Jaramillo BE 2003

8/13/2019 Jaramillo BE 2003

1/44

Progress in Biophysics & Molecular Biology 81 (2003) 144

Review

Tight junction proteins

L. Gonz!alez-Mariscal*, A. Betanzos, P. Nava, B.E. Jaramillo

Department of Physiology, Biophysics and Neuroscience, Center for Research and Advanced Studies (CINVESTAV),

Ave. Polit!ecnico Nacional 2508, M!exico DF, 07000, Mexico

Abstract

A fundamental function of epithelia and endothelia is to separate different compartments within the

organism and to regulate the exchange of substances between them. The tight junction (TJ) constitutes the

barrier both to the passage of ions and molecules through the paracellular pathway and to the movement of

proteins and lipids between the apical and the basolateral domains of the plasma membrane. In recent years

more than 40 different proteins have been discovered to be located at the TJs of epithelia, endothelia and

myelinated cells. This unprecedented expansion of information has changed our view of TJs from merely a

paracellular barrier to a complex structure involved in signaling cascades that control cell growth and

differentiation. Both cortical and transmembrane proteins integrate TJs. Among the former are scaffolding

proteins containing PDZ domains, tumor suppressors, transcription factors and proteins involved in vesicle

transport. To date two components of the TJ filaments have been identified: occludin and claudin. The latter

is a protein family with more than 20 members. Both occludin and claudins are integral proteins capable of

interacting adhesively with complementary molecules on adjacent cells and of co-polymerizing laterally.

These advancements in the knowledge of the molecular structure of TJ support previous physiological models

that exhibited TJ as dynamic structures that present distinct permeability and morphological characteristics

in different tissues and in response to changing natural, pathological or experimental conditions.

r 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Tight junctions; Claudin; Occludin; JAM, MAGUK, ZO; Cingulin

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2. Transmembrane proteins of the TJ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1. TJ tetraspan proteins found in epithelial and endothelial cells . . . . . . . . . . . . . . . 5

2.1.1. Occludin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1.2. Claudins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

*Corresponding author. Tel.: +52-55-57-47-70-00 x 5110 or 5155; fax: +52-55-57-47-70-00 x 5702.

E-mail address: [email protected] (L. Gonz!alez-Mariscal).

0079-6107/03/$ - see front matterr 2003 Elsevier Science Ltd. All rights reserved.

PII: S 0 0 7 9 - 6 1 0 7 ( 0 2 ) 0 0 0 3 7 - 8


2/44

1. Introduction

In multicellular organisms fluids with different molecular compositions (e.g. urine, milk, gastricjuice, blood, etc.) are contained in compartments delineated by epithelia (e.g. renal tubules) and

endothelia (blood vessels). These cellular sheets constitute the frontier between the organism

internal milieu and the compartments contents.

Epithelia and endothelia have tight junctions (TJ) that regulate the passage of ions, water and

molecules through the paracellular pathway. This characteristic is generally referred to as the gate

property of the TJ. The establishment of TJ at the uppermost portion of the lateral plasma

membranes is the result of a polarized insertion of proteins, yet TJ act as a fence that maintains

cell polarity, by blocking the free diffusion of proteins and lipids between the apical and

basolateral domains of the plasma membrane.

2.2. TJ tetraspan proteins found within myelin sheaths. . . . . . . . . . . . . . . . . . . . 10

2.2.1. OSP/claudin 11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.2.2. PMP22/gas-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.2.3. OAP-1/TSPAN-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.3. TJ proteins that belong to the immunoglobulin superfamily . . . . . . . . . . . . . . . 11

2.3.1. JAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.3.2. CAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.3.3. P0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3. Plaque proteins of the TJ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.1. PDZ-containing proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.1.1. The MAGUK proteins of the TJ . . . . . . . . . . . . . . . . . . . . . . . . 14

3.1.2. MAGI, the MAGUK inverted proteins of the TJ . . . . . . . . . . . . . . . . 20

3.1.3. PAR proteins of the TJ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.1.4. MUPP1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3.1.5. AF-6/Afadin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3.1.6. PATJ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.2. TJ proteins lacking PDZ domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.2.1. Cingulin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.2.2. Symplekin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3.2.3. 7H6 antigen/barmotin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3.2.4. Rab proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3.2.5. Pilt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.2.6. JEAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.2.7. huASH1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.2.8. Heterotrimeric G proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

4. Molecular assembly of the TJ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

4.1. Assembly of TJ molecules in epithelial and endothelial monolayers . . . . . . . . . . . 28

4.2. Molecular maturation of TJ during trophoectoderm differentiation . . . . . . . . . . . 294.3. TJ assembly during earlyXenopus development . . . . . . . . . . . . . . . . . . . . . 29

5. Conclusions and future perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

L. Gonz !alez-Mariscal et al. / Progress in Biophysics & Molecular Biology 81 (2003) 1442


3/44

On ultrathin section electron micrographs, TJs are viewed as a series of fusion points between

the outer leaflets of the membrane of adjacent cells. At these kissing points, the intercellular

space is completely obliterated. On freeze-fracture replica electron micrographs TJs appear as a

network of continuous and anastomosing filaments on the protoplasmic face (P) of the plasmamembrane, with complementary grooves on the exoplasmic face (E) (Gonzalez-Mariscal et al.,

2001) (Fig. 1).

Two models have been developed to explain the chemical nature of TJ strands. In the protein

model, the filaments are formed by integral membrane proteins that associate with a partner in the

apposing membrane of the adjacent cell. In the lipid model instead, strands are proposed to be

formed of cylindrical micelles with the polar groups of the lipids directed inwards, and the

hydrophobic tails immersed in the lipid matrix of the plasma membrane of both contacting cells

(Kachar and Reese, 1982; Pinto and Kachar, 1982). The lipid model is however inconsistent with

the observation that changing the total composition of phospholipids, sphingolipids and

cholesterol in epithelial cells does not alter the appearance of TJ strands nor the gate or fencefunction of TJ (Calderon et al., 1998). Nevertheless, a rapid reduction of cell cholesterol by

methyl-b-cyclodextrin has been reported to generate a decline in transepithelial electrical

resistance (TER), an increased mannitol flux and an augmented number of TJ particles associated

with the E face (Francis et al., 1999). These results therefore suggest that although the TJ strands

might be composed of proteins, the structure is sensitive to rapid changes on its lipidic

environment.

In recent years, a constellation of cortical and integral proteins of the TJ has been

discovered. Of the former, 16 different molecules have so far been identified. Some function

as scaffolds, that link the integral proteins of the TJ to the actin cytoskeleton, while others

act as cross linkers of transmembrane junctional proteins. Still others are involved in

vesicular trafficking to the TJ or in cell signaling through their association to kinases andRas. Exciting new information has revealed that some submembranous junctional proteins even

have a role in gene expression due to their nuclear shuttling and specific binding to transcription

factors.

At the TJ three integral proteins are found: occludin, claudins and JAM. The former

two constitute the backbone of TJ strands while JAM appears to be important for the

routine trafficking of T-lymphocytes, neutrophiles and dentritic cells from the lymphoid

and vascular compartments to the tissues during immune surveillance and inflammatory

responses.

TJs were initially described in epithelia and endothelia. However recent observations have

demonstrated that they are also present in myelinated cells. Thus, during development of thecentral (CNS) and peripheral nervous system (PNS), oligodendrocyte and Schwann cells stop

dividing and respectively start to wrap axons in a loose spiral, resulting in the typical

multilamellar structure that electrically isolates the axon and allows saltatory conduction to

proceed. TJs present between the glia and the axons and within the myelin sheaths have proteins

found in the TJs of epithelia and endothelia as well as other nervous system specific molecules that

will also be described in this review.

Since a wealth of information on TJ proteins has emerged in recent times, and keeping updated

on so many different proteins has become a hard task, we have written this review, to serve as a

guide for readers interested in the field of cellcell adhesion.

L. Gonz !alez-Mariscal et al. / Progress in Biophysics & Molecular Biology 81 (2003) 144 3


4/44

Fig. 1. Structure of tight junctions. (A) Freeze-fracture electron microscopic image of TJs in an epithelial monolayer.

TJs appear as a network of anastomosing strands on the P face (P) with complementary grooves on the E face (E). Scale

bar, 100 nm:(B) Ultrathin section of a TJ. Ruthenium red added to the apical surface of epithelial monolayers cannotpass beyond the tight junction (arrow). Scale bar, 10 nm (courtesy of Dr. Bibiana Ch !avez de Ram!rez).



5/44

2. Transmembrane proteins of the TJ

2.1. TJ tetraspan proteins found in epithelial and endothelial cells

In this section we will refer to integral TJ proteins whose structure predicts four transmembrane

regions with two extracellular domains, and with both amino and carboxyl terminal ends oriented

towards the cytoplasm. While certain TJ tetraspans have both extracellular loops of

approximately the same size (e.g. occludin), some have one extracellular loop larger than the

other (e.g. claudins and tetraspanins) (Fig. 2).

2.1.1. Occludin

The name of this integral protein of the TJ derives from the Latin word occludere which

means to occlude (Furuse et al., 1993). Two crucial lines of evidence have shown that occludin is a

Occludin TetraspaninClaudin

254 aaCOOH

21-42 aaCOOH149 aa

NH2

10 aa

45 aa 44 aa

NH2

51 aa14 aa

13 aa

9-24 aaNH2

C

4-40aaCOOH

4 aa

13-30aa

78-150aa

C

C

C

G

X

XP

Y G

G

G Y

YG

G

-

-

-

++

+

+

+-

Y

Y

Y

Y

Y

Y

C

Fig. 2. Schematic representation of tetraspan proteins of the TJ. Tetraspan proteins have four transmembrane regions,

two extracellular domains, and their amino and carboxyl terminal ends are oriented towards the intracellular region.

Both extracellular loops of occludin are of approximately the same size, lack charged residues and are very rich in

tyrosine (Y). More than half of the first loop residues are tyrosines and glycines (G). In claudins the first extracellular

loop is longer than the second one. Both claudin loops display a number of charged residues ;which are expectedto influence the passage of ions through the paracellular space. Tetraspanins have a longer second extracellular loop

and can be further differentiated from claudins, by the presence within the long loop of CCG and PXXCC motifs as

well as two or four cysteine residues (C), one of which is consistently located 11 residues from the predicted start of the

fourth transmembrane domain.



6/44

constituent of TJ filaments: (A) Occludin has the capacity of forming TJ-like filaments when

transfected into cells that lack TJ (e.g. L-fibroblasts) (Furuse et al., 1998b). (B) Freeze-fracture

immunoreplica electron microscopy (EM) has revealed the presence of occludin within TJ fibrils

(Fujimoto, 1995). Occludin comprises four transmembrane domains, two extracellular loops ofsimilar size, and three cytoplasmic domains: one intracellular short turn, a small amino terminal

domain and a long carboxyl terminal region. Both extracellular loops are enriched in tyrosine

residues, and in the first one more than half of the residues are tyrosines and glycines (Fig. 2).

Several lines of evidence assign occludin an important role at TJs. Thus, the over-expression of

mutant forms of occludin in epithelial cells leads to changes in the gate and fence function of TJs

(Balda et al., 1996b; McCarthy et al., 1996; Bamforth et al., 1999) as in the transepithelial

migration of neutrophils (Huber et al., 2000). In addition, the administration of synthetic peptides

corresponding to the extracellular loops of occludin, to epithelial cells, results in the

disappearance of TJs, inhibition of cell adhesion and up regulation of b-catenin and of the

b-catenin/TCF downstream target gene c-myc (Lacaz-Vieira et al., 1999; Medina et al., 2000; VanItallie and Anderson, 1997; Vietor et al., 2001; Wong and Gumbiner, 1997).

Occludin migrates as a tight cluster of 6282 kDa bands on SDS gels as a result of

phosphorylation on serine, threonine (Sakakibara et al., 1997; Wong, 1997; Wong and Gumbiner,

1997) and tyrosine residues (Chen et al., 2002b; Tsukamoto and Nigam, 1999). In these

phosphorylations PKC (Andreeva et al., 2001), CK2 (Cordenonsi et al., 1999b), p34cdc2=cyclin Bcomplex (Andreeva et al., 2001; Cordenonsi et al., 1997; 1999b) and the non-receptor tyrosine

kinase c-Yes (Chen et al., 2002b) are involved. In epithelial cell lines highly phosphorylated

occludin molecules are selectively concentrated at TJs whereas non- or less-phosphorylated

occludin molecules localize in the cytoplasm (Andreeva et al., 2001; Sakakibara et al., 1997;

Tsukamoto and Nigam, 1999). Studies inXenopusembryos, have however revealed opposite data,

as occludin dephosphorylation correlates with de novo assembly of TJs (Cordenonsi et al., 1997).Furthermore, in endothelial cells, shear stress significantly reduces occludin content and increases

its tyrosine phosphorylation with a concomitant increase in hydraulic conductivity (DeMaio et al.,

2001). Therefore occludin phosphorylation may play opposite roles in distinct biological systems

or alternatively, phosphorylation of different residues may have dissimilar consequences.

The last 150 amino acids of the carboxyl tail of occludin interact directly with F-actin (Wittchen

et al., 1999). This constitutes a property not shared by other TJ integral proteins that require the

mediation of scaffolding proteins for actin association. Occludin is also capable of binding

directly through this carboxyl segment, to the MAGUK proteins ZO-1 (Furuse et al., 1994), ZO-2

(Itoh et al., 1999b; Wittchen et al., 1999). and ZO-3 (Haskins et al., 1998). These terminal 150

amino acids of the occludin tail, are remarkably conserved among interspecies and are predictedto form a typical a-helical coiled-coil structure (Ando-Akatsuka et al., 1996). Employment of a

novel bait peptide method, revealed that occludin could interact with itself through this domain.

Additionally this region associates with the regulatory proteins PKC-z; non-receptor tyrosinekinase c-Yes and phosphatidylinositol 3-kinase as with the gap junction component connexin 26

(Nusrat et al., 2000). However since these interactions were detected by an in vitro procedure their

physiological significance remains unclear.

The type I WW binding motif (PPXY) present in the amino terminal portion of occludin

interacts with four WW motifs present in Itch, a E3 ubiquitin protein ligase involved in occludin

degradation (Traweger et al., 2002).



7/44

Recently an occludin related gene (ORG) has been identified on the Y chromosome of

Drosophila melanogaster (Carvalho et al., 2001), and two alternatively splicing forms of occludin

have lately been described (Muresan et al., 2000). Isoform 1B contains a unique N terminal

sequence of 56 amino acids, whose function remains unknown.Although occludin is a clear constituent of TJ filaments, and its abundance is related to the

degree of sealing of epithelia (e.g. more in distal than in the proximal segments of the nephron)

(Gonzalez-Mariscal et al., 2000), its precise role in TJ remains unclear, specially after observing

that occludin knock out mice display well developed TJ (Saitou et al., 2000).

2.1.2. Claudins

The paradoxical results obtained with occludin deficient mice described above, led Tsukita and

co-workers to search for other integral components of TJ. Using the same liver fraction employed

to identify occludin, and by means of a sucrose step gradient, a single 22 kDa band was

discovered as a putative novel TJ integral protein. Peptide sequencing revealed two proteins in thisband that were subsequently named claudin 1 and 2 (Furuse et al., 1998a). The name claudin

derives from the Latin word claudere which means to close.

By data base searching and cDNA and genomic cloning the claudin family has expanded to 24

members (Table 1) (Tsukita et al., 2001). All claudins encode 2027 kDa proteins with four

transmembrane domains, two extracellular loops where the first one is significantly longer than

the second one, and a short carboxyl intracellular tail (Fig. 2). The last amino acids of this tail are

highly conserved within the family and constitute PDZ binding motifs: claudins 19 and 17 S/

TYV, claudins 10 and 15 AYV, claudin 11 AHV, claudin 12 HTT, claudin 13 LDV, claudins 14,

18 and 20 DYV, claudin 16 TRV, and claudin 19 DRV. Through these motifs claudins are linked

to the TJ PDZ containing proteins ZO-1, ZO-2, ZO-3 (Itoh et al., 1999a), PATJ (Roh et al.,

2002a) and MUPP1 (Hamazaki et al., 2002).When individual claudins 13, 5 or 11 were introduced into mouse L fibroblasts,

intramembrane strands appeared in freeze-fracture replicas (Furuse et al., 1998b; Morita et al.,

1999a, b). Thus suggesting that claudins constitute the backbone of TJ strands. Different claudin

species are capable of generating different freeze-fracture patterns. Thus, transfection of claudins

1 or 3 generates continuous smooth intramembrane strands on the protoplasmic surface (P face)

of the replicas (Furuse et al., 1999), whereas claudins 2 or 5 form discontinuous chains of particles

associated to the exoplasmic face (E face) (Morita et al., 1999b). Transfection with claudin 11

generates instead parallel intramembrane strands on the P face that scarcely branch (Morita et al.,

1999a).

Heterogeneous claudins can interact within a single TJ strand. For example, by immunoreplicaEM the co-incorporation of distinct transfected claudins into individual intramembrane strands

has been confirmed. The particular combination of claudins within a TJ strand might give rise to

different freeze-fracture patterns. Thus strands formed with claudins 1 and 3 are continuous and

associated to the P face, while strands formed with claudins 1 and 2 or 3 and 2 have evenly

scattered particles in the E face grooves. At the paracellular space the extracellular loops of

different species of claudins belonging to neighboring cells can also interact, except in some

combinations. Thus, when L transfectant singly expressing claudin 1, 2 or 3 were co-cultured,

claudin 3 strands associated with claudin 1 and 2 strands of the apposing cell, whereas claudin 1

did not interact with claudin 2 strands (Furuse et al., 1999).



8/44

Claudins depict a differential distribution in distinct tissues, supporting the idea that they are

responsible for the ample variety in electrical resistance and paracellular ionic selectivity displayed

by epithelia and endothelia. The nephron is a good model to exemplify this point, since it is

integrated by tubules with low and high TER (6 O cm2 in proximal segments vs. 8702000 Ocm2

in the collecting duct) that specialize in the resorption of specific ions. Northern blot analysis hasrevealed the expression of all claudins except 6, 9, 13 and 14 in a total kidney extract. However,

the study of kidney cryosections (Enck et al., 2001; Kiuchi-Saishin et al., 2002) or of micro-

dissected tubules (Reyes et al., 2002) has revealed the differential distribution of these proteins:

claudins 5 and 15 in endothelia, claudins 2, 10 and 11 at the proximal segment, claudins 1, 3 and 8

at the distal tubule and claudins 1, 3, 4 and 8 at the collecting segment.

The expression of different type of claudins appears to be finely tuned during development. For

example: (A) Claudin 6 is present in embryonic epithelia (Turksen and Troy, 2001) and its over

expression in transgenic mice generates a defective epidermal permeability barrier (Turksen and

Troy, 2002). It has been proposed that the unstable temperature control and dehydration

Table 1

Characteristic features of different claudins

Claudin Distinctive characteristics

1 Present in high-resistance epithelia (e.g. collecting segment) and absent in leaky epithelia (e.g. proximal

tubule) (Reyes et al., 2002). Crucial for the mammalian epidermal barrier (Furuse et al., 2002). Absent in

most human breast cancer cell lines (Hoevel et al., 2002).

2 Present in leaky epithelia (e.g. proximal tubule) and absent in tight epithelia (e.g. collecting segment)

(Reyes et al., 2002; Enck et al., 2001). Present in the choroids plexus epithelium (Wolburg et al., 2001).

3 Also known as RVP1 (Briehl and Miesfeld, 1991). Present in the tighter segments of the nephron (Kiuchi-

Saishin et al., 2002). Its expression is elevated in regressing ventral prostate and in prostate

adenocarcinomas (Long et al., 2001). Capable of CPE binding (Sonoda et al., 1999).

4 Its expression decreases paracellular conductance through a selective decrease in sodium permeability

(Van Itallie et al., 2001). Present in the tighter segments of the nephron (Kiuchi-Saishin et al., 2002). Over

expressed in pancreatic and gastrointestinal tumors (Michl et al., 2001). The selective CPE binding gave

rise to its alternative name CPE-R (Sonoda et al., 1999).

5 Receives the alternative name of TMVCF, as it is frequently deleted in Velo cardio facial syndrome

(Sirotkin et al., 1997). Constitutes TJ strands in endothelial cells (Morita et al., 1999b). Transiently

expressed during the development of the retinal pigment epithelium (Kojima et al., 2002).

6 Present in embryonic epithelia (Turksen and Troy, 2001). Its over expression in transgenic mice generates

a defective epidermal permeability barrier (Turksen and Troy, 2002).

7 Down regulated in head and neck squamous cell carcinomas (Al Moustafa et al., 2002).

8 Present in the tighter segments of the nephron (Kiuchi-Saishin et al., 2002).

11 Also named OSP. Present in oligodendrocytes and Sertoli cells (Morita et al., 1999a).

14 Expressed in the sensory epithelium of the organ of Corti. Mutations in the gene cause autosomal

recessive deafness (Wilcox et al., 2001).

15 Present in endothelial cells (Kiuchi-Saishin et al., 2002).

16 Also known as Paracellin-1. Critical for Mg2 and Ca2 resorption in the human thick ascending limb of

Henle (Blanchard et al., 2001; Simon et al., 1999).18 A downstream target gene for the T/EBP/NKX2.1 homeodomain transcription factor. Expressed in lung

and stomach (Niimi et al., 2001).

Claudins 9, 10, 12, 13, 17 and 1924 have not been yet well characterized.



9/44

frequently observed in premature infants, might be related to the expression of this claudin in their

epidermis. (B) Claudin 5 is transiently expressed during the development of the retinal pigment

epithelium (Kojima et al., 2002). (C) Claudin 11 is expressed in Sertoli cells, immediately after the

peak of expression of the sex determining region in the Y gene (Hellani et al., 2000).Claudin 16 is mutated in human patients with hypomagnesemia hypercalciuria syndrome

(HHS) (Simon et al., 1999). These patients manifest a selective defect in paracellular Mg2 and

Ca2 reabsorption in the thick ascending limb of Henle (TAL), while maintaining an intact NaCl

resorption ability (Blanchard et al., 2001). Hence claudin 16 might function as a paracellular

channel selective for Mg2 and Ca2 (Goodenough and Wong, 1999). Other claudins have also

proved to be ionic selective. Such is the case of claudin 4, that when transfected into epithelial

cells, decreases the paracellular conductance through a selective decrease in Na permeability

without a significant effect on Cl permeability (Van Itallie et al., 2001). The proposal of ion

channels or pores within the TJ strands is more than two decades old, and arose with Claudes

observation that TER increases with the number of TJ strands present in the epithelia, not in alinear fashion as would be expected from the addition of resistors in series, but exponentially

(Claude, 1978; Gonzalez-Mariscal et al., 2001). The ionic selectivity at the TJ could therefore be

determined by the specific claudins that constitute the pore. On analyzing the extracellular loops

of claudins an enormous variability in distribution and number of charged residues is found. For

example the isoelectric points of the first loop range from 4.17 in claudin 16 to 10.49 in claudin 14,

and in the second extracellular loop from 4.05 in claudins 2, 7, 10 and 14 to 10.5 in claudin 13.

Based on the pKIs of the extracellular loops sequences, claudin 16 is predicted to act as a cation

pore, whereas claudins 4, 11 and 17 should function as anionic channels (Mitic and Van Itallie,

2001).

Variations in the tightness of the TJ appear to be determined by the combination and mixing

ratios of different claudin species. Thus when MDCK cells expressing claudin 1 and 4 wereincubated with the claudin 4 binding protein,Clostridium perfringensenterotoxin (CPE), claudin 4

was selectively removed from TJs, generating a significant decrease in TER (Sonoda et al., 1999).

When claudin-2 instead was introduced into high-resistance MDCK cells (MDCK I), their TJs

became leaky and were similar functionally and morphologically to those in low-resistance cells

(MDCK II), which normally contain high levels of claudin 2 (Furuse et al., 2001).

The role of claudins in carcinogenesis is controversial. Thus, claudin 4 is over-expressed in

pancreatic cancer and gastrointestinal tumors. Treatment with TGFb or CPE, leads to a

significant reduction of tumor growth (Michl et al., 2001), thus suggesting that proteins involved

in cellcell contacts such as claudins may facilitate processes of invasion and migration. On the

other hand certain claudins remain low or undetectable in a number of tumors and cancer celllines. For example claudin 1 expression is lost in most human breast cancers without presenting

alterations in its promoter or coding sequences (Hoevel et al., 2002; Kramer et al., 2000), and

claudin 7 is down regulated in head and neck squamous cell carcinomas (Al Moustafa et al.,

2002).

The crucial task of claudins in the gate function of TJs is highlighted by the following evidence:

(A) In the mammalian epidermis, claudin-1 co-localizes with occludin in the most apical regions

of the second layer of the stratum granulosum, while claudin 4 is present in deeper layers of the

stratum. In claudin 1 deficient mice, the epidermal barrier is severely affected leading to

dehydration, wrinkled skin and death of the animals within 1 day of birth. In these mice the



10/44

occludin positive and claudin-1 deficient skin layers allow the passage of paracellular tracers,

suggesting that the combination of claudin-1 and occludin is needed for the establishment of an

effective paracellular barrier (Furuse et al., 2002). (B) In human breast cancer cells that have lost

the expression of claudin-1, transfection of this claudin decreases the paracellular flux of tracersdespite the absence of occludin (Hoevel et al., 2002).

2.2. TJ tetraspan proteins found within myelin sheaths

2.2.1. OSP/claudin 11

OSP/claudin 11 is a crucial component of TJs in CNS myelin and between Sertoli cells (Gow

et al., 1999; Spector et al., 1998). Claudin 11 null mice have no TJ in their oligodendrocytes and

Sertoli cells, show slow CNS conductance, hind limb weakness, and sterility in male animals (Gow

et al., 1999).

In contrast to conventional TJ that are formed by networks of anastomosing strands, thosefound in the CNS myelin and in Sertoli cells are comprised by parallel filaments (Southwood and

Gow, 2001), thus suggesting that claudin 11 polymerization restricts the formation of branching

TJ fibrils.

OSP/claudin 11 is the third most abundant central nervous system (CNS) myelin protein

(Bronstein et al., 1997). During prenatal development OSP/claudin 11 is profuse in developing

meninges and mesenchymal cells, especially around regions of chondrocyte formation (Bronstein

et al., 2000). Postnatally, it is only expressed in oligodendrocytes and testis. In adult animals,

expression of OSP/claudin 11 by the testis is inhibited by the hormone FSH and the cytokine

TNFa (Hellani et al., 2000).

Although abundant evidence supports a major role for claudin 11 at the TJ of Sertoli cells, the

participation of occludin in these junctions cannot be ruled out. In fact, the administration of asynthetic peptide corresponding to the second extracellular loop of occludin perturbs the blood

testis barrier and reversibly disrupts spermatogenesis (Chung et al., 2001). Therefore both

occludin and claudin 11 might be needed for Sertoli TJ to develop.

2.2.2. PMP22/gas-3

Theperipheralmyelinprotein PMP22/gas-3 was originally identified as a growtharrestspecific

protein of fibroblasts (Schneider et al., 1988). PMP22/gas-3 is a 22 kDa tetraspan glycoprotein of

160 amino acids. PMP22/gas-3 expression is closely synchronized with Schwann cells

differentiation and localizes in the myelin sheath (Baechner et al., 1995). It has been characterized

as a strong adhesive component for compact myelin formation. Deletions, duplications ormutations of PMP22/gas-3 account for the majority of heritable demyelinating peripheral

neuropathies in mice (Trembler) and humans including CharcotMarie-Tooth disease type IA,

Dejerine Sottas syndrome and heredity neuropathy with liability to pressure palsies (HNPP)

(Suter and Nave, 1999). PMP22/gas-3 mRNA has been detected in a variety of non-neural tissues,

the epithelial cells of the lungs and intestine being the higher expressers (Baechner et al., 1995).

In epithelial cells PMP22/gas-3 co-localizes with occludin and ZO-1 at the TJs, and its over-

expression in L cell fibroblasts mediates the formation of ZO-1 positive intercellular junctions

(Notterpek et al., 2001). Therefore it is tempting to speculate that PMP22/gas-3 plays a role in the

establishment and maintenance of TJs in epithelia and within the Schwann cell membrane. The



11/44

amino acid sequence and predicted structure have posed the question of whether PMP22/gas-3 is

a claudin family member that functions as the peripheral nervous system (PNS) homologue of

claudin 11. In this respect it should be pointed that: (1) the sequence of PMP22/gas-3 is fairly

shorter compared to those of claudins (160 vs. 207264 amino acids); (2) PMP22/gas-3 has a weakhomology even with claudin-10 to which sequence it resembles the most (25% identity); (3)

claudin family members display at their carboxyl termini the PDZ binding motifs YV or fXf

(where X denotes any amino acid and f a hydrophobic one), while PMP22/gas-3 ends with RE,

which is not a consensus for PDZ binding; and (4) PMP22/gas-3 expressing cells do not show any

homophilic cell adhesion (Takeda et al., 2001). Therefore it has been suggested that PMP22/gas-3

does not contribute by itself to form and maintain the compact myelin sheath, and instead does so

via its heterophilic interaction with P0, a member of the immunoglobulin superfamily

(Berditchevski, 2001), known to form compact myelin sheaths by homophilic adhesion (DUrso

et al., 1990).

PMP22/gas-3 has recently been described as a member of the evolving epithelial membraneprotein family (EMP13) which appears to function in regulating cell growth and differentiation.

Since members of this family, are somehow similar to the claudin family, it has been suggested

that they derive from a common ancestor (Jetten and Suter, 2000).

2.2.3. OAP-1/TSPAN-3

OAP-1/TSPAN-3 is a tetraspanin of 28 kDa with 254 amino acids. Tetraspanins can be

differentiated from other tetraspan proteins such as claudins, for having their first extracellular

loop shorter than the second one. They are also characterized by the presence within the long

extracellular loop, of CCG and PXXCC motifs as well as of two or four cysteine residues, one of

which is consistently located 11 residues from the predicted start of the fourth TM domain(Berditchevski, 2001) (Fig. 2).

OAP-1/TSPAN-3 forms a complex with b1 integrin and OSP/claudin-11 within myelin sheaths

that regulates proliferation and migration of oligodendrocytes (Tiwari-Woodruff et al., 2001).

2.3. TJ proteins that belong to the immunoglobulin superfamily

2.3.1. JAM

The junctional adhesion molecule JAM is a glycosilated 43 kDa protein found at the TJs of

epithelial and endothelial cells. It has three distinct structural domains: an extracellular region of

215 amino acids that contains two variable type Ig domains; a single transmembrane domain, anda short intracellular tail (45 aa) that features a classical type II PDZ binding motif (Martin-Padura

et al., 1998). Through its carboxyl termini JAM interacts with the PDZ domains of AF6 (Ebnet

et al., 2000), ASIP/Par-3 (1st domain) (Itoh et al., 2001; Ebnet et al., 2001) and ZO-1 (domains 2

and 3) (Bazzoni et al., 2000). JAM co-immunoprecipitates with cingulin, and this association

requires the amino terminal globular head of cingulin (Bazzoni et al., 2000; Ebnet et al., 2000).

The X-ray structure of JAM suggests a homophilic adhesion model in which U-shaped JAM

dimmers stick out almost perpendicular to the cell surface. Contact is established between the first

variable type amino terminal loops that lie almost parallel to the cell surface (Fig. 3) (Kostrewa

et al., 2001).



12/44

Antibodies against JAM inhibit TER recovery in a transient calcium depletion assay,

suggesting the participation of JAM in TJ sealing. The same antibodies have no effect when

applied to confluent monolayers with well-formed TJ, thus indicating the inaccessibility of JAM

within the sealed TJs (Liu et al., 2000).

In freeze-fracture immunoreplicas JAM shows an intimate spatial relationship with TJ strands.However, JAM transfection into fibroblasts does not generate the appearance of TJ fibrils (Itoh

et al., 2001). This can be explained by the observation that integral membrane proteins with a

single membrane-spanning domain like JAM, cannot be detected as intramembrane particles

(IMP) in freeze-fracture replicas. Therefore JAM molecules in epithelial cells may associate

laterally as dimmers, that in turn could aggregate with TJ strands made of linear polymers of

claudin and occludin.

JAM transfection instead generates the appearance of IMP devoid areas in the freeze-fracture

replicas. This pattern is remarkable, as it resembles the in vivo appearance of the membrane

during the beginning of TJ assembly, described by several groups more than two decades ago

COOH COOH

COOH COOH COOH

NH2 NH2

NH2NH2

NH2NH2

215a

a

45

aa COOH

*

*

*

Fig. 3. Homophilic model of interaction of JAM molecules. U-shaped JAM dimmers (indicated with a discontinuousgreen line) stick out almost perpendicular to the cell surface, while their first amino terminal loops (red) lie almost

parallel to the cell surface and contact each other in a common central plane (asterisks). Paracellular JAM interactions

occur between the first loops of JAM molecules located in apposing cell membranes (arrows). JAM network is thus

constructed by repeating the structural motif of the U-shaped dimmers over several neighboring cells.



13/44

(Humbert et al., 1976; Montesano et al., 1975; Tice et al., 1977). The development of this pattern

could speculatively suggest a role for JAM in restricting the free diffusion of proteins within the

membrane. This is a fundamental characteristic of TJs and was described long before the

molecular components of the TJ were first identified (Dragsten et al., 1981; Mandel et al., 1993).Endothelial TJs in addition to their role in regulating solute permeability, serve to impede

leukocyte egress. However, during inflammation leukocytes traverse the microvasculature. The

role of JAM in this process is revealed by the ability of a neutralizing antibody to modulate

monocyte transmigration through the vessel wall (Lechner et al., 2000; Martin-Padura et al.,

1998). New JAMs have recently been described, suggesting the existence of a JAM protein family.

JAM2/VE-JAM present at the cellular borders of venules and vessels, functions as an adhesion

protein capable of capturing human T cells (Cunningham et al., 2000; Palmeri et al., 2000). It

displays homo and heterotypic interactions. The latter happen on T cells, when JAM3 functions

as the counter receptor of JAM2 (Arrate et al., 2001).

Most recently JAM has also been described as a receptor for Reovirus attachment protein s1(Barton et al., 2001; Tyler et al., 2001).

2.3.2. CAR

Thecoxsackievirus andadenovirusreceptor (CAR) is a 46 kDa integral membrane protein with

one transmembrane region, a long cytoplasmic tail, and an extracellular region composed of two

Ig-like domains (Tomko et al., 1997).

CAR seems to be a functional component of TJs since: (A) In epithelial cells it co-

immunoprecipitates with ZO-1 and co-localizes with it at the TJ. The carboxyl terminal domain of

CAR contains the type I PDZ binding motif SXV, that could account for the observed JAM/ZO-1

interaction. (B) In transfected fibroblasts, CAR mediates homotypic cell aggregation and recruitsZO-1 to cellcell contacts. (C) CAR over expression in epithelial cells leads to an increase in TER

accompanied by a reduced passage of macromolecules through the paracellular pathway (Cohen

et al., 2001).

CAR binds to IgG and IgM present in serum (Carson and Chapman, 2001) and is over

expressed at sites of inflammation (Ito et al., 2000). Therefore it is tempting to speculate that CAR

like JAM might participate in the transmigration of cells of the immune system. However, an

identity lower than 30% is maintained between the extracellular regions of CAR and JAM.

2.3.3. P0

Protein 0 (P0) is the major myelin protein of the PNS. Mutations in the P0 gene, cause thedemyelinating peripheral neuropathy CharcotMarie-Tooth disease, the more severe Dejerine

Sottas syndrome and congenital hypomyelination.

In transfected epithelial cells, P0 behaves as a homophilic adhesion molecule (DUrso et al.,

1990; Filbin et al., 1990). This PNS protein is capable of triggering epithelial reversion in

carcinoma cells, highlighting its importance as a cell adhesion molecule (Doyle et al., 1995).

Together with the tetraspan PMP22, P0 is involved in the formation and compaction of myelin.

These two proteins co-localize at the intercellular borders of transfected epithelial cells and when

PMP22 and P0 are expressed in separate but neighboring epithelial cells, P0 is recruited at the

apposed plasma membrane of the PMP22 expressor cell (DUrso et al., 1999). Crystallographic



14/44

studies of P0 explain this PMP22/P0 heterotypic interaction, proposing a tetrameric arrangement

of P0 molecules arranged around a central hole that accommodates PMP22 (Shapiro et al., 1996).

The interaction between two distinct types of myelin proteins, opens the possibility that other

heterotypic contacts might be present in epithelial TJ, for example between JAM and claudins oroccludin. However it should be pointed that the crystallographic structure predicted for JAM is of

a U-shaped dimmer instead of a tetramer.

3. Plaque proteins of the TJ

3.1. PDZ-containing proteins

PDZ are 8090 amino acid modules that bind to specific motifs [e.g. S/TXV, FXF;for a review

see (Bezprozvanny and Maximov, 2001; Songyang et al., 1997)] found at the carboxyl terminalend of several proteins, although some PDZ domains are capable of recognizing internal motifs

(Shieh and Zhu, 1996). The PDZ motif also mediates interactions with PDZ motifs in other

proteins, thus this module in the neural NO synthase (nNOS) binds to the PDZ domain of PSD-

95 (Brenman et al., 1996). At the TJ, ZO-1 associates through its second PDZ to the second PDZ

present in ZO-2 and ZO-3 as will be described below in further detail. PDZ domains are critical

for the clustering and anchoring of transmembrane proteins (Kim et al., 1995). Thus proteins that

contain multiple PDZ domains (PSD95/DLG/ZO-1) function as scaffolds that bring together

cytoskeletal, signaling, and integral proteins at specific regions of the plasma membrane (Fig. 4).

3.1.1. The MAGUK proteins of the TJThe race for the discovery of the molecular components of the TJ started with the identification

by Daniel Goodenough and Mark Mooseker groups of a 225 kDa protein associated to the TJ,

and consequently named ZO-1 (zonula occludens 1) (Stevenson et al., 1986). When the cDNA of

ZO-1 was unraveled, its homology with the tumor suppressor protein disc large (Dlg) ofDrosophilaand with the postsynaptic density protein PSD95/SAP90 was recognized (Itoh et al.,

1993; Willott et al., 1993). Later, when the sequence of the other ZO molecules of the TJ, namely

ZO-2 (Jesaitis and Goodenough, 1994) and ZO-3 (Haskins et al., 1998) was acknowledged, it

became clear they too belonged to a protein family named MAGUK (membrane associated

guanylate kinase homologues). Proteins in this family are recognized for having structurally

conserved PDZ, SH3 and GK domains.SH3 are 5070 amino acid and non-catalytic protein domains that bind to GK modules or to

ligands at least seven residues in length that contain a PXXP sequence. The GK module is

homologous to the enzyme guanylate kinase that catalyzes the conversion of GMP to GDP at the

expense of ATP. However since the sequence of ZO proteins does not predict binding neither to

GMP nor to ATP, the GK module in these proteins is assumed to be enzymatically inactive.

Instead protein binding properties have been ascribed to this module (Kim et al., 1997). It has also

been hypothesized that the GK domain could activate G-protein coupled pathways. In this

respect it should be mentioned that TJ assembly is regulated by G proteins (Balda et al., 1991;

Saha et al., 2001).



15/44

TJ proteins, particularly ZO-1 and ZO-2, also contain a long carboxyl terminal region with an

acidic module, a proline-rich domain and several alternative splicing sites. This area absent in

other MAGUK proteins might be responsible for the unique properties of MAGUK TJ

molecules. In fact transfection with ZO-1 mutants that maintain the MAGUK core but lack the

PDZ1 PDZ2 PDZ3+ SH3 GK - PR

claudins ZO-1occludin

actin, 4.1

cingulin, atypical PKC, AP-1 and C/EBP

PDZ1 PDZ2 PDZ3+ SH3 GK -PR

claudins ZO-1

occludin and cinguin

ZO-1

ZO-2

ZO-3

MAGI-1

MAGI-2

PDZ1 PDZ2 PDZ3 PDZ5PDZ4PDZ0 GK WW

GEP

PDZ1 PDZ2 PDZ3 PDZ5PDZ4PDZ0 GK WW

PTEN

MAGI-3PDZ1 PDZ2 PDZ3 PDZ5PDZ4PDZ0 GK WW

PTEN

MUPP1PDZ1

AF-6PDZmyosin VDRBD1 PRRBD2 kinesin D PRPR actin BD

ZO-1cingulin

JAM

PATJ

CAR, cingulin and AF-6

PDZ1 PDZ2 PDZ3+ SH3 GK - PR

claudins ZO-2, ZO-3 ZONAB ZAK

occludin actinJAM 4.1

claudinsJAM

PDZ2 PDZ3 PDZ4 PDZ5 PDZ6 PDZ7 PDZ8 PDZ9 PDZ10 PDZ11 PDZ12 PDZ13

PDZ1 PDZ2 PDZ3 PDZ4 PDZ5 PDZ6 PDZ7 PDZ8

actin

PAR-3

PAR-6PDZ1

atypical PKC

CR1 CR2 CRIB

PDZ1 PDZ2 PDZ3

atypical PKC

CR3CR1

JAM

ZO-3 claudins

PATJ

PDZ9 PDZ10MRE

Pals1

Pals1PDZ1 SH3 GKU1 L27N L27C 4.1 B

PATJ, MUPP1 CRB1

MRE

Pals1

Fig. 4. PDZ-containing proteins found at the TJ. PDZ domains are represented by ovals while the other domains are

all schematized by dotted boxes. Intermolecular associations with other TJ and cytoskeletal proteins are indicated with

brackets.



16/44

carboxyl region generates a transformation from epithelia to a mesenchymal like type (Ryeom

et al., 2000).

3.1.1.1. ZO-1. ZO-1 is a 210225 kDa protein found at the submembranous domain of TJs inepithelia and endothelia. Cells that do not form TJs such as fibroblasts show ZO-1 disperse in the

cytoplasm and concentrated at cadherin-based adherens junctions (Itoh et al., 1993) through

interactions with a-catenin and the nectinafadin system (Yokoyama et al., 2001).

At the TJ ZO-1 is associated through its first PDZ domain to the carboxyl terminal end of

claudins (Itoh et al., 1999a), by the second and third PDZs to JAM (Ebnet et al., 2000) and by its

GK module to occludin (Fanning et al., 1998; Schmidt et al., 2001). ZO-1 immunoprecipitates

with CAR, a protein that contains PDZ and SH3 recognition motifs (Cohen et al., 2001). ZO-2

and ZO-3 independently associate to ZO-1 through a PDZ-2/PDZ-2 interaction (Wittchen et al.,

1999). ZO-1 binds to the actin cytoskeleton (Fanning et al., 1998; Itoh et al., 1997; Wittchen et al.,

1999) and to actin binding protein 4.1 (Mattagajasingh et al., 2000) through its carboxyl terminalend. Other cortical proteins of the TJ such as AF-6 (Yamamoto et al., 1997) and cingulin

(Cordenonsi et al., 1999a) bind to ZO-1. ZO-1 associates to the adherens junction proteins a-

catenin (Itoh et al., 1997) and to the gap junction proteins connexins 43 (Barker et al., 2002;

Toyofuku et al., 1998) and 45 (Kausalya et al., 2001).

ZO-1 is a phosphoprotein, however the effect of phosphorylation will remain controversial,

until studies on the participation of different kinases over distinct residues on the protein, clarify

the results so far obtained. ZO-1 in low resistance cells is significantly more phosphorylated than

in high-resistance monolayers (Stevenson et al., 1989), and hypoxia in brain micro-vessels induces

an enhanced phosphorylation of ZO-1 that correlates with a decreased expression and

dislocalization of ZO-1 (Fischer et al., 2002). However, a low phosphorylated ZO-1 has been

detected in cells that lack TJs or have them disassembled due to lack of calcium (Howarth et al.,1994). With regards to tyrosine phosphorylation, some recent studies have demonstrated that

vascular endothelial growth factor increases paracellular permeability and augments ZO-1

tyrosine phosphorylation (Antonetti et al., 1999). However, in A431 cells, epidermal growth

factor induces tyrosine phosphorylation of ZO-1 and concentrates this protein at TJs (Van Itallie

et al., 1995) and during TJ assembly ZO-1 becomes tyrosine phosphorylated (Chen et al., 2000).

ZO-1 associates and is a substrate of ZAK, a serine/threonine kinase (Balda et al., 1996a) and

of PKC (Avila-Flores et al., 2001). MAPK signaling pathway regulates tyrosine phosphorylation

of ZO-1, as MEK1 inhibition in Ras transformed epithelial cells restores epithelial morphology

and increases tyrosine phosphorylation of ZO-1 and occludin (Chen et al., 2000).

Three alternative splicing domains have been identified in ZO-1, all of which are located at thecarboxyl region of the molecule. The first named motifa is an 80 amino acid domain (Balda and

Anderson, 1993). In epithelia and endothelia both thea anda isoforms are expressed. Yet, the

a is quantitatively more abundant in epithelia while the opposite is true for endothelia (Balda

and Anderson, 1993; Underwood et al., 1999). These isoforms seem to perform different roles.

For example, the a isoform is present in cells that display no TER like podocytes (Balda and

Anderson, 1993; Kurihara et al., 1992), in Sertoli cells (Balda and Anderson, 1993), whose

junctions can be described as dynamic since they move along the entire lateral cell border and

break and reseal around migrating spermatocytes, and in mouse blastomeres that lack TJs.

Instead, the a isoform is expressed later upon the formation of the blastocoele and the



17/44

development of TJs (Sheth et al., 1997). Therefore the a isoform appears to be related to the

establishment of functional TJs (Table 2), while the a is related to structurally dynamic

junctions.

The other alternative splicing domains identified in ZO-1 areb1;b2and g;with respective motifsof 7, 20 and 45 amino acids. Although they are expressed in a variety of tissues their functional

significance still remains unclear (Gonzalez-Mariscal et al., 1999).

The sequence of ZO-1 contains two putative nuclear export signals (NES) and three nuclear

export signal (NES) (Gonzalez-Mariscal et al., 1999), thus suggesting shuttling of ZO-1 between

the nucleus and the plasma membrane (Islas et al., 2002). Furthermore, cells with decreased

cellcell contact, such as those in sparse or mechanically injured monolayers, display a strong

presence of ZO-1 at the nuclei (Gottardi et al., 1996). ZO-1 specifically interacts through its

SH3 domain, with a Y box transcription factor named ZONAB, which binds to promoter

sequences of cell cycle regulators. This interaction modulates paracellular permeability and gene

expression in reporter assays (Balda and Matter, 2000), speculatively suggesting that ZOmolecules establish a cross talk between the nucleus and the TJ that balances epithelial cell

differentiation and growth.

Numerous studies employing cytokines, hormones and growth factors have been done, that

relate ZO-1 abundance with the degree of tightness of the junction. For example, IL-15 up-

regulates ZO-1 and fastens intestinal monolayers (Nishiyama et al., 2001), while IL-3 and IL-4 in

lung epithelia, decrease ZO-1 expression and the barrier function of TJs (Ahdieh et al., 2001).

Pathogens and their toxins also modify ZO-1 expression an epithelial permeability. Thus,

Entamoeba hystolytica, alters the TER and paracellular flow of enteric monolayers and induces

degradation of ZO-1 (Leroy et al., 2000), and Clostridium difficile toxin A increases paracellular

permeability of colonic epithelia and delocalizes ZO-1 from the TJ (Chen et al., 2002a). In

contrast, glycoprotein E of Varicella-Zoster increases translocation of ZO-1 to the cell membraneand augments the TER (Mo et al., 2000). Many of these correlations should however be taken

with caution as changes in ZO-1 expression do not necessarily imply that this protein is the direct

target of the treatment employed. Instead alterations in ZO-1 could arise as a consequence of the

modification of another key TJ component. One of the few cases in which the direct participation

of ZO-1 in the development of tighter junctions has been demonstrated is constituted by the study

of glucocorticoid treatment in trabecular endothelial cells of the eye. In this case, inhibition of

ZO-1 expression with an specific antisense, abolished the dexamethasone-induced increase in

resistance, supporting the idea that ZO-1 is involved in development and maintenance of TER

(Underwood et al., 1999).

The role played by ZO-1 in tumorigenesis remains widely unexplored. However, ZO-1 isstarting to be considered a tumor suppressor since deletions or mutations in its gene produce

overgrowth, and down regulation of its expression is found coupled to breast cancer progression

(Hoover et al., 1998). The participation of ZO-1 in tumor suppression is complex, as many

additional factors appear to be intertwined. For example, in breast cancer cells, insulin-like

growth factor I receptor (IGF-IR) induces E-cadherin mediated cellcell adhesion by up-

regulating ZO-1. The expression of IGF-IR and ZO-1 increase growth and survival of the

primary tumor but in contrast, may reduce cell metastasis (Mauro et al., 2001). Vitamin D3

promotes the differentiation of colon carcinoma cells by the induction of E-cadherin and ZO-1,

inhibition of b-catenin signaling and translocation of ZO-1 from the nucleus to the plasma



18/44

membrane (Palmer et al., 2001). In opposition to the above observations, ZO-1 is over expressed

in primary and metastasic pancreatic cells (Kleeff et al., 2001). The reason for this astonishing

difference is not known, but reveals that ZO-1 may act as a tumor suppressor only in specific

cancers.

Table 2

Appearance of TJ proteins during embryonic development

Protein Stage at which first detected Assembly into TJ Model References

Claudin-1 ND 32-cell stage Mouse Fleming et al. (2001)

Claudin-5 Barely detectable on

embryonic day 5 (early

stage)

Embryonic day 10 (near the

beginning of the

intermediate stage)

Chick

RPE

Kojima et al. (2002)

Xcla Throughout all embryonic

stages

Blastula stage Xenopus Brizuela et al. (2001)

Occludin 7275 kDa band,

throughout all embryonic

stages, decreases from late

blastocyst onwards 65

67 kDa band, throughout

all embryonic stages,

increases from early

blastocyst onwards 58 kDa

band, throughout all

embryonic stages, decreases

from compact 8-cell

embryos onwards

Early 32-cell stage, just

prior to blastocele

cavitation

Mouse Sheth et al. (2000b)

Throughout all embryonic

stages

2-cell stage after cingulin

and ZO-1 incorporation

Xenopus Fesenko et al. (2000)

JAM ND 8-cell stage Mouse Fleming et al. (2001)

ZO-1 a Throughout all embryonic

stages

Punctuate staining at

compact 8-cell embryos

Mouse Sheth et al. (1997)


stages

2-cell stage, after cingulin

incorporation


ZO-1 a Beginning of the blastocyst

stage

32-cell stage, just prior to

the early blastocyst stage

Mouse Sheth et al. (1997)


stages

2-cell stage, after cingulin

incorporation


AF-6 ND Observed at 7.5 days post-

coitum

Mouse Zhadanov et al. (1999)

Cingulin Throughout all embryonic

stages

16-cell stage Mouse Javed et al. (1993)


stages

First cell division, 2-cell

stage

Xenopus Cardellini et al. (1996)

Rab13 Throughout all embryonic

stages

Punctuate staining at

compact 8-cell embryos

Mouse Sheth et al. (2000a)

ND, not determined; RPE, retinal pigment epithelium; Xcla, Xenopus claudin.



19/44

In Drosophila, a ZO-1 homologue named Tamou, is involved in the signaling pathway that

activates the expression of the repressor gene emc which participates in neural development

(Takahisa et al., 1996).

3.1.1.2. ZO-2. ZO-2, a 160 kDa molecule, was originally identified as a TJ protein due to its co-

immunoprecipitation with ZO-1 (Gumbiner et al., 1991). Further studies demonstrated that

this association proceeds through the respective second PDZ domains of each molecule (Wittchen

et al., 1999). ZO-2 interacts as well with other tight and adherens junction associated mole-

cules: with claudin by its first PDZ module (Itoh et al., 1999a), with occludin by its GK

region (Itoh et al., 1999b), and with cingulin (Cordenonsi et al., 1999a; DAtri et al., 2002) and

a-catenin (Itoh et al., 1999b). The proline-rich domain of ZO-2, located at the carboxyl terminal

end of the protein binds to actin (Wittchen et al., 1999) and to protein 4.1 (Mattagajasingh et al.,

2000).

The sequence of ZO-2 contains NLS (Gonzalez-Mariscal et al., 1999) and NES (Islas et al.,2002). In sparse monolayers ZO-2 is conspicuously present at the nucleus in speckles where it co-

localizes with splicing factor SC35 (Islas et al., 2002). Recent evidence has indicated that ZO-2

associates both at the nuclei and TJ with transcription factors Fos, Jun and C/EBP (Betanzos

et al., 2001). These results thus suggest a role for ZO-2 in signaling to the nucleus the adhesion

state of the monolayer.

Although tyrosine phosphorylation of ZO-2 has been reported in v-src transfected epithelial

cells (Takeda and Tsukita, 1995), and six putative tyrosine phosphorylation sites are present

in ZO-2 sequence, the two-dimensional phosphoaminoacid analysis of native ZO-2 does

not reveal tyrosine phosphorylated residues in confluent monolayers nor in those with

disassembled junctions due to calcium chelation (Avila-Flores et al., 2001). Instead ZO-2 is

significantly phosphorylated in serine and threonine residues specially when TJs are eitherabsent or disassembled due to Ca2 removal. This increased phosphorylation is due to the action

of both cAMP-dependent protein kinase (PKA) and PKC, particularly by the atypical isoforms l

and z:ZO-2 has recently been identified as a candidate tumor suppressor protein. This assertion

responds to the observation that ZO-2 expression is either lost or significantly decreased in the

majority of breast cancer lines and adenocarcinomas, although it is mostly present in colon

cancers and prostate carcinomas (Chlenski et al., 2000). The ZO-2 gene employs two alternative

promoters that give rise to two ZO-2 isoforms that differ at their amino terminal portion by 23

amino acids. Although both isoforms are present in normal tissues, the longer one is absent in

most pancreatic cancers (Chlenski et al., 1999a, b). Moreover, over-expression of ZO-2 suppressesthe neoplastic growth of cells activated by Ras V12, polyomavirus middle T protein and

adenovirus type 9 oncogenic determinant E4. The mechanism underlying this tumor growth arrest

is still poorly understood, however, sequestration of ZO-2 in the cytoplasm with tumorigenic

proteins is observed (Glaunsinger et al., 2001).

3.1.1.3. ZO-3. ZO-3 was originally identified as a 130 kDa phosphoprotein which co-

immunoprecipitates with the ZO-1/ZO-2 complex (Balda et al., 1993). Further studies however

demonstrated that this interaction proceeds through association between the second PDZ

domains of ZO-1 and ZO-3, while no direct binding appears to take place, at least under low



20/44

stringency conditions, between ZO-3 and ZO-2 (Wittchen et al., 1999). ZO-3 associates by its first

PDZ to claudins (Itoh et al., 1999a) and via both its amino and carboxyl terminal halves to

occludin (Haskins et al., 1998) and cingulin (Cordenonsi et al., 1999a; Wittchen et al., 2000). The

carboxyl terminal end of ZO-3 contains the class I PDZ binding motif TDL, that binds to the 6thPDZ domain of PATJ (Roh et al., 2002a). In contrast to ZO-1 and ZO-2, the amino terminal half

of ZO-3 associates to actin (Wittchen et al., 2000). ZO-3 associates to the PDZ binding motif

present in Connexin 45 (Kausalya et al., 2001), suggesting it might have a role in the targeting or

localization of gap junctions to specialized domains of the plasma membrane. ZO-3 does not

posses the long carboxyl tail that characterizes ZO-1 and ZO-2. Instead in ZO-3 the proline-rich

region typical of ZO proteins is located between the second and third PDZ domains (Haskins

et al., 1998).

The sequence of ZO-3 contains two putative bipartite NLS (Gonzalez-Mariscal et al., 1999)

and one NES (Islas et al., 2002), although no studies have yet reported its presence at the

nuclei.Transfection with the amino terminal half of ZO-3 (13 PDZ domains) delays the assembly of

tight and adherens junction (Wittchen et al., 2000), suggesting that the carboxyl terminal half that

associates with occludin and cingulin is crucial for TJ to assemble.

3.1.1.4. Pals1. InCaenorhabditis elegans, three PDZ containing proteins Lin-2, Lin-7 and Lin-10,

are necessary for the basolateral targeting of the Let-23 growth factor receptor (Kaech et al.,

1998). Pals1 is a recently discovered protein associated with Lin-7, that localizes at epithelial TJ

(Kamberov et al., 2000; Roh et al., 2002b). It is a MAGUK protein that contains one PDZ

module, SH3 and GK regions. Between the latter two a 4.1 binding domain is found. Pals1 is

different from other TJ MAGUKs, as it lacks the acidic and proline-rich domains present in ZO

proteins, and instead contains two Lin-7 binding modules termed L27 domains. The L27Cmodule binds Lin-7, while the L27N domain targets Pals1 to TJ by binding to the MRE region of

PATJ and MUPP1, both recently discovered PDZ-containing proteins of the TJ. The extreme

amino terminal region of Pals1 contains a 125 amino acid domain that bears no similarity to other

proteins and is thus referred to as unknown 1 (U1).

At the TJ, Pals1 exists in a ternary complex with PATJ and the human Crumbs homo-

logue CRB1 (Roh et al., 2002b). Crumbs together with Disc lost (Dlt) functions as an apical

polarity determinant in Drosophila. In mammalian epithelia, Pals1 interacts directly with CRB1

and thus serves as an adaptor protein, mediating the indirect interaction between CRB1 and

PATJ.

3.1.2. MAGI, the MAGUK inverted proteins of the TJ

TheMAGUK inverted proteins named MAGI have three unique structural features: (A) they

contain six PDZ domains, one located at the amino terminus and the rest at the carboxyl terminal

domain, (B) the SH3 region is replaced by two WW domains, and (C) the GK domain is found

after the first PDZ module at the amino terminus of the protein. Three members of this family

have been so far described.

3.1.2.1. MAGI-1/BAP-1. This protein co-localizes with ZO-1 at the TJ of epithelial cells (Ide

et al., 1999b). MAGI-1 appears to be a tumor suppressor, since the tumorigenic potential of the



21/44

viral oncoproteins of the human adenovirus type 9 (E4-ORF1) and the high-risk human

papillomaviruses (E6) depends on their ability to sequester or to target MAGI-1 for degradation

(Glaunsinger et al., 2000). MAGI-1 interacts at the TJ with the signaling molecule GEP. The

latter is aGDP/GTPexchangeprotein specific for the small G protein RAP (Mino et al., 2000). Atthe glomerular podocytes MAGI-1 associates with the transmembrane glycoprotein megalin

(Patrie et al., 2001). Three splicing variants of MAGI-1 have been characterized: MAGI-1a, -1b

and -1c. MAGI-1a is found in soluble and insoluble cellular fractions. MAGI-1b localizes to the

basolateral membrane of epithelial cells and forms complexes with b-catenin and E-cadherin

during junction formation (Dobrosotskaya et al., 1997). MAGI-1c contains three bipartite

nuclear localization signals and is predominantly found at the nucleus of epithelial cells

(Dobrosotskaya and James, 2000). At the neuromuscular junction MAGI-1c interacts with

MuSK, a tyrosine kinase receptor active in differentiation (Strochlic et al., 2001).

3.1.2.2. MAGI-2. MAGI-2 is found at synaptic junctions where it functions as a scaffoldingprotein (MAGI-2/S-SCAM,synapticscaffoldingmolecule). The identification of MAGI-2 ligands

is important for the elucidation of the structure of synaptic junctions. It interacts through its GK

region, with SAPAP (SAP90associatedprotein), and by its PDZ domains with NMDA receptors,

neuroligins, MAGUIN (membrane associated guanylate kinase-interacting protein-1), b1-

adrenergic receptor, and GEP (Hirao et al., 1998; Ohtsuka et al., 1999; Xu et al., 2001; Yao

et al., 1999). MAGI-2 associates with atrophin-1 (MAGI-2/AIP1, atrophininteractingprotein), a

protein with a polyglutamine repeat expansion, which is responsible for dentatorubral and

pallidoluysian atrophy (Wood et al., 1998). In epithelial cells MAGI-2 localizes the TJ, where it

forms a complex through its second PDZ repeat with the carboxyl terminal end of PTEN. The

latter is a tumor suppressor that functions as a catalyst for the removal of 3-phosphate from

phosphatidylinoditol 3,4,5,-triphosphate. This phospholipid is a product of PI3-kinase, and isinvolved in the activation of the protooncogene AKT/PKB that suppresses apoptosis (Wu et al.,

2000a). Phosphorylation of PTEN tail causes a conformational change that results in the masking

of the PDZ binding domain (Vazquez et al., 2001). MAGI-2 also bindsb and d catenin (Ide et al.,

1999a; Kawajiri et al., 2000). Three isoforms of MAGI-2, a; b and g have been characterized(Hirao et al., 2000).

3.1.2.3. MAGI-3. MAGI-3 localizes in epithelial cells at the TJs where it binds through its second

PDZ domains to the tumor suppressor PTEN. In the synaptic junction MAGI-3 associates

through its fifth PDZ repeat to the NMDA receptor (Wu et al., 2000b).

3.1.3. PAR proteins of the TJ

PAR are partitioning-defective proteins, required for embryonic polarity. In C. elegans for

example, PAR-3 localizes asymmetrically at the anterior periphery of one cell embryos. Mutations

in PAR-3 alter the polarized distribution of other proteins involved in cell fate determination and

the orientation of the mitotic spindles in successive cell cycles (Bowerman et al., 1997; Ebnet et al.,

2001; Guo and Kemphues, 1996).

In epithelial cells PAR-3, a three PDZ containing protein, localizes at TJs, where it directly

binds through its second PDZ domain to the carboxyl terminus of JAM (Itoh et al., 2001). PAR-3

forms a complex with PAR-6, a one PDZ possessing molecule with a CRIB domain (Johansson



22/44

et al., 2000), and atypical PKCs l and z (PAR-3/ASIP, atypical PKC isotype-specific

interacting protein) (Izumi et al., 1998). Expression of a dominant negative mutant atypical

PKC causes mislocalization of PAR-3 and cell surface polarity impairments, suggesting a role

for the PAR protein complex in the apico-basal polarization of epithelial cells (Suzuki et al.,2001).

Par-6 inhibits TJ reassembly after junctional disruption induced by Ca2 depletion, but does

not inhibit adherens junction formation. The amino terminal fragment of PKC z;which binds toPAR-6, also inhibits TJ assembly (Gao et al., 2002). In accordance, we have observed that the

MAGUK protein ZO-2 is a phosphorylation target for atypical PKCs, and that the

phosphorylated state of ZO-2 restrains its capacity to operate at the junctional complex (Avila-

Flores et al., 2001).

PAR-6 is a binding partner for the Rho GTPases Cdc42-GTP and Rac1 (Johansson et al.,

2000). Hence PAR-6 is a key adaptor that links Rac1, Cdc42 and atypical PKCs to PAR-3.

Binding of Cdc42-GTP to PAR-6 enhances the activity of the atypical PKCs (Yamanaka et al.,2001), and thus the activated Cdc42 disrupts TJs (Gao et al., 2002).

Recruitment of the PAR-3/PAR-6/CDC-42/aPKC complex to the TJs appears to be mediated

by the tethering of PAR-3 to JAM (Itoh et al., 2001).

3.1.4. MUPP1

MUPP1 is a protein that contains 13 PDZ domains (multi-PDZ domainprotein 1); therefore it

is suggested to function as a multivalent scaffold protein. It is exclusively concentrated at TJs

where it interacts through its PDZ 10 with claudins and by its PDZ 9 with JAM (Hamazaki et al.,

2002). Thus, MUPP1 functions as a cross linker between claudin-based TJ strands and JAM

oligomers in TJs. Other integral proteins might also be tethered to the claudin-based TJ strands

through MUPP1 molecules. In fact, a serotonin receptor (Ullmer et al., 1998), the protooncogenethat encodes the receptor for the stem cell factor named c-Kit (Mancini et al., 2000) and the

membrane spanning proteoglycan NG2 (Barritt et al., 2000) respectively bind to MUPP1s PDZ

domains 10, 10 and 1, although their recruitment to TJs remains unclear.

Within the amino terminus of MUPP1 a novel proteinprotein interaction domain has been

found. Since this domain has the ability to bind and recruit MAGUK protein Pals1 to TJ, it has

been named MAGUK recruitment domain (MRE) (Roh et al., 2002b).

The major oncogenic determinant for human adenovirus type 9 (E4-ORF1) aberrantly

sequesters MUPP1 within the cellular cytoplasm, whereas the high-risk human papilloma viruses

determinant HPV-18 E6 targets MUPP1 for degradation. Consequently MUPP1 is proposed to

be negatively involved in regulating cellular proliferation (Lee et al., 2000).

3.1.5. AF-6/Afadin

AF-6 is the ALL-1 fusion partner at chromosome 6. The ALL-1/AF6 chimeric protein is a

critical product associated with acute human leukemia (Prasad et al., 1993).

AF-6 is a 205 kDa multidomain protein that contains two Ras-binding domains within the

amino terminus, followed by kinesin and myosin like domains, and a PDZ module at the middle

of the protein. At the carboxyl terminal AF-6 contains three proline-rich domains followed by a

F-actin binding region. Af-6 is a component of tight (Yamamoto et al., 1997) and adherens

(Mandai et al., 1997) junctions.



23/44

AF-6 is a target of Ras and of several related proteins (Rap1A, Rit, Rin and M-Ras) (Boettner

et al., 2000; Quilliam et al., 1999; Shao et al., 1999; Yamamoto et al., 1997). ZO-1 interacts with

the Ras binding domains of AF-6 and this association is inhibited by activated Ras. The over

expression of activated Ras perturbs cellcell contacts and decreases the accumulation of AF-6and ZO-1 at the cell borders (Yamamoto et al., 1997). AF-6 also interacts with cingulin, another

TJ plaque protein (Cordenonsi et al., 1999a).

AF-6 can directly associate through its PDZ domain with the TJ integral protein JAM. Since

both AF-6 and ZO-1 associate to the PDZ type II binding motif of JAM (Ebnet et al., 2000), these

complexes are mutually exclusive and should therefore be involved in different functions. The fact

that AF-6 is transiently expressed at the cell contacts of epithelial lines suggests, that this protein

might be more important for the formation of junctions that for the maintenance of stable

junctional complexes.

AF-6 is also a constituent of a novel cellcell adhesion system named NAP, which localizes at

adherens junctions (Asakura et al., 1999). The NAP complex is composed ofnectin, afadin andponsin. Nectin, whose name derives from the Latin word necto meaning to connect is a

molecule member of the immunoglobulin superfamily, previously identified as a poliovirus and

alpha herpes virus receptor. It associates to afadin through its carboxyl terminal PDZ binding

motif (Takahashi et al., 1999). Ponsin, whose name derives from the Latin word pons, meaning

bridge, is capable of binding to the proline-rich regions of AF-6 through its SH3 domain, and to

vinculin. Thus ponsin connects the NAP system with the cadherincatenin adhesion junctions

(Mandai et al., 1999).

AF-6 associates with profilin, a protein that activates monomeric actin units for subsequent

polymerization and participates in cortical actin assembly (Boettner et al., 2000). Thus Af-6

through its interaction with profilin could modulate actin modeling at the adhesion complexes.

AF-6 forms a complex with and serves as a substrate for Fam, a deubiquitinating enzymeproduct of thefat facetsgene (Taya et al., 1998). Fam probably maintains the stability of cellcell

contacts by deubiquitinating the components of intercellular adhesions, therefore its recruitment

to tight and adherens junctions through AF-6 might be crucial.

AF-6 is a critical regulator of cellcell junctions during development. Thus a null mutation in

the Af6 locus disrupts epithelial intercellular junctions and cell polarity during mouse

development (Zhadanov et al., 1999). Furthermore, during Drosophila embryogenesis, the

reduced expression of ZO-1 and the AF-6 homologue, Canoe, generates a failure in the embryonic

dorsal closure (Takahashi et al., 1998).

AF-6 has a splicing variant of 190 kDa (s-AF-6) that lacks the F-actin binding domain and the

third proline-rich domains, and is abundantly expressed in neuronal tissues (Mandai et al., 1997).AF-6 localizes at post-synaptic densities where it interacts and clusters with some Ephrin receptor

tyrosine kinases (RTKs) (Buchert et al., 1999). AF-6 is a phosphorylation substrate of the Eph

receptor. Since in neural tissues AF-6 accumulates in post-synaptic densities whereas ZO-1

localizes at pre-synaptic terminals, it is assumed that in this system they fulfill different roles.

3.1.6. PATJ

PATJ contains 10 PDZ domains and concentrates at the TJ of epithelial cells (Roh et al.,

2002b), although it is also found at the apical plasma membrane. Over expression of PATJ

disrupts the TJ localization of ZO-1 and ZO-3, thus suggesting that it might be involved in



24/44

regulating the integrity of TJs. In epithelia antibodies against PATJ recognize proteins of 230,

200, 135 and 75 kDa bands (Lemmers et al., 2002).

PATJ was cloned by homology to Drosophila INAD and thus named human INAD-like

protein (hINADl) (Philipp and Flockerzi, 1997). INAD is a protein with 5 PDZ domains thatparticipates in photo-transduction in Drosophila. Recent findings however, have shown that its

similarity is stronger for Dlt. The latter is a 4 PDZ containing protein that co-localizes at the

apical region with dCrumbs, and together play a crucial role in regulating cell polarity in

Drosophila. The organization of PATJ is similar to MUPP1s, thus it has been suggested that they

are paralogues.

PATJ recruits Pals1 to TJ (Pals1 associated tight junctions protein) through the specific

interaction between the L27N domain of latter and the MRE domain of PATJ. Pals1/PATJ

interaction is not crucial for PATJ targeting to TJs. The 6th and 8th PDZ modules of PATJ

interact with ZO-3 and claudin-1, respectively, through the type I PDZ binding domains present

in the carboxyl terminal ends of these proteins. While the interaction with the former appears tobe crucial for PATJ targeting to TJ, deleting the 8th domain has little effect on PATJ localization.

Therefore PATJ is proposed to be recruited to TJ by its association to ZO-3 (Roh et al., 2002a).

3.2. TJ proteins lacking PDZ domains

Several submembranous proteins of the TJ do not contain PDZ domains. Some are involved in

vesicular trafficking to the TJ, other bridge integral TJ proteins to the actin myosin cytoskeleton,

certain are transcription factors or proteins with nuclear functions, while the task of others still

remains unclear (Table 3).

3.2.1. CingulinCingulin was named from the Latin word cingere which means to encircle. It is a 140

160 kDa protein that localizes at the TJ submembranous region of epithelial and endothelial cells

(Citi et al., 1988). Cingulin has globular head and tail domains and a central a-helical rod region.

The latter is responsible for the formation of coiled-coil parallel dimmers which can further

aggregate though intermolecular interactions. The globular head of cingulin interacts with ZO-2

(residues 150295 ofXenopuscingulin), ZO-3, AF-6, JAM, F-actin and myosin. ZO-3 and myosin

are also capable of interaction with the rest of the cingulin molecule (Bazzoni et al., 2000;

Cordenonsi et al., 1999a; DAtri and Citi, 2001; DAtri et al., 2002). The interaction with ZO-1 is

complex, since a sequence remarkably conserved at the head ofXenopus(residues 4155), mouse

(residues 4357) and human (residues 4862) cingulin denoted as ZIM, is required for ZO-1binding in pulldown experiments (DAtri et al., 2002). Yet thisZO-1interactionmotif appears not

to be sufficient for ZO-1 binding, as GST fusion proteins containing only such cingulin residues

fail to interact with ZO-1. The amino terminal region of cingulin, containing the ZIM domain, is

capable of targeting transfected cingulin to the junctions only when fused to rod-tail sequences.

Furthermore, deletion of ZIM does not abolish junctional recruitment of cingulin. These results

therefore suggest the requirement of multiple protein interactions for junctional localization to

proceed. In fibroblasts that lack the molecular context of TJ (e.g. occludin and claudins) but

contain cadherin-based cellcell adhesion sites with ZO-1, the ZIM domain of cingulin is required

for recruitment to cellcell adhesion sites. The interaction between cingulin and ZO-1 might be



25/44


26/44

3.2.2. Symplekin

The name symplekin derives from Greek and means to weave or tie together. This 127 kDa

protein, is found at the TJ of epithelial cells, but is absent from endothelia. Symplekin can also be

detected in the nucleoplasm of a wide range of cells including those devoid of any stable cellcellcontact (Keon et al., 1996). In Xenopus oocytes symplekin has been located at Cajal bodies.

Symplekin contains 4 putative NLS and interacts at the nuclei with CstF and CPSF, both

constituents of the 30-end cleavage and polyadenylation complex of mRNA precursors. Symplekin

also has a significant similarity with the yeast protein PTA1, that is another component of the

polyadenylation machinery (Hofmann et al., 2002; Takagaki and Manley, 2000). Surprisingly,

symplekin is found in the cytoplasm associated to CPSF proteins, suggesting their involvement in

cytoplasmic polyadenylation and in the regulation of translation.

3.2.3. 7H6 antigen/barmotin

A monoclonal antibody generated against a rat bile canaliculus rich membrane fraction,recognized a novel TJ cortically associated protein named 7H6 antigen (Zhong et al., 1993). This

155 kDa protein shows homology to the SMC family. These are alpha helical coiled-coil proteins

with a putative ATPase domain (Ezoe et al., 1995). Due to these structural characteristics antigen

7H6 was also named barmotin (Muto et al., 2000). Phosphorylation of 7H6/barmotin is closely

related to its localization at the TJs, thus cellular ATP depletion reversibly dissociates 7H6/

barmotin from the junction (Zhong et al., 1994a). 7H6/barmotin appearance at TJs correlates

with the maintenance of the paracellular barrier function in epithelial (Zhong et al., 1993),

endothelial (Satoh et al., 1996) and mesothelial (Tobioka et al., 1996) cells. Furthermore, in HGF-

induced cell spreading (Muto et al., 2000), liver carcinogenesis (Zhong et al., 1994b), primary

biliary cirrhosis (Sakisaka et al., 2001), under treatment with a bacterial lipopolysacharide

(Kimura et al., 1997) and upon exposure to Helicobacter pylori(Suzuki et al., 2002), a reducedexpression of 7H6/barmotin is found that correlates with disruption of cellular polarity,

adhesiveness and an increased paracellular permeability.

The appearance of 7H6/barmotin at developing TJs suffers a transition from a dotted

arrangement to a continuous honeycomb linear appearance only after ZO-1 and occludin have

completely surrounded the cellular borders (Kimura et al., 1996). Therefore instead of being a

constituting protein of the TJ, 7H6/barmotin might play a role in TJ maintenance and

maturation.

3.2.4. Rab proteins

Rab proteins are monomeric molecules that bind GDP/GTP and exhibit an intrinsic GTPaseactivity. They constitute a branch of the Ras superfamily of G proteins and are closely related to

the yeast Ypt1 and Sec4 gene products. Rab proteins are involved in vesicle trafficking and thus

cycle between cytosolic and membrane bound forms (Novick and Zerial, 1997).

3.2.4.1. Rab13. In fibroblasts Rab13 associates with vesicles throughout the cytoplasm, while in

epithelial cells it accumulates at the TJ (Zahr

Date post:	03-Jun-2018
Category:	Documents
Upload:	blanca-estela-jaramillo-loranca
View:	218 times
Download:	0 times

Jaramillo BE 2003

Documents