Oncogene

https://doi.org/10.1038/s41388-019-0751-4

ARTICLE

Mitochondrial GPT2 plays a pivotal role in metabolic adaptation to

the perturbation of mitochondrial glutamine metabolism
1,2,3
Minjoong Kim ●
Jihye Gwak1,2,3 Sunsook Hwang1,2,3 Seungyeon Yang1,2,3 Seung Min Jeong1,2,3
● ● ●

Received: 19 October 2018 / Revised: 11 January 2019 / Accepted: 31 January 2019

© Springer Nature Limited 2019

Abstract
Cancer cells exhibit metabolic dependence on mitochondrial glutamine metabolism that provides them with the substrates
required for rapid proliferation. Despite the extensive efforts to target this glutamine addiction for therapeutic purposes, the
adaptive metabolic responses and the mechanisms whereby cells maintain their unlimited growth remain areas of active
investigation. Here we report that mitochondrial glutamate-pyruvate transaminase 2 (GPT2) contributes to cell survival and
growth by sustaining the tricarboxylic acid (TCA) cycle anaplerosis after the inhibition of glutaminase (GLS), the first
enzyme for mitochondrial glutamine metabolism. We found that elevated reactive oxygen species upon GLS inhibition
1234567890();,:
1234567890();,:

induce GPT2 expression via activating transcription factor 4. Moreover, inhibition of GPT2 synergized with suppression of
GLS activity to induce a pronounced reduction in proliferation and an increase in cell death of cancer cells. Our data uncover
GPT2 as an important component of the adaptive metabolic response for glutamine deprivation and indicate that targeting
this pathway in combination with GLS inhibition may be an effective therapeutic approach for cancer treatment.

Introduction and lipids for their unlimited proliferation [3]. However,

The dysregulation of cell metabolism is a defining factor of
many cancers [1]. For example, cancer cells undergo gly-
colysis even in the presence of ample oxygen and this
preferential use of aerobic glycolysis has emerged as a
hallmark of cancer [2]. Cancer cells also use precursors
derived from the tricarboxylic acid (TCA) cycle to produce
biomass building blocks including amino acids, nucleotides,
Supplementary information The online version of this article (https://
doi.org/10.1038/s41388-019-0751-4) contains supplementary
material, which is available to authorized users.
* Seung Min Jeong
smjeong@catholic.ac.kr
1
Department of Biochemistry, College of Medicine, The Catholic
University of Korea, 222, Banpo-daero, Seocho-gu, Seoul 06591,
Republic of Korea
2
Department of Biomedicine & Health Sciences, College of
Medicine, The Catholic University of Korea, 222, Banpo-daero,
Seocho-gu, Seoul 06591, Republic of Korea
3 activity [3, 5]. As GLS is the most proximal enzyme in
Institute for Aging and Metabolic Diseases, College of Medicine,
The Catholic University of Korea, 222, Banpo-daero, Seocho-gu,
Seoul 06591, Republic of Korea
continuous use of TCA cycle intermediates would lead to
the loss of mitochondrial functions and limit the pro-
liferative capacity of cells. Accruing evidence suggests that
glutamine (Gln) provides a carbon source to refill the TCA
cycle, which is critical for the maintenance of mitochondrial
integrity, cell growth and survival [3, 4].
In mitochondria, Gln is converted to glutamate via glu-
taminase (GLS) and further metabolized to the TCA cycle
intermediate α-ketoglutarate (αKG) via glutamate dehy-
drogenase 1 (GLUD1) or transaminases such as glutamate-
oxaloacetate transaminase 2 (GOT2) or glutamate-pyruvate
transaminase 2 (GPT2). The incorporation of αKG into the
TCA cycle through Gln metabolism is the major anaplerotic
step in proliferating cells [5]. Recent studies have demon-
strated that mitochondrial Gln metabolism is also an
important regulator of cellular redox state and thus its
defects lead to the generation of reactive oxygen species
(ROS) [6–8].
Given the importance of mitochondrial Gln metabolism
to support anabolic processes that fuel cell proliferation, it is
not surprising that many cancer cells are addicted to Gln
and exhibit an increased Gln anaplerosis and high GLS

mitochondrial Gln metabolism, drugs targeting GLS are
being developed and tested for cancer treatment [9, 10].
However, these drugs suffer from poor bioavailability and
cytotoxicity at higher doses [9]. Moreover, many cancer
cells retain the capacity for metabolic adaptation to sustain
cell growth and survival periods of impaired Gln metabo-
lism by inducing alternative metabolic pathways. For
example, Gln withdrawal activates serine biosynthesis
pathway, which is important for cell survival by facilitating
glutathione (GSH) production and nucleic acid synthesis
[11, 12]. It has been also shown that asparagine synthetase
(ASNS) suppresses Gln deprivation-induced cell death by
maintaining intracellular asparagine (Asn) levels [13–15].
However, the adaptive molecular events that maintain Gln
anaplerosis upon GLS inhibition are still unclear.
In this study, we describe a role of mitochondrial GPT2
in the regulation of cell growth and survival after the
inhibition of mitochondrial Gln metabolism, and provide
evidence that targeting this compensatory response may be
an important therapeutic strategy for multiple cancer cells.
Results
Mitochondrial GPT2 is induced by Gln deprivation
Gln, the most abundant amino acid in the body, plays an
essential role for cell proliferation [4]. To investigate how
cells might cope with impaired Gln metabolism, we first
examined the expression of metabolic enzymes which are
involved in mitochondrial Gln metabolism after Gln
deprivation. We found that GPT2 messenger RNA (mRNA)
and protein levels were highly induced upon Gln with-
drawal, whereas the expressions of other mitochondrial Gln
metabolism-related enzymes were not changed (Fig. 1a–c
and Supplementary Fig. 1A, B). Notably, we observed no
obvious change in GPT2 expression under glucose-deprived
conditions (Fig. 1b, c and Supplementary Fig. 1A), indi-
cating that the induction of GPT2 is a specific response
under Gln-starved conditions and not an indirect con-
sequence of the metabolic crisis. In mitochondria, Gln is
catabolized into αKG to replenish the TCA cycle. When
mitochondrial Gln metabolism was impaired by adding bis-
2-(5-phenylacetoamido-1,2,4-thiadiazol-2-yl)ethylsulfide
(BPTES), an inhibitor of GLS [9], we found that GPT2 was
highly induced (Fig. 1d, e and Supplementary Fig. 1C). We
also observed a similar induction of GPT2 with another
GLS inhibitor, CB-839 (Supplementary Fig. 1D) [10].
Moreover, GLS knockdown using short hairpin RNAs
(shRNAs) also markedly increased GPT2 expression (Fig.
1f, g). These data hint that GPT2 may have an important,
previously undetermined role when mitochondrial Gln
metabolism is suppressed.
M. Kim et al.
Mitochondrial GPT2 is required for cell growth and
survival upon GLS inhibition
Given the significant induction of GPT2 upon GLS inhi-
bition, we speculated that GPT2 may be required for cell
growth and/or survival when mitochondrial Gln metabolism
is impaired. To test this concept, we inhibited GPT2 by
treating cells with aminooxyacetate (AOA), a transaminase
inhibitor [16], and examined whether this would affect cell
growth after BPTES treatment. First, we found that BPTES
or AOA treatment alone had a mild growth-suppressive
effect (Fig. 2a, b and Supplementary Fig. 2A, B).
Remarkably, when these two inhibitors are combined, this
effect was notably augmented, indicating that cells are
markedly more sensitive to GLS inhibition when GPT2
activity is suppressed (Fig. 2a, b and Supplementary Fig.
2A, B). When we measured cell viability and clonogenic
growth, similar synergistic effects were observed (Fig. 2c, d
and Supplementary Fig. 2C, D). Near-identical results were
also observed with CB-839 or with another GPT2 inhibitor,
cycloserine (Supplementary Fig. 2E, F) [17]. Consistent
with these results, GLS knockdown also significantly atte-
nuated AOA treated cell viability (Fig. 2e). As further
confirmation of the importance of GPT2 upon GLS inhi-
bition, we suppressed GPT2, GLUD1 and GOT2 expression
using shRNAs and assessed cell viability under GLS
inhibited conditions. In line with our hypothesis, GLUD1 or
GOT2 knockdown had no effect on cell viability (Supple-
mentary Fig. 2G, H). In contrast, GPT2 knockdown
robustly diminished cell viability, when mitochondrial Gln
metabolism is repressed by BPTES treatment (Fig. 2f).
Mitochondrial Gln metabolism refuels the TCA cycle,
which is essential for cell survival [4]. Thus, we next
explored the role of GPT2 on cell survival. We found that
AOA treatment or knockdown of GPT2 significantly
induced cell death, when GLS activity is inhibited by
BPTES treatment (Fig. 2g, h). Similarly, we found that
GPT2 impaired cells exhibited significantly increased cell
death in Gln-free media compared to control cells (Sup-
plementary Fig. 2I, J). Together, these data demonstrate that
GPT2 contributes to cell growth and survival upon GLS
inhibition.
ATF4 regulates the transcription of GPT2 upon GLS
inhibition
We next examined the mechanisms underlying increased
GPT2 expression after GLS inhibition. Given the essential
role of mitochondrial Gln metabolism in maintaining the
cellular redox homeostasis [6], we speculated that elevated
ROS production by GLS inhibition may affect this adaptive
metabolic response. Indeed, cellular ROS production was
significantly induced by Gln deprivation or GLS inhibition
Mitochondrial GPT2 plays a pivotal role in metabolic adaptation to the perturbation of mitochondrial. . .

A B
***

Relative mRNA expression levels

5 Control 6
***

Relative GPT2 mRNA levels

-Gln 12h
-Gln 24h
4

4
3

2
2
1

0 0
GLS1 GLUD1 GOT2 GPT2
Glc + + -
Gln + - +
C D E
3
**

Relative GPT2 mRNA levels GPT2

GPT2 2

β-actin β-actin

Glc + + - 1
BPTES - +

Gln + - +

0
BPTES - +

F G
*
Relative mRNA expression levels

3 shGFP
shGLS1 #1 ***
shGLS1 #2
GLS1
2
GPT2

β-actin
1
shRNA GFP GLS1 #1 GLS1 #2

0
GLS1 GPT2

Fig. 1 Glutamine starvation or GLS inhibition induces expression of
GPT2. a Relative mRNA levels of GLS1, GLUD1, GOT2, and GPT2
in HeLa cells incubated with or without Gln for 12 or 24 h (n = 3). β-
actin was used as an endogenous control for qRT-PCR. b, c Relative
GPT2 mRNA (b) and protein (c) levels in HeLa cells incubated with or
without Gln or Glc for 12 h. β-actin serves as a loading control.
d, e Relative GPT2 mRNA (d) and protein (e) levels in HeLa cells
treated with BPTES (10 μM) for 24 h. f, g Relative mRNA (f) and
protein (g) levels of GLS1 and GPT2 in HeLa cells expressing a
control shRNA or two independent shRNAs to GLS1. All error bars,
± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001

(Supplementary Fig. 3A). Thus, we first treated cells with significantly reduced GPT2 expression (Fig. 3a, b). NAC
the antioxidant N-acetylcysteine (NAC) in order to probe also inhibited GPT2 induction by Gln withdrawal (Fig. 3c
the model that suppressing ROS could block the induction and Supplementary Fig. 3B). To directly test whether GLS
of GPT2 upon GLS inhibition. Indeed, whereas GLS- inhibition increases GPT2 gene transcription through ROS,
impaired cells had higher levels of GPT2, NAC treatment we cloned the human GPT2 promoter region into a
 M. Kim et al.

A B C
15 DM SO 1.2
BPTES
AOA
***

Relative cell viability

Cell number x 104 BPTES + AOA
10 0.8

***
5 0.4

0 0.0
Day 0 Day 1 Day 2 Day 3 Day 4
BPTES - + - +
AOA - - + +
D E F
1.8 1.2 *** 1.2 ***
*** ***
***
Relative colony number

Relative cell viability

Relative cell viability

1.2 0.8 0.8

0.6 0.4 0.4

0.0 0.0 0.0

BPTES - + - + AOA - + - + - + BPTES - + - + - +
AOA - - + + shGFP shGLS1 #1 shGLS1 #2 shGFP shGPT2 #1 shGPT2 #2

GLS1 GPT2

β-actin β-actin

shRNA GFP GLS1 #1 GLS1 #2 shRNA GFP GPT2 #1 GPT2 #2

G H ***
40 30 shGFP
*** shGPT2 #1
shGPT2 #2
30
Cell death (%)
Cell death (%)

20

20

10
10

0 0
DMSO
BPTES - + - +
AOA - - + +
Fig. 2 GPT2 knockdown sensitizes cells to the inhibition of mito-
chondrial Gln metabolism. a Growth curves of HeLa cells treated with
or without BPTES and/or AOA (125 μM). Statistical differences were
determined using a two-way ANOVA. Data represent the mean ± SD
(n = 3). b HeLa cells were treated with or without BPTES and/or AOA
as indicated for 72 h. Phase microscopy was used to observe cell
proliferation. c Relative cell viability of HeLa cells treated with or
without BPTES (5 μM) and/or AOA for 72 h (n = 4). d Relative clo-
nogenic growth of HeLa cells treated with or without BPTES and/or
AOA. The number of colonies was counted (n = 3). e Relative cell
viability of HeLa cells expressing a control shRNA or two independent
BPTES
shRNAs to GLS1 treated with or without AOA (top) (n = 4). Western
blot confirmed the knockdown of GLS1 expression (bottom). f Rela-
tive cell viability of HeLa cells expressing a control shRNA or two
independent shRNAs to GPT2 treated with or without BPTES (top)
(n = 4). Western blot confirmed the knockdown of GPT2 expression
(bottom). g Survival of HeLa cells cultured with or without BPTES
and/or AOA for 72 h (n = 3). h Survival of HeLa cells expressing a
control shRNA or two independent shRNAs to GPT2 incubated with
or without BPTES for 72 h (n = 3). All error bars (except growth
curves), ± SEM. ***p < 0.001
Mitochondrial GPT2 plays a pivotal role in metabolic adaptation to the perturbation of mitochondrial. . .

luciferase reporter construct. We found that the increased A B

5

Relative GPT2 mRNA levels

GPT2 promoter activity upon GLS inhibition was restored **
4
by NAC treatment (Fig. 3d). Similar results were observed
under Gln-free conditions (Fig. 3e). Thus, these data 3
GPT2

demonstrate that increased ROS production by GLS inhi- β-actin

2
bition induces GPT2 expression.
BPTES - - + +
As GLS inhibition positively regulates the transcription 1
NAC - + - +
of GPT2, we next attempted to identify the transcription
0
factor that might be responsible for GPT2 induction. Recent
BPTES - - + +
studies reported that activating transcription factor 4 (ATF4) NAC - + - +
is a critical regulator of genes involved in amino acid
metabolism including GPT2 and an enhanced ROS gen- C D
15 4

Relative GPT2 mRNA levels

eration upregulates ATF4 expression [18–21]. Thus, we

Relative GPT2 Luc activity

** ***
reasoned that ATF4 may regulate GPT2 transcription upon 3
10
GLS inhibition. First, we observed that GLS inhibition
resulted in a marked increase in ATF4 expression at the 2

transcriptional level, which was completely reverse by NAC 5

treatment (Fig. 3f). Next, we found that mutating a motif 1

matching ATF4 consensus-binding site, located in a GPT2

0 0
promoter region (Supplementary Fig. 3C), significantly
Gln + + - - BPTES - - + +
abrogated BPTES or Gln deprivation-induced increase of NAC - + - + NAC - + - +
GPT2 promoter activity (Fig. 3g, h). Taken together, these
data demonstrate that an increased ROS production caused E F
10 ** 4

Relative ATF4 mRNA levels

Relative GPT2 Luc activity

by GLS inhibition induces GPT2 expression via ATF4. ***

8
3
Maintaining TCA cycle anaplerosis by GPT2 supports 6
cell survival during GLS inhibition 2
4

Given the important role of GPT2 in converting glutamate 1

2
into αKG to support mitochondrial carbon pool, we
0 0
speculated that GPT2-mediated TCA cycle anaplerosis
Gln + + - - BPTES - - + +
could support cell growth and survival upon GLS inhibi- NAC - + - + NAC - + - +
tion. To test this concept, we treated cells with membrane
permeable dimethyl-αKG (DMKG) and examined whether G H
5 10
***
this would affect cell growth and survival of GPT2 impaired
Relative GPT2 Luc activity

Relative GPT2 Luc activity

***
4 8
cells upon GLS inhibition. Importantly, DMKG supplement
almost completely restored decreased viability and increase 3 6
cell death in GLS inhibited or Gln-deprived cells upon
2 4
AOA treatment (Fig. 4a, b and Supplementary Fig. 4A–C).
Consistent with these results, DMKG treatment also 1 2
recovered cell survival upon GPT2 knockdown (Fig. 4c).
0 0
Similar results were observed when we treated another TCA BPTES - + - + Gln + - + -
cycle intermediate, oxaloacetate (OAA) (Supplementary WT Mut WT Mut

Fig. 4D), demonstrating that GPT2 is required for sustain-

ing TCA cycle anaplerosis upon GLS inhibition.
Several groups have provided evidence that the addition
of Asn rescued cell viability in the absence of Gln [13, 15].
However, we found that Asn supplement failed to restore
cell survival when GPT2 activity is suppressed (Supple-
mentary Fig. 4E). Additionally, we found that alanine (Ala),
another byproduct of GPT2, or antioxidants treatment had
minor effect on cell survival under these conditions (Fig. 4d
and Supplementary Fig. 4F–H).
As GPT2 converts glutamate to αKG, glutamate sup-
plement is necessary for GPT2 function. Recently, it has
been reported that the re-accumulation of glutamate by
alternative metabolic pathways after GLS inhibition [22].
Indeed, we found that ASNS, which can produce glutamate
in a GLS independent manner, is highly elevated upon GLS
inhibition (Fig. 4e). Moreover, non-Gln derived glutamate
producing enzymes, including branched chain

Fig. 3 Mitochondrial Gln metabolism inhibition induces GPT2 via
ROS-ATF4 pathway. a, b Relative GPT2 mRNA (a) and protein (b)
levels in HeLa cells treated with or without BPTES and/or NAC (5
mM) for 24 h. c Relative GPT2 mRNA levels in HeLa cells treated
with or without Gln and/or NAC for 24 h (n = 3). d, e GPT2 promoter
luciferase activity. HeLa cells were co-transfected with the pGL3-
GPT2 promoter and Renilla reporter constructs. Renilla luciferase
activity was measured to normalize the transfection efficiency. Luci-
ferase activity was measured in cells treated with or without BPTES
and/or NAC (d), or with or without Gln and/or NAC (e) for 24 h (n =
3). f Relative ATF4 mRNA levels in HeLa cells treated with or without
BPTES and/or NAC for 24 h (n = 3). g, h HeLa cells were co-
transfected with ATF4-binding-site-intact (WT) or ATF4-binding-site-
mutated (Mut) GPT2 promoter. Luciferase activity was measured in
cells incubated with or without BPTES (g), and with or without Gln
(h) for 24 h (n = 3). All error bars, ± SEM. **p < 0.01 and ***p <
0.001
aminotransferase 1 (BCAT1), Gamma-Glutamyl hydrolase
(GGH) and 5-oxoprolinase (OPLAH), were also increased
in these cells (Fig. 4e). Furthermore, BPTES treatment had
only a modest impact on intracellular glutamate levels,
indicating that alternate metabolic pathways can recover
cellular glutamate upon GLS inhibition (Fig. 4f). Taken
together, these results demonstrate the essentiality of GPT2
in cell survival by supplying the TCA cycle anaplerosis
upon GLS inhibition.
Inhibition of GPT2 synergizes with GLS inhibitors to
suppress cancer cell survival
Our data indicate that GPT2 has an important role in the
adaptive metabolic response to GLS inhibition. Until
recently, drugs targeting GLS are being developed for
cancer therapy [9, 10]. However, poor bioavailability and
low potency mitigate its use as a viable therapeutic strategy
[22]. Thus, these observations suggest that inhibition of
GPT2 may have therapeutic implications by sensitizing
cancer cells to GLS inhibition. To test this idea, we inhib-
ited GPT2 activity and examined whether this would
synergize with BPTES treatment in multiple cancer cell
lines including MiaPaCa2, HCT116, A549 and MCF-7.
Although each cancer line exhibited different sensitivity to
either AOA or BPTES, the combinational treatment pro-
duces synthetic lethality in all tested cancer cell lines (Fig.
5a–d). To further confirm the importance of GPT2 activity
upon GLS inhibition in tumorigenic properties of cancer
cells, we assessed their clonogenic growth. As expected, the
treatment with AOA potently synergized with BPTES to
diminish the clonogenic properties of cancer cells
(Supplementary Fig. 5A–D). In conclusion, our results
demonstrate that the combination treatments aimed at
inhibiting GLS and GPT2 potently suppress cancer cell
survival.
M. Kim et al.
Discussion
In this study, we demonstrate a novel dependence on
mitochondrial GPT2 for cell survival and growth after the
repression of mitochondrial Gln metabolism. We dis-
covered that Gln deprivation or GLS inhibition-mediated
ROS production induces the transcription of GPT2 via
ATF4 (Fig. 5e). Significantly, the induction of GPT2 is
required for supplying cellular αKG upon GLS inhibition,
highlighting the potential importance of GPT2-mediated
adaptive metabolic response in cell survival and growth.
This idea is further validated by the finding that GPT2
inhibition sensitizes cancer cells to the repression of mito-
chondrial Gln metabolism.
Many therapeutic approaches have been developed to
target metabolic pathways of cancer cells [1]. As GLS is the
most proximal enzyme in mitochondrial Gln metabolism,
GLS inhibition has been used to treat cancers [9, 10].
However, cancer cells have the abilities to adapt to meta-
bolic challenges by activating alternative metabolic path-
ways or by scavenging for fuel sources [15, 22, 23]. Indeed,
we found that chronic GLS inhibition had a minor growth-
suppressive effect on several cancer cells. We demonstrate
that GPT2 is intimately implicated in adaptive metabolic
responses to exposure of GLS inhibitors. Importantly, AOA
treatment even at low doses significantly sensitizes cells to
GLS inhibition-induced growth arrest and cell death. This
finding may have important therapeutic implications, given
that its cellular toxicity at higher doses limits the clinical use
of AOA [4]. The essentiality of GPT2 on cell survival upon
GLS inhibition and the fact that GLUD1 and GOT2 are
dispensable under these conditions may provide potential
therapeutic strategies for cancer treatment.
Our studies reveal the important impact of a mitochon-
drial transaminase GPT2 on cell survival and proliferation
by supplying αKG in GLS-impaired cells. It has been
shown that a transaminase GOT2 generates aspartate,
allowing Kras-driven cancer cells to avoid cell cycle arrest
upon Gln starvation by producing nucleotides [24].
Although GOT2 also can fulfill TCA cycle by producing
αKG, it consumes OAA, another TCA cycle intermediate,
to convert glutamate to αKG. Thus, cells may prefer to use
pyruvate to produce αKG via GPT2 upon GLS inhibition.
In support of this idea, we show that GPT2, but not GOT2,
is markedly induced in GLS-impaired cells (Fig. 1).
GLUD1, which produces αKG without the need for other
amino acids, has been shown to be required for cell survival
when glucose is scarce [25]. Thus, it will be interesting for
future work to investigate whether the relative abundance of
pyruvate is a determinant of these metabolic choices.
Recent studies have shed light on the mechanisms
through which mitochondrial Gln metabolism regulates
Mitochondrial GPT2 plays a pivotal role in metabolic adaptation to the perturbation of mitochondrial. . .

A B
1.2 50

***
40
Relative cell viability ***

Cell death (%)

0.8
30

20
0.4

10

0.0 0
BPTES - + - + + BPTES - + - + +
AOA - - + + + AOA - - + + +
DM KG - - - - + DM KG - - - - +

C ***
D
30 DM SO 1.2
BPTES
BPTES + DM KG

Relative cell viability

**
Cell death (%)

20 0.8

10 0.4
**
ns

0 0.0
shGFP shGPT2 #1 shGPT2 #2
BPTES - + - + + +
AOA - - + + + +
NAC - - - - - +
Ala - - - - + -

E F ns
1.2 ns
Relative mRNA expression levels

15 DM SO
*** BPTES
Relative glutamate levels

10 0.8

5 0.4
**
ns *

0 0.0
ASNS BCAT1 GGH OPLAH
BPTES 0 5 10
(μM )

Fig. 4 αKG rescue cell survival treated with BPTES and AOA. a 3). d Relative cell viability of HeLa cells treated with BPTES, AOA,
Relative cell viability of HeLa cells treated with BPTES, AOA and/or NAC and/or Ala (4 mM) (n = 4). e Relative mRNA levels of ASNS,
DMKG (4 mM) for 72 h (n = 4). b Survival of HeLa cells cultured BCAT1, GGH and OPLAH in HeLa cells treated with or without
with BPTES, AOA and/or DMKG for 72 h (n = 3). c Survival of HeLa BPTES for 24 h (n = 3). f Relative Glu levels in HeLa cells treated
cells expressing a control shRNA or two independent shRNAs to with or without BPTES (n = 4). All error bars ± SEM. n.s., not sig-
GPT2 incubated with or without BPTES and/or DMKG for 72 h (n = nificant. *p < 0.05, **p < 0.01 and ***p < 0.001
 M. Kim et al.

A MiaPaCa2 B HCT116 mitochondrial integrity [5], which can increase mitochon-

1.2 1.2 *** drial ROS production. Given the importance of Gln ana-
plerosis in maintaining cellular redox state, it has been

Relative cell viability

Relative cell viability

reported that growth defects caused by Gln deprivation were

0.8 ** 0.8 rescued by the treatment of antioxidants such as NAC or
*** GSH [6]. However, we found that elevated ROS upon GLS
inhibition induces GPT2 expression, which is important for
0.4 0.4 cell growth and survival. Moreover, when GPT2 is
impaired, NAC treatment has minimal effects on GLS
inhibition-mediated growth suppression. These findings are
0.0 0.0 in line with a previous work showing that an enhanced
BPTES - + - + BPTES - + - + ROS production contributes to cell survival of Gln-
AOA - - + + AOA - - + + depleted cancer cells by promoting transcriptional
activation of Sestrin2 [23]. Thus, further studies are
C A549 E needed for delineating the roles of ROS in Gln-deprived
1.2
conditions.
Relative cell viability

***
0.8
Materials and methods

Cell culture
0.4

HeLa and HCT116 cell lines used in this work have been
described before [27]. MiaPaCa2 and MCF-7 cell lines were
0.0
provided by Dr. Alec C. Kimmelman (NYU Langone
BPTES - + - +
Medical Center, New York, NY, USA). A549 cell line was
AOA - - + +
provided by Dr. Jeong-Hwa Lee (The Catholic University
D MCF-7 of Korea, Seoul, Korea). Primary MEFs were immortalized
1.2 with retrovirus expression SV40 large T antigen at passage
3. MEFs, HeLa, MiaPaCa2, HCT116, MCF-7, and A549
Relative cell viability

were cultured in Dulbecco’s modified Eagle’s medium

***
0.8 (Biowest, Nuaillé, France) supplemented with 10% fetal
bovine serum (FBS, Biowest) and penicillin/streptomycin
(Biowest). All cell lines were tested routinely for myco-
0.4 plasma contamination by using a commercially available kit
(Takara Bio Inc, Shinga, Japan).

0.0 Constructs and reagents

BPTES - + - +
AOA - - + + The following antibodies were used: GLS1 (ab93434,
Fig. 5 Multiple cancer cell lines suffer the consequence of synthetic Abcam, Cambridge, MA, USA), GLUD1 (H00002746-
lethality by combined BPTES and AOA. a–d Relative cell viability of M01, Abnova, Taipei, Taiwan), GOT2 (CSB-PA912447,
MiaPaCa2 (a), HCT116 (b), A549 (c), and MCF-7 (d) cell lines Cusabio biotech, Wuhan, China), GPT2 (16757-1-AP,
treated with or without BPTES and/or AOA (n = 4). e A proposed
model illustrating the regulation of cell survival via ROS-ATF4-GPT2
Proteintech Group, Rosemont, IL, USA) and β-actin
during mitochondrial Gln metabolism-inhibited conditions. All error (A4700, Sigma, St. Louis, MO, USA). Ala, glutamate, Asn,
bars, ± SEM. **p < 0.01 and ***p < 0.001 OAA, cycloserine, AOA, DMKG, GSH and NAC were
purchased from Sigma. BPTES was purchased from Cay-
man Chemical (Ann Arbor, MI, USA). CB-839 was
cellular ROS production [6]. For example, Gln metabolism obtained from Selleckchem (Houston, TX, USA). shRNAs
provides the metabolic precursors necessary for de novo for GPT2 (TRCN0000035026 and TRCN0000035028)
synthesis of GSH, an intracellular antioxidant [26]. Addi- were purchased from Sigma. Other shRNAs were described
tionally, inhibition of Gln anaplerosis results in the loss of previously [6].
Mitochondrial GPT2 plays a pivotal role in metabolic adaptation to the perturbation of mitochondrial. . .
Quantitative RT-PCR
Total RNA was prepared with Ribospin kit (GeneAll,
Seoul, Korea) according to the manufacturer’s instructions.
In all, 0.5 μg of total RNA was reverse-transcribed using the
iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA).
Diluted cDNAs were analyzed by real-time PCR using
SYBR Green I Mastermix on a Lightcycler 480 (Roche,
South San Francisco, CA, USA). The level of gene
expression was normalized to β-actin. The primer sequences
were: TTCCAGAAGGCACAGACATG and GGCTCAG
TACTCTTTCACCAG for human GLS1; AGGAATGA
CACCAGGGTTTG and TCAGACTCACCAACAGCAA
TAC for human GLUD1; GTTTGCCTCTGCCAATCA
TATG and GAGGGTTGGAATACATGGGAC for human
GOT2; CATGGACATTGTCTGAACC and TTACCCAG
GACCGACTCCTT for human GPT2; CAGCTGAAA
GAAGCCCAAGT and AGAGCCTGAATGCCTTCCTC
for human ASNS; TGAGGCTTGGCTTTTGTGAA and
GGCTCTGGTGTAACAAAGCC for human BCAT1;
CCCAGTGCTACTTTGAGGGG and CTGTTACTGTC
GATGATGAGGC for human OPLAH; GAGTCTG
CAGGTGCGAGAGTTGTA and TTTGGCCACTTTAG
CATAATCTGAGC for human GGH; CCAACAA
CAGCAAGGAGGAT and AGTGTCATCCAACGTG
GTCA for human ATF4; CTACGTCGCCC TGGA
CTTCGAGC and GATGGAGCCGCCGATCCACACGG
for human β-actin; ATTATGACTCCAGAACAGCCC and
AGTGTAGTGCTTCATCCATGG for mouse GLS1;
AGGAATGACACCAGGCTTTG and CTCCAACACC
AACACATTTAGC for mouse GLUD1; GATCGG
CATGTTCTGTTTCAC and CATGTAGACCGAGAAC
TCCTTG for mouse GOT2; CTGCACCTACCCAAACC
TAC and TGCCACATCTTCACGGATAC for mouse
GPT2; and AGCCATGTACGTAGCCATCC and CTCTC
AGCTGTGGTGGTGAA for mouse β-actin.
Western blotting
Cells were lysed with EZ-RIPA lysis buffer (ATTO, WSE-
7420, Tokyo, Japan) supplemented with protease inhibitor
cocktail (Roche) and phosphatase inhibitors (Sigma). Cell
lysates were separated by sodium dodecyl sulfate-
polyacrylamide gel electrophoresis (SDS-PAGE) and
immunoblotting.
Clonogenic assay
Cells were plated in 6-well plates at 300 cells per well in 2
ml of media and treated with drugs the day after seeding.
After 8–12 days, colonies were fixed with 80% methanol
and stained with 0.2% crystal violet. Media was not chan-
ged throughout the course of the experiment.
ROS determination
The levels of cellular ROS were assessed by using DCFDA
(Invitrogen, Grand Island, NY, USA) according to the
manufacturer’s instructions. Cells were washed with PBS
and then incubated at 37 °C for 30 min in Hank’s balanced
salt solution (Gibco, Big Cabin, OK, USA) containing 5 μM
DCFDA. Cells were trypsinized, pelleted by centrifugation
and resuspended in Hank’s balanced salt solution. Fluor-
escence was measured by flow cytometry using a FACS
Canto (BD Biosciences, San Jose, CA, USA).
Cell viability assay
Cells were plated into 96-well plates at 1000 cells per well
in 100 μl of growth media. The following day, the cells
were treated with drugs or were replaced by growth media.
Parallel plates were analyzed at 3 days by Cell Titer Glo
analysis (Promega, Madison, Wisconsin, USA), per the
manufacturer’s instruction.
Flow cytometric measurement of cell death
Cells at less than 80% confluence were treated with drugs.
After 72 h, cells were harvested by trypsinization, pelleted
by centrifugation, and resuspended in PBS. The measure-
ment of cell death was performed by flow cytometry using
propidiumiodide (PI) staining, as previously described [28].
Reporter assay
The human GPT2 promoter was cloned into pGL3 basic
vector to obtain a pGL3-GPT2 promoter plasmid. pGL3-
GPT2 promoter was co-transfect with Renilla luciferase-
expressing pRL-TK. After 18 h of transfection, the medium
was changed to complete or Gln-free medium for 24 h.
Luciferase activity was then assayed according to the
manufacturer’s instructions (Promega). Firefly luciferase
activity was normalized to Renilla luciferase activity.
Mutations in the GPT2 promoter were generated with the
Quickchange kit (Agilent Technologies, CA, USA) and
primer sequences were: CCGTGTGGCCTTGGACCCT
GAAACTCGGGGCG and CGCCCCGAGTTTCAGGGT
CCAAGGCCACACGG.
Intracellular glutamate detection
Cells were plated into 96-well plates in 100 μl of growth
media. The following day, the cells were supplemented
media with or without BPTES. Parallel plates were ana-
lyzed at 3 days by Glutamate-Glo™ Assay Kit (Promega),
per the manufacturer’s instruction. Glutamate levels nor-
malized to the number of cells in each well.

Statistical analysis
Statistical analysis was performed using GraphPad Prism
5 software (GraphPad Software, Inc, San Diego, CA, USA).
The graphs show the mean from independent experiments
unless otherwise specified (SD is standard deviation and
SEM is standard error of the mean). Sample size and
number of independent experiments for each experiment is
indicated in the appropriate figure legend. Unless otherwise
noted, n equals the number of independent biological
replicates. All experiments were performed at least twice.
Statistically significant differences were calculated using
unpaired two-tailed Student’s t-test or two-way ANOVA. A
p-value of < 0.05 was considered statistically significant.
Acknowledgements This research was supported by Basic Science
Research Program through the National Research Foundation of Korea
(NRF) funded by the Ministry of Education (2015R1C1A1A01052548
and 2018R1D1A1B07040961).
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of
interest.
Publisher’s note: Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
References
1. Tennant DA, Duran RV, Gottlieb E. Targeting metabolic trans-
formation for cancer therapy. Nat Rev Cancer. 2010;10:267–77.
2. Vander Heiden MG, Cantley LC, Thompson CB. Understanding
the Warburg effect: the metabolic requirements of cell prolifera-
tion. Science. 2009;324:1029–33.
3. Wise DR, Thompson CB. Glutamine addiction: a new therapeutic
target in cancer. Trends Biochem Sci. 2010;35:427–33.
4. Altman BJ, Stine ZE, Dang CV. From Krebs to clinic: glutamine
metabolism to cancer therapy. Nat Rev Cancer. 2016;16:619–34.
5. DeBerardinis RJ, Mancuso A, Daikhin E, Nissim I, Yudkoff M,
Wehrli S, et al. Beyond aerobic glycolysis: transformed cells can
engage in glutamine metabolism that exceeds the requirement for
protein and nucleotide synthesis. Proc Natl Acad Sci USA.
2007;104:19345–50.
6. Son J, Lyssiotis CA, Ying H, Wang X, Hua S, Ligorio M, et al.
Glutamine supports pancreatic cancer growth through a KRAS-
regulated metabolic pathway. Nature. 2013;496:101–5.
7. Yang S, Hwang S, Kim M, Seo SB, Lee JH, Jeong SM. Mito-
chondrial glutamine metabolism via GOT2 supports pancreatic
cancer growth through senescence inhibition. Cell death & Dis.
2018;9:55.
8. Panieri E, Santoro MM. ROS homeostasis and metabolism: a
dangerous liason in cancer cells. Cell death & Dis. 2016;7:e2253
9. Robinson MM, McBryant SJ, Tsukamoto T, Rojas C, Ferraris
DV, Hamilton SK, et al. Novel mechanism of inhibition of rat
kidney-type glutaminase by bis-2-(5-phenylacetamido-1,2,4-thia-
diazol-2-yl)ethyl sulfide (BPTES). Biochem J. 2007;406:407–14.
10. Gross MI, Demo SD, Dennison JB, Chen L, Chernov-Rogan T,
Goyal B, et al. Antitumor activity of the glutaminase inhibitor CB-839
in triple-negative breast cancer. Mol Cancer Ther. 2014;13:890–901.
M. Kim et al.
11. Polet F, Corbet C, Pinto A, Rubio LI, Martherus R, Bol V, et al.
Reducing the serine availability complements the inhibition of the
glutamine metabolism to block leukemia cell growth. Oncotarget.
2016;7:1765–76.
12. Sun L, Song L, Wan Q, Wu G, Li X, Wang Y, et al. cMyc-
mediated activation of serine biosynthesis pathway is critical for
cancer progression under nutrient deprivation conditions. Cell
Res. 2015;25:429–44.
13. Toda K, Kawada K, Iwamoto M, Inamoto S, Sasazuki T, Shir-
asawa S, et al. Metabolic Alterations Caused by KRAS Mutations
in Colorectal Cancer Contribute to Cell Adaptation to Glutamine
Depletion by Upregulation of Asparagine Synthetase. Neoplasia.
2016;18:654–65.
14. Ye J, Kumanova M, Hart LS, Sloane K, Zhang H, De Panis DN,
et al. The GCN2-ATF4 pathway is critical for tumour cell survival
and proliferation in response to nutrient deprivation. EMBO J.
2010;29:2082–96.
15. Zhang J, Fan J, Venneti S, Cross JR, Takagi T, Bhinder B, et al.
Asparagine plays a critical role in regulating cellular adaptation to
glutamine depletion. Mol Cell. 2014;56:205–18.
16. Wise DR, DeBerardinis RJ, Mancuso A, Sayed N, Zhang XY,
Pfeiffer HK, et al. Myc regulates a transcriptional program that
stimulates mitochondrial glutaminolysis and leads to glutamine
addiction. Proc Natl Acad Sci USA. 2008;105:18782–7.
17. Wong DT, Fuller RW, Molloy BB. Inhibition of amino acid
transaminases by L-cycloserine. Adv Enzym Regul. 1973;11:
139–54.
18. Ferguson J, Smith M, Zudaire I, Wellbrock C, Arozarena I.
Glucose availability controls ATF4-mediated MITF suppression
to drive melanoma cell growth. Oncotarget. 2017;8:32946–59.
19. Hao Y, Samuels Y, Li Q, Krokowski D, Guan BJ, Wang C, et al.
Oncogenic PIK3CA mutations reprogram glutamine metabolism
in colorectal cancer. Nat Commun. 2016;7:11971.
20. Salgado MC, Meton I, Anemaet IG, Baanante IV. Activating
transcription factor 4 mediates up-regulation of alanine amino-
transferase 2 gene expression under metabolic stress. Biochim
Biophys Acta. 2014;1839:288–96.
21. Wang SF, Chen MS, Chou YC, Ueng YF, Yin PH, Yeh TS, et al.
Mitochondrial dysfunction enhances cisplatin resistance in human
gastric cancer cells via the ROS-activated GCN2-eIF2alpha-
ATF4-xCT pathway. Oncotarget. 2016;7:74132–51.
22. Biancur DE, Paulo JA, Malachowska B, Quiles Del Rey M, Sousa
CM, Wang X, et al. Compensatory metabolic networks in pan-
creatic cancers upon perturbation of glutamine metabolism. Nat
Commun. 2017;8:15965.
23. Byun JK, Choi YK, Kim JH, Jeong JY, Jeon HJ, Kim MK, et al.
A Positive feedback loop between Sestrin2 and mTORC2 Is
required for the survival of glutamine-depleted lung cancer cells.
Cell Rep. 2017;20:586–99.
24. Patel D, Menon D, Bernfeld E, Mroz V, Kalan S, Loayza D, et al.
Aspartate rescues S-phase arrest caused by suppression of gluta-
mine utilization in KRas-driven cancer cells. J Biol Chem.
2016;291:9322–9.
25. Shanware NP, Mullen AR, DeBerardinis RJ, Abraham RT. Glu-
tamine: pleiotropic roles in tumor growth and stress resistance. J
Mol Med. 2011;89:229–36.
26. Lu SC. Regulation of glutathione synthesis. Mol Asp Med.
2009;30:42–59.
27. Jeong SM, Hwang S, Park K, Yang S, Seong RH. Enhanced
mitochondrial glutamine anaplerosis suppresses pancreatic cancer
growth through autophagy inhibition. Sci Rep. 2016;6:30767.
28. Jeong SM, Xiao C, Finley LW, Lahusen T, Souza AL, Pierce K,
et al. SIRT4 has tumor-suppressive activity and regulates the
cellular metabolic response to DNA damage by inhibiting mito-
chondrial glutamine metabolism. Cancer Cell. 2013;23:450–63.