Bioorganic & Medicinal Chemistry Letters 22 (2012) 6460–6468

Contents lists available at SciVerse ScienceDirect

Bioorganic & Medicinal Chemistry Letters

journal homepage: www.elsevier.com/locate/bmcl

1-(sulfonyl)-5-(arylsulfonyl)indoline as activators of the tumor cell specific

M2 isoform of pyruvate kinase
Avihai Yacovan ⇑, Rachel Ozeri, Tzofit Kehat, Sima Mirilashvili, Daniel Sherman, Alex Aizikovich,
Alina Shitrit, Efrat Ben-Zeev, Nili Schutz, Osnat Bohana-Kashtan, Alexander Konson, Vered Behar,
Oren M. Becker
Dynamix Pharmaceuticals, Rehovot 76385, Israel

a r t i c l e i n f o a b s t r a c t

Article history: Cancer cells preferentially use glycolysis rather than oxidative phosphorylation for their rapid growth.
Received 8 July 2012 They consume large amount of glucose to produce lactate even when oxygen is abundant, a phenomenon
Revised 5 August 2012 known as the Warburg effect. This metabolic change originates from a shift in the expression of alterna-
Accepted 13 August 2012
tive spliced isoforms of the glycolytic enzyme pyruvate kinase (PK), from PKM1 to PKM2. While PKM1 is
Available online 23 August 2012
constitutively active, PKM2 is switched from an inactive dimer form to an active tetramer form by small
molecule activators. The prevalence of PKM2 in cancer cells relative to the prevalence of PKM1 in many
Keywords:
normal cells, suggests a therapeutic strategy whereby activation of PKM2 may counter the abnormal cel-
PKM2
Pyruvate kinase
lular metabolism in cancer cells, and consequently decreased cellular proliferation. Herein we describe
Cellular metabolism the discovery and optimization of a series of PKM2 activators derived from the 2-((2,3-dihydro-
Anti-cancer strategies benzo[b][1,4] dioxin-6-yl)thio)-1-(2-methyl-1-(methylsulfonyl)indolin-5-yl) ethanone scaffold. The syn-
Small molecule activators thesis, SAR analysis, enzyme active site docking, enzymatic reaction kinetics, selectivity and
pharmaceutical properties are discussed.
Ó 2012 Elsevier Ltd. All rights reserved.

Most cancer cells undergo major metabolic changes to enable
their massive proliferation and energy production.1 Their altered
metabolism is represented by increased glucose breakdown and
lactate production.2–5 A key mediator of glycolysis is pyruvate ki-
nase (PK), a rate-limiting enzyme that catalyzes the last step in gly-
colysis. There are four PK isoforms in mammalian cells6–10: the M1
isoform (PKM1) is expressed in many differentiated tissues (skele-
tal muscle, heart and brain), PKM2 is expressed during embryonic
development, PKL and PKR are expressed in liver and erythrocytes,
respectively. All four isoforms of PK catalyze the transformation of
phosphoenolpyruvate (PEP) and ADP to pyruvate and ATP.11,12
While PKM2, PKL and PKR are activated by the binding of fructose
1-6-bis-phosphate (FBP) at an allosteric site, PKM1 does not re-
quire allosteric activation and is continuously active. Furthermore,
the isoform that is over expressed in cancer cells, PKM213, facili-
tates their altered metabolism by switching from a highly active
tetramer form to low-activity monomer or dimer form.13,14 This
process is regulated by the upstream glycolytic intermediate, FBP.
The change in the expressed PK isoform, enables cancer cells to
balance their use of glucose carbon backbones, whether for ATP
production or for biomass generation, including the synthesis of
amino acid, nucleotide, and lipid, according to their changing
⇑ Corresponding author. Tel.: +972 8 939 6010; fax: +972 8 939 6010.
E-mail address: ayacovan@dynxp.com (A. Yacovan).
demands.15 This metabolic difference, between cancer cells and
normal cells, provides an attractive new target for cancer therapy.
Furthermore, the observed down-regulation of PKM2 activity in
cancer cells relative to the high PKM1 activity present in many nor-
mal16 cells suggests a therapeutic strategy whereby activation of
PKM2 may perturb cellular metabolism, and consequently de-
crease cellular proliferation. Thus small molecule PKM2 activators,
which stabilize the tetramer form, are expected to affect cancer
metabolism, offering a novel anti-cancer therapeutic strategy.
The first small molecules capable of activating PKM2 were re-
ported by the NIH chemical genomics center and include a series
of diarylsulfoamides.17 Since then, very few additional PKM2 acti-
vator chemotypes have been reported.18–20 To identify novel acti-
vators with selectivity for PKM2 over the other isoforms we used
DynamixFit™ in silico screening technology.21 Based on available
PKM2 crystal structures, a 3D DynamixFit™ model of PKM2 was
generated and used to screen for compounds that target the
PKM2 dimer interface region.12 A library of 13,000,000 drug-like
compounds (collected from 30 different vendors and continuously
updated) was screened in silico against the PKM2 structure. The
top 200 selected compounds (‘‘virtual hits’’) were tested in vitro
in an enzymatic PKM2 activation assay. The screening identified
9 novel scaffolds of PKM2 activators12, including 2-((2,3-dihydro-
benzo[b][1,4]dioxin-6-yl)thio)-1-(2-methyl-1-(methylsulfonyl)
indolin-5-yl) ethanone, compound 1 (Fig. 1). Here, we describe the

0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.08.054
 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468 6461

O reagent), yielding the mono-brominated sulfonamide.22 Finally,

S O
N O
O Given that compound 1, discovered using in silico screening, is
S
O
Figure 1. Chemical structure of the PKM2 activator 1
development and optimization of a series of PKM2 activators de-
rived from 1.
The cell-free enzymatic assay of PKM2 activity was performed
as described (Supplementary data). Using this assay for PKM2 acti-
vators, compound 1 (as a racemic mixture) was found to have an
AC50 = 550 nM and a maximal activation of 164% relative to the en-
zyme’s basal activity level, an activation level that is comparable to
that achieved in this same assay by the natural activator FBP
(AC50 = 80 nM, max. activity 148%). Resynthesis of this hit was
achieved according to the methodology presented in Scheme 1.
The synthetic route selected for the chemical synthesis of 1 was
based on an assembly of literature procedures.
At the first stage the indoline-ethanone derivative was dis-
solved in THF in the presence of pyridine, then an equimolar
amount of methanesulfonyl chloride was added with stirring. The
reaction mixture was stirred at 40 °C for about 5 h to give the de-
sired sulfonamide product. Then, the sulfonamide ketone was sub-
mitted to a bromination reaction in THF using a bromination agent,
for example, phenyltrimethylammonium tribromide (Jacques
base-mediated coupling was performed with the commercially
available 2,3-dihydrobenzo[b][1,4]dioxine-6-thiol in acetone, to
give the desired product, compound 1.
commercially available, additional close analogs of 1 were also
purchased and screened in vitro for their PKM2 activation potency.
Results are presented in Table 1. All tested analogs showed activity
between 0.5–4 lM. Notably, altering the alkoxy groups as R1 sub-
stituents with di-fluoro (5) decreased the activity. Cyclic moieties
such as indoline (1) and tetrahydroquinoline (2) were more potent
than the acyclic sulfone amide (6). The AC50 range of these analogs
suggests that this scaffold is susceptible to further optimization.
The next step for validating compound 1 as a lead, and moving
towards the optimization stage, was to test its cell permeability
(Caco-2 assay) and in vitro metabolic stability (in human liver
microsoms, hLM). These tests (Table 2) showed that although 1
demonstrated good Caco-2 permeability with no efflux (efflux ratio
<1), it suffered from poor metabolic stability in human liver micro-
soms (t1/2 = 3.4 min). Analogs 3 and 6 demonstrated similarly poor
metabolic stability in human liver microsoms.
At this stage it was clear that structural modifications were nec-
essary in order to address this issue and generate metabolically
stable compounds that would be suitable for further optimization.
We began by looking at the importance of the thioethanone group
as the linker between the two aryl moieties of compound 1
(Table 3). As structural analysis suggested that shorter linkers
could still maintain key binding interactions, four different linkers

O
Cl S S O Br
Br
S O
H O O
N N +
O N N
+ Br

Pyridine THF Br
O O O

S O S O
O O
N O K2CO3 O
N
+
Br HS O Acetone S O
O O

Scheme 1. General synthetic methodology used for compound 1.

Table 1
SAR of 1-Aryl-2-(arylthio)ethanone

# R1 R2 R3 n hPKM2 AC50a (lM) hPKM2 Max Activityb (%)

O 1 3,4-ethylendioxy Me MeC 1 0.55 164

S 2 3,4-ethylendioxy Me H 2 1.3 107
R 3 n R 1 3 3,4-propylendioxy Et H 1 0.51 149
N 4 3-methoxy Et H 1 2.4 110
O S 5 2,4-difluoro Et H 1 4.1 171
O
R2 1-5
O
S
H R1
N 6 3,4-ethylendioxy Me — — 2.5 140
O S
O
2
R
6
a
AC50 values were determined using enzymatic PKM2 activity assay (Supplementary data).
b
Max Activation represents the maximal% activation of the enzymatic activity above its basal level (where 0% is enzyme activity in the absence of compound).
c
Racemic mixture.
6462 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468

Table 2 were prepared: two sulfonamides, sulfone and carbonyl. It should

In vitro metabolic stability in hLM and Caco-2 permeability for compound 1, 3 and 6 be noted that all compounds bearing a methyl substituent on the
Compound In vitro Caco-2 permeabilitya 2-indoline moiety represent racemic mixtures.
microsomal Papp (A B) Papp (B A) Effluxratio Both sulfonamides linkers, compounds 7 and 8, showed com-
stabilitya t1/2 (10 6, cm/s) (10 6, cm/s) plete loss of activity, as did the carbonyl group when used as the
(min)
hLM
linker (10). On the other hand, the sulfone linker in compound 9
preserved most of the activity of the parent compound 1. Fortu-
1 3.4 3.68 1.54 0.42
3 1.4 ND ND ND
nately, 9 also demonstrated a 10-fold improvement in metabolic
6 3.3 ND ND ND stability (hLM t1/2 = 31.5 min), a 10-fold improvement in Caco-2
a
permeability (Papp (A B) = 20.4 10 6 cm/s, Papp (B A) = 12.4 10
See Supplementary data for experimental procedures. 6
cm/s, and an efflux ratio of 0.6). Schemes 2–5 give the reaction
conditions for the preparation of compounds 7, 8, 9 and 10.
Sulfonamide analog 7 was prepared through coupling of the
Table 3 commercially available N-protected indoline, 1-Acetyl-2-
Linker variations between the two aryl moieties
methylindoline-5-sulphonyl chloride, and 2,3-dihydrobenzo
O [b][1,4]dioxin-6-amine followed by acidic deacetylation and meth-
O
ansulfonation of the indoline amine. The ‘‘opposite’’ sulfonamide,
N Linker O
S Compound 8, was prepared using the same methodology but in
O this case the amine group was on the indoline moiety and the sul-
fonylchloride on the benzodioxine building block.
Compounds 9 and 10, with either sulfone or carbonyl linkers,
# Linker hPKM2 AC50a (lM) hPKM2 Max Activityb (%)
were prepared using Friedel–Crafts acylation conditions.
7 -SO2-NH 51 25.5 Based on the improved properties of 9 over 1, compound 9 was
8 -NH-SO2- >100 32.5
selected as the new lead. It was then crystallized together with the
9 -SO2- 2 149.5
10 -CO- Not active — PKM2 enzyme and the structure of the complex was determined
using X-ray crystallography. The resulting 3D structure of the pro-
a
AC50 values were determined using enzymatic PKM2 activity assay (Supple-
tein-compound complex (PDB 4G58) confirmed that compound 9
mentary data).
b
Max Activation represents the maximal% activation of the enzymatic activity
indeed binds to the PKM2 dimer interface, exactly at the site pre-
above its basal level (where 0% is enzyme activity in the absence of compound). dicted by DynamixFit™, and in the predicted binding mode.12

O O

N
H2N O i N
O ii
+ O
O O
S S
N O
O Cl O H

O
H S
O
N N
O
O iii O
+ O
O
S
S Cl S
N O O N O
O H O H

7
Scheme 2. Reagents and conditions: (i) DIEA, THF, 20 h, rt; (ii) HCl, THF, 2 h rt; (iii) Pyridine, 20 h, rt.

O
H O
N O i S
ii
+ S
Cl
N
NO2 O
NO2

O
S O S
O
O O O
N
Cl
S O iii N
O
+ O S O
NH2 N
O H O

8
Scheme 3. Reagents and conditions: (i) Pyridine, 20 h, rt; (ii) H2/Pd-C 10%, EtOH, 20 h, rt; (iii) Pyridine, 20 h, rt.
 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468 6463

S O O
H O Cl
N O i N
S O
O
+ S
Cl +
O O

O
S
O
N
ii O
S O
O

O
9
Scheme 4. Reagents and conditions: (i) Pyridine, 20 h, rt; (ii) AlCl3, CH2Cl2, 20 h, 70 °C (Pressure tube).

O O O
S S
O O
N
Cl
O i N O

+
O O
O

10
Scheme 5. Reagents and conditions: (i) AlCl3, CH2Cl2, 20 h, rt.

Exploring the interactions of 9 with the PKM2 binding site

(Fig. 2) revealed that two Phe26 residues, from both monomers,
made p–p interaction with the phenyl ring of compound 9. One
of these Phe26 residues also formed a T-shape interaction with
the benzene-ethylenedioxy moiety of compound 9, while its sul-
fone group formed an H-bond interaction with Tyr390 backbone,
coordinated by the conserved water molecules. With these insights
we also docked into the binding site the two less active sulfon-
amide linked compounds (7 and 8) in an attempt to understand
their loss of activity compared to 9 (Fig. 3).
As seen in Figure 2, the binding site at the PKM2 dimer interface
includes two conserved water molecules, at least one of which was
observed in other PKM2 crystal structures as well.23 It is hypothe-
sized that compounds that replace one or two of the conserved
water molecules while maintaining their network of interactions
will retain potency, as these are assumed to be the most important
interactions between the ligand and the target protein. On the
other hand, compounds that do not maintain these interactions
either eliminate them altogether or alter their electrostatic charac-
teristics, are predicted to lose potency.
When docked into the binding site (Fig. 3(A)) Compound 7,
which lost most of its activity (AC50 = 51 lM; Max%Act = 25.5%), Figure 2. Compound 1 in the PKM2 binding site as determined by X-ray
followed the above hypothesis. Having the additional NH group crystallography
elongated the compound, caused it to clash with the walls of the
binding site and disrupted the interaction with the water molecule Based on the experimental data and structural analysis, 9 was
(Fig. 3(B)). Furthermore, this additional NH group also changed the selected as the new lead compound due to its improved human
electrostatic characteristics of the adjacent benzene ring, modify- metabolic stability (as measured by hLM half-life) relative to hit
ing the T-shape interaction of this ring with Phe26 (Fig. 3(C); for 1 while it maintained much of its PKM2 activity. Lead optimization
comparison the bioactive conformation of compound 9 from the started with structural modifications of compound 9. Due to the
X-ray structure is colored orange). A similar analysis was per- beneficial impact of the ethylenedioxy moiety and the sulfone lin-
formed with compound 8, which was completely inactive ker, we kept these motifs and performed a SAR exploration focus-
(AC50 > 100 lM Max%Act = 32.5%; Figure 3(D)), and showed the ing on the indoline moiety, hypothesizing that this region could be
same disruptive effect of the additional NH group (Fig. 3(E)). How- used to modulate activity. All suggested modifications were ana-
ever, in this case we also saw that the planarity between the two lyzed, prioritized and selected before actual synthesis, using our
Phe 26 groups was disrupted with compound 8 (Fig. 3(F)). To over- DynamixFit™ technology and the experimentally determined
come the steric clashes, the sulfone moiety was flipped in the binding mode of compound 9.
docked conformation, leading to a loss of the important interaction We began by looking at potential modification of sulfonamides
with the conserved water (enlargement in Fig. 4). substituents (Table 4) with benzene rings.
6464 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468

A O
D
O
S S
O O O
N N
O
O
O
S O
S N
N H O
O
O H

7 8
B E

C F

Figure 3. Compound 7 and 8 docked into the PKM2 dimer interface binding site; (A) Chemical structure of 7; (B) space fill representation of 7 in the binding site; (C) stick
representation of 7 in the binding site; (D) Chemical structure of 8; (E) space fill representation of 8 in the binding site; (F) stick representation of 8 in the binding site;

Table 4
2-methyl indoline sulfonamides variations
O O

S
O
N
O O
S
O
R

# R hPKM2 AC50a (lM) hPKM2 Max Activityb (%)

9 Me 2.0 149
11 3-methoxybenzne 0.80 119
12 4-fluorobenzne 0.60 159
13 2,6-difluorobenzne 0.50 168
Figure 4. Overlay of the docked conformation of compound 8 and crystallography 14 Ph 0.60 134
determined conformation of compound 9 (colored orange). 15 benzyl 1.6 111
a
AC50 values were determined using enzymatic PKM2 activity assay (Supple-
mentary data).
In general, all of the aromatic modifications improved activity b
Max Activation represents the maximal% activation of the enzymatic activity
compared to the methyl derivative of lead compound 9. above its basal level (where 0% is enzyme activity in the absence of compound).
 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468 6465

A B

Figure 5. Docking of compound 14 into the PKM2 dimer interface binding site (gray): (A) compared to the experimentally determined bond conformation of compound 9
(orange); (B) compared to the docking pose of compound 15 (green).

Table 5
Acyclic sulfonamides variations
the additional benzene ring improved potency because it better
O O filled the volume of the binding site, added a second T-shape inter-
S action with the other Phe 26 residue from monomer B, and posi-
R1
O tioned the central ring more symmetrically between the two Phe
N
O O 26 rings. On the other hand, replacing the benzene ring of 14 with
S R3
R2
O a benzyl (compound 15; AC50 = 1.6 lM,% Max Act = 111%) resulted
in the loss of most of this extra potency, leaving 15 only slightly
# R1 R2 R3 hPKM2 AC50a (lM) hPKM2 Max Activityb (%) more potent than 9. Figure 5(B) shows that compound 15 retained
many of the beneficial interactions of compound 14: a better fill of
16 Et Me H 41 100
17 i-Pr Me H 2.8 194 the binding site volume, an additional T-shape interaction with the
18 Et Me Me 2.7 160 second Phe 26 from monomer B, and the symmetrical positioning
a of the central aromatic ring between the two Phe 26 groups. How-
AC50 values were determined using enzymatic PKM2 activity assay (Supple-
mentary data). ever, it seems that these advantages were countered by steric hin-
b
Max Activation represents the maximal% activation of the enzymatic activity drance into the binding site, possibly clashing with the site
above its basal level (where 0% is enzyme activity in the absence of compound). boundaries.
Compounds bearing acyclic sulfonamides (Table 5) were also
examined. Generally, acyclic sulfonamide analogs did not improve
potency compared to the respective cyclic analogs, in agreement
Table 6
with what was found earlier with the acyclic thioethanone linked
Indoline sulfonamides variations
compound 6 (Table 1). Compound 17, which is the direct acyclic
O O analog of 9, exhibited some loss of activity (AC50 = 2.8 lM vs.
S 2.0 lM, respectively).
O Given that heterocyclic sulfonamides seemed to have better
N
O O activity relative to the acyclic analogs, we further considered mod-
S
R2
O R1 ifications of the indoline sulfonamides (Table 6).
Removing the 2-methyl group (19), eliminating the chiral cen-
ter, gave an almost 100-fold improvement in activity to the low
# R1 R2 hPKM2 AC50b (lM) hPKM2 Max Activityc (%) nM range. Since the chiral methyl disrupted the planarity of the
19 H Me 0.045 106 indoline, removing it regains the indoline planarity and improves
20 H Et 0.027 89 its stacking between the two Phe 26 residues. Ethyl and cyclopro-
21 H c-Pr 0.017 105 pyl substituents (20 and 21) further improved the potency. Aro-
22 H Ph 0.093 122
matic sulfonamide substituents of 19 also retained this new level
23 di- Me 0.079 78
Fa of potency (compounds 22 AC50 = 93 nM; compound 24
24 H 3- 0.03 87 AC50 = 30 nM).
fluorobenzne To test the importance of the indoline five member ring, a six
a
di-ortho relative to the sulfone linker. member ring tetrahydroquinoline compound was prepared (com-
b
AC50 values were determined using enzymatic PKM2 activity assay (Supple- pound 25; Table 7). This compound showed significant loss of po-
mentary data). tency, in agreement with the loss of potency in 2, the
c
Max Activation represents the maximal% activation of the enzymatic activity tetrahydroquinoline analog of 1 (Table 1). We also tested the effect
above its basal level (where 0% is enzyme activity in the absence of compound).
of changing the non-aromatic five member ring of the indoline to
the aromatic five member ring on an indole. Compounds 26 and
Specifically, the substituted benzene (11–13) and the 27 were prepared to test this effect. In general, the activity of the
unsubstituted phenyl (14) moieties improved the compounds’ indole analogs was similar to that of the indoline compounds. Spe-
activity, while adding the bulkier benzyl derivative (15) led only cifically, compound 26 was as active as compound 19, while adding
to a modest improvement in activity over the lead compound 9. a methyl substituent (27) resulted in lower activity.
Figure 5(A) compares the docked conformation of compound 14 The next SAR step explored altering the sulfonamide position on
(AC50 = 0.6 lM,% Max Act = 134%) in the PKM2 binding site and the the five member hetrocyclic ring, from indoline to isoindoline. Iso-
experimentally determined bound conformation of lead compound indoline analogs were prepared according to the reaction route
9 (AC50 = 2.0 lM,% Max Act = 148%). This comparison showed that presented in Scheme 6.
6466 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468

Table 7
Sulfonamide heterocyclic ring variation

# hPKM2 AC50a (lM) hPKM2 Max Activityb (%)

O O

S
O
N O 2.3 152
O
S
O

25
O O

S
O
N 0.023 98
O O
S
O
26
O O

S
O 2.72 154
N
O O
S
O
27
a
AC50 values were determined using enzymatic PKM2 activity assay (Supplementary data).
b
Max Activation represents the maximal% activation of the enzymatic activity above its basal level (where 0% is enzyme activity in the absence of compound).

O
HO S Cl
O O O
i O
N S
NH + Cl S N S O
O O ii S O
Cl
O

O
O
O S N O
O S O
iii O

Scheme 6. Reagents and conditions: (i) Pyridine, 20 h, rt. (ii) 6 h rt; (iii) AlCl3, CH2Cl2, 20 h, rt.

with the methyl and ethylsulfonamides indoline analogs (19 and

Table 8
Isoindoline sulfonamides variations 20), the methyl and ethylsulfonamides isoindoline analogs (28
and 30) were highly potent (compound 28: AC50 = 17 nM,% Max
O O
O Activity = 125%; compound 30: AC50 = 20 nM, although with low%
R1 S N S Max Activity of 35%).
O As mentioned above, PKM2 is naturally activated by the up-
O O
stream glycolysis intermediate FBP (fructose 1,6-bis-phosphate),
which binds at an allosteric site and modulates the kinetics of
# R1 hPKM2 AC50a (lM) hPKM2 Max Activityb (%)
the enzymatic reaction (PEP + ADP?pyruvate + ATP). To check
whether our compounds, which also activate this enzyme albeit
28 Me 0.017 125
29 i-Bu 0.189 91.5
at a different non-FBP allosteric site, are able to modulate the enzy-
30 Et 0.02 35 matic kinetics of the PEP reaction, we checked the reaction kinetics
31 Ph 0.061 140 in the presence and absence of compound. Figure 6 compares the
a effect of compound 19 and of FBP on this enzymatic kinetics. In
AC50 values were determined using enzymatic PKM2 activity assay (Supple-
mentary data). the absence of activator (FBP or 19), PKM2 shows low affinity for
b
Max Activation represents the maximal% activation of the enzymatic activity PEP (KM = 254 lM). In the presence of either compound 19 or FBP
above its basal level (where 0% is enzyme activity in the absence of compound). a decreased KM for PEP was observed, KM of 124 lM with 19 and
53 lM with FBP, with no significant effect on Vmax (calculated Vmax
values of 0.34 lM/min without FBP, 0.26 lM/min with FBP, and
Commercially available isoindolines were reacted with substi- 0.36 lM/min with compound 19). Overall, this study demonstrated
tuted chlorosulfonates giving the sulfonamides, which were then that the kinetic effect of compound 19 was similar to the effect of
chlorosulfonated on the benzene ring followed by a Friedel–Craft the natural activator FBP, that is, compound 19 activated PKM2 by
reaction with 1,2-ethylenedioxybenzene. As shown in Table 8, increasing the enzyme’s affinity for PEP in agreement to previous
the isoindoline analogs also exhibied low nM range activity. As reports for activation with FBP11, and also with previously reported
 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468 6467

Figure 6. (A) Initial velocity of PKM2 reaction as a function of PEP concentration in the presence (filled circles) or absence (open circles) of FBP (10 lM). (B) Initial velocity of
PKM2 reaction as a function of PEP concentration in the presence (filled circles) or absence (open circles) of 19 (10 lM). V0, initial rate, defined as ATP production from PEP
and ADP by pyruvate kinase M2. Kinetic assays were carried out with 0.05 nM PKM2 using [ADP] = 5 mM (see Supplementary data).

O O

S
O
N
O O
S
O

19

Figure 7. Pyruvate kinase isoform selectivity profile of Compound 19. PKM2 (filled circles), PKM1 (circles), PKL (filled squares), PKR (squares).% activity was measured and
analyzed as described (Supplementary data).

10. All compounds tested exhibited good metabolic stability in hu-

Table 9 man liver microsoms, excellent permeability with no efflux, no
Compound 19 selectivity against all PK isoforms. hERG inhibition, and favorable protein binding.
In conclusion, we have developed a novel chemotype capable of
Compound PKM2 PKM1 PKL PKR
activating PKM2. These molecules, which are based on the 5-((2,3-
19 AC50 (lM) 0.045 NOT ACTIVE >100 >100
dihydrobenzo[b][1,4]dioxin-6-yl)sulfonyl)-2-methyl-1-(methyl-
% Max Act 106 NOT ACTIVE 33 34
sulfonyl) indoline scaffold, have low nanomolar PKM2 potency with
excellent selectivity over the other PK isoforms and good ADME
properties (Caco-2 permeability, metabolic stability in human liver
activators.19,20 In contrast, varying of the concentration of ADP in microsoms, protein binding, no efflux, and no hERG inhibition).
the presence or absence of activators has no significant effect on Following the discovery of this novel chemotype, the aim is to devel-
the kinetics, as previously observed. op these molecules into pharmacological agents able to inhibit tu-
Compound 19 was also tested for selectivity against the other mor growth in vivo. New lead compounds will be characterized in
PK isoforms. As is shown in Figure 7, and summarized in Table 9, cancer cell line assays and in vivo cancer models. The presented data
compound 19 has no significant activity against the other three demonstrate that selective targeting of PKM2 with a drug-like mol-
PK isoforms, R, L and M1, maintaining a selectivity window greater ecule is possible and suggest that efforts to target PKM2 may yield
than 2000-fold over the next PK isoforms (PKL and PKR), and with compounds suitable for targeting cancer metabolism for cancer
virtually no effect on the common, and widely prevalent isoform, therapy.
PKM1.
A number of analogs with low nM activity were tested for their Supplementary data
ADME properties, including cell permeability (Caco-2), in vitro
metabolic stability in human liver microsoms (hLM), hERG inhibi- Supplementary data associated with this article can be found, in
tion, and serum protein binding. This data is summarized in Table the online version, at http://dx.doi.org/10.1016/j.bmcl.2012.08.
054.
Table 10
Microsomal stability, caco-2 permeability hERG inhibition and protein binding for References and notes
compounds 9, 19 and 28.
1. Vander Heiden, M. G.; Cantley, L. C.; Thompson, C. B. Science 2009, 1029, 324.
Compound In vitro Caco-2 permeabilitya hERGa% Protein 2. Warburg, O.; Posener, K.; Negelein, E. Biochem Zeitschr. 1924, 152, 309.
metabolic Papp Papp Effuxratio inhibition Binding 3. Warburg, O. Science 1956, 123, 309.
stabilitya (A B) (B A) at 1 lM (%)a 4. Hsu, PP; Sabatini, DM Cell 2008, 134, 703.
t1/2 (min) (10 6, (10 6, 5. Ferreira, L. M. Exp. mol. pathol. 2010, 89, 372.
hLM cm/s) cm/s) 6. Kenzaburo, T.; Yoshida, M. C.; Hitoshi, S.; Keiji, M.; Tamio, N.; Takehiko, T.;
Hisaichi, F.; Shiro, M. Gene 1988, 73, 509.
9 31.5 20.4 12.1 0.59 ND 92.5
7. Satoh, H.; Tani, K.; Yoshida, M. C.; Sasaki, M.; Miwa, S.; Fujii, H. Cytogenet. Cell
19 26.0 37.5 43.0 1.14 9.0 94.1
Genet. 1988, 47, 132.
28 >45 30.4 32.4 1.06 7.3 90.9 8. Takenaka, M.; Yamada, K.; Lu, T.; Kang, R.; Tanaka, T.; Noguchi, T. Eur. J.
a Biochem. 1996, 235, 366.
See Supplementary data for experimental procedures.
6468 A. Yacovan et al. / Bioorg. Med. Chem. Lett. 22 (2012) 6460–6468
9. Noguchi, T.; Inoue, H.; Tanaka, T. J Biol. Chem. 1986, 261, 13807.
10. Noguchi, T.; Yamada, K.; Inoue, H.; Matsuda, T.; Tanaka, T. J Biol. Chem. 1987,
262, 14366.
11. Dombrauckas, J. D.; Santarsiero, B. D.; Mesecar, A. D. Biochem. 2005, 44, 9417.
12. Shitrit, A., et. al. Submitted for publication.
13. Mazurek, S.; Boschek, CB; Hugo, F; Eigenbrodt, E Semin. Cancer Biol. 2005, 15,
300.
14. Mazurek, S. Int. J. Biochem. Cell Biol. 2011, 43, 969.
15. Wolf, A.; Agnihotri, S.; Guha, A. Oncotarget. 2010, 1, 552.
16. Christofk, H. R. et al Nature 2008, 452(7184), 230.
17. Boxer, M. B.; Jiang, J-K; Vander Heiden, M. G; Shen, M; Skoumbourdis, A. P.;
Southall, N.; Veith, H.; Leister, W.; Austin, C. P.; Park, H.; Inglse, J.; Cantley, L. C.;
Auld, D. S.; Thomas, C. J. J. Med. Chem. 2010, 53, 1018.
18. Jiang, J-K; Boxer, M. B.; Vander Heiden, M. G; Shen, M; Skoumbourdis, A. P.;
Southall, N.; Veith, H.; Leister, W.; Austin, C. P.; Park, H.; Inglse, J.; Cantley, L. C.;
Auld, D. S.; Thomas, C. J. J. Bioorg. Med. Chem. 2010, 20, 3387.
19. Walsh, M. J.; Brimacombe, K. R.; Veith, H.; Bougie, J. M.; Daniel, T.; Leister, W.;
Cantley, L. C.; Israelsen, W. J.; Vander Heiden, M. G.; Shen, M; Auld, D. S;
Thomas, C. J; Boxer, M. B. J. Bioorg. Med. Chem. Lett. 2011, 21, 6322.
20. Salituro, F. G.; Saundress, J. O. PCT Int. Appl WO2010118063.
21. Bloch, I., et al. Submitted for publication.
22. Malecha, J.; Noble, Stewart.; Hassig, Chritian.; Wash, Paul.; Wiley, Brandon.;
Lawrence, Chales.; Hoffman, Timothy. PCT Int. Appl WO2005120515.
23. PDB codes: 3GQY, 3GR4 and 3H6O.