Reengineered tricyclic anti-cancer agents

Article abstract of DOI:10.1016/j.bmc.2015.07.007

The phenothiazine and dibenzazepine tricyclics are potent neurotropic drugs with a documented but underutilized anti-cancer side effect. Reengineering these agents (TFP, CPZ, CIP) by replacing the basic amine with a neutral polar functional group (e.g., RTC-1, RTC-2) abrogated their CNS effects as demonstrated by in vitro pharmacological assays and in vivo behavioral models. Further optimization generated several phenothiazines and dibenzazepines with improved anti-cancer potency, exemplified by RTC-5. This new lead demonstrated efficacy against a xenograft model of an EGFR driven cancer without the neurotropic effects exhibited by the parent molecules. Its effects were attributed to concomitant negative regulation of PI3K-AKT and RAS-ERK signaling.

Full text of DOI:10.1016/j.bmc.2015.07.007

Bioorganic & Medicinal Chemistry xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Reengineered tricyclic anti-cancer agents
David B. Kastrinsky a, Jaya Sangodkar b, Nilesh Zaware a, Sudeh Izadmehr b, Neil S. Dhawan b,
Goutham Narla b,c, Michael Ohlmeyer a,
⇑
a Department of Structural and Chemical Biology, Icahn School of Medicine at Mt. Sinai, 1425 Madison Avenue, New York, NY 10029, United States
b Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mt. Sinai, 1425 Madison Avenue, New York, NY 10029, United States
c Department of Medicine, Institute for Transformative Molecular Medicine, Case Western Reserve University, 2103 Cornell Road, Cleveland, OH 44106, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The phenothiazine and dibenzazepine tricyclics are potent neurotropic drugs with a documented but
underutilized anti-cancer side effect. Reengineering these agents (TFP, CPZ, CIP) by replacing the basic
amine with a neutral polar functional group (e.g., RTC-1, RTC-2) abrogated their CNS effects as demon-
strated by in vitro pharmacological assays and in vivo behavioral models. Further optimization generated
several phenothiazines and dibenzazepines with improved anti-cancer potency, exemplified by RTC-5.
This new lead demonstrated efficacy against a xenograft model of an EGFR driven cancer without the
neurotropic effects exhibited by the parent molecules. Its effects were attributed to concomitant negative
regulation of PI3K-AKT and RAS-ERK signaling.
Received 12 May 2015
Revised 26 June 2015
Accepted 4 July 2015
Available online xxxx
Keywords:
Phenothiazine
Dibenzazepine
Clomipramine
Tricyclic
Ó 2015 Published by Elsevier Ltd.
Neuroleptic
1. Introduction
activity. Thus, while kinase inhibitors turn off the ‘on switch,’ there
is a corresponding requirement to restore the ‘off switch’ engen-
The genesis and progression of cancers requires the coordinated
activation of oncogenes via activating mutations or amplifications
and the simultaneous loss of function in a tumor suppressor. A
prominent oncogene, the Epidermal Growth Factor Receptor
(EGFR) is a receptor tyrosine kinase (RTK) that responds to mito-
genic signals by inducing multiple intracellular kinase networks.
Many of these are aberrantly activated in lung adenocarcinomas
including PI3K-AKT and RAS-RAF-MEK-ERK which are induced by
activated EGFR to stimulate cell growth and replication, respec-
tively. The inhibition of kinases represented a promising strategy
for the treatment of some forms of cancer demonstrated by the
clinical success of drugs targeting BCR-ABL (gleevec, Chronic
Myeloid Leukemia),1 EGFR (erlotinib, gefitinib, Non-Small Cell
Lung Carcinoma (NSCLC)),2,3 and B-Raf (vemurafenib, B-Raf
V600E mutant melanoma).4 However, as a generalized treatment
strategy, the inhibition of a single pathway does not provide a dra-
matic increase over the standard of care, cytotoxic chemotherapy.
Compensatory activation mechanisms in transformed cells, path-
way crosstalk, and the emergence of resistant mutants limits effi-
cacy. Furthermore, for cancer cells to sustain the pro-survival and
growth promoting output of these networks, activated kinase sig-
naling typically pairs with a concomitant loss of phosphatase
dered by tumor suppressors.5,6
Generating leads for a specific disease indication without a pri-
ori screens or target information is an insurmountable task. An
encouraging strategy, using existing drug molecules as leads offers
several advantages.7 These starting points exhibit drug like proper-
ties and information about sites of metabolism and tissue distribu-
tion is available. This approach also emphasizes the segregable
nature of the chemical fragments used to build such molecules. It
piqued our interest that a chemical genetic screen selected several
tricyclic neuroleptics (thioridazine, chlorpromazine (CPZ), and tri-
fluoperazine (TFP), Fig. 1) as modulators of FoxO1 localization.8
FoxO1 is a transcription factor and tumor suppressor that is active
when unphosphorylated and localized to the nucleus. FoxO1 is
inactivated by cytoplasmic sequestration when it is phosphory-
lated at multiple sites by activated kinases including AKT and
ERK.9–12 Therefore, FoxO1 cellular localization provides a surrogate
marker for the activation status of oncogenic signaling. This anti-
cancer effect of the tricyclics, restoring FoxO1 nuclear localization
in transformed, PTEN deficient cells, was attributed to negative
regulation of PI3K-AKT signaling. The hits from this screen pos-
sessed an advantageous property as whole pathway modulators
distinct from individual kinase inhibitors.
The tricyclics selected in this screen are members of a large
class of drugs, first developed in the 1950s as antagonists of mono-
amine receptors and transporters.13 Their uses are myriad for
⇑
Corresponding author. Tel.: +1 (212) 659 8630; fax: +1 (212) 849 2456.
E-mail address: michael.ohlmeyer@mssm.edu (M. Ohlmeyer).
http://dx.doi.org/10.1016/j.bmc.2015.07.007
0968-0896/Ó 2015 Published by Elsevier Ltd.
Please cite this article in press as: Kastrinsky, D. B.; et al. Bioorg. Med. Chem. (2015), http://dx.doi.org/10.1016/j.bmc.2015.07.007

2
D. B. Kastrinsky et al. / Bioorg. Med. Chem. xxx (2015) xxx–xxx
FoxO1 Modulating Tricyclics
Neurotransmitters
OH
OH
OH
Aromatic
Linker
NH2
S
N
S
N
S
CF3
N
SCH3
N
Cl
HN
serotonin
OH
NH2
dopamine
N
N
N
Amine
NH2
Trifluoperazine (TFP)
norepinephrine
Thioridazine
Cl
Chlorpromazine (CPZ)
HO
OH
X
N
HO
Cl
O
N
Anti-Cancer
CNS Pharmacophore
S
N
Pharmacophore
Tricycle
Haloperidol
Undefined and
Mechanistically
Unrelated to the Parent
Molecules
Clomipramine (CIP)
F
N
CF3
N
Linker
Figure 1. FDA approved neuroleptic medications.
N
Amine
diseases of the peripheral and central nervous systems (CNS)
including numerous psychiatric conditions. They are notably unse-
lective with a strong side effect profile due to binding to multiple
receptor classes and subtypes in different tissues. This potent, pri-
mary activity, both on and off target, would interfere with studying
their anti-cancer properties and precludes their use in animal
models. It is conceivable that an appropriately selected agent could
be used clinically for treating specific cancers.14 However, the pro-
nounced sedative, extrapyramidal, and anti-cholinergic effects are
severely dose limiting.
Figure 2. Pharmacophore of the tricyclics: CNS vs anti-cancer.
receptors than TFP,29 we utilized the 2-chlorophenothiazine tricy-
cle (I, Fig. 3). We also included the 3-chloro-10,11-dihydro-5H-
dibenzo[b,f]azepine (IIa, Fig. 3) tricycle exemplified in the anti-de-
pressant Clomipramine (CIP, Fig. 1). Its two carbon bridge is an iso-
stere of the phenothiazine sulfur.30 Simple derivatives of CPZ and
CIP were prepared with one key modification: exchange of the
dimethylamine base for a neutral polar substituent (sulfonamide,
carbamate, amide, urea: e.g., RTCs (1–4): Tables 1 and S2–S3).
Our first series included two (A), three (B), and four (C) carbon
linker variants. In the phenothiazine series, the B and C-linker ami-
nes were accessed by direct alkylation with a bromoalkyl phthal-
imide followed by phthaloyl deprotection. The A-linker required
an alternative approach of acylation with chloroacetyl chloride, fol-
lowed by azide substitution and combined reduction to the satu-
rated amine. This acylation approach31 was adapted to the
dibenzazepine series which resisted most attempts at direct alky-
lation. Here the B-linker and C-linker analogues were prepared
by acylation with chloroacetyl chloride and 3-chloroproprionyl
chloride, respectively, followed by conversion to the nitrile, and
combined reduction to the amine precursors. These precursors
were derivatized with substituted sulfonyl chlorides, chlorofor-
mates, and acyl chlorides (Schemes 1 and 2).32
The second objective was to optimize these resulting com-
pounds to improve their anti-cancer potency and physiochemical
properties. Systematic, iterative rounds of optimization introduced
alterations to the tricycle, linker, and to the pendant N-linked side
chain (Fig. 3).
Due to their unrivaled biological activity, subsequent efforts
converged mainly on sulfonamides. The 4-trifluoromethoxyben-
zenesulfonamide of the 3-linker variant proved especially potent
(RTC-5). Aiming to improve its properties, we introduced varia-
tions to the tricyclic moiety (Fig. 3). These include the thioxan-
thene (III) and dibenzocycloheptene (IV) heterocyclic tricycles
present in thioxithene and amitriptyline, respectively. Their syn-
theses rely on modifications to published routes (Scheme S1).33,34
The next set of analogues defined the minimally potent phar-
macophore. These included removing the bridging atom(s) in the
tricyclic moiety (acyclic, V), and one of the fused benzene rings
We proposed to reengineer the tricyclic antipsychotics to opti-
mize their anti-cancer side effect.15,16 This effect has been docu-
mented in a number of basic research studies17–24 and clinical
epidemiological studies.25 Phenothiazines are frequently observed
in high throughput screens of FDA approved compounds and inter-
act with numerous biological targets.26 The CNS pharmacophore of
these drugs is a consequence of their structural similarity to the
monoamine neurotransmitters whose binding they obstruct
(dopamine, serotonin, norepinephrine).27 From both the tricyclics
and the natural substrates of these receptors and transporters it
is clear that the heterocycle, a 2–3 carbon linker, and an amine
are essential for the CNS effects (Fig. 2). In the screen discussed
earlier,8 the antipsychotic haloperidol was examined to rule out
the possibility that dopamine receptor antagonism played a role
in FoxO1 modulation. Haloperidol is a powerful antagonist and
inverse-agonist to a number of neurotransmitter receptors.28 It is
structurally unrelated to the tricyclics but it contains an amine, a
key component for binding dopamine receptors. Haloperidol did
not affect FoxO1 localization and thus does not perturb oncogenic
signaling. This led us to speculate that the chemical fragments in
the tricyclics that were responsible for the CNS versus the anti-can-
cer properties were not identical. Here, we systematically probe
the tricyclics’ structure to determine the anti-cancer pharma-
cophore. Our first efforts examined alternative functionalization
of the dimethylamine portion. Deleting this functional group
would conceivably eliminate the CNS activity of these molecules,
facilitating their development for a specific anti-cancer purpose.
2. Chemistry
In synthesizing reengineered tricyclics (RTCs) we resolved to
accomplish two disparate goals. The first was to eliminate the
CNS pharmacology. Since CPZ is less potent at its target CNS
Please cite this article in press as: Kastrinsky, D. B.; et al. Bioorg. Med. Chem. (2015), http://dx.doi.org/10.1016/j.bmc.2015.07.007

D. B. Kastrinsky et al. / Bioorg. Med. Chem. xxx (2015) xxx–xxx
3
(a)
(b)
(c)
(d)
Figure 3. Tricyclic, linker, and side chain variants.
Table 1
Cell viability screen of reengineered tricyclics
Scheme S3)35–37 and a nitrogen substituted dibenzazepine series
(VIII–X, Scheme S4).38,39 Several analogues included constrained
cyclic linkers intended to increase drug likeness by limiting rotat-
able bonds.40 These utilized the dibenzazepine (IIb) and dibenzocy-
cloheptene (IV) tricycles and were prepared via reductive
amination,41 McMurry coupling,42 and alkylation sequences,
respectively (Schemes S5–S7). Three additional series included:
RTC
Tricyclic
Linker
Side chain
GI50 (lM)
1
2
3
4
5
6
7
8
I
I
B
B
B
B
B
B
A
A
A
A
A
A
C
C
C
C
B
B
B
B
B
B
B
B
B
B
B
B
B
N
N
ii
i
ii
i
iii
iv
ii
>40
>40
>40
>40
12.6
20.0
24.4
Inactive
25.0
20.0
IIa
IIa
IIa
IIa
I
I
I
I
modified linkers (K, L) and
(Schemes S8 and S9).
a spirocyclic ether (M) series
i
A special case unto itself, RTC-30 contains a hydroxylated linker
(N) that confers increased oral bioavailability. Its synthesis began
with the alkylation of iminodibenzyl with S-epichlorohydrin.43
Epoxide opening with sodium azide followed by reduction pro-
vides the amine precursor. Quantitative sulfonylation of the amine
provided RTC-30. Installation of this hydrophilic linker adds a
9
iii
iv
iii
iv
ii
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Ent-30
IIa
IIa
I
22.0
19.1
>40
I
i
>40
IIa
IIa
I
I
I
iii
iv
v
vi
vii
iii
viii
vii
v
Inactive
Inactive
>40
>40
>40
>40
12.6
20.0
24.4
Inactive
25.0
20.0
22.0
19.1
>40
15.0
degree of complexity because it introduces
a chiral center.
However, both enantiomers were accessed from the commercially
available and optically pure forms of epichlorohydrin (Scheme 3).
I
IIa
IIa
IIa
IIa
IIa
IIa
IIb
VII
VII
IIb
IIb
3. Results and discussion
All compounds were subjected to an MTT (3-[4,5-dimethylthia-
zol-2-yl]-2,5 diphenyl tetrazolium bromide) cell viability screen44
using H1650 lung adenocarcinoma cells. This cell line harbors an
activating mutation in EGFR and inactivated PTEN, and thus is
characterized by constitutively activated AKT and ERK signaling.
These lesions confer resistance to single pathway RTK inhibitors
such as erlotinib.45 As a guide, the cells were treated with com-
ix
x
xi
iii
iii
iv
iii
iii
18.0
pound at five concentration points bracketing 1–40
lM in incre-
ments of 10 M. Cells were exposed to drug for 48 h. Growth
l
Inhibition (GI50) potencies are reported only in ranges where the
data exhibit a sigmoidal dose response. If the data did not converge
on a sigmoidal response curve but still showed a response, they are
(truncated, VI) (Scheme S2). Once established, our next efforts
focused on improving the potency and pharmacokinetic properties
of RTC-5. We introduced heteroatoms: dibenzoxazepine (VII,
Please cite this article in press as: Kastrinsky, D. B.; et al. Bioorg. Med. Chem. (2015), http://dx.doi.org/10.1016/j.bmc.2015.07.007

4
D. B. Kastrinsky et al. / Bioorg. Med. Chem. xxx (2015) xxx–xxx
1. chloroacetyl chloride
toluene, 100 oC, 76%
S
N
S
2. NaN3, DMF, 96%
3. BH3THF;
HCl-dioxane, 84%
N
H
Cl
Cl
Cl
S
N
S
N
Et3N, ClCO2R1,
or ClSO2Ar-R2
NH2HCl
Cl
Cl
DMF
1. NaH, N-(3-Bromopropyl)-
phthalimide, DMF, 73%
(CH2)n
NH
(CH2)n
NH
S
S
N
or NaH, N-(4-Bromobutyl)-
phthalimide, DMF
O
S O
N
H
Cl
O
O
R1
n = 0, 1, 2
(CH2)n
NH2HCl
R2
2. NH2NH2, EtOH;
HCl-dioxane
n = 1 (ref)
n = 2, 42%
Scheme 1. Synthesis of phenothiazine derivatives with varied linker lengths and N-modifications.
R1 = Cl
NaN3, DMF
74%
BH3-THF
R1 = N3
THF, 70 oC;
HCl-dioxane
57%
N
Cl
NH2HCl
chloroacetyl chloride
toluene, 100 o
93%
C
N
N
H
Cl
Cl
O
R1
N
R1 = Cl
Cl
Et3N, ArSO2Cl
DMF
NaCN, DMF
61%
BH3-THF
(CH2)n
NH
R1 = CN
THF, 70 oC;
HCl-dioxane
40%
N
N
Cl
O
S O
n = 0, 1, 2
NH2HCl
R2
3-chloroproprionyl chloride
BH3-THF
THF, 70 oC;
HCl-dioxane
62%
toluene, 100 o
68%
C
N
N
H
Cl
Cl
Cl
O
R1
R1 = Cl
NH2HCl
NaCN, DMF
93%
R1 = CN
Scheme 2. Synthesis of dibenzazepine analogues with varied linker lengths and N-modifications.
O
Cl
1. NaN3, NH4Cl
EtOH, 92%
Et3N, ArSO2Cl
DMF, 91%
NaNH2, toluene
37-45% (ref. 43)
N
N
N
N
2. PPh3, H2O
THF, 61%
H
OH
O
OH
O
O
S
N
H
NH2
RTC-30
OCF3
Scheme 3. Synthesis of RTC-30.
presumed >40
absorbance >80% at 40
(Fig. 4).
A panel of tricyclic drugs was surveyed and most exhibited
weak reductions on cell viability (Table S1). TFP was the most
l
M. If the response was negligible with residual
Despite this lackluster potency, we believed this set to be a signif-
icant departure and conducted a series of in vitro and in vivo phar-
macological assays. Emphatically, this set did not possess the
neurotropic properties of the parent molecules (vide infra). With
this required departure confirmed, we pursued a more diverse ser-
ies of analogues.
Next, we evaluated the A and C-linker variations (Table 1). All of
the C-linker analogues decidedly lost potency (13–16), while some
of the A-linker analogues (7, 9–12) were comparable or slightly
inferior to the B-linker versions. Attempts to optimize the poten-
cies of carbamate derivates akin to RTC-2 did not prove fruitful
in either phenothiazine or dibenzazepine series, and aliphatic
lM, the compound is deemed ‘inactive’
active and provided a benchmark (12.2
potent (>40 M). The first set of compounds including RTCs (1–4),
were weakly potent (>40 M, Table 1). Cell cycle analysis of RTC-1
and RTC-2 on H1650 cells revealed a dose dependent sub-G1 accu-
mulation indicative of apoptosis at 20 and 40 M (Fig. S1).
Clonogenic assays verified some loss of cell viability, which became
evident at 20 and 40 M after seven days of treatment (Fig. S2).
lM) while CIP was weakly
l
l
l
l
Please cite this article in press as: Kastrinsky, D. B.; et al. Bioorg. Med. Chem. (2015), http://dx.doi.org/10.1016/j.bmc.2015.07.007

D. B. Kastrinsky et al. / Bioorg. Med. Chem. xxx (2015) xxx–xxx
5
Figure 4. MTT assay of selected compounds, H1650 cells.
Figure 5. Western blot analysis of AKT and ERK levels.
and aryl amides were similarly inactive. Several sulfonamides,
however, exhibited potencies comparable with TFP.
10
8
The search for an optimal sulfonamide yielded reproducible
trends. Aliphatic sulfonamides were inactive (Tables S2 and S3).
In the aryl series, electron donating groups at the 4 position (17,
OCH3), (18, OPh) and substituents at the 2 position (24, 25) reliably
generated weakly active or inactive compounds (Table 1). The most
potent phenothiazines and dibenzazepines possess electron with-
drawing substituents: halogen, trifluoromethyl, or trifluo-
6
4
2
romethoxy substituents at
3
and
4
positions of
a
benzenesulfonamide core. A number of analogues emerged from
the dibenzazepine B-linker aryl sulfonamides series: RTC-21 (4-
CF3), 22 (4-CN), 6 (4-Cl), 24 (3,4-dichloro), and 5 (4-OCF3). RTC-5
represented a significant breakthrough and was subjected to a
much more comprehensive biological and pharmacological
evaluation.
In concordance with the results obtained for RTC-1 and RTC-2,
RTC-5 induced a sub-G1 accumulation and cell cycle arrest, sim-
ilar to but more pronounced than TFP and CIP (Fig. S3). A further
confirmation of its apoptotic mechanism, treatment of cells with
RTC-5 resulted in an increase in Annexin V staining. Additional
treatment with the pan caspase inhibitor Z-VAD reversed
Annexin V staining (Fig. S4), indicating that apoptosis was cas-
pase-mediated. Seeking its molecular target, the effects of RTC-
5 treatment on intracellular kinases were evaluated by Western
blot. Both PI3K-AKT and RAS-ERK pathways are simultaneously,
negatively regulated by RTC-5 as indicated by decreases in phos-
pho-AKT and phospho-ERK levels (Fig. 5). RTC-10, a potent phe-
0
DMSO
TFP
RTC-5
Figure 6. H1650 xenograft study in mice.
Inspired by these results, we continued synthesizing molecules
with conservative variations. Introducing an oxygen to the central
ring of the tricycle (dibenzoxazepines, 28–29, Table 1) afforded less
potent compounds. Eliminating the bridging atoms or removing
one of the fused benzene rings provided inactive analogues illus-
trating that the intact tricycle is required for activity (68–71,
Table S6).
The tricyclics contain a charged amine that confers aqueous sol-
ubility and oral bioavailability. As a necessary consequence, its
removal diminishes these desirable qualities. The resulting com-
pounds are lipophilic, less water soluble, and less orally bioavail-
able. Fortunately, the development of RTC-30 imparted a major
solution to this problem (vide infra).
nothiazine analogue, and TFP exhibit similar effects.
A
This concern was addressed in other ways by incorporating
polar modifications and restricting rotatable bonds. Two series
containing modified linkers (Table S7) and three series containing
nitrogen substituted tricyclic systems (62–67, Table S5) proved
less potent than RTC-5. In another series, several commercially
available heterocyclic arylsulfonyl chlorides were obtained and
introduced. Of these, some thiophene analogues (76, 77) provided
potent compounds, not surprisingly since thiophenes are benzene
ring isosteres. However, most of these heterocyclic analogues were
weakly potent, inactive, or unlikely to exhibit improved pharma-
cokinetic properties (Table S8).
A dibenzocycloheptene (IV) series in which the tricyclic nitro-
gen is replaced with a double bond provided several active com-
pounds (56–59, Table S5). In a similar vein, several series of
analogues with cyclic linkers were investigated. Those in which
the sulfonamide nitrogen is endocyclic (F, J), were decidedly inac-
tive (94–99). Those in which the nitrogen is exocyclic (D, E, H, I)
DiscoveRx kinome screen confirmed that RTC-5 was not inhibit-
ing any target relevant kinase in an ATP competitive manner
(unpublished results). Recent literature reports implicate the acti-
vation of the tumor suppressor Protein Phosphatase 2A (PP2A) as
the source of the anti-cancer effects demonstrated by the tri-
cyclics.23 PP2A is a ubiquitously expressed tumor suppressor that
negatively regulates multiple oncogenic signaling pathways
including AKT and ERK.46–48 Unpublished experiments have con-
firmed this mechanism for our RTCs and these results will be pre-
sented shortly.
Probing its in vivo effects, we performed a xenograft study
using the H1650 lung cancer cell line. Treatment with 100 mg/kg
RTC-5 caused a statistically significant decrease in mean fold
change in tumor volume (1.49 0.26, n = 9) compared to vehicle
control (3.46 0.95, n = 7) (p <0.004, student’s t test). In this same
study, TFP could not be dosed higher than 10 mg/kg due to marked
CNS effects (Fig. 6).
Please cite this article in press as: Kastrinsky, D. B.; et al. Bioorg. Med. Chem. (2015), http://dx.doi.org/10.1016/j.bmc.2015.07.007

6
D. B. Kastrinsky et al. / Bioorg. Med. Chem. xxx (2015) xxx–xxx
N
N
F3CO
OCF3
OH
Cl
O
O
RTC-5
RTC-30
S
O
S
N
N
O
H
H
X
Tricycle Required
X = S, CH2CH2
(Y-Z) = (N-C) or (C=C)
W
Y
Z
Linker Hydroxyl Improves
Oral Bioavailability
OH
NH
W = Cl, H
Sulfonamide N-H required
O
S O
R = 3- or 4-
electron
withdrawing group
R
Figure 8. Monoamine transporter inhibition (CIP vs RTC-5). Antagonist radioligand
assays: DT: [3H]BTCP; 5-HTT: [3H]imipramine; NET: [3H]nisoxetine; each bar
represents the average of two independent determinations.
Figure 7. Consensus structure.
proved comparable in potency to RTC-5 (100–107), as did several
members of a genus containing a spirocyclic ether (M, 108-114)
(Table S9).
Sulfone, urea, and sulfonyl urea analogues were deemed inac-
tive confirming a general preference for the sulfonamide moiety
(115–117, Table S10). Moreover, making N-Me versions of five of
the most potent compounds (118–122, Table S11) rendered them
inactive demonstrating an absolute requirement for the sulfon-
amide N–H. A consensus structure activity relationship picture
emerged at the end of these studies (Fig. 7).
It deserves final mention that the compounds prepared in this
study exhibit relatively modest, micromolar potencies. To address
this issue, we highlight that the MTT assay uses a transformed,
metastatic NSCLC cell line with multiple, activated tumor promot-
ing pathways. The RTCs possess a different mechanism of action
from cytotoxics and even from kinase inhibitors, and their advan-
tages have only become apparent in vivo. For comparison, some
FDA approved kinase inhibitors including vemurafenib49 and
sorafenib50 require micromolar serum concentrations for effica-
cious dosing. Using the MTT assay as a guide, RTC-5, RTC-30 and
numerous analogues strikingly display activity in a model of meta-
static NSCLC for which there are no treatments.51 Comparable to
In pharmacokinetic studies in mice, RTC-1 exhibits significant
absorption by intra-peritoneal (IP) route (50%), orally (17%) and
low clearance (11.1 ml/min/kg, t1/2 = 0.61 h). RTC-2 exhibits signif-
icant absorption IP (47%), and orally (16%), and moderate clearance
(52.5 ml/min/kg t1/2 = 0.61 h). In both cases, the reduced oral expo-
sure is likely due to poor solubility in the aqueous vehicle (10%
DMSO, 30% Cremophor EL, 60% water). The poor exposure is likely
exacerbated for RTC-2 due to its increased rate of clearance. This
lack of oral bioavailability was a concern addressed in subsequent
synthetic efforts (Tables S12 and S13).
Because of its increased potency, RTC-5 was compared to CIP
against a much more comprehensive panel of amine transporters,
GPCRs, other receptors, and ion channels. Tested at three concen-
trations, 0.1 mM, 1.0 mM, and 10 mM, CIP completely inhibits
the serotonin transporter (5-HT, Ki = 0.28 nM,53) the source of its
on-target biological effects, as well as the dopamine (DT, 14%
(1.0 mM), 60% (10 mM), and norepinephrine (NET, 5%, (0.1 mM)
90% (1.0 mM), 100% (10 mM)) transporters. In contrast, RTC-5 does
not bind the serotonin (5-HTT, <5%) and binds the dopamine (DT)
and norepinephrine (NET) transporters weakly and only at the ele-
vated concentrations (Fig. 8).
RTC-5, the GI50 of RTC-30 in the MTT assay was 15 lM, slightly bet-
CIP binds several dopamine and serotonin subtypes. RTC-5 does
not bind most dopamine or serotonin receptor subtypes except for
some residual binding at 5-HT5A (24% (1.0 mM), 68% (10 mM))
(Figs. S8 and S9). Several tricyclics have been associated with QT
interval prolongation.54 These cardiovascular liabilities of CIP are
attributed to ion channel inhibition. RTC-5 does not bind a panel
of calcium and potassium channels (Figs. S10 and S11). RTC-5 exhi-
ter than its enantiomer. However, its improved pharmacokinetic
properties are what prompted and enabled its subsequent in vivo
evaluation.
4. Pharmacology
bits negligible effects on a voltage gated sodium channel (Nav1.5
)
RTCs 1–3 were evaluated for binding to a panel of dopamine
receptors (D1–D5). TFP, CPZ, and CIP bind this panel indiscrimi-
nately. D2 is the most clinically relevant subtype and its antago-
nism is attributed to the antipsychotic effects associated with
these drugs. At 0.1 mM ligand concentration, none of our com-
pounds displayed significant binding to D2 or any of the dopamine
receptor subtypes (D1–D4 <5%, D5 <10%) (Fig. S5). At 1.0 mM and of
less physiological relevance, RTC-2 displayed minimal binding to
D1–D3 (14%, 21%, 22%) as did RTC-1 for D1 and D2 (12%, 14%).
TFP, in contrast exhibits >95% binding at 1.0 mM to all five sub-
types, and at 0.1 mM, >95% towards D1–D3 with predominant bind-
ing to D4 (73%) and D5 (81%) (Fig. S6). Clearly, this single
modification, conversion of an amine to a neutral polar functional
group diminished the neurotropic properties of these molecules.
We also examined the neurotropic effects of compound admin-
istration in vivo using a scoring test for lethargy.52 The effects on
mouse behavior, notably sedation, were rather striking with TFP.
RTC-1 and RTC-2 do not induce lethargy in this system and are
comparable to control (Fig. S7).
compared to CIP in a patch clamp assay (Fig. S12). It also does
not bind a panel of receptors (M2, H1, H2, hERG) localized to heart
tissues and linked to QT interval prolongation (Fig. S13).
The pharmacokinetic properties in mice of RTC-5 were reason-
able with significant absorption by IP (34–38%), orally (15–18%)
and moderate clearance (42 ml/min/kg t1/2 = 0.61 h) (Table S14).
The properties of RTC-30 were dramatically improved with IP
absorption (62–94%) and oral absorption (36–50%) nearly doubled
with similar clearance (44 ml/min/kg t1/2 = 0.81 h) (Table S15).
5. Conclusion
The identification of molecular drivers involved in oncogenesis
played an instrumental role in redefining the study and treatment
of cancer. This information permits the precise targeting of dis-
eases with a defined molecular mechanism and resulted in the suc-
cessful development of pharmaceuticals for a small but increasing
number of aggressive, untreatable cancers. While much of this
Please cite this article in press as: Kastrinsky, D. B.; et al. Bioorg. Med. Chem. (2015), http://dx.doi.org/10.1016/j.bmc.2015.07.007

D. B. Kastrinsky et al. / Bioorg. Med. Chem. xxx (2015) xxx–xxx
7
early work focused on RTKs like EGFR and its immediate down-
Supplementary data
stream signaling partners, there are many more targets to be
addressed. Capitalizing on and evolving with this trend, the utiliza-
tion of chemical genetic screens enables the dissection of compli-
cated signaling pathways involved in these diseases. Small
molecule hits from these screens possess a mode of action not dis-
cernible from classical phenotypic assays. Translation of these dis-
coveries will provide new tools for modulating these undefined
and previously deemed undruggable pathways.
Prior to this work, numerous reports indicated the anti-cancer
properties of the tricyclics. On the surface, this was a finding of
dubious importance. The tricyclics are an ancient class and not
an active area of development. Achieving appreciable serum con-
centrations to observe the anti-cancer effects would cause dose-
limiting effects on the CNS of experimental animals. Although
the tricyclics are FDA approved drugs, we recognized that these
compounds were not optimized for this anti-cancer purpose.
Towards this end, we introduced an important modification:
changing the amine to a sulfonamide. This eliminated the CNS
associated pharmacology and enabled further optimization of the
anti-cancer effect. We also recognized the importance of their
novel mechanism of action: dual pathway inhibition of PI3K-AKT
and RAS-ERK signaling via putative activation of the tumor sup-
pressor PP2A.
Supplementary data (experimental data) associated with this
article can be found, in the online version, at http://dx.doi.org/10.
1016/j.bmc.2015.07.007.
References and notes
1. Druker, B. J. et al N. Engl. J. Med. 2001, 344, 1038.
2. Maemondo, M. et al N. Engl. J. Med. 2010, 362, 2380.
3. Giaccone, G. et al Clin. Cancer Res. 2006, 12, 6049.
4. Sosman, J. A. et al N. Engl. J. Med. 2012, 366, 707.
5. Hahn, W. C. et al Nature 1999, 400, 464.
6. Hahn, W. C. et al Mol. Cell. Biol. 2002, 22, 2111.
7. Siegel, M. G.; Vieth, M. Drug Discovery Today 2007, 12, 71.
8. Kau, T. R. et al Cancer Cell 2003, 4, 463.
9. van Gorp, A. G. et al Cancer Res. 2006, 66, 10760.
10. Terragni, J. et al BMC Cell Biol. 2008, 9, 6.
11. Tran, H. et al Science’s STKE : signal transduction knowledge environment 2003,
172, RE5.
12. Asada, S. et al Cell. Signal. 2007, 19, 519.
13. Owens, D. G. A guide to the extrapyramidal side-effects of antipsychotic drugs;
Cambridge University Press: Cambridge, New York, 1999; p 351.
14. Jahchan, N. S. et al Cancer Discovery 2013, 3, 1364.
15. Wermuth, C. G. Med. Chem. Res. 2001, 10, 431.
16. Wermuth, C. G. J. Med. Chem. 2004, 47, 1303.
17. Aftab, D. T. et al Mol. Pharmacol. 1991, 40, 798.
18. Ford, J. M.; Prozialeck, W. C.; Hait, W. N. Mol. Pharmacol. 1989, 35, 105.
19. Choi, J. H. et al Ann. N. Y. Acad. Sci. 2008, 1138, 393.
20. Yde, C. W. et al Anticancer Drugs 2009, 20, 723.
21. Tzadok, S. et al Int. J. Oncol. 2010, 37, 1043.
22. Gil-Ad, I. et al J. Mol. Neurosci.: MN 2004, 22, 189.
23. Gutierrez, A., et al., Phenothiazines induce apoptosis in T-cell acute
lymphoblastic leukemia by activating the phosphatase activity of the PP2A
tumor suppressor. Submitted, 2012.
Efforts to reengineer the tricyclics produced a series of potent
compounds exemplified by RTC-5 and RTC-30. Their effects on cells
and animal models have been documented and here we delineate
the structural features important for their activity. RTC-5 and RTC-
30 will continue to be studied in selected models of cancers sus-
ceptible to dual pathway inhibition. The results of these studies
will be presented in due course.
24. Yeh, C. T. et al Am. J. Respir. Crit. Care Med. 2012, 186, 1180.
25. Walker, A. J. et al Br. J. Cancer 2011, 104, 193.
26. Jaszczyszyn, A. et al Pharmacol. Rep. 2012, 64, 16.
27. Horn, A. S.; Snyder, S. H. Proc. Natl. Acad. Sci. U.S.A. 1971, 68, 2325.
28. Schaus, J. M.; Bymaster, F. P. Annu. Rep. Med. Chem. 1998, 33, 1.
29. Savelyeva, M. V.; Baldenkov, G. N.; Kaverina, N. V. Biomed. Biochim. Acta 1988,
47, 1085.
Competing interests statement
30. Craig, P. N. et al J. Org. Chem. 1961, 26. 135-&.
31. Sinning, S. et al J. Biol. Chem. 2010, 285, 8363.
The Icahn School of Medicine at Mount Sinai on behalf of the
authors G.N., M.O., N.S.D., D.B.K. have filed patents covering com-
position of matter on the small molecules disclosed herein for
the treatment of human cancer and other diseases. Dual
Therapeutics LLC has licensed this intellectual property. The
authors G.N., M.O., N.S.D., D.B.K. have an ownership interest in
Dual Therapeutics LLC. G.N. and M.O. are consultants for Dual
Therapeutics LLC.
32. Ohlmeyer, M. et al Preparation of phenothiazine derivatives and analogs for use as
antitumor agents; Mt. Sinai School of Medicine: USA, 2013; p 152.
33. Oligomer-Tricyclic Conjugates. PCT/US2008/013221 [00212], 2009.
34. 9-Substituted thioxanthenes and their acid addition salts. Dansk Patent NR.
88606, 1960.
35. Zaware, N.; Ohlmeyer, M. Heterocycl. Commun. 2014, 251.
36. Kubota, K. et al Bioorg. Med. Chem. 2011, 19, 3005.
37. Margolis, B. J.; Swidorski, J. J.; Rogers, B. N. J. Org. Chem. 2003, 68, 644.
38. Villani, F. J.; Mann, T. A. J. Med. Chem. 1968, 11, 895.
39. Ohlmeyer, M.; Kastrinsky, D. Tricyclic heterocycles as anticancer agents; Icahn
School of Medicine at Mount Sinai: USA, 2014; p 84.
40. Veber, D. F. et al J. Med. Chem. 2002, 45, 2615.
Acknowledgments
41. Abdel-Magid, A. F. et al J. Org. Chem. 1996, 61, 3849.
42. Process for the preparation of 10,11-dihydro-5h-dibenzo[a,d]cyclohept-5-enes and
derivatives thereof, WO 1998038166 A1, 1998.
43. Levy, O. et al Bioorg. Med. Chem. Lett. 2001, 11, 2921.
44. Denizot, F.; Lang, R. J. Immunol. Methods 1986, 89, 271.
45. Sos, M. L. et al Cancer Res. 2009, 69, 3256.
46. Zhang, Q.; Claret, F. X. Enzyme Res. 2012, 2012, 659649.
47. Kuo, Y. C. et al J. Biol. Chem. 2008, 283, 1882.
48. Letourneux, C.; Rocher, G.; Porteu, F. EMBO J. 2006, 25, 727.
49. Bollag, G. et al Nature 2010, 467, 596.
50. Wilhelm, S. M. et al Mol. Cancer Ther. 2008, 7, 3129.
51. Sharma, S. V. et al Nat. Rev. Cancer 2007, 7, 169.
52. Gerhard, H.; Vogel, E. Drug discovery and evaluation pharmacological assays, 3rd
Ed., 2008; p 217.
53. Tatsumi, M. et al Eur. J. Pharmacol. 1997, 340, 249.
54. Zemrak, W. R.; Kenna, G. A. Am. J. Health-System Pharm. 2008, 65, 1029.
We gratefully acknowledge the Icahn School of Medicine at Mt.
Sinai for startup funding (M. Ohlmeyer), the New York City
Investment
Fund
(Bioaccelerate
Prize),
and
Dual
Therapeutics/Biomotiv for continued support. Goutham Narla
received a Howard Hughes Medical Institute (HHMI) Physician-
Scientist Early Career Award and a Case Comprehensive Cancer
Center Pilot Award. He is a Harrington Distinguished Scholar
(Early Career Award). We would like to thank Dr. Lei Zeng (Mt.
Sinai) for assistance with NMR spectroscopy and Dr. Michael
Takase (Yale University Department of Chemistry X-ray Facility)
for X-ray analysis. We would also like to thank Heather Melville
and Kasey Grewe for performing the MTT assays.
Please cite this article in press as: Kastrinsky, D. B.; et al. Bioorg. Med. Chem. (2015), http://dx.doi.org/10.1016/j.bmc.2015.07.007