PUBLISHED VERSION
http://hdl.handle.net/2440/101545
Dawei Liu, Zeeshan Shaukat, Tianqi Xu, Donna Denton, Robert Saint, Stephen Gregory
Autophagy regulates the survival of cells with chromosomal instability
Oncotarget, 2016; 7(39):63913-63923
© Creative Commons License All site content, except where otherwise noted, is licensed under a
Creative Commons Attribution 3.0 License.
Originally published at:
http://doi.org/10.18632/oncotarget.11736
PERMISSIONS
http://creativecommons.org/licenses/by/3.0/
Oncotarget63913
www.impactjournals.com/oncotarget
www.impactjournals.com/oncotarget/
Oncotarget, Vol. 7, No. 39
Autophagy regulates the survival of cells with chromosomal
instability
Dawei Liu
1
, Zeeshan Shaukat
1
, Tianqi Xu
2
, Donna Denton
2
, Robert Saint
3
, Stephen
Gregory
1
1
Department of Genetics, University of Adelaide, Adelaide, SA, Australia
2
Centre for Cancer Biology, University of South Australia, Adelaide, SA, Australia
3
Flinders University, Adelaide, SA, Australia
Correspondence to: Stephen Gregory, email: stephen.gregory@adelaide.edu.au
Keywords: chromosomal instability, autophagy, mitophagy, parkin, drosophila
Received: October 08, 2015 Accepted: August 21, 2016 Published: August 31, 2016
ABSTRACT
Chromosomal instability (CIN) refers to genomic instability in which cells have
gained or lost chromosomes or chromosomal fragments. A high level of CIN is common
in solid tumours and is associated with cancer drug resistance and poor prognosis.
The impact of CIN-induced stress and the resulting cellular responses are only just
beginning to emerge. Using proliferating tissue in Drosophila as a model, we found
that autophagy is activated in CIN cells and is necessary for their survival. Specically,
increasing the removal of defective mitochondria by mitophagy is able to lower levels
of reactive oxygen species and the resultant cellular damage that is normally seen in
CIN cells. In response to DNA damage, CIN is increased in a positive feedback loop,
and we found that increasing autophagy by Tor depletion could decrease the level
of CIN in proliferating cells. These ndings underline the importance of autophagy
control in the development of CIN tumours.
INTRODUCTION
Chromosomal instability (CIN) refers to the process
by which cells are unable to maintain chromosomal
integrity or number [1]. Chromosomal instability (CIN)
or genomic instability (GIN) has been suggested as a
pivot hallmark of cancer which facilitates the acquisition
of other cancer hallmarks required for tumorigenesis [2].
CIN is seen in most human solid tumours and the genetic
variation it generates can account for the development
of drug resistance and the poor prognosis of CIN cancer
patients [3, 4]. It has been proposed that CIN itself could
be an attractive target for chemotherapy, as it is a relatively
cancer-specic phenotype [5–7]. However, little is known
about which properties of CIN cells might allow CIN
tumours to be efciently killed.
Autophagy is a normal cellular pathway for the
degradation and recycling of unnecessary or dysfunctional
cellular components [8–10]. The process of autophagy
involves the sequestration of cytoplasmic material by double-
membrane phagophores to form autophagosomes that then
fuse with lysosomes to enable degradation of their cargo
[8]. Autophagy is induced in response to various stresses
to maintain metabolic homeostasis and prevent the build-
up of dysfunctional cellular components [9]. The aberrant
regulation of autophagy has been seen in several diseases,
especially in neurodegenerative disease and cancer [10–12],
as well as in cells in which aneuploidy has been induced
[13–15]. However, whether autophagy is protective or
deleterious in the development of cancer has been widely
debated [16]. The information currently available from
clinical trials and mouse models suggests that a lack of
autophagy predisposes tissue to develop tumours, possibly
because autophagy normally moderates oxidative stress
and DNA damage by removing defective mitochondria.
However, in some model systems, autophagy is essential for
the growth of the tumour [17, 18]. Consequently there are
now ongoing clinical trials evaluating the combination of
inhibition of autophagy with chemotherapeutics [19, 20]. The
expectation is that tumours may need autophagy to tolerate
the metabolic demands of proliferation, to avoid excessive
oxidative stress and consequently an unmanageable level of
genome instability. Thus reduced autophagy may promote
tumorigenesis by increasing DNA damage rates, but for
tumours to thrive they may need to increase their autophagic
ux to prevent deleterious levels of oxidative damage.
Research Paper
Oncotarget63914
www.impactjournals.com/oncotarget
In characterizing pathways which facilitate the
survival of CIN cells, we have previously reported
that CIN cells are sensitive to changes in glycolysis
or antioxidant enzymes and generate elevated levels
of reactive oxygen species [21]. Based on that study,
we carried out further screening for candidates whose
depletion can specically kill CIN cells. In this process,
we found that when CIN is induced in otherwise normal
cells, they become sensitive to the depletion of autophagy.
Here we show that CIN leads to an increase in autophagy,
and that autophagy is needed to limit reactive oxygen
species, DNA damage and cell death in CIN cells.
Furthermore, elevated levels of autophagy promote the
survival of CIN cells.
Altogether, our research highlights the signicance
of understanding autophagy pathways as a potential
therapeutic target for the treatment of CIN tumours.
RESULTS
Autophagy is activated when CIN is induced in
proliferating cells
We have previously used RNA interference
knockdown of the spindle assembly checkpoint gene mad2
or cohesin gene rad21 to generate inducible CIN models
with different CIN levels [22]. From this work, and that
of others [23] it has become clear that aneuploidy is
associated with elevated levels of reactive oxygen species
(ROS). We expected that in response, CIN cells would
induce autophagy to recycle damaged macromolecules.
To test autophagy levels in cells with induced CIN, we
initially used lysotracker staining, which was elevated
in both mad2 and rad21 CIN cells relative to normally
proliferating cells (Figure 1A–1C). To conrm this result
we examined the levels of a tagged form of Atg8a [24]. In
line with the lysotracker staining, we found robust Atg8a
puncta formation in rad21 CIN cells indicating autophagy
activation (Figure 1F). Stronger induction of autophagy
was seen in rad21 CIN cells than in mad2 CIN cells
(Figure 1A–1F), consistent with the higher level of CIN
generated in the rad21 model [22].
Reducing autophagy leads to increased oxidative
stress and apoptosis in CIN cells
Having found that autophagy is activated in CIN
cells, we hypothesized that robust autophagy activation
might be particularly needed for the survival of CIN
cells. In order to address this hypothesis, we depleted
the essential autophagy genes Atg1 or Atg18a [25, 26]
by RNA interference in CIN cells. Atg1 is needed for a
functional autophagy induction complex and leads to
the recruitment of Atg18/WIPI2, which is needed for
Atg8 recruitment and phagosome function [25, 27]. We
found that knocking down either Atg1 or Atg18 led to
dramatically increased levels of oxidative stress and DNA
damage in CIN cells (Figure 2, Supplementary Figure S1).
Furthermore, depletion of Atg1 or Atg18 in CIN cells
resulted in a signicant increase in apoptosis as detected
by active caspase staining (Figure 3). Elevated levels of
cell death were seen when autophagy was blocked in
either CIN model (Supplementary Figure S2). However,
depleting Atg1 or Atg18 in normal proliferating cells
had no detectable effect on ROS levels, DNA damage or
apoptosis. These results are consistent with a protective
role for autophagy in response to cellular stresses [28], and
showed that that autophagy activation was required for the
survival of CIN cells.
Enhancing autophagic ux rescues oxidative
stress levels and apoptosis in CIN cells
Having observed that CIN cells required autophagy
to avoid cell death, we wished to see whether enhancing
autophagic ux could improve the survival of CIN cells.
Autophagy induction is regulated by conserved upstream
signalling pathways that converge on the target of
rapamycin (TOR) kinase, which prevents autophagy by
inhibiting Atg1 [26, 29]. By the removal of the autophagy
inhibitor Tor using RNAi, we found that enhancing
autophagic ux (Supplementary Figure S4) could rescue
the oxidative stress and apoptosis phenotype in CIN cells
(Figure 4, Supplementary Figure S3). This suggested that
autophagy is not normally induced enough to protect cells
with high levels of CIN, and that elevated autophagy,
which is often seen in cancer [7, 8], can improve the
survival of these cells.
Autophagy of mitochondria is needed in CIN
cells to prevent ROS and cell death
One function of autophagy activation is the
removal of defective mitochondria through pink1/
parkin-mediated mitophagy [30]. CIN is known to cause
defective mitochondria and increased oxidative stress in
cells [21, 22], therefore, we checked whether mitophagy
is involved in the response to CIN. We found that
overexpression of the essential mitophagy gene parkin
reduced the level of ROS and apoptosis in CIN cells at
least as effectively as increasing general autophagy by
Tor depletion (Figure 4). Consistent with this, depletion
of Parkin signicantly increased apoptosis in CIN cells,
but not normal cells (Supplementary Figure S5). If
removal of defective mitochondria is an essential function
in CIN cells, we would expect to detect mitochondria
being processed by autophagy in CIN cells. To test this
we visualized autophagosomes with mCherry-Atg8 and
mitochondria with mito-GFP (Figure 5). In CIN cells we
observed cytoplasmic accumulations of Atg8, marking
the autophagosomes, and in approximately 20% of cases
(127 of 600) they contained mito-GFP. In some cells the
Oncotarget63915
www.impactjournals.com/oncotarget
mitochondrial network appeared to be interrupted by the
presence of the autophagosomes, or the mito-GFP signal
was compromised where it fell in an autophagosome
relative to the adjacent mitochondria. Control cells did not
have large autophagosomes or any striking co-localization
with mito-GFP. These results suggest that mitochondria
in CIN cells can be removed by autophagy and that this
process is necessary for the survival of CIN cells.
Enhancing autophagic ux reduces the level of
CIN in proliferating cells
It has been reported that defective autophagy
increases the level of CIN in cancer cells due to increased
DNA damage and gene amplication [31]. Conversely, we
would expect treatments that decrease DNA damage to
lower CIN levels. As enhancing autophagic ux reduced
the level of ROS (Figure 4), and we have previously shown
that DNA damage in CIN cells is caused by elevated ROS
[21], we wished to test whether increasing autophagy
could moderate the CIN level. In order to address this
hypothesis, we checked the frequency of aneuploid
metaphases after autophagy enhancement and compared
them with the frequency seen in CIN cells or CIN cells
in which cell death had been blocked by expressing the
apoptosis inhibitor p35. Blocking apoptosis allowed the
retention of many more aneuploid cells in CIN tissue, but
by contrast we found that enhanced autophagic ux could
signicantly reduce the CIN level in a proliferating tissue
(Figure 5).
DISCUSSION
Autophagy can function as a pro-survival protective
pathway in cancer cells to full their metabolic demands
for rapid cell proliferation and to respond to cellular
stresses that may include genomic instability and
metabolic stress [31–34]. Therefore, we assessed the level
of autophagy in cells with induced CIN and found a robust
activation of autophagy (Figure 1). As would be expected
for a tissue with random mitotic defects, not every cell
showed elevated autophagy. The frequency of elevated
Figure 1: Autophagy is activated in tissues with Chromosomal Instability (CIN). CIN was induced in the posterior half of
each wing disc as indicated by the dotted line (marked by the expression of UAS-CD8-GFP) while the rest of each disc was wild type.
(A–C) Lysotracker staining of third instar larval wing discs. Wing discs with CIN induced by either Mad2 depletion (B, engrailed > Gal4,
UAS-CD8-GFP, UAS-mad2
RNAi
) or Rad21 depletion (C, engrailed > Gal4, UAS-CD8-GFP, UAS-rad21
RNAi
UAS-Dicer2) showed increased
lysosome staining relative to the control (A, engrailed > Gal4, UAS-CD8-GFP). Representative discs are shown; the phenotype was
observed all Mad2 depleted discs tested (11) and all Rad21 depleted discs (21) but no control discs (0 from 7). (D–F) The level of mCherry-
Atg8a in third instar larval wing discs. Wing discs with CIN induced by either Mad2 depletion (e, engrailed > Gal4, UAS-CD8-GFP,
UAS-mCherry-Atg8 UAS-mad2
RNAi
) or Rad21 depletion (F, engrailed > Gal4, UAS-CD8-GFP, UAS-mCherry-Atg8, UAS-rad21
RNAi
UAS-
Dicer2) showed increased induction of autophagy relative to the control (D, engrailed > Gal4, UAS-CD8-GFP) as indicated by the level of
mCherry-Atg8a puncta. This phenotype of differing puncta from the wild type half was observed in all Mad2 depleted discs tested (5) and
all Rad21 depleted discs (39) but no control discs (0 from 8).
Oncotarget63916
www.impactjournals.com/oncotarget
autophagy was signicantly higher in the high-CIN cells
generated by Rad21 depletion, which also have a high
frequency of elevated ROS generation, DNA damage
and aneuploidy [22]. This is consistent with data from
human cells showing that increased levels of aneuploidy
correlate with elevated Atg8/LC3 and p62 [13, 14]. In that
work, the effect of ongoing karyotypic variation (CIN) on
autophagy was not tested, possibly because it is difcult
to maintain proliferation in vertebrate CIN cells without
additional changes such as p53 loss [1], which would itself
impact autophagy. In the case of CIN cells, we found that
p62 levels are decreased, indicating effective clearance by
higher autophagic ux, as opposed to stable aneuploids in
which p62 has been reported to accumulate [13]. Because
Figure 2: Blocking autophagy causes redox stress in CIN cells. CellRox staining was used to detect the level of oxidative stress.
The indicated genes were knocked down in the posterior half of each wing disc as indicated by the dotted line while the rest of each
disc was wild type. Knocking down either Atg1 ((A) engrailed > Gal4, UAS-CD8-GFP, UAS-atg1
RNAi
) or Atg18 ((C) engrailed > Gal4,
UAS-CD8-GFP, UAS-atg18
RNAi
) did not give oxidative stress, and the CellRox signal was low or absent in mad2
RNAi
CIN cells
((E) engrailed > Gal4, UAS-CD8-GFP, UAS-mad2
RNAi
). However, when Atg1 ((B), engrailed > Gal4, UAS-CD8-GFP, UAS-mad2
RNAi
,
UAS-atg1
RNAi
) or Atg18 ((D), engrailed > Gal4, UAS-CD8-GFP, UAS-mad2
RNAi
, UAS-atg18
RNAi
) were depleted in CIN cells, an elevated
level of oxidative stress was observed. Depletion of rad21 (F, engrailed > Gal4, UAS-CD8-GFP, UAS-rad21
RNAi
, UAS-Dicer2) shows for
comparison the elevated ROS generated by a high CIN rate. Representative discs are shown; a clear difference from the wild type anterior
half was observed in all discs when Atg1 (11 discs tested) or Atg18 (7 tested) were depleted with Mad2, but none of the control discs
(0 from 10 for Atg1 alone; 0 from 9 for Atg18 alone; 0 from 13 for Mad2 alone).