A pairwise chemical genetic screen identifies new inhibitors of glucose transport

Article abstract of DOI:10.1016/j.chembiol.2010.12.015

Oxidative phosphorylation (OXPHOS) and glycolysis are the two main pathways that control energy metabolism of a cell. The Warburg effect, in which glycolysis remains active even under aerobic conditions, is considered a key driver for cancer cell proliferation, malignancy, metastasis, and therapeutic resistance. To target aerobic glycolysis, we exploited the complementary roles of OXPHOS and glycolysis in ATP synthesis as the basis for a chemical genetic screen, enabling rapid identification of novel small-molecule inhibitors of facilitative glucose transport. Blocking mitochondrial electron transport with antimycin A or leucascandrolide A had little effect on highly glycolytic A549 lung carcinoma cells, but adding known glycolytic inhibitors 2-deoxy-D-glucose, iodoacetate or cytochalasin B, rapidly depleted intracellular ATP, displaying chemical synthetic lethality. Based on this principle, we exposed antimycin A-treated A549 cells to a newly synthesized 955 member diverse scaffold small-molecule library, screening for compounds that rapidly depleted ATP levels. Two compounds potently suppressed ATP synthesis, induced G1 cell-cycle arrest and inhibited lactate production. Pathway analysis revealed that these novel probes inhibited GLUT family of facilitative transmembrane transporters but, unlike cytochalasin B, had no effect on the actin cytoskeleton. Our work illustrated the utility of a pairwise chemical genetic screen for discovery of novel chemical probes, which would be useful not only to study the system-level organization of energy metabolism but could also facilitate development of drugs targeting upregulation of aerobic glycolysis in cancer.

Full text of DOI:10.1016/j.chembiol.2010.12.015

Chemistry & Biology
Article
A Pairwise Chemical Genetic Screen Identifies
New Inhibitors of Glucose Transport
Olesya A. Ulanovskaya,1 Jiayue Cui,1 Stephen J. Kron,2 and Sergey A. Kozmin1,
*
1Department of Chemistry
2Department of Molecular Genetics and Cell Biology
The University of Chicago, Chicago, IL 60637, USA
*Correspondence: skozmin@uchicago.edu
DOI 10.1016/j.chembiol.2010.12.015
SUMMARY
Most recently, the advent of high-throughput screening has
accelerated chemical probe discovery (Bredel and Jacoby,
Oxidative phosphorylation (OXPHOS) and glycolysis 2004). However, while significant progress toward identification
of compounds perturbing many key pathways has been made,
developing highly specific chemical probes remains challenging.
A particularly powerful approach has been to exploit synthetic
are the two main pathways that control energy
metabolism of a cell. The Warburg effect, in which
glycolysis remains active even under aerobic condi-
tions, is considered a key driver for cancer cell prolif-
eration, malignancy, metastasis, and therapeutic
resistance. To target aerobic glycolysis, we exploited
the complementary roles of OXPHOS and glycolysis
in ATP synthesis as the basis for a chemical genetic
lethality, where a defined genetic defect sensitizes the cell to
small molecules that target compensatory pathways (Hartwell
et al., 1997). By analogy with classical genetic analysis of inter-
acting genes (Tong et al., 2001, 2004), only via combining the
mutation with the proper small molecule can one observe the
phenotype, as either perturbation alone is insufficient. This
screen, enabling rapid identification of novel small-
molecule inhibitors of facilitative glucose transport.
approach is limited, however, by the availability of mutant cell
lines and RNAi may not offer a satisfactory alternative. Alterna-
Blocking mitochondrial electron transport with anti- tively, a chemical probe can substitute for the mutation, and
the compensatory response of the system might then be tar-
mycin A or leucascandrolide A had little effect on
geted by a second small molecule, which can be selected from
a chemical library. Here, the pairwise chemical perturbation
can result in a unique phenotype and enable the discovery of
highly glycolytic A549 lung carcinoma cells, but add-
ing known glycolytic inhibitors 2-deoxy-D-glucose,
iodoacetate or cytochalasin B, rapidly depleted intra-
cellular ATP, displaying chemical synthetic lethality.
Based on this principle, we exposed antimycin
A-treated A549 cells to a newly synthesized 955
new chemical probes. Particularly where prior genetic analysis
has identified the compensatory cellular pathway, linking the
small molecules to their targets is highly feasible.
Oxidative phosphorylation (OXPHOS) and glycolysis are the
member diverse scaffold small-molecule library,
screening for compounds that rapidly depleted ATP
levels. Two compounds potently suppressed ATP
two main pathways that control energy metabolism in the cell.
The interdependence of the two metabolic pathways has been
known since Pasteur’s pioneering work, which demonstrated
synthesis, induced G1 cell-cycle arrest and inhibited that yeast consumed more glucose anaerobically than
aerobically (Racker, 1974). Recent systematic analysis of all
single and double knockouts of 890 metabolic genes in
lactate production. Pathway analysis revealed that
these novel probes inhibited GLUT family of facilita-
Saccharomyces cerevisiae demonstrated that genetic perturba-
tions of OXPHOS aggravated disruption of glycolysis, because
either fermentation or respiratory function were needed for
tive transmembrane transporters but, unlike cyto-
chalasin B, had no effect on the actin cytoskeleton.
Our work illustrated the utility of a pairwise chemical
`
ATP synthesis (Segre et al., 2005). Pairwise chemical perturba-
genetic screen for discovery of novel chemical
probes, which would be useful not only to study the
system-level organization of energy metabolism but
could also facilitate development of drugs targeting
upregulation of aerobic glycolysis in cancer.
tion of OXPHOS and glycolysis has also been explored in human
cancer cell lines. The combination of small-molecule inhibitors of
mitochondrial electron transport chain and glucose catabolism
synergistically suppressed ATP production and impaired cellular
viability (Ulanovskaya et al., 2008; Liu, et al., 2001). However, the
ability to carry out chemical genetic studies of energy metabo-
lism is currently limited by the availability of potent, specific,
and stable chemical inhibitors of glycolysis (Pelicano et al.,
2006). Such compounds would be useful not only to study the
INTRODUCTION
In chemical genetics, small-molecule probes rather than muta- systems-level organization of metabolism in real time, but could
tions are used to modulate cellular phenotypes, thereby offering also open new directions for discovery of drugs targeting the up-
regulation of aerobic glycolysis in cancer discovered by Warburg
(Warburg, 1956; Vander Heiden et al., 2009; Tennant et al., 2010;
access to biological insights that may not be obtained by
conventional genetics (Stockwell, 2000; Lehar et al., 2008).
222 Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
Gatenby and Gillies, 2004; DeBerardinis et al., 2008; Gohil et al.,
2010).
Here, we exploited dual contribution of the two main energy-
producing cellular pathways to production of ATP for the devel-
opment of a practical chemical genetic screen, which enabled
rapid identification of new small-molecule inhibitors of facilitative
glucose transport. This approach was based on the initial
suppression of OXPHOS in A549 cells with a potent and specific
small-molecule inhibitor of complex III. This treatment alone did
not result in any observable defects in cellular viability or ATP
production within the first 30 min of drug incubation. Subse-
quently, a second chemical perturbation of the system with
a small-molecule inhibitor of glycolysis or glucose transport re-
sulted in synergistic, rapid depletion of intracellular ATP levels.
Having validated this synthetic effect using a series of known
inhibitors, we subjected antimycin A-treated A549 cells to
a newly synthesized 955 member small-molecule library and
measured effects of each library member on ATP production.
The screen identified two new compounds that potently sup-
pressed ATP synthesis only in the presence of a mitochondrial
inhibitor, induced G1 cell-cycle arrest and inhibited lactate
production, which is highly indicative of blocking glycolytic
pathway. We further demonstrated that the two newly identified
chemical agents potently inhibited glucose cellular uptake,
which is mediated by the GLUT family of facilitative transmem-
brane transporters. Our investigation demonstrated the utility
of a pairwise chemical genetic screen for the discovery of novel
chemical probes, and identified two previously unknown chemo-
types that enable effective suppression of facilitative glucose
transport in mammalian cells.
RESULTS
Synergistic Suppression of ATP Synthesis
Glycolysis is a series of metabolic reactions that convert each
glucose to two pyruvic acids and two equivalents of ATP (Fig-
ure 1A). OXPHOS produces up to 36 additional equivalents of
ATP by coupling reduction of molecular oxygen to oxidation of
NADH and FADH2. We have previously observed (Ulanovskaya,
et al., 2008) that combining small-molecule inhibitors of OX-
PHOS and glycolysis resulted in rapid depletion of cellular ATP
while the action of a mitochondrial inhibitor alone did not
substantially impact the ATP level under the same conditions
(see Figure S1 available online). Based on this observation, we
profiled several pairwise combinations of known small-molecule
inhibitors of OXPHOS and glycolysis to establish the generality of
synthetic suppression of ATP production in mammalian cells.
2-Deoxy-D-glucose (3) is an established inhibitor of glycolysis
that acts by blocking the activity of phosphoglucose isomerase
following its initial phosphorylation by hexokinase (Brown,
1962). Treatment of A549 cells with variable concentrations of
3 resulted in partial suppression of ATP levels (Figure 1B) within
the first 30 min of incubation with the drug. This effect was
substantially enhanced (Figures 1B and 1C) in the presence of
a chemical inhibitor of complex III – either the synthetic analog
Figure 1. Synergistic Suppression of ATP Synthesis
(A) Schematic representation of the contribution of OXPHOS and glycolysis to
ATP production and structures of known small-molecule inhibitors of the two
energy-producing pathways, including a synthetic analog of leucascandrolide
A (1), antimycin A (2), 2-deoxy-D-glucose (3), sodium iodoacetate (4), and cyto-
chalasin B (5).
(B) Suppression of ATP synthesis by 3 in the absence and presence of 1.
(C) Suppression of ATP synthesis by 3 in the absence and presence of 2.
(D) Suppression of ATP synthesis by 4 in the absence and presence of 2.
(E) Suppression of ATP synthesis by 5 in the absence and presence of 2.
All values are presented as percentage of vehicle treated sample. Each value is
mean SEM of duplicate values from a representative experiment. This figure
is supported by Figure S1.
of leucascandrolide A (1) developed in our laboratory (Ulanov- doacetate (4) (Figure 1D), which irreversibly inhibits glyceroalde-
skaya, et al., 2008) or antimycin A (2) (Ohnishi, and Trumpower, hydephosphate dehydrogenase (Sabri and Ochs, 1971). The
1980). Similarly, a more effective suppression of ATP synthesis most pronounced effect on synergistic suppression of ATP
was observed using a combination of antimycin A and sodium io- synthesis was observed in the case of cytochalasin B (5),
Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved 223

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
Figure 2. Synthesis of 955 Member Chemical Library
The assembly process entailed a four-step synthetic sequence shown using 5 ketoesters (I), 16 amines II, which were employed during the first diversification
step, and 12 amines (V), which were used for the final amidation. For each 96-compound set, 12 randomly selected compounds were analyzed by 500 MHz 1H
NMR to determine chemical purity and yield as described in Experimental Procedures. Also shown are the structures of five representative library members 6–10,
which were randomly selected and fully characterized by 1H NMR, 13C NMR and MS. This figure is supported by Figure S2.
a known inhibitor of facilitative glucose transport (Deves and vinylogous amides III to a cyclic anhydride (Figure 2), followed
Krupka, 1978). While treatment of A549 cells with 5 did not by intramolecular amidation to give highly substituted five-
substantially impact the ATP level within the first 30 min of incu- membered lactams IV, which could be further diversified by acti-
bation, the addition of a mitochondrial inhibitor 2 resulted in vation of carboxylic acid and coupling with amines V en route to
highly effective blockage of ATP synthesis at low concentrations the target small-molecule library VI. The use of cyclic vinylogous
of both 2 and 5 (Figure 1E).
amides III with varied connectivity between R1 and R2 groups
These results confirmed that inhibition of mitochondrial respi- would result in scaffold diversification and production of the final
ration could enhance the suppression of ATP synthesis by glyco- library with the increased level of structural diversity.
lytic inhibitors independently of their mechanism of action. The
The assembly process began with condensation of 5 ketoest-
detailed pattern of ATP suppression, however, appeared to be ers I with 16 primary amines II, which readily occurred at 70ꢀC in
uniquely associated with a specific mode of action of each CHCl3 and afforded the desired 80 vinylogous amides III in 60%–
glycolytic inhibitor tested, providing a unique phenotypic 90% yield after chromatographic purification (Figure 2).
readout for each compound. These data establish the potential Construction of pyrrolidinones was achieved by treatment of vi-
´
for discovery of new glycolytic inhibitors by screening chemical nylogous amides III with maleic anhydride (Cave et al., 1997) in
libraries for small molecules that would act in concert with a mito- CHCl3 at 20ꢀC, followed by conversion of the resulting carboxylic
chondrial inhibitor to block ATP synthesis, as well as identifica- acids into the corresponding N-hydroxysuccinimide esters IV
tion of their cellular targets within the glycolytic pathway by using resin-bound carbodiimide to facilitate chromatographic
examining the pattern of inhibition for each agent.
purification of activated esters. Each of the esters IV was
produced as a single diasteromer. The relative stereochemical
relationship of the newly created stereogenic centers was veri-
Chemical Library Synthesis
High-throughput organic synthesis enables construction of fied by X-ray crystallography. The final stage of the synthesis en-
biogenic, structurally diverse small-molecule libraries for broad tailed condensation of 12 amines V with 80 activated esters IV.
biological screening (Tan, 2005). We selected a five-membered The resulting library VI was prepared in solution on 2.5 mmol
pyrrolidinone subunit as a structural element for rapid introduc- scale (1.0–1.5 mg of final products) and was rapidly purified by
tion of molecular diversity due to the prevalence of this chemo- parallel preparative thin layer chromatography. Chromato-
type in various biologically active compounds, i.e., doxapram graphic analysis of all final compounds (see Supplemental
and lactacystin (Yost, 2006; Omura et al., 1991). Selection of Experimental Procedures) established that 955 out of 960
such bioactive privileged structures for library design is attrac- compounds were produced successfully. In addition, NMR
tive to yield collections of compounds with favorable physico- spectroscopy was employed to validate high chemical purity
chemical properties in order to enable new lead discovery (Cui and quantify chemical yields of 120 randomly selected library
et al., 2010; Lee et al., 2010). The tandem reaction sequence members. Structures of five fully characterized library members
for library synthesis entailed intermolecular Michael addition of 6–10 (Figure 2) are representative of the notable skeletal
224 Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
Figure 3. Effects of 11 and 12 on ATP Synthesis,
Cell Proliferation, and Lactate Production
(A) Chemical structure of 11.
(B) Chemical structure of 12.
(C) Inhibition of intracellular ATP production upon treat-
ment of A549 and CHO-K1 cells with 11 in the presence
or absence of 10 nM antimycin A.
(D) Inhibition of intracellular ATP production upon treat-
ment of A549 and CHO-K1 cells with 12 in the presence
or absence of 10 nM antimycin A.
(E) Effect of 11 on growth of A549 and CHO-K1 cells.
(F) Effect of 12 on growth of A549 and CHO-K1 cells.
(G) Inhibition of lactate production in CHO-K1 cells upon
treatment with 11.
(H) Inhibition of lactate production in CHO-K1 cells upon
treatment with 12.
All values are presented as percentage of vehicle treated
sample. Each value is mean SEM of duplicate values
from a representative experiment. This figure is supported
by Figure S3.
and subjected to a series of detailed cell-based
and biochemical studies. Initial structure-activity
analysis (Figures S3A and S3B) revealed that the
naphthyl moiety in 11 and 12 can be replaced
with other aromatic groups without a substantial
decrease in activity. However, replacement of
the N-acyl piperazine with other groups in 11
could not be tolerated. Similarly, the dimethoxy-
phenylethyl subunit proved to be uniquely
responsible for the activity of 12 within the initial
collection of compounds tested.
Suppression of ATP Synthesis, Cell
Growth, and Lactate Production
Dose-dependent inhibition of ATP production by
11 and 12 was next evaluated in several
mammalian cell lines in the absence and pres-
diversity, which was enabled by the selection of our unique
synthetic strategy. Favorable physicochemical properties of all
newly produced compounds were further confirmed by desired
molecular weight and lipophilicty distributions (Figure S2).
ence of a mitochondrial inhibitor, such as antimycin A. We found
that each of the two compounds did not substantially alter the
level of ATP in A549 cells with normal mitochondrial activity.
However, in the presence of antimycin A, the amount of ATP
was depleted in A549 cells within 30 min of incubation with 11
or 12 with IC50 of 10 and 3 mM, respectively (Figures 3C and
Pairwise High-Throughput Chemical Genetic Screen
In order to identify new inhibitors of glycolysis, we subjected anti- 3D). Notably, ATP levels in Chinese hamster ovary cells (CHO-
mycin A-treated A549 cells to each member of the 955 member K1) were highly sensitive to both compounds 11 and 12 with
small-molecule library and measured the effects of each IC50 values of 2 and 1 mM, respectively (Figures 3C and 3D).
compound on ATP production after 30 min of drug incubation. The same profile of synergistic ATP suppression was observed
We anticipated that synergistic suppression of ATP synthesis in PC3 (prostate) and U373 (glioma) cell lines (Figures S3C–
under such conditions could be achieved only by compounds S3F). In addition, the synergistic effect of compounds 11 and
that inhibited either glucose metabolism or its transport into 12 on suppression of ATP synthesis in the presence of antimycin
the cells. The primary screen was performed with 10 nM concen- A was similar to that observed by replacing glucose with pyru-
tration of antimycin A and a 20 mM average concentration of each vate (Figure S3G). Importantly, the ATP suppression activity
individual library member. In order to identify the most potent patterns elicited by 11 and 12 were almost identical to that
small molecules, the initial set of active compounds was sub- observed for cytochalasin B (5) including both recapitulating
jected for a follow-up dose-dependent analysis of suppression the negligible effect on ATP production under normal respiration
of ATP synthesis in the presence and absence of antimycin A. in A549 (Figure 1E), as well as the enhanced sensitivity of CHO-
Structures of the two most potent compounds 11 and 12 are K1 to 5 (Figure S3H). These results suggested that the two newly
shown in Figures 3A and 3B. Each was resynthesized on prepar- identified compounds might similarly act by inhibiting glucose
ative scale, purified by conventional silica gel chromatography uptake into mammalian cells. Considering the supersensitivity
Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved 225

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
Figure 4. Inhibition of Glucose Transport by
11 and 12
(A) Inhibition of 3-O-methylglucose uptake in CHO-K1
cells by 11.
(B) Inhibition of 2-deoxy-D-glucose uptake in CHO-K1
cells by 11.
(C) Inhibition of D-glucose uptake in erythrocyte ghosts by
11.
(D) Inhibition of D-glucose uptake in erythrocyte ghosts by
12.
(E) Kinetics of the inhibition of D-glucose uptake in eryth-
rocyte membranes by 11. Kinetic parameters: Vm
0.052 0.004 mM/min, Km = 0.19 0.04 mM, Vm(app)
0.028 0.002 mM/min, Km(app) = 0.14 0.03 mM.
=
=
(F) Kinetics of the inhibition of D-glucose uptake in erythro-
cyte membranes by 12. Kinetic parameters: Vm = 0.068
0.005 mM/min, Km = 0.28 0.05 mM, Vm(app) = 0.030
0.005 mM/min, Km(app) = 0.45 0.16 mM.
Figure 4 is supported by Figure S4. Each value is mean
SEM of duplicate values from a representative experi-
ment.
production with IC50 values of 3 and 1.5 mM,
respectively (Figures 3G and 3H). Similar effect
was observed in PC3 cells (Figure S3J).
Inhibition of facilitative Glucose
Transport
The similarity of the ATP suppression patterns of
11 and 12 to that observed for cytochalasin B (5)
in two representative cell lines, as well as the
absence of major glycolytic metabolites
observed upon treatment of cells with 11 or 12
strongly suggested that the two compounds eli-
cited their activities by blocking transport of
of CHO-K1 cells to 11 and 12 and its history of use as an in vitro glucose into the cells. Expression of the GLUT1 facilitative
model for investigation of facilitative glucose transport (Faik glucose transporter in A549 and CHO-K1 cell lines was verified
et al., 1989), we selected this cell line for further study.
by western blot (Figure S3L). Direct drug action on mammalian
We next examined the effects of 11 and 12 on cell growth and glucose transporters was measured by monitoring rapid uptake
viability. As expected for inhibitors of energy metabolism, both of 3H-labeled 3-O-methylglucose, a radiotracer that is efficiently
compounds inhibited growth of A549 and CHO-K1 cell lines taken up via glucose transporters but not metabolized further,
(Figures 3E and 3F). The potency of 11 and 12 in growth inhibitory allowing the assessment of initial rate of sugar uptake. Under
assays and cell-line sensitivity correlated well with the initially such conditions, 11 inhibited uptake of this labeled glucose
observed effects of the two compounds on ATP synthesis. analog by 50% at concentration of 2 mM (Figure 4A). The uptake
Furthermore, the effects of 11 and 12 on cell viability appeared of 3H-labeled 2-deoxy-D-glucose, a glucose analog that can be
to be cytostatic at a range of concentrations tested. We also phosphorylated by hexokinase, was blocked by 11 with similar
evaluated the effect of the two compounds on cell-cycle regula- potency (Figure 4B). Similar effects were observed in highly
tion. Treatment of A549 cells with 11 or 12 for 24 hr resulted in sensitive U373 glioma cell line (Figure S4A).
a notable increase in G1 subpopulation to 71% and 73%,
We next evaluated the effects of 11 and 12 on uptake of
respectively, compared with 62% of the G1 peak in the negative D-glucose in purified sealed erythrocyte membranes (Figures
control (Figure S3K). This result was again in agreement with the 4C and 4D), which exclusively express GLUT1 (Carruthers
impairment of energy metabolism in cells exposed to 11 or 12.
et al., 2009). We employed cytochalasin B (5), a noncompetitive
The level of excreted lactate provides a validated measure of inhibitor of glucose uptake (Deves and Krupka, 1978), and gen-
the efficiency of glycolytic flux of mammalian cells. Indeed, effi- istein, a competitive inhibitor of glucose uptake (Vera et al.,
cient inhibition of lactate production was observed in CHO-K1 1996), as positive controls. Each elicited the expected poten-
cells treated for 4 hr (Eichner et al., 2010) with either 2-deoxy- cies and kinetic profiles (Figure S4B and S4C). The effects of
D-glucose or cytochalasin B compared with negative controls 11 and 12 on facilitative glucose transport by the erythrocyte
using either DMSO or dihydrocytochalasin B (Figure S3I). Sub- membranes were consistent with a noncompetitive mode of
jecting CHO-K1 cells to either 11 or 12 under the same condi- inhibition with Ki values of 1.2 and 0.8 mM, respectively (Figures
tions resulted in dose-dependent inhibition of cellular lactate 4E and 4F).
226 Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
Cytochalasin B (5) is known to depolymerize F-actin (Bonder pathway-based screen was designed to target multiple proteins
and Mooseker, 1986) at concentrations similar to those used involved in cellular glucose consumption and metabolism.
to inhibit glucose transport (Figure S4E). While this compound Having initially validated the synthetic suppression of ATP
has been used extensively for biochemical characterization of synthesis by a combination of small-molecule inhibitors of
glucose transport, the existence of alternative cellular targets OXPHOS and glycolysis or glucose transport, we screened
of cytochalasin B substantially limits its applications of this agent a newly constructed diverse chemical library for small molecules
in cell-based and in vivo studies of glucose transport. We exam- that would act in concert with antimycin A to suppress cellular
ined depolymerization of the actin cytoskeleton in A549 cells and ATP production. This effort resulted in identification of two
found that neither 11 nor 12 altered F-actin distribution even at chemotypes 11 and 12 that potently inhibited ATP synthesis in
the highest concentrations tested (Figures S4F and S4G). While A549 cells with chemically suppressed mitochondrial function.
other off-target effects of 11 and 12 cannot be ruled out at this As expected, inhibition of lactate production by the two newly
point, the absence of cytoskeletal effects of the two compounds identified compounds was highly indicative of substantial reduc-
uniquely distinguishes them from cytochalasin-based inhibitors tion of glycolytic flux. The pattern of ATP synthesis suppression
of facilitative glucose transport.
and differential cellular sensitivity strongly suggested that 11 and
12 blocked glucose transport in mammalian cell lines. Detailed
examination of this effect revealed that both compounds in-
hibited uptake of 3-O-methylglucose and 2-deoxy-D-glucose
DISCUSSION
Glucose is an essential metabolic energy source for all living in CHO-K1 cells. Furthermore, 11 and 12 suppressed uptake
organisms and a structural precursor for cellular biosynthesis of labeled D-glucose in sealed erythrocyte membranes. The
of proteins, lipids, and nucleic acids. A common feature of kinetic studies suggested a noncompetitive inhibition of facilita-
cancer cells is a dramatic increase in glucose uptake associated tive glucose transport through erythrocyte membranes with low
with the Warburg effect, providing a basis tumor imaging based micromolar inhibitory constants. Such potent inhibition of facili-
on the accumulation of the PET tracer 18fluorodeoxyglucose by tative glucose transport was fully consistent with the ability of
many tumors (Gambhir, 2004). Most vertebrate cells must trans- 11 and 12 to synthetically suppress ATP production, reduce
port glucose across the cell membrane, which is mediated by glycolytic flux and inhibit mammalian cell growth at similarly
active and passive glucose transporters. While active glucose low concentrations. It is noteworthy that the two newly identified
transport is sodium-dependent and relies on an electrochemical chemotypes do not share significant structural similarity with
gradient, passive glucose transport is controlled by facilitative known inhibitors of cellular glucose transport and can be effi-
glucose transporters (GLUTs), which respond to the gradient of ciently accessed by short, efficient and fully diastereoselective
glucose concentration across the cell membrane (Manolescu synthetic sequences. This new class of glucose transport inhib-
et al., 2007). Among 13 members of the mammalian GLUT family, itors provides a foundation for subsequent development of
individual isoforms are characterized by their unique tissue highly potent compounds that can modulate cellular glucose
distributions and distinct affinities for various substrates that uptake in a selective manner, as might be necessary to target
they transport. In addition to variable capacity to transport of aerobic glycolysis in cancer.
D-pentoses and D-hexoses, GLUTs have also been shown to
transport the oxidized form of vitamin C, dehydroascorbic acid SIGNIFICANCE
(Vera et al., 1993).
Overexpression of glucose transporters and other glycolytic The Warburg effect, in which glycolysis remains active even
enzymes is one of the key alterations associated with the high under aerobic conditions, is considered a key driver for
glycolytic rate of malignant cells. GLUT1 is particularly highly ex- cancer cell proliferation, malignancy, metastasis, and thera-
pressed in several cancers including breast, brain, renal, colon, peutic resistance. Currently, there is a limited arsenal of
ovarian, and cervical carcinoma (Yamamoto et al., 1990; Nish- small molecules that can target aerobic glycolysis with
ioka et al., 1992; Brown and Wahl, 1993; Haber, et al., 1998; Can- high potency and specificity. We developed a new approach
tuaria et al., 2001; Rudlowski et al., 2003). Several studies have for rapid discovery and subsequent target identification of
demonstrated a link between GLUT1 expression and chemore- glycolytic inhibitors from structurally diverse chemical
sistance, tumor aggressiveness, and poor survival (Hatanaka, libraries. Such compounds are identified by their ability to
1974; Evans et al., 2008). Therefore, inhibition of GLUT1 function synergistically suppress ATP production in cancer cells in
may be a promising therapeutic strategy for sensitization of the presence of a mitochondrial inhibitor. We initially vali-
highly glycolytic cancer cells to chemotherapeutic agents and dated this synthetic effect by evaluating a number of known
for direct targeting of cancer cells. However, progress has re- inhibitors of glycolysis and oxidative phosphorylation. We
mained limited by the currently available small-molecule inhibi- then synthesized a chemical library and analyzed effects of
tors, which lack potency and/or exhibit substantial off-target each library member on ATP production in the presence of
effects. Thus, there is a significant need for the development of a mitochondrial inhibitor. This effort identified two new
new inhibitors of glucose transport that can potently modulate compounds that potently suppressed ATP synthesis by in-
this process in vitro and in vivo.
hibiting GLUT family of facilitative transmembrane trans-
We have described a practical and general strategy for porters. Such compounds would be useful not only to study
discovery of new inhibitors of glycolysis and glucose transport, the system-level organization of energy metabolism but
which can be rapidly identified by their ability to deplete cellular could also facilitate development of drugs targeting upregu-
ATP in cells with suppressed mitochondrial function. This lation of aerobic glycolysis in cancer.
Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved 227

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
EXPERIMENTAL PROCEDURES
Cellular uptake of Radiolabeled Glucose Analogs
CHO-K1 cells (2 3 106) were washed in PBS and incubated for 15 min at
37ꢀC in glucose uptake buffer (8.1 mM Na2HPO4, 1.4 mM KH2PO4, 0.5 mM
MgCl2, 2.6 mM KCl, 136 mM NaCl, and 0.9 mM CaCl2 [pH 7.4]). Twenty-
five microliters of a 20% perchloric acid/ 8% sucrose solution was added
to the bottom of a 0.5 ml microfuge tube. 1-Bromododecane (200 ml) was
overlaid above the perchloric acid/sucrose solution. Fifty microliters of triti-
ated sugar (25 mCi/ml) was overlaid above the 1-bromododecane layer. After
the 37ꢀC incubation, 100 ml of cells were added to the sugar layer and pulsed
with 3-O-methylglucose and 2-deoxy-D-glucose for 30 and 120 s, respec-
tively. The reaction was stopped by centrifugation at 14,000 rpm for
10 min. Tubes were then snap-frozen in an ethanol/dry ice bath. The tips of
the tubes were cut just above the perchloric acid/sucrose/bromododecane
interface and transferred into scintillation vials containing 100 ml of 1% Triton
X-100. Optiphase Supermix (PerkinElmer) was added to each vial and
counted using an LS6500 scintillation counter (Beckman Coulter). Each
sample was measured in triplicate. Nonspecific uptake was assessed by
treatment of cells with 5–50 mM cytochalasin B and was excluded from all
samples. All values are presented as percentage of vehicle treated sample.
Each value is mean SEM of at least duplicate values from a representative
experiment.
Chemical Library Synthesis
The following procedure represents the synthesis of the first set of 96
compounds. Eight 1.5 ml polypropylene centrifuge tubes were charged with
CHCl3 (0.8 ml each), methyl acetoacetate (0.5 mmol each), and treated individ-
ually with eight primary amines (II, 0.5 mmol) at 70ꢀC in a sand bath. The vinyl-
ogous amides III were purified by preparative TLC (ethyl acetate: hexanes =
1:5 to 1:1), dissolved in CHCl3 (0.8 ml), and treated with maleic anhydride
(0.3–0.5 mmol) at 20ꢀC. Upon completion of each reaction, the mixtures
were diluted with CHCl3 (1.6 ml) and THF (4.8 ml), followed by treatment
with N-hydroxysuccinimide (0.38–0.63 mmol) and PS-carbodiimide resin
(1.1 mmol/g, 345–573 mg). The reaction mixtures were stirred for 2–4 hr at
20ꢀC, filtered, concentrated, and purified by preparative TLC (ethyl acetate:
hexanes = 2:1) to give the eight corresponding succinimide esters IV, which
were diluted with CH2Cl2 to final concentrations of 0.1 M. Aliquots (25 ml) of
each resulting stock solutions were transferred into a polypropylene 96-well
PCR plate and treated with 12 amines V (4 mmol per well) and CH2Cl2 (30 ml
per well) using the plate maps shown in Supplemental Experimental Proce-
dures. Upon completion, the reaction mixtures were transferred onto prepar-
ative TLC plates using a multichannel pipettor with adjustable gaps. The plates
were developed in ethyl acetate/hexanes (3:2). The products VI were detected
under UV light and removed from TLC plates as circular silica gel pellets. Each
compound was eluted from silica gel with ethyl acetate (0.6 ml). Following
removal of the solvent in vacuo, 12 randomly selected compounds were dis-
solved in CD3OD (0.5 ml) and analyzed by 1H NMR. The amount of material
in each sample was determined by integration using residual MeOH as a pre-
calibrated internal standard. This protocol was used next to prepare all the re-
maining library members.
Isolation of Pink Ghosts from Red Blood Cells
Outdated red cells were obtained from blood bank. Cells were washed once in
KCl buffer (10 mM Tris-HCl, 150 mM KCl, 5 mM MgCl2, 4 mM EGTA [pH 7.4]),
and membranes were prepared by dispersing 1 volume of cells in 24 ml ice-
cold lysis buffer (10 mM Tris-HCl, 1 mM EGTA [pH 7.4]). After 10 min of incu-
bation on ice, membranes were collected at 22,000 3 g/10 min. Red cell
membranes were resealed by gently resuspending the pellet in 20 volumes
of KCl medium followed by a 1 hr incubation at 37ꢀC. The resulting ghosts
were collected by centrifugation at 22,000 3 g/20 min, resuspended in KCl
medium, and placed on ice until use.
Cell Culture
A549, CHO-K1, PC3, U373, and U87 cell lines were purchased from ATCC and
were maintained in either F12-K or DMEM medium supplemented with 10%
FBS and 1% penicillin-streptomycin-glutamine solution.
D-Glucose Uptake in Erythrocyte Ghosts
All samples were done in duplicate and at ice temperature. Ghosts were
diluted in KCl medium to 0.4 mg/ml, treated with small molecules and
exposed to D-glucose (100 mM + 10 mCi/ml). Initial rates of glucose uptake
were measured after 30 s of incubation, where uptake was linear with time
and did not exceed 17% of equilibrium uptake assessed by exposing
membranes to sugar for 1 hr at room temperature. Then, 10 volumes (rela-
tive to assay volume) of stopper solution (KCl medium containing 10 mM
cytochalasin B and 100 mM phloretin) were added to the assay mixture fol-
lowed by centrifugation at 14,000 g/1 min and a second wash. Membranes
were dissolved in 0.5 ml 0.1 N NaOH, mixed with 5 ml Optophase Supermix
scintillation liquid and counted using an LS-6000IC scintillation counter
(Beckman Coulter). Nonspecific uptake assessed by treatment of
membranes with 10 mM cytochalasin B did not exceed 6% of uptake in
vehicle treated membranes and was excluded from all samples. Uptake
was calculated as previously described (Helgerson et al., 1989). Initial
velocity of D-glucose uptake in ghosts treated with vehicle control was
0.028 mM/min.
Quantification of Cellular ATP Levels
Cells were seeded in 96-well plates at a density of 2500 cells/well. The next
day, cells were treated with either medium or medium containing antymicin
A (final concentration = 10 nM) followed immediately by addition of graded
concentrations of small molecules and incubated for 30 min. The ATPlite assay
kit (PerkinElmer) was used to measure ATP levels. All assays were performed
using two replicate wells for each small-molecule concentration tested.
High-Throughput Library Screen
The effects of small molecules on ATP levels in A549 cells in the presence of
10 nM antimycin A were measured as described above. Each compound
was tested at a single concentration of 20 mM. Concentration of DMSO did
not exceed 1% of the total assay volume.
Cellular Growth Inhibition
Cells were seeded in 6-well plates at a density of 100,000 cells/well and were
allowed to attach to plate surface. Then cells were washed with PBS and
treated with fresh medium containing graded concentrations of small mole-
cules. After 48 hr of incubation, cells were trypsinized, stained with trypan
blue, and counted using Countess automated cell counter from Invitrogen.
All assays were performed using two replicate wells for each small molecule
concentration test.
Kinetics of D-Glucose Uptake in Erythrocyte Ghosts
Sealed membranes were treated with small molecules and exposed to vari-
able concentrations of D-glucose (0.0625 O 1000 mM plus 10 mCi/ml). Initial
rates of glucose uptake were measured as described above over intervals of
30–120 s where uptake did not exceed 18% of equilibrium uptake. Michae-
lis-Menten parameters were estimated by nonlinear curve fitting (Prism soft-
ware). Kinetic constants were calculated using a single concentration of
inhibitor.
Lactate Production
CHO-K1 or PC3 cells were seeded in 6-well plates at a concentration of
400,000 cells/well and incubated overnight. Old medium was removed; cells
were washed with PBS, and then treated with 1 ml of medium containing
different concentrations of compounds. After 4 hr of incubation, an aliquot of
the cell culture medium was removed from each well and kept at À80ꢀC until
processed. The amount of lactate excreted into the cell culture medium was
measured using the Lactate assay kit from Biovision. All samples were done
in duplicate.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
four figures, and the analytical characterization of chemical library and can
be found with this article onlineonline at doi:10.1016/j.chembiol.2010.12.015.
228 Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
ACKNOWLEDGMENTS
Haber, R.S., Rathan, A., Weiser, K.R., Pritsker, A., Itzkowitz, S.H., Bodian, C.,
Slater, G., Weiss, A., and Burstein, D.E. (1998). GLUT1 glucose transporter
expression in colorectal carcinomas: a marker for poor prognosis. Cancer
83, 34–40.
This work was funded by NIH P50 GM086145 and the Chicago Biomedical
Consortium with support from the Searle Funds at the Chicago Community
Trust. We thank Gregory Driessens for assistance with glucose uptake exper-
iments, and Jamie Cabrera-Pardo for preparation of compounds 11 and 12 on
preparative scale. We also thank Paul Schumacker and Theodore Steck for
helpful discussions.
Hartwell, L.H., Szankasi, P., Roberts, C.J., Murray, A.W., and Friend, S.H.
(1997). Integrating genetic approaches into the discovery of anticancer drugs.
Science 278, 1064–1068.
Hatanaka, M. (1974). Transport of sugars in tumor cell membranes. Biochim.
Biophys. Acta 355, 77–104.
Received: August 27, 2010
Revised: November 26, 2010
Accepted: December 1, 2010
Published: February 24, 2011
Helgerson, A.L., Hebert, D.N., Naderi, S., and Carruthers, A. (1989).
Characterization of two independent modes of action of ATP on human eryth-
rocyte sugar transport. Biochemistry 28, 6410–6417.
Lee, H., Suzuki, M., Cui, J., and Kozmin, S.A. (2010). Synthesis of an Azide-
Tagged Library of 2,3-dihydro-4-quinolones. J. Org. Chem. 75, 1756–1759.
REFERENCES
Lehar, J., Stockwell, B.R., Giaever, G., and Nislow, C. (2008). Combination
Bonder, E.M., and Mooseker, M.S. (1986). Cytochalasin B slows but does not
prevent monomer addition at the barbed end of the actin filament. J. Cell Biol.
102, 282–288.
chemical genetics. Nat. Chem. Biol. 4, 674–681.
Liu, H., Hu, Y.P., Savarai, N., Priebe, W., and Lampadis, T. (2001).
Hypersensitization of tumor cells to glycolytic inhibitors. Biochemistry 40,
5542–5547.
Bredel, M., and Jacoby, E. (2004). Chemogenomics: An emerging strategy for
rapid target and drug discovery. Nat. Rev. Genet. 5, 262–275.
Manolescu, A.R., Witkowska, K., Kinnaird, A., Cessford, T., and Cheeseman,
C. (2007). Facilitated hexose transporters: new perspectives on form and func-
tion. Physiology (Bethesda) 22, 234–240.
Brown, J. (1962). Effects of 2-deoxyglucose on carbohydrate metabolism:
review of the literature and studies in the rat. Metabolism 11, 1098–1112.
Brown, R.S., and Wahl, R.L. (1993). Overexpression of GLUT-1 glucose trans-
porter in human breast cancer. An immunohistochemical study. Cancer 72,
2979–2985.
Nishioka, T., Oda, Y., Seino, Y., Yamamoto, T., Inagaki, N., Yano, H., Imura, H.,
Shigemoto, R., and Kikuchi, H. (1992). Distribution of glucose transporters in
human brain tumors. Cancer Res. 52, 3972–3979.
Cantuaria, G., Fagotti, A., Ferrandina, G., Magalhaes, A., Nadji, M., Angioli, R.,
Penalver, M., Mancuso, S., and Scambia, G. (2001). GLUT-1 expression in
ovarian carcinoma: association with survival and response to chemotherapy.
Cancer 92, 1144–1150.
Ohnishi, T., and Trumpower, B.L. (1980). Differential effects of antimycin on
ubisemiquinone bound in different environments in isolated succinate-
cytochrome c reductase complex. J. Biol. Chem. 255, 3278–3284.
Omura, S., Fujimoto, T., Otoguro, K., Matsuzaki, K., Moriguchi, R., Tanaka, H.,
and Sasaki, Y. (1991). Lactacystin, a novel microbial metabolite, induces neu-
ritogenesis of neuroblastoma cells. J. Antibiot. 44, 113–116.
Carruthers, A., DeZutter, J., Gaunguly, A., and Devaskar, S.U. (2009). Will the
original glucose transporter isotype stand up!. Am. J. Physiol. Endocrinol.
Metab. 297, E836–E848.
Pelicano, H., Martin, D.S., Xu, R., and Huang, P. (2006). Glycolysis inhibition for
Cave´ , C., Gassama, A., Mahuteau, J., d’Angelo, J., and Riche, C. (1997).
Condensation of chiral imines and chiral b-enaminoesters with maleic and cit-
raconic anhydrides. Tetrahedron Lett. 38, 4773–4776.
anticancer treatment. Oncogene 25, 4633–4646.
Racker, E. (1974). History of the Pasteur effect and its pathobiology. Mol. Cell.
Biochem. 5, 17–23.
Cui, J., Matsumoto, K., Wang, C.Y., Peter, M.E., and Kozmin, S.A. (2010).
Synthesis of high-purity chemical library reveals a potent inducer of oxidative
stress. ChemBioChem 11, 1224–1227.
Rudlowski, C., Becker, A.J., Schroder, W., Rath, W., Buttner, R., and Moser,
M. (2003). GLUT1 messenger RNA and protein induction relates to the malig-
nant transformation of cervical cancer. Am. J. Clin. Pathol. 120, 691–698.
DeBerardinis, R.J., Lum, J.J., Hatzivassiliou, G., and Thompson, C.B. (2008).
The biology of cancer: metabolic reprogramming fuels cell growth and prolif-
eration. Cell Metab. 7, 11–20.
Sabri, M.I., and Ochs, S. (1971). Inhibition of glyceraldehydes-3-phoshate
dehydrogenase in mammalian nerve by iodoacetic acid. J. Neurochem. 18,
1509–1514.
Deves, R., and Krupka, R.M. (1978). Cytochalasin B and the kinetics of inhibi-
tion of biological transport. A case of asymmetric binding to the glucose
carrier. Biochim. Biophys. Acta 510, 339–348.
`
Segre, D., Deluna, A., Church, G.M., and Kishony, R. (2005). Modular epistasis
in yeast metabolism. Nat. Genet. 37, 77–83.
Stockwell, B.R. (2000). Chemical genetics: ligand-based discovery of gene
Eichner, L.J., Perry, M.C., Dufour, C.R., Bertos, N., Park, M., St-Pierre, J., and
Giguere, V. (2010). miR-378* mediates metabolic shiftin breast cancer cells via
the PGC-1b/ERRg transcriptional pathway. Cell Metab. 12, 352–361.
function. Nat. Rev. Genet. 1, 116–125.
Tan, D.S. (2005). Diversity-oriented synthesis: exploring the intersections
between chemistry and biology. Nat. Chem. Biol. 1, 74–84.
Evans, A., Bates, V., Troy, H., Hewitt, S., Holbeck, S., Chung, Y., Phillips, R.,
Stubbs, M., Griffiths, J., and Airley, R. (2008). GLUT-1 as a therapeutic target:
increased chemoresistance and HIF-1-independent link with cell turnover is
revealed through COMPARE analysis and metabolomic studies. Cancer
Chemother. Pharmacol. 61, 377–393.
Tennant, D.A., Dura´ n, R.V., and Gottlieb, E. (2010). Targeting metabolic trans-
formation for cancer therapy. Nat. Rev. Cancer 10, 267–277.
´
Tong, A.H., Evangelista, M., Parsons, A.B., Xu, H., Bader, G.D., Page, N.,
Robinson, M., Raghibizadeh, S., Hogue, C.W., Bussey, H., et al. (2001).
Systematic genetic analysis with ordered arrays of yeast deletion mutants.
Science 294, 2364–2368.
Faik, P., Morgan, M., Naftalin, R.J., and Rist, R.J. (1989). Transport and accu-
mulation of 2-deoxy-D-glucose in wild-type and hexokinase-deficient cultured
Chinese-hamster ovary (CHO) cells. Biochem. J. 260, 153–155.
Tong, A.H., Lesage, G., Bader, G.D., Ding, H., Xu, H., Xin, X., Young, J., Berriz,
G.F., Brost, R.L., Chang, M., et al. (2004). Global mapping of the yeast genetic
interaction network. Science 303, 808–813.
Gambhir, S.S. (2004). Molecular imaging of cancer with positron emission
tomography. Nat. Rev. Cancer 4, 891–899.
Gatenby, R.A., and Gillies, R.J. (2004). Why do cancers have high aerobic
Ulanovskaya, O., Janjic, J., Matsumoto, K., Schumacker, P.T., Kron, S.J., and
Kozmin, S.A. (2008). Synthesis enables identification of the cellular target of
leucascandrolide A and neopeltolide. Nat. Chem. Biol. 4, 418–424.
glycolysis? Nat. Rev. Cancer 4, 891–899.
Gohil, V.M., Sheth, S.A., Nilsson, R., Wojtovich, A.P., Lee, J.H., Perocchi, F.,
Chen, W., Clish, C.B., Ayata, C., Brookes, P.S., et al. (2010). Nutrient-sensi-
tized screening for drugs that shift energy metabolism from mitochondrial
respiration to glycolysis. Nat. Biotechnol. 28, 249–255.
Vander Heiden, M.G., Cantley, L.C., and Thompson, C.B. (2009).
Understanding the Warburg effect: the metabolic requirements of cell prolifer-
ation. Science 324, 1029–1033.
Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved 229

Chemistry & Biology
Small-Molecule Inhibition of Glucose Transport
Vera, J.C., Ricas, C.I., Fischbarg, J., and Golde, D.W. (1993). Mammalian facil-
itative hexose transporters mediate the transport of dehydroascorbic acid.
Nature 364, 79–82.
Warburg, O. (1956). On the origin of cancer cells. Science 123, 309–314.
Yamamoto, T., Seino, Y., Fukumoto, H., Koh, G., Yano, H., Inagaki, N.,
Yamada, Y., Inoue, K., Manabe, T., and Imura, H. (1990). Over-expression of
facilitative glucose transporter genes in human cancer. Biochem. Biophys.
Res. Commun. 170, 223–230.
Vera, J.C., Reyes, A.M., Carcamo, J.G., Velasques, F.V., Rivas, C.I., Zhang,
R.H., Strobel, P., Iribarren, R., Scher, H., Slebe, J.C., and Golde, D.W.
(1996). Genistein is a natural inhibitor of hexose and dehydroascorbic acid
transport through the glucose transporter, GLUT1. J. Biol. Chem. 271, 8719–
8724.
Yost, C.S. (2006). A new look at the respiratory stimulant doxapram. CNS Drug
Rev. 12, 236–249.
230 Chemistry & Biology 18, 222–230, February 25, 2011 ª2011 Elsevier Ltd All rights reserved