carboxylic acid
3
3
clogP=4.53) and higher sp character (Im-2–Im-11: fsp =
methyl esters 10–12 were selectively reduced with borane–tet-
3
[19]
0
.39–0.48 vs Im-1: fsp =0.24).
rahydrofuran to the desired hydroxy esters 15–17. Alcohols
Our initial attempts to synthesize the targeted imatinib ana-
13 and 14 were oxidized in one step to the corresponding car-
[
20]
logues through literature procedures were met with difficulties
associated with contamination and reactivity issues of inter-
mediates and final products. Specifically, the reported se-
quence for the conversion of 1a to imatinib (Im-1) suffered
from problems due to palladium impurities in the final step
boxylic acids, which were then methylated with (trimethylsi-
lyl)diazomethane and desilylated with triethylamine trihydro-
fluoride to afford hydroxy esters 18 and 19, respectively. Hy-
droxy methyl ester 20 was prepared by methanolysis of d-va-
[
21]
lerolactone, whereas hydroxy methyl ester 21 was obtained
from trans-4-(hydroxymethyl)cyclohexanecarboxylic acid (21a)
by methylation using (trimethylsilyl)diazomethane.
[9]
(
4+5!Im-1, Scheme 1A), while the reductive amination of
[9]
1
a (1a+2!1b, Scheme 1A) was not easily applicable to the
[
22]
aldehydes involved in the present work due to harsh condi-
tions (i.e., anticipated decomposition and, in some cases, epi-
merization). In addition, the expected decreased stability and
reactivity of the required alkyl halides in the projected nucleo-
philic substitutions according to the reported routes for acti-
vated benzylic and/or benzoyl chlorides (7+1c or 1d; then
Alcohols 8, 9 and 15–21 were oxidized by Swern or Dess–
[
23]
Martin periodinane protocols, and the resulting crude alde-
hydes were reductively aminated with 1-methylpiperazine (2),
furnishing amino esters 22a–i as shown in Scheme 3. Addition-
ally, reductive amination of commercially available methyl 6-
oxohexanoate [MeO C(CH ) CHO] furnished amino ester 22j.
2
2 4
ChemMedChem 2016, 11, 31 – 37
www.chemmedchem.org
32
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Scheme 2. Synthesis of key building blocks 8, 9 and 15–21: Reagents and
Scheme 1. Improved synthesis of imatinib (Im-1); A and B) previous work
and C) this work. Reagents and conditions: a) Pt/C, H (5 bar), MeOH, 808C,
0 h, 70%; b) Me Al, toluene, 408C, 0.5 h, 75%; c) Pd (dba) ·CHCl , rac-BINAP,
sodium tert-butylate, xylene, reflux, 5 h, 72%; d) 1c, pyridine, 238C, 24 h,
yield not reported; e) 1d, Et N, THF, 08C, 3 h, 93%; then 1-methylpiperazine
2), reflux, 3 h, 91%; f) 2 (1.0 equiv), NaBH(OAc) (1.6 equiv), ClCH CH Cl,
58C, 2 h, 98%; g) Me Al (3.0 equiv), 7 (1.0 equiv), toluene, 408C, 0.5 h, 74%
unoptimized).
conditions: a) BH
for 16, 84% for 17; b) NaIO
O (1:1:1.5 v/v), 258C, 2 h, 65%; c) (CH
3:2), 258C, 0.5 h, quant.; d) HF·Et N (3.0 equiv), THF, 258C, 6 h, 71% for 18,
0% for 19.
3
·THF (1.0 equiv), THF, À20!258C, 16 h, 92% for 15, 92%
(4.0 equiv), RuCl ·xH O (3% mol), CH CN/CCl
SiCHN (1.2 equiv), toluene/MeOH
2
4
/
4
3
2
3
2
3
2
3
3
H
2
3
)
3
2
(
3
3
7
(
2
(
3
2
2
3
The synthesized compounds were evaluated for their bio-
pharmaceutical properties (the calculated and experimental
data along with the presented abbreviations are shown
Table S2 in the Supporting Information). Thus, the Pfizer in-
house data confirmed the low aqueous solubility of Im-1,
which was determined to be 30.7 mm at pH 7.4. This is in line
Amino esters 22a–j were then coupled with amine 7 in the
presence of trimethylaluminum in toluene at 408C, affording
the target imatinib analogues Im-2–Im-11 in high overall
yields as shown in Scheme 3.
À1
The absolute stereochemical configurations of analogues
Im-2 and Im-3, and their respective precursors 22a and 22b
with the previously published data of 0.01 mgmL (~20 mm,
[
5]
see above). Pleasingly, the key bicyclo[1.1.1]pentane analogue
(Im-4) exhibited greater than 80-fold improved aqueous solu-
bility (2506 mm at pH 7.4), a result that confirms the initial find-
(
Scheme 3), were determined through X-ray crystallographic
[
24]
analysis (Scheme 3C) of bromo-derivative 23. The latter was
prepared and converted to imatinib analogue Im-2 as shown
in Scheme 3B. Thus, reaction of methyl ester 22a with trime-
[
7]
ing of the g-secretase inhibitor program and expectation of
this study. This improvement in solubility is in accordance with
that observed with compounds of similar three-dimensionality,
[9]
thylaluminum, as described above, yielded crystalline amide
3 (m.p. 175–1778C, CH CN), in which the relative and abso-
3
2
as indicated by the fraction sp and polarity, represented by
3
[
4]
lute configuration was proved to be the one shown, thereby
leading to the indicated stereochemical assignments for ana-
logues Im-2 and Im-3 and their precursors.
both the calculated (clogP) and measured shake flask logD
(SFlogD) lipophilicity values (Table S2 in the Supporting Infor-
mation). For example, cyclobutyl analogue Im-3 exhibited
ChemMedChem 2016, 11, 31 – 37
www.chemmedchem.org
33
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
hybridized equivalents due to increased p-stacking, an intrinsi-
[
25]
cally higher lipophilicity, or a combination of both factors. It
3
is assumed that sp density and hydrophobicity correlate, re-
spectively, with crystal packing and hydration free-energy con-
[
26]
tributions to aqueous solubility. Deconvolution of the rela-
tive importance of these contributions allows the understand-
ing of solubility-limiting features for effective optimization of
solubility.
To understand whether the observed solubility changes in
the case of our imatinib analogues are likely due to changes in
lattice energy and/or lipophilicity, we also included melting
points (m.p.) in our analysis. Using lipophilicity and melting
[
27]
point as the key parameters, Jain and Yalkowsky developed
the general solubility equation (GSE) as a tool for approximat-
ing thermodynamic solubility. The GSE equation is defined as
logS=0.5ÀlogPÀ0.01 (m.p.À25), and it has been previously
proposed that the low solubility of a compound with a m.p. of
ꢀ
2508C and logP value of ꢁ2 is likely limited due to a solid-
state effect, while compounds with a logP value of ꢀ3.5 usual-
ly fall into the solvation-limited solubility category. As indicated
by a clogP value of 4.5 and a m.p. of 2048C, the relatively low
thermodynamic solubility of imatinib free base (30.7 mm at
pH 7.4) is presumed to be a result of a lipophilicity-limited sol-
vation. Improvements in solubility over the parent imatinib
will, therefore, likely be driven by decreases in lipophilicity and
not disruption of the intermolecular forces stabilizing the crys-
talline state (as represented by the melting points). This is con-
firmed with the set of analogues Im-2–Im-11. Thus, while our
structural modifications were designed to impact both the
three-dimensionality and lipophilicity of the parent compound
(
Im-1) as outlined above (Figure 2), the thermodynamic solubil-
ity tracks with the respective lipophilicities, rather than the
melting points. For example, replacing the central phenyl
group in imatinib (Im-1) with the bicyclo[1.1.1]pentane moiety
3
(
(
i.e., Im-4) leads to an almost twofold increase in fraction sp
Scheme 3. A) Completion of the synthesis of imatinib analogues Im-2–Im-
1: a) oxalyl chloride (5.0 equiv), DMSO (10 equiv), Et N (25 equiv), CH Cl
À78!258C, 2 h, 63–68% yield; or Dess–Martin periodinane (1.2 equiv),
NaHCO (4.0 equiv), CH Cl , 258C, 1 h; b) 1-methylpiperazine (2, 1.0 equiv),
NaBH(OAc) (1.6 equiv), ClCH CH Cl, 258C, 2 h, 60–82% (two steps); c) Me
3.0 equiv), 7 (1.0 equiv), toluene, 408C, 0.5 h, 70–88%. B) Alternative synthe-
sis of Im-2: d) Me Al (3.0 equiv), aniline (3, 1.0 equiv), toluene, 408C, 0.5 h,
8%; e) 5 (1.0 equiv), NaOtBu (1.5 equiv), rac-BINAP (5% mol),
Pd (dba) ·CHCl (1.3% mol), xylene, reflux, 5 h, 50%. C) X-ray-derived ORTEP
representation of compound 23 showing the atom numbering scheme.
Thermal ellipsoids at 60% equiprobability envelopes; C=grey, O =red,
N=blue, H=green. X-ray crystallographic analysis of 23 provides proof of
relative configurations of 23, Im-2, and by deduction Im-3.
3
3
Im-1: fsp =0.24 vs Im-4: fsp =0.43), but this replacement
1
3
2
2
,
does not translate into a significant decrease in melting point
[
28]
(
Im-1: mp=2048C vs Im-4: m.p.=195–1988C). The thermo-
dynamic solubility at pH 7.4, however, increased by >80-fold
Im-1: 30.7 mm at pH 7.4 vs Im-4: 2500 mm at pH 7.4), which
3
2
2
3
2
2
3
Al
(
(
3
can be rationalized with a decrease in lipophilicity [Im-1:
clogP=4.53, SFlogD (pH 7.4)=2.45 vs Im-4: clogP=1.93,
SFlogD (pH 7.4)=1.51]. As noted above, this solubility im-
provement imparted by the unusual bicyclo[1.1.1]pentane
moiety is in line with the changes observed upon replacing
the phenyl group with bioisosteres commonly used by medici-
nal chemists. For example, cyclopropyl- and cyclobutyl ana-
logues Im-7, Im-2 and Im-3 all have significantly decreased lip-
ophilicity (clogP of Im-7, Im-2 and Im-3 is greater than twofold
decreased relative to Im-1) and similar melting points, but
thermodynamic solubility values more than 50-fold higher
than that of the parent molecule (Im-1). As indicated by both
the calculated and measured values, the lipophilicity of the un-
usual cubane imatinib analogue (Im-6) is in line with the con-
ventional aliphatic analogues (Im-6: clogP=1.46, SFlogD
(pH 7.4)=1.67), despite the high carbon count of the cubane
moiety, and more than twofold lower than that of Im-1 [Im-1:
7
2
3
3
[
24]
a thermodynamic solubility (1840 mm at pH 7.4) within twofold
3
of that of Im-4 and a fraction sp and clogP/SFlogD in line
3
with that of Im-4 (Im-3: fsp =0.41, clogP/SFlogD=2.18/1.33
3
vs Im-4: fsp =0.43, clogP/SFlogD=1.93/1.51). The same ap-
plies to cyclopropyl analogue Im-7 (thermodynamic solubility:
3
3
720 mm at pH 7.4, fsp =0.39, clogP/SFlogD=1.85/1.35) and,
to some extent, cubane analogue Im-6 (thermodynamic solu-
3
bility: 383 mm at pH 7.4, fsp =0.48, clogP/SFlogD=1.46/1.67).
2
As outlined previously, compounds with high sp character
3
are thought to suffer from solubility issues relative to their sp -
ChemMedChem 2016, 11, 31 – 37
www.chemmedchem.org
34
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
clogP=4.53, SFlogD (pH 7.4)=2.45]. The decreased lipophilici-
ty of analogue Im-5 translates into an increased thermodynam-
ic solubility relative to Im-1, although the extent of the im-
provement (10-fold) is smaller relative to that of the previously
tration tested (Table S3). Overall, the cytotoxicity profiles of Im-
6 and Im-9 resembled the activity of imatinib the closest in
highly imatinib-sensitive cell lines and in the mildly sensitive
SUP-B15 cell line. Only moderate levels of activity were
reached in K562 CML cell line. The observed differences in ac-
tivities between the cytotoxicity and ABL1 assays for Im-6 and
Im-9 may be indicative of off-target binding that contributes
to the overall effect.
[29]
discussed analogues (>50-fold).
In this study, it was also found that the impact of a decrease
in the lipophilicity on endpoints other than thermodynamic
solubility is not significant, and changes in the intrinsic perme-
ability and resistance towards metabolic turnover (measured in
both human liver microsomes and cryopreserved human hepa-
tocytes) lie within the experimental error of the assay (~2–3-
The significant loss of binding affinity for the targeted kinase
observed for compounds Im-2–Im-11 may be explained con-
sidering the binding mode of the parent imatinib (Im-1) within
the binding pocket of ABL1. Thus, as shown in Figure 3, in the
X-ray crystallographic structure of the ABL1–imatinib complex
[
30,31]
fold).
For example, the intrinsic permeability of analogues
Im-2–Im-11 in the Ralph Russ canine kidney (RRCK) permeabili-
[
31]
ty assay, conducted in Madin–Darby canine kidney cells, was
(PDB ID: 2HYY), Im-1 is stabilized through a number of hy-
drogen-bonding interactions (such as the ones indicated with
dotted lines in Figure 3A,B; see also Supporting Information):
À6
À1
found to be between 4.62 and 11.9”10 cms , lying in the
same range as that of the parent compound (Im-1: RRCK=
À6
À1
pyr
8
.28”10 cms ). This is in contrast to the previous observa-
i) the pyridine nitrogen (N ) and the backbone of Met318 in
[7]
an
tions in the g-secretase inhibitor program in which the corre-
sponding bicyclo[1.1.1] analogue exhibited a greater than
threefold increase in both RRCK permeability and metabolic
the hinge region; ii) the anilinic amine (N ) and the hydroxy
group of the “gatekeeper” residue Thr315; iii) the amide NH
am
(N ) and the carboxylate group of Glu286; iv) the amide car-
[30,31]
stability relative to the parent compound.
bonyl (C=O) and the main-chain NH of Asp381 (DFG motif);
pip
Compounds Im-1–Im-11 were profiled with regards to their
inhibitory potencies against ABL1 kinase, the target of imati-
nib. The IC50 value of Im-1 in our in-house (Pfizer) assay was
and v) the protonated N-methylpiperazine (N ) and the car-
bonyl groups of Ile360 and/or His361. Complementary to these
hydrogen bonds are extensive hydrophobic interactions, espe-
cially between the phenyl ring of imatinib and Met290 as well
as around its pyridine-pyrimidine moieties, albeit less for N-
methylpiperazine that is partially exposed to the solvent.
To provide a rationale for the loss of their potency for ABL1
kinase, molecular docking calculations were employed for all
analogues (Im-2–Im-11; see Supporting Information). As illus-
trated in Figure 3, upon replacing the planar phenyl group
determined to be 371 nm (ATP concentration at the K of ATP)
m
while all imatinib analogues (Im-2–Im-11) exhibited significant-
ly lower potency, with all IC50 values greater than 1 mm (see
Supporting Information). Furthermore, none of the analogues
Im-2–Im-11 exhibited significant inhibition of any kinases in-
cluded in the Invitrogen panel (40 kinases).
Compounds Im-1–Im-11 were evaluated in a 72 hour cyto-
toxicity assay (Table S3 in the Supporting Information) in four
human cell lines carrying the BCR-ABL translocation, including
the three CML cell lines K562, KU-812 and MEG-01, as well as
the B-ALL cell line SUP-B15 (for further details, see Supporting
Information). BCR-ABL negative human OCI-AML3 cells were
used as a control cell line. Imatinib (Im-1) achieved maximum
cytotoxicity in all tested cell lines, except in SUP-B15 where
only up to 52% cell killing was observed (Table S3 in the Sup-
3
with sp linkers, there is a potential disruption of favorable in-
teractions, especially with the key Met290, Asp381, Glu286 and
lle360 residues. Specifically, substitution of the phenyl ring by
the cubane moiety in Im-6 is predicted to weaken the hydro-
am
gen bond between the N and Glu286 and disrupt the favora-
ble alkyl–aryl interaction operating for Im-1 with the side-chain
of Met290. In addition to the latter, the different lengths of the
linkers, like the shorter bicyclo[1.1.1]pentane of Im-4, place the
protonated terminal piperazine group in positions that are
likely to disrupt the key hydrogen-bonding interaction of Im-
1 with Ile360 of the kinase (Figure 3B; see also Supporting In-
formation). Similar effects are predicted for all other imatinib
analogues (Table S1 in the Supporting Information).
porting Information). Interestingly, SUP-B15 was the only cell
BCR–ABL
line tested expressing the p190
fusion transcript, whereas
BCR–ABL
the three CML cell lines tested express the p210
fusion
protein. The activity of Im-1 was consistently in the sub-micro-
molar range, thus paralleling its potency against its primary
target ABL1.
To summarize, the described chemistry provides a convenient
synthesis of imatinib (Im-1), which was applied to the con-
struction of a series of novel imatinib analogues (Im-2–Im-11)
Among the analogues tested, Im-6 exhibited the most
prominent activity, reaching 94% cell killing in KU-812 cells
with an estimated EC50 value of 1.4 mm, and 77% maximum cy-
totoxicity in MEG-01 cells with an estimated EC50 value of
3
where the 1,4-disubstituted phenyl ring was replaced with sp
structural motifs, including bicyclo[1.1.1]pentane, cubane and
cyclohexane. Biopharmaceutical evaluation of the synthesized
compounds confirmed the beneficial effect of the bicy-
clo[1.1.1]pentane moiety (analogue Im-4) over its phenyl coun-
terpart (Im-1) in terms of thermodynamic solubility (Im-1:
31 mm at neutral pH; Im-4 improved >80-fold). The study also
1
.8 mm (Table S3 in the Supporting Information). Cell lines K562
and SUP-B15 were significantly less sensitive to Im-6 with max-
imum cytotoxicity of 38% and 50%, respectively (Table S3).
Similarly, Im-9 showed maximum cell killing of 80% in KU-812
with an EC50 value of 1.4 mm, and 67% maximum killing in
MEG-01 with an EC50 value of 1.6 mm. K562 and SUP-B15 cell
lines were significantly less sensitive to Im-9 with only 40%
and 45% maximum killing, respectively, at the highest concen-
3
demonstrated that the sp substitution of aryl rings for improv-
ing aqueous solubility, introduced with the bicyclo[1.1.1]
3
system, is valid for a variety of other sp structural motifs, in-
ChemMedChem 2016, 11, 31 – 37
www.chemmedchem.org
35
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
commodate these changes by appropriately adjusting its con-
formation.
In conclusion, given the rather drastic change of the geome-
try and binding properties of such structural motifs as the
ones employed in this study, it is not surprising that, while
such substitutions can safely be predicted to improve certain
characteristics (e.g., aqueous solubility), they may also lead to
loss of other desirable properties (e.g., potency, target selectiv-
ity), thereby necessitating further optimization studies in order
to achieve the desirable goal of overall improvement of the
pharmacological profile of a particular molecular entity. This
may not, however, always be possible. Further investigations
with these and other novel structural motifs as substitutes of
excess aromatic moieties within the structures of potential
drug candidates are warranted.
Experimental Section
Experimental procedures and characterization data, data on molec-
ular modeling, powder X-ray diffraction (PXRD) and differential
scanning calorimetry (DSC) results of key compounds, re-crystalliza-
1
13
tion protocols, biological data, and H and C NMR spectra can be
found in the Supporting Information.
Acknowledgements
The authors thank Drs. Lawrence B. Alemany and Quinn K. Kleer-
ekoper (Rice University, Houston, USA) for NMR spectroscopic as-
sistance, C. Pennington for Mass spectroscopic assistance and Dr.
James D. Korp (University of Houston, USA) for X-Ray crystallo-
graphic analysis. The authors also thank Dr. Bethany L. Kormos,
Dr. Andreas Maderna and Dr. Yuriy Abramov for valuable discus-
sions in preparing this article, Dr. Steven O’Neil for help with
sample submission, Dr. Paul Meenan and Dr. Suman Luthra for
help with differential scanning calorimetry and melting point
measurements, and Dr. Laurence Philippe for thermodynamic sol-
ubility measurements. The infrastructure used for the computa-
tional work performed at the “Laboratory of Chemical Biology of
Natural Products & Designed Molecules” is associated with the
Greek National Research Infrastructure in Structural Biology (In-
struct-EL) of NCSR Demokritos. This work was supported by the
Cancer Prevention & Research Institute of Texas (CPRIT), USA and
The Welch Foundation, Houston, USA (grant C-1819).
Figure 3. Molecular models of the protonated forms of A) Im-6 and B) Im-4
superimposed with the X-ray structure of imatinib (Im-1) inside the catalytic
cleft of Abl (PDB ID: 2HYY) indicating the hydrogen-bonding interactions
[
32]
(
(
dotted lines). The corresponding distances (Š) are shown in the insert table
see also Supporting Information). Im-1 is shown with light brown C atoms,
Im-6 and Im-4 with cyan C atoms, the interacting residues with green C
atoms, all N atoms in blue, O in red, and S in yellow. Red color distances in
the table for Im-6 and Im-4 indicates significant deviations from those for
Im-1.
[6]
cluding the rarely used cubane moiety (Im-6) and the
straight C5 chain (Im-11). However, significant inhibition of the
target ABL1 kinase was lost in all analogues tested (Im-2–Im-
11), although some compounds were found to possess consid-
erable cytotoxicity against imatinib-sensitive cancer cells (e.g.,
cubane analogue Im-6: 94% cell killing, estimated EC =
50
1
.4 mm against cell line KU-812 and cyclohexane analogue Im-
: 80% cell killing, estimated EC =1.4 mm against cell line KU-
Keywords: anticancer agents · biopharmaceuticals · cubane
and bicyclo[1.1.1]pentanes · gleevec · imatinib analogues
9
5
0
8
12). The apparent discrepancy between the inhibitory poten-
cies of the latter compounds and their cytotoxicity potencies
suggest additional biological target(s) for these molecules.
Overall, and while the substitutions of the aromatic ring in
these imatinib analogs resulted in the desired improvement in
solubility, they also led to diminished cytotoxicity potencies,
pointing to caution with such structural motif changes. Al-
though molecular modeling (Figure 3) appears to be predictive
of the decrease of kinase binding, and thereby cytotoxicity po-
tency, it was not clear whether the biological target could ac-
[1] a) A. Levitzki, E. Mishani, Annu. Rev. Biochem. 2006, 75, 93–109; b) T.
Hunter, J. Clin. Invest. 2007, 117, 2036–2043.
[
[
[
2] A. Levitzki, Annu. Rev. Pharmacol. Toxicol. 2013, 53, 161–185.
3] F. Lovering, J. Bikker, C. Humblet, J. Med. Chem. 2009, 52, 6752–6756.
4] Values for clogP were calculated using the clogP program (version 4.3)
from BioByte Corp. (Claremont, USA); www.biobyte.com.
5] Martindale: The Complete Drug Reference, 36th ed. (Ed.: S. C. Sweetman),
Pharmaceutical Press, London, 2009, p. 733.
[
[
6] For a recent Essay on this topic, see: K. C. Nicolaou, Angew. Chem. Int.
Ed. 2014, 53, 9128–9140; Angew. Chem. 2014, 126, 9280–9292 and ref-
erences therein.
ChemMedChem 2016, 11, 31 – 37
www.chemmedchem.org
36
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
[
7] A. F. Stepan, C. Subramanyam, I. V. Efremov, J. K. Dutra, T. J. O’Sullivan,
K. J. DiRico, W. S. McDonald, A. Won, P. H. Dorff, C. E. Nolan, S. L. Becker,
L. R. Pustilnik, D. R. Riddell, G. W. Kauffman, B. L. Kormos, L. Zhang, Y. Lu,
S. H. Capetta, M. E. Green, K. Karki, E. Sibley, K. P. Atchison, A. J. Hallgren,
C. E. Oborski, A. E. Robshaw, B. Sneed, C. J. O’Donnell, J. Med. Chem.
[24] CCDC 1421073 contains the supplementary crystallographic data for
this paper. These data are provided free of charge by The Cambridge
Crystallographic Data Centre.
[25] a) M. A. Walker, Expert Opin. Drug Discovery 2014, 9, 1421–1433; b) T. J.
Ritchie, S. J. F. Macdonald, Drug Discovery Today 2009, 14, 1011–1020.
[26] R. Docherty, K. Pencheva, Y. A. Abramov, J. Pharm. Pharmacol. 2015, 67,
2012, 55, 3414–3424.
8
47–856.
[
[
8] J. Wlochal, R. D. M. Davies, J. Burton, Org. Lett. 2014, 16, 4094–4097.
9] O. Loiseleur, D. Kaufmann, S. Abel, H. M. Buerger, M. Meisenbach, B.
Schmitz, G. Sedelmeier, (Novartis Pharma GmbH, Germany), PCT Pat.
Appl. WO/2003/066613 A1, 2003.
[
[
27] N. Jain, S. H. Yalkowsky, J. Pharm. Sci. 2001, 90, 234–252.
28] The melting point of imatinib determined in our laboratory (2048C) lies
in the same range as those previously reported: a) 207–2108C: Ref. [11];
b) 206–2108C: A. S. Ivanov, S. V. Shishkov, Monatsh. Chem. 2009, 140,
[
[
[
[
[
10] J. Zimmermann, E. Buchdunger, H. Mett, T. Meyer, N. B. Lydon, P. Traxler,
Bioorg. Med. Chem. Lett. 1996, 6, 1221–1226.
11] Y. F. Liu, C. L. Wang, Y. J. Bai, N. Han, J. P. Jiao, X. L. Qi, Org. Process Res.
Dev. 2008, 12, 490–495.
12] A. F. Abdel-Magid, K. G. Carson, B. D. Harris, C. A. Maryanoff, R. D. Shah,
J. Org. Chem. 1996, 61, 3849–3862.
13] Y. F. Liu, Y. J. Bai, J. Zhang, Y. Y. Li, J. P. Jiao, X. L. Qi, Eur. J. Org. Chem.
619–623; c) 210–2128C: H.-Y. Liu, W.-P. Xia, Y. Luo, W. Lu, Monatsh.
Chem. 2010, 141, 907–911; d) 208–2098C: K.-E. Kil, Y.-S. Ding, K.-S. Lin,
D. Alexoff, S. W. Kim, C. Shea, Y. Xu, L. Muench, J. S. Fowler, Nucl. Med.
Biol. 2007, 34, 153–163.
[
29] Although each analogue was determined to be crystalline by powder
X-ray diffraction studies (see Supporting Information), it is important to
note that characterization of the crystal form of each analogue was not
undertaken. Although it is known that different crystal forms can
impact thermodynamic solubility, the thermodynamic solubility of dif-
ferent polymorphs is typically within two- to fivefold, see: M. Pudiped-
di, A. T. M. Serajuddin, J. Pharm. Sci. 2005, 94, 929–939.
2007, 6084–6088.
14] J. B. Aggen, T. Church, R. Kasar, M. S. Linsell, Q. Lu, H. E Moser, P. A.
Patten, D. Wang, (Achaogen, Inc., USA), PCT Pat. Appl. WO/2012/
1
54204 A1.
[
[
[
15] R. H. Perry, Jr., J. Org. Chem. 1959, 24, 829–833.
[
[
[
30] E. Callegari, B. Malhotra, P. J. Bungay, R. Webster, K. S. Fenner, S. Kemp-
shall, J. L. Laperle, M. C. Michel, G. G. Kay, Br. J. Clin. Pharmacol. 2011,
16] R. ChÞnevert, R. Martin, Tetrahedron: Asymmetry 1992, 3, 199–200.
17] M. L. Ingalsbe, J. D. St. Denis, J. L. Gleason, J. P. Savage, R. Priefer, Syn-
thesis 2010, 98–102.
18] Y. Kazuta, A. Matsuda, S. Shuto, J. Org. Chem. 2002, 67, 1669–1677.
19] N. M. Yoon, C. S. Pak, J. Org. Chem. 1973, 38, 2786–2792.
20] A. B. Charette, B. CôtØ, J. Am. Chem. Soc. 1995, 117, 12721–12732.
21] E. Enholm, A. Joshi, D. L. Wright, Bioorg. Med. Chem. Lett. 2005, 15,
262–5265.
22] a) A. J. Mancuso, S.-L. Huang, D. Swern, J. Org. Chem. 1978, 43, 2480–
482; b) K. Omura, D. Swern, Tetrahedron 1978, 34, 1651–1660.
23] D. B. Dess, J. C. Martin, J. Org. Chem. 1983, 48, 4155–4156.
7
2, 235–246.
31] N. A. Hosea, W. T. Collard, S. Cole, T. S. Maurer, R. X. Fang, H. Jones, S. M.
[
[
[
[
Kakar, Y. Nakai, B. J. Smith, R. Webster, K. Beaumont, J. Clin. Pharmacol.
2
32] S. W. Cowan-Jacob, G. Fendrich, A. Floersheimer, P. Furet, J. Liebetanz,
G. Rummel, P. Rheinberger, M. Centeleghe, D. Fabbro, P. W. Manley, Acta
Crystallogr. Sect. D 2007, 63, 80–93.
009, 49, 513–533.
5
[
[
2
Received: October 29, 2015
Published online on November 20, 2015
ChemMedChem 2016, 11, 31 – 37
www.chemmedchem.org
37
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim