The apical transmembrane protein Crumbs functions as a tumor suppressor that regulates Hippo signaling by binding to Expanded.

Abstract

The Hippo signaling pathway regulates organ size and tissue homeostasis from Drosophila to mammals. At the core of the Hippo pathway is a kinase cascade extending from the Hippo (Hpo) tumor suppressor to the Yorkie (Yki) oncoprotein. The Hippo kinase cascade, in turn, is regulated by apical membrane-associated proteins such as the FERM domain proteins Merlin and Expanded (Ex), and the WW- and C2-domain protein Kibra. How these apical proteins are themselves regulated remains poorly understood. Here, we identify the transmembrane protein Crumbs (Crb), a determinant of epithelial apical-basal polarity in Drosophila embryos, as an upstream component of the Hippo pathway in imaginal disk growth control. Loss of Crb leads to tissue overgrowth and target gene expression characteristic of defective Hippo signaling. Crb directly binds to Ex through its juxtamembrane FERM-binding motif (FBM). Loss of Crb or mutation of its FBM leads to mislocalization of Ex to basolateral domain of imaginal disk epithelial cells. These results shed light on the mechanism of Ex regulation and provide a molecular link between apical-basal polarity and tissue growth. Furthermore, our studies implicate Crb as a putative cell surface receptor for Hippo signaling by uncovering a transmembrane protein that directly binds to an apical component of the Hippo pathway.

Full text

Page 1

The apical transmembrane protein Crumbs functions
as a tumor suppressor that regulates Hippo signaling
by binding to Expanded
Chen Linga,1, Yonggang Zhenga,1, Feng Yina, Jianzhong Yua, Juan Huangb, Yang Hongb, Shian Wua,c,2, and Duojia Pana,2
aDepartment of Molecular Biology and Genetics, Howard Hughes Medical Institute, Johns Hopkins University School of Medicine, Baltimore, MD 21205;
bDepartment of Cell Biology and Physiology, University of Pittsburgh Medical School, Pittsburgh, PA 15261; andcDepartment of Genetics and Cell Biology,
School of Life Sciences, Nankai University, Tianjin 300071, China
Edited* by Jeremy Nathans, Johns Hopkins University, Baltimore, MD, and approved May 4, 2010 (received for review March 30, 2010)
The Hippo signaling pathway regulates organ size and tissue
homeostasis from Drosophila to mammals. At the core of the Hippo
pathway is a kinase cascade extending from the Hippo (Hpo) tumor
suppressor to the Yorkie (Yki) oncoprotein. The Hippo kinase cas-
cade, in turn, is regulated by apical membrane-associated proteins
such as the FERM domain proteins Merlin and Expanded (Ex), and
the WW- and C2-domain protein Kibra. How these apical proteins
are themselves regulated remains poorly understood. Here, we
identify the transmembrane protein Crumbs (Crb), a determinant
of epithelial apical-basal polarity in Drosophila embryos, as an up-
stream component of the Hippo pathway in imaginal disk growth
control. Loss of Crb leads to tissue overgrowth and target gene
expression characteristic of defective Hippo signaling. Crb directly
binds to Ex through its juxtamembrane FERM-binding motif (FBM).
Loss of Crb or mutation of its FBM leads to mislocalization of Ex to
basolateral domain of imaginal disk epithelial cells. These results
shed light on the mechanism of Ex regulation and provide a molec-
ular link between apical-basal polarity and tissue growth. Further-
more, our studies implicate Crb as a putative cell surface receptorfor
Hippo signaling by uncovering a transmembrane protein that di-
rectly binds to an apical component of the Hippo pathway.
apical-basal polarity|development|Drosophila|organ size|signal
transduction
T
that regulates organ size in diverse species, including mammals
(1, 2). In Drosophila, four tumor suppressor proteins including
Hippo (Hpo), Salvador (Sav), Warts (Wts) and Mats form a ki-
nase cascade that ultimately phosphorylates and inactivates the
oncoprotein Yki (3). Although the molecular organization of this
core kinase cascade is well established, signaling events upstream
of Hpo remains less understood. Studies in Drosophila have im-
plicated several tumor suppressor proteins as upstream regulators
of the Hippo kinase cascade. Several apical membrane-associated
cytoplasmic proteins, including the FERM-domain proteins Ex-
panded (Ex) and Merlin (Mer) (4), and the WW- and C2-domain
protein Kibra (5–7), have been suggested to function together in
a protein complex that regulates Hippo signaling, at least in part,
by recruiting the Hpo-Sav kinase complex to cell membrane (5).
The atypical cadherin Fat (Ft) has been proposed as a cell surface
receptor that regulates Hippo signaling by controlling the mem-
brane localization and/or stability of Ex (8–10). However, no
physical interactions have been reported between Ft and Ex, and
this model was challenged with an alternative model wherein Ft
and Ex function in parallel to regulate Wts stability and activity,
respectively (11, 12). Overall, our understanding of these up-
stream pathway components remains incomplete, and additional
cell surface proteins may exist that physically link the apical
components such as Ex to the cell membrane.
Crumbs (Crb) is an apical transmembrane protein that is re-
quired for organizing apical-basal polarity and adherens junc-
he Hippo signaling pathway, first discovered in Drosophila, has
recently emerged as an evolutionarily conserved mechanism
tions (AJs) in gastrulating Drosophila embryos (13). It contains
28 EGF-like and four laminin AG-like repeats in its extracellular
domain and a short intracellular domain including a juxtamem-
brane FERM-binding motif (FBM) and a C-terminal PDZ-
binding motif (PBM). Through its PBM, Crb forms a complex
with the PDZ domain proteins Stardust (Sdt) and Patj (14–16).
In regulating apical-basal polarity in Drosophila embryos, the
Crb-Sdt-Patj complex functions together with another apical
protein complex containing Bazooka (Baz), Par-6, and aPKC,
and these apical complexes are antagonized by a basolateral
protein complex containing Scribble (Scrib), Discs large (Dlg),
and Lethal giant larvae (Lgl) (17, 18). Strikingly, the intracellular
domain of Crb is required and sufficient for Crb’s function in
apical-basal polarity in Drosophila embryos (19). In contrast to its
essential role in establishing apical-basal polarity in Drosophila
embryos, Crb is not essential for apical-basal polarity or AJ in-
tegrity in third instar larval imaginal discs (20, 21). Rather, Crb is
required for rhabdomere elongation and stalk membrane for-
mation during photoreceptor morphogenesis (20, 21).
Here, we report the identification of Crb as a tumor suppressor
and an upstream component of the Hippo signaling pathway.
We further show that Crb directly binds to Ex through Crb’s FBM
and is required for apical localization of Ex in imaginal epithelium.
Our studies uncover a transmembrane protein that directly binds
to an apical component of the Hippo pathway and implicate Crb
as a potential cell surface receptor for Hippo signaling.
Results
Identification of crb as a Tumor Suppressor Gene. In a screen for
growth regulators using the eyeless-FLP/recessive cell lethal
technique (22), we identified two lethal mutations (82–04 and
82–16) defining a single complementation group on chromosome
3R that caused mild overgrowth of adult eyes and increased
representation of mutant tissues (Fig. 1 A and B). To further
compare the growth property of the mutant versus wild-type
cells, we generated mosaic eyes using eyeless-FLP without the
recessive cell lethal. Such mosaic eyes also showed increased
representation of mutant tissues (Fig. 1 C and D). Adult wings
comprised predominantly of mutant cells were also larger than
normal wings (Fig. 1 E and F). Consistent with the overgrowth of
adult tissues, mutant clones in larval eye and wing imaginal discs
Author contributions: C.L., Y.Z., and D.P. designed research; C.L. and Y.Z. performed re-
search; F.Y., J.Y., J.H., Y.H., and S.W. contributed new reagents/analytic tools; C.L., Y.Z.,
and D.P. analyzed data; and C.L., Y.Z., and D.P. wrote the paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
1C.L. and Y.Z. contributed equally to this work.
2To whom correspondence may be addressed. E-mail: djpan@jhmi.edu or wusa@nankai.
edu.cn.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1004279107/-/DCSupplemental.
10532–10537
| PNAS
| June 8, 2010
| vol. 107
| no. 23www.pnas.org/cgi/doi/10.1073/pnas.1004279107

Page 2

were larger than their twin spots (no cell lethal used) (Fig. 1 G
and H).
We mapped 82–04 and 82–16 to molecular coordinates 20096927;
20124398 based on their noncomplementation with two overlapping
deficiencies, Df(3R)Exel8178 (20096927;20353553) and Df(3R)
BSC317 (20046218;20124398). Tests with available mutations in
this interval showed that they failed to complement three in-
dependent alleles of crumbs (crb), crb1, crb2, and crb07207. 82–04
and 82–16 contained nonsense mutations that were predicted
to truncate Crb in its extracellular domain: 1252Gluto STOP in
82–04 and 68Trpto STOP in 82–16 (Fig. 1I). As 82–16 trun-
cates Crb close to its N terminus, and both alleles displayed simi-
lar overgrowth phenotypes, we consider both as null alleles.
Loss of crb Produces Signaling Defects Similar to Mutants of Up-
stream Regulators of the Hippo Pathway. The overgrowth pheno-
type observed in crb mosaic eyes resemble that of defective Hippo
signaling. To explore the relationship between Crb and Hippo
signaling, we examined diap1 and ex, two well characterized
Hippo target genes, in eye discs. crb clones showed a modest
increase in diap1 transcription and Diap1 protein levels (Fig. 2
A–B″). crb clones also showed a significant increase in Ex protein
levels (Fig. 2 C–C″), although we could not detect a visible in-
crease in ex transcription (Fig. 3 A–A″). Such modest effects
on target genes resemble upstream components of Hippo sig-
naling but are weaker than the core components of the Hippo
kinase cascade.
Another characteristic of defective Hippo signaling is alteration
of interommatidial cell numbers in pupal retina. Inactivation of
core components of the Hippo kinase cascade (such as hpo and
sav) results in a massive increase in interommatidial cells [>40
extra cells per cluster, (ECPC)], whereas mutants of upstream
regulators of the Hippo pathway show a much milder phenotype,
with loss of ex, mer, and kibra resulting in 1.4, 8.4, and 5.8 ECPC,
respectively (5). crb mutant pupal retina showed an average of 1.2
ECPC (Fig. 4 A–A″), which resembles ex but is milder than mer or
kibra. To examine the relationship between crb and the known
upstream regulators, we examined kibra crb, ex;crb and mer;crb
double mutant clones. Although crb, mer, and kibra mutant eyes
had an average of 1.2, 8.4, and 5.8 ECPC, respectively, mer;crb,
and kibra crb eyes had 14.1 and 21.4 ECPC, respectively (Fig. 4).
In contrast, ex;crb mutant eyes had 1.5 ECPC, which resembled crb
or ex single mutants (Fig. 4). These observations strongly suggest
that Crb might act through Ex to regulate Hippo signaling.
Crb Directly Binds to Ex Through Its FBM. Tosubstantiateamolecular
link between Crb and Hippo signaling, we examined the effect of
Crb on Wts T1077 phosphorylation (5). Because a truncated Crb
construct encoding the membrane-bound cytoplasmic domain of
Crb is sufficient for Crb’s function in Drosophila embryos (19),
we examined the activity of an analogous construct (CrbMyc-Intra)
in S2R+ cells. Interestingly, although CrbMyc-Intraalone did not
activate Wts phosphorylation, it enhanced Ex-mediated activa-
tion of Wts phosphorylation (Fig. 5A), thus revealing a positive
influence of Crb on Hippo kinase cascade. In the course of
this experiment, we noted that CrbMyc-Intrapromoted Ex phos-
crbcrb
WT
A
crb
B
WT
C
crb
D
GH
I
FBM
PBM
Lamin AG-likeLamin AG-like
EGF-like
Transmembrane
* **
****
82-1682-04
Hs LREKRQTEGTYRPSSEEQVGARVPPTPNLKLPPEERLI
Dm RN-KRATRGTYSPSAQEYCNPRLEMDNVLKPPPEERLI
E
WT
F
crb
Fig. 1.
a mutant eye composed predominantly of crb82-04cells (B). The genotypes
are y w ey-flp; FRT 82B/FRT 82B P[w+], L(3)c1-R3 (A) and y w ey-flp; FRT 82B
crb82-04/FRT 82B P[w+], L(3)c1-R3 (B). (C and D) A control (C) and a mutant eye
containing crb82-04clones (D). The genotypes are y w ey-flp; FRT 82B /FRT 82B
Ubi-GFP (C) and y w ey-flp; FRT 82B crb82-04/FRT 82B Ubi-GFP (D). Note that
crb82-04mosaic eyes (D) contained predominantly mutant tissues (white),
whereas mosaic eyes for a control chromosome contained far less white
tissues (C). (E and F) A control (E) and a mutant wing composed pre-
dominantly of crb82-04cells (F). The genotypes are y w MS1096-Gal4, UAS-flp;
FRT 82B /FRT 82B P[w+], L(3)c1-R3 (G) and y w MS1096-Gal4, UAS-flp; FRT 82B
crb82-04/FRT 82B P[w+], L(3)c1-R3 (H). crb82-04wings were 112% of normal size
(n = 20, t test: P = 0.0001). (G and H) An eye (G) or wing (H) disk containing
crb82-04clones (GFP-negative). Mutant clones were induced at 36 h after
egg deposition. The average area of crb eye clones versus twin spots is 2.5
(n = 30). (I) Schematic diagram of the Drosophila Crb protein. The nonsense
mutations in crb82-04and crb82-16predicted to truncate the protein are also
shown. Sequence alignment shows the cytoplastic domains of Crb orthologs
from Drosophila (Dm) and human (Hs).
Loss of crb results in tissue overgrowth. (A and B) A control (A) and
crb
B
diap1-lacZ
B’
crb
C
Diap1
C’
crb
Ex
A A’A”
B”
C”
Fig. 2.
negative) are marked by arrows, and arrowhead marks the morphogenetic
furrow (MF). (A–A″) An eye disk showing elevated diap1-lacZ levels in crb
clones posterior to the MF. (B–B″) An eye disk showing elevated Diap1 pro-
tein levels in crb clones posterior to the MF. (C–C″) An eye disk showing el-
evated Ex protein levels in crb clones both anterior and posterior to the MF.
Crb regulates Hippo target genes. In all panels, crb clones (GFP-
Ling et al. PNAS
| June 8, 2010
| vol. 107
| no. 23
| 10533
DEVELOPMENTAL
BIOLOGY

Page 3

phorylation accompanied by a decrease of Ex protein levels
(Fig. 5 A and B). Interestingly, mutation of the FBM (by mutating
Y10P12E16of the FBM to alanines), but not the PBM (by deleting
the C-terminal residues ERLI), abrogated Crb-induced Ex phos-
phorylation (Fig. 5C). Given the presence of FERM domain in
Ex, we hypothesized that in our S2R+ cell assay, Crb may recruit
Ex to membrane via its FBM. Upon reaching the cell membrane,
Ex may be phosphorylated by unknown kinase(s). Consistent with
this hypothesis, targeting Ex to cell membrane by attaching
a myristylation signal to it N terminus also induced Ex phos-
phorylation (Fig. 5D).
To investigate physical interactions between Crb and Ex, we
used subcellular fractionation to examine the effect of CrbMyc-Intra
on the relative distribution of Ex between cytosolic and membrane
fractions. When expressed alone in S2R+ cells, Ex was equally
distributed in membrane and cytosolic fractions (Fig. 5E). Strik-
ingly, coexpression with CrbMyc-Intraresulted in a near complete
relocation of Ex to membrane fraction (Fig. 5E). This interaction
is specific, because CrbMyc-Intradid not affect the relative dis-
tribution of FERM-domain proteins Mer and Moesin (Moe)
(Fig. S1 A and B). Of note, a similar construct expressing the
membrane-bound intracellular domain of Ft (FtΔECD) (9) failed
to recruit Ex to membrane fraction in our assay (Fig. S2). To
further explore the specificity of Crb-Ex interaction, we con-
ducted similar assays using CrbMyc-Intrawith mutations in its
FBM or PBM. Consistent with a critical role for Crb’s FBM in Ex
binding, mutation of the FBM, but not the PBM, abrogated Ex
membrane recruitment (Fig. 5E). In contrast, Sdt, which is
known to bind Crb through its PBM (14, 15), was recruited to
the membrane by CrbMyc-Intrain a PBM-dependent, but FBM-
independent manner (Fig. 5F). Further supporting a role for
Crb’s FBM in Ex binding, we mapped the Crb-interacting region
in Ex to Ex’s N-terminal half, which contains the FERM domain
(Fig. 5G).
To examine direct binding between Ex and Crb, we conducted
GST pull-down assays between GST-Crb-intra (with and without
mutation of FBM or PBM) and cell lysates expressing epitope-
tagged Ex (Fig. 5H) or bacterially purified Ex proteins (Fig. 5I).
In both assays, GST-CrbIntrainteracted with Ex in a FBM-
dependent, but PBM-independent, manner (Fig. 5 H and I). In
contrast, GST-CrbIntradid not interact with Mer (Fig. S1 C and
D). Taken together, we suggest that the intracellular domain
of Crb may be separated into two functional motifs, with the
FBM and the PBM mediating Ex- and Sdt-binding, respectively.
Crb Is Required for Apical Membrane Localization of Ex in Epithelial
Cells. Given the apical localization of Crb (13) and Ex (23) and
the Crb-Ex binding aforementioned, we examined whether Crb is
required for the apical localization of Ex in imaginal discs. In-
deed, in crb mutant clones, Ex staining was no longer restricted
to apical surface; instead, Ex staining extended more basolat-
A
A’
A”
crb
B
ex-lacZ
B’
C
C’
C”
kibra crb
D
ex-lacZ
D’
D”
kibra ΔFBM
E
ex-lacZ
E’
ex-lacZ
kibra ΔPBM
E”
B”
kibra
ex-lacZ
Fig. 3.
In all panels, mutant clones (GFP-negative) are indicated with arrows. Eye
discs containing crb (A–A″), kibra (B–B″), kibra crb (C–C″), kibra crbΔFBM(D–D″),
and kibra crbΔPBM(E–E″) clones were stained for ex-lacZ (red). Note elevated
ex-lacZ levels in kibra crb and kibra crbΔFBM, but not in crb, kibra, or kibra
crbΔPBMclones.
The FBM is required for Crb’s function in regulating Hippo signaling.
CD
EFG
mer
kbr ex
mer;crb
kibra crbex;crb
B
A
A’
A”
Fig. 4.
ommatidial cell number. (A–A″) A pupal eye containing crb82-04clones (GFP-
negative) (A), and stained for Dlg (A’). Superimposed image is shown in A″.
(B–G) Pupal eyes of the indicated genotype and stained for Dlg. The geno-
types are: (B) y w ey-flp Ubi-GFP FRT19A /mer4FRT19A, (C) y w ey-flp; FRT
82B kibradel/FRT 82B Ubi-GFP, (D) y w ey-flp; FRT40A exe1/FRT40A Ubi-GFP,
(E) y w ey-flp Ubi-GFP FRT19A /mer4FRT19A; FRT 82B crb82-04/FRT 82B
Ubi-GFP, (F) y w hsp-flp; FRT 82B kibradelcrb82-04/FRT 82B Ubi-GFP, (G) y w
ey-flp; FRT40A exe1/FRT40A Ubi-GFP; FRT 82B crb82-04/FRT 82B Ubi-GFP.
Twenty ommatidial clusters of each genotype were used for counting
interommatidial cells.
Crb and Ex function in a common pathway to regulate inter-
10534
| www.pnas.org/cgi/doi/10.1073/pnas.1004279107 Ling et al.

Page 4

erally (Fig. 6 A–B”). Of note, loss of crb did not affect the apical
localization of the related FERM protein Mer (Fig. S3). Because
increased Ex levels per se do not perturb the apical restriction of
Ex (4, 12), we conclude that the mislocalization of Ex in crb
mutant clones reflects a specific requirement for Crb in localizing
Ex to apical membranes of epithelial cells.
Recently, we used genomic engineering to introduce mutation
of the FBM or the PBM into the endogenous crb locus (24).
These crb alleles (crbY10AP12AE16Aand crbdelERLI) carry mutations
identical to the FBM and the PBM mutations used in our bio-
chemical analysis, and for simplicity, they will be referred to as
crbΔFBMand crbΔPBM, respectively. These designer alleles allowed
us to examine the respective contribution of Crb’s FBM and
PBM to Ex localization. Consistent with a role for Crb’s FBM in
concentrating Ex to apical membranes, Ex was mislocalized to
basolateral domains in crbΔFBMclones in the wing (Fig. 6 C–D”).
In contrast, crbΔPBMclones in the wing showed a relatively nor-
mal apical localization of Ex (Fig. 6 E–F”). In the eye disk,
however, crbΔPBMclones did show a consistent mislocalization of
Ex to basolateral domains, although this mislocalization was
weaker than that observed in crbΔFBMclones (Fig. S4). The un-
derlying reason for such tissue specificity is unclear and warrants
further investigation. Taken together with our biochemical anal-
ysis, we conclude that Crb is required to concentrate Ex in the
apical domain of epithelial cells, predominantly through direct
binding between Crb’s FBM and Ex’s FERM domain.
FBM Is Required for Crb’s Function in Hippo Signaling and Growth-
Suppression. Besides revealing a differential requirement for
FBM and PBM in Ex localization, the crbΔFBMand crbΔPBM
Ex(HA)
Vector
WT
ΔFBM
ΔPBM
V5-Ex
Myr-V5-Ex
CIP
+
-
++
-
-
-
-
+
Crb (Myc)
Myc-Intra
Ex(V5)
Wts(HA)
Ex(HA)
Crb(Myc)
P-Wts
HA-Wts + + + +
HA-Ex - - + +
Crb - + - +
Myc-Intra
A
Crb
HA-Ex
Myc-Intra
Ex(HA)
CIP
--+--
-
+
+
+
+
+
+
+
-
+
IP-HALysate
B
C
D
T C M
Vector
T C M T C M T C M
Crb
Myc-Intra
Crb
Myc-IntraΔFBM
Crb
Myc-IntraΔPBM
Ex(HA)
β-Tubulin
Crb(Myc)
E
α-ExN(V5)
GST input
ExN
GST input
Input (Ex)
GST-Crb-intra
WT
ΔFBM
1 2 3 4
1 2 3 4 5
1 2 3 4
1 2 3
1 2 3 4 5 6 7 8 9 10 11 12
H
I
1 2 3 4
1 2 3 4
TCMTCM
Vector
Crb (Myc)
ExN(V5)
β-Tubulin
Crb(Myc)
ExC(V5)
TCMTCM
Vector
β-Tubulin
Crb
Myc-Intra
Crb
Myc-Intra
G
1 2 3 4 5 6 7 8 9 10 11 12
Sdt(FLAG)
Crb(Myc)
β-Tubulin
T C M
Vctor
T C M T C M T C M
Crb
Myc-Intra
F
Crb
Myc-IntraΔFBM
Crb
Myc-IntraΔPBM
1 2 3 4 5 6 7 8 9 10 11 12
GST
ΔPBM
GST-Crb-intra
WT
ΔFBM
GST
ΔPBM
Input (Ex)
Crb
Myc-Intra
Fig. 5.
phosphorylation in conjunction with Ex. S2R+ cells expressing the indicated constructs were probed with the indicated antibodies. Note the increased P-Wts
signal in lane 4 compared with lanes 2 and 3. (B) HA-Ex immunoprecipitates were treated with or without Alkaline Phosphatase (CIP) (lanes 1–3). Note the
reversal of CrbMyc-Intra-induced Ex mobility shift by CIP treatment. Ex in S2R+ cell lysates also showed Crb-induced mobility shift (lanes 4–5). (C) The FBM, but not
the PBM, is required for CrbMyc-Intrato promote Ex phosphorylation. (D) Membrane targeting of Ex via fusion with a myristylation signal induced Ex phos-
phorylation (lanes 1–2), which was reversed by CIP treatment (lane 3). (E) The FBM is required for CrbMyc-Intrato recruit Ex to plasma membrane. S2R+ cells
expressing HA-Ex with the indicated constructs were subjected to cell fractionation. Cytosol (C), membrane (M), and a portion of the total lysate (T) were
probed with indicated antibodies. Note that CrbMyc-Intra(lanes 4–6) or CrbMyc-IntraΔPBM(lanes 10–12), but not CrbMyc-IntraΔFBM(lanes 7–9), relocated Ex from
cytosol to membrane (compare with lanes 1–3). (F) Similar to (E) except that Sdt was analyzed. Note that CrbMyc-Intra(lanes 4–6) or CrbMyc-IntraΔFBM(lanes 7–9),
but not CrbMyc-IntraΔPBM(lanes 10–12), relocated Sdt from cytosol to membrane (compare with lanes 1–3). (G) CrbMyc-Intrainteracts with Ex’s N-terminal half (ExN:
aa1-709; lanes 1–6) but not with Ex’s C-terminal half (ExC: aa710-1427; lanes 7–12) in recruitment assay. (H) Physical association between Crb and Ex in vitro. Cell
lysates containing V5-ExN were incubated with glutathione beads containing purified GST (as a control), GST-Crbintra, GST-CrbintraΔFBMor GST-CrbintraΔPBM. V5-
ExN associated with the beads was probed with α-V5 antibody. Note that GST-Crbintraand GST-CrbintraΔPBMbound to Ex, but GST and GST-CrbintraΔFBMdid not. (I)
Purified Crb and Ex bind to each other in vitro. Similar to H except that bacterially purified Myc-ExN was incubated with glutathione beads containing the
respective GST fusion proteins, and Ex associated with the beads (pull-down) was probed with α-Myc antibody. Note that GST-Crbintraand GST-CrbintraΔPBM
bound to Ex, but GST and GST-CrbintraΔFBMdid not.
Crb binds to Ex through its FBM. (A–D) Hyperphosphorylated Ex is indicated by a filled circle next to the protein band. (A) CrbMyc-Intrapromotes Wts
Ling et al.PNAS
| June 8, 2010
| vol. 107
| no. 23
| 10535
DEVELOPMENTAL
BIOLOGY

Page 5

alleles allowed us to dissect the relative contribution of each
motif in regulating Hippo signaling. For this purpose, we took
advantage of the fact that although neither kibra nor crb single
mutant clones exhibited visible up-regulation of ex transcription
(Fig. 3 A–B”), kibra crb mutant clones showed robust elevation
of ex transcription (Fig. 3 C–C”). Given the modest effect of crb
mutation on Hippo target genes (Fig. 2), the kibra crb double
mutant combination provided a more robust and reliable assay
for crb function. Consistent with a critical role for the FBM in
mediating Crb’s input into the Hippo pathway, kibra crbΔFBM
clones showed increased ex transcription similar to that observed
in kibra crb clones (Fig. 3 D–D”). In contrast, kibra crbΔPBM
clones did not show significant increase of ex transcription (Fig. 3
E–E”). These findings reinforce our biochemical and localization
studies implicating Crb in Ex binding and subcellular localization.
Consistent with a role for Crb’s FBM in Hippo signaling, we
found that adult wings composed predominantly of crbΔFBM
mutant cells were larger than normal (116% of normal; n = 20,
t test: P = 0.01), whereas crbΔPBMhad no significant effect on
wing size (104% of normal; n = 20, t test: P = 0.19). These
observations further support our conclusion that the growth-
suppressive function of Crb is mediated predominantly through
FBM-Ex interactions.
Discussion
Compared with the core kinase cascade leading from Hpo to Yki
phosphorylation, signaling events upstream of Hpo are less well
understood. Several apical membrane-associated cytoplasmic
proteins function as tumor suppressors that act synergistically to
regulate the Hippo kinase cascade, suggesting that the apical
membrane represents a “subcellular niche” for Hippo pathway
regulation. Understanding how these apical tumor suppressor
proteins are targeted to the apical membranes may shed light on
the molecular regulation of the Hippo signaling pathway.
Although the transmembrane protein Ft has been proposed as
a potential receptor for the Hippo pathway by controlling the
stability/localization of Ex (8–10), no direct physical interactions
between Ft and Ex have been reported. Indeed, our own assays
did not support a direct physical interaction between Ex and Ft,
at least in S2R+ cells (Fig. S2). Using cell fractionation, in vitro
binding and in vivo subcellular localization, we demonstrate di-
rect binding between Crb and Ex. The physiological significance
of this interaction is further supported by tissue overgrowth and
elevated Hippo target genes in crbΔFBMclones. The specific
biochemical and genetic link between Crb and Ex, but not be-
tween Crb and Mer or Crb and Kibra, is consistent with the
view that these apical membrane-associated proteins may inte-
grate distinct upstream signals. We also note that like kibra, ex,
and mer, transcription of crb is increased in Hippo pathway
mutant cells (25). Thus, negative-feedback regulation of up-
stream regulators appears to be a general feature of this pathway
and may provide an important mechanism to maintain a constant
level of Hippo signaling activity in vivo.
How does Crb regulate Ex and Hippo pathway activity? Crb
may function as a scaffold that is merely required to recruit Ex to
apical membrane, making it available for activation by another
receptor such as Ft. Alternatively, Crb may function as a re-
ceptor that directly modulates Hippo signaling via its interaction
with Ex. Yet another possibility is that Crb may function as
a coreceptor for Ft. Although we cannot distinguish between the
first two models at present, we can largely eliminate the third
possibility as we found that the polarized distribution of Dachs,
the most immediate known response to Ft signaling (26), is not
abrogated by loss of Crb (Fig. S5).
Overexpression of full-length Crb or its intracellular domain
was reported to drive tissue overgrowth characterized by loss of
apical-basal polarity, elevated Yki activity and decreased Ex
protein levels (27, 28). Although these observations are sugges-
tive of an oncogenic role for Crb, our current study using loss-of-
function analysis clearly uncovers a tumor suppressor function
for Crb. This discrepancy may be due to distinct mutant-neighbor
interactions encountered in the different studies. In the over-
expression studies, Crb was expressed in whole compartments
(27, 28), whereas our study examined isolated clones—the latter
situation, but not the former, allows interactions between cells
with differential Crb activity. It is also possible that Crb over-
expression causes a dominant-negative or neomorphic pheno-
type. It is worth noting that Crb3, a mammalian homolog of Crb,
has been implicated as a tumor suppressor in mammals (29). It
will be interesting to examine whether Crb3 or other Crb ho-
mologs are required for Hippo signaling in mammals.
We note that Crb’s FBM has been shown to mediate an in-
hibitory interaction wherein binding to the FERM-domain pro-
tein Yurt (a basolateral cytoskeleton protein) is required to re-
strict Crb to the apical domain (30). Accordingly, loss of Yurt
results in expansion of the apical domain (30). The Crb-Ex in-
teractions revealed in the current study is clearly distinct from
the Crb-Yurt interaction as the former represents positive inter-
actions. Besides the inhibitory Crb-Yurt interaction, Crb also
positivelyregulatestheCrb-Sdt-PatjcomplexviaitsPBM(14–16).
How Crb coordinately interacts with thegrowth-regulatory Hippo
crb
B
ΔFBM
D
ΔPBM
F
A
A’A”
Ex
B’B”
CC’C”
D’D”
E’EE”
Ex
Ex
F’ F”
Fig. 6.
cells. In all panels, wing discs were stained with α-Ex antibody (red) and
mutant clones (GFP-negative) are indicated with arrows. (A–A″) A horizontal
section through a wing disk containing crb clones. The optical section was
captured midway down from the apical surface of epithelium. Note in-
creased Ex staining in crb clones. The dotted line in A″ indicates the position
of vertical section in B–B″. (B–B″) A vertical section through the wing disk in
A–A″ (apical is to the top). Ex localization extended more basolaterally in crb
clones. (C–D″) Similar to A–B″ except that crbΔFBMclones were analyzed.
Note the mislocalization of Ex to more basolateral position in crbΔFBMclones.
(E–F″) Similar to A–B″ except that crbΔPBMclones were analyzed. Note the
relatively normal apical localization of Ex in crbΔPBMclones.
Loss of crb results in mislocalization of Ex in imaginal disk epithelial
10536
| www.pnas.org/cgi/doi/10.1073/pnas.1004279107Ling et al.