Department of Physiology, Biophysics and Neuroscience, Center for Research and Advanced Studies (CINVESTAV), Mexico City, MexicoDepartment of Experimental Pathology, Center for Research and Advanced Studies (CINVESTAV), Mexico City, Mexico
A R T I C L E I N F O R M A T I O N
⁎ Corresponding author. Department of P(CINVESTAV), Ave. IPN 2508, Mexico D.F., 073
Article Chronology:Received 5 September 2006Revised version received19 January 2007Accepted 22 January 2007Available online 15 February 2007
ZO-2 is a tight junction (TJ) protein that shuttles between the plasma membrane and thenucleus. ZO-2 contains several protein binding sites that allow it to function as a scaffoldthat clusters integral, adaptor and signaling proteins. To gain insight into the role of ZO-2 inepithelial cells, ZO-2 was silenced in MDCK cells with small interference RNA (siRNA). ZO-2silencing triggered: (A) changes in the gate function of the TJ, determined by an increase indextran flow through the paracellular route of mature monolayers and achievement oflower transepithelial electrical resistance values upon TJ de novo formation; (B) changes inthe fence function of the TJ manifested by a non-polarized distribution of E-cadherin on theplasma membrane; (C) altered expression of TJ and adherens junction proteins, determinedby a decreased amount of occludin and E-cadherin in mature monolayers and a delayedarrival to the plasma membrane of ZO-1, occludin and E-cadherin during a calcium switchassay; and (D) an atypical monolayer architecture characterized by the appearance ofwidened intercellular spaces, multistratification of regions in the culture and an alteredpattern of actin at the cellular borders.
In multicellular organisms the external environment isseparated from the internal milieu by epithelia that coverthe body and the different internal compartments. In order forepithelia to function as a barrier and to display a vectorialtransport, two fundamental characteristics have to be ful-filled: first epithelial cells need to establish tight junctions(TJs) that function as gates that regulate the passage of ionsand molecules through the paracellular pathway [1], andsecond they have to exhibit a polarized phenotype. The latter
implies that the apical and basolateral plasma membranesurface display distinct composition and morphology. In themaintenance of epithelial polarity, TJs play a crucial role sincethey function as a fence that blocks the free diffusion withinthe membrane of lipids and proteins between the apical andbasolateral surfaces and vice versa [2].
TJs have a complex molecular composition [3]. They areconstituted by three types of integral proteins responsible forestablishing cell–cell contacts in the paracellular space: thetetraspan claudins and occludin and several members of theJAM family characterized for presenting two immunoglobulin
d Neuroscience, Center for Research and Advanced Studies0 61 37 54.
1534 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
repeats in their extracellular portion. These integral proteinsof the TJ associate through their cytoplasmic domains toseveral adaptor proteins that bind to the actin cytoskeletonand to a diverse group of signaling proteins that includekinases, phosphatases and transcription factors. Numerous TJadaptor proteins have as a common feature the presence ofPDZ domains. Among these molecules some share otherregions in common. Such is the case of the MAGUK proteinsthat display PDZ, SH3 and GuK domains [4].
ZO-2 is a MAGUK protein that was first identified as a ZO-1 binding protein by coimmunoprecipitation [5]. This 160-kDa protein is present at the TJ of epithelial and endothelialcells and at the adherens junction (AJ) of nonepithelial cellssuch as fibroblasts and cardiac muscle cells [6]. ZO-2 is ascaffold protein since it is capable of associating diverseproteins at the particular region where it is located. Thus, atthe TJ ZO-2 associates with claudins through its first PDZdomain [7], to ZO-1 by the second PDZ regions of both ZOproteins [8], and to cingulin [9], a TJ adaptor protein thatlacks PDZ domains and instead contains globular and coiled-coil domains. ZO-2 associates to actin [8] and the actinbinding protein 4.1 [10] through the proline-rich domainpresent in its carboxyl terminal end [6]. Occludin as well asthe AJ adaptor protein α-catenin, binds directly to the aminoterminal portion of ZO-2 [6]. Although the precise site ofinteraction of ZO-2 with occludin has not been explored [8],it is presumed to take place through the GuK domain of themolecule, as appears to be the case for ZO-1 [11].
The localization of ZO-2 within the cell is sensitive to thedegree of cell–cell contact. Hence in confluent monolayers ZO-2 is observed at TJs while in sparse cultures it concentrates atthe nuclei where it displays a speckled distribution and co-localizes with splicing factor SC-35 and SAF-B [12,13], achromatin component [14] that in breast cancer cells func-tions as estrogen receptor corepressor and growth inhibitor[15]. At the nucleus ZO-2 is present in the nuclear matrix andassociates to lamina B1 and actin [16].
Shuttling of ZO-2 between the plasma membrane and thenucleus is possible due to the presence of several nuclearlocalization and exportation signals in its sequence [12,16,17].
Fig. 1 – siRNA induced ZO-2 silencing in MDCK cells. (A) siRNA o3525 nt mRNA of canine ZO-2. The arrows indicate the sites of tharrowheads the sites of the start and stop codons. The sequence ois shown. (B) Quantitative detection of ZO-2mRNA inMDCK cells.or 10 nM of negative control (NC) siRNA, the monolayers were lyQuantiGene® System. The amount of luminescence reported inmRNA expressed. For normalization, we gave a value of 1 to thenon-transfected monolayers. The media±standard error was obStudent's T test. (C)Western blot detection of ZO-2 3 days after thZO-2 siRNA plus full-length ZO-2. The upper image shows a repdensitometric analysis obtained from three independent experimdensitometric measures obtained in each experimental plate forstandard error is shown. *P<0.05; **P<0.005 in a Student's T test.days post-transfection. Actin was used as a loading control for thcontrol cells and in monolayers treated with ZO-2 siRNA, NC siRsiRNA and full-length canine ZO-2. (F) Determination of cell viabnon-transfected,mock transfected or transfectedwith 10 nMof NClater, cells were harvested and viable cells were counted by Tryp
The dual localization of ZO-2 suggests the possible participa-tion of this protein in diverse functions. This idea has beenreinforced by the observation that ZO-2 associates to tran-scription factors Jun, Fos and C/EBP and is capable of mod-ulating gene transcription in reporter gene assay that employpromoters under the control of AP-1 sites [18]. Furthermore,ZO-2 appears to be a candidate tumor suppressor protein dueto its high homology to Dlg [19], a well-known tumor sup-pressor [20], its absence or decreased expression in themajority of breast cancers [21] and the observation thathuman pancreatic cancers lack the longer ZO-2 isoform Athat contains 23 additional amino acids at its amino terminalportion [22]. Further insight into the perception of ZO-2 as atumor suppressor has been gained by the observation that ZO-2 over expression can inhibit in vitro transforming activities ofdifferent oncoproteins, including activated Ras, polyomavirusmiddle T protein and adenovirus type 9 oncogenic determi-nant E4 [23].
To gain further insight into the role of ZO-2, here we havechosen to silence ZO-2 gene expression in epithelial cellsutilizing the recently developed strategy of small interferenceRNAs (siRNAs). We analyze the impact of ZO-2 knockdown onthe gate and fence function of theTJ, on the expressionof otherTJ proteins and on epithelial proliferation and morphology.
Material and methods
Cell culture
Epithelial MDCK cells clone A10 (Madin Darby kidney cells)between the 60th and 90th passage were grown as previouslydescribed [24]. Cells were harvested with trypsin–EDTA andplated either sparse (1×105 cells/cm2) or at confluency (3×105
cells/cm2). The synchronized cultures in the sparse conditioncells were plated at an initial density of 0.5×105 cells/cm2.
In the case of co-cultures of control and ZO-2 siRNA-transfected cells, the latterwere labeled for 1 hwith 6 μMof thefluorescent marker Cell Tracker™ Orange CMTMR (Molecular
ligonucleotide sites in cZO-2. The top line represents thee sequences chosen for the siRNA constructs and thef the complimentary oligonucleotide pairs of siRNA employedThree days after transfectionwith 2, 5 or 10 nM of ZO-2 siRNAsed and the amount of ZO-2 mRNA quantitated with therelative light units (RLUs), is proportional to the level of ZO-2RLU obtained, in each experimental plate, for thetained from three independent experiments. *P<0.05 in aemonolayers were transfected with ZO-2 siRNA, NC siRNA, orresentative example, while the graph bellow illustrates theents. For normalization, we gave a value of 1 to theZO-2/actin in the non-transfected monolayers. The media±(D) ZO-2 silencing with 10 nM of siRNA 1 persists for at least 7e Western blot. (E) Immunofluorescence detection of ZO-2 inNA, mock transfected, and cultures co-transfected with ZO-2ility after siRNA transfection. MDCK cells were leftsiRNA or 2 and 10 nMof siRNAnumber 1. Seventy-two hoursan blue exclusion.
1535E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
Probes, Eugene, Ore., C-2927). Cells were then washed threetimes with PBS, re-incubated for 1 h in CDMEM, treated withtrypsin, and the suspension mixed with control cells. Theresulting mixed cell suspension was then plated on glasscoverslips, and observed 24 h later.
Expression plasmids and transfection assays
Full-length canine ZO-2 introduced into the CMV expressionplasmidpGW1 (pGW1-HA-ZO-2)waskindlyprovidedbyRonaldJavier (Baylor College of Medicine, Houston, TX). Transfections
1536 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
of HA-ZO-2-pGW1were done with Lipofectamine 2000 (Cat. No11668-019; Invitrogen, Life Technologies) as indicated by themanufacturer.
siRNA design and transfection
In order to obtain an effective and highly specific knockdownhere we employed Stealth™ RNAi from Invitrogen LifeTechnologies (Carlsbad, CA, USA). The latter are chemicallymodified dsRNAs that result in greater longevity in cellculture, minimize the induction of non-specific cellular stressresponse pathways and avoid unwanted off-target effects,since only the antisense strand can participate in RNAi. Thedesign of the two Stealth™ RNAis herein employed was donewith the Invitrogen RNAi Designer (https://rnaidesigner.invi-trogen.com/sirna/) that uses a proprietary algorithm to elimi-nate target sequences that have homology to other genes,follows published rules on RNAi design and a consensus fromfunctionally tested successful knockdown sequences. Fig. 1Ashows the sequence and the localization along the canine ZO-2(cZO-2) sequence of the Stealth™ RNAi herein employed.
Stealth™ RNAi was delivered to MDCK cells by transfectionwith Lipofectamine™ 2000 transfection reagent (InvitrogenLife Technologies) following the manufacturer's instructions.A negative control (NC) siRNA from Invitrogen, not homo-logous to anything in vertebrate transcriptosome wasemployed. In the experiments shown in Fig. 9, sparse cultures
Fig. 2 – ZO-2 down regulation by siRNA delays cell aggregation. Tcells were trypsinized and suspended as hanging drops for 3, 6 ashows a representative microscope field for each experimental cresults, obtained in three independent experiments, counting thedifferent fields per experiment. *P<0.05; **P<0.005.
were transfected twice with siRNA: once on the second dayafter plating and again on the third day. These cultures wereanalyzed on the fourth day.
Quantitation of mRNA
For mRNA quantitation the QuantiGene Explore Kit (Cat. No.QG-000-002; Genospectra, Fremont, CA. USA)) was employedaccording to the manufacturer's instructions. This assay usesbranched DNA technology. In it the mRNA that is releasedfrom lysed cells is captured to a microwell by syntheticoligonucleotide probes called capture extenders specificallydesigned against cZO-2 mRNA. A second set of probes calledlabel extenders hybridizes to different sequences on the cZO-2mRNAs and to sequences on the amplifier probe. The latterhas fifteen branches, and on each branch are three copies of asequence that is complementary to the label probe. Alkalinephosphatase is covalently attached to each of the label probes.The labeled amplifier complexes are detected by the alkalinephosphatase mediated degradation of the chemilumigenicsubstrate dioxetane. Luminescence is reported as relative lightunits (RLUs) on a microplate luminometer and the amount ofchemiluminescence is directly proportional to the quantity ofinput RNA. This assay eliminates the biases introduced byreverse transcription and PCR amplification, and has beensuccessfully employed by researchers interested in siRNAtechnology [25,26].
hree days after transfection with 10 nM siRNA number 1, thend 9 h, from the lid of a 24-well culture dish. The right panelondition, while the left panel illustrates the quantitationnumber of isolated cells and of those forming aggregates, in 3
Fig. 3 – ZO-2 silencing affects the gate function of TJs. (A) Twodays after plating on Transwell-Clear inserts, the TER of themonolayers was first determined. One day later, afterobtaining a similar TER value, the monolayers were mocktransfected, transfected with ZO-2 siRNA number 1 or NCsiRNA, or left untreated. TER was then measured for thefollowing 5 days. (B) In the third day post-transfection themonolayers cultured on Transwell filters were incubated inlow calcium media for 20 h (time 0). Then the monolayerswere transferred to calcium containing media and TER wasmeasured at different time points. The data of fivemonolayers are reported as the mean±SE for eachexperimental condition. *P<0.05 in a Student's T test. (C) Theparacellular flux of 70 kDa FITC–dextran, measured in theapical to basolateral direction, was determined on the fifthday post-transfection.
1537E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
Protein blotting
MDCK cells were lysed under gentle rotation for 15 min at4 °C with RIPA buffer (40 mM Tris–HCl, pH 7.6, 150 mM NaCl,2 mM EDTA pH 8.0, 10% glycerol, 1% Triton X-100, 0.5%sodium deoxycholate, 0.2% SDS, 1 mM PMSF) containing theprotease inhibitor cocktail Complete™. The lysate was thensonicated three times for 30 s each in a high intensityultrasonic processor (Vibra cell, Sonics and Materials Inc.;Danbury, CT).
The proteins in the cellular extracts were quantified andthe samples were diluted (1:1) in sample buffer (125 mM Tris–HCl, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, pH 6.8), runin 8, 10 or 12% polyacrylamide gels and transferred to PVDFmembranes (Hybond RPN303F; AmershamPharmacia Biotech,Little Chalfont, Buckinghamshire, UK).
Blotting was performed with polyclonal antibodies againstZO-2 (Cat. No. 71-1400, dilution 1:1000; Zymed Laboratories,San Francisco, CA, USA), ZO-1 (Cat. No. 61-7300, dilution 1:500;Zymed Laboratories) or with a monoclonal against actin,generated and generously provided by Dr. Jose ManuelHernandez from the Department of Cell Biology at CINVES-TAV, México (dilution 1:50). Peroxidase-conjugated goat IgGagainst rabbit IgG or against mouse IgG (Cat. Nos. 62-6120 and62-6520, respectively; dilution 1:3000; Zymed Laboratories)were used as secondary antibodies, followed by a chemilumi-nescence detection system (ECL+Plus, RPN 2132; AmershamPharmacia Biotech).
Immunofluorescence
The immunofluorescence experiments were done followingstandard procedures previously described [16]. The first anti-bodies employed were rabbit polyclonals against ZO-2 (Zymed71-1400, dilution 1:100), mouse monoclonals for occludin(Zymed 33-1500, dilution 1:100, dilution 1:100) and GP135 (agenerous gift of Dr. George Ojakian, NewYork State University)and rat monoclonals against ZO-1 (R26.4C, dilution 1:100,Developmental Studies Hybridoma Bank, University of Iowa,IA, USA) and E-cadherin (Sigma U-3254, dilution 1:50). Assecondary antibodieswe employed FITC-conjugated goat anti-rabbit IgG (Zymed 62-6111; dilution 1:100) and goat anti-rat IgG(Zymed 62-9511; dilution 1:100), a TRITC-conjugated goat anti-rabbit IgG (Zymed 81-6114; dilution 1:100) and a CY5-conju-gated goat anti-mouse (Zymed 81-6516; dilution 1:100).
Actin was detected with 1 μm rhodaminated phalloidin(Molecular Probes R415). Fluorescence was examined using aconfocal microscope (Leica SP2) with argon and helium–neonlasers and employing the Leica confocal software.
Cell viability and apoptosis determination
The proportion of viable cells in the culture after siRNAtransfection was determined using the vital stain Trypan blue(Cat. No. T 6146, Sigma, St. Louis, USA), according to themanufacturer's instructions.
Apoptotic cells in the culture were visualized with DeadEnd™ Fluorometric TUNEL staining (PROMEGA G3250). Cellsgrown on coverslips were fixed with 4% paraformaldehydefor 20 min at room temperature, rinsed in PBS, and
permeabilized with 0.1% Triton X-100 in PBS for 10 min.TUNEL staining was performed according to the manufac-turer's instructions. For the detection and quantification ofapoptosis and differentiation from necrosis at single cell
1538 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
level we utilized the Annexin-V-FLUOS staining kit (Roche 1-858-777). In these experiments the percentage of earlyapoptotic and apoptotic lysed cells was determined on aBecton-Dickinson FACS Vantage SE BD system.
Aggregation assay
Cells were tested for their ability to aggregate in hanging dropsuspension cultures [27]. Seventy-two hours after transfec-tion with siRNA number 1, the cell culture was trypsinized inthe presence of EDTA, washed twice in PBS and resuspendedin DMEM. 1.5×104 cells in 30 μl of media were suspended ashanging drops from the lid of a 24-well culture dish. Thecorresponding wells were filled with sterile water to preventdrying of the drops. The culture dish was kept in a humid 5%CO2 incubator at 37 °C. Aggregation was determined 4, 8 and12 h after plating. For this purpose, the cells in each dropwere passed 10 times through a standard 200 μl Gilson pipettip, and photographed within 20 min through a Nikon E600microscope 10× phase contrast objective. The number ofisolated cells and of cells forming aggregates was counted.
Measurement of transepithelial electrical resistance (TER)
MDCK cells were plated at subconfluent density on Transwell-clear inserts (Cat. No. 3452, Costar, Cambridge, MA, USA). Twodays after plating, after confirming by visual inspection in theinverted microscope, that the monolayers had reached con-fluency, the transepithelial electrical resistance (TER) wasmeasured using the EVOM epithelial voLtohmmeter (WorldPrecision Instruments, Sarasota, FL, USA). TER values,expressed as Ω cm2, were obtained by subtracting the cell-free filter readings. Twenty-four hours later TER measure-ments were repeated (time 0 in Fig. 3A) and if the valueremained stable, the monolayers were then left untreated,mock transfected or transfected with siRNA number 1. TERmeasurements were then done on several consecutive days.
For the Ca2+ switch assay the confluentmonolayers grownonTranswell filters were transferred to low Ca2+ (LC) medium (1–5 μMCa2+) 3 days after being transfectedwith siRNA. After 20 h ofincubation in LCmedium, the inserts were culturedwith normal
Fig. 4 – ZO-2 silencing perturbs the distribution of E-cadherin. Tof the monolayers was first determined. One day later, after obtawith NC siRNA, ZO-2 siRNA, or full-length canine ZO-2 plus ZO-2processed for the immunofluorescence detection of the apical prindicates the apical localization of E-cadherin, in the monolayers
calciummedium (1.8mMCa2+) for different periods of time afterwhich TER was measured as described above.
Paracellular flux assay of non-ionic molecules
This assay was done as previously described [28]. Briefly, 1.5 mlof the tracer solution (10 μg/ml FITC–dextran of 70 kDa; Sigma-Aldrich, Cat. No. FD-70S) was added to the apical side of con-fluentmonolayers grownonTranswell-clear inserts. After 1 h ofincubation at 37 °C the media from the bottom chamber werecollected and the FITC–dextran was measured in a fluorometer(excitation 492 nm; emission 520 nm). These experiments weredoneon themonolayerswhoseTERmeasurements aredepictedin Fig. 3A at day 5.
Transmission electron microscopy
MDCK monolayers were fixed with 2.5% glutaraldehyde in0.1 M sodium cacodylate buffer, pH 7.2 during 60 min andpostfixed with 1% osmium tetroxide. After dehydration withcrescent concentrations of ethanol and propylene oxide, thesamples were embedded in eponate resin and polymerized at60 °C during 24 h. Thin sections (60 nm) were contrasted withuranyl acetate and lead citrate and the observation was donein a Zeiss EM 910 electron microscope.
Semithin sections
These sections were stained with toluidine blue and observedin an Axiophot Zeiss light microscope using an AxioCam MRcZeiss.
Synchronization of cell cultures and cell proliferation assay
MDCK cells were plated at confluent (3×105 cells/cm2) andsparse density (0.5×105 cells/cm2) in Dulbecco's modifiedEagle's basal medium (DMEM; D1152 Sigma Co., St. Louis,MO) with penicillin (100 IU/ml; Eli Lilly, Mexico) and 10% ironsupplemented certified calf serum (Gibco BRL, 10371-029,Grand Island, NY). Twenty-four hours later the cultures weretransferred to DMEM containing 0.1% of the serum for 2 days.
wo days after plating on Transwell-Clear inserts, the TERining a similar TER value, the monolayers were transfectedsiRNA. Three days post-transfection, the monolayers were
otein gp135 and the lateral marker E-cadherin. The arrowtransfected with ZO-2 siRNA.
1539E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
Cell cycle entry was then triggered by addition of 10% serum.Twenty-four hours later a group of cells was transfected withZO-2 siRNA, mock transfected or left untreated (control).
The entry into the S-phase was quantified by determiningthe incorporation of [3H]thymidine. For such a purpose,synchronized cells were incubated for 24 h with 0.45 μCi/ml[3H]thymidine. The incorporation was stopped with twowashes with cold PBS, precipitation with 5% TCA, and celllysis with 0.4 M NaOH. After 1 h of gentle agitation at roomtemperature, 100 μl of each sample was taken for proteinquantification and other 100 μl was solubilized in 25 μl of 10%acetic acid, 400 μl of 0.4MNaOH and 2ml of scintillation liquid.Samples were counted in a Beckman β-scintillation counter(LS 6000 TA).
Fig. 5 – ZO-2 silencing modifies the monolayer architecture. Phaof sparse and confluent cultures treated with ZO-2 siRNA (A) whesections processed for electronmicroscopy (B) show cells growingsilenced monolayers. This effect is abolished in cells co-transfec
Results
Treatment with ZO-2 siRNA silences the expression of ZO-2 inMDCK cells
The initial aim of this study was to determine if the expressionof ZO-2 could be silenced with siRNA. The effect of thetransfection of the ZO-2 siRNA sequences described in Fig. 1Ais analyzed at the mRNA level in Fig. 1B. Three days aftertransfection, both ZO-2 siRNAs employed at a concentration of10nM, areable to significantly reduce theamountofZO-2mRNAin MDCK monolayers in comparison to untreated monolayers(control), to mock-transfected cultures and to cells transfected
se contrast microscopy reveals no change in cell morphologyreas semithin sections observed by light microscopy and thinon top of each other andwidened intercellular spaces in ZO-2ted with ZO-2 siRNA and full-length canine ZO-2.
1540 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
with 10nMof thenon-effective siRNA (NC). In Fig. 1CweanalyzebyWestern blot, the effect of ZO-2 siRNA on the amount of ZO-2protein expression. Observe how even with 2 nM of ZO-2 siRNAnumber 1, the amount of ZO diminishes significantly (54%)whencompared tomonolayers transfectedwithNCsiRNA.WithZO-2 siRNAnumber2,only theconcentrationof10nMiscapableof significantly reducing (48%) the amount of ZO-2 protein. Toconfirm, that the decreased expression of ZO-2 protein is notcaused by off-target effects of the siRNA, we transfected full-length canine ZO-2 together with 10 nM ZO-2 siRNA 1, andobserved that this treatment rescued ZO-2 expression to a leveleven higher than the one detected in control monolayers.
Since with 10 nM of ZO-2 siRNA number 1, we obtain both asignificant reduction in ZO-2 mRNA (45%) and protein (65%),we selected it for the rest of the experiments here presented.ZO-2 silencing with 10 nM siRNA 1 is apparent within 24 h andpersists for a week (Fig. 1D).
Next we analyzed by confocal immunofluorescence theexpression of ZO-2 in confluent monolayers. Fig. 1E showshow the intensity of ZO-2 staining at the cellular borderssignificantly diminishes in the monolayers treated with ZO-2siRNA as compared to control cultures, mock- and NC siRNA-transfected monolayers. Next, we co-transfected ZO-2 siRNAtogether with full-length canine ZO-2, and observed that theexpression of ZO-2 at the cellular borders was rescued.
In order to determine if the administration of ZO-2 siRNAnumber 1 generated a deleterious effect on the epithelial mono-layer, we analyzed the cell viability of the culture, by Trypan blueexclusion, and observed no change compared to mock-trans-fectedculturesandcellstransfectedwith10nMNCsiRNA(Fig.1F).
Fig. 6 – Actin distribution is altered in ZO-2 siRNA-treatedcells. The apical ring of actin was visualized withrhodaminated phalloidin in control monolayers and incultures transfected with NC siRNA, ZO-2 siRNA, and ZO-2siRNA plus full-length canine ZO-2.
ZO-2 silencing transiently delays cell aggregationin a hanging drop assay
To determine whether ZO-2 down regulation affected cell–celladhesion,weevaluated the impactofZO-2 siRNAtransfectiononcell aggregationusing ahanging drop assay. Fig. 2 showshow3hafter the start of the assay no significant difference could bedetected either by microscopic appreciation or by counting thepercentage of isolated cells and of those forming aggregates,between the ZO-2 siRNA-transfected cells, and control mono-layers, cultures transfected with NC siRNA and cells co-transfectedwithZO-2 siRNAand full-lengthcanineZO-2. Insteadby the 6th hour, in ZO-2 siRNA-transfected cultures, the numberof isolated cells is higher and a decreased number of cells arefound forming aggregates with 9 to 15 and 16 to 20 cells. Thisdifference is abolished in the cultures co-transfected with ZO-2siRNA and full-length ZO-2. At the 9th hour, no detectabledifferences could beobservedamong the conditions tested, sinceat this time the number of cells forming the aggregates was sohigh, that it hindered the possibility of correctly counting thenumber of cells constituting them.
ZO-2 silencing diminishes the TER developed during a Ca2+
switch assay and increases the paracellular permeability ofepithelial monolayers
The graph in Fig. 3A shows the TER values at post-transfectiondays 1 to 5. Cultures treated with ZO-2 siRNA have TER values
similar to those of control monolayers, mock- and NC-trans-fected cultures.
In order to study if ZO-2 silencing could alter de novo TJformation, we measured the TER of monolayers during a Ca2+
switch assay. Fig. 3B shows how ZO-2 siRNA-transfected cellsachieve lower values of TER than control monolayers andthose treated with negative control (NC) siRNA.
Since theTJ regulates thepassageof both ionsandunchargedmolecules throughtheparacellularpathway,wenextproceededto analyze if the down regulation of ZO-2 affected the flux of a70-kDa FITC–dextran, anon-ionicparacellularmarker. As canbeobserved in Fig. 3C, ZO-2 siRNA treatment induced a 10-foldincrease in the paracellular permeability of the 70-kDa FITC–dextran in comparison to the rest of the experimental condi-tions. We therefore conclude that although the electricalresistance remains unchanged in mature monolayers (Fig. 3A),ZO-2 silencing induces a leakage of small non-ionic tracersthrough the paracellular pathway.
ZO-2 silencing induces the appearance of E-cadherin at theapical epithelial surface
Once we had observed that the gate function of the TJ wascompromised during ZO-2 silencing, we proceeded to explore ifthe fenceproperty of theTJwasmaintained. For this purposeweexamined by immunofluorescence z sections, if ZO-2 knock-down altered the distribution of the apical plasma membraneprotein gp135 and the basolateral protein E-cadherin.
In ZO-2 siRNA-transfected cells, GP135 remained confinedto the luminal surface (Fig. 4 middle panel) while E-cadherinsurprisingly appeared along the apical surface (Fig. 4 leftpanel). Transfection of the ZO-2 construct together with theZO-2 siRNA prevents the apical distribution E-cadherin.
We further observed that ZO-2 siRNA-transfected mono-layers display an atypical profile characterized for the presenceof regions where the apical surface appears overgrown.
1541E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
ZO-2 down regulation generates profound alterations inmonolayer architecture
Next we studied in further detail the morphology of ZO-2silenced cultures. By phase contrast microscopy we observedno differences between ZO-2 siRNA and the rest of themonolayers in both sparse and confluent cultures (Fig. 5A). Inthe latter we even detected the presence of domes. These are
Fig. 7 – ZO-2 siRNA treatment has no effect on cell density, prolifecells extruded from the culture. (A) The number of cells present icultures was determined by counting the number of cells in a Neincorporation of [3H]thymidine in sparse cultures of MDCK cells sserum. For normalization we gave a value of 1 to the counts obtaculture show green immunofluorescence, while the red dye correcells released per hour into the medium was determined by couchange in 3-day confluent cultures plated in 28-cm2 petri dishesindependent experiments.
blister-like structures formed in polarized epithelia as a resultof vectorial transport of ions that induces a localized fluidaccumulation between the basal cell surface and the plasticsubstratum [29,30].
The semithin sections and transmission electron micro-scopy (TEM) images shown in Fig. 5B, illustrate the conspic-uous intercellular space developed between most of the cellsof the monolayer in the ZO-2 siRNA-transfected cultures. It is
ration and apoptosis although it decreases the low amount ofn control monolayers, NC siRNA and ZO-2 siRNA-transfectedubauer chamber. (B) Cell proliferation was determined byynchronized by incubation for 48 h with media with 0.1%ined in the control condition. (C) TUNEL, apoptotic cells in thesponds to propidium iodide stained nuclei. (D) The amount ofnting the number of cells present 48 h after the last medium. Results are reported as media±standard error, from 3
1542 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
also noteworthy the observation that in these cultures,scattered along the monolayer, cells are found growing ontop of each other. These alterations are not observed in mock-transfected monolayers and in cultures treated with NCsiRNA. The transfection of full-length ZO-2 together withZO-2 siRNA rescues the characteristic epithelial phenotypeindicating that ZO-2 silencing profoundly alters the mono-layer architecture.
An altered actin pattern is found in ZO-2 silenced cultures
When rhodaminated phalloidin was employed to visualizeactin, an altered apical ring was observed in ZO-2 siRNA-transfected cultures (Fig. 6). In them the actin ring thatencircles the cell acquires a punctuated pattern that contrastswith the continuous staining detected in control and NCsiRNA-transfected monolayers. This effect appears to be dueto ZO-2 silencing as it is absent in the monolayers co-transfected with ZO-2 siRNA and full-length ZO-2.
Fig. 8 – ZO-2 silencing delays the arrival of ZO-1, occludin and E-cMonolayers in control condition, transfected with NC siRNA, ZO-transferred to low calcium media. Twenty hours later (time 0) thdifferent time periods and processed for the immunofluorescenc
ZO-2 silencing has no effect on cell proliferation and apoptosis
The perturbed monolayer profiles observed in Figs. 4 and 5B,prompted us to determine if cell density is altered in ZO-2silenced cultures. Fig. 7 shows how ZO-2 and NC siRNA-transfected cultures contain a similar number of cells (Fig. 7A).In accordance, the amount of [3H]thymidine incorporation inZO-2 siRNA-transfected cells is similar to that of NC-transfectedcultures (Fig. 7B). Next we searched if the number of apoptoticcellswasdifferent. Fig. 7CshowshowemployinganFITC-TUNELdetection system, very few apoptotic cells were present in bothNC siRNA-transfected cultures (2.65%±2.1, having counted 1106cells), and in those treatedwith ZO-2 siRNA (1.46%±1.18, havingcounted 1140 cells). As a positive control we included culturesincubated for 8 hwith 1 μMof staurosporine, a potent apoptosisinducer. Observe how in this culture many cells have detachedand the remaining ones exhibit a strong FITC signal. Weconfirmed these results quantifying apoptosis in control, NC-and ZO-2 siRNA-transfected cells performing an annexin V
adherin to the plasmamembrane during a Ca2+ switch assay.2 siRNA, and ZO-2 siRNA plus ZO-2, were 3 days latere cultures were incubated in calcium containing media fore detection of ZO-1, occludin and E-cadherin.
Fig. 8 (continued).
1543E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
labeling experiment (data not shown). Finally we determined ifthe number of cells extruded from the culture changedwith ZO-2 siRNA transfection. Fig. 7D shows how the number of cellsextruded from the monolayer is lower in ZO-2 siRNA-trans-fected cells than in control and NC siRNA-transfected cultures.However, the amount of cells released in the three cultures is solow that the difference found among themhas no impact in thenumber of cells that remain in the monolayer (Fig. 7A).
The arrival of AJ and TJ proteins to the membrane during aCa2+ switch assay, is delayed in monolayers transfected withZO-2 siRNA
In the Ca2+ switch assay, ZO-2 silenced monolayers reachedlower values of TER (Fig. 3B). This prompted us to analyze ifunder this condition the arrival to the plasma membrane ofdifferent cell–cell adhesion proteins was being affected. Fig. 8showshowZO-2 siRNA-transfected cultures, display even at the5th hour of the Ca2+ switch, a discontinuous pattern of ZO-1 (A),occludin (B) and E-cadherin (C), that contrasts with the clearstaining that surrounds control cultures and monolayerstransfected with NC siRNA and ZO-2 siRNA plus full-lengthcanine ZO-2.
In themonolayerswhereZO-2was silenced, certainareas ofthe culture were practically devoid of occludin and exhibited adiffuse E-cadherin staining (asterisks).
ZO-2 siRNA decreases occludin and E-cadherin expression butexerts no effect on ZO-1 and claudin-1
Next we proceeded to analyze if ZO-2 siRNA treatment affectsthe expression level of proteins present at the tight andadherens junctions. In order to test this point in a conditionwhere maximal knockdown of ZO-2 can be attained, MDCKcells were transfected in 2 consecutive days with siRNA. Theeffect that this double transfection exerts on the expressionof ZO-2, ZO-1, occludin, claudin-1 and E-cadherin is shown inFig. 9. Western blot analysis reveals a 91% decreased ex-pression of ZO-2 when cells are twice transfected with ZO-2siRNA. This profound silencing of ZO-2 impacts the expres-sion of other cell–cell adhesion proteins, as the level of E-cadherin and occludin diminishes 58 and 83% respectively.However, the expression level of ZO-1 and claudin-1 remainsunchanged (Fig. 9A). Occludin and E-cadherin expression isrestored by co-transfecting full-length ZO-2 together with theZO-2 siRNA.
Fig. 8 (continued).
1544 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
Next we analyzed by immunofluorescence the expressionof these cell–cell adhesion proteins. In order to make a moreaccurate comparison of transfected versus non-transfectedcells, we mixed control with ZO-2 knockdown cells which hadbeen additionally labeled with the fluorescent reagent Cell-tracker™ Orange CMTMR. Fig. 9B illustrates how the ZO-2siRNA-transfected cells in the co-culture, distinguished forexhibiting a fluorescent red stain, have a barely detectable ZO-2 stain at the cellular borders (arrow) that contrasts to thatobserved in non-transfected cells. In agreement with theWestern blot results, the expression of ZO-1 and claudin-1 isunaffected (emptyarrowheads),while thepresenceof occludinand E-cadherin at the cellular borders diminishes in ZO-2silenced cells (arrows). In accordance with Figs. 4 and 5C, E-cadherin displays a diffuse apical staining in ZO-2 silencedcells (full arrowhead).
Discussion
In this work we have been able to demonstrate that byemploying siRNAs directed against ZO-2 it is possible to silencethe expression of this protein in cultured epithelial cells.
To discard the possibility of having phenotypic changesprovoked by off-target effects of ZO-2 siRNA, in this study wetook the following precautions: first we employed StealthsiRNAs (Invitrogen) as with them only the antisense strandcan participate in RNAi; second, we employed as a negativecontrol an NC siRNA that is not homologous to anything in thevertebrate transcriptosome, and third we showed that thetransfection of a canine ZO-2 can reverse the phenotypecaused by ZO-2 siRNA.
The first evidence indicating that ZO-2 is involved in theestablishment of cell–cell adhesions appeared in the cellaggregation assay. The fact that the differences foundbetween ZO-2 siRNA and mock-transfected cells is transitory,suggests that the presence of ZO-2 is important in the earlystages of cell adhesion and that later its absence can beovercome by other cell adhesion proteins.
The observation that ZO-2 siRNA diminishes the amountof occludin is particularly interesting, and can even bedetected, albeit to a lower extent, with the 65% level of ZO-2silencing obtained with a single transfection of ZO-2 siRNA(data not shown). Although occludin is capable of directlyinteracting with actin [8], it also associates indirectly with thecytoskeleton through the ZO proteins, since the carboxyl
1545E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
terminal region of these MAGUKs interacts directly with actin[8,11]. Therefore, we suspect that ZO-2 silencing diminishesoccludin stability giving rise to the decreased amount of thisprotein detected in ZO-2 siRNA-treated cultures.
Further evidence confirming the importance of ZOproteins in occludin arrival at the TJ has been obtained inexperiments with the mammary epithelial cell line Eph4 andMDCK, where cells respectively lacking ZO-1 expression orhaving ZO-1 transiently silenced, display a retardation of therecruitment of occludin to junctional areas during a Ca2+
switch [31,32]. However, the absence of ZO-1 in thoseexperiments did not diminish the total amount of occludinpresent in the cells. A recent work indicated that in Eph4cells stably expressing siRNAs against ZO-2, recruitment tothe plasma membrane of occludin, JAM and claudin 3, is notaffected. Instead the arrival of these proteins to the cellborders is blocked if the expression of ZO-1 and ZO-2 issimultaneously suppressed [33]. Furthermore, in Eph4 cells,ZO-2 silencing exerts no effect on the paracellular flux ofdextran, cell morphology and the polarized distribution of E-cadherin and instead a blockade of both ZO-1 and ZO-2expression is needed in order to inhibit the formation of TJs[33]. These results suggest that in Eph4 cells ZO-1 and ZO-2are functionally redundant, and hence it is necessary to
Fig. 9 – ZO-2 silencing diminishes the expression of occludin andwere transfected twice in consecutive days with one of the followlength canine ZO-2. Twenty-four hours after the second transfecclaudin was analyzed with specific antibodies by Western blot. Rindependent experiments (A). The distribution of ZO-1, ZO-2, E-cimmunofluorescence in mixed cultures formed by control and twred fluorescent dye CMTMR (B).
suppress the expression of both proteins in order to generatea phenotypic change. Since our results demonstrate that thesole suppression of ZO-2 in MDCK cells is sufficient todiminish occludin expression, to delay its arrival to theplasma membrane, and to alter the gate functions of the TJ,we suspect that in MDCK cells ZO-1 and ZO-2 displayindependent functions. This point is further reinforced byobserving for example how in MDCK cells, ZO-1 associates toZO-3 while ZO-2 does not [8], and how ZO-2 and not ZO-1 isable to down regulate the activity of gene promotersregulated by AP-1 sites [18].
The controversy is however not over, since a recent workperformed with MDCK cells showed that the transientsilencing of ZO-2 has no affect on occludin expression [32].This discrepancy could again be due to differences in thecellular context that arise with the employment of differentclones of MDCK cells. Ample evidence has shown howdifferent clones of this line, exhibit diverse sets of claudins[34], display contrasting levels of TER [35,36], vary in theircapacity to form domes and tubules [37] and exhibit differentplasma membrane distributions of the same protein [38–41].
The increased paracellular flow of Dextran in MDCKmonolayers that suffer no change in their TER upon ZO-2silencing, might be related to the decreased expression of
E-cadherin but has no effect on ZO-1 and claudin-1. Culturesing constructs: ZO-2 siRNA, NC siRNA or ZO-2 siRNA plus full-tion, the amount of ZO-1, ZO-2, E-cadherin, occludin andesults are reported as media±standard error, from 2adherin, occludin and claudin was analyzed byice ZO-2 siRNA-transfected cells additionally labeled with the
1546 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
occludin, as this protein lacks the charged residues present inthe extracellular loops of claudins responsible for the ionicselectivity of the TJ [42–44]. Hence, the diminished expressionof occludin is expected to have a stronger impact on theparacellular permeability of non-ionic tracers than in theTER.
The Ca2+ switch experiment revealed how ZO-2 silencingdelays the arrival of ZO-1, occludin and E-cadherin to theplasmamembrane. In contrast to ZO-1 and occludinwhich areknown or presumed, to associate directly with ZO-2,E-cadherin is not expected to bind to it, since other MAGUKslike ZO-1 do not associate to the cytoplasmic domain ofE-cadherin [45].
The apical expression of E-cadherin generated by ZO-2knockdown, suggests a role for ZO-2 in maintaining AJconstituents at cell–cell contact regions. This is not unex-pected since α-catenin binds directly to the amino terminalportion of ZO-2 [6] and the catenin complex provides thebridge for the interaction between E-cadherin and the actincytoskeleton ring that surrounds epithelial cells.
ZO-2 silencing does not alter the number of cells in themonolayer, the proliferation or the apoptotic rate. It onlydiminishes the number of cells released from the monolayer.However, since the latter is very small, it exerts no significantimpact in the number of cells that constitute a monolayer.Therefore, the altered architecture of the ZO-2 silencedmonolayers, characterized for the presence of regions wherecells are growing on top of each other cannot be explained byan uncontrolled cell proliferation, and instead might berelated to the non-polarized expression of E-cadherin.Hence, in these monolayers the non-polarized presence of E-cadherin at the plasma membrane might encourage thegrowth of neighboring cells on top of each other, by apical tobasal interactions of E-cadherin.
The expanded intercellular spaces found in ZO-2 silencedmonolayers, somehow resemble the intercellular blebsdescribed in small intestine segments perfused with glucose,alanine or leucine solutions [46]. In that case it was demon-strated that such an effect was due to the contraction of theperijunctional actomyosin ring. In our ZO-2 silenced mono-layers, we observe a punctuated actin staining at the cellperiphery and suspect that this pattern might be related to anaffected anchorage of actin to the plasmamembrane, probablydue to the diminished amount of ZO-2 expressed in theculture, since this protein functions as a bridge between thecytoskeleton and the integral proteins of the AJ and the TJ.
In summary, these results suggest that the presence of ZO-2 is crucial for maintaining occludin and E-cadherin at theintercellular junctional complex, and for keeping a continuousapical ring of actin. These conditions are important for theestablishment of non-stratified cultures, capable of restrictingthe paracellular passage of molecules.
Acknowledgments
This work was supported by grant 45691-Q from the MexicanNational Council for Science and Technology (CONACYT).Sandra Hernandez was a recipient of a doctoral fellowshipfrom CONACYT (No. 144231).
R E F E R E N C E S
[1] J.M. Anderson, M. Cereijido, Introduction. Evolution of ideason the tight junction, in: M. Cereijido, J.M. Anderson (Eds.),Tight Junction, CRC Press, Boca Raton, 2001.
[2] M. Cereijido, L. Gonzalez-Mariscal, R.G. Contreras, Tightjunction: barrier between higher organisms andenvironment, NIPS 4 (1989) 72–75.
[3] L. Gonzalez-Mariscal, A. Betanzos, P. Nava, B.E. Jaramillo,Tight junction proteins, Prog. Biophys. Mol. Biol. 81 (2003)1–44.
[4] L. Gonzalez-Mariscal, A. Betanzos, A. Avila-Flores, MAGUKproteins: structure and role in the tight junction, Semin. CellDev. Biol. 11 (2000) 315–324.
[5] B. Gumbiner, T. Lowenkopf, D. Apatira, Identification of a160-kDa polypeptide that binds to the tight junctionprotein ZO-1, Proc. Natl. Acad. Sci. U. S. A. 88 (1991)3460–3464.
[6] M. Itoh, K. Morita, S. Tsukita, Characterization of ZO-2 as aMAGUK family member associated with tight as well asadherens junctions with a binding affinity to occludin andalpha catenin, J. Biol. Chem. 274 (1999) 5981–5986.
[7] M. Itoh, M. Furuse, K. Morita, K. Kubota, M. Saitou, S. Tsukita,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 (1999) 1351–1363.
[8] E.S. Wittchen, J. Haskins, B.R. Stevenson, Protein interactionsat the tight junction. Actin hasmultiple binding partners, andZO-1 forms independent complexes with ZO-2 and ZO-3,J. Biol. Chem. 274 (1999) 35179–35185.
[9] M. Cordenonsi, F. D'Atri, E. Hammar, D.A. Parry, J.Kendrick-Jones, D. Shore, S. Citi, Cingulin contains globularand coiled-coil domains and interacts with ZO-1 formsindependent complexes with ZO-2, ZO-3, and myosin, J. CellBiol. 147 (1999) 1569–1582.
[10] S.N. Mattagajasingh, S.C. Huang, J.S. Hartenstein, E.J. BenzJr., Characterization of the interaction between protein4.1R and ZO-2. A possible link between the tight junctionand the actin cytoskeleton, J. Biol. Chem. 275 (2000)30573–30585.
[11] A.S. Fanning, B.J. Jameson, L.A. Jesaitis, J.M. Anderson, Thetight junction protein ZO-1 establishes a link between thetransmembrane protein occludin and the actin cytoskeleton,J. Biol. Chem. 273 (1998) 29745–29753.
[12] S. Islas, J. Vega, L. Ponce, L. Gonzalez-Mariscal, Nuclearlocalization of the tight junction protein ZO-2 in epithelialcells, Exp. Cell Res. 274 (2002) 138–148.
[13] A. Traweger, R. Fuchs, I.A. Krizbai, T.M. Weiger, H.C. Bauer, H.Bauer, The tight junction protein ZO-2 localizes to the nucleusand interacts with the heterogeneous nuclearribonucleoprotein scaffold attachment factor-B, J. Biol. Chem.278 (2003) 2692–2700.
[14] A. Renz, F.O. Fackelmayer, Purification andmolecular cloningof the scaffold attachment factor B (SAF-B), a novel humannuclear protein that specifically binds to S/MAR-DNA, NucleicAcids Res. 24 (1996) 843–849.
[15] S. Oesterreich, Scaffold attachment factors SAFB1 and SAFB2:innocent bystanders or critical players in breasttumorigenesis? J. Cell. Biochem. 90 (2003) 653–661.
[16] B.E. Jaramillo, A. Ponce, J. Moreno, A. Betanzos, M. Huerta, E.Lopez-Bayghen, L. Gonzalez-Mariscal, Characterization of thetight junction protein ZO-2 localized at the nucleus ofepithelial cells, Exp. Cell Res. 297 (2004) 247–258.
[17] L. Gonzalez-Mariscal, S. Islas, R.G. Contreras, M.R.Garcia-Villegas, A. Betanzos, J. Vega, A. Diaz-Quinonez, N.Martin-Orozco, V. Ortiz-Navarrete, M. Cereijido, J. Valdes,Molecular characterization of the tight junction protein ZO-1in MDCK cells, Exp. Cell Res. 248 (1999) 97–109.
1547E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 5 3 3 – 1 5 4 7
[18] A. Betanzos, M. Huerta, E. Lopez-Bayghen, E. Azuara, J.Amerena, L. Gonzalez-Mariscal, The tight junction proteinZO-2 associates with Jun, Fos and C/EBP transcription factorsin epithelial cells, Exp. Cell Res. 292 (2004) 51–66.
[19] E. Willott, M.S. Balda, A.S. Fanning, B. Jameson, C. VanItallie,J.M. Anderson, The tight junction protein ZO-1 is homologousto the Drosophila discs—large tumor suppressor protein ofseptate junctions, Proc. Natl. Acad. Sci. U. S. A. 90 (1993)7834–7838.
[20] D.F. Woods, P.J. Bryant, The discs—large tumorsuppressor gene of Drosophila encodes a guanylate kinasehomolog localized at septate junctions, Cell 66 (1991)451–464.
[21] A. Chlenski, K.V. Ketels, G.I. Korovaitseva, M.S. Talamonti, R.Oyasu, D.G. Scarpelli, Organization and expression of thehuman zo-2 gene (tjp-2) in normal and neoplastic tissues,Biochim. Biophys. Acta 1493 (2000) 319–324.
[22] A. Chlenski, K.V. Ketels, M.S. Tsao, M.S. Talamonti, M.R.Anderson, R. Oyasu, D.G. Scarpelli, Tight junction protein ZO-2 is differentially expressed in normal pancreatic ductscompared to human pancreatic adenocarcinoma, Int. J.Cancer 82 (1999) 137–144.
[23] B.A. Glaunsinger, R.S. Weiss, S.S. Lee, R. Javier, Link of theunique oncogenic properties of adenovirus type 9 E4-ORF1 toa select interaction with the candidate tumor suppressorprotein ZO-2, EMBO J. 20 (2001) 5578–5586.
[24] L. Gonzalez-Mariscal, D.R. Chavez, M. Cereijido, Tightjunction formation in cultured epithelial cells (MDCK),J. Membr. Biol. 86 (1985) 113–125.
[25] S.J. Wu, D.C. Spink, B.C. Spink, L.S. Kaminsky, Quantitation ofCYP1A1 and 1B1 mRNA in polycyclic aromatichydrocarbon-treated human T-47D and HepG2 cells by amodified bDNA assay using fluorescence detection, Anal.Biochem. 312 (2003) 162–166.
[26] J. Soutschek, A. Akinc, B. Bramlage, K. Charisse, R. Constien,M. Donoghue, S. Elbashir, A. Geick, P. Hadwiger, J. Harborth,M. John, V. Kesavan, G. Lavine, R.K. Pandey, T. Racie, K.G.Rajeev, I. Rohl, I. Toudjarska, G. Wang, S. Wuschko, D.Bumcrot, V. Koteliansky, S. Limmer, M. Manoharan, H.P.Vornlocher, Therapeutic silencing of an endogenous gene bysystemic administration of modified siRNAs, Nature 432(2004) 173–178.
[27] M.A. Thoreson, P.Z. Anastasiadis, J.M. Daniel, R.C. Ireton, M.J.Wheelock, K.R. Johnson, D.K. Hummingbird, A.B. Reynolds,Selective uncoupling of p120(ctn) from E-cadherin disruptsstrong adhesion, J. Cell Biol. 148 (2000) 189–202.
[28] P. Nava, S. Lopez, C.F. Arias, S. Islas, L. Gonzalez-Mariscal, Therotavirus surface protein VP8 modulates the gate and fencefunction of tight junctions in epithelial cells, J. Cell Sci. 117(2004) 5509–5519.
[29] J. Leighton, Z. Brada, L.W. Estes, G. Justh, Secretory activityand oncogenicity of a cell line (MDCK) derived from caninekidney, Science 163 (1969) 472–473.
[30] S. Fernandez-Castelo, J.J. Bolivar, R. Lopez-Vancell, G. Beaty,M. Cereijido, Ion transport in MDCK cells, in: M. Taub (Ed.),Tissue Culture of Epithelial Cells, Plenum Press, New York,1985.
[31] K. Umeda, T. Matsui, M. Nakayama, K. Furuse, H. Sasaki, M.Furuse, S. Tsukita, Establishment and characterization of
cultured epithelial cells lacking expression of ZO-1, J. Biol.Chem. 279 (2004) 44785–44794.
[32] E. McNeil, C.T. Capaldo, I.G. Macara, Zonula occludens-1function in the assembly of tight junctions in Madin-Darbycanine kidney epithelial cells, Mol. Biol. Cell 17 (2006)1922–1932.
[33] K. Umeda, J. Ikenouchi, S. Katahira-Tayama, K. Furuse, H.Sasaki, M. Nakayama, T. Matsui, S. Tsukita, M. Furuse, S.Tsukita, ZO-1 and ZO-2 independently determine whereclaudins are polymerized in tight-junction strand formation,Cell 126 (2006) 741–754.
[34] M. Furuse, K. Furuse, H. Sasaki, S. Tsukita, Conversion ofzonulae occludentes from tight to leaky strand type byintroducing claudin-2 into Madin-Darby canine kidney I cells,J. Cell Biol. 153 (2001) 263–272.
[35] B.R. Stevenson, J.M. Anderson, D.A. Goodenough, M.S.Mooseker, Tight junction structure and ZO-1 content areidentical in two strains of Madin-Darby canine kidney cellswhich differ in transepithelial resistance, J. Cell Biol. 107(1988) 2401–2408.
[36] L. Gonzalez-Mariscal, D.R. Chavez, A. Lazaro, M. Cereijido,Establishment of tight junctions between cells from differentanimal species and different sealing capacities, J. Membr.Biol. 107 (1989) 43–56.
[37] R.J. Klebe, A. Grant, G. Grant, P. Ghosh, Cyclic-AMPdeficient MDCK cells form tubules, J. Cell. Biochem. 59(1995) 453–462.
[38] M.J. Caplan, H.C. Anderson, G.E. Palade, J.D. Jamieson,Intracellular sorting and polarized cell surface delivery of(Na+,K+)ATPase, an endogenous component of MDCK cellbasolateral plasma membranes, Cell 46 (1986) 623–631.
[39] R.W. Hammerton, K.A. Krzeminski, R.W. Mays, T.A. Ryan,D.A. Wollner, W.J. Nelson, Mechanism for regulating cellsurface distribution of Na+, K(+)-ATPase in polarizedepithelial cells, Science 254 (1991) 847–850.
[40] C. Zurzolo, E. Rodriguez-Boulan, Delivery ofNa+,K(+)-ATPase in polarized epithelial cells, Science 260(1993) 550–552.
[42] O.R. Colegio, C.M. VanItallie, H.J. McCrea, C. Rahner, J.M.Anderson, Claudins create charge-selective channels in theparacellular pathway between epithelial cells, Am. J. Physiol.:Cell Physiol. 283 (2002) C142–C147.
[43] O.R. Colegio, C. VanItallie, C. Rahner, J.M. Anderson, Claudinextracellular domains determine paracellular chargeselectivity and resistance but not tight junction fibrilarchitecture, Am. J. Physiol.: Cell Physiol. 284 (2003) C1346–C1354.
[44] C.M. VanItallie, A.S. Fanning, J.M. Anderson, Reversal ofcharge selectivity in cation or anion-selective epithelial linesby expression of different claudins, Am. J. Physiol.: Renal.Physiol. 285 (2003) F1078–F1084.
[45] M. Itoh, A. Nagafuchi, S. Moroi, S. Tsukita, Involvement ofZO-1 in cadherin-based cell adhesion through its direct bindingto alpha catenin and actin filaments, J. Cell Biol. 138 (1997)181–192.
[46] J.L. Madara, J.R. Pappenheimer, Structural basis forphysiological regulation of paracellular pathways inintestinal epithelia, J. Membr. Biol. 100 (1987) 149–164.
