Current Biology 24, 19–28, January 6, 2014 ª2014 Elsevier Ltd All rights reservedhttp://dx.doi.org/10.1016/j.cub.2013.11.029
Autophagy in Oncogenic K-Ras
Promotes Basal Extrusion
of Epithelial Cells by Degrading S1P
Gloria Slattum,1Yapeng Gu,1Roger Sabbadini,2
and Jody Rosenblatt1,*
1Huntsman Cancer Institute, University of Utah, 2000 Circle of
Hope, Salt Lake City, UT 84112, USA
2Lpath, Inc., 4025 Sorrento Valley Boulevard, San Diego, CA
Background: To maintain a protective barrier, epithelia
extrude cells destined to die by contracting a band of actin
and myosin. Although extrusion can remove cells triggered
to die by apoptotic stimuli, to maintain constant cell numbers,
epithelia extrude live cells, which later die by anoikis. Because
transformed cells may override anoikis and survive after extru-
sion, the direction of extrusion has important consequences
for the extruded cell’s fate. As most cells extrude apically,
they are typically eliminated through the lumen; however, cells
with upregulated survival signals that extrude basally could
potentially invade the underlying tissue and migrate to other
sites in the body.
Results: We found that oncogenic K-Ras cells predominantly
extrude basally, rather than apically, in a cell-autonomous
manner and can survive and proliferate after extrusion.
Expression of K-RasV12downregulates the bioactive lipid
sphingosine 1-phosphate (S1P) and its receptor S1P2, both
of which are required for apical extrusion. Surprisingly, the
S1P biosynthetic pathway is not affected because the S1P
precursor, sphingosine kinase, and the degradative enzymes
S1P lyase and S1PP phosphatase are not significantly altered.
Instead, we found that high levels of autophagy in extruding
RasV12cells leads to S1P degradation. Disruption of auto-
phagy chemically or genetically in K-RasV12cells rescues
S1P localization and apical extrusion.
Conclusions: Oncogenic K-Ras cells downregulate both S1P
andits receptorS1P2topromote basal extrusion.Because live
basally extruding cells can survive and proliferate after extru-
sion, we propose that basal cell extrusion provides a novel
mechanism for cells to exit the epithelium and initiate invasion
into the surrounding tissues.
Epithelia provide a protective barrier for the organs they
encase, yet the cells comprising epithelia are constantly
turning over via cell death and celldivision. Tomaintain a func-
tional barrier, cells destined to die are squeezed out of the
epithelium by a mechanism that we have termed ‘‘cell extru-
sion’’ . In previous work, we have shown that this process
is mediated by the bioactive sphingolipid, sphingosine
1-phosphate (S1P), which is produced by the extruding cell
and binds to a G protein-coupled receptor (S1P2) in the neigh-
boring cells to trigger the guanosine triphosphatase (GTPase)
Rho to form and contract an intercellular actomyosin band .
This contraction squeezes the cell out of the epithelial sheet
while simultaneously closing the gap that may have resulted
from the cell’s exit, thus preserving the epithelial barrier
Although extrusion is activated whenever cells are targeted
to die by apoptotic stimuli, we have found that normally during
homeostasis, extrusion drives cell death [3, 4]. To maintain cell
numberhomeostasis,epithelia extrude livecells atsites where
epithelial cells are most crowded both in vivo and in vitro. Live
extruded cells generally die by anoikis, a type of cell death
caused by the loss of survival signals from cell matrix .
Blocking extrusion leads to epithelial cell masses, supporting
the ideathat cell extrusion promotes death. On the other hand,
override anoikis and enable survival after extrusion [6–8].
Should cells survive after extrusion, the direction they
extrude could have important consequences for their fate.
Typically, epithelia extrude cells apically into the lumen, so
that even transformed live cells would be eliminated into
essentially dead space. Less frequently, however, cells can
extrude basally into the tissue the epithelium encases. Should
transformed cells that can no longer die extrude basally, they
may have the potential to invade the underlying tissue and
initiate metastasis. The direction a cell extrudes depends on
if the actomyosin ring formed in its neighboring cells contracts
out above or below the epithelium, respectively [9, 10]. Where
the actomyosin ring contracts depends, at least in part, upon
microtubule dynamics and the tumor suppressor adenoma-
tous polyposis coli (APC), which target where the ring forms
To investigate the fate of extruded cells that upregulate sur-
vival signals and override anoikis, we expressed a commonly
occurring oncogenic allele of K-Ras (K-RasV12), which renders
it constitutively active, leading to downregulation of apoptosis
and increased cell survival. Increased survival in K-Ras-
transformed cells is thought to be from not only increased
expression of survival signals but also enhanced protective
autophagy [11–14]. The importance of mutations in the
K-RAS proto-oncogene is well established in epithelial-based
carcinogenesis, especially in lung, pancreatic, and colon
Here, we investigated if cells expressing oncogenic K-Ras
could still extrude, and if so, could they survive after extrusion.
We found that K-RasV12cells not only survive and proliferate
after extrusion but, surprisingly, also preferentially extrude
basally, beneath epithelia. Moreover, we found that K-RasV12
that high levels of autophagy in extruding oncogenic K-Ras
cells disrupt S1P production and signaling required for apical
extrusion. S1P normally forms puncta at the interface between
an apically extruding cell and its neighboring cells and is
required only for apical but not basal extrusion. In extruding
oncogenic K-Ras cells, however, S1P is greatly decreased
despite the fact that the pathways required for its synthesis
anddegradation are unaltered.We found thatmarkers ofauto-
phagy, typically upregulated in oncogenic K-Ras cells, are
even more pronounced in extruding K-Ras cells. Blocking
autophagy rescues S1P localization and apical extrusion.
Thus, K-Ras transformation can promote basal extrusion and
enable cells with higher survival and proliferation potential to
exit the epithelia and initiate invasion.
Epithelial Cells Expressing K-RasV12Extrude Basally
in a Cell-Autonomous Manner
Because we previously found that epithelia normally extrude
cells that later die due to loss of survival signaling, or anoikis,
we wondered if transformed cells that block anoikis could
still extrude and survive after extrusion. To test this hypothe-
sis, we expressed oncogenic K-Ras, which upregulates sur-
vival signals that override anoikis. To determine whether cells
expressing oncogenic K-Ras can still extrude, we induced
extrusion in Madin-Darby canine kidney (MDCK) II cell mono-
layers that stably express either yellow fluorescent protein
(YFP)-tagged K-RasWTor GFP-tagged K-RasV12by irradiating
with UV254to induce apoptosis. Surprisingly, we found that
(w75%) basally rather than apically, the direction typically
seen in wild-type (WT) MDCK monolayers or MDCK cells
expressing WT K-Ras (Figures 1A–1C and quantified in Fig-
ure 1D). To demonstrate the direction a cell extrudes in a sin-
gle picture, we overlaid the xy plane of the actin-extruding
ring with the xy plane of the DNA of the extruded cell.
Thus, during apical extrusion in WT and YFP-K-RasWT, the
apoptotic nucleus lies above the plane of the actin ring so
that the nuclei of the surrounding cells are not in focus
(also see xz insets in Figures 1A and 1B and schematics
below). GFP-K-RasV12basally extruding cells, conversely,
contract an actin ring apically above the apoptotic nucleus,
which lies in the same xy plane as the nuclei of the surround-
ing cells (Figure 1C, xz insets, schematic below; Movie S1
available online). Although these extruding cells did not sur-
vive such harsh apoptotic stimulus, we found that cells
extruding without apoptotic stimulus from overgrown mono-
layers also did so predominantly basally (Figure S1).
To determine if K-RasV12expression controls basal extru-
sion within the extruding cell or the cells neighboring it, we
induced extrusion by UV-irradiating monolayers comprised
of a 1:100 ratio of GFP-K-RasV12to WT MDCK cells. GFP-
K-RasV12expression in the extruding cell was sufficient to
drive extrusion basally at rates similar to homogeneously ex-
pressed GFP-K-RasV12(Figure 1E, with quantification in Fig-
ure 1F; Movie S2). Thus, K-RasV12drives basal extrusion
cell autonomously, suggesting that tumor cells in which
this mutation arises could use basal extrusion to invade
beneath the epithelium. Alternatively, cells could be induced
to extrude and invade from a mosaic patch of K-Ras-trans-
Percent extruded cells
Figure 1. Epithelial Cells Expressing K-RasV12Extrude Basally in a Cell-Autonomous Manner
(A–C) Confocal projections and xz cross-sections (below) showing extruding cells (white arrowheads). Control MDCK cells (A) and those expressing YFP-
K-RasWT(B) extrude apically (DNA of extruding cellis out of planewith neighboring cells), whereas those expressingGFP-K-RasV12(C)extrude basally (DNA
of extruding cell is in same plane as neighboring cells).
(D) Quantification of cell extrusion events from five independent experiments with >1,000 extruding cells per cell line.
(E) MDCK GFP-K-RasV12cells mixed with WT MDCK cells 1:100 indicate that K-RasV12cells (green arrow) drive basal extrusion cell autonomously, where
arrowhead indicates basally extruded K-RasV12cell, and asterisks show WT cells. Pink and blue arrows indicate actin ring and DNA, respectively, and
dashed line where the xz section was taken.
(F) Quantification of three independent experiments, where n = 600 extruding cells.
Scale bars represent 10 mm. Error bars indicate the SEM. p < 0.001 by Student’s t test comparing control to RasV12. See also Figure S1 and Movies S1
Current Biology Vol 24 No 1
S1P Signals Apical but Not Basal Extrusion and Is
Misregulated in K-RasV12-Expressing Cells
To determine how oncogenic K-Ras induces basal extrusion,
we investigated if the cytoskeletal mechanism or the
signaling controlling extrusion was altered in K-RasV12-
expressing monolayers. We previously found that the direc-
tion a cell extrudes depends upon where the actin and
myosin ring in the surrounding cells contracts [9, 10]. If it con-
tracts at the bottom of the cell, the cell exits apically, whereas
if it contracts apically, the cell extrudes basally. The site of
because altering microtubule dynamics with either taxol or
nocodazole or disruption of a protein that regulates microtu-
bules, APC, shifts the direction of extrusion from apical to
basal [9, 10]. Therefore, we examined if microtubules were
altered in K-RasV12-extruding cells. Unlike other situations
where the direction of extrusion was linked to misregulation
of microtubules, we found no obvious changes in microtu-
bules in K-RasV12-extruding cells or cells surrounding them
(Figures 2A and 2D). Next, we examined whether K-RasV12
expression could alter the intrinsic polarity of cells to alter
extrusion direction by immunostaining extruding monolayers
for ZO-1, an apical marker, and b-catenin, a component of
adherens-type junctions, used here as a basal marker. How-
ever, both of these polarity markers were also unaltered in
RasV12-compared to WT-extruding cells, suggesting that
K-RasV12expression does not affect the direction of extru-
sion by disrupting the intrinsic polarity (Figures 2B–2F).
Therefore, we examined if K-RasV12acted cell autonomously
to alter extrusion signaling.
Figure 2. K-RasV12Does Not Affect Direction of Extrusion by Altering Microtubule Dynamics or Cell Intrinsic Polarity
Confocal projections andxz cross-sections (below) of control MDCK monolayers apically extruding (A–C) andK-RasV12monolayers basally extruding (D–F).
Whitearrowspointtothemicrotubulespointingbasallyduring apicalextrusion (A)andapicallyduringbasalextrusion(D),whereasterisksindicateextruding
cells location. White arrows show the apical marker ZO-1 (B and E) and the basal marker b-catenin (Cat) (E and F) are both localized correctly. White arrow-
head indicates extruding cell, pink and blue arrows indicate actin ring and DNA, respectively, and dashed line where the xz section was taken. Scale bars
represent 10 mm.
Oncogenic K-Ras Drives Epithelial Basal Extrusion
We next tested if extrusion signaling through S1P and the
S1P2 receptor were altered in oncogenic K-Ras-extruding
cells. We previously found that a cell destined to extrude emits
S1P, which binds the S1P2receptor in neighboring cells to
activate Rho-mediated assembly and contraction of an inter-
cellular actomyosin ring that squeezes it out [1, 2]. Because
our previous data showed that disrupting S1P2disrupts extru-
sion, we assessed S1P2levels in K-RasV12cells and found that
they had w3-fold lower levels than WT MDCK cells by immu-
noblotting (Figure 3A). S1P is normally amplified only in
extruding cells; therefore, we immunostained for this lipid us-
ing a specific monoclonal antibody to S1P [2, 15] in extruding
monolayers. Whereas S1P forms prominent puncta at the
interface of extruding/neighboring cells along the basolateral
surface in control apically extruding cells (Figure 3B), S1P
was completely absent in basally extruding K-RasV12cells
induced to extrude with UV (Figure 3E) or under homeostatic
conditions (Figure S1). By quantifying the number of extruding
cells that have S1P, we found that 89% of apically extruding
WT cells had S1P, whereas only 7% of K-RasV12basally
extruding cells had S1P, where n = 100 for each cell type. In
the minor populations of K-RasV12cells that extruded apically,
S1P was diffuse and mislocalized (Figure 3C). Importantly, the
minor population of control cells that basally extrudes also
lacked S1P puncta, suggesting that the S1P-S1P2signaling
pathway signals apical but not basal extrusion (Figure 3D).
To test if the S1P pathway controls only apical and not basal
extrusion, we blocked extrusion inMDCK controlcells withthe
Figure 3. S1P and the S1P Receptor 2 Signal Apical but Not Basal Extrusion and Are Misregulated in K-RasV12-Expressing Cells
(A) Immunoblot analysis comparing the expression levels of S1P2in control MDCK cells, and MDCK cells expressing K-RasWTor K-RasV12, where lamin B2
serves as a loading control, and quantified ratio compared to control is below.
(B–E) Confocal projections and xz cross-sections (below) showing an apically extruding control MDCK produces high S1P levels (B), whereas an apically
extruding K-RasV12cell does not (C). Basally extruding control (D) and K-RasV12(E) MDCK cells do not produce any S1P. White arrowheads point to dying,
extruding cells at live-dead cell interfaces, pink arrows indicate actin ring, blue arrows show DNA, and dashed line where the xz section was taken. Scale
bars represent 10 mm.
(F–G0) Apical extrusion is disrupted in control MDCK cells treated with S1P2antagonist (JTE-013), where (F) shows comparative rates, and (G) shows total
rates from three independent experiments analyzing 300 extrusion events for each experiment. Error bars indicate the SEM. p < 0.001 by Student’s t test
comparing control versus each treatment. (G0) shows examples of blocked extrusion, scored by the presence of a faint actin ring that does not contract
around a late-staged caspase-3-positive dying cell, apical extrusion, and basal extrusion. Arrowheads show extruding cell.
Current Biology Vol 24 No 1
S1P2antagonist, JTE-013. JTE-013 increased the percentage
of blocked extrusions, scored by the presence of a faint actin
ring that does not contract around a late-staged caspase-3-
positive dying cell, and seen previously in . However, it
increased the relative percentage of basal extrusions (Fig-
ure 3F). Although blocking S1P2signaling shifts the direction
of extrusion from apical to basal, it actually does so by block-
ing apical extrusion without affecting basal extrusion, as indi-
cated when total numbers of each type of extrusion were
quantified (Figure 3G). When total rates of apical versus basal
extrusion were similarly quantified in K-RasV12-extruding
monolayers, instead of relative percentages, we found that
they mirrored the effects of blocking S1P-S1P2 signaling
pathway on extrusion.
K-RasV12-Extruding Cells Express High Levels of the
Autophagy Marker LC3A/LC3B
To determine how S1P is downregulated in K-RasV12-
extruding cells, we analyzed the S1P synthetic pathway in
WT versus oncogenic K-Ras monolayers induced to extrude
with UV254. We found that sphingosine kinase 1 (SphK1), the
enzyme that converts sphingosine to S1P, was 1.5-fold upre-
gulated in K-RasV12-expressing cells (Figure 4A). Furthermore,
enzymes that downregulate S1P, by converting it back to
sphingosine or degrading it into phosphoethanolamine and
hexadecanal, S1P phosphatase 1 (S1PP), and S1P lyase,
respectively, were not significantly altered in K-RasV12cells
(Figure 4A, where both are 0.9-fold that of WT cells). Immuno-
fluorescence also confirmed that S1P synthetic and degrada-
tive enzymes were not significantly altered in K-RasV12-
extruding cells compared to WT-extruding MDCK cells (data
not shown). Because neither S1P synthetic nor degradative
pathways were significantly altered in K-RasV12-expressing
cells, we next considered if S1P depletion in extruding
K-RasV12cells was due to an alternative degradation pathway.
α LC3 A/B-I
α LC3 A/B-II
α S1P Lyase
F-Actin LC3 A/BDNAMerge
α Lamin B2
Figure 4. K-RasV12-Extruding Cells Express High
Levels of the Autophagy Marker LC3A/LC3B
(A) SIP synthetic pathway and immunoblots
comparing the expression levels of key regula-
tors of S1P in control versus K-RasV12cell lysates
where tubulin and lamin serve as loading
(B) LC3A/LC3B immunoblots of MDCK, K-RasWT,
and K-RasV12cells, where a-tubulin serves as a
loading control, LC3A/LC3B-I is the cytosolic
form, whereas LC3A/LC3B-II is associated with
autophagosomes. Ratios compared to the con-
trol are below.
(C) Confocal projections of control MDCK cells
and K-RasV12basal-extruding cell; white arrow-
heads show increased LC3 puncta in extruding
cell. Scale bars represent 10 mm.
See also Figure S2 and Movies S3 and S4.
Because other studies have found
that both oncogenic K-Ras expression
[11–14] and SK1 levels  can promote
autophagy, we investigated if auto-
phagy may act to degrade S1P during
K-RasV12cell extrusion. Autophagy is
where components of the cytoplasm
and organelles are sequestered into au-
tophagosomes, which then fuse with
lysosomes to become degraded. Although cells use auto-
phagy to conserve energy and resources during starvation,
oncogenic Ras cells have been found to be addicted to auto-
phagy to increase their survival rates [11–14]. Although auto-
phagy has not been previously reported to degrade S1P, the
fact that other membranous compartments of the cell can be
degraded by autophagy suggests that S1P could also be
degraded in this way. Additionally, because several sphingoli-
pids have been found to promote autophagy [17, 18], it
seemed possible that a sphingolipid also could be regulated
by autophagy. Autophagosomes form by converting the auto-
phagymarkerMicrotubule-Associated Protein1-Light-Chain 3
(LC3) forms A and B from a cytosolic form (LC3-I) to a phos-
(LC3-II). Thus, autophagic activity can be identified by LC3
puncta formation using immunofluorescence and LC3-II by a
we found that compared to control cells, K-RasV12cells had
elevated levels of the autophagy marker LC3-II (Figure 4B).
LC3A/LC3B was even more enriched in K-RasV12-extruding
cells compared to neighboring K-RasV12cells when viewed
by immunofluorescence (Figure 4C). Quantification of LC3A/
LC3B fluorescent intensity showed that K-RasV12-extruding
cells had consistently higher numbers of puncta (w2,300
mean fluorescent intensity per 20 cells) compared to their non-
extruding neighboring cells (w800 mean fluorescent intensity
per 20 cells). The LC3A/LC3B increase in extruding cells may
be similar to that seen in extruding cells in Drosophila amnio-
serosa prior to extrusion . Extruding K-RasV12may have
higher levels of autophagy than either WT-extruding or unex-
truding K-RasV12cells due to the fact that both K-RasV12
signaling and extrusion signaling promote autophagy (as
seen in Figure 4B). Our findings that autophagy is especially
prominent in K-RasV12cells targeted to extrude suggests a
mechanism for how these cells downregulate S1P to promote
Oncogenic K-Ras Drives Epithelial Basal Extrusion
basal extrusion. To determine if inducing autophagy in control
MDCK cells alone could switch the direction of extrusion from
predominantly apical to basal, we treated MDCK monolayers
with Torin-2 (a potent ATP-competitive inhibitor of mammalian
target of rapamycin [mTOR]) that induces autophagy. We
found that inducing autophagy in otherwise WT cells was
sufficient to cause cells to extrude basally (Figure S2).
Blocking Autophagy in K-RasV12Cells Rescues S1P
Localization and Apical Extrusion
To test if the increased autophagy in K-RasV12cells disrupts
S1P-mediated apical extrusion, we blocked autophagy to
assess if it would rescue both S1P and apical extrusion. We
pretreated control and K-RasV12monolayers with commonly
used small molecule inhibitors of autophagy, induced extru-
sion, and assayed for both S1P expression (Figures 5A and
5B) and the direction cells extrude (Figure 5C). By blocking
autophagy with the phosphoinositide-3 kinase inhibitor wort-
mannin, which blocks autophagosome formation , or with
bafilomycin A1 or chloroquine , which both block auto-
phagosome degradation by preventing fusion with the lyso-
some, we found that inhibition of autophagy increased the
percentage of cells undergoing apical extrusion compared to
untreated K-RasV12cells (Figures 5A and 5B and quantified
in Figure 5C). We expressed the tandem mCherry-EGFP-
LC3B reporter in oncogenic K-Ras cells to confirm that auto-
phagic flux to the lysosome was occurring in basally extruding
cells. This reporter indicated that LC3 becomes targeted to
lysosomes, inactivating GFP and turning red when a
yellow when fusion to the lysosome is blocked with chloro-
quine and the cell extrudes apically (Movie S4; Figure S3B).
K-RasV12cells (Figure 5B). On the other hand, blocking auto-
phagy did not affect S1P2receptor levels, as measured by
immunoblotting or immunostaining (Figure S4), suggesting
that enough S1P2remains in the K-RasV12to rescue apical
extrusion if S1P levels are increased.
Because these inhibitors can also affect other cellular func-
tions, we confirmed if blocking autophagy could rescue S1P
accumulation and apical extrusion in K-RasV12by knocking
down two essential autophagy genes: Atg7 and Atg5. Simi-
larly, Atg7 or Atg5 knockdown rescued both S1P puncta
formation (Figure 5E) and apical extrusion (Figure 5F) in
K-RasV12-extruding cells (Figure S5). The decrease in auto-
phagy flux by diminished expression of LC3-II and accumula-
tion of the p62/SQSTM1-scaffolding protein that is degraded
by autophagy confirmed that Atg knockdown blocked auto-
phagy (Figures 5D and S5). Therefore, K-RasV12cells drive
cells to extrude basally by targeting the proextrusion signal
S1P for degradation, and both S1P and apical extrusion can
be rescued by simply blocking autophagy.
Although these experiments yielded the surprising finding
that oncogenic K-Ras drives extrusion basally rather than
cells could survive and proliferate after extrusion for several
reasons. First, to increase the rates of extrusion, we typically
induce extrusion with the strong apoptotic stimulus, UV254,
ment) extrude predominantly basally (Figure S1) and WT cells
extrude predominantly apically at the same ratios as when
apoptosis is induced, the extrusion rates are greatly reduced,
making them harder to score. Second, if we allow cells to
naturally extrude and accumulate over time, it is extremely
hard to identify quantifiable numbers of live extruded cells in
two-dimensional (2D) cultures: live apically extruded cells are
lost within the medium, and live basally extruded cells may
intercalate into the normal surrounding cells within the mono-
layer. Moreover, live basally extruded K-RasV12cells would be
far more likely to die after basal extrusion in cell culture than in
real tissue because they could become trapped between the
cover glass and the monolayer, which would prevent their
access to growth factors. In vivo, live basally extruded cells
might still have access to growth factors in the matrix and un-
derlying stroma. Therefore, we decided to investigate the fate
of basally extruded cells in 3D cultures.
Basally Extruded K-RasV12Cells Survive and Proliferate
in 3D Cultures
To test if GFP-K-RasV12cells can survive after extrusion, we
, where we would be able to score viability after extrusion.
MDCK cells grown in Matrigel form a clonal acinar structure
containing a hollow lumen that faces the apical surface (Fig-
ure 6A). Cysts comprised of oncogenic K-Ras MDCK cells,
however, extruded live cells that persisted and often divided
(Figure 6B). Eventually, these K-RasV12cells developed
lumens filled with apoptotic cells, live cells, and smaller cysts,
similar to results found with ErbB2 expression  (Figure S6).
Additionally, approximately 20%of K-RasV12MDCK cysts also
formed mini cysts attached to the outside of the main cyst
To determine the frequency of survival of the basally
extruded K-RasV12compared to WT control cells, and those
expressing WT K-Ras in 3D cultures, we immunostained cysts
grown under homeostatic conditions for 3 days for the
apoptotic marker antiactive caspase-3. We found that 67%
of K-RasV12cells extrude live cells, whereas WT MDCK cells
or MDCK cells expressing WT K-Ras only extrude 2% live
cells, where the remainder are apoptotic (Figure S6). Attached
basal mini cysts could arise either from other live migrating
cells attaching to cysts or from proliferation of live basally
extruded cells to form new smaller cysts. To test between
these two models, we imaged live-cell extrusion by phase
time-lapse video microscopy of control MDCK or GFP-K-
RasV12cysts grown under homeostatic conditions. Whereas
cells from MDCK cysts extruded into the lumen, underwent
apoptosis, and were engulfed by neighboring cells (Figure 6D;
Movie S5, left), apically extruded K-RasV12cells were not en-
extruded K-RasV12cells either migrated away from the cyst or
proliferated into a mini cyst (Figure 6E; Movie S5, right). Onco-
genic K-Ras cells expressing fluorescently tagged myosin
light chain (MLC) or actin clearly show that cells escaping
the cyst do so by extrusion, as characterized by contraction
of an actin and myosin II ring (Figure 6F; Movie S6). These
extrude basally and, instead of dying, proliferate, suggesting
that basal extrusion could provide a mechanism for invading
matrix and tissue underlying epithelia.
We have found that epithelial cells stably expressing onco-
genic K-Ras extrude predominantly basally from both epithe-
lial monolayers and 3D cysts in culture. Basal extrusion is
cell autonomous because single K-RasV12cells surrounded
Current Biology Vol 24 No 1
ATG 5/7 KD
Figure 5. Blocking Autophagy in K-RasV12Cells Rescues S1P Localization and Apical Extrusion
(Aand B) Confocal section andxz cross-sections (below) of a K-RasV12monolayer that extrudes primarily basally and lacks SIP (A) butextrudes apically and
accumulates S1P puncta with 30 mM chloroquine, which blocks autophagy (B), where white arrowheads indicate extruding cells at the live-dead cell inter-
faces, pink arrows indicate actin ring, blue arrows show DNA, and dashed line the xz section. Scale bars represent 10 mm.
(C)Quantifications ofcellextrusion directionwiththreeautophagy inhibitors fromfourindependentexperimentswhere>1,000extrusionswereanalyzedper
treatment. Error bars indicate the SEM.
(D) Immunoblots show siRNA-mediated knockdown of Atg7 and LC3-II reduction and p62 upregulation. Ratios compared to the control (Ctr.) are below.
Atg7 knockdown also rescues apical extrusion and S1P. Tub, tubulin.
(E and F) Quantification from <500 total extrusion events from three siRNA experiments. p < 0.001 by Student’s t test comparing control versus each treat-
ment. The scale bar represents 10 mm.
(G) Model for how apical extrusion is misregulated in K-RasV12-expressing cells. S1P normally binds S1P2in surrounding cells to contract basally and
squeeze the cell out. However, in K-RasV12-extruding cells, S1P is targeted for degradation by increased autophagy. Without S1P, basolateral contraction
in neighboring cells is not activated, and instead, only apical contraction of the dying cell occurs, driving extrusion basally. S1P (blue), autophagosome
(yellow), lysosome (green), extruding cell actomyosin contraction (red), and neighboring cell actomyosin contraction (maroon). KD, knockdown.
See also Figures S3–S5.
Oncogenic K-Ras Drives Epithelial Basal Extrusion
by WT cells also drive extrusion basally. Importantly, we found
that although oncogenic K-Ras cells are typically autophagic,
autophagy is even higher when these cells extrude, which dis-
rupts S1P critical for apical extrusion. The normal S1P synthe-
sis pathway is not altered in K-RasV12cells, however, because
blocking autophagy either by knockdown of Atg5 or Atg7 or by
chemical inhibitors rescues S1P puncta formation and apical
extrusion. Together, our results suggest a model where
K-RasV12cells targeted for normal extrusion produce S1P,
which due to high levels of autophagy, becomes degraded in
the lysosomes (Figure 5G). When S1P is absent, it can no
longer signal neighboring cells to contract basally and extrude
the cell apically. Instead, the basally extruding K-RasV12cell
contracts at the apical junctions where actin and myosin II
already exist, driving the cell out basally. In WT cells or when
autophagy is blocked, S1P produced in the cell targeted for
extrusion activates S1P2in the neighboring cells to contract
basally and shove the cell out apically.
Although we previously identified that extrusion requires the
S1P-S1P2-Rho pathway, here, we find that this signaling axis
controls only apical but not basal extrusion. This was not
directly apparent because most cells extrude apically. Our
finding that K-RasV12cells switch the extrusion direction
from apical to basal by degrading S1P and downregulating
S1P2also demonstrated that S1P signaling is only required
for apical extrusion. Rescuing S1P by blocking autophagy res-
cues apical extrusion in K-RasV12cells. K-RasV12cells may
affect extrusion also by downregulating the S1P2receptor by
an autophagy-independent mechanism. However, the fact
thatblocking autophagy issufficienttorescueapicalextrusion
suggests that there is enough S1P2present in K-Ras-trans-
formed cells to enable apical extrusion when S1P levels are
Our results showing that oncogenic K-Ras cells extrude
basally contrast somewhat to those found by Hogan et al.
that show H-Ras transformation causes apical extrusion .
We believe that the main difference in our results lies in the
length of time K-RasV12is expressed. In experiments like
ours where K-RasV12is expressed constitutively, Liu et al.
found that a single oncogenic H-Ras cell within a cyst either
extruded basally or migrated , suggesting that basal
extrusion may be a common feature of expressing any form
of oncogenic Ras long term. Hogan et al. , instead, use
an inducible expression system to activate H-RasV12acutely.
surrounding WT cells within the monolayer by what appears to
be differential adhesion, rather than the extrusion process we
have defined . We have found similar results if we express
K-RasV12acutely (datanotshown), suggesting thatexpressing
any isoform of activated Ras transiently will cause delamina-
tion apically. Although the ultimate effect is the same—that
these cells delaminate from the epithelium—they do not do
so through the canonical extrusion pathway that requires
S1P-mediated activation of an intercellular actomyosin cable
Figure 6. Basally Extruded K-RasV12Cells Survive and Proliferate in 3D Cultures
(A and B) Cysts from MDCK and K-RasV12-MDCK cells cultured in Matrigel for 3 days. Confocal sections show that WT MDCK cysts have clear lumens (A),
whereas MDCK K-RasV12cysts extrude live cells basally (B). Yellow arrowhead, filled lumen; red arrows, live basal extruded cells.
(C) Quantification of phenotypes over time from three independent experiments (n > 100 cysts per experiment). Error bars indicate the SEM. p < 0.001 by
Student’s t test comparing control to RasV12.
(D and E) Stills from movies of cysts (in Movie S5) filmed at day 3 for 24 hr without UV treatment. A control MDCK cyst extrudes cells apically into the lumen
(yellow arrow) (D), whereas a K-RasV12cyst extrudes cells basally into the matrix, which later proliferate (E) (red arrows).
(F)StillsfromamovieofaUV-treatedK-RasV12cystexpressing MLC-redfluorescentprotein(RFP) (inMovie S6)show thatbasalextrusion of acelloccursby
contraction of a myosin ring.
Scale bars represent 20 mm. Time is indicated as hr:min. See also Figure S6 and Movies S5 and S6.
Current Biology Vol 24 No 1
that contracts in surrounding cells. Therefore, alternative
downstream signaling may lead to apical delamination versus
basal extrusion depending on the length of oncogenic K-Ras
Although we focused mainly on the mechanism that drives
K-RasV12epithelial cells to extrude basally when targeted for
cell death, it is important to note that live cells also extrude
basally when no apoptotic stimulus is present. Because most
cells extrude live prior to dying, the direction live cells extrude
has important consequences for their later fate. Apically
extruded cells will essentially be eliminated through the lumen
or outside the body, even if transformed by K-RasV12, whereas
basally extruded transformed cells could potentially invade
and initiate metastasis. Thus, apical elimination of cells that
become mutated with K-RasV12could act as a self-regulating
mechanism to rid the body of transformed precancerous cells.
On the other hand, oncogenic K-Ras-driven tumors, such as
lung or pancreatic carcinomas, serve as testament to the fact
that this mechanism sometimes fails. In these cases, shifting
extrusion from apical to basal coupled with increased cell sur-
vival could act to promote tumor progression and metastasis.
This newly defined potential mechanism for invasion, which is
enhanced specifically in K-Ras-transformed cells, could
explain why K-Ras-driven tumors are so metastatic and
deadly. Our results suggest that simply blocking autophagy
with chloroquine, a drug already in clinical use [28, 29], could
prevent invasion of K-RasV12-driven tumors and better target
them for apoptosis. Therefore, future studies will need to
determine if basal extrusion of K-Ras cells can promote their
invasion and metastasis in vivo and, if so, whether inhibition
of autophagy can reverse this invasive phenotype.
Control MDCK II cells and cells expressing oncogenic K-Ras (gift from
K. Matlin, University of Chicago, Chicago, IL) were cultured in Dulbecco’s
modified Eagle’s medium (DMEM) high glucose with 5% FBS (all from
HyClone) and 100 mg/ml penicillin/streptomycin (Invitrogen) at 5% CO2,
37?C. MDCK II cells stably expressing GFP-oncogenic or YFP-WT K-Ras
were generated by transfecting pEGFP C3 containing K-RasV12or pEYFP
C1 containing K-RasWT(gift from Channing J. Der, University of North
Carolina, Chapel Hill, NC), respectively, by using Lipofectamine 2000 (Invi-
trogen) according to manufacturer’s instructions. To monitor autophagy
facturer Lonza Biologics) with pBABE-puro mCherry-EGFP-LC3B tandem
biosensor (Addgene plasmid 22418 deposited by Jayanta Debnath).
MDCK cysts were generated as previously described in  and the
Supplemental Experimental Procedures. For live fluorescent imaging of
MLC (gift from Thomas Marshall, University of Utah, Salt Lake City, UT) or
mApple-actin-7 (a gift from Michael W. Davidson, Florida State University,
Tallahassee, FL), oncogenic K-Ras cells were transduced or transfected,
respectively, before culturing in Matrigel.
UV and Drug Treatment
MDCK II cells grown to confluence on glass coverslips were exposed to
1,200 uJ/cm2UV254for 43 s using a Spectrolinker (Spectroline) to induce
apoptotic extrusion and incubated for 2 hr before fixation. For complete cell
drug treatment, please refer to the Supplemental Experimental Procedures.
Cells were either fixed with ice-cold 100% methanol for 45 s or 4% parafor-
maldehyde in PBS at 37?C for 20 min. For protocols and antibodies, please
refer to the Supplemental Experimental Procedures.
siRNAs against the canine ATG7 and ATG5 sequences were synthesized at
the University of Utah Oligo and Peptide Synthesis Core. For complete se-
quences, refer to the Supplemental Experimental Procedures.
Image, Video Acquisition, and Immunoblot Analysis
Microscopy and immunoblots were performed according to standard pro-
tocols. For details, please see the Supplemental Experimental Procedures.
All statistical analyses were performed using Prism 5 (GraphPad) and
employing unpaired t test.
Supplemental Information includes Supplemental Experimental Proce-
dures, six figures, and six movies and can be found with this article online
We thank Katie Ullman and Thomas Marshall for helpful comments on our
manuscript. We also thank Karl Matlin for his MDCK cell lines and Jayanta
Debnath for first suggesting that we investigate autophagy. This work was
supported by a National Institute of Health Innovator Award numbers DP2
OD002056-01 and 1R01GM102169-01 to J.R., a NIH Developmental Biology
Training Grant 5T32 HDO7491 to G.S., and P30 CA042014 awarded to the
Huntsman Cancer Institute for core facilities. R.A.S. has stock options in
Received: January 11, 2013
Revised: October 11, 2013
Accepted: November 14, 2013
Published: December 19, 2013
1. Rosenblatt, J., Raff, M.C., and Cramer, L.P. (2001). An epithelial cell
destined for apoptosis signals its neighbors to extrude it by an actin-
and myosin-dependent mechanism. Curr. Biol. 11, 1847–1857.
2. Gu, Y., Forostyan, T., Sabbadini, R., and Rosenblatt, J. (2011). Epithelial
cell extrusion requires the sphingosine-1-phosphate receptor 2
pathway. J. Cell Biol. 193, 667–676.
3. Eisenhoffer, G.T., Loftus, P.D., Yoshigi, M., Otsuna, H., Chien, C.B.,
Morcos, P.A., and Rosenblatt, J. (2012). Crowding induces live cell
extrusion to maintain homeostatic cell numbers in epithelia. Nature
4. Marinari, E., Mehonic, A., Curran, S., Gale, J., Duke, T., and Baum, B.
(2012). Live-cell delamination counterbalances epithelial growth to limit
tissue overcrowding. Nature 484, 542–545.
5. Frisch, S.M., Vuori, K., Ruoslahti, E., and Chan-Hui, P.Y. (1996). Control
of adhesion-dependent cell survival by focal adhesion kinase. J. Cell
Biol. 134, 793–799.
6. Douma, S., Van Laar, T., Zevenhoven, J., Meuwissen, R., Van Garderen,
E., and Peeper, D.S. (2004). Suppression of anoikis and induction of
metastasis by the neurotrophic receptor TrkB. Nature 430, 1034–1039.
7. Horbinski, C., Mojesky, C., and Kyprianou, N. (2010). Live free or die:
tales of homeless (cells) in cancer. Am. J. Pathol. 177, 1044–1052.
8. Taddei, M.L., Giannoni, E., Fiaschi, T., and Chiarugi, P. (2012). Anoikis:
an emerging hallmark in health and diseases. J. Pathol. 226, 380–393.
9. Marshall,T.W.,Lloyd,I.E.,Delalande,J.M.,Na ¨thke,I.,andRosenblatt,J.
(2011). The tumor suppressor adenomatous polyposis coli controls the
direction in which a cell extrudes from an epithelium. Mol. Biol. Cell 22,
10. Slattum, G., McGee, K.M., and Rosenblatt, J. (2009). P115 RhoGEF and
microtubules decide the direction apoptotic cells extrude from an
epithelium. J. Cell Biol. 186, 693–702.
11. Mathew, R., and White, E. (2011). Autophagy, stress, and cancer meta-
bolism: what doesn’t kill you makes you stronger. Cold Spring Harb.
Symp. Quant. Biol. 76, 389–396.
12. Guo, J.Y., Chen, H.Y., Mathew, R., Fan, J., Strohecker, A.M., Karsli-
Uzunbas, G., Kamphorst, J.J., Chen, G., Lemons, J.M., Karantza, V.,
et al. (2011). Activated Ras requires autophagy to maintain oxidative
metabolism and tumorigenesis. Genes Dev. 25, 460–470.
13. Mirzoeva, O.K., Hann, B., Hom, Y.K., Debnath, J., Aftab, D., Shokat, K.,
and Korn, W.M. (2011). Autophagy suppression promotes apoptotic cell
death in response to inhibition of the PI3K-mTOR pathway in pancreatic
adenocarcinoma. J. Mol. Med. 89, 877–889.
Oncogenic K-Ras Drives Epithelial Basal Extrusion
14. Lock, R., and Debnath, J. (2011). Ras, autophagy and glycolysis. Cell Download full-text
Cycle 10, 1516–1517.
15. Swaney, J.S., Moreno, K.M., Gentile, A.M., Sabbadini, R.A., and Stoller,
G.L. (2008). Sphingosine-1-phosphate (S1P) is a novel fibrotic mediator
in the eye. Exp. Eye Res. 87, 367–375.
16. Lavieu, G., Scarlatti, F., Sala, G., Carpentier, S., Levade, T., Ghidoni, R.,
Botti, J., and Codogno, P. (2006). Regulation of autophagy by sphingo-
sine kinase 1 and its role in cell survival during nutrient starvation.
J. Biol. Chem. 281, 8518–8527.
17. Young, M.M., Kester, M., and Wang, H.G. (2013). Sphingolipids: regula-
tors of crosstalk between apoptosis and autophagy. J. Lipid Res. 54,
18. Bedia, C., Levade, T., and Codogno, P. (2011). Regulation of autophagy
by sphingolipids. Anticancer. Agents Med. Chem. 11, 844–853.
19. Tanida, I., Ueno, T., and Kominami, E. (2008). LC3 and autophagy.
Methods Mol. Biol. 445, 77–88.
20. Cormier, O., Mohseni, N., Voytyuk, I., and Reed, B.H. (2012). Autophagy
can promote but is not required for epithelial cell extrusion in the amnio-
serosa of the Drosophila embryo. Autophagy 8, 252–264.
21. Blommaart, E.F., Krause, U., Schellens, J.P., Vreeling-Sindela ´rova ´, H.,
and Meijer, A.J. (1997). The phosphatidylinositol 3-kinase inhibitors
wortmannin and LY294002 inhibit autophagy in isolated rat hepato-
cytes. Eur. J. Biochem. 243, 240–246.
22. Yoshimori, T., Yamamoto, A., Moriyama, Y., Futai, M., and Tashiro, Y.
(1991). Bafilomycin A1, a specific inhibitor of vacuolar-type H(+)-
ATPase, inhibits acidification and protein degradation in lysosomes of
cultured cells. J. Biol. Chem. 266, 17707–17712.
23. Paludan, C., Schmid, D., Landthaler, M., Vockerodt, M., Kube, D.,
Tuschl, T., and Mu ¨nz, C. (2005). Endogenous MHC class II processing
of a viral nuclear antigen after autophagy. Science 307, 593–596.
24. Zegers, M.M., O’Brien, L.E., Yu, W., Datta, A., and Mostov, K.E. (2003).
Epithelial polarity and tubulogenesis in vitro. Trends Cell Biol. 13,
25. Leung, C.T., and Brugge, J.S. (2012). Outgrowth of single oncogene-
expressing cells from suppressive epithelial environments. Nature
26. Hogan, C., Dupre ´-Crochet, S., Norman, M., Kajita, M., Zimmermann, C.,
Pelling, A.E., Piddini, E., Baena-Lo ´pez, L.A., Vincent, J.P., Itoh, Y., et al.
(2009). Characterization of the interface between normal and trans-
formed epithelial cells. Nat. Cell Biol. 11, 460–467.
27. Liu, J.S., Farlow, J.T., Paulson, A.K., Labarge, M.A., and Gartner, Z.J.
(2012). Programmed cell-to-cell variability in Ras activity triggers emer-
gent behaviors during mammary epithelial morphogenesis. Cell Rep 2,
28. Krishna, S.,andWhite, N.J. (1996).Pharmacokinetics of quinine, chloro-
quine and amodiaquine. Clinical implications. Clin. Pharmacokinet. 30,
29. Solomon, V.R., and Lee, H. (2009). Chloroquine and its analogs: a new
promise of an old drug for effective and safe cancer therapies. Eur. J.
Pharmacol. 625, 220–233.
Current Biology Vol 24 No 1