Báo cáo khoa học: A novel four transmembrane spanning protein, CLP24 A hypoxically regulated cell junction protein pdf

A novel four transmembrane spanning protein, CLP24A hypoxically regulated cell junction protein Jonathan Kearsey, Silvere Petit, Catherine De Oliveira and Fabien Schweighoffer ExonHit Th

A novel four transmembrane spanning protein, CLP24

A hypoxically regulated cell junction protein

Jonathan Kearsey, Silvere Petit, Catherine De Oliveira and Fabien Schweighoffer

ExonHit Therapeutics, Paris, France

A novel hypoxically regulated intercellular junction protein

(claudin-like protein of 24 kDa, CLP24) hasbeen identified

that shows homology to the myelin protein 22/epithelial

membrane protein 1/claudin family of cell junction proteins,

which are involved in the modulation of paracellular

per-meability The CLP24 protein containsfour predicted

transmembrane domains and a C-terminal protein–protein

interaction domain These domains are characteristic of the

four transmembrane spanning (tetraspan) family of

pro-teins, which includes myelin protein 22, and are involved

in cell adhesion at tight, gap and adherens junctions

Expression profiling analyses show that CLP24 is highly

expressed in lung, heart, kidney and placental tissues

Cellular studies confirm that the CLP24 protein localizes

to cell–cell junctionsand co-localizeswith the b-catenin adherens junction-associated protein but not with tight junctions Over-expression of CLP24 results in decreased adhesion between cells, and functional paracellular flux studies confirm that over-expression of the CLP24 protein modulatesthe junctional barrier function These data therefore suggest that CLP24 is a novel, hypoxically regu-lated tetraspan adherens junction protein that modulates cell adhesion, paracellular permeability and angiogenesis Keywords: adherens; angiogenesis; claudin; DATAS; hyp-oxia

Endothelial and epithelial cell sheetsline all the cavitiesof

the body and are linked by specialized adhesive junctions

that provide a selective barrier for the passage of plasma

proteins, circulating cells, water and/or solutes Two types

of adhesive junctions, namely tight junctions and adherens

junctions, play a major role in controlling this paracellular

barrier function [1,2] Tight junctionsare required at the

apical face of the cell junctionsin order to maintain a

selective paracellular barrier Adherens junctions are located

below the tight junction at the apical junction and are

required for tight junction formation and the maintenance

of barrier integrity Adhesion junctions also contribute to

vascular morphogenesis in endothelial cells [3] Adherens

junctionsundergo changesfollowing a reduction in oxygen

levels(hypoxia), in order to allow the initiation of an

angiogenic response that requires increased vascular

per-meability, endothelial cell proliferation and migration [3,4]

ln addition to providing barrier and morphological

func-tions, these cell junctions are targeted by a variety of

signaling processes involved in normal physiology (cell

growth and differentiation) and pathology [2,5]

All cell junctions, including tight and adherens junctions, are composed of transmembrane proteins that show structural, but often little sequence, homology These transmembrane proteins comprise four transmembrane domains, together with extracellular loop regions that interact adhesively with complementary molecules in adja-cent cellsto form the junction Thisstructural family of proteins (tetraspan proteins) includes connexins/innexins and peripheral myelin protein 22 (PMP22)/epithelial membrane protein 1 (EMP1)/claudin members, which are involved in gap and tight junctions, respectively [6,7] Claudins have been shown to be one of the structural adhesive componentsof tight junctions[2] PMP22 was originally isolated as a growth arrest specific transcript (Gas 3), induced following serum deprivation of fibroblasts [8] PMP22 was subsequently shown to be a major component

of myelinated fibersin the peripheral nervoussystem and associated with tight junctions [6,9,10] Mutations of the gene encoding PMP22 cause Charcot–Marie–Tooth disease Type 1A, Dejerine–Sottassyndrome and hereditary neuro-pathy [11] Both Charcot–Marie–Tooth disease and Dejerine–Sottas syndrome are sensorineural peripheral polyneuropathies, the most commonly inherited disorder

of the peripheral nervous system [12] Sequence similarity and co-localization studies show that PMP22 is a tight junction associated transmembrane protein in both neur-onal and non-neurneur-onal cells[6] Thus, the PMP22 gene product isa dual-function protein, involved in both tight junctionsadhesion and cell proliferation

This study describes the identification and characteriza-tion of a novel transmembrane junccharacteriza-tional protein with structural homology to the tetraspan family of proteins Thisgene wasidentified in a screen for hypoxically regulated genesin endothelial cellsthat could provide angiogenic therapeutic targets Sequence analyses show that this novel

Correspondence to F Schweighoffer, ExonHit Therapeutics, 65 Bd

Masse´na, 75013 Paris, France Fax: + 33 1 53 94 77 04,

Tel.: + 33 1 53 94 77 69, E-mail: fabien.schweighoffer@exonhit.com

Abbreviations: CLP24, claudin-like protein of 24 kDa; EGFP, green

fluorescent protein; EMP1, epithelial membrane protein 1; EST,

expressed sequence tag; HKG, housekeeping gene; HMEC, human

microvascular endothelial cells; PDZ, protein–protein interaction

domain; PMP22, myelin protein 22; TMHMM, transmembrane

hidden Marckoff model: ZO-1, Zona Occluden-1.

Note: a website is available at http://www.exonhit.com/

(Received 11 February 2004, revised 19 April 2004,

accepted 26 April 2004)

protein is most closely related to PMP22, a claudin cell

junction-associated family member This novel gene

prod-uct hastherefore been called claudin-like protein of 24 kDa

(CLP24) The protein product of the CLP24 gene contains

four transmembrane spanning domains together with a

C-terminal protein–protein interaction (PDZ) domain CLP24

is most highly expressed in lung, heart, kidney and placenta,

showing little similarity to the expression patterns of other

PMP22/EMP1/claudin members However, this is not

unexpected as distinct tissue-distribution profiles are

observed for all the PMP22/EMP1/claudin family members,

which allowsregulation of paracellular specificity between

different endothelial cell types[2,13]

Expression studies using recombinant CLP24-enhanced

green fluorescent protein (EGFP) demonstrated that CLP24

localizes to cell junctions Co-localization studies were

performed using recombinant CLP24-EGFP together with

antibodies against either the cell adhesion molecule,

b-catenin, or the tight junction Zona Occluden-1 (ZO-1)

associated protein These experiments demonstrated that

CLP24 waslocalized to regionsof the membrane associated

with adherens junctions, but with little association to the

tight junction componentsat the apical face

Over-expres-sion of CLP24 increased paracellular permeability across an

endothelial monolayer, confirming that CLP24 actsasa

structural component in cell junctions These data therefore

suggest that a novel, although distantly related, member of

the claudin/PMP22 family of proteinshasbeen identified

However, CLP24 appearsto be distinct from many PMP22/

EMP1/claudin members, in that CLP24 influences

paracel-lular permeability through itsinteraction with adherens,

rather than tight junction components

Materials and methods

Cell culture

Immortalized human microvascular endothelial cells

[HMEC-1; CDC (Centre for Disease Control and

Preven-tion), Atlanta, GA, USA) were cultured in MCDB-131

medium (Sigma) supplemented with 15% (w/v)

heat-inactivated fetal bovine serum (Invitrogen), 2 mM

L-glut-amine (Invitrogen), 100 UÆmL)1 penicillin (Invitrogen),

100 lgÆmL)1streptomycin (Invitrogen), 10 ngÆmL)1human

recombinant EGF (Invitrogen) and 1 lgÆmL)1

hydrocorti-sone MDCK (a canine normal kidney cell line; ATCC) and

Calu-6 (a human, lung carcinoma cell line; ATCC) cells

were cultured in Dulbecco’smodified Eagle’smedium

supplemented with Glutamax (Invitrogen), 10% (w/v)

heat-inactivated fetal bovine serum, 100 UÆmL)1penicillin,

and 100 lgÆmL)1 streptomycin For hypoxic treatments,

cellswere grown in an atmosphere of 3% O2in an IG750

incubator (Jouan, France), or in the presence of

100 lgÆmL)1 desferrioxamine The HMEC-1 and Calu-6

cell lines were chosen as they both express CLP24 mRNA

(data not shown for Calu-6)

Differential analysis of transcripts with alternative

splicing (DATAS)

Thistechnology hasbeen previously described by

Sch-weighoffer et al [14] Briefly, first-strand cDNAs were

reverse transcribed from HMEC-1 total RNA (treated either with normoxia or hypoxia) using the Superscript II

RT kit (Invitrogen) and an anchored biotinylated oligo-dT25 cDNAswere then treated with RNAse I followed

by proteinase K and phenol/chloroform extraction mRNA from normoxic HMEC-1 cellsand cDNA from hypoxic HMEC-1 cellswere mixed at a 1 : 1 molar ratio and precipitated using sodium acetate and ethanol The recip-rocal experiment with mRNA from hypoxic HMEC-1 cells and cDNA from normoxic HMEC-1 cellswasalso performed The pellet was redissolved in 80% formamide/ 0.1% SDS, and heteroduplexeswere allowed to form by denaturation at 85C and gradual cooling to 40 C Heteroduplexes were then isolated using streptavidin beads (no 112.06; Dynal) and the single-stranded RNA released

by the action of RNAse H (no 18021-071; Invitrogen) Residual cDNA was removed by extraction using strept-avidin beads and treatment with DNAse I The isolated single-stranded RNA molecules were reverse transcribed using the Superscript II RT kit and random hexamer primers The cDNAs were amplified by PCR (using the DOP PCR methodology [15]) with six different anchored degenerate primers, purified using spin columns and cloned (pCR II-TOPO vector, no 45-0640; Invitrogen) Following transformation and plating onto LB (Luria–Bertani) plates containing ampicillin (100 lgÆmL)1), recombinant bacterial colonieswere isolated and the cloned cDNA wassequenced Thislibrary wasdesignated HMEC-EHT1

cDNA array generation Each cDNA wasamplified from a bacterial colony by performing PCR amplificationsusing SP6-T7 primers Each PCR product (600 lL final volume) wasconcentrated and visualized on an agarose gel before spotting The PCR productswere arrayed on 8· 12 cm nylon filtersthat were spotted in duplicate using a QPix robot (GenePix 4000; Axon) with the nonredundant human HMEC-EHT1 cDNA library Arrayed nylon filterswere stored at 4C until use

Probe labelling and hybridizations

To make a single probe, 50 ng of total RNA (from normoxic or hypoxic HMEC cells) was used to generate double-stranded cDNA with the MMLV RT (Invitrogen) using an anchored oligo-dT primer [5¢-CCTATTGTTTGT GTGTGTCC-3¢ RN1-oligo(dT)25] PCR amplification, using an RN1 primer, was performed and the amplified productswere measured by a fluorometric method (Pico-Green quantitation kit; Interchim) The cDNAswere labeled with redivue, stabilized [33P]dCTP[aP] (Rediprime

II, Amersham) and hybridized according to the manufac-turer’sinstructions After hybridization, the filterswere quantified by scanning densitometry using a Biorad Mole-cular Imager and evaluated using biostatistical analyses cDNA cloning and construct generation

Total RNA from HMEC-1 and Calu-6 cell lineswere reverse transcribed using the multiscribe RNA polymerase and random hexamer primers(Archive kit; Applera) PCR

amplification of the full-length open reading frame of

CLP24 wasachieved using a proofreading DNA

poly-merase (Platinum Pfx DNA polypoly-merase; Invitrogen), and

by using the sense primer TTTGAATTCCCACCATG

ACCGTGCAGAGACTC (containing the ATG start

codon of CLP24, together with a Kozak sequence and an

EcoRI restriction site), and the antisense primer, AAAG

GATCCAGGCATGGTGACTCCACGTA (containing a

BamHI restriction site) PCR conditions were: 94C for

30 seconds, 58C for 30 seconds, 68 C for 1 min, for 35

cycles The PCR product was cloned in the pCRII TOPO

vector (Invitrogen) and sequenced using an automatic

sequencer (Applied Biosystems model 3100) A C-terminal

EGFP-CLP24 fusion construct was then generated by

cloning the CLP24 open reading frame into vector

pEGFP-N1 (BD Biosciences) using the EcoRI and BamHI

restric-tion sites Recombinant HMEC-1/CLP24-EGFP and

MDCK/CLP24-EGFP cells were established after

transfec-tion with the full-length CLP24 cDNA in pEGFP-N1 (BD

Biosciences) and selection of stably transfected cells with

150 lgÆmL)1and 400 lgÆmL)1geneticin, respectively

Bioinformatic analysis

Bioinformatic analyses were performed using Genetics

Computer Group (GCG) software, including BLAST, the

multiple sequence alignment tool CLUSTALW, and the

pairwise alignment toolGAP(BLOTSUM 55) In addition,

membrane protein prediction (TMHMM),SCANSITEand

PRO-SITEsoftware have been used to characterize CLP24 [16,17]

PCR

The expression of CLP24 in different tissues and cells was

determined by PCR The cDNA from a number of different

human tissues (Clonetech), together with human epithelial

[Calu-6 (a lung carcinoma cell line), RCC4 (a renal

carcinoma cell line), NTERA-2 (a neuronal precursor

epithelial cell line), H1299 (a nonsmall cell carcinoma cell

line), HepG2 (a hepatocellular carcinoma cell line) and the

breast cell lines MDA-MD231, MDA-MB-435, MCF7,

BT549 and T-47D (ATCC) and endothelial (HMEC

and HUVEC) cell lines, were characterized PCR was

performed, using standard PCR conditions

[Amp-litaq (0.075 UÆmL; Applied Biosystems), anti-taq Ig

(0.075 UÆmL; Invitrogen), 15 mM MgCl2, 1· buffer 1

(Applied Biosystems), dNTPs (0.2 mM, Invitrogen) and

0.5 mM of each primer] Thirty-five cyclesof PCR were

performed using an annealing temperature of 60C The

primers selected for the specific expression of CLP24 were

5¢-CCCTAGCAGCGTCGGCT-3¢ and 5¢-CGTTGCGCT

AACCAGGAAAG-3¢, which give an amplicon size of 1002

bp PCR products were visualized following separation on a

1% agarose gel

Real-time quantitative RT-PCR

CLP24differential gene expression has been monitored by

quantitative real-time RT-PCR (Q-RT-PCR) using

Taq-Man technology In brief, total RNA from HMEC-1 and

Calu-6 cells were isolated using the Trizol kit (Invitrogen),

then 5 lg of total RNA was reverse transcribed using the

multiscribe RNA polymerase and random hexamer primers (Archive kit; Applera) Real-time PCR wasperformed on

an ABI Prism 7700 Sequence Detector machine (Applera) and analyzed us ingSDS, version 1.6.3, software The primers (Life Technologies, Inc.) and TaqMan probes (Eurogentec) for the quantification of the CLP24 transcripts were designed using the primer design software,PRIMER EXPRESS (Applera) except for human b-actin where commercially available assay reagents were used (Applera) Initial experi-mentswere performed to define a housekeeping gene (HKG) whose level remained constant following hypoxic treatment Two HKG were tested (b-actin and b2-micro-globulin) and two separate primer sets were tested for b-actin The b-actin primerswere purchased from Applied Biosystems (b-actin control reagent 401846) b2-microglob-ulin primerswere: 5¢-GGACTGGTCTTTCTATCTCTTG TACTAC and 3¢-AGTCACATGGTTCACACGGC Only minor variation acrossthe HKG primer setswasobserved between treated and nontreated samples, and the b-actin HKG was selected for further experimental analyses Primer sequences for CLP24 were: forward primer, 5¢-CGTTTACTGTTATGTCGGTCATAT-3¢ and reverse primer, 5¢- GTTGCGCTAACCAGGAAAGC-3¢; probe sequence: CLP24, 5¢-TGTCGTGGGCCAACCTCGTT CTG-3¢ Specificity of the PCR amplification wasconfirmed

on an agarose gel The PCR reactions were carried out using TaqMan universal PCR master mix (Applera) For CLP24, both primerswere used at 150 nMand the probe at 100 nM The TaqMan PCR reaction conditionswere: 2 min at

50C, 10 min at 95 C, then 40 cycleseach of 15 sat 95 C and 1 min at 60C Each plate contained triplicatesof the test cDNA templates, a standard curve for the individual amplicon, and no-template controlsfor each reaction mix Standard curveswere generated for each amplicon in order

to determine PCR amplification efficiency The Ct value is defined asthe number of PCR cyclesrequired for the fluorescence signal to exceed the detection threshold value The fold difference (F) wascalculated and normalized to the levels for the established HKG, b-actin, according the following formula that isa derivation of that defined by Plaffl et al [18]:

F¼ f½ð1 þ EtargetÞCt target=ð1 þ EHKGÞCt HKGconditionAg= f½ð1 þ EtargetÞCt target=ð1 þ EHKGÞCt HKGconditionBg where: F, fold induction; Etarget, PCR amplification effi-ciency of the target gene; Ct target, threshold cycle of the PCR amplification of the target gene; EHKG, PCR ampli-fication efficiency of the housekeeping gene; Ct HKG, threshold cycle of the PCR amplification of the housekeep-ing gene; condition A, normoxic; condition B, hypoxic Northern blot analysis

A human Northern blot (no 7780-1; Clontech ) wasused and a32P-labeled random-primed DNA probe wasgener-ated using the full-length CLP24 transcript and the Strip-EZ RNA Ambion kit (no 1360) The blot wasprehybridized

in ULTRAhyb hybridization buffer (no 8670, Ambion) for 2 h at 68C and the probe denatured at 95 C for

10 min The blot washybridized overnight at 68C, washed

as described by the manufacturer, and exposed to

Kodak Biomax MS film with a transcreen-HE intensifying

screen

Immunofluorescence confocal microscopy

The following antibodieswere obtained commercially:

mouse against b-catenin (no 610153; BD Biosciences

Pharmingen) and rabbit against ZO-1 (no 61-7300;

Zymed) Cells were plated on coverslips, rinsed twice with

NaCl/Pi(phosphate-buffered saline) and subsequently fixed

in 4% paraformaldehyde for 20 min After washing with

NaCl/Pi, cellswere permeabilized with 0.1% Triton for

5 min Then, cellswere blocked with Powerblock (universal

blocking agent; Biogenex) for 15 min to minimize

nonspe-cific binding Cellswere incubated with primary and

secondary antibodies in blocking buffer for 30 min,

fol-lowed by three washes with NaCl/Piafter each incubation

The primary antibody was visualized using the

appropri-ate Cy3-conjugappropri-ated anti-mouse or anti-rabbit secondary

antibodies[Cy3-conjugated F(ab¢)2 fragments; Jackson

Immuno Research) Finally, the cells were mounted with

fluorescence mounting medium (Fluosave, Molecular

Probes) and viewed with a confocal imaging system (model:

Leica TCS SPZ and software: LCS Software, Leica,

Heidelburg, Germany) Output wavelengthswere 488, 543

and 633 nm, EGFP fluorescence was imaged at 510 nm A

cross-sectional image (x–z) through the junction was

computer generated in order to assess the localization of

proteins within tight or adherens junctions Phase-contrast

images were taken using a Nikon Elispe TE300 microscope

together with a Nikon CoolPix 990 camera

Paracellular tracer flux measurement of a 3 kDa

FITC-labeled dextran molecule

For paracellular tracer flux assay, MDCK cells expressing

either EGF alone or the CLP24-EGF fusion protein were

seeded onto a cell culture 8 lm pore size insert (no 3097;

Becton Dickinson), at a density of 105 cellsper filter,

and cultured for 5–6 days[19] To measure paracellular

flux, Dulbecco’smodified Eagle’smedium, containing

0.5 mgÆmL)1of 3 kDa FITC-labeled dextran (Molecular

Probes), was added to the apical compartment and aliquots

from the basal compartment were collected over an 8 h

period (37C) The amount of FITC-dextran that diffused

from the apical to the basolateral side of the cellular layer

was measured by using a fluorimeter (Fluoroscan Ascent

FL; Thermolabsystem) The permeability coefficient was

calculated using the following formula:

DF=Dt where DF/Dt isthe rate of the increase of the fluorescence

signal in the basal chamber, P is the permeability (cmÆmin)1)

coefficient, and S is the surface of the insert

Results

Hypoxic regulation ofCLP24

The aim of thisstudy wasto identify novel genesregulated

by hypoxia Macroarray expression profiling studies were

performed using RNA isolated from hypoxic (3% O, 16 h)

and normoxic HMEC-1 in order to identify differentially regulated cDNA transcripts within the HMEC-EHT1 cDNA library One of the differentially expressed tran-scripts identified in this study was found to be homologous

to an expressed sequence tag (EST) sequence that represents

a novel uncharacterized gene (GenBank accession number: NM_024600) Thistranscript wasupregulated in HMEC-1 cellstreated with hypoxia An induction of 2.93-fold (± 0.85) wasobserved in hypoxic (3% O2, 16 h)

HMEC-1 cellscompared to normoxic HMEC-HMEC-1 cells Further confirmation of hypoxic regulation wasobtained through independent experimentsusing the chemical agent desfer-rioxamine (100 lMfor 16 h) to induce a hypoxic response in HMEC-1 cultures This treatment resulted in a 3.16-fold (± 0.30) induction of CLP24 in hypoxic cellscompared to sham-treated HMEC-1 cells

Independent validation of the macroarray data for CLP24wasobtained using real-time quantitative PCR on total RNA from normoxic and hypoxic HMEC-1 and Calu-6 (lung carcinoma) cell lines Gaseous hypoxia treat-ment induced a strong upregulation of CLP24 expression in the endothelial cell line, HMEC-1, and the epithelial cell line, Calu-6, with an induction factor of 5.3- and 10.4-fold, respectively, thus confirming the hypoxic induction of the CLP24transcript in two different cell types

Bioinformatic characterization of CLP24 RT-PCR wasused to clone the full-length open reading frame of human CLP24, using cDNA from the HMEC-1 and Calu-6 human cell lines Cloning and sequencing of independent clonesfrom both these cell linesrevealed the insertion of a cytidine residue within the open reading frame

of CLP24 (position 1102) compared to the sequence deposited in GenBank (NM_024600) This extra base was found in every amplified cDNA from the two cell linesand bioinformatic analysis confirmed the presence of this extra cytidine residue in both the human genomic sequence (accession numbers AC046159 and AC096995) and in all homologous Homo sapiens ESTsanalyzed The insertion of this cytidine residue results in a frameshift in the protein coding sequence and a divergent C-terminal sequence compared with NM_024600 The correct cDNA and amino acid sequence of CLP24 is shown in Fig 1

The full-length human CLP24 cDNA is1871 nucleotides

in length, containing a 226-amino acid open reading frame (621–1301) with a calculated molecular mass of 24.54 kDa and a predicted pI of 8.10 An in-frame stop codon is present 21 nucleotides upstream of the methionine start codon, and a kozak consensus sequence is also present (Fig 1) The us e of membrane protein prediction s oftware (TMHMM) [17] predicted the CLP24 protein to be composed

of four alpha-helical transmembrane domains TheTMHMM analyses also revealed that the first extracellular loop of CLP24 islonger than the second and that CLP24 contains only a very short N-terminal cytoplasmic tail, but a longer C-terminal tail (Fig 1) Furthermore, motif-scanning ana-lysis (scansite.mit.edu [20]), revealed a PMP22/EMP1/ claudin homologousdomain within the CLP24 sequence (Fig 1B) and a potential class 1 PDZ protein–protein interaction motif in the last 10 amino acids of the CLP24 protein [20] A glycosylation site within the second

extracellular loop of CLP24 isalso predicted An amino acid comparison between PMP22 and claudin family members showed only a weak homology with CLP24; however, this low level of sequence homology is characteristic of a number

of PMP22/EMP1/claudin family members, including

VAB-9 and MP20 [21,22] Even though the amino acid sequence

of CLP24 isonly distantly related to that of PMP22/EMP1/ claudin, the predicted structure of the CLP24 protein shows

a number of similaritiesto thistight junction protein family that support the notion that CLP24 is a cell junction-associated protein The PMP22/EMP1/claudin family all have four transmembrane segments, most family members contain a C-terminal PDZ domain-interacting motif and the first extracellular loop is larger than the second and is believed to bridge the extracellular space (Fig 2) [17,23,24] These structural motifs are all present within CLP24 and, together with the observation that CLP24 is expressed in both endothelial and epithelial cell lines, provides further support for the suggestion that CLP24 is a novel member of the PMP22/EMP1/claudin four transmembrane junctional protein family; thusit hasbeen designated CLP24 for claudin-like protein of 24 kDa

Comparison of rattus, murine, gallus, porcine and bovine ESTsrevealsa high level of conservation of CLP24 nucleic acid sequence between species The translated protein sequences were aligned usingCLUSTALWsoftware (Fig 3) The rat, mouse and chicken sequences were found to be 89%, 87% and 81% identical to that of human, respectively CLP24 tissue-specific expression

The tissue distribution of CLP24 was characterized by Northern blot analyses (Fig 4) and showed the presence of

a 1.9kb transcript in lung, heart, kidney and placenta

aaaaaacaaccatttcctctctgctgagagccagggaaggcgagctctgc

gcacacgggcgtccctgcagcagccactctgctttccaggaccggccaac

tgccctggaggcatccacacaggggcccaggcagcacagaggagctgtga

acccgctccacaccggccaccctgcccggagcctggcactcacagcaggc

cggtgctaaggagtgtggcgcgggctcgactcccactgctgccggcctcc

cgagtgactctgttttccactgctgcaggcgagaagaggcacgcgcggca

caggccggcctccgcttcccgggaagacggcgcactcctggccctgggtt

cttgctgctgcccaccctctgctccctgggatgggccccgaggcgagcag

cttcagcacaggcctggccctgctccaggtgcaggaaggaggataaggcc

gggccgagaggcggcacacctggaccatcccatgggcctccgcccgcgcc

gccccgaggatgagtggtgatgtcctctagccacccctagcagcgtcggc

tctccctggacgtgcggccgcggactgggacttggctttctccggataag

cggcggcaccggcgtcagcgATGACCGTGCAGAGACTCGTGGCCGCGGCC

GTGCTGGTGGCCCTGGTCTCACTCATCCTCAACAACGTGGCGGCCTTCAC

CTCCAACTGGGTGTGCCAGACGCTGGAGGATGGGCGCAGGCGCAGCGTGG

S N W V C Q T L E D G R R R S V G

GGCTGTGGAGGTCCTGCTGGCTGGTGGACAGGACCCGGGGAGGGCCGAGC

L W R S C W L V D R T R G G P S

CCTGGGGCCAGAGCCGGCCAGGTGGACGCACATGACTGTGAGGCGCTGGG

P G A R A G Q V D A H D C E A L G

CTGGGGCTCCGAGGCAGCCGGCTTCCAGGAGTCCCGAGGCACCGTCAAAC

W G S E A A G F Q E S R G T V K L

TGCAGTTCGACATGATGCGCGCCTGCAACCTGGTGGCCACGGCCGCGCTC

Q F D M M R A C N L V A T A A L

ACCGCAGGCCAGCTCACCTTCCTCCTGGGGCTGGTGGGCCTGCCCCTGCT

GTCACCCGACGCCCCGTGCTGGGAGGAGGCCATGGCCGCTGCATTCCAAC

S P D A P C W E E A M A A A F Q L

TGGCGAGTTTTGTCCTGGTCATCGGGCTCGTGACTTTCTACAGAATTGGC

A S F V L V I G L V T F Y R I G

CCATACACCAACCTGTCCTGGTCCTGCTACCTGAACATTGGCGCCTGCCT

TCTGGCCACGCTGGCGGCAGCCATGCTCATCTGGAACATTCTCCACAAGA

GGGAGGACTGCATGGCCCCCCGGGTGATTGTCATCAGCCGCTCCCTGACA

E D C M A P R V I V I S R S L T

GCGCGCTTTCGCCGTGGGCTGGACAATGACTACGTGGAGTCACCATGCTG

Agtcgcccttctcagcgttccatcgatgcacacctgctatcgtggaacag

cctagaaaccaagggactccaccaccaagtcacttcccctgctcgtgcag

aggcacgggatgagtctgggtgacctctgcgccatgcgtgcgagacacgt

gtgcgtttactgttatgtcggtcatatgtctgtacgtgtcgtgggccaac

ctcgttctgcctccagctttcctggttagcgcaacgcggctccacgacca

cacgcacttcagggtggaagctggaagctgagacacaggttaggtggcgc

gaggctgccctgcgctccgctttgctttgggattaatttattctgcatct

gctgagaggggcaccccagccatatcttacactttggtaaagcagaaaac

caggaaaattttcttaaaatatccacaatattccttgagtgagtcagaat

ctatagccggttagtgatggtttcaggcagaatcgtgttcgtgtctgttt

tgctcgattcctttcctaagttaaataaatgcaagcctctgaacttctgt

ctataaaaaaaaaaaaaaaaa

1

51

101

151

201

251

301

351

401

451

501

551

601

1

651

11

701

28

751

45

801

61

851

78

901

95

951

111

1001

128

1051

145

1101

161

1151

178

1201

195

1251

211

1301

1351

1401

1451

1501

1551

1601

1651

1701

1751

1801

1851

50 100 150 200 250 300 350 400 450 500 550 600 650

10

700

27

750

44

800

60

850

77

900

94

950

110

1000

127

1050

144

1100

160

1150

177

1200

194

1250

210

1300

226

1350 1400 1450 1500 1550 1600 1650 1700 1750 1800 1850 1871

PMP22 1 MLLLLLSIIVLHVAVLVL.LF VSTI-VSQWIVGNG -HATDLWQNC 42

ml+lLl iivlh+a L LF Vsti qW v +LW +C

Consensus mlvlLlgiivlhiawviL.Lf.VsTiPtdqWkvsdyvgdniiTaaaasaGLWrnC

m v l + +a v L L V t W s GLWr+C

CLP24 1 MTVQRLVAAAVLVALVSLILnnVAAF-TSNWVCQTLED -GRRRSVGLWRSQ 49

B

A

Fig 1 Nucleotide sequence and deduced amino acid sequences of

human claudin-like protein of 24 kDa (CLP24) (A) Uppercase letters

represent the coding sequence, lowercase letters represent 5¢- and

3¢-untranslated regions (UTR) The CLP24 sequence has an

inser-tion of a C at posiinser-tion 1102 compared with NM_024600 (bold) This

insertion within the protein coding sequence results in a divergent

amino acid sequence as compared with NM_024600 The potential

N-glycosylation site is marked with a circle The protein

trans-membrane domainsare underlined The protein–protein interaction

domain-interacting motif isboxed The italic region in the 3¢-UTR

isa consensuspolyadenylation sequence *, Stop codon (B)

Align-ment of the conserved motifs of the CLP24 and myelin protein 22

(PMP22) amino acid sequence against the PMP22/epithelial

mem-brane protein 1 (EMP1)/claudin family consensus sequence using

motif-scanning analysis (http://scansite.mit.edu) Uppercase letters

show conserved amino acids between PMP22/EMP1/claudin family

members.

N

C

Extracellular loop Claudins 10-21 aa CLP21 6 aa

Extracellular loop Claudins 41-55 aa CLP24 68 aa

Intracellular loop Claudins 21-42 aa CLP24 35 aa PDZ domain

Fig 2 Membrane folding model of claudin-like protein of 24 kDa (CLP24) and myelin protein 22 (PMP22)/epithelial membrane protein 1 (EMP1)/claudin family members Schematic representation of the tetraspan structure of the CLP24 protein compared with PMP22/ EMP1/claudin family members Both CLP24 and PMP22/claudin members show four transmembrane domains, a C-terminal protein– protein interaction domain (PDZ), and a characteristic extracellular loop structure Approximate sizes (number of amino acids) of the extracellular and intracellular domainsare given for the claudin/ PMP22 family and CLP24.

Transcripts were also detected in thymus, spleen and liver,

but at lower levels A similar tissue distribution was

observed by RT-PCR using a panel of cDNAs from human

tissues (multiple tissue cDNA panel; Clonetech) This

analysis showed CLP24 to be expressed in testis, spleen

and ovary, but not in prostate (data not shown)

CLP24 is expressed in a number of different tissues that

contain epithelial cell tight junctions, including lung and

kidney, together with tissue containing high levels of blood

vessel endothelial cells, including the placenta and lung

Further confirmation of endothelium- and

epithelium-specific expression was sought through expression profiling

of human cell lines As described above, CLP24 is expressed

within the HMEC-1 and primary endothelial cells

(HUVEC) RT-PCR expression analysis showed the

CLP24gene to be expressed in a number of the human

epithelial cell lines, including the Calu-6 lung carcinoma cell

line, the RCC4 renal carcinoma cell line and the NTERA-2

(NT2) neuronal precursor epithelial cell line CLP24 is not, however, expressed ubiquitously in all epithelium-contain-ing tissues and cell lines No CLP24 expression was detected

in epithelial cell lines, including the nonsmall cell carcinoma cell line, H1299, the hepatocellular carcinoma, HepG2, or the breast carcinoma cell lines, 231,

MDA-MB-435, MCF7, BT-549, or T-47D (data not shown) Thes e data therefore demonstrate that CLP24 is expressed within specific epithelial and endothelial cell populations, which is consistent with that expected for a cell junction-associated protein

Localization of exogenously expressed CLP24

To further assess whether CLP24 is a member of the PMP22/EMP1/claudin family, the subcellular localization

of CLP24 was determined using recombinant CLP24 fused

to a green fluorescent protein (CLP24-EGFP) Transfection and translation of this plasmid construct was used to monitor the subcellular location of CLP24 in a number of cell lines, including MDCK and HMEC-1 Analysis of stably transfected clones demonstrated that the CLP24-EGFP fluorescence signal was localized to intercellular junctions(Fig 5A), in a similar manner to b-catenin and ZO-1

To confirm the association of CLP24 with tight junction components, co-localization experiments, using the tight junction ZO-1 or adherensjunction b-catenin markers, were performed MDCK cells were stably transfected with a CLP24-EGFP fusion construct and immunocytochemistry wasperformed using either anti-ZO-1 or anti-b-catenin Ig Figure 5A shows good co-localization of CLP24-EGFP with ZO-1 and b-catenin to the intercellular junctions b-catenin and ZO-1 are, however, associated with different compartmentsof the intercellular junction, namely the adherens and tight junctions, respectively The tight junction isfound at the apical face, whereasthe adherensjunction is more basal The view shown in Fig 5A looks vertically through the intercellular junction and istherefore unable to distinguish tight from adherens junction components because the tight junction is directly above the adherens junction More precise localization studies were therefore

H sapiens 1 MTVQRLVAAAVLVALVSLILNNVAAFTSNWVCQTLEDGRRRSVGLWRSCWLVDRTRGGPSPGARAGQVDAHDCEALGWGSEAAGFQESRGTVKLQFDMMR

R norvegicus 1 MTVQKLVATAVLVALVSLILNNAAAFTPNWVYQTLEDGRKRSVGLWKSCWLVDRGKGGTSPGTRTGQVDTHDCEVLGWGSESAGFQESRGTVKLQFDMMR

M musculus 1 MTLQKLVATAVLVALVSLILNNAAAFTPNWVYQTLEDGRKRSVGLWKSCWLVDRGKGVTSPGTRTGQVDTHDCEVLGWGSESAGFQESRGTVKLQFDMMR

G gallus 1 MTVQKLVATAVLVALVSLILNNAAAFTPNWVYQTLEDGRKRSVGLWKMCWLAERSRAGASTSSRHGQGEERECEALGWGSESAGFQESRSTVKLQFDMMR

S scrofa 1 MTVXKVVATAVLVALVSLVLNNVAALTPNWVYQTLEDGRRRSVGLWRSCWLLDRXXXXXXXXXXXXXXXXXXXXXXXXXXXXXGFQESRGTVKLQFDMMR

B taurus 1 -DVRDCEALGWGSEAAGFQESRGTVKLQFDMMR

H sapiens 101 ACNLVATAALTAGQLTFLLGLVGLPLLSPDAPCWEEAMAAAFQLASFVLVIGLVTFYRIGPYTNLSWSCYLNIGACLLATLAAAMLIWNILHKREDCMAP

R norvegicus 101 ACNLVATAALAVGQITFILGLTGLPLMSPESQCWEEAMAAAFQLASFVLVIGLVTFYRIGPYTNLSWSCYLNIGACLLATLAAAMLIWNILHRREDCMAP

M musculus 101 ACNLVATAALVVGQITFILGLTGLPLMSPESQCWEEAMAAAFQLASFVLVIGLVTFYRIGPYTNLSWSCYLNIGACLLATLAVAMLIWNILHRREDCMAP

G gallus 101 ACNLIATVALTAGQLIFVLGLVEIPIISQDTQWWEEAIAAVFQLASFVLVIGLVTFYRIGPYTNLSWSCYLNIGACLLATLAAAILIWNILHRREDCMAP

S Scrofa 101 ACNLVATAALAAGQLTFVLGLTGLPLMSPDSQCWEEAMAAAFQLASFVLVIGLVTFYRIGPYTNLSWSCYLDIGACLLATLAAAMLIWNVLHRREDCMAP

B taurus 35 ACNLVATAALAAGQLTFVLGLTGLPLLSPDAQCWEEAMAAAFQLASFVLVIGLVTFYRIGPYTSLSWSCYLNIGACLLATLAAAMLIWNVLHRREDCTAP

H sapiens 201 RVIVISRSLTARFRRGLDNDYVESPC

R norvegicus 201 RVIVISRSLTARFRRGLDNDYVESPC 88.9% identity

M musculus 201 RVIVISRSLTARFRRGLDNDYVESPC 88.0% identity

G gallus 201 RVIVISRTLTARFRRGLENDYVESPC 81.4% identity

S Scrofa 201 RVIVISRSLAARFRRGLDXXXXXXXX

B taurus 135 RVIVISRSLTARFRRGLDNDYVESPC

Fig 3 Alignment of the amino acid sequences of mammalian claudin-like protein of 24 kDa (CLP24) orthologues Alignment of the amino acid sequences of human CLP24 with rat, mouse and chicken orthologues, together with partial amino acid sequences of porcine and bovine ortho-logues The human sequence is 89%, 87% and 81% identical to the rat, mouse and chicken sequences, respectively.

2.4

1.35

spleen kidne

placenta lung

kb

Fig 4 Northern blot analysis of human claudin-like protein of 24 kDa

(CLP24) Human multiple tissue Northern blots (Clonetech) were

probed with a radiolabelled DNA fragment of CLP24 ske muscle,

skeletal muscle; small intes., small intestine; PBL, peripheral blood

leukocytes.

performed using a computer-generated cross-section (x–z

scan) through the cell junction (Fig 5B) The merged

images of the x–z scans (Fig 5B) show that the ZO-1

protein signal is located at the apical face of the MDCK

monolayer, asexpected within polarized MDCK cells, whilst the CLP24-EGFP signal is localized throughout the intercellular junctionsof the MDCK monolayer and shows co-localization with the adherensjunction protein b-catenin

To further confirm thisinitial observation, co-localization studies were performed using an anti-claudin 1 tight junction-associated antibody As observed with ZO-1, claudin 1 waslocalized at the apical surface and did not localize with EGFP-CLP24 (data not shown)

CLP24 mediates cell junction interactions Overexpression of CLP24-EGFP in MDCK cells results in the expression of CLP24 at cellular junctions; however, MDCK cells expressing CLP24 display a different mor-phology to the nontransfected cells (Fig 5C) The expres-sion of the CLP24-EGFP protein led to the disappearance

of the typical cobblestone structure that results from the adhesion of confluent MDCK cells In mixed cultures containing both CLP24-EGFP expressing cells (detectable

by fluorescence microscopy) and nonexpressing cells, the nontransfected cells organized themselves into cobblestone structures, whereas the cells expressing the CLP24–EGFP protein displayed a more fibroblastic morphology, charac-teristic of a decrease in cellular adhesion

Overexpression of CLP24 alters cellular permeability Studieswere performed to characterize the potential effect

of CLP24 on the paracellular barrier function Paracellular flux measurements were undertaken across a confluent monolayer of cells expressing CLP24-EGFP, compared to those obtained using control cell lines, and showed a markedly higher paracellular flux of a 2 kDa FITC-dextran tracer molecule than MDCK control cells(Fig 6A) The observed total permeability coefficient of MDCK/CLP24-EGFP monolayerswasthreefold higher than those of MDCK or MDCK/EGFP monolayers(Fig 6B) This demonstrates that the CLP24 protein is able to modulate junctional barrier function and shows both structural and functional properties that are consistent with CLP24 being a novel PMP22/EMP1/claudin family member

Discussion

Angiogenesis, the growth of new blood vessels out of pre-existing capillaries, is fundamental to many physiological and pathological processes, such as cancer, ischemic diseases and chronic inflammation Hypoxia isone of the physio-logical signals that promote angiogenesis During this process, adherens junctions are involved in the control of vascular permeability and in vascular remodeling [25–27] Regulation of proteinsinvolved in adherensjunctionsis essential in the detachment of endothelial cells from the vessel wall and invasion into the underlying tissues, which are, in turn, essential for new vessel formation [3] The identification of CLP24 asa claudin-related protein, which isa hypoxically regulated adherensjunction component that isable to influence vascular permeability, isan important observation that documentsroutesthrough which adherens junctionsmay be regulated during normal pathology and disease

A

CLP24-EGFP ββββ-catenin overlay

B

CLP24-EGFP ββββ-catenin overlay

C

Phase contrast CLP24-EGFP

Fig 5 Localization of claudin-like protein of 24 kDa (CLP24).

MDCK cells were stably transfected with full-length CLP24 fused to

green fluorescent protein (EGFP) The transfected clones

(CLP24-EGFP in green), still mixed with nontransfected cells, were stained by

indirect immunofluorescence for either Zona Occluden-1 (ZO-1) or

b-catenin (both in red) (A) CLP24–EGFP expres s ion is localized at

intercellular borders, as observed for the ZO-1 and b-catenin junction

proteins (B) Co-localization studies, using an x–z scan, demonstrate

that CLP24 istargeted to the intracellular junctionsand co-localizes

with b-catenin (top), but not with ZO-1 (bottom) Bar: 10 lm.

(C) Morphological differencesobs erved between co-culturesof

CLP24-EGFP, expressed and not expressed in MDCK cells

Expres-sion of CLP24-EGFP results in an altered morphology compared to

the cells not expressing CLP24 and a loss of the cobblestone structure

that results from the formation of epithelial cell junctions.

Bioinformatic characterization suggested that CLP24 is a

member of the PMP22/EMP1/claudin family, which are

tight junction associated proteins However, recent studies

have shown that VAB-9, a PMP22/EMP1/claudin family

member from Caenorhabditis elegans (nematode),

isinvol-ved in adherensjunctions[21] Therefore, co-localization

experiments were performed to more precisely assess the

location of CLP24 expression at the apical cell junction

These studies showed that CLP24 co-localizes with adherens

junction protein b-catenin, but not with the ZO-1 tight

junction protein

In agreement with the fact that endothelial cell

interac-tionshave to be reorganized during vessel formation,

over-expression of CLP24 induced morphological changes in

MDCK cells that are characteristic of a decreased adhesion

between cells A similar phenotype has also been observed

for the over-expression of Z0-1 adherens junction

trunca-tion mutants, which results in disruptrunca-tion of the cobblestone

structure to that of a fibroblastic morphology [28]

Func-tional studies showed that CLP24 is able to influence

paracellular permeability, asobserved for other PMP22/

EMP1/claudin members Taken together, these results

indicate that CLP24 isinvolved in cell–cell interactions

through adherens, rather than tight, junctions

The expression pattern of CLP24 is distinct from other

claudin family members, showing expression in lung, kidney

heart and placenta It is noticeable that CLP24 is expressed

within the heart, which isunusual for the claudin family

members This observation suggests that CLP24, like

PMP22/EMP1/claudin family members, has a

tissue-speci-fic distribution, which is linked to cell junction specitissue-speci-ficity

[23,29] Both endothelial and epithelial cells possess tight

junction and adhesion structures to seal intercellular spaces

Here we show, using RT-PCR, that CLP24 is expressed in

endothelial cellsand in only a restricted population of

epithelial cell lines These observations suggest that CLP24

hasa specific barrier function within different cell typesand

confirm the observationsof otherswhich show that complex

interplay between different claudin family membersis

required to control paracellular permeability

Work by Leach [26,30] hasshown dynamic regulation of

both adherensand tight junction componentsduring

angiogenesis This, together with the observation that

claudin family membersare deregulated during hypoxia and cancer [31–35], suggests that CLP24 could be also implicated in both normal and tumor angiogenesis Knockout experiments in mice show that loss of the adenomatouspolyposiscoli (APC) or b-catenin results in reduction of cell–cell adherensjunctionsadhesion [36] and the b-catenin–APC complex hasbeen shown to have a role

in the proliferation and migration of vascular endothelial cellsduring neovascularization Asclaudin family members interact with b-catenin, it can be therefore envisioned that CLP24 isinvolved in b-catenin signaling, and that up-regulation of CLP24 upon hypoxia might participate

to the pro-antigenic deregulation of thiscascade Hypoxic stimulation results in the induction of vascular endothelial growth factor (VEGF) and endothelial hyperpermeability Both tight junction and adherensjunction moleculeshave been shown to influence paracellular permeability Differ-ential expression of claudin family members results in the different permeability propertiesobserved in different epithelial and endothelial membranes [23], whilst loss of b-catenin results in decreased cell–cell adhesion and increased paracellular permeability [25] The identification

of CLP24 asa novel adherensjunction component that is induced by hypoxia and isable to reduce adhesion, thereby increasing intracellular permeability, provides additional insight into the understanding of how hypoxic stimuli induce morphological changesin endothelial cellsrequired for an angiogenic response A valuable additional insight into the functional role of CLP24, and a clearer under-standing of its therapeutic potential, will be provided by further characterization of CLP24 in pathologiesincluding cardiovascular disease, neurological disorders and tumori-genesis

Acknowledgements

We are grateful for the support of Prof P Corvol at the College de France (INSERM) for his support with laser confocal microscopy.

References

1 Nagafuchi, A (2001) Molecular architecture of adherensjunc-tions Curr Opin Cell Biol 13, 600–603.

100 200

300

0 0.5 1.0 1.5 2.0

*

CLP24-EGFP EGFP

CLP24-EGFP EGFP

Fig 6 Stable expression of claudin-like

protein of 24 kDa (CLP24) in MDCK cells

increases the paracellular flux of 3 kDa

FITC-dextran (A) The paracellular flux of 3 kDa

FITC-dextran, across a monolayer of MDCK

cells, was monitored over time The results

show that over-expression of CLP24 results in

an increase in paracellular permeability

com-pared to controls(mean ± standard error of

three independent experiments) (B)

Per-meability coefficientswere calculated from the

data in Fig 6A and demonstrate a three-fold

increase in permeability in MDCK

mono-layers expressing CLP24 compared to controls.

2 Gonzalez-Mariscal, L., Betanzos, A., Nava, P & Jaramillo, B.E.

(2003) Tight junction proteins Prog Biophys Mol Biol 81, 1–44.

3 Dejana, E (1996) Endothelial adherensjunctions: implicationsin

the control of vascular permeability and angiogenesis J Clin.

Invest 98, 1949–1953.

4 Park, J.H., Okayama, N., Gute, D., Krsmanovic, A., Battarbee,

H & Alexander, J.S (1999) Hypoxia/aglycemia increases

endo-thelial permeability: role of second messengers and cytoskeleton.

Am J Physiol 277, C1066–C1074.

5 Zahraoui, A., Louvard, D & Galli, T (2000) Tight junction, a

platform for trafficking and signaling protein complexes J Cell

Biol 151, F31–F36.

6 Notterpek, L., Roux, K.J., Amici, S.A., Yazdanpour, A., Rahner,

C & Fletcher, B.S (2001) Peripheral myelin protein 22 isa

con-stituent of intercellular junctions in epithelia Proc Natl Acad Sci.

USA 98, 14404–14409.

7 Evans, W.H & Martin, P.E (2002) Gap junctions: structure and

function [Review] Mol Membr Biol 19, 121–136.

8 Schneider, C., King, R.M & Philipson, L (1988) Genes

specifi-cally expressed at growth arrest of mammalian cells Cell 54, 787–

793.

9 Taylor, V., Welcher, A A., Program, A E & Suter, U (1995)

Epithelial membrane protein-1, peripheral myelin protein 22, and

lensmembrane protein 20 define a novel gene family J Biol.

Chem 270, 28824–28833.

10 Welcher, A.A., Suter, U., De Leon, M., Snipes, G.J & Shooter,

E.M (1991) A myelin protein isencoded by the homologue of

a growth arrest-specific gene Proc Natl Acad Sci USA 88,

7195–7199.

11 Brancolini, C., Edomi, P., Marzinotto, S & Schneider, C (2000)

Exposure at the cell surface is required for gas3/PMP22 to regulate

both cell death and cell spreading: implication for the

Charcot-Marie-Tooth type 1A and Dejerine-Sottas diseases Mol Biol.

Cell 11, 2901–2914.

12 Skre, H (1974) Genetic and clinical aspects of

Charcot-Marie-Tooth’s disease Clin Genet 6, 98–118.

13 Simon, D.B., Lu, Y., Choate, K.A., Velazquez, H., Al-Sabban, E.,

Praga, M., Casari, G., Bettinelli, A., Colussi, G.,

Rodriguez-Soriano, J., McCredie, D., Milford, D., Sanjad, S & Lifton, R.P.

(1999) Paracellin-1, a renal tight junction protein required for

paracellular Mg 2+ resorption Science 285, 103–106.

14 Schweighoffer, F., Ait-Ikhlef, A., Resink, A.L., Brinkman, B.,

Melle-Milovanovic, D., Laurent-Puig, P., Kears ey, J & Bracco, L.

(2000) Qualitative gene profiling: a novel tool in genomicsand in

pharmacogenomics that deciphers messenger RNA isoforms

diversity Pharmacogenomics 1, 187–197.

15 Sanchez-Cespedes, M., Cairns, P., Jen, J & Sidransky, D (1998)

Degenerate oligonucleotide-primed PCR (DOP-PCR): evaluation

of its reliability for screening of genetic alterations in neoplasia.

Biotechniques 25, 1036–1038.

16 Falquet, L., Pagni, M., Bucher, P., Hulo, N., Sigrist, C.J.,

Hofmann, K & Bairoch, A (2002) The PROSITE database, its

status in 2002, Nucleic Acids Res 30, 235–238.

17 Krogh, A., Larsson, B., von Heijne, G & Sonnhammer, E.L.

(2001) Predicting transmembrane protein topology with a hidden

Markov model: application to complete genomes J Mol Biol.

305, 567–580.

18 Pfaffl, M.W (2001) A new mathematical model for relative

quantification in real-time RT-PCR Nucleic Acids Res 29, e45.

19 Lennon, P.F., Taylor, C.T., Stahl, G.L & Colgan, S.P (1998)

Neutrophil-derived 5¢-adenosine monophosphate promotes

endothelial barrier function via CD73-mediated conversion to

adenosine and endothelial A2B receptor activation J Exp Med.

188, 1433–1443.

20 Yaffe, M.B., Leparc, G.G., Lai, J., Obata, T., Volinia, S &

Cantley, L.C (2001) A motif-based profile scanning approach for

genome-wide prediction of signaling pathways Nat Biotechnol.

19, 348–353.

21 Simske, J.S., Koppen, M., Sims, P., Hodgkin, J., Yonkof, A & Hardin, J (2003) The cell junction protein, VAB-9, regulates adhesion and epidermal morphology in C elegans Nat Cell Biol.

5, 619–625.

22 Louis , C.F., Hur, K.C., Galvan, A.C., TenBroek, E.M., Jarvis , L.J., Eccleston, E.D & Howard, J.B (1989) Identification of an 18,000-dalton protein in mammalian lensfiber cell membranes.

J Biol Chem 264, 19967–19973.

23 Heiskala, M., Peterson, P.A & Yang, Y (2001) The roles of claudin superfamily proteins in paracellular transport Traffic 2, 93–98.

24 Itoh, M., Furuse, M., Morita, K., Kubota, K., Saitou, M & Tsukita, S (1999) Direct binding of three tight junction-associated MAGUKs, ZO-1, ZO-2, and ZO-3, with the COOH termini of claudins J Cell Biol 147, 1351–1363.

25 Cattelino, A., Liebner, S., Gallini, R., Zanetti, A., Balconi, G., Cors i, A., Bianco, P., Wolburg, H., Moore, R., Oreda, B., Kemler,

R & Dejana, E (2003) The conditional inactivation of the beta-catenin gene in endothelial cellscausesa defective vascular pattern and increased vascular fragility J Cell Biol 162, 1111–1122.

26 Leach, L., Lammiman, M.J., Babawale, M.O., Hobson, S.A., Bromilou, B., Lovat, S & Simmonds, M.J (2000) Molecular organization of tight and adherensjunctionsin the human placental vascular tree Placenta 21, 547–557.

27 Sund, M., Ylonen, R., Tuomisto, A., Sormunen, R., Tahkola, J., Kvist, A.P., Kontusaari, S., Autio-Harmainen, H & Pihlajaniemi,

T (2001) Abnormal adherence junctionsin the heart and reduced angiogenesis in transgenic mice overexpressing mutant type XIII collagen EMBO J 20, 5153–5164.

28 Ryeom, S.W., Paul, D & Goodenough, D.A (2000) Truncation mutantsof the tight junction protein ZO-1 disrupt corneal epi-thelial cell morphology Mol Biol Cell 11, 1687–1696.

29 Furuse, M., Sasaki, H & Tsukita, S (1999) Manner of interaction

of heterogeneousclaudin specieswithin and between tight junction strands J Cell Biol 147, 891–903.

30 Leach, L., Babawale, M.O., Anderson, M & Lammiman, M (2002) Vasculogenesis, angiogenesis and the molecular organisa-tion of endothelial juncorganisa-tionsin the early human placenta J Vasc Res 39, 246–259.

31 Brown, R.C., Mark, K.S., Egleton, R.D., Huber, J.D., Burroughs, A.R & Davis, T.P (2003) Protection against hypoxia-induced increase in blood–brain barrier permeability: role of tight junction proteinsand NFkappaB J Cell Sci 116, 693–700.

32 Mark, K.S & Davis, T.P (2002) Cerebral microvascular changes

in permeability and tight junctionsinduced by hypoxia-reoxy-genation Am J Physiol Heart Circ Physiol 282, H1485–H1494.

33 Witt, K.A., Mark, K.S., Hom, S & Davis , T.P (2003) Effects of hypoxia-reoxygenation on rat blood–brain barrier permeability and tight junctional protein expression Am J Physiol Heart Circ Physiol 285, H2820–H2831.

34 Kominsky, S.L., Argani, P., Korz, D., Evron, E., Raman, V., Garrett, E., Rein, A., Sauter, G., Kallioniemi, O.P & Sukumar, S (2003) Lossof the tight junction protein claudin-7 correlateswith histological grade in both ductal carcinoma in situ and invasive ductal carcinoma of the breast Oncogene 22, 2021–2033.

35 Michl, P., Barth, C., Buchholz, M., Lerch, M.M., Rolke, M., Holzmann, K.H., Menke, A., Fensterer, H., Giehl, K., Lohr, M., Leder, G., Iwamura, T., Adler, G & Gress, T.M (2003) Claudin-4 expression decreases invasiveness and metastatic potential of pancreatic cancer Cancer Res 63, 6265–6271.

36 Carothers, A.M., Melstrom, K.A Jr, Mueller, J.D., Weyant, M.J.

& Bertagnolli, M.M (2001) Progressive changes in adherens junction structure during intestinal adenoma formation in Apc mutant mice J Biol Chem 276, 39094–39102.