Polarized deposition of basement membrane proteins
depends on Phosphatidylinositol synthase and the
levels of Phosphatidylinositol 4,5-bisphosphate
Olivier Devergne, Karen Tsung, Gail Barcelo, and Trudi Schüpbach1
Howard Hughes Medical Institute and Department of Molecular Biology, Princeton University, Princeton, NJ 08544
Contributed by Trudi Schüpbach, April 22, 2014 (sent for review January 23, 2014)
The basement membrane (BM), a specialized sheet of the extra-
cellular matrix contacting the basal side of epithelial tissues, plays
an important role in the control of the polarized structure of epithelial
cells. However, little is known about how BM proteins themselves
achieve a polarized distribution. Here, we identify phosphatidylino-
sitol 4,5-bisphosphate (PIP2) as a critical regulator of the polarized
secretion of BM proteins. A decrease of PIP2 levels, in particular
through mutations in Phosphatidylinositol synthase (Pis) and other
members of the phosphoinositide pathway, leads to the aberrant
accumulation of BM components at the apical side of the cell with-
out primarily affecting the distribution of apical and basolateral
polarity proteins. In addition, PIP2 controls the apical and lateral
localization of Crag (Calmodulin-binding protein related to a Rab3
GDP/GTP exchange protein), a factor specifically required to prevent
aberrant apical secretion of BM. We propose that PIP2, through the
control of Crag’s subcellular localization, restricts the secretion of
BM proteins to the basal side.
embryonic and adult organisms. Epithelia exhibit a profound
apical–basal polarity that is manifested in the cytoplasmic and
surface organization of individual cells (1–3). Loss of apical–
basal cell polarity is often associated with carcinoma progression
and tumor metastasis (4, 5). The establishment and maintenance
of cell polarity relies on the transport of newly synthesized and
recycled proteins to their correct destinations (6, 7). The lipid
composition of the transport vesicles and of the plasma mem-
brane is crucial for the establishment and maintenance of cell
polarity (6–9). In particular, in 3D in vitro culture of Madin–Darby
canine kidney (MDCK) cells, phosphatidylinositol 4,5-bisphosphate
(PIP2) and phosphatidylinositol 3,4,5-trisphosphate (PIP3), two
phosphoinositides (PtdIns) have been shown to play critical roles
in polarized vesicle trafficking by mediating the recruitment of
proteins to these different domains (10, 11).
To set up a correct cell polarity, membrane asymmetry needs
to be established. In cell culture, and likely during development
of many tissues in multicellular organisms, this process is
achieved by two external cues: one provided by the adjacent cells
via cadherin-dependent adhesion and the other by interaction
with the basement membrane (BM), a specialized sheet of the
ECM secreted basally by the epithelial cells (12, 13). The main
components of the BM are secreted glycoproteins, such as colla-
gen IV (Coll IV), laminin, and the heparan sulfate proteoglycan
perlecan (Pcan) (14), which interact with different membrane
receptors, including integrin and dystroglycan (14, 15). Previous
studies in model organisms and 3D culture models have shown
that BM secreted by the epithelial cells at their basal side plays
a role as an initial determinant in basal polarity (13, 16, 17).
Moreover, Pcan, through its cellular receptor dystroglycan, is in-
volved in the maintenance of epithelial polarity (18). Despite its
important roles in the establishment and maintenance of polarity,
pithelial cells are characterized by their polarized architec-
ture, which enables them to exert their varied functions in
it is not well understood how the BM achieves its own polarized
accumulation on the basal side of the cells.
Recently, two new factors have been shown to be critical for
the polarized deposition of BM proteins: Crag (Calmodulin-
binding protein related to a Rab3 GDP/GTP exchange protein)
and Rab10 (19, 20). The loss of either Crag or Rab10 leads to
anomalous accumulation of BM components on both the apical
as well as the basal side of epithelial cells without directly dis-
rupting the distribution of apical or basolateral proteins (19, 20).
This finding indicates that these two factors are specifically re-
quired for the restriction of the BM to the basal side and that this
process is independently regulated from apical and basolateral
secretion. It has been shown that Crag can act as a GTP exchange
factor (GEF) for Rab10 in Drosophila (21). In addition, in Dro-
sophila embryos, Scarface, a protease-like protein, has been shown
to act as a specific regulator of laminin A-polarized deposition (22).
To further elucidate the molecular mechanism leading to the
polarized secretion of BM components, we identified additional
genes involved in this process using the Drosophila follicular
epithelium (FE) as a model system. The FE is composed of a
monolayer of somatic cells, the follicle cells (FCs), which sur-
round the central germ-line cells during Drosophila oogenesis
(Fig. 1 A and B) (23). The FE is a classical epithelium, with
a distinct apical–basal polarity where the apical domain of the
FCs faces the germ line and the BM is secreted at the basal side
(Fig. 1 A and B). We identified Phosphatidylinositol synthase
(Pis), an evolutionarily conserved enzyme involved in the syn-
thesis of PtdIns, as a factor critical for restricted secretion of BM
proteins. In addition, we show that the level of PIP2 is critical for
this process. FCs with a reduced level of PIP2 allow an aberrant
accumulation of BM components on both their apical and basal
sides. Interestingly, a decrease of PIP2 significantly reduces the
The cellular functions of epithelia rely on their polarized ar-
chitecture. Loss of their polarity is often associated with car-
cinoma progression and tumor metastasis. The basement
membrane (BM), a specialized sheet of the extracellular matrix,
secreted basally by most epithelia, plays an important role in the
establishment and maintenance of epithelial cell polarity. How-
ever, the process of BM-polarized secretion is not well un-
derstood. In this study, we show that a significant decrease
of the phospholipid phosphatidylinositol 4,5-bisphosphate (PIP2)
results in BM protein secretion on both the apical and the basal
side of the epithelium. Together, our data indicate a specific role
for PIP2 in the organization of epithelial architecture by restricting
the deposition of BM proteins to the basal side.
Author contributions: O.D. and T.S. designed research; O.D., K.T., and G.B. performed
research; O.D. and T.S. analyzed data; and O.D. and T.S. wrote the paper.
The authors declare no conflict of interest.
1To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| May 27, 2014
| vol. 111
| no. 21
localization of Crag at the apical and lateral plasma membrane,
indicating that the level of PIP2 also controls the subcellular
distribution of Crag.
Pis Regulates the Polarized Deposition of BM Proteins. To find new
genes specifically involved in the basal secretion of BM proteins,
we performed a secondary screen of a collection of X-chromo-
some lethal mutations (19) using an α2–Coll IV protein trap line
(Coll IV–GFP) (24) as a reporter for BM localization. Apart
from the Crag alleles that had previously been described, we
identified four other mutant lines affecting Pis (Fig. S1 A–F) in
which we observed a mislocalization of BM proteins when FCs
were homozygous mutant (Fig. 1). In addition to the mislocalization
of BM proteins, we observed that these mutant FCs frequently
formed multiple layers or gaps in the epithelial sheet instead of
maintaining a strictly monolayered epithelium, indicating a role for
the affected gene in epithelial morphogenesis (Fig. S1 B and C).
Pis is a highly conserved transmembrane protein with its cat-
alytic domain present on the cytoplasmic side of the endoplasmic
reticulum (25). It has an indispensable role in the synthesis of
phosphatidylinositol (PI) (26). Pis is required for the incorporation
of an inositol group in cytidine diphosphate–diacylglycerol (CDP-
DAG) to produce PI (Fig. 2A). The phosphorylated derivatives of
PI, known as PtdIns, are crucial regulators of calcium homeostasis,
membrane trafficking, secretory pathways, and signal transduction
(26). In particular, Pis has been shown to be essential for PIP2
regeneration after its hydrolysis by phospholipase C (PLC) (27).
To confirm our initial observation that BM proteins are mis-
localized in Pis mutant cells, we visualized the distribution of two
main components of the BM membrane: Coll IV (α1 and α2
chains of Coll IV) and Pcan. In wild-type (WT) FC, Coll IV and
Pcan are localized in vesicles and secreted exclusively to the
basal side of the cell (Fig. 1 and ref. 18). Pis mutant FCs accu-
mulate α1 and α2 Coll IV apically (Fig. 1E). We obtained a
previously described null allele of Pis (Pis1) (27) and confirmed
that the abnormal apical accumulation of BM proteins, including
Coll IV (α1 and α2 chains) and Pcan–GFP (28) (Fig. 1F), is also
observed in FCs mutant for this null allele. Although, Crag and
Pis mutant cells showed the same apical mislocalization of BM
proteins, we observed some differences in their phenotypes. BM
components accumulated in a uniform, compact extracellular
layer in Crag mutant FCs (Fig. 1C) and in extracellular aggregates in
Pis mutant FCs (Fig. 1 D–F). By 3D reconstruction of Pis or Crag
apically to the outside of the cells (Fig. S2 and Movies S1 and S2).
Together,thesedataindicate thatwehaveidentified Pisasaregu-
lator ofpolarized BM deposition in the FE.
PIP2 Levels Are a Crucial Determinant for the Polar Secretion of BM
Proteins. Pis has been shown to be involved in PI production, and
therefore in the regulation of all PtdIns levels (26). Among the
various forms of PtdIns, PIP2 can mediate distinct biological
bution of BM proteins. (A) Schematic representation
of the FE with the different polarity domains in-
dicated. AJ, adherens junction. (B–B′′′′) Longitudinal
(Lg) section through a WT ovariole containing dif-
ferent egg chambers (B) and through the FC layer of
a WT egg chamber (magnification of B indicated
with the rectangular area, B′–B′′′′) expressing Coll
IV–GFP (a α2–Coll IV–GFP protein-trap; green),
a major component of the BM, which accumulates
at the basal side of the cell. The egg chambers are
stained with aPKC (an apical domain marker; red)
and Discs Large (Dlg; a lateral domain marker;
white), revealing the polarized structure of the ep-
ithelium, and DNA (blue). (B′) Apical (a) and basal
(b) sides of FCs are marked. (C and D) Lg section
through egg chambers containing CragCJ101(C) and
PisFM18(D) mutant FC clones expressing Coll IV–GFP
(green) and stained for F-Actin (red) and DNA
(blue). (C, C′) In Crag mutant FCs, marked by the
absence of intracellular GFP (green), the polarized
distribution of BM is disrupted, as revealed by the
strong accumulation of Coll IV–GFP at the apical
side of the FCs (arrows). (D, D′) The same phenotype
is observed in Pis mutant FCs, where Coll IV–GFP
accumulates apically in aggregates (arrows). (E and
F) Lg section through the FC layer of an egg
chamber containing Pis clones, marked by the ab-
sence of intracellular GFP (green; clonal boundary
indicated by a dashed line; WT and −/− homozy-
gous mutant FCs are specified), coexpressing Coll
IV–GFP (green, E) or Pcan–GFP (a Pcan–GFP protein-
trap; green F), and stained for α1–Coll IV (red, E, E″),
F-Actin (red, F and F″), and DNA (blue). In Pis mutant
FCs, Coll IV (E–E″) and Pcan (F, F′) accumulate apically,
indicating that Pis is required for restriction of BM
deposition to the basal side. (Bars, 10 μm.)
Pis mutant FCs disrupt the polarized distri-
| www.pnas.org/cgi/doi/10.1073/pnas.1407351111Devergne et al.
activities, including cell polarity, secretion, vesicular trafficking,
and cell adhesion (8–11, 29). We therefore investigated whether,
among all of the various PtdIns, PIP2 levels might be specifically
involved in the regulation of polarized BM deposition. Conse-
quently, we assayed the intracellular levels and distribution of
PIP2 in Pis mutant FCs using the PIP2 reporter, Ubi–PH–PLCδ–
GFP, which has been shown to specifically recognize PIP2 (30).
As in mammalian epithelial cells, PIP2 is localized to the plasma
membrane of WT FCs and shows a distinct polarized distribution
(Fig. 2B). In the Drosophila FE, PIP2, as revealed by the PIP2
reporter, is clearly detected at the apical and lateral domains of
the FCs (Fig. 2B). No specific accumulation of PIP2 is detectable
at the basal side of the cell using this PIP2 reporter and taking
the diffuse cytoplasmic staining into account. In Pis mutant FCs
of late-stage egg chambers, we observed a clear decrease in in-
tracellular PIP2 on apical and lateral sides, confirming a critical
role of Pis in the regulation of PIP2 levels (Fig. 2C).
To determine whether the BM mislocalization phenotype ob-
served in Pis mutant cells is due to the decrease of PIP2, we ex-
amined the role of other enzymes implicated in PtdIns production
in BM-polarized deposition. Downstream of Pis, the production of
PIP2 requires two different steps, which are catalyzed by two
distinct classes of PI kinases (Fig. 2A) (26). First, PI is phos-
phorylated by PI4K to generate PI4P, which is then phosphory-
lated by PIP5K to produce PIP2. In the same original screen of
X-linked lethal mutations, we also isolated PIP4K (PI4KIIIα)
mutants and previously showed that cells mutant for this gene
exhibit a loss of the PIP2 reporter from the apical plasma mem-
brane, thereby indicating a role of PI4KIIIα in the distribution of
PIP2 (31). Here, we tested these mutations for mislocalization of
BM proteins and found that FCs mutant for PI4KIIIα show an
apical accumulation of Coll IV (Fig. 3B). Similarly, we found that
when skittles (sktl), a Drosophila PIP5K, is knocked down using
RNAi in FCs, the BM protein Pcan accumulates at the apical side
(Fig. 3C; compare with Fig. 3A). Together, these data suggest that
the loss of PIP2 at the plasma membrane resulting from
the mutations in PI4KIIIa and sktl is the crucial factor for the
abnormal secretion of BM proteins on the apical side of the
To further confirm that PIP2 is involved in BM polarity, we
perturbed its production independently of the Pis/PI4K/PIP5K
pathway by using mutants in Phosphatase and Tensin homolog on
Chromosome 10 (PTEN) (Fig. 3D and Fig. S3) as well an RNAi
line (Fig. S4). PTEN is a phosphatase that catalyzes the con-
version of PIP3 to PIP2 (26). In FCs mutant for the loss-of-
function allele PTENC494, we again observed apically localized
Pcan in distinct aggregates (Fig. 3D), indicating that in the ab-
sence of PTEN, BM proteins are also not correctly restricted to
the basal side. We also isolated a separate mutation in PTEN
(JH59) in a screen on chromosome 2L and observed the same
apical mislocalization phenotype in homozygous mutant FCs
(Fig. S3). In addition, we were able to reproduce the same
phenotype using an RNAi knockdown against PTEN (Fig. S4).
The severity of the BM mislocalization observed in PTEN
mutant cells, using two different mutants and one RNAi line, was
less pronounced than in Pis mutant cells (Fig. 3D and Figs. S3 and
S4; compare with Fig. 1). Mutations in PTEN are, in fact, not
expected to block the production of PIP2 as severely, as long as
Pis and the PI kinase pathways are still functioning (see pathway
in Fig. 2A). Together, therefore, our data clearly indicate that
PIP2 levels are critical in the polarized deposition of BM proteins
in the FE.
Loss of PIP2 Does Not Lead to a General Loss of Polarity in the FE.
Although, to our knowledge, this is the first report indicating that
PIP2 has a crucial role in the polarized secretion of BM proteins,
PIP2 has been previously shown to be important for the estab-
lishment of cell polarity in mammalian MDCK cells (10, 11).
PTEN has also been shown to localize early to the apical domain,
metabolic cycle. Pis plays a critical role catalyzing the incorporation of inositol
and CDP-DAG to produce PI, the backbone of all PtdIns. Some of the PtdIns’
metabolic steps and their respective products are shown with the specific
Phosphoinositide kinase (PIK) and PtdIns phosphatase implicated. (B) Lg sec-
tion through the FE expressing the PIP2 reporter, Ubi–PH–PLCδ–GFP and
stained for GFP (green), F-actin (white), and DNA (blue). In WT FCs, PIP2 is
localized at the plasma membrane on the apical and lateral domains. (C, C′)
Lg section through FCs containing PisFM18mutant cells, expressing Ubi–PH–
PLCδ–GFP. Staining for GFP (green), F-actin (white), and DNA (blue) reveals
that the PIP2reporter is absent from the membrane of Pis mutant cells, which
are marked by the absence of the RFP clonal marker, in contrast to the WT
cells. The clonal boundary is indicated with a dashed line. WT and −/− ho-
mozygous FCs are specified. (Bar, 10 μm.)
Pis and PIP2 pathway. (A) Schematic representation of the PtdIns
WT FCs (tj–Gal4) expressing Pcan–GFP (green), stained for F-Actin (red) and
DNA (blue). (B, B′) Lg section through the FC layer of an egg chamber con-
taining PI4KIIIαGS27clones, marked by the absence of intracellular GFP, coex-
pressing Coll IV–GFP (green) and stained for F-Actin (red) and DNA (blue).
Clonal boundaries are indicated by a dashed line. WT and −/− homozygous FCs
are specified. In PI4KIIIα mutant FCs, BM proteins accumulate apically in
aggregates. (C, C′) Lg section through FCs expressing Pcan–GFP (green),
stained for F-Actin (red) and DNA (blue), and expressing an RNAi construct
against sktl (PIP5K) using tj–Gal4. The knockdown of PIP5K leads to the apical
localization of Pcan–GFP. (D, D′) Lg section through the FC layer of an egg
chamber containing PTENC494clones, marked by the absence of intracellular
GFP, coexpressing Pcan–GFP (green) and stained for F-Actin (red) and DNA
(blue). In PTEN mutant FCs, Pcan–GFP accumulates apically in aggregates. To-
gether, these data indicate that PIP2 controls the polarized secretion of BM
proteins. (Bars, 10 μm.)
PIP2 levels control polarized BM deposition. (A, A′) Lg section through
Devergne et al.PNAS
| May 27, 2014
| vol. 111
| no. 21
and its activity is required for the segregation of PIP2 and PIP3
(10). Furthermore, recent studies have established that the ratio
of PTEN and PIP2/PIP3 is important for apical polarity in the
Drosophila embryo (32). It therefore seemed possible that the BM
mislocalization phenotype we observed could be due to a more
general perturbation of cell polarity rather than to a specific effect
on BM secretion.
To examine whether the aberrant mislocalization of BM pro-
teins in Pis and PTEN mutant FCs is specific or whether it is due
to a global loss of epithelial cell polarity, we analyzed the distri-
bution of other major apical, basolateral, and adherens junctional
markers. All of the examined mutant genotypes produced occa-
sional multilayered epithelia. We therefore focused our analysis
on mutant FCs that maintain normal monolayer organization, to
avoid analyzing secondary effects due to an eventual loss of epi-
thelial architecture caused by the mislocalized BM and its sub-
sequent effects on cell polarity possibly via integrin signaling (Fig.
4 and Figs. S3 and S4). We used atypical protein kinase C (aPKC)
and Crumbs (Fig. S5A for Crumbs) for the apical domain; Discs
Large (Dlg) and Fasciclin II (Fig. S5B) for the lateral domain;
and, finally, DE–cadherin (DE-Cad) and Armadillo (Fig. S5C) for
the junction complex. Using these different markers, we did not
observe any extension or diminution of the different domains. We
therefore conclude that reducing the levels of PIP2 does not di-
rectly alter the polarized trafficking of these apical and lateral
polarity proteins. Instead, our data suggest that the exclusion of
BM components from the apical side of the FE is particularly
sensitive to a reduction in PIP2.
PIP2 Levels Are Required for the Proper Localization of Crag at the
Plasma Membrane. To determine whether there is an effect of the
decrease of PIP2 on known factors involved in BM polarized
secretion, we analyzed the distributions of Crag and Rab10 in Pis
In FCs, Crag localizes principally at the plasma membrane and
also colocalizes in intracellular puncta that overlap with Rab11
and Rab5 endosomal compartments (19). Because antibodies are
no longer available to detect endogenous Crag, we used a full-
length HA–Crag construct as a readout for Crag localization. This
construct has been shown to serve as a faithful marker of endoge-
nous Crag (19). Using this construct, we observed that in WT FCs,
HA–Crag localized at the plasma membrane and had a cytoplasmic
localization (Fig. 5, WT FC). However, in Pis mutant FCs, Crag no
longer accumulated at the plasma membrane, but instead showed
only a diffuse cytoplasmic localization (Fig. 5A). In Pis mutant FCs,
Crag levels were also reduced overall compared with WT FCs. It
should be noted that this phenotype is not fully penetrant, and some
Pis mutant FCs still showed normal Crag localization. However,
larger clones in late egg chambers showed a higher tendency for loss
of Crag at the plasma membrane, suggesting that cells need to be
severely depleted of PIP2 to lose Crag from the plasma membrane.
Similarly, in FCs mutant for PI4KIIIα, an enzyme required for PIP2
production downstream of Pis, Crag was lost from the plasma
membrane, and its overall level was reduced (Fig. 5B).
Another regulator of BM-polarized deposition is Rab10. It has
been shown that Crag can act as a GEF for Rab10 (20, 21). In
addition, Crag physically interacts with Rab10 and is critical for
the correct localization of Rab10 to the basal side of the FE (20).
Because we found that Pis activity is important for the sub-
cellular localization of Crag, we similarly assayed the localization
of Rab10 in Pis mutant FCs. However, we could not detect any
differences in Rab10 localization between Pis mutant and WT
cells (Fig. S5D), suggesting that PIP2 levels are not crucial for
the observed Rab10 accumulation in the cytoplasm and at the
basal side of the epithelial cells. It seems that the decrease and
loss of Crag from the apical and lateral membranes is not suf-
ficient to affect the basal accumulation of Rab10. Together, our
data suggest that PIP2 levels are essential for proper polarized
BM deposition, possibly through the regulation of the subcellular
localization of Crag, in epithelial cells.
The polarized organization of epithelial cells is normally estab-
lished and maintained by external cues (12). In particular, it has
been shown that the BM can direct the orientation of the apico-
basal axis of epithelial cells (15–17, 33, 34). In addition, in-
cubating MDCK cells with BM proteins such as collagen or
laminin leads to polarity reversal (13, 35, 36). Despite its im-
portant roles, little is known about the mechanism by which BM
proteins are specifically secreted at the basal side of the cell.
Recently, two regulators of the process, Crag and Rab10, were
shown to be critical in the polarized accumulation of BM
in FCs depleted of PIP2. Lg section through the FC
layer of egg chambers stained for markers of epi-
thelial polarity, such as aPKC (apical domain marker;
A and B), Dlg (lateral domain marker; C and D), and
DE-Cad (adherens junction marker; E and F), and
stained for DNA (blue). Egg chambers containing Pis
(A, C, and E) or PTEN (B, D, and F) mutant clones,
marked by the absence of intracellular GFP; clonal
boundaries are indicated by a dashed line. WT
and −/− homozygous FCs are specified. Pis and PTEN
mutant FCs do not show any difference in the dis-
tribution of the apical marker aPKC (A, A′ and B, B′,
red), the lateral marker Dlg (C, C′ and D, D′, red), or
the adherens junction marker DE-Cad (E, E′ and F, F′,
red). Note that in PTEN mutant FCs, Pcan–GFP
accumulates apically. Together, these data indicate
that disrupting the function of components in-
volved in PIP2 production has no direct effect on the
maintenance of polarity domains. (Bars, 10 μm.)
Normal apical–basal polarity is maintained
| www.pnas.org/cgi/doi/10.1073/pnas.1407351111 Devergne et al.
proteins by preventing aberrant apical secretion (19, 20). Here,
we report a critical role for PIP2 in the control of the polarized
secretion of BM components such as Coll IV and Pcan, which, to
our knowledge, has not been previously recognized.
In Pis mutant FCs, Pcan and Coll IV accumulated on both
sides of the epithelium. Similarly, when we affected the function
of different enzymes implicated in PIP2 production downstream
of Pis, such as PI4K and PIP5K, we also observed BM proteins to
accumulate apically. The same phenotype was also observed in
cells mutant for PTEN, which encodes an enzyme required for
the production of PIP2 from PIP3. These results indicate that
affecting the production of PIP2 in different ways leads to the
mislocalization of BM components. In contrast, the distributions
of other polarity proteins to the apical, basolateral, and junc-
tional domains were not immediately affected in epithelial cells
that showed a clear decrease of intracellular PIP2.
Although FCs with a decrease of PIP2 showed a clear apical
mislocalization of BM proteins, the observed accumulation was
patchier and less coherent than in Crag mutant FCs. This
somewhat weaker phenotype can be explained by the fact that
there are multiple ways to produce PIP2 in the cells, and it
therefore may take a longer time after the cells become mutant
for PIP2 levels to be sufficiently exhausted before a mislocaliza-
tion of BM components occurs.
PIP2 has been shown to be important in the establishment and
maintenance of epithelial polarity. In particular, the ratio of
PIP2/PIP3 was found to be critical for the establishment and
maintenance of the apical and lateral domains (10, 11). In non-
polarized MDCK cells, PIP2 and PIP3 are evenly distributed along
the plasma membrane. However, in early stages of the polariza-
tion, PIP2 becomes concentrated at the apical domain, whereas
PIP3 remains more abundant at the basolateral domain (10). This
process is initiated by PTEN, which localizes to the apical domain
and mediates the segregation of PIP2 to the apical domain and
PIP3 to the basolateral surface, thus recruiting apical determi-
nants to the apical domain. In Drosophila, PIP2 has also been
implicated in the establishment of the apical domain and the re-
cruitment of apical polarity proteins (32). Here, we show that
higher levels of PIP2 are not required primarily to maintain po-
larity domains, but instead control the polarized secretion of BM
proteins. In addition, we found that PIP2 is required for the
proper localization of Crag to the apical and lateral domains of
the plasma membrane, further suggesting that the control of the
polarized secretion of BM proteins by PIP2 is mediated in part
by the regulation of Crag localization. It seems likely that, similar
to the establishment and maintenance of the apical domain,
where PIP2 interacts with different polarity-promoting proteins,
PIP2, possibly indirectly, interacts apically with Crag, a compo-
nent necessary to prevent apical accumulation of BM compo-
nents. In addition, the decrease in intracellular PIP2 levels had
no visible effect on the cytoplasmic localization of Rab10, despite
the fact that Crag has been reported to be required for the proper
distribution of Rab10 (20). This result can be explained best by
the fact that Crag accumulates both at the apical and lateral
membrane and in intracellular compartments (19). The loss of
PIP2 affects the overall level of Crag and, in particular, strongly
reduces Crag localization to the membrane, but presumably still
supports sufficient levels of intracellular Crag to maintain the
subcellular localization of Rab10 in basal regions of the cell.
A recent study suggested that basal secretion of the BM pro-
teins could in part be explained by the localization of the re-
spective mRNAs to a basal compartment within the FCs and the
resulting proteins being targeted to basally localized Golgi
clusters (20). However, our results show that this localization
cannot be the determining factor for the exclusion of the BM
proteins from the apical side. In the absence of either Crag or
Rab10, or with low levels of PIP2, the BM proteins are still se-
creted on the basal side, but in addition, they now also accu-
mulate apically. It has been shown that the FCs themselves
produce the BM proteins found on their basal side (37). There-
fore, Crag and Pis are not required for the basal secretion of BM
proteins, but specifically for the prevention of the abnormal apical
accumulation. The mutant phenotypes also show that localization
of the respective mRNAs and targeting of the proteins to basal
Golgi compartments are not sufficient to prevent abnormal se-
cretion of these proteins on the apical side of the cell. Instead, the
phenotypes that we report show that exocytic vesicles containing
BM protein must be able to reach the apical side and deliver their
cargo in this location. It requires the activity of Crag and sufficient
levels of PIP2 to block this apical path of secretion. Although
targeting secretion toward the basal side may certainly facilitate
the basal accumulation of the BM proteins, the negative role of
PIP2 and of Crag are the important factors in assuring that BM
proteins are normally only secreted on the basal side. In summary,
our results show that PIP2, Crag, and Rab10 are primarily in-
volved in the blocking of apical secretion of BM proteins and not
in their basal targeting.
Although loss of PIP2 and Crag from the FCs, in general, does
not result in an overall loss of cell polarity, we occasionally ob-
serve more severe phenotypes in mutant epithelia with multi-
layering or gaps (see also ref. 19). In addition, at later stages of
egg chamber development, the mutant cells often die, and such
egg chambers do not form normal eggs. Because the more general
epithelial abnormalities are usually associated with later stages
and larger mutant clones, it is possible that the abnormal se-
cretion of BM proteins to the apical side may eventually lead to
such general epithelial defects. These abnormalities may well
correspond to the observed loss of cell polarity in mammalian
epithelial cells, when they are treated with exogenous collagen or
laminin (13, 35, 36). Alternatively, it is also possible that the loss of
PIP2 has other, more direct consequences on the long-term
maintenance of the epithelial polarity complexes without involving
Crag. Nevertheless, Crag appears to be particularly sensitive to the
loss of PIP2, because it is lost from the membranes, whereas other
polarity proteins are still maintained at apparently normal levels.
Our results allow us to propose a model in which PIP2 plays
a previously unidentified, essential role in the polarized secretion
of BM proteins. In epithelial cells, PIP2, which assumes high levels
at the apical and lateral plasma membrane, prevents the secretion
of BM components possibly via direct or indirect recruitment of
Crag and other factors to those domains. Crag prevents the
abnormal accumulation of BM proteins at these sites through
an as-yet-unknown mechanism. In Crag mutants, BM proteins
section through the FC layer of an egg chamber containing PisGB73(A) and
PI4KIIIαGS27clones (B), marked by the absence of intracellular GFP, coex-
pressing HA–Crag (UAS HA–Crag, red) and stained for DNA (blue); clonal
boundaries are indicated by dashed lines. WT and −/− homozygous FCs are
specified. In Pis (A, A′) and PI4KIIIα (B, B′) mutant cells, HA–Crag loses its
plasma membrane localization, and its levels are reduced overall, suggesting
that PIP2 levels are required for targeting or maintaining Crag at the
membrane. (Bars, 10 μm.)
PIP2 levels control the plasma membrane localization of Crag. Lg
Devergne et al. PNAS
| May 27, 2014
| vol. 111
| no. 21
are unselectively secreted to both the basal and the apical side of
the epithelial cells. In mutants in the PtdIns pathway, where the
intracellular level of PIP2 is low, Crag is no longer localized prop-
erly at the plasma membrane, and its overall levels are reduced.
This process, in turn, also leads to the loss of polarized secretion of
BM proteins and an unselective accumulation of BM proteins at
both sides of the cells. In summary, we have discovered a crucial
role for PIP2 in regulating cell polarity by controlling the polarized
accumulation of BM proteins, which is ultimately an essential
component in the establishment or maintenance of cell polarity.
Materials and Methods
Fly Stocks and Genetics. Duplication lines used for mapping were obtained
from the Bloomington Stock Center. The following Drosophila stocks were
used: CragCJ101(19); PisGB73, PisFM18, PisFJ78, and PisFV4(this study); Pis1and hs-
Pis (27) (a gift from C. Montell, University of California, Santa Barbara, CA);
PTENJH59(this study); PTENC494(38) (a gift from J. A. Brill, University of
Toronto, Toronto), PHPLCδ–GFP [PIP2 probe (30)]; UAS–mCD8–RFP (Bloo-
mington); tj–Gal4 (a gift from D. Bilder, University of California, Berkeley,
CA); UAS–RNAi–Sktl (TRIP line JF02796; Bloomington); UAS–RNAi–PTEN (TRIP
line JF01987; Bloomington); UAS–dicer 2 (Bloomington); UAS–HA–CragA (19);
and UASp–YFP–Rab10 (39). The protein trap lines Vkg–GFP [CC00791 (24)] and
Pcan–GFP [ZCL1700 (28)] were obtained from Flytrap. FC clones for Pis, PTEN,
PI4KIIIα, and Crag were induced by using the Flipase/FRT method (19).
Mapping and Rescue of Pis Mutations. Mapping and sequencing of PisGB73and
PisFM18were performed as described (19). PisFJ78and PisFV4were not sequenced,
but give equally strong phenotypes. Rescue experiments were performed by
subjecting PisGB73; hs–Pis and PisFM18; hs–Pis flies to daily heat-shock treatments
for 1 h from the embryonic to the adult stage.
Immunofluorescence Stainings. Ovaries were dissected and immunostained
as described (19). Ovaries, mounted in Aquapolymount (Polysciences), were
visualized by using a Zeiss LSM510 or Nikon A1 laser-scanning confocal mi-
croscope. The following primary antibodies were used: mouse anti-Coll IV
[6G7; 1:10 (40)], rabbit anti-aPKC (1:1,000; Santa Cruz), rat anti–DE-Cad [DCAD2;
1:20; Developmental Studies Hybridoma Bank (DSHB)], mouse anti-Dlg (4F3;
1:100; DSHB), rat anti-HA (3F10; 1:50; Roche), rabbit anti-GFP (AB3080P; 1:500;
Millipore), mouse anti-Crb (cq4; 1/5; DSHB), mouse anti-FasII (1D4; 1/1000;
DSHB), and mouse anti-Arm (N2 7A1; 1/50; DSHB). Secondary antibodies
used were Alexa Fluor 488-, 568-, and 647-conjugated (1:400; Life Tech-
nologies). Alexa Fluor 546 Phalloidin (1:500; Life Technologies) and Alexa
Fluor 647 Phalloidin (1:50; Life Technologies) were used to stain F-actin.
DNA was stained by using Hoechst (10 μg/mL; Life Technologies).
ACKNOWLEDGMENTS. We thank Natalie Denef and Yan Yan for conducting
the original mutant screen that led to the isolation of the Pis mutations and
Elena Domanitskaya and Yi Sun for conducting the original mutant screen
that led to the isolation of the PTEN mutation. We thank Craig Montell, Julie
Brill, the Developmental Studies Hybridoma Bank, and the Bloomington
stock center for providing flies and antibodies; Joe Goodhouse and Gary
Laevsky for advice with confocal microscopy; and members of the T.S. and
E. Wieschaus laboratories for feedback and advice. We also thank Wei Li,
Shawn Little, Julie Merkle, and Jean Schwarzbauer for helpful comments on
manuscript. This work was supported by the Howard Hughes Medical Insti-
tute and U.S. Public Health Service Grant R01 GM077620.
1. Thompson BJ (2013) Cell polarity: Models and mechanisms from yeast, worms and
flies. Development 140(1):13–21.
2. Müller HA, Bossinger O (2003) Molecular networks controlling epithelial cell polarity
in development. Mech Dev 120(11):1231–1256.
3. Roignot J, Peng X, Mostov K (2013) Polarity in mammalian epithelial morphogenesis.
Cold Spring Harb Perspect Biol 5(2):5.
4. Ellenbroek SI, Iden S, Collard JG (2012) Cell polarity proteins and cancer. Semin Cancer
5. Thiery JP (2002) Epithelial-mesenchymal transitions in tumour progression. Nat Rev
6. Rodriguez-Boulan E, Kreitzer G, Müsch A (2005) Organization of vesicular trafficking
in epithelia. Nat Rev Mol Cell Biol 6(3):233–247.
7. Apodaca G, Gallo LI, Bryant DM (2012) Role of membrane traffic in the generation of
epithelial cell asymmetry. Nat Cell Biol 14(12):1235–1243.
8. Leslie NR, Batty IH, Maccario H, Davidson L, Downes CP (2008) Understanding PTEN
regulation: PIP2, polarity and protein stability. Oncogene 27(41):5464–5476.
9. Di Paolo G, De Camilli P (2006) Phosphoinositides in cell regulation and membrane
dynamics. Nature 443(7112):651–657.
10. Martin-Belmonte F, et al. (2007) PTEN-mediated apical segregation of phosphoino-
sitides controls epithelial morphogenesis through Cdc42. Cell 128(2):383–397.
11. Gassama-Diagne A, et al. (2006) Phosphatidylinositol-3,4,5-trisphosphate regulates
the formation of the basolateral plasma membrane in epithelial cells. Nat Cell Biol
12. Nelson WJ (2003) Adaptation of core mechanisms to generate cell polarity. Nature
13. Yu W, et al. (2005) Beta1-integrin orients epithelial polarity via Rac1 and laminin. Mol
Biol Cell 16(2):433–445.
14. Yurchenco PD (2011) Basement membranes: Cell scaffoldings and signaling platforms.
Cold Spring Harb Perspect Biol 3(2):3.
15. Yurchenco PD, Amenta PS, Patton BL (2004) Basement membrane assembly, stability
and activities observed through a developmental lens. Matrix Biol 22(7):521–538.
16. Li S, Edgar D, Fässler R, Wadsworth W, Yurchenco PD (2003) The role of laminin in
embryonic cell polarization and tissue organization. Dev Cell 4(5):613–624.
17. O’Brien LE, Zegers MM, Mostov KE (2002) Opinion: Building epithelial architecture:
Insights from three-dimensional culture models. Nat Rev Mol Cell Biol 3(7):531–537.
18. Schneider M, et al. (2006) Perlecan and Dystroglycan act at the basal side of the
Drosophila follicular epithelium to maintain epithelial organization. Development
19. Denef N, Chen Y, Weeks SD, Barcelo G, Schüpbach T (2008) Crag regulates epithelial
architecture and polarized deposition of basement membrane proteins in Drosophila.
Dev Cell 14(3):354–364.
20. Lerner DW, et al. (2013) A Rab10-dependent mechanism for polarized basement
membrane secretion during organ morphogenesis. Dev Cell 24(2):159–168.
21. Xiong B, et al. (2012) Crag is a GEF for Rab11 required for rhodopsin trafficking and
maintenance of adult photoreceptor cells. PLoS Biol 10(12):e1001438.
22. Sorrosal G, Pérez L, Herranz H, Milán M (2010) Scarface, a secreted serine protease-
like protein, regulates polarized localization of laminin A at the basement membrane
of the Drosophila embryo. EMBO Rep 11(5):373–379.
23. Horne-Badovinac S, Bilder D (2005) Mass transit: Epithelial morphogenesis in the
Drosophila egg chamber. Dev Dyn 232(3):559–574.
24. Buszczak M, et al. (2007) The carnegie protein trap library: A versatile tool for Dro-
sophila developmental studies. Genetics 175(3):1505–1531.
25. Kim YJ, Guzman-Hernandez ML, Balla T (2011) A highly dynamic ER-derived phos-
phatidylinositol-synthesizing organelle supplies phosphoinositides to cellular mem-
branes. Dev Cell 21(5):813–824.
26. Maffucci T (2012) An introduction to phosphoinositides. Curr Top Microbiol Immunol
27. Wang T, Montell C (2006) A phosphoinositide synthase required for a sustained light
response. J Neurosci 26(49):12816–12825.
28. Morin X, Daneman R, Zavortink M, Chia W (2001) A protein trap strategy to detect
GFP-tagged proteins expressed from their endogenous loci in Drosophila. Proc Natl
Acad Sci USA 98(26):15050–15055.
29. Schill NJ, Anderson RA (2009) Out, in and back again: PtdIns(4,5)P(2) regulates cad-
herin trafficking in epithelial morphogenesis. Biochem J 418(2):247–260.
30. Gervais L, Claret S, Januschke J, Roth S, Guichet A (2008) PIP5K-dependent production
of PIP2 sustains microtubule organization to establish polarized transport in the
Drosophila oocyte. Development 135(23):3829–3838.
31. Yan Y, Denef N, Tang C, Schüpbach T (2011) Drosophila PI4KIIIalpha is required
in follicle cells for oocyte polarization and Hippo signaling. Development 138(9):
32. Chartier FJ, Hardy EJ, Laprise P (2011) Crumbs controls epithelial integrity by in-
hibiting Rac1 and PI3K. J Cell Sci 124(Pt 20):3393–3398.
33. Miner JH, Yurchenco PD (2004) Laminin functions in tissue morphogenesis. Annu Rev
Cell Dev Biol 20:255–284.
34. Rasmussen JP, Reddy SS, Priess JR (2012) Laminin is required to orient epithelial po-
larity in the C. elegans pharynx. Development 139(11):2050–2060.
35. O’Brien LE, et al. (2001) Rac1 orientates epithelial apical polarity through effects on
basolateral laminin assembly. Nat Cell Biol 3(9):831–838.
36. Zuk A, Matlin KS (1996) Apical beta 1 integrin in polarized MDCK cells mediates
tubulocyst formation in response to type I collagen overlay. J Cell Sci 109(Pt 7):
37. Haigo SL, Bilder D (2011) Global tissue revolutions in a morphogenetic movement
controlling elongation. Science 331(6020):1071–1074.
38. Huang H, et al. (1999) PTEN affects cell size, cell proliferation and apoptosis during
Drosophila eye development. Development 126(23):5365–5372.
39. Zhang J, et al. (2007) Thirty-one flavors of Drosophila rab proteins. Genetics 176(2):
40. Murray MA, Fessler LI, Palka J (1995) Changing distributions of extracellular matrix
components during early wing morphogenesis in Drosophila. Dev Biol 168(1):150–165.
| www.pnas.org/cgi/doi/10.1073/pnas.1407351111 Devergne et al.
Devergne et al. 10.1073/pnas.1407351111
SI Materials and Methods
For the 3D reconstructions, a series of optical cross-sections
through mutant clones for Crag or Pis, coexpressing Collagen IV
(Coll IV)–GFP and mCD8–RFP, were taken with a 0.3-μm space
between each section. Then, the 3D views were reconstituted by
using Volocity software (Perkin-Elmer). The colocalization be-
tween Coll IV–GFP and mCD8–RFP was detected by using the
colocalization tool of Volocity.
Phosphatidylinositol synthase (Pis). (A) Schematic representation of Pis protein and mapping of Pis mutations are shown. Transmembrane domain (TM; blue)
and cytidine diphosphate (CDP)–alcohol phosphatidyltransferase enzymatic domain (red). Missense mutations in GB73 and FM18 (red) and the deleted region
in Pis1(blue) are shown. Both missense mutations change conserved amino acids localized in the CDP–alcohol phosphatidyltransferase catalytic domain of Pis.
(B and C) Ovariole (B) and egg chamber (C) containing Pis mutant FCs marked by the absence of intracellular GFP (green) and stained for F-Actin (red) and DNA
(blue). In some clones, the Pis mutant FCs are irregular in shape, pile up in multiple layers (arrows), and form gaps in the epithelium (arrowheads). (D and E)
Ovariole (D) and egg chamber (E) containing PisFM18mutant clones encompassing the entire FE and expressing Pis under the control of the heat-shock pro-
moter (hs–Pis). No epithelial defects are observed. The egg chambers are stained for F-Actin (red) and DNA (blue). Clones are marked by the absence of in-
tracellular GFP (green). (F, F′) Egg chamber with PisFM18mutant clones coexpressing Pis and stained for α1–Coll IV (red). No apical accumulation of Coll IV is
detected in Pis mutant cells, marked by the absence of intracellular GFP (clonal marker, green), when Pis is expressed. (D–F) Together, these data indicate that
the observed phenotypes affecting FC architecture and the polarized deposition of BM proteins are due to the loss of Pis activity. In addition, we were able to
rescue the lethality of Pis mutants using hs–Pis. The clonal boundary is indicated with a dashed line. WT and −/− homozygous FCs are specified. (Bars, 10 μm.)
Epithelial and basement membrane (BM) polarized deposition defects in GB73 and FM18 mutant follicle cells (FCs) are caused by mutation in
Devergne et al. www.pnas.org/cgi/content/short/14073511111 of 6
and D–D″) mutant epithelium expressing Coll IV–GFP (green). All FCs express the membrane marker mCD8–RFP (red). The colocalization of red voxels (mCD8–
RFP) and green voxels (Coll IV–GFP), as determined by the colocalization tool of Volocity (Perkin-Elmer), is indicated in yellow. In mutant cells for Crag (B) or Pis
(D), Coll IV–GFP accumulates at the level of the apical membrane, as revealed by the green and red voxel colocalization (yellow). To determine whether Coll IV–
GFP is apically secreted outside the FCs, as it is known to be at the basal side, we compared the distribution of Coll IV with the colocalization reference (yellow).
In Crag and Pis mutant FCs, Coll IV–GFP is apically localized outside of the FC (arrows), because we observed some Coll IV–GFP on top of the yellow (arrows). The
results indicate that in Crag mutant FCs, Coll IV–GFP is secreted apically and tends to form a continuous BM sheet (arrows, B). In Pis mutant FCs, Coll IV–GFP is
also secreted apically; however, it accumulates in patches at the surface, leading to the formation of a noncontiguous, aggregate-like structure (arrows, D). The
difference between Crag and Pis mutant FCs appears, therefore, to be due to a difference in the amount of BM mislocalization and not to a difference in
secretion. These data indicate that both Crag and Pis are required to prevent secretion of BM proteins at the apical side of the FC and thus, indirectly, restrict
the BM to the basal side of the FCs. In the mutant cells, BM proteins are no longer exclusively secreted on the basal side, but now also become indiscriminately
deposited on the apical surface of the cells. Orientation reference is shown. Anterior (A) and posterior (P) are indicated. See Movies S1 and S2 for different
views of the 3D reconstruction of Crag mutant FE and Pis mutant FE, respectively.
Crag and Pis mutant cells secrete Coll IV apically. Longitudinal (Lg) section (A and C) and 3D reconstruction (B and D) of a Crag (A and B–B″) and Pis (C
Devergne et al. www.pnas.org/cgi/content/short/1407351111 2 of 6
through the FC layer of an egg chamber containing PTENJH59clones, marked by the absence of intracellular GFP, coexpressing Pcan-GFP (green), and stained for
DNA (blue) and markers of epithelial polarity, such as aPKC (apical marker; A, A′), Discs Large (Dlg; lateral marker; B, B′), and DE–Cadherin (DE-Cad; adherens
junction marker; C, C′). Mutant cells for PTENJH59accumulate Pcan–GFP apically, but do not have altered polarity domains. (Bars, 10 μm.)
PTENJH59mutant FCs accumulate Pcan apically, but do not exhibit any defects in the distribution of other polarity markers. Shown are Lg section
Devergne et al. www.pnas.org/cgi/content/short/14073511113 of 6
WT FCs expressing Pcan–GFP (green) and stained for DNA (blue) and aPKC (apical marker, A, A′), Dlg (lateral marker, C, C′), or DE-Cad adherens junction
marker, E, E′). (B, D, and F) Egg chambers expressing Pcan–GFP (green) and RNAi against PTEN using tj–Gal4 in the FCs. In knockdown PTEN FCs, Pcan–GFP
accumulates apically. However, the other polarity markers, such as aPKC (B′), Dlg (D′), and DE-Cad (F′), are distributed normally. No distinguishable differences
are observed with the control. (Bars, 10 μm.)
FCs with RNAi knockdown for PTEN accumulate Pcan apically, but do not exhibit any defects in the distribution of other polarity markers. (A, C, and E)
Devergne et al. www.pnas.org/cgi/content/short/14073511114 of 6
chamber containing Pis clones, marked by the absence of intracellular GFP, stained for DNA (blue) and markers of epithelial polarity, such as Crumbs (Crb;
apical marker; A, A′), Fasciclin II (Fas II; lateral marker; B, B′), and Armadillo (Arm; adherens junction marker; C, C′). The analysis of these polarity markers
further confirmed that cells mutant for Pis do not have altered polarity domains. (D, D′) Lg section through the FC layer of an egg chamber containing PisGB73
clones, marked by the absence of intracellular RFP (red), coexpressing YFP–Rab10 (green), and stained for DNA (blue). No differences in Rab10 subcellular
localization between Pis mutant and WT cells could be detected, indicating that Pis does not control the localization of Rab10. Clonal boundaries are indicated
by dashed lines. WT and −/− homozygous FCs are specified. (Bars, 10 μm.)
Pis mutant FCs do not exhibit any defects in the distribution of various polarity markers and Rab10. (A–C) Lg section through the FC layer of an egg
Devergne et al. www.pnas.org/cgi/content/short/14073511115 of 6
Movie S1. Download full-text
FCs express Coll IV–GFP (green) and the membrane marker mCD8–RFP (red). The colocalization between Coll IV–GFP and mCD8–RFP, as determined by the
colocalization tool of Volocity, is indicated in yellow. Orientation reference is shown. Note the green accumulation of Coll IV outside (“on top of”) the yellow
(red) labeled membrane region.
The 3D reconstruction of Crag mutant FCs shown in Fig. S2. Different views of the 3D reconstruction of Crag mutant FCs are shown in the movie.
express Coll IV–GFP (green) and the membrane marker mCD8–RFP (red). The colocalization between Coll IV–GFP and mCD8–RFP, as determined by the co-
localization tool of Volocity, is indicated in yellow. Orientation reference is shown. Note the green accumulation of Coll IV outside (on top of) the yellow (red)
labeled membrane region.
The 3D reconstruction of Pis mutant FCs shown in Fig. S2. Different views of the 3D reconstruction of Pis mutant FCs are shown in the movie. FCs
Devergne et al. www.pnas.org/cgi/content/short/14073511116 of 6