Chemistry & Biology

Article

Family-wide Investigation of PDZ Domain-MediatedProtein-Protein Interactions Implicates b-Cateninin Maintaining the Integrity of Tight JunctionsTaranjit S. Gujral,1,2,3 Ethan S. Karp,2,3 Marina Chan,1 Bryan H. Chang,2 and Gavin MacBeath1,2,*1Department of Systems Biology, Harvard Medical School, Boston, MA 02115, USA2Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA3These authors contributed equally to this work*Correspondence: gavin_macbeath@harvard.edu

http://dx.doi.org/10.1016/j.chembiol.2013.04.021

SUMMARY

b-catenin is amultifunctional protein that plays a crit-ical role in cell-cell contacts and signal transduction.b-catenin has previously been shown to interact withPDZ-domain-containing proteins through its C termi-nus. Using protein microarrays comprising 206mouse PDZ domains, we identified 26 PDZ-domain-mediated interactions with b-catenin andconfirmed them biochemically and in cellular lysates.Many of the previously unreported interactionsinvolved proteins with annotated roles in tight junc-tions. We found that four tight-junction-associatedPDZ proteins—Scrib, Magi-1, Pard3, and ZO-3—colocalize with b-catenin at the plasma membrane.Disrupting these interactions by RNA interference,overexpression of PDZ domains, or overexpressionof the b-catenin C terminus altered localization ofthe full-length proteins, weakened tight junctions,and decreased cellular adhesion. These results sug-gest that b-catenin serves as a scaffold to establishthe location and function of tight-junction-associ-ated proteins.

INTRODUCTION

b-catenin is a ubiquitously expressed, multifunctional protein

that plays important roles in cell adhesion and signal transduc-

tion (Morin, 1999). It consists of an amino-terminal region con-

taining 12 armadillo repeats and a long carboxy-terminal tail. It

is frequently found at the plasma membrane, where it interacts

with E-cadherin to regulate cell adhesion, but it is also found in

the cytoplasm, where it plays a role in Wnt signaling, and in the

nucleus, where it acts as a transcriptional activator (Kumar

Kundu et al., 2006). Aberrant expression and/or localization of

b-catenin have been implicated in a wide variety of cancers,

most notably colorectal cancer (Abbosh and Nephew, 2005;

Harris and Peifer, 2005; Herynk et al., 2003; Morin, 1999).

Although b-catenin interacts with E-cadherin, adenomatous

polyposis coli (APC), and axin via its armadillo repeats, it also in-

teracts with PDZ-domain-containing proteins. PDZ domains

816 Chemistry & Biology 20, 816–827, June 20, 2013 ª2013 Elsevier

mediate protein-protein interactions by binding the C termini of

their target proteins. It was initially found that the PDZ domain

of dishevelled homolog 1 (Dvl-1) recognizes the C-terminal tail

of b-catenin and that this interaction is required to induce nuclear

accumulation of b-catenin (Axelrod et al., 1998). Subsequently,

several other PDZ-domain-mediated interactions were reported.

For example, TIP-1, which features a single PDZ domain, binds

b-catenin and inhibits b-catenin-dependent transcription (Kana-

mori et al., 2003). Similarly, MAGI-1b, which features six PDZ

domains, complexes with b-catenin at the plasma membrane

during the formation of cell-cell junctions (Dobrosotskaya and

James, 2000).

PDZ domains constitute one of the largest families of protein

interaction modules found in nature. In mammals, they facilitate

a wide range of cellular processes, including protein trafficking,

neuronal signaling, the establishment of cell polarity, and nutrient

uptake in the gut (Nourry et al., 2003). PDZ-domain-containing

proteins often act as molecular scaffolds to colocalize their bind-

ing partners and facilitate signaling (Kim and Sheng, 2004). As

b-catenin is able to interact with a variety of PDZ domains and

as these interactions affect the localization and activity of b-cat-

enin, we asked if a broad and unbiased screen for additional PDZ

domains that recognize b-catenin could uncover previously

unrecognized biological roles for this protein. To this end, we

prepared protein microarrays comprising 206 mouse PDZ do-

mains and probed them with a fluorescently labeled peptide

derived from the C terminus of b-catenin. Our arrays highlighted

nine previously reported interactions, as well as 17 additional in-

teractions, all of which we confirmed and quantified using a so-

lution-phase florescence polarization (FP) assay. In addition, all

26 PDZ domains captured b-catenin from cellular lysates, and

many of the full-length proteins from which these domains

were derived coimmunoprecipitated with b-catenin upon co-

transfection of HEK293 cells.

Interestingly, a survey of gene ontology revealed that many

of the previously unreported interactions involved proteins

that have previously been implicated in tight junctions. By

overexpressing isolated PDZ domains, by overexpressing the

PDZ-binding motif at the carboxyl terminus of b-catenin, or by

knocking down b-catenin, we found that these interactions

play a causal role in forming and maintaining the overall strength

and integrity of tight junctions. In addition, they inhibit b-catenin-

mediated cell adhesion, proliferation, and migration. These ob-

servations are consistent with a model in which b-catenin plays

Ltd All rights reserved

0

40

80

120

160

0 25 50 75 100 125 150

MUPP1

[PDZ Domain] (μM)

0

40

80

120

160

200

0 25 50 75Fluo

resc

ence

pol

ariz

atio

n (m

P)

Scrib

[PDZ Domain] (μM)

0

40

80

120

160

0 25 50 75 100 125Fluo

resc

ence

pol

ariz

atio

n (m

P)

Magi-1

[PDZ Domain] (μM)

0

40

80

120

0 10 20 30 40 50 60

[PDZ Domain] (μM)

Pard-3

Fluo

resc

ence

pol

ariz

atio

n (m

P)

Fluo

resc

ence

pol

ariz

atio

n (m

P)

Magi-1

ScribPard-3

MUPP1

A

B

KD = 13 μMKD = 1 μM

KD = 100 μM KD = 49 μM

Figure 1. Identification of b-Catenin/PDZ-

Domain Interactions Using Protein Microar-

rays and Fluorescence Polarization

(A) Images of PDZ domain microarrays,

probed with a fluorescently labeled peptide

derived from the C terminus of b-catenin. The

Cy5 image (green) shows the placement of the

spots on one array and the 5(6)-TAMRA images

(red) show binding of the peptide to immobilized

PDZ domains. The PDZ domains were spotted in

duplicate from top to bottom as shown. In-

teractions with PDZ domains derived fromMagi-1,

Scrib, Pard3, and Mupp1 are highlighted in white

boxes.

(B) Confirmation and quantification of b-catenin/

PDZ-domain interactions by fluorescence polari-

zation. Saturation binding curves are shown for

PDZ domains derived from Magi-1, Scrib, Pard3,

and Mupp1.

See also Figure S1 and Table S1.

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

an integral role in maintaining cell-cell contacts, participating not

only in adherens junctions but also in tight junctions. Misregula-

tion of these functions affects adhesion, proliferation, andmigra-

tion, consistent with the established role of b-catenin in cellular

transformation.

RESULTS

Family-wide Discovery of PDZ Domains that Recognizeb-CateninWe have previously shown that protein microarrays can be

used to uncover novel interactions between PDZ domains and

synthetic peptides representing the C-terminal tails of proteins

(Stiffler et al., 2006, 2007). In order to obtain a broad and

unbiased view of PDZ-domain-mediated interactions with

b-catenin, we prepared protein microarrays comprising 206

distinct PDZ domains. We then synthesized a fluorescently

labeled peptide that includes the six C-terminal residues of

b-catenin and used this peptide to probe our PDZ domain

microarrays. Upon washing and scanning for fluorescence, our

arrays highlighted 26 distinct interactions (Figure 1A; Figure S1;

Table S1 available online). To our knowledge, only nine interac-

tions between PDZ domains and b-catenin have previously

been reported (Table 1). All nine of these interactions were

Chemistry & Biology 20, 816–827, June 20, 2013

observed on ourmicroarrays and 17 addi-

tional interactions were also uncovered

(Table 1).

As a secondary assay, we used FP

to retest all 26 domain-peptide inter-

actions (Figure 1B). To quantify binding

affinities, we held the fluorescent peptide

at a constant concentration (20 nM)

and introduced purified PDZ domains at

16 concentrations, ranging from low

nanomolar to high micromolar. In all 26

cases, saturation binding was observed.

The domains spanned a wide range of

affinities, with many previously unre-

ported interactions exhibiting moderate to high affinities (KD <

50 mM; Table 1).

Interactions between Full-Length ProteinsBuilding on these domain-peptide interactions, we next asked if

isolated PDZ domains could recognize full-length b-catenin in

the context of complex cellular lysates. Each of the His6-tagged

PDZ domains was incubated with lysate derived from HEK293

cells, which naturally express b-catenin. Upon recovery of the

domains by immobilized metal affinity chromatography (IMAC),

we found that endogenous b-catenin copurified with 24 of the

26 domains (Figure 2A). No b-catenin was recovered when the

PDZ domain was omitted from the pull-down or when noncog-

nate PDZ domains were used (PDZ domains that did not recog-

nize the b-catenin peptide on our microarrays; Figure 2A).

To investigate interactions between full-length b-catenin and

full-length PDZ proteins, we focused on eight proteins for which

full-length human complementary DNAs (cDNAs) were readily

available: Mals2, Grasp55, Tiam2, PDZk7, Mast1, Dvl3,

OMP25, and Scrib. The coding regions for these eight proteins

were subcloned into a mammalian expression vector that ap-

pends an amino-terminal hemagglutinin (HA) tag. HEK293 cells

were then cotransfected with vectors expressing the HA-tagged

proteins and Myc-tagged b-catenin (Figure 2B). PDZ proteins

ª2013 Elsevier Ltd All rights reserved 817

Table 1. Biophysical Interactions between b-Catenin and PDZ

Domains

Previously Reported Interactions KD (mM)

Unreported

Interactions KD (mM)

Tip1 (Kanamori et al., 2003) 0.3 Magi-3 1

Magi-2 (Nishimura et al., 2002) 0.6 Tiam2 1

Magi-1 (Dobrosotskaya and James,

2000)

1 ZO-3 4

Erbin (Ress and Moelling, 2008) 6 Shank3 9

NHERF-1 (Theisen et al., 2007) 20 Scrib 13

LIN7 (Perego et al., 2000) 25 PSD95 31

Dvl1 (Song and Gelmann, 2005) 79 GRASP55 38

Dvl3 (Song and Gelmann, 2005) 155 Mast1 39

Grip1 (Song and Gelmann, 2005) 369 MUPP1 49

PTP-BL 79

Pard3 100

PDZk7 103

Neurabin-1 176

PDZk3 207

Synip 276

PDZ11 333

SAP97 342

See also Tables S2 and S3.

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

were immunoprecipitated using an anti-HA antibody. Empty

Myc-tagged vector served as a negative control. b-catenin has

previously been shown to interact with Dvl3 and Mals2 (Perego

et al., 2000; Song et al., 2000). Both of these interactions were

observed, as expected (Figure 2C). In addition, all six previously

unreported interactions were observed (Figure 2D). A lesser

nonspecific interaction was also observed in negative controls

of Mals2 and OMP25 immunoprecipitates (Figure 2D).

b-Catenin Interacts with Tight-Junction-AssociatedProteinsHaving established that the domain-peptide interactions we

identified on our microarrays are also observed using full-length

proteins, we asked if our microarray results could provide new

insight into the biological role of b-catenin. Using the DAVID inte-

grated data mining environment (Dennis et al., 2003) and the

Gene Ontology (GO) database (http://www.geneontology.org/),

we classified the proteins in Table 1 according to biological pro-

cess, molecular function, and cellular components associated

with each protein (Table S2). Not surprisingly, annotation of bio-

logical process and molecular function varied widely among the

nonhomologous PDZ domain-containing proteins. When all of

the cellular components associated with each PDZ protein

were grouped, however, we observed significant overrepresen-

tation of proteins associated with either tight junctions or their

neurological equivalent, the synaptosome (pz10�5) (Figure 3A).

It is well established that components of adherens junctions,

such as E-cadherin, a-catenin, and p120 catenin, play a role in

the formation of tight junctions (Itoh et al., 1997; Schneeberger

and Lynch, 2004). To date, however, it has only been speculated

that b-catenin is associated with tight junctions; how this occurs

818 Chemistry & Biology 20, 816–827, June 20, 2013 ª2013 Elsevier

has not yet been explored (Dobrosotskaya and James, 2000;

Shin et al., 2006).

To investigate this role for b-catenin, we chose four proteins

in Table 1 that have well-documented roles in the formation or

maintenance of tight junctions (Bilder and Perrimon, 2000; Tsu-

kita et al., 2001): one that has previously been shown to interact

with b-catenin (Magi-1) and three that have not (Pard3, ZO-3,

and Scrib). Each of these proteins features three or more PDZ

domains (Figure 3B). To establish which of these domains recog-

nize b-catenin, we transfected HEK293 cells with the coding re-

gions for Myc-tagged versions of each of the individual PDZ do-

mains as well as with several clusters of adjacent PDZ domains.

Upon immunoprecipitating each domain and immunoblotting for

endogenous b-catenin, we found that b-catenin was bound by

PDZ domains 4 and 5 of Magi-1; PDZ domain 3 of Pard3; PDZ

domains 1, 2, and 3 of Scrib; and PDZ domain 2 of ZO-3. PDZ

domain 6 of Magi-1 and PDZ domain 1 of Pard3 bound b-catenin

weakly (Figure 3C). All of the PDZ domain clusters that included

these cognate domains also bound b-catenin in the coimmuno-

precipitation assay (Figure 3C). Importantly, these biochemical

data are consistent with the biophysical data obtained from the

protein microarray and FP assays (Table 1).

b-Catenin Colocalizes with Tight-Junction-AssociatedProteinsAll four PDZ proteins—Magi-1, Pard3, ZO-3, and Scrib—as well

as b-catenin are endogenously expressed in Madin-Darby

canine kidney (MDCK) cells. Using confocal fluorescence micro-

scopy, we found that b-catenin is largely confined to the plasma

membrane and colocalizes with all four proteins (Figure 4A). In

SW480 colorectal cancer cells, however, b-catenin escapes

the APC/Axin degradation complex and accumulates in the

cytosol and nucleus (Nath et al., 2003; Smith et al., 1993). As ex-

pected, we observed strong localization of b-catenin in both the

cytosol and nucleus in these cells (Figure 4B). Interestingly, all

four tight-junction-associated PDZ proteins do not localize at

the plasma membrane in these cells but are instead found

primarily in the cytoplasm (Figure 4B). This suggests that locali-

zation of b-catenin may affect localization of its PDZ-domain-

containing binding partners. Consistent with this hypothesis,

we found that knocking down the levels of b-catenin in

HEK293 cells resulted in higher cytosolic levels of exogenously

expressed Scrib and Pard3 (Figure S2). Similarly, knocking

down the levels of b-catenin in Caco-2 epithelial cells resulted

in decreased membranous expression of endogenous Scrib

(Figure 4C). Further, consistent with previous studies and the

documented role of Scrib in tight junction assembly, knocking

down levels of Scrib reduced localization of Occludin, a marker

for tight junctions (Figure S3). Knocking down Scrib in MDCK

cells (2-fold), however, did not affect the membranous localiza-

tion of b-catenin, consistent with the notion that b-catenin drives

localization of its tight junction protein binding partners rather

than the other way around (Figure 4D).

Overexpressing Isolated PDZ Domains Affects TightJunctionsTo investigate the functional role of PDZ-domain-mediated inter-

actions in tight junction biology, we performed a series of exper-

iments in which we transfected MDCK cells with constructs

Ltd All rights reserved

IP: Myc

HA-Dvl3Myc-β-catenin

IB: HA

IB: Myc

Myc-β-cateninHA-Tiam2

IB: HA

IB: Myc

IP: Myc

IP: Myc

HA-OMP25Myc-β-catenin

IB: HA

IB: Myc

IP: Myc

IP: Myc

HA-Grasp55Myc-β-catenin

IB: HA

IB: Myc

IP: Myc

IP: Myc

HA-Pdzk7Myc-β-catenin

IB: HA

IB: Myc

IP: Myc

IP: Myc

HA-Mast1Myc-β-catenin

IB: HA

IB: Myc

IP: Myc

IP: Myc

IB: HA

IB: Myc

IP: Myc

IP: Myc

Myc-β-cateninHA-ScrbHA-Dvl3

HA-Grasp55HA-Mals2HA-Mast1HA-OMP25

HA-Pdzk7HA-Tiam2

Myc-β-catenin

HA-Scrb

IB: Myc

Lysates

Lysates

150 100

75

250

50

25

kDa

IB: HA

A

B D

C

+ + + + + + + + + -+ - - - - - - - - -- + - - - - - - - -- - + - - - - - - -- - - + - - - - - -- - - - - + - - - -- - - - + - - - - -- - - - - - + - - -- - - - - - - + - -

+ - + -+- + - Myc-β-catenin

HA-Mals2

IB: HA

IB: MycIP: Myc

+ - + -+- + -

+ - + -+- + -

+ - + -+- + -

+ - + -+- + -

+ - + -+- + -

+ - + -+- + -

+ - + -+- + -

PS

D95

(3/3

)

PD

Zk11

(1/1

)O

MP

25(1

/1)

Scr

ib(3

/4)

PTP

-BL(

2/5)

Mag

i-3(5

/5)

Grip

1(1/

7)M

agi-3

(3/5

)P

DZk

3(3/

6)D

vl1(

1/1)

Dvl

3(1/

1)N

eura

bin-

1(1/

1)

Mag

i-1(6

/6)

Mas

t1(1

/1)

Sha

nk3(

1/1)

Cha

psyn

-110

(3/3

)S

ynte

nin-

2(1/

2)H

is6-

Thio

redo

xin

Syn

ip(1

/1)

PD

Zk7(

1/1)

Gra

sp55

(1/1

)Ti

am2(

1/1)

SA

P(9

7(1,

2,3/

3)M

UP

P!(1

,2,3

/13)

NH

ER

F1(1

,2/2

)P

ard3

(1,2

,3/3

)ZO

-3(1

,2,3

/3)

Cha

psyn

-110

(3/3

)S

ynte

nin-

2(1/

2)H

is6-

Thio

redo

xin

IB:β-catenin

FP Pos

itive

FP Pos

itive

FP Neg

ative

FP Neg

ative

7878

100

IP: Myc IP: Myc

7510075

10075

10075

10075

10075

10075

10075

250 250

10025

5025

50150

Figure 2. Validation of b-Catenin/PDZ-Domain Interactions by His6 Pull-Down and Co-IP Assays

(A) Western blots showing that full-length b-catenin fromHEK293 cells binds to purified PDZ domains in a His6 pull-down assay. Noncognate PDZ domains serve

as negative controls. The numbers in parentheses refer to the PDZ domain number (counting from the N terminus) and the total number of PDZ domains in the full-

length sequence. Data shown are representative images from two independent experiments.

(B) Western blot showing expression of full-length Myc-tagged b-catenin (top) and full-length HA-tagged PDZ-domain-containing proteins (bottom) in whole-cell

lysates derived from transiently transfected HEK293 cells. Data shown are representative images from three independent experiments.

(C and D) Full-length b-catenin associates with full-length HA-tagged PDZ-domain-containing proteins. HEK293 cells were transiently cotransfected with Myc-

tagged b-catenin and full-length HA-tagged PDZ-domain-containing proteins. Myc-tagged b-catenin was immunoprecipitated with an anti-Myc antibody and

copurifying proteins were detected by immunoblotting with an anti-HA antibody. (C) Previously reported interactions between b-catenin and both Dvl3 andMals2

were confirmed. (D) Previously unreported interactions were observed between b-catenin and Scrib, Tiam2, Grasp55, Omp25, Mast1, and Pdzk7. Data shown

are representative images from three independent experiments.

See also Figure S1.

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

expressing isolated Myc-tagged PDZ domains derived from

Magi-1, Pard3, ZO-3, and Scrib. We reasoned that isolated

PDZdomains,whenoverexpressed,would competewith endog-

enous PDZ proteins for binding to b-catenin. They would there-

fore act as dominant negatives. We started by using confocal

microscopy to visualize the subcellular localization of occludin,

a tight junction marker protein (Furuse et al., 1993), as well as

the four PDZ proteins and b-catenin. In untransfected MDCK

cells, Scrib localized at the plasmamembrane (Figure 5A). In cells

transfected with Myc-tagged PDZ3 of Scrib, however, increased

levels of cytosolic Scrib were observed (Figure 5A). In addition,

only the cells expressing PDZ3 exhibited truncated and discon-

tinuous occludin staining at the cell borders (Figure 5A). Similar

results were observed when cognate PDZ domains from Magi-

1, Pard3, and ZO-3 were overexpressed (Figure S4). As a nega-

Chemistry & Biology 20,

tive control, we used PDZ4 of Scrib, which does not recognize

b-catenin. Overexpression of this domain had no appreciable

affect on the localization of either occludin or Scrib (Figure 5B).

Interestingly, none of these PDZ domains had an appreciable

effect on the localization of b-catenin itself, suggesting that the

localization of b-catenin is dominant and affects the localization

of its PDZ domain-containing binding partners.

One way in which the integrity of tight junctions can be quan-

tified is to measure the length of occludin in individual cells (Chen

and Macara, 2006; Wan et al., 1999). Longer continuous

stretches of occludin indicate well-formed tight junctions. To

quantify the effect of disrupting PDZ-protein/b-catenin interac-

tions on tight junction integrity, we measured the mean occludin

length in untransfected MDCK cells and in cells overexpressing

isolated PDZ domains. In every case, overexpressing cognate

816–827, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 819

PDZ1PDZ2

PDZ3PDZ1

-2-3

Negati

ve

PDZ1PDZ2

PDZ3PDZ1

-2-3

Negati

ve

PDZ1PDZ2

PDZ3PDZ4

PDZ1-2

PDZ3-4

Negati

ve

PDZ1PDZ2

PDZ3PDZ4

PDZ6PDZ4

-5

Negati

ve

ZO-3 PDZ1

32-110 196-264 431-504

N C

Guanylate kinase-like

domainSH3PDZ2 PDZ3

ZO-3

Pard-3282-361 469-548 599-684

PDZ1 PDZ2 PDZ3N C

Pard-3

Scrib722-801 856-936 998-1079 1095-1178

PDZ1 PDZ2 PDZ3 PDZ4Leucine-rich

repeatsN C

Scrib

Magi-1

Magi-1

643-715 841-917 949-1065 1140-1214472-54826-105

WWPDZ1 PDZ2 PDZ3 PDZ4 PDZ5 PDZ6Guanylate kinase-like

domainN C

Adherens Junction

Synaptosome

Tight Junction Synapse

PostsynapticMembrane

Cytoskeleton/Cytoplasm

Cell-Cell Junction/Membrane

Endoplasmic Membrane

A B

C

Lysates

Lysates

IP: Myc

IP: Myc

IB: β-catenin

IB: Myc

IB: Myc

IB: β-cateninLysates

Lysates

IP: Myc

IP: Myc

IB: β-catenin

IB: Myc

IB: Myc

IB: β-catenin

Lysates

Lysates

IP: Myc

IP: Myc

IB: β-catenin

IB: Myc

IB: Myc

IB: β-catenin Lysates

Lysates

IP: Myc

IP: Myc

IB: β-catenin

IB: Myc

IB: Myc

IB: β-catenin

5625

7878

2516

56

1616

78

78

78 7878

78

16

56

16

25

16

56

16

16

25

Figure 3. b-Catenin Interacts with Several Tight-Junction-Associated PDZ-Domain-Containing Proteins

(A) Pie chart showing the cellular component classification as reported in the Gene Ontology database (http://www.geneontology.org) of all PDZ-domain-

containing proteins that were identified as binding partners of b-catenin. PDZ-domain-containing proteins that are known to localize in tight junctions are

highlighted in green.

(B) Domain structures of Magi-1, Pard3, Scrib, and ZO-3.

(C) Investigation of which PDZ domains in Magi-1, Pard3, Scrib, and ZO-3 bind full-length endogenous b-catenin in a co-IP assay. HEK293 cells were transiently

transfected with Myc-tagged versions of isolated PDZ domains or PDZ domain clusters derived from Magi-1, Pard3, Scrib, and ZO-3. PDZ domains were

immunoprecipitated using an anti-Myc antibody and interactions were detected by immunoblotting for b-catenin. Whole-cell lysates showing expression of the

Myc-tagged PDZ domains and endogenous b-catenin are also displayed. Data shown are representative images from at least three independent experiments.

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

PDZ domains caused a significant decrease in the length of oc-

cludin per cell (p < 0.05; Figure 5C), whereas overexpressing the

noncognate PDZ4 of Scrib had no effect. The largest effect was

observed when PDZ6 of Magi-6 was overexpressed, consistent

with the observation that this is the highest affinity interaction

(KD = 1 mM).

Tight junction integrity can also be assessed by measuring

transepithelial electrical resistance (TEER). To assess the effect

of disrupting PDZ domain-mediated interactions with b-catenin,

we measured TEER in untransfected MDCK cells and in cells

transfected with isolated PDZ domains. As with the occludin

assay, overexpressing PDZ domains that recognized b-catenin

resulted in significant decreases in tight junction integrity (p <

0.05; Figure 5D), whereas overexpressing PDZ4 of Scrib had

no effect.

To delineate whether b-catenin plays a role in maintaining tight

junctions or forming them,we performed a calcium switch assay,

820 Chemistry & Biology 20, 816–827, June 20, 2013 ª2013 Elsevier

commonly used to monitor the development of tight junctions in

MDCK cells. In the Ca2+-switch assay, we maintained MDCK

monolayers in low calcium medium and subsequently trans-

ferred them to normal calcium-containing medium, which in-

creases cell-cell contacts and initiates assembly of junctional

complexes (Figure 5E). At selected times, cells were fixed and

the presence of tight junctions was monitored by visualizing

localization of occludin by immunofluorescence. Before the

Ca2+-switch (0 hr), the tight junction protein occludin showed a

discontinuous staining pattern in both parental MDCK cells

and in cells overexpressing isolated PDZ domains (Figures 5E

and 5F). In every case, overexpressing cognate PDZ domains

caused a significant decrease in the length of occludin per cell

(p < 0.05), whereas overexpressing the noncognate PDZ4 of

Scrib had no effect (Figure 5F).

b-catenin has been implicated in wide variety of cellular pro-

cesses, including cell adhesion, proliferation, and migration

Ltd All rights reserved

Figure 4. Localization of b-Catenin and Tight-Junction-Associated PDZ Proteins in Epithelial Cells

(A) Confocal images showing that b-catenin (red) colocalizes with the PDZ-domain-containing tight junction proteins Magi1, Pard3, Scrib, and ZO-3 (green) in

MDCK kidney epithelial cells. Scale bar is 10 mm.

(B) In SW480 colon cancer cells, b-catenin (red) is mostly localized in the cytoplasm and nucleus, whereas the PDZ domain-containing tight junction proteins

Magi1, Pard3, Scrib, and ZO-3 are localized in the cytoplasm. Themerged images showing both b-catenin and the PDZ-domain-containing tight junction proteins

are displayed to the right. Yellow indicates colocalization. Nuclei are stained in blue (Hoechst dye). Data shown are representative images from at least two

independent experiments. Scale bar is 10 mm.

(C) Confocal images showing b-catenin (red) colocalizes with the PDZ-domain-containing tight junction protein Scrib (green) in Caco-2 epithelial cells (top).

Knocking down the levels of b-catenin results in decreased membranous expression of endogenous Scrib (bottom).

(D) Knocking down the levels of Scrib in MDCK cells does not affect the membranous localization of b-catenin.

See also Figures S2 and S3.

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

(Blankesteijn et al., 2000; Gavert and Ben-Ze’ev, 2007). To

further explore the functional relevance of our interactions, we

assessed the ability of MDCK cells expressing dominant nega-

tive PDZ domains to attach to extracellular matrix. Untransfected

MDCK cells adhere more strongly to Collagen I than to Fibro-

nectin or Laminin I (Figure 6A). Consistent with the loss of tight

junction integrity, MDCK cells expressing b-catenin-binding

PDZ domains exhibited a decreased ability to attach to all three

matrices (Figure 6A). When grown on Fibronectin, MDCK cells

overexpressing PDZ domains from Pard3, Scrib, and ZO-3

showed significantly less adhesion whereas MDCK cells overex-

pressing PDZ domains from Magi-1, Pard3, Scrib, and ZO-3

Chemistry & Biology 20,

all showed loss of adhesion on Collagen I. Overexpression of

PDZ domains from Magi-1, Pard3, and ZO-3 displayed loss of

adhesion on Laminin I, whereas MDCK cells expressing the non-

cognate PDZ4 of Scrib showed no change in their adhesion

properties relative to untransfected cells.

In a similar fashion, we also assessed the effect of expressing

isolated PDZ domains on cell proliferation using an MTT viability

assay and on cell migration using aBoyden chamber assay. Both

proliferation and migration increased when the dominant-

negative PDZ domains were overexpressed; no change was

observed when the control PDZ domain was overexpressed

(Figure 6B; Figure S5).

816–827, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 821

Figure 5. Overexpression of b-Catenin-Interacting PDZ Domains Affects the Strength of Tight Junctions

(A) Confocal images showing membranous localization of endogenous Scrib MDCK cells (i) and cytosolic localization of Myc-tagged PDZ3 of Scrib (ii, v, and viii).

MDCK cells expressing PDZ3 of Scrib also showed altered expression of the tight junction marker occludin (iii and iv) and endogenous Scrib (vi and vii). The white

arrows indicate localization of occludin in cells that do not expressMyc-tagged PDZ3 of Scrib or PDZ6 ofMagi-1. The yellow arrows indicate occludin localization

in cells expressing PDZ3 of Scrib. Themembranous localization of endogenous b-catenin in cells expressing PDZ3 of Scrib is also shown (ix and x). Blue indicates

nuclei (Hoechst stain), and yellow indicates colocalization. Scale bars are 10 mm.

(B) MDCK cells expressing PDZ4 of Scrib (i, iv, and vii) do not show altered expression or localization of either occludin (ii and iii) or endogenous Scrib (v and vi).

The membranous localization of endogenous b-catenin in cells expressing PDZ4 of Scrib is also shown (viii and ix).

(legend continued on next page)

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

822 Chemistry & Biology 20, 816–827, June 20, 2013 ª2013 Elsevier Ltd All rights reserved

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

Overexpressing the PDZ Domain-Binding Motif at theCarboxyl Terminus of b-Catenin Affects Tight JunctionsAs PDZ domains are often able to recognize more that one pro-

tein, it is possible that overexpressing isolated PDZ domains also

affects interactions with proteins other than b-catenin. As the

converse of the experiments detailed in Figure 5, we sought to

perturb specific interactions between b-catenin and tight-junc-

tion-associated PDZ proteins by overexpressing GFP displaying

the carboxyl terminus of b-catenin (-GWFDTDL-CO2H) (Fig-

ure 6A). Our biochemical data show that this PDZ-domain-bind-

ing motif is capable of binding to tight-junction-associated PDZ

proteins (Figure 1). Thus, overexpressing this GFP construct

(GFP-bcat-Cterm) should act as a dominant negative by

competing with endogenous full-length b-catenin in binding to

tight-junction-associated PDZ proteins. We performed the

same series of experiments as described above, but this time

transfecting MDCK cells with GFP-bcat-Cterm or with GFP-

mut-Cterm, a nonbinding mutant version of GFP-bcat-Cterm in

which the last three residues were mutated to alanine (NH2-

GWFDAAA). In untransfected MDCK cells, endogenous Scrib

localized at the plasma membrane, as observed by confocal mi-

croscopy (Figure 6B). In cells transfected with GFP-bcat-Cterm,

however, increased levels of cytosolic Scrib were observed (Fig-

ure 6B). Further, GFP-bcat-Cterm and endogenous Scrib were

observed to colocalize in the cytosol, consistent with the hypoth-

esis that the C-terminal tail of b-catenin drives the localization of

Scrib (Figure 6B). In contrast, overexpressing GFP-mut-Cterm

had no appreciable affect on the subcellular localization of Scrib

and did not colocalize with endogenous Scrib (Figure 6B).

Next, we analyzed the de novo assembly of tight junctions using

the calcium switch assay, as described above. Overexpressing

GFP-bcat-Cterm caused a significant decrease in the average

length of occludin per cell, whereas overexpressing GFP-mut-

Ctermhadnoeffect (p<0.05; Figure6C).Consistentwith these re-

sults, transmission electron microscopy (TEM) images of MDCK

cells overexpressing GFP-mut-Cterm showed the presence of

typical tight junctions, as well as adherens junctions, in sites of

cell-cell contact (Figure 6D). In contrast, cells overexpressing

GFP-bcat-Cterm showedweaker or no tight junctions (Figure 6D).

Consistent with the loss of tight junction integrity, MDCK cells

expressing GFP-bcat-Cterm also exhibited a decreased ability

to attach to extracellular matrices (Figure 6E). When grown on

Fibronectin,Collagen I, or Fibrinogen,MDCKcells overexpressing

GFP-bcat-Cterm exhibited significantly less adhesion, whereas

MDCK cells expressing GFP-mut-Cterm showed little or no

(C) Quantification of occludin localization in parental MDCK cells or in cells expr

Occludin length was measured using ImageJ (http://rsbweb.nih.gov/ij/). Error ba

(D) Transepithelial electrical resistance (TEER), measured in parental MDCK cells a

least three biological replicates, error bars represent the SEM, and asterisks den

(E) A Ca2+-switch assay was performed with parental or MDCK cells expressing M

stained for occludin and Myc tag. Confocal images of occludin (green) and Myc

(F) Graphs show the results of quantification of occludin recruitment to cell-cell con

is the mean of at least three biological replicates, and error bars represent SEM.

(G) Relative adhesion of parental MDCK cells and cells expressing isolated PDZ

measured 48 hr posttransfection. Adhesion values are the means of at least thre

significant difference from mock-transfected parental cells (p < 0.05).

(H) Relative proliferation of parental MDCK cells and cells transfected with isolate

replicates, error bars represent the SEM, and asterisks denote a significant diffe

See also Figures S4 and S5.

Chemistry & Biology 20,

change in their adhesion properties relative to untransfected cells

(Figure 6E). Finally, overexpressing GFP-bcat-Cterm caused sig-

nificant (p < 0.05) increases in migration and invasion; no signifi-

cant change was observed with the mutant control (Figures 6F

and 6G). Taken together, these results show that PDZ-domain-

mediated interactionsbetweenb-cateninand tight-junction-asso-

ciated proteins affect a variety of b-catenin-mediated functions.

DISCUSSION

b-catenin is a central component of the cadherin cell adhesion

complex and plays an essential role in Wnt signaling. Previous

studies have shown that b-catenin is a physiological ligand of

PDZ domains. Here, we report a family-wide, unbiased examina-

tion of PDZ-domain-mediated interactions with b-catenin. Using

PDZ domain microarrays, we rediscovered all nine previously re-

ported b-catenin/PDZ-domain interactions and uncovered 17

additional interactions. All 24 PDZ domains were able to pull

down full-length b-catenin from cell lysates. In addition, when

we tested a representative sample of eight full-length PDZ

domain-containing proteins, all eight copurified with b-catenin

from cotransfected cells. It is therefore likely that most, if not

all, of the previously unreported interactions can occur in the

complex environment of the cell.

The full set of PDZ-domain-mediated interactions with b-cate-

nin can be visualized by constructing a b-catenin-centric

network diagram (Figure S6). This diagram includes all of the

PDZ proteins that were found to interact with b-catenin as well

as other proteins that have previously been identified as binding

partners for the PDZ proteins. This diagram offers further insight

into the biological role of b-catenin. For example, b-catenin has

previously been associated indirectly with PSD95 and Tiam2

through Lin7 and Src (Gottardi and Gumbiner, 2004; Piedra

et al., 2003). Here, we see that PDZ domains in PSD95 and

Tiam2 directly recognize b-catenin. Given the high connectivity

of this network, along with the fact that many of these proteins

are expressed in the same cells or tissues (Fanning and Ander-

son, 1999) (Table S3), it is likely that several different proteins

are interacting with b-catenin in the same cell at the same

time. One way in which these competing interactions could be

regulated is through subcellular localization of b-catenin. Some

PDZ proteins are confined to the cytoplasm, where they could

affect the signaling function of b-catenin, whereas others are

confined to the plasma membrane, where they could affect its

ability to stabilize cell-cell junctions.

essing isolated PDZ domains derived from tight-junction-associated proteins.

rs indicate the SEM of at least four measurements.

nd in cells expressing individual PDZ domains. TEER values are the mean of at

ote a significant difference from mock-transfected parental cells (p < 0.05).

yc-Scrib (PDZ3) and Myc-Scrib (PDZ4) constructs. Cell layers were fixed and

tag-scrib PDZ domain (red) are shown.

tacts in parental or MDCK cells overexpressing isolated PDZ domains. The bar

domains grown on Fibronectin, Collagen I, and Laminin I. Cell adhesion was

e biological replicates, error bars represent the SEM, and asterisks denote a

d PDZ domains. Each proliferation value is the mean of at least three biological

rence from mock-transfected parental cells (p < 0.05).

816–827, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 823

Rel

ativ

e m

igra

tio (A

.U)

1.6

1.2

0.8

0.4

0.0

control

2.4

2.0

0.0

0.4

0.8

1.2

1.6

Rel

ativ

e in

vasi

on (A

.U)

Rel

ativ

e ad

hesi

on (A

.U) 1.0

0.8

0.6

0.4

0.2

0.0

Fibron

ectin

Collag

en I

Fibrino

gen

*

*

*

*

GFPN WFDTDL CPDZ binding

motif

GFPN WFDAAA C

GFP-β-cat-Cterm

GFP-mut-CtermPDZ binding

motif

Scrib GFP Merge

β-ca

teni

nc-

term

β-

cate

nin

mut

c-te

rm

Colocolization

50

100

150

300

0

Occ

ludi

n le

ngth

/cel

l (pi

xels

)

250

200

1hr Ca2+ 2hr Ca2+

GFP-β-cat-CtermGFP-mut-Cterm

0hr Ca2+

** **

parental

GFP-β-cat-Cterm GFP-β-cat-CtermGFP-mut-Cterm

GFP-β-cat-CtermGFP-mut-Cterm

A B

D

F G

C

E

Figure 6. Overexpressing the PDZ-Domain-Binding Motif of b-Catenin Affects Tight Junctions

(A) Schematic showing the carboxyl terminus of b-catenin (GFP-bcat-Cterm: NH2-GWFDTDL-COOH) andGFP-mut-Cterm, a nonbindingmutant version in which

the last three residues were mutated to alanine (NH2-GWFDAAA-COOH).

(B) Confocal images showing MDCK cells expressing either GFP-bcat-Cterm or GFP-mut-Cterm and endogenous subcellular localization of Scrib. GFP-bcat-

Cterm and endogenous Scrib were observed to colocalize in the cytosol (yellow); GFP-mut-Cterm, on the other hand, had no appreciable affect on the subcellular

localization of Scrib and did not colocalize with endogenous Scrib.

(C) Bar graph showing quantification of occludin length in a Ca2+-switch assay. Overexpressing GFP-bcat-Cterm caused a significant decrease in the average

length of occludin per cell, whereas overexpressing GFP-mut-Cterm had no effect (p < 0.05). The bar is the mean of at least three biological replicates, and error

bars represent SEM.(legend continued on next page)

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

824 Chemistry & Biology 20, 816–827, June 20, 2013 ª2013 Elsevier Ltd All rights reserved

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

Nevertheless, the PDZ proteins that interact with b-catenin

each exhibit distinct expression profiles across different cell

types and tissues (Table S3). It is therefore likely that the tissue-

specific expression of these PDZ proteins affects the role that

b-catenin plays in each cell type. For example, Magi-1, Magi-2,

Magi-3, and Scrib help recruit synaptic vesicles to the neuronal

synapse (Ito et al., 2012; Oliva et al., 2012; Roche et al., 2002)

and all four of these proteins interact with b-catenin through their

PDZ domains (Table 1). As they are all expressed at moderate to

high levels in brain tissue (Table S3), these interactionsmay guide

the role that b-catenin plays at the synaptic junction. Indeed,

Magi-1 and Scrib have previously been shown to colocalized at

synapses with b-catenin in cultured rat hippocampal neurons

and b-catenin coimmunoprecipitates with Magi-1 and Scrib in

neuronal cell lysates (Nishimura et al., 2002; Perego et al.,

2000; Sun et al., 2009). Additionally, overexpressing the C-termi-

nal region of b-catenin blocks synaptic targeting of Magi-1,

suggesting that the PDZ-domain-mediated interaction with

b-catenin is critical for its synaptic targeting. In a similar fashion,

Dvl-1, Dvl-3, and TIP-1 play important roles inWnt signaling (Gao

and Chen, 2010; Zhang et al., 2008) and each of these proteins

interacts with b-catenin through their PDZ domains (Table 1).

We would therefore expect that these proteins would be coex-

pressed with b-catenin in cells regulated by Wnt signaling. It is

interesting to note that Dvl-1 is highly expressed in intestinal tis-

sue, Dvl-3 in colon tissue, and TIP-1 in bone (Table S3), where

Wnt signaling plays prominent and well-documented roles (Hae-

gebarth and Clevers, 2009; Kramer et al., 2010).

Based on the overrepresentation of tight junction proteins in

our interaction data set (MUPP1, Pard3, Scrib, Magi-1, Magi-3,

ZO-3, and ZO-1), we focused on four specific PDZ domain-con-

taining proteins—Magi-1, Pard3, ZO-3, and Scrib—to determine

if their interactions with b-catenin play a functional role in tight

junction biology. Tight junctions provide contacts between

neighboring epithelial cells and act as the primary barriers to

the diffusion of solutes through the intercellular space (Tsukita

et al., 2001). They differ both in composition and purpose from

adherens junctions, which usually occur basal to tight junctions.

PDZ-domain-containing proteins have been shown to play

crucial roles in the proper functioning of tight junctions. For

example, Mupp1 serves as a multivalent scaffold to recruit

several essential tight junction proteins, including claudin and

junctional adhesion molecule (Hamazaki et al., 2002).

To delineate the role of b-catenin in tight junction biology, we

first determined which PDZ domains in each protein interact

directly with b-catenin (Figures 3B–3D). Then, using these iso-

lated domains as dominant negatives, we assessed the effect

of disrupting these interactions on tight junction function. This

strategy has previously been used to study protein function:

overexpressing the PDZ domain of ZO-1 causes the full-length

(D) TEM images ofMDCK cells overexpressingGFP-mut-Cterm showed the prese

arrow), in sites of cell-cell contact. In contrast, cells overexpressing GFP-bcat-C

(E) MDCK cells expressing GFP-bcat-Cterm also exhibited a decreased ability t

Fibrinogen, MDCK cells overexpressing GFP-bcat-Cterm exhibited significantly le

no change in their adhesion properties relative to untransfected cells. The bar is

(F and G) Overexpressing GFP-bcat-Cterm caused significant increases in migrat

control. Each bar is the mean of at least three biological replicates, error bars

transfected parental cells (p < 0.05).

Chemistry & Biology 20,

protein to mislocalize and results in an epithelial to mesenchymal

transition (Reichert et al., 2000; Ryeom et al., 2000). Using

confocal microscopy, we found that overexpressing cognate

PDZ domains from all four proteins increased the cytoplasmic

localization of their endogenous, full-length proteins and caused

aberrant localization of occludin. In addition, TEERwas reduced,

indicating a loss in tight junction integrity. These effects were not

observed when a PDZ domain that does not recognize b-catenin

was overexpressed. From these studies, we conclude that PDZ-

domain-mediated interactions with b-catenin determine the sub-

cellular localization of these four tight junction proteins, that

these interactions play a direct role in maintaining the integrity

of tight junctions, and that several different interactions are prob-

ably occurring in the same cell at the same time.

Interestingly, the localization of b-catenin was not affected in

these experiments, suggesting that its localization to the plasma

membrane is not determined by its interactions with these PDZ

proteins. As membrane recruitment of tight junction proteins is

a necessary step in the formation or maintenance of tight junc-

tions, it is likely that specific interactions with b-catenin are

required (Tsukita et al., 2001). Moreover, as b-catenin is associ-

ated with the cytoskeleton, it may act as a scaffold that links tight

junction-associated PDZ proteins to the cytoskeleton.

The functional importance of these interactions is further un-

derscored by our investigations of adhesion, proliferation, and

migration. Using the same dominant-negative strategy, we

found that overexpressing PDZ domains derived from Magi-1,

Pard3, Scrib, and ZO-3 resulted in decreased adhesion to a va-

riety of extracellular matrices and increased rates of proliferation

and migration.

Taken together, our data show that b-catenin plays a role in

defining the subcellular localization of Magi-1, Pard3, ZO-3, and

Scrib and that these interactions contribute to the integrity of tight

junctions, promote cellular adhesion, and inhibit cellular prolifer-

ationandmigration. Theseobservationsareparticularly intriguing

because many colon cancer cells exhibit aberrant accumulation

of b-catenin in the cytosol. In keeping with this, we found that

Magi-1, Pard3, ZO-3, and Scrib were predominantly localized

at theplasmamembrane in normalMDCKcells butwere localized

in the cytosol of SW480 colon cancer cells. This suggests that the

buildup of cytoplasmic b-catenin in colon cancer cells causes the

observed mislocalization of tight-junction-associated PDZ pro-

teins and that this, in turn, promotes decreased adhesion and

increased proliferation and migration. This model is consistent

with a previous study showing that increased tight junction

permeability and decreased epithelial barrier function precede

the development of colon tumors (Soler et al., 1999). We submit

that the mislocalization of tight junction proteins caused by the

misregulation of b-cateninmay play a functional role in the devel-

opment of colon cancer and warrants further investigation.

nce of typical tight junctions (yellow arrow), as well as adherens junctions (white

term showed weaker or no tight junctions.

o attach to extracellular matrices. When grown on Fibronectin, Collagen I, or

ss adhesion, whereas MDCK cells expressing GFP-mut-Cterm showed little or

the mean of at least two biological replicates, and error bars represent SEM.

ion and invasion (p < 0.05); no significant change was observed with the mutant

represent the SEM, and asterisks denote a significant difference from mock-

816–827, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 825

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

SIGNIFICANCE

b-catenin has previously been shown to interact with a vari-

ety of PDZ domain-containing proteins through its C termi-

nus, linking it to a diverse set of biological processes. Using

protein microarrays comprising virtually every mouse PDZ

domain, we identified and validated 26 PDZ domain-

mediated interactions with b-catenin, 17 of which had not

previously been described. Most notably, four tight-junc-

tion-associated PDZ proteins—Scrib, Magi-1, Pard3, and

ZO-3—interact with b-catenin in vitro and colocalize with it

at the plasmamembrane inMDCKepithelial cells. Disrupting

these interactions affects cellular adhesion, proliferation,

and migration, consistent with a model in which b-catenin

plays an integral role in maintaining cell-cell contacts,

participating not only in adherens junctions but also in tight

junctions. These studies show that biochemical interactions

identified in vitro can be used to uncover biology, suggesting

that constructing microarrays of protein interaction do-

mains provides a viableway to segment the problem of iden-

tifying biologically meaningful protein-protein interactions

on a proteome-wide scale.

EXPERIMENTAL PROCEDURES

Expression Constructs

Myc- and HA-tagged expression vectors were constructed using the Invitro-

gen Gateway system. Full-length cDNA for human b-catenin (ID 6151332),

LIN7b (ID 430017), Grasp55 (ID 4298251), Tiam2 (ID 9021272), ZO-3 (ID

40008232), PDZk7 (ID 5179973), Pard3 (ID 30344732), Mast1 (ID 5200610),

Dvl3 (ID 5200610), OMP25 (ID 3952366), and Scrib (ID 100015198) were ob-

tained from Open Biosystems. PCR products were prepared for each coding

region beginning with the sequence CACCand endingwith a stop codon (UAA)

using PFU Ultra polymerase (Stratagene) and the following thermocycling pa-

rameters: 95�C, 5 min; followed by 38 cycles of 95�C, 30 s; 54�C, 30 s; 72�C,1 min; followed by a final 10 min incubation at 72�C. The resulting products

were transferred into the vector pENTR/D-TOPO by topoisomerase-I-medi-

ated directional cloning (Invitrogen). Each clone was verified by DNA

sequencing. The coding region for each PDZ domain or full-length protein

was transferred into Gateway compatible mammalian expression vectors

by l-recombinase-mediated directional subcloning and confirmed by

sequencing. Expression vectors that append amino-terminal Myc and HA

tags were based on pCMV-Myc and pCMV-HA (Clontech Laboratories).

Cell Lines and Reagents

MDCK, HEK293, and SW480 cells were obtained from American Type Culture

Collection and maintained in Dulbecco’s modified Eagle’s medium supple-

mented with 10% (v/v) fetal bovine serum, 2 mM glutamine, 100 IU/mL

penicillin, and 100 mg/mL streptomycin. The expression constructs were intro-

duced into cells using Lipofectamine 2000 (Invitrogen) according to the man-

ufacturer’s instructions.

Primary antibodies were obtained from the following sources: rabbit anti-

b-catenin (Cell Signaling Technology; cat. #9562), mouse anti-b-catenin (Cell

Signaling Technology; cat #2677), mouse anti-Myc (Cell Signaling Technology;

cat #2276), rabbit anti-HA (Cell Signaling Technology; cat #3724), mouse anti-

b-actin (Sigma-Aldrich; cat. #A1978), rabbit anti-Magi-1 (Santa Cruz Bio-

technology; cat. #sc-25663), rabbit anti-Pard3 (Santa Cruz Biotechnology;

cat #sc-98509), rabbit anti-Scrib (Santa Cruz Biotechnology; cat #sc-28737),

and rabbit anti-ZO-3 (Invitrogen, cat #36-4000).

Pull-Down Assays and Western Blotting

Cell extracts for His6 pull-down assays or coimmunoprecipitation (co-IP) as-

says were prepared by rinsing cells with cold PBS and then introducing cell

lysis buffer (Cell Signaling Technology; cat #9803) supplemented with 1 mM

826 Chemistry & Biology 20, 816–827, June 20, 2013 ª2013 Elsevier

phenylmethylsulfonyl fluoride and a cocktail of protease inhibitors (0.5 ml per

well of a six-well plate). Samples were incubated at 4�Con a rotating shaker for

15min, and insoluble material was removed by centrifugation at 12,0003 g for

15min at 4�C. For pull-down experiments, the supernatant wasmixed with pu-

rified His6-tagged PDZ domains immobilized on Ni-NTA Agarose (QIAGEN)

and incubated on a rotating shaker for 2 hr at 4�C. For co-IP experiments,

the supernatant was mixed either with an anti-Myc antibody (Santa Cruz

Biotechnology; cat #SC40) or an anti-HA antibody (Santa Cruz Biotechnology;

cat #SC7392) for 2 hr at 4�C. Protein A/G PLUS-Agarose (Santa Cruz Biotech-

nology) was then added and the slurry incubated for 1 hr on a rotating shaker at

4�C. For both the pull-down and co-IP experiments, the agarose beads were

washed four times with lysis buffer and then boiled for 5 min in NuPAGE SDS

sample buffer (Invitrogen) prior to analysis by western blotting.

For quantitative immunoblots, primary antibodies were detected using IR-

Dye 680-labeled goat-anti-rabbit immunoglobulin G (IgG) or IRDye 800-

labeled goat-anti-mouse IgG (LI-COR Biosciences) at 1:10,000 dilutions.

Bands were visualized and quantified using an Odyssey Infrared Imaging Sys-

tem (LI-COR Biosciences).

TEER Measurement

TEER of the MDCK cell monolayers was monitored using a MillicellERS (Milli-

pore) connected to a pair of chopstick electrodes according to manufacturer’s

instructions. MDCK cells expressing Myc-tagged PDZ domains or empty vec-

tor were seeded in 24-well millicell plates (Millipore) at a density of 13 105 cells

per well. After 3 days, electrical potentials obtained from blank inserts were

subtracted from those obtained from inserts with confluent monolayers.

Monolayer resistance values were multiplied by the membrane area and aver-

aged to calculate TEER (Udcm2).

SUPPLEMENTAL INFORMATION

Supplemental Information includes Supplemental Experimental Procedures,

six figures, and four tables and can be found with this article online at http://

dx.doi.org/10.1016/j.chembiol.2013.04.021.

ACKNOWLEDGMENTS

We thank Daniel Goodenough and Marc Kirschner for helpful discussions and

suggestions and Peter Sorger for generous use of his facilities. We also thank

the staff of the Harvard Center for Biological Imaging and the Nikon Imaging

Center at Harvard Medical School for help and support. This study was sup-

ported by awards from the WM Keck Foundation and the Camille and Henry

Dreyfus Foundation and by grants from the National Institutes of Health (R01

GM072872, R33 CA128726, and P50 GM068762). T.S.G. is a Human Frontier

Science Program fellow, and E.S.K. is a recipient of a National Science Foun-

dation graduate fellowship. The funders had no role in study design, data

collection and analysis, decision to publish, or preparation of the manuscript.

T.S.G., E.S.K., and G.M. conceived and designed the research; T.S.G. and

G.M. analyzed the data and wrote the manuscript; and T.S.G., E.S.K., M.C.,

and B.H.C. performed the experiments and analyzed the data.

Received: January 13, 2013

Revised: March 26, 2013

Accepted: April 18, 2013

Published: June 20, 2013

REFERENCES

Abbosh, P.H., and Nephew, K.P. (2005). Multiple signaling pathways converge

on beta-catenin in thyroid cancer. Thyroid 15, 551–561.

Axelrod, J.D., Miller, J.R., Shulman, J.M., Moon, R.T., and Perrimon, N. (1998).

Differential recruitmentofDishevelledprovidessignalingspecificity in theplanar

cell polarity and Wingless signaling pathways. Genes Dev. 12, 2610–2622.

Bilder, D., and Perrimon, N. (2000). Localization of apical epithelial determi-

nants by the basolateral PDZ protein Scribble. Nature 403, 676–680.

Blankesteijn, W.M., van Gijn, M.E., Essers-Janssen, Y.P., Daemen, M.J., and

Smits, J.F. (2000). Beta-catenin, an inducer of uncontrolled cell proliferation

Ltd All rights reserved

Chemistry & Biology

b-Catenin Helps Maintain Tight Junctions

and migration in malignancies, is localized in the cytoplasm of vascular endo-

thelium during neovascularization after myocardial infarction. Am. J. Pathol.

157, 877–883.

Chen, X., andMacara, I.G. (2006). Par-3 mediates the inhibition of LIM kinase 2

to regulate cofilin phosphorylation and tight junction assembly. J. Cell Biol.

172, 671–678.

Dennis, G., Jr., Sherman, B.T., Hosack, D.A., Yang, J., Gao, W., Lane, H.C.,

and Lempicki, R.A. (2003). DAVID: Database for Annotation, Visualization,

and Integrated Discovery. Genome Biol. 4, 3.

Dobrosotskaya, I.Y., and James, G.L. (2000). MAGI-1 interacts with beta-cat-

enin and is associated with cell-cell adhesion structures. Biochem. Biophys.

Res. Commun. 270, 903–909.

Fanning, A.S., and Anderson, J.M. (1999). PDZ domains: fundamental building

blocks in the organization of protein complexes at the plasma membrane.

J. Clin. Invest. 103, 767–772.

Furuse, M., Hirase, T., Itoh, M., Nagafuchi, A., Yonemura, S., Tsukita, S., and

Tsukita, S. (1993). Occludin: a novel integral membrane protein localizing at

tight junctions. J. Cell Biol. 123, 1777–1788.

Gao, C., and Chen, Y.-G. (2010). Dishevelled: The hub of Wnt signaling. Cell.

Signal. 22, 717–727.

Gavert, N., and Ben-Ze’ev, A. (2007). beta-Catenin signaling in biological con-

trol and cancer. J. Cell. Biochem. 102, 820–828.

Gottardi, C.J., and Gumbiner, B.M. (2004). Distinct molecular forms of beta-

catenin are targeted to adhesive or transcriptional complexes. J. Cell Biol.

167, 339–349.

Haegebarth, A., and Clevers, H. (2009). Wnt signaling, lgr5, and stem cells in

the intestine and skin. Am. J. Pathol. 174, 715–721.

Hamazaki, Y., Itoh, M., Sasaki, H., Furuse, M., and Tsukita, S. (2002). Multi-

PDZ domain protein 1 (MUPP1) is concentrated at tight junctions through its

possible interaction with claudin-1 and junctional adhesion molecule. J. Biol.

Chem. 277, 455–461.

Harris, T.J., and Peifer, M. (2005). Decisions, decisions: beta-catenin chooses

between adhesion and transcription. Trends Cell Biol. 15, 234–237.

Herynk, M.H., Tsan, R., Radinsky, R., and Gallick, G.E. (2003). Activation of c-

Met in colorectal carcinoma cells leads to constitutive association of tyrosine-

phosphorylated beta-catenin. Clin. Exp. Metastasis 20, 291–300.

Ito, H., Morishita, R., Sudo, K., Nishimura, Y.V., Inaguma, Y., Iwamoto, I., and

Nagata, K.i. (2012). Biochemical and morphological characterization of MAGI-

1 in neuronal tissue. J. Neurosci. Res. 90, 1776–1781.

Itoh, M., Nagafuchi, A., Moroi, S., and Tsukita, S. (1997). Involvement of ZO-1

in cadherin-based cell adhesion through its direct binding to alpha catenin and

actin filaments. J. Cell Biol. 138, 181–192.

Kanamori, M., Sandy, P., Marzinotto, S., Benetti, R., Kai, C., Hayashizaki, Y.,

Schneider, C., and Suzuki, H. (2003). The PDZ protein tax-interacting pro-

tein-1 inhibits beta-catenin transcriptional activity and growth of colorectal

cancer cells. J. Biol. Chem. 278, 38758–38764.

Kim, E., and Sheng, M. (2004). PDZ domain proteins of synapses. Nat. Rev.

Neurosci. 5, 771–781.

Kramer, I.,Halleux,C.,Keller,H., Pegurri,M.,Gooi, J.H.,Weber,P.B.,Feng,J.Q.,

Bonewald, L.F., and Kneissel, M. (2010). Osteocyte Wnt/b-catenin signaling is

required for normal bone homeostasis. Mol. Cell. Biol. 30, 3071–3085.

Kumar Kundu, J., Choi, K.Y., and Surh, Y.J. (2006). beta-Catenin-mediated

signaling: A novel molecular target for chemoprevention with anti-inflamma-

tory substances. Biochim. Biophys. Acta 1765, 14–24.

Morin, P.J. (1999). beta-catenin signaling and cancer. Bioessays 21, 1021–

1030.

Nath, N., Kashfi, K., Chen, J., and Rigas, B. (2003). Nitric oxide-donating

aspirin inhibits beta-catenin/T cell factor (TCF) signaling in SW480 colon can-

cer cells by disrupting the nuclear beta-catenin-TCF association. Proc. Natl.

Acad. Sci. USA 100, 12584–12589.

Nishimura, W., Yao, I., Iida, J., Tanaka, N., and Hata, Y. (2002). Interaction of

synaptic scaffolding molecule and b -catenin. J. Neurosci. 22, 757–765.

Chemistry & Biology 20,

Nourry, C., Grant, S.G., and Borg, J.P. (2003). PDZ domain proteins: plug and

play! Sci. STKE 2003, RE7.

Oliva, C., Escobedo, P., Astorga, C., Molina, C., and Sierralta, J. (2012). Role of

the MAGUK protein family in synapse formation and function. Dev. Neurobiol.

72, 57–72.

Perego, C., Vanoni, C., Massari, S., Longhi, R., and Pietrini, G. (2000).

Mammalian LIN-7 PDZ proteins associate with beta-catenin at the cell-cell

junctions of epithelia and neurons. EMBO J. 19, 3978–3989.

Piedra, J., Miravet, S., Castano, J., Palmer, H.G., Heisterkamp, N., Garcıa de

Herreros, A., and Dunach, M. (2003). p120 Catenin-associated Fer and Fyn

tyrosine kinases regulate beta-catenin Tyr-142 phosphorylation and beta-cat-

enin-alpha-catenin Interaction. Mol. Cell. Biol. 23, 2287–2297.

Ress, A., and Moelling, K. (2008). The PDZ protein erbin modulates catenin-

dependent transcription. Eur. Surg. Res. 41, 284–289.

Reichert, M., Muller, T., and Hunziker, W. (2000). The PDZ domains of zonula

occludens-1 induce an epithelial to mesenchymal transition of Madin-Darby

canine kidney I cells. Evidence for a role of beta-catenin/Tcf/Lef signaling.

J. Biol. Chem. 275, 9492–9500.

Roche, J.P., Packard, M.C., Moeckel-Cole, S., and Budnik, V. (2002).

Regulation of synaptic plasticity and synaptic vesicle dynamics by the PDZ

protein Scribble. J. Neurosci. 22, 6471–6479.

Ryeom, S.W., Paul, D., and Goodenough, D.A. (2000). Truncation mutants of

the tight junction protein ZO-1 disrupt corneal epithelial cell morphology.

Mol. Biol. Cell 11, 1687–1696.

Schneeberger, E.E., and Lynch, R.D. (2004). The tight junction: a multifunc-

tional complex. Am. J. Physiol. Cell Physiol. 286, C1213–C1228.

Shin, K., Fogg, V.C., and Margolis, B. (2006). Tight junctions and cell polarity.

Annu. Rev. Cell Dev. Biol. 22, 207–235.

Smith, K.J., Johnson, K.A., Bryan, T.M., Hill, D.E., Markowitz, S., Willson, J.K.,

Paraskeva, C., Petersen, G.M., Hamilton, S.R., Vogelstein, B., et al. (1993). The

APC gene product in normal and tumor cells. Proc. Natl. Acad. Sci. USA 90,

2846–2850.

Soler, A.P., Miller, R.D., Laughlin, K.V., Carp, N.Z., Klurfeld, D.M., and Mullin,

J.M. (1999). Increased tight junctional permeability is associated with the

development of colon cancer. Carcinogenesis 20, 1425–1431.

Song, L., and Gelmann, E. (2005). Interaction of b-catenin and TIF2/GRIP1 in

transcriptional activation by the androgen receptor. J. Biol. Chem. 280,

37853–37867.

Song, D.H., Sussman, D.J., and Seldin, D.C. (2000). Endogenous protein ki-

nase CK2 participates in Wnt signaling in mammary epithelial cells. J. Biol.

Chem. 275, 23790–23797.

Stiffler, M.A., Grantcharova, V.P., Sevecka, M., and MacBeath, G. (2006).

Uncovering quantitative protein interaction networks for mouse PDZ domains

using protein microarrays. J. Am. Chem. Soc. 128, 5913–5922.

Stiffler, M.A., Chen, J.R., Grantcharova, V.P., Lei, Y., Fuchs, D., Allen, J.E.,

Zaslavskaia, L.A., and MacBeath, G. (2007). PDZ domain binding selectivity

is optimized across the mouse proteome. Science 317, 364–369.

Sun, Y., Aiga, M., Yoshida, E., Humbert, P.O., and Bamji, S.X. (2009). Scribble

interacts with b-catenin to localize synaptic vesicles to synapses. Mol. Biol.

Cell 20, 3390–3400.

Theisen, C., Wahl, J., III, Johnson, K., and Wheelock, M. (2007). NHERF links

the N-cadherin/catenin complex to the platelet-derived growth factor receptor

to modulate the actin cytoskeleton and regulate cell motility. Mol. Biol. Cell 18,

1220–1232.

Tsukita, S., Furuse, M., and Itoh, M. (2001). Multifunctional strands in tight

junctions. Nat. Rev. Mol. Cell Biol. 2, 285–293.

Wan, H., Winton, H.L., Soeller, C., Tovey, E.R., Gruenert, D.C., Thompson,

P.J., Stewart, G.A., Taylor, G.W., Garrod, D.R., Cannell, M.B., and

Robinson, C. (1999). Der p 1 facilitates transepithelial allergen delivery by

disruption of tight junctions. J. Clin. Invest. 104, 123–133.

Zhang, J., Yan, X., Shi, C., Yang, X., Guo, Y., Tian, C., Long, J., and Shen, Y.

(2008). Structural basis of b-catenin recognition by Tax-interacting protein-1.

J. Mol. Biol. 384, 255–263.

816–827, June 20, 2013 ª2013 Elsevier Ltd All rights reserved 827