ARTICLE
Received 25 Apr 2016
| Accepted 23 Nov 2016 | Published 9 Jan 2017
Pyruvate kinase type M2 promotes tumour cell
exosome release via phosphorylating
synaptosome-associated protein 23
Yao Wei
1,2,
*, Dong Wang
1,
*, Fangfang Jin
1,
*, Zhen Bian
3,
*, Limin Li
1
, Hongwei Liang
1
, Mingzhen Li
1
, Lei Shi
3
,
Chaoyun Pan
1
, Dihan Zhu
1
, Xi Chen
1
, Gang Hu
2
, Yuan Liu
3
, Chen-Yu Zhang
1
&KeZen
1,3
Tumour cells secrete exosomes that are involved in the remodelling of the tumour–stromal
environment and promoting malignancy. The mechanisms governing tumour exosome
release, however, remain incompletely understood. Here we show that tumour cell exosomes
secretion is controlled by pyruvate kinase type M2 (PKM2), which is upregulated and
phosphorylated in tumours. During exosome secretion, phosphorylated PKM2 serves as a
protein kinase to phosphorylate synaptosome-associated protein 23 (SNAP-23), which
in turn enables the formation of the SNARE complex to allow exosomes release. Direct
phosphorylation assay and mass spectrometry confirm that PKM2 phosphorylates SNAP-23
at Ser95. Ectopic expression of non-phosphorylated SNAP-23 mutant (Ser95-Ala95)
significantly reduces PKM2-mediated exosomes release whereas expression of selective
phosphomimetic SNAP-23 mutants (Ser95-Glu95 but not Ser20-Glu20) rescues
the impaired exosomes release induced by PKM2 knockdown. Our findings reveal a non-
metabolic function of PKM2, an enzyme associated with tumour cell reliance on aerobic
glycolysis, in promoting tumour cell exosome release.
DOI: 10.1038/ncomms14041
OPEN
1
State Key Laboratory of Pharmaceutical Biotechnology, Jiangsu Engineering Research Center for MicroRNA Biology and Biotechnology, Nanjing Advanced
Institute for Life Sciences, School of Life Sciences, Nanjing University, Nanjing, Jiangsu 210093, China.
2
School of Medicine and Life Sciences, Nanjing
University of Chinese Medicine, Nanjing, Jiangsu 210023, China.
3
Center for Immunology, Inflammation and Infectious Diseases & Department of Biology,
Georgia State University, Atlanta, Georgia 30302, USA. * These authors contributed equally to this work. Correspondence and requests for materials should
be addressed to G.H. (email: ghu@njutcm.edu.cn) or to Y.L. (email: yliu@gsu.edu) or to C.-Y.Z. (email: cyzhang@nju.edu.cn) or to K.Z. (email:
kzen@nju.edu.cn).
NATURE COMMUNICATIONS | 8:14041 | DOI: 10.1038/ncomms14041 | www.nature.com/naturecommunications 1
A
s a mechanism to communicate with the microenviron-
ment, tumour cells actively release large quantity of
extracellular vesicles (EVs), including exosomes, micro-
vesicles (MVs) or microparticles, and apoptotic bodies. These
tumour-released EVs, which are abundant in the body fluids of
patients with cancer, play a critical role in promoting tumour
growth and progression
1,2
. For example, NCI-H460 tumour cells
actively release MVs containing EMMPRIN, a transmembrane
glycoprotein highly expressed by tumour cells, MV-encapsulated
EMMPRIN that facilitates tumour invasion and metastasis via
stimulating matrix metalloproteinase expression in fibroblasts
3
.
Tumour cell exosomes also deliver active Wnt proteins to regulate
target cell b-catenin-dependent gene expression
4
. Cancer cell-
derived microparticles bearing P-selectin glycoprotein ligand 1
accelerate thrombus formation in vivo, and by targeting P-selectin
glycoprotein ligand 1 researchers were able to prevent
thrombosis
5
. While these studies are exciting and suggest
potential strategies for blocking metastasis, the mechanism
underlying the active exocytosis of exosomes by tumour cells,
however, remains unclear. Previous studies suggest that cellular
exosome secretion activity is increased during tumorigenesis
6,7
,
but the molecular basis for switching on the exocytosis process in
tumour cells requires further clarification.
The mechanisms that govern cell endosomal secretion have
been extensively studied. Exosomes share structural and bio-
chemical characteristic with intraluminal vesicles of multi-
vesicular endosomes (MVEs). Studying trafficking of proteolipid
protein in Oli-neu cells, Trajkovic et al.
8
reported that the
sphingolipid ceramide played a key role in triggering budding of
exosomes into MVEs, and the release of exosomes was reduced
after the inhibition of ceramide synthesis. Furthermore, Kosaka
et al.
9
found that neutral sphingomyelinase 2 was directly
involved in promoting tumour cell endosomal secretion. Using an
RNAi screen, Ostrowski et al.
10
identified the role of Rab
GTPases in promoting exosome secretion: among the small
GTPases, Rab27a and Rab27b were involved in MVE docking to
the plasma membrane. Like other cells, tumour cells employ the
soluble N-ethylmaleimide-sensitive fusion factor attachment
protein receptor (SNARE) complex that many cell types utilize
in the exocytic release of exosomes
11
. The SNARE complex is
comprised of proteins on membrane of budding vesicles
(v-SNAREs) and proteins on the cell’s membrane (t-SNAREs).
The v-SNAREs and t-SNAREs enable the apposition of the vesicle
and cell membranes and the subsequent fusion of the two
membranes thereby mediating vesicle exocytosis. In tumour cells,
the SNARE complex includes syntaxin-4 (ref. 12) and SNAP-23
(ref. 13) serving as t-SNAREs, while VAMP-2 (ref. 14) and
VAMP-8 (refs 12,15) represent candidates for v-SNAREs.
Phosphorylation of SNAP-23 not only directly increases cell
exocytosis
16,17
but also promotes association with other SNARE
proteins, thereby allowing the formation of the stable SNARE
complex to enhance cell exocytosis
18
. In mast cells, SNAP-23 has
been reported to be phosphorylated by IkB kinase (IKK) to
promote exocytosis
19,20
. However, the kinase that phosphorylates
SNAP-23 in the tumour cell has not been identified.
In the present study, we demonstrate that PKM2, an enzyme
involved in the tumour cell’s reliance on aerobic glycolysis
(Warburg effect), plays a critical role in promoting the release
of exosomes from the tumour cell. Specifically, we identify
SNAP-23, which controls the docking and release of secretory
granules or exosome-containing multivesicular bodies, is a
substrate of PKM2 in tumour cells. During exosome secretion,
phosphorylated PKM2 forms a dimer structure with low pyruvate
kinase activity but high protein kinase activity
21
and then
associates with SNAP-23 near cell’s membranes, leading to
SNAP-23 phosphorylation at Ser95 and upregulation of tumour
cell exosome release. We conclude that PKM2, following
phosphorylation and dimerization, plays an essential role in
not only switching tumour cell metabolism from oxidative
phosphorylation to aerobic glycolysis, but also promoting
tumour cell exosome secretion via directly phosphorylating
SNAP-23.
Results
Exosome secretion requires high level of aerobic glycolysis.
Exosomes released by various cell types were isolated from cell
culture medium using exosome isolation kit and analysed by
transmission electron microscopy and western blot (WB) using
antibodies against exosomal marker proteins. As shown in Fig. 1a,
tumour cells actively release exosomes, a double membrane
vesicle with 50–100 nm size, into the culture medium. The
immune-gold label showed that exosomes expressed membrane
protein CD63 (Fig. 1a, right lower panel). Expression of
additional marker proteins such as CD63, Tsg101 and CD9 was
validated by WB analysis (Fig. 1b). To monitor the concentration
of exosomes released by tumour cells and non-tumour primary
culture cells, a Nanosight NS 300 system (NanoSight) was used to
track the release of exosomes (Fig. 1c). Nanoparticle tracking
analysis (NTA) confirmed that the sizes of released exosomes are
around 100 nm. In agreement with previous reports, we found
that tumour cells (SW480, Hela, A549 and HepG2 cells) generally
displayed more active exosome secretion than non-tumour
mammalian cells, mouse primary myoblast and mammary
epithelial cell (MEC) (Fig. 1d). Interestingly, the increased
exosome release by tumour cells is positively correlated to the
higher aerobic glycolysis (Fig. 1e). In line with the positive
correlation between aerobic glycolysis and exosome secretion
observed in Fig. 1f, we found that glycolysis level was positively
correlated with the amount of exosome release in tumour cells
(Supplementary Fig. 1). To further examine the potential link
between the exosome secretion and the aerobic glycolysis flow in
the tumour cells, we treated A549 cells and HepG2 cells with
aerobic glycolysis inhibitor shikonin
22
or aerobic glycolysis
promoter tumour necrosis factor a (ref. 23), and then
determined the alteration of exosome release from tumour cells.
The results clearly showed that suppressed aerobic glycolysis by
shikonin inhibited, whereas activated aerobic glycolysis by TNFa
enhanced, the release of exosomes by tumour cells (Fig. 1g).
Identification of the dependence of exosome secretion process on
cell glycolysis is in agreement with previous findings in other cell
types
22,24
.
Previous reports showed that epidermal growth factor (EGF)
and oleanolic acid (OA) can enhance or inhibit exosome secretion
process
25,26
, respectively. We next treated the A549 tumour cells
with EGF or OA and examined whether the effect of these
reagents was mediated through alteration of tumour cell aerobic
glycolysis. As shown in Fig. 1h, EGF and OA significantly
enhanced and suppressed aerobic glycolysis in A549 cells,
respectively. We next examined whether the effect of EGF and
OA was mediated through alteration of tumour cell aerobic
glycolysis. As shown in Fig. 1i, EGF and OA significantly
enhanced and suppressed aerobic glycolysis in A549 and HepG2
cells, respectively. Moreover, the effect EGF and OA on
promoting or inhibiting A549 and HepG2 cells exosome
exocytosis was abolished by decreasing or increasing aerobic
glycolysis, respectively. These results collectively suggest that
release of exosomes in tumour cells is dependent on cellular
aerobic glycolysis.
PKM2 plays a critical role in tumour cell exocytosis. PKM2
expression has been widely regarded as an important molecular
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14041
2 NATURE COMMUNICATIONS | 8:14041 | DOI: 10.1038/ncomms14041 | www.nature.com/naturecommunications
Lactate level (nM µl
–1
)
0.2
0.4
0.6
0.8
1.0
1.2
1.4
PBS
EGF
OA
**
g
d
ba
Exosome size (nm)
0 200 400 600 800
Exosome concentration
(10
6
particles per ml)
0
2
4
6
8
10
12
A549 exosomes
c
ih
Glycolysis level
0.5 1.0 1.5 2.0 2.5 3.0 3.5
Exosome concentration
(10
8
particles/5×10
6
cells)
0
5
10
15
20
25
30
35
Exosome vs Glucose (r
1
2
=0.83)
Exosome vs Lactate (r
2
2
=0.81)
A549 exosomes
200 nm 100 nm
Myoblast
MEC
SW480
Hela
A549
HepG2
Myoblast
MEC
SW480
Hela
A549
HepG2
Myoblast
MEC
SW480
Hela
A549
HepG2
Myoblast
MEC
SW480
Hela
A549
HepG2
Exosome concentration
(10
6
particles/5×10
6
cells)
0
10
20
30
40
PBS
Shikonin
TNF α
Relative glucose uptake (%)
0
100
200
300
400
**
NS
**
***
***
Lactate level (nM µl
–1
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
*
NS
*
**
**
e
Exosome concentration
(10
8
particles/5×10
6
cells)
0
5
10
15
20
25
30
*
**
**
***
NS
**
A549 HepG2
A549 HepG2
Relative glucose uptake (%)
50
100
150
200
250
300
350
400
*
Exosome concentration
(10
8
particles/5×10
6
cells)
0
5
10
15
20
25
30
*
*
*
**
**
**
*
**
**
*
*
*
**
*
*
*
PBS
PBS
OA
OA+TNF
α
EGF
EGF+shikonin
*
*
*
f
Cell
Exosome
Tsg101
CD63
CD9
55 kDa
45 kDa
25 kDa
35 kDa
Figure 1 | Release of exosomes by tumour cells depends on aerobic glycolysis. (a–c) Isolated exosomes from A549 cells assessed by transmission
electron microscopy (a), WB (b) and NTA (c). The lower right image in a represents the immune-gold labelling of CD63 in an exosome. (d) Tumour cells
release more exosomes than non-tumour cells. (e) Positive correlation between exosome secretion and aerobic glycolysis. (f) Linear regression between
glucose uptake (r
1
2
¼ 0.83) and lactate level (r
2
2
¼ 0.81). (g) Release of exosomes is dependent on cellular aerobic glycolysis. Cells were treated with
shikonin (1 mM) or tumour necrosis factor a (5 ng ml
1
) to inhibit or promote aerobic glycolysis. Exosome concentration was measured 24 h
post-treatment. (h) Effect of EGF and OA on cell aerobic glycolysis. Note that EGF (10 ng ml
1
), an enhancer of exosome release, increases aerobic
glycolysis, while OA (10 mgml
1
), an inhibitor of exosome release, decreases aerobic glycolysis. (i) EGF and OA regulate A549 cell exosome release via
altering cellular aerobic glycolysis. Data are presented as the mean
±
s.e.m. and represent at least three independent experiments with three replicates per
data point. NS, no significance. *Po0.05, **Po0.01, **Po0.001 as determined by the one-way ANOVA test.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14041 ARTICLE
NATURE COMMUNICATIONS | 8:14041 | DOI: 10.1038/ncomms14041 | www.nature.com/naturecommunications 3
feature of tumour development
27
. In tumour cells, PKM2 forms
a dimer, which is catalytically inactive for conversion of
phosphoenolpyruvate (PEP) to pyruvate and production of
ATP
28,29
. Lowering pyruvate formation provides a growth
advantage for tumour progression as blocking production of
pyruvate helps to channel the glycolytic intermediates to
biosynthesis to meet the demands for tumour cell proliferation.
Given that PKM2 plays a key role in switching tumour
cell metabolic status from oxidative phosphorylation to aerobic
glycolysis and PKM2 level is increased during tumorigenesis
30,31
,
we next tested whether tumour cell exosome release is correlated
with the PKM2 expression. In agreement with previous reports,
tumour cells contained significantly higher PKM2 level than
non-tumour cells (Fig. 2a). In line with the correlation between
aerobic glycolysis and exocytosis observed in Fig. 1, we found that
level of PKM2 was positively correlated with the amount of
exosome release in tumour cells (Fig. 2b). In a similar manner,
tumour cells also showed significantly higher phosphorylated
PKM2 (p-PKM2) level than non-tumour cells (Fig. 2c), and the
p-PKM2 level was positively correlated with the amount of
exosome release in tumour cells (Fig. 2d). Interestingly, treating
A549 lung carcinoma cells with OA (Fig. 2e) or EGF (Fig. 2f),
which enhanced or suppressed tumour cell exosome release,
respectively, we found that the level of PKM2 in A549 cells was
dose-dependently decreased by OA or increased by EGF.
To determine whether PKM2 particularly p-PKM2 plays a role
in modulating the release of exosomes from tumour cells, we
assessed exosome release after knocking down PKM2 level in
A549 and Hela tumour cells via PKM2 siRNA or overexpressing
PKM2 in myoblasts and MEC via HA-tagged PKM2-expressing
plasmid. As shown in Supplementary Fig. 2, knockdown or
overexpression of cellular PKM2 levels decreased or enhanced
58 kDa
0
2
4
6
8
10
12
Myoblast CTL-vector
Myoblast HA-PKM2
MEC CTL-vector
MEC HA-PKM2
Myoblast CTL-vector
Myoblast HA-PKM2
MEC CTL-vector
MEC HA-PKM2
PKM2/actin ratio
0.0
0.2
0.4
0.6
0.8
1.0
r
2
= 0.85
PKM2 level
0
5
10
15
20
25
30
Myoblast
MEC
SW480
Hela
A549
HepG2
p-PKM2/actin ratio
0.0
0.5
1.0
1.5
2.0
2.5
EGF (ng ml
–1
)
PKM2/actin ratio
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
OA (µg ml
–1
)
0 20 40 60 80 100
0 1 10 50 100
0 10 25 50 100
PKM2/actin ratio
0.0
0.2
0.4
0.6
0.8
1.0
1.2
PKM2/actin ratio
0.0
0.2
0.4
0.6
0.8
1.0
1.2
PKM2
Actin
PKM2
Actin
PKM2
Actin
PKM2
Actin
PKM2
Actin
g
i
hfe
l
j
A549 ncRNA
0
2
4
6
8
10
12
14
16
18
20
A549 ncRNA
A549 PKM2 siRNA
Hela ncRNA
Hela PKM2 siRNA
A549 PKM2 siRNA
Hela ncRNA
Hela PKM2 siRNA
PKM2/actin ratio
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
**
**
r
2
= 0.88
PKM2 level
0 102030405060
0
5
10
15
20
25
30
Myoblast
MEC
SW480
Hela
A549
HepG2
Myoblast
MEC
SW480
Hela
A549
HepG2
Myobl
ast
MEC
SW480
Hela
A549
HepG2
**
**
*
NS
NS
**
**
***
k
WB: anti-PKM2
Crosslinking
–
+
PBS
FBP
Serine
pTyr
Tetrameric
PKM2 (232 kDa)
Dimeric
PKM2 (116 kDa)
Monomeric
PKM2 (58 kDa)
PKM2 (58 kDa)
ba dc
p-PKM2
Actin
***
***
**
**
**
**
**
**
***
***
**
**
***
**
**
**
PBS
pTyr
FBP
Ser
0
5
10
15
20
25
*
**
**
43 kDa
58 kDa
43 kDa
58 kDa
43 kDa
58 kDa
43 kDa
58 kDa
43 kDa
58 kDa
43 kDa
Exosome concentration
(10
8
particles/5×10
6
cells)
Exosome concentration
(10
8
particles/5×10
6
cells)
Exosome concentration
(10
8
particles/5×10
6
cells)
Exosome concentration
(10
8
particles/5×10
6
cells)
Exosome concentration
(10
8
particles/5×10
6
cells)
Figure 2 | PKM2 plays a critical role in release of exosomes in tumour cells. (a) Relative level of PKM2 in tumour or non-tumour cells. (b) Linear
regression represents a positive correlation between PKM2 levels in different cell lines with exosome secretion. (c) Phosphorylated PKM2 level in tumour
or non-tumour cells. (d) Linear regression represents a positive correlation of phosphorylated PKM2 level in different cell lines with exosome secretion.
(e) OA, an inhibitor of exosome release, decreases PKM2 level. (f) EGF, an enhancer of exosome release, increases PKM2 level. (g,h) Knockdown of PKM2
in A549 and HeLa tumour cells via PKM2 siRNA (g) reduces the release of exosomes (h). (i,j) Overexpression of PKM2 in mouse primary myoblast cells
and mammary epithelial cells (MEC) via transfection with HA-PKM2-expressing plasmid (i) increases the release of exosomes (j). (k) Effect of pTyr, FBP
and serine on the switch of PKM2 from tetrameric formation to dimeric formation in A549 tumour cells. (l) Effects of pTyr (100 mM), FBP (500 mM) and
serine (5 mM) on exosome release in A549 tumour cells. Data are presented as the mean
±
s.e.m. of three independent experiments. *Po0.05. **Po0.01.
***Po0.001 as determined by the one-way ANOVA test (two-tailed t-test for g–j).
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14041
4 NATURE COMMUNICATIONS | 8:14041 | DOI: 10.1038/ncomms14041 | www.nature.com/naturecommunications
p-PKM2 levels. NTA results clearly showed that knockdown of
PKM2 in A549 and Hela cells (Fig. 2g) strongly reduced the
release of exosomes (Fig. 2h). In contrast, overexpression of
HA-tagged PKM2 (HA-PKM2) in non-tumour cells (Fig. 2i)
markedly enhanced the release of exosomes (Fig. 2j). Given that
apoptotic cells secrete more exosomes than healthy cells and
aerobic glycolysis inhibitor or PKM2 knockdown may affect cell
apoptosis, we next determined the effects of shikonin and PKM2
knockdown on cell apoptosis using flow cytometry. As shown in
Supplementary Fig. 3, both shikonin treatment and PKM2
knockdown significantly increased early or late apoptosis of
A549 cells. Considering that total exosome release from A549
cells is decreased after shikonin treatment or PKM2 knockdown,
increase of cell apoptosis by shikonin treatment or PKM2
knockdown further demonstrates that PKM2-mediated aerobic
glycolysis promotes tumour cell exosome release. Furthermore,
given that switching the behaviour of PKM2 from a tetramer
form to a dimer form increases the initial steps of tumour cell
aerobic glycolysis and promotes tumour progression
21,32
,
we treated A549 cells with pTyr, a phosphotyrosine peptide
that can promote PKM2 dimeric formation
33
, or fructose
1,6-bisphosphate (FBP) and serine, two molecules that enhance
PKM2 tetrameric formation
34
. To assess the dimeric or
tetrameric formation of PKM2, chemical crosslinking reaction
was carried out to maintain the polymer structure before WB
analysis
21
. Parallel samples without crosslinking treatment were
included as loading controls. As expected, pTyr treatment
increased the level of PKM2 dimer (116 kDa), while FBP and
serine enhanced tetrameric formation (232 kDa) in A549 cells
(Fig. 2k). Consistent with the configuration of PKM2 either
facilitating or reducing exosome exocytosis, pTyr, induced
dimeric PKM2, increased tumour cell exosome exocytosis, while
FBP and serine, which induced tertrameric PKM2, significantly
decreased tumour cell exosome exocytosis (Fig. 2l).
In addition, through assaying the level change and the effect on
secretion exosomes of PKM1, we found that pyruvate kinase
activity of PKM might be not relevant to tumour cell exosome
secretion. As shown in Supplementary Fig. 4, overexpression or
knockdown of PKM1 in Hela and A549 cells displayed no effect
on the release of exosomes from tumour cells. Taken together,
these results strongly argue that PKM2, particularly phosphory-
lated PKM2 which easily dimerizes, plays an essential role in
promoting the release of exosomes.
PKM2-promoted exosome release is dependent on SNAP-23.
As a critical component of general cell exocytosis machinery,
SNAP-23 has been widely reported to be involved in controlling
cell exocytosis
13,16,17
. We isolated exosomes released by SW480,
A549, Hela, 293 T and LLC cells, and then performed mass
spectra analysis and an isobaric tags for relative and absolute
quantitation (iTRAQ) assay for protein expression profiling. As
expected, protein profiling analysis showed that SNAP-23 was the
only component of SNARE complex detected in exosomes
derived from all the five cell lines (Supplementary Table 1). To
test whether PKM2-promoted tumour cell exocytosis is through
SNAP-23-mediated exocytic machinery, we first compared the
levels of SNAP-23 and phosphorylated SNAP-23 (p-SNAP-23) in
tumour or non-tumour cells (Fig. 3a). As shown, tumour cells
c
b
WB: SNAP23
WB: HA
WB: actin
HA-PKM2
–
–
–
–
SNAP23 siRNA
SNAP-23/actin ratio
0.0
0.5
1.0
1.5
2.0
2.5
ncRNA
SNAP23 siRNA
ncRNA + HA-PKM2
SNAP23 siRNA + HA-PKM2
ncRNA
SNAP23 siRNA
ncRNA + HA-PKM2
SNAP23 siRNA + HA-PKM2
Exosome concentration
(10
8
particles/5×10
6
cells)
0
5
10
15
20
25
+Phos-Tag
–Phos-Tag
Non-phosphorylated SNAP-23
SNAP-23 (23 kDa)
Actin (43 kDa)
Phosphorylated SNAP-23
Myoblast
MEC
SW480
Hela
A549
HepG2
a
*
*
**
**
Myoblast
MEC
SW480
Hela
A549
HepG2
p-PKM2/actin ratio
0
1
2
3
4
5
***
***
***
***
**
23 kDa
55 kDa
43 kDa
+
+
+
+
Figure 3 | PKM2-promoted exosome release in tumour cells is dependent on SNAP-23. (a) Relative level of SNAP-23 and phosphorylated SNAP-23 in
tumour or non-tumour cells. (b) Knockdown of SNAP-23 and overexpression of PKM2 in A549 tumour cells via SNAP-23 siRNA and HA-PKM2-expressing
plasmid, respectively. (c) Knockdown of SNAP-23 decreases PKM2-promoted release of exosomes in A549 cells. Data are presented as the mean
±
s.e.m.
of three independent experiments. *Po0.05. **Po0.01. ***Po0.001 as determined by the one-way ANOVA test.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14041 ARTICLE
NATURE COMMUNICATIONS | 8:14041 | DOI: 10.1038/ncomms14041 | www.nature.com/naturecommunications 5