Hypertonic stress increases claudin-4 expression andtight junction integrity in association withMUPP1 in IMCD3 cellsMiguel A. Lanaspa, Ana Andres-Hernando, Christopher J. Rivard, Yue Dai, and Tomas Berl*

Division of Renal Diseases and Hypertension, School of Medicine, University of Colorado Health Sciences Center, Denver, CO 80262

Edited by Maurice B. Burg, National Institutes of Health, Bethesda, MD, and approved August 6, 2008 (received for review June 13, 2008)

We reported that the multiple PDZ protein 1 (MUPP1) is an osmoticresponse protein in kidney cells. This up-regulation was found to benecessary for the maintenance of tight epithelial properties in thesecells. We investigated whether an interaction with one or moremembers of the claudin family is responsible for this observation. Inresponse to hypertonicity, the up-regulation of claudin-4 (Cldn4)expression, and not other claudins, was initially identified in innermedullary collecting duct (IMCD3) cells by gene array and furtherverified by quantitative PCR and Western blotting. In kidney tissues,Cldn4 expression was substantial in the papilla and absent in thecortex. Furthermore, Cldn4 expression significantly increased in thepapilla of mice after 36 h of thirsting. Cldn4 immunofluorescence inhypertonically stressed cells revealed colocalization with MUPP1 atthe tight junctions. Interaction between Cldn4 and MUPP1 was alsodemonstrated by coimmunoprecipitation of both proteins fromIMCD3 cells chronically adapted to hypertonicity. In IMCD3 cells stablysilenced for MUPP1 expression under hypertonic conditions, a signif-icant decrement in Cldn4 expression was observed that was restoredafter inhibition of lysosome activity. Immunofluorescence detectionidentified that in these MUPP1-silenced cells Cldn4 was mistargetedto the lysosomes. Functionally, silencing Cldn4 expression in IMCD3cells resulted in a decrease in the transepithelial resistance to the samedegree as observed when MUPP1 expression was silenced, suggest-ing that MUPP1 contributes to the maintenance of a tight epitheliumin the medulla of the kidney under hypertonic stress by correctlylocalizing Cldn4 to the tight junctions.

osmotic stress response kidney cells inner medulla

The cells that inhabit the hypertonic environment of the innermedulla possess a number of adaptive mechanisms that allow

them to survive this harsh environment. This survival is medi-ated initially by the activation of ion transport systems andthereafter by the cellular accumulation of a number of organicosmolytes (1–4). It has become increasingly evident that inaddition to the proteins required for the cellular uptake and/orsynthesis of these osmolytes (transporters and enzymes), hyper-tonic stress brings about a coordinated response involving theup- and down-regulation of hundreds of genes, many of whichmay be critical to cell viability and adaptation (2, 5).

We recently described the up-regulation of multiple PDZprotein 1 (MUPP1) under hypertonic stress in inner medullarycollecting duct (IMCD3) cells (6). Under these conditions,silencing MUPP1 reduced monolayer transepithelial resistance(TER). MUPP1 through its 13 PDZ domains acts as scaffoldingfor attaching different proteins at the proper location in themembrane (7–12). It has become apparent that in polarizedepithelial cells MUPP1 plays a key role in assembling tightjunctions by recruiting and anchoring different members ofjunctional adhesion molecules (JAM) and claudins (9, 11), thattogether with occludin, form tight junctions strands (for excel-lent reviews, see refs. 13–17). In mammals, claudins represent alarge superfamily of at least 24 different proteins that areconsidered as the principal barrier-forming proteins (16). Clau-dins are transmembrane proteins that span the bilayer four

times, contain two extracellular loop domains, and have bothends oriented toward the cytoplasm (13, 16). The C terminus ofnearly all claudins ends in a PDZ-binding motif that interactswith PDZ domains in the cytoplasmic scaffolding proteins ZO-1,-2, and -3, and MUPP1. Interaction between claudins and PDZproteins is important to form well organized tight junctionstrands at the apical border of the lateral membrane.

To date, two claudins, claudin-1 and claudin-8, have beenidentified to interact with MUPP1 (9, 11). In the kidney, variousclaudins are expressed in different segments of the nephron, butinterestingly, claudin-4 (Cldn4) expression is localized only insegments with high resistance, including the medullary part ofthe collecting ducts (18). We therefore postulated that theobservations we reported for MUPP1 expression in the adaptiveresponse to hypertonicity (6) could involve an interaction withone or more members of the claudin family. Thus, the aim of thiswork was to explore the expression of claudins, their relationshipto MUPP1, and their functional relevance in the response tohypertonic stress in kidney cells.

ResultsEffect of Hypertonicity on the Expression of Claudins in IMCD3 cells.The effect of chronic hypertonic stress on claudin family geneexpression in IMCD3 cells was evaluated by gene chip analysis.As depicted in Fig. 1A, expression of most claudins was absentin IMCD3 cells chronically adapted to hypertonicity (600 mOsm/kgH2O) whereas a number of claudins were down-regulated(claudins-1, -2, -3, and -8). In contrast, the expression of Cldn4mRNA was uniquely and robustly up-regulated (5, P 0.01)under hypertonic stress. This increase was further verified byquantitative PCR using specific primers for mouse Cldn4 (Fig.1B). Expression of Cldn4 protein in IMCD3 cells exposed toacute and chronic hypertonicity was also evaluated. As shown inFig. 1C, Cldn4 protein expression is nearly absent in cellsmaintained at isotonic conditions (300 mOsm/kgH2O). How-ever, in cells acutely stressed (550 mOsm/kgH2O for 24 h) orchronically adapted (600 mOsm/kgH2O) to hypertonicity, Cldn4expression is substantially up-regulated (6 and 11, respec-tively, P 0.01), further validating gene chip data.

Immunocytochemical Localization of Cldn4 in IMCD3 Cells. Immuno-cytochemical staining by confocal microscopy was undertaken toassess the presence and localization of Cldn4 in IMCD3 cellsexposed to hypertonic stress compared with cells maintained at

Author contributions: M.A.L., A.A.-H., C.J.R., and T.B. designed research; M.A.L., A.A.-H.,C.J.R., and Y.D. performed research; M.A.L., A.A.-H., C.J.R., and T.B. analyzed data; andM.A.L., A.A.-H., C.J.R., and T.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

*To whom correspondence should be addressed at: Division of Renal Diseases and Hyper-tension, School of Medicine, University of Colorado Health Sciences Center, 4200 East 9thAvenue, Denver, CO 80262. E-mail: tomas.berl@uchsc.edu.

© 2008 by The National Academy of Sciences of the USA

www.pnas.orgcgidoi10.1073pnas.0805761105 PNAS October 14, 2008 vol. 105 no. 41 15797–15802

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

17, 2

021

isotonic conditions. Fig. 2A Left demonstrates little or nostaining for Cldn4 in IMCD3 cells grown under isotonic condi-tions and only minor intracellular staining. In contrast, inIMCD3 cells stressed to 550 mOsm/kgH2O for 6 h, Cldn4staining increased substantially (Fig. 2 A Right). Confocal Z-scananalysis reveals that Cldn4 immunocytochemical staining oc-curred uniformly in the lateral membrane as expected forrepresentative tight junction proteins. To assess whether Cldn4expression corresponded specifically to tight junctions, we per-formed a line Z-scan of the lateral membrane by using antibodiesto the 1 subunit of the Na/K-ATPase as a basolateral marker.As shown in Fig. 2B, Cldn4 staining was restricted to the top part

of the lateral membrane, indicating that its localization underhypertonic stress is likely to be at the tight junctions.

Kinetics of Cldn4 Protein Expression After Acute Exposure to Hyper-tonicity. Cldn4 protein expression in IMCD3 cells subjected toacute sublethal osmotic stress was evaluated to determine therelative onset of expression. To this end, we undertook Westernblot analysis for protein at numerous time points after exposureto sublethal hypertonicity. Data shown in Fig. 3 demonstrate asignificant increase in Cldn4 protein after 6–8 h of hypertonicstress (550 mOsm/kgH2O) with a maximum level of proteindetermined after 24–36 h.

Expression of Cldn4 in Renal Cortex and Medulla of Rodent Kidneys.To assess whether the changes seen in cultured cells are alsoobserved in renal tissues, protein expression was examined inkidney tissues of mice subjected to water ad libitum and afterthirsting for 36 h (urine osmolality increased from 1,424 211 to3,105 524 mOsm/kgH2O, n 6). Western blotting data shown inFig. 4 indicate a relative absence of Cldn4 protein expression in thecortex and substantial protein in the papilla. Furthermore, Cldn4protein expression significantly increased (P 0.05) in the cortexand papilla tissues after thirsting the animals.

Cldn1

Cldn2Cldn

3Cldn4

Cldn8

Cldn 5-7, 9

-240

1

2

3

4

5

6 300 mOsm/kgH2O600 mOsm/kgH2O

Claudin Superfamily Members

Gen

echi

pm

RN

A le

vels

(arb

itrar

y un

its)

Abs

ent

A

Genechip QPCR0

1

2

3

4

5

6 300 mOsm/kgH 2O600 mOsm/kgH 2O

Cld

n4 m

RN

A le

vels

(arb

itrar

y un

its)

Method

B C

300 550, 24 h 600

Isotonic Hypertonic

0

2500

5000

7500

10000

12500

Wes

tern

blo

t den

sito

met

ry(a

rbitr

ary

units

)

Tonicity (mOsm/kgH2O)

Claudin-4 25 kDaβ-actin 42 kDa

Fig. 1. Comparison of claudin expression by Affymetrix gene chip, quantitative PCR, and Western blotting in IMCD3 cells exposed to hypertonic stress. (A andB) Cells kept at isotonic conditions (300 mOsm/kgH2O) or chronically adapted to 600 mOsm/kgH2O were harvested for mRNA isolation and analyzed by gene array(A) and quantitative PCR (B). Data were collected from three identical experimental replicates and represent the mean SEM for two independent experiments(n 6). (C) Cells kept at isotonic conditions (300 mOsm/kgH2O) or acutely (550 mOsm/kgH2O, 24 h) and chronically stressed to 600 mOsm/kgH2O were harvestedfor protein and analyzed by Western blotting. Western blotting data indicate a substantial increase in Cldn4 protein expression with greater levels of tonicity.A representative Western blot is shown that indicates the -actin loading control.

A

B

Fig. 2. Effect of acute or chronic exposure of IMCD3 cells to hypertonic stress onCldn4 protein localization. (A) Cells acutely exposed to hypertonicity (550 mOsm/kgH2O, 6 h) were analyzed by immunofluorescence against Cldn4. Results showthat Cldn4 localization under hypertonic conditions is restricted to the lateralmembrane of the IMCD3 cells as shown in the Z-scan section. (B) Counterstainingcell monolayers with antibodies to the 1 subunit of the Na/K-ATPase (red) isshown. Line Z-scan analysis indicates that Cldn4 (green) is located near the top ofthe lateral membrane and does not colocalize with the 1 subunit.

Fig. 3. Effect of acute sublethal osmotic stress (550 mOsm/kgH2O) on Cldn4protein expression in IMCD3 cells. Cell lysates (0–48 h) were analyzed by Westernblotting. Data depict the mean SEM from three Western blots performed withduplicate independent samples (80 g of total protein per lane, n 6). Arepresentative Western blot including the -actin loading control is shown.

15798 www.pnas.orgcgidoi10.1073pnas.0805761105 Lanaspa et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

17, 2

021

Cldn4 and MUPP1 Interact and Colocalize in IMCD3 Cells UnderHypertonic Stress. Because the last 2 aa of nearly all claudins,including Cldn4, are considered PDZ-binding domains, we exam-ined a possible interaction between MUPP1 and Cldn4. To stabilizeprotein interactions better, cells were treated with the liposolublecross-linker, dithiobis(succinimidyl)propionate (DSP). It is impor-tant to note that initial coimmunoprecipitation (CoIP) experimentswere performed without the cross-linking agent, resulting in asimilar although less intense signal. As shown in Fig. 5A, MUPP1coprecipitates with Cldn4. In contrast, little or no signal for MUPP1is found in the supernatant obtained from the unbound protein toCldn4 or from further washes (SN1 and SN2, respectively). Further,we colocalized MUPP1 and Cldn4 in IMCD3 cells acutely exposed

to hypertonicity. As shown in Fig. 5B, Cldn4 (green) and MUPP1(red) colocalize (yellow) in these cells. The Z-scan view localizesboth proteins at the lateral membrane, suggesting that this inter-action occurs at the tight junctions.

MUPP1 Is Necessary for Normal Expression and Localization of Cldn4Under Hypertonic Stress. To assess the physiologic role of MUPP1on the osmotic up-regulation of Cldn4, we undertook to silenceprotein expression. As described in ref. 6, we generated stableIMCD3 clones silenced for MUPP1 expression employingshRNA (pSM2-MUPP1; Open Biosystems). We therefore ex-amined the expression of Cldn4 under hypertonic stress in theabsence of MUPP1 and in empty-vector control cells expressingMUPP1 (pSM2-EV). As shown in Fig. 6A by Western blotting,after 24 h of hypertonic stress, Cldn4 expression is substantiallyreduced in MUPP1-silenced cells compared with empty-vectorcontrol cells (59.2% reduction, P 0.01). In contrast, silencingMUPP1 does not affect the expression of other constituentproteins of the tight junctions including ZO-1, indicating thatMUPP1 is important for normal expression of Cldn4. Further-more, immunofluorescence studies (Fig. 6B) performed in thesecells reveal not only a decrease in Cldn4 expression but also a lossin localization at the membrane, suggesting that MUPP1 isinvolved in the proper localization of Cldn4 at the tight junctions.Because mislocalization of Cldn4 in MUPP1-silenced cells maylead to enhanced degradation and thereby an overall decrementin protein expression, we performed experiments inhibitinglysosome activity with 1 M bafilomycin A1, a specific inhibitorof the H-ATPase proton pump. As shown in Fig. 7A, Cldn4expression is essentially restored (open bars) after lysosomalinhibition [86.4% P 0.05 compared with empty-vector controlcells (filled bars)], indicating that MUPP1 absence promoteslysosomal Cldn4 degradation. To evaluate Cldn4 degradation bylysosomes further, MUPP1-silenced cells were stained for Cldn4and LAMP1, a well known lysosome marker. As shown in Fig.7B, Cldn4 (green) colocalized with LAMP1 (red) after 8 h ofhypertonic stress, indicating that Cldn4 is mistargeted to lyso-somes in the absence of MUPP1. In contrast, Cldn4 was not

Fig. 4. Cldn4 protein expression in mouse kidney tissues. Mice kidney tissues(papilla and cortex) were harvested after 36 h of water restriction (thirst) orwith ad libitum water, and protein homogenates were analyzed by Westernblotting. Expression of Cldn4 was substantial in mouse kidney papilla com-pared with little or no expression in cortex tissues. Water restriction in miceleads to a 40% increase in Cldn4 protein expression in papilla (P 0.05). Datarepresent the mean SEM from three Western blots performed with dupli-cate independent samples (80 g of total protein per lane, n 6). A repre-sentative Western blot including the -actin loading control is shown.

A

B

Fig. 5. MUPP1 and Cldn4 interaction in IMCD3 cells under hypertonic stress.(A) MUPP1 and Cldn4 coprecipitate (IP) from IMCD3 cells chronically adaptedto 600 mOsm/kgH2O. Protein interactions were cross-linked before immuno-precipitation with anti-Cldn4. Immunoprecipitates were run on 4–20% gra-dient acrylamide gels and analyzed for MUPP1 and Cldn4 expression. Arepresentative Western blot including the supernatants obtained from theunbound proteins and after a series of washes is shown (SN1, SN2). (B) Cldn4(green) and MUPP1 (red) colocalize (yellow) in IMCD3 cells acutely stressed tohypertonic conditions (550 mOsm/kgH2O, 24 h). A Z-scan view depicts that thecolocalization is restricted to the lateral membrane of these cells.

Fig. 6. MUPP1 expression is necessary for correct expression and localizationof Cldn4 at the membrane of IMCD3 cells under hypertonic stress. (A) MUPP1-silenced cells (Right) induced a significant decrease in Cldn4 protein expres-sion (59.2%, P 0.01) compared with empty-vector control cells (Left). Otherproteins of the tight junctions such as ZO-1 were not affected. Data representthe mean SEM from three Western blot analyses performed with duplicateindependent samples (80 g of total protein per lane, n 6). A representativeWestern blot including the -actin loading control is shown. (B) Cldn4 (green)is mislocalized in IMCD3 cells silenced for MUPP1 expression (Lower) comparedwith IMCD3 cells that express MUPP1 (Upper).

Lanaspa et al. PNAS October 14, 2008 vol. 105 no. 41 15799

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

17, 2

021

colocalized to other organelles including the Golgi apparatus(anti-GS15) or early endosomes (anti-Rab5) (data not shown).

MUPP1/Cldn4 Confers a Tight Epithelia Phenotype in IMCD3 CellsUnder Hypertonic Stress. To determine whether the absence ofCldn4 at the tight junctions of IMCD3 under hypertonic stressaffects the functionality of these epithelia, MUPP1-silenced cellswere analyzed for TER and compared with control cells. Wereported an association between MUPP1 expression and TER,indicating that MUPP1 was necessary for high TER in IMCD3 cells(6). In similar experiments, monolayers were analyzed for themaximum value of TER over time. The maximum TERs formonolayers from MUPP1-silenced cells adapted to 550 mOsm/kgH2O compared with control cells occurred at day 6, and dataidentified a 28.5 3.4% (P 0.01) reduction in monolayer TERcompared with control values (Fig. 8A). Despite restoration ofCldn4 expression, treatment of the monolayers with bafilomycin A1did not restore TER to empty-vector control values because theprotein is mislocalized to the lysosome in the absence of MUPP1.This finding strongly suggests that MUPP1 expression is requiredfor correct localization of Cldn4 to the tight junctions (Fig. 7B). Todetermine better whether the difference in TER values was causedby an absence of Cldn4 at the tight junctions, we stably silenced theexpression of Cldn4 in IMCD3 cells (pSM2-Cldn4; Open Biosys-tems). In Fig. 8B, a representative Western blot is shown, demon-strating two different clones in which Cldn4 expression was partially(clone 1) and totally (clone 2) silenced compared with empty-vectorcontrols. As shown in Fig. 8C, a 12 0.8% (P 0.01) reduction inTER was observed in the partially silenced clone 1. In contrast, a24 2.1% (P 0.001) reduction in TER was observed in the totallysilenced clone 2 compared with empty-vector control cells. Thesedata are similar to results obtained from MUPP1-silenced cells andtherefore strongly suggest that the reduction in TER with MUPP1-silenced cells is caused by the absence of Cldn4 at the tight junctions.

DiscussionChanges in the cellular genome and proteome are required for cellsof the renal inner medulla to survive and adapt to extreme changesin hypertonicity. A large body of information already exists forchanges in proteins that are involved in organic osmolyte accumu-lation; ion transport, and DNA repair (1, 2, 4, 19, 20). However, itis expected that a larger set of genes is involved that have wideranging effects of critical importance. We have reported (6) theup-regulation of the tight junction scaffolding protein, MUPP1,under hypertonic stress in IMCD3 cells and in the inner medulla ofthe kidney. MUPP1 up-regulation under hypertonic conditionsultimately leads to the maintenance of a tighter epithelia monolayeras demonstrated by an increase in monolayer TER. In this work, we

Fig. 7. Cldn4 is targeted for lysosomal degradation in MUPP1-silenced cells.(A) Western blot analysis of Cldn4 protein in empty-vector and MUPP1-silenced cells in the presence and absence of the lysosomal inhibitor bafilo-mycin A1 (BafA1). Cldn4 expression is significantly reduced in MUPP1-silencedcells compared with empty-vector controls under hypertonic stress. The Cldn4protein level is nearly complete restored (86.2 2.4%) in MUPP1-silenced cellstreated with bafilomycin A1. Data represent the mean SEM from threeWestern blot analyses performed with duplicate independent samples (80 gof total protein per lane, n 6). A representative Western blot including the-actin loading control is shown. (B) Cldn4 (green) colocalizes with the lyso-somal marker LAMP-1 (red) in IMCD3 cells silenced for MUPP1 expression.(Bottom) Colocalization is shown in yellow.

pSM2-EVpSM2-MUPP1pSM2-MUPP1

+ BafA1

Maximum TER value at day 6 after seeding on filters

A

0

250

500

750

1000

Tran

sEpi

thel

ialR

esis

tanc

e(Ω

*cm

2 )

0

1000

2000

3000

4000

5000

6000

7000

Wes

tern

blo

t den

sito

met

ry(a

rbitr

ary

units

)

pSM2-EV

300

#1 #2pSM2-Cldn4

550 mOsm/kg, 24h

B

Claudin-4

β-actin

25 kDa47 kDa

Clones

C

1 2 3 4 5 6350

450

550

650

750pSM2-EV

pSM2-Cldn4#2pSM2-Cldn4#1

Days

Tran

sEpi

thel

ialR

esis

tanc

e(Ω

*cm

2 )

Fig. 8. MUPP1 expression is required for tight epithelial properties in IMCD3 cellsunderhypertonic stress. (A)ComparisonofTER inMUPP1-silencedandempty-vectorcontrolmonolayers.Cells thatdonotexpressMUPP1exhibitasubstantialdecreaseinTER values (28.5 3.4%, P 0.01) at day 6 after seeding on filters compared withempty-vector control cells.Nochanges inTERwereobserved inMUPP1-silencedcellstreated with bafilomycin A1 (BafA1) compared with nontreated cells. (B) Represen-tativeWesternblotshowingpartiallyandtotally silencedIMCD3clones (clones1and2) for Cldn4 expression. (C) TER values obtained from empty-vector and clones 1 and2 during the 6 days after seeding the clones.

15800 www.pnas.orgcgidoi10.1073pnas.0805761105 Lanaspa et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

17, 2

021

describe the up-regulation under hypertonic stress of another tightjunction protein, Cldn4, a member of the claudin superfamily.Previous reports indicate that overexpression of Cldn4 in epithelialcells (Madin–Darby canine kidney II and Lilly Laboratories Celland Porcine Kidney 1) increased monolayer TER with a concom-itant decrease in the permeability for sodium (21, 22). This workdescribes the up-regulation of Cldn4 under hypertonic stress. Genechip analysis for IMCD3 cells exposed to increasing levels oftonicity demonstrated a 5-fold increase in Cldn4 mRNA levels. Incontrast, mRNA levels of Cldn2, a well known claudin associatedwith ‘‘leaky’’ epithelia (23, 24), was substantially down-regulated,indicating that these cells coordinate a change in phenotype forenhanced tight junction integrity under hypertonic stress. Changesin Cldn4 mRNA levels were further verified for protein expressionby Western blotting. Cldn4 protein expression increased rapidlyupon exposure of cells to an acute sublethal osmotic stress in a timeframe similar to early response proteins including MUPP1. Underisotonic conditions, IMCD3 cells were found to express very lowlevels of Cldn4 that increased substantially during acute or chronicexposure to hypertonicity (6 and 11, respectively, P 0.01). Theexpression of Cldn4 protein in response to hypertonicity was alsodemonstrated in kidney tissues of mice with substantial Cldn4protein levels in the hypertonic papilla tissues and nearly absent inthe isotonic cortex, corroborating previous reports showing Cldn4presence in the kidney restricted mostly to collecting duct cells (18).Thirsting mice and thereby further increasing kidney tissue tonicityresulted in an additional increase in the expression of Cldn4 inpapilla tissues of mice. Confocal immunofluorescence in IMCD3cells identified localization of Cldn4 protein by Z-scan analysis atthe upper level of the basolateral membrane, consistent with thisprotein involvement in tight junction complexes. Because Cldn4and MUPP1 are up-regulated under hypertonic stress in a verysimilar time frame, we examined a potential mechanism by whichthey interact as a consequence of the osmotic stress response.Knowing that the C terminus of most claudins including Cldn4binds to PDZ proteins such as ZO-1, -2, -3, and MUPP1, weevaluated coprecipitation of both proteins from IMCD3 cellschronically adapted to hypertonicity. After cross-linking the inter-actions with DSP, we could coprecipitate Cldn4 and MUPP1,indicating that these proteins do indeed interact in IMCD3 cells.Also, confocal analysis revealed that both proteins colocalized atthe tight junctions of these cells under hypertonic stress. Severalprevious studies reported that PDZ proteins including MUPP1 mayact as scaffolding and transport proteins for correct targeting,expression, and localization of claudins in the tight junctions (16,25). In this regard, inappropriate localization of claudins has beendescribed to induce several diseases and pathologies, includinghypomagnesemia hypercalciuria with nephrocalcinosis, in which amutation in the claudin-16 gene leads to disruption of its PDZ-binding site, abolishing its interaction with ZO-1 and mistargetingto the lysosome (26, 27). A more dramatic example is that viraloncoproteins such as the adenovirus E4-ORF protein are able tosequester tight junction PDZ proteins (MUPP1, MAGI-1, ZO-2) inthe cytoplasm of epithelial cells, avoiding their localization at thetight junctions and ultimately disrupting the tight junction integritycausing polarity defects. These cellular changes enhance the tu-morigenic potential of these viruses (28, 29). A similar effect ofsequestering MUPP1 in the cytosol and tight junction disruptionhas been described for the high-risk human papillomavirus type 18E6 oncoprotein, thus suggesting that MUPP1 is a key target foroncogenic virus and important in assembling tight junctions (29–31). To determine whether MUPP1 is important in the expressionand proper localization of Cldn4, we silenced MUPP1 by employinga stable siRNA vector. In these cells, we have found that Cldn4protein expression is substantially blunted by Western blottingunder hypertonic stress. Experiments also revealed that Cldn4protein expression in MUPP1-silenced cells under hypertonic stresscould be restored by inhibition of lysosome activity using bafilo-

mycin A1. Furthermore, Cldn4 protein was found to colocalize withthe lysosomal marker LAMP1 in bafilomycin A1-treated cells.These data thus strongly suggest that MUPP1 is necessary for thecorrect expression and localization of Cldn4 at the tight junction inkidney cells under hypertonic stress. Conversely, without MUPP1protein expression, Cldn4 is misrouted to lysosomes for degrada-tion. Functionally, MUPP1 silencing leads to a reduction in TER inIMCD3 cell monolayers under hypertonic stress as we described(6). Interestingly, treatment of MUPP1-silenced cells with bafilo-mycin A1 under hypertonic stress essentially restored overall Cldn4protein expression but not proper localization because localizationto the tight junction is necessary for the generation of TER. Thistreatment also failed to restore monolayer TER. As indicated, lossof monolayer TER is likely to be caused by decreased localizationof Cldn4 at the tight junctions because silencing Cldn4 expressionin IMCD3 cells led to a similar reduction in TER values comparedwith empty-vector control cells. This disruption of tight junctioncomplexes at the inner medulla of the kidney may result in a lossof the permeability characteristics of this usually tight epithelium.

In conclusion, this report identifies Cldn4 as an osmotic responseprotein in IMCD3 cells and the papilla of mouse kidneys. We alsodemonstrate the critical role of MUPP1 to target Cldn4 to the tightjunction. Our data suggest that Cldn4/MUPP1 up-regulation playsa critical role in maintaining a tight epithelial phenotype in the innermedulla under hypertonic stress.

Experimental ProceduresMaterials. Cell culturemedium,FCS,andantibioticswerefromGIBCO.Antibodiesto Cldn4, ZO-1, Rab5, and the 1 subunit of Na/K-ATPase were from Santa CruzBiotechnology. Antibodies to MUPP1 and GS15 were from Clontech. Antibody toLAMP1 was from Abcam, and antibodies to -actin were from Cell Signaling. Allother chemicals, including bafilomycin A1, were from Sigma.

Cell Culture. The established murine IMCD3 cell line originally developed byRauchman (32) was provided by Steve Gullans (Rx Gen, Hamden, CT). Cellstocks were maintained as described in ref. 33. IMCD3 clones silenced forMUPP1, Cldn4, and empty-vector control cells were developed as described inref. 6. In experiments involving hypertonic stress, the media in culture disheswere exchanged for that with added NaCl to the specified osmolality depend-ing on the experiment. Osmolality was determined with a microosmometer(model 3300; Advanced Instruments).

Gene Arrays. The mouse gene array 430-2.0 from Affymetrix contains 39,000transcripts. RNA transcription to cDNA, biotinylation to cRNA, fragmentation,hybridization to the chip, and chip analysis with a Hewlett–Packard gene arrayscanner were performed according to the manufacturer’s recommendationsby the University of Colorado Gene Array Core. Gene array data were analyzedby using both the Affymetrix Microarry Suite 5.0 and GeneSpring (SiliconGenetics) software analysis programs.

TER Measurements. TER measurements on cell monolayers were performedafter seeding cells on Millicel-PCF filters (PIHP01250; Millipore) as described inref. 6. An EVOM voltohmmeter (World Precision Instruments) was used for TERmeasurements. Monolayer TER was measured over a 6-day period or untilmaximum measurement was determined.

Mouse Kidney Tissues. C57BL/6 mice were obtained from Jackson Laboratories.Mice were subjected to food and ad libitum water or thirsted for 36 h. Mice wereharvested by cervical dislocation; urine samples were collected from the bladder forosmolality analysis, and kidneys were removed and papilla and cortex tissues dis-sected and snap frozen in liquid nitrogen. Tissues were homogenized with a glasstissue grinder on ice with lysis buffer and were analyzed as described in ref. 34.

RNA Extraction, Analysis, and Message Quantification. Cytosolic RNA wasisolated from confluent cultures by using the RNeasy kit (Qiagen). RNAintegrity was assessed by capillary electrophoresis by using a 2100 Bioanalyzer[using the 28 S to 18 S rRNA ratio (Agilent)]. RNA was converted to cDNA byusing the Omniscript reverse transcriptase kit (Qiagen) as described by themanufacturer. Quantitative PCR primers specific to Cldn4 were designed byusing Beacon Designer 5.0 software (Premier Biosoft International). Quanti-tative PCR was performed by using 5-CCACTCTGTCCACATTGC-3 forwardprimer and 5-TGCCTTCAGCCCATATCC-3 reverse primer (70 nM each) with a

Lanaspa et al. PNAS October 14, 2008 vol. 105 no. 41 15801

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

17, 2

021

SYBR Green master mix (JumpStart Taq Readymix; Sigma) on a Bio-Rad I-Cy-cler. Quantitative PCR runs were analyzed by agarose gel electrophoresis andmelt curve to verify that the correct amplicon was produced. -Actin RNA wasused as internal control, and the amount of RNA was calculated by thecomparative CT method as recommended by the manufacturer.

CoIP and Cross-Linking. Before performing CoIP, cells were treated for 30 min atroom temperature with 2 mM DSP (Pierce) to stabilize protein interactions. Thecross-linking reaction was stopped with 20 mM Tris (pH 7.5) for 15 min, and cellswere resuspended in ice-cold RIPA buffer (Pierce) and disrupted by repeatedpassage through a 21-gauge needle. Cellular debris was pelleted by centrifuga-tion, and soluble proteins obtained from the supernatant were precleared byincubation with 25% protein A–agarose slurry in PBS at 4°C (Roche). Sampleswere centrifuged at low rpm, and 500 g of the supernatant was incubatedovernight at 4°C with 1 g of rabbit polyclonal anti-Cldn4 antibody. Unboundproteins were collected for analysis (SN1), and immunoprecipitated proteinswere obtained by adding 20 l of protein A–agarose beads and a series ofwashes/centrifugations with PBS (SN2). After the final wash, the pellet wasresuspended in 2 sample buffer containing 2-mercaptoethanol to cleave cross-linked interactions and boiled just before electrophoresis.

Protein Extraction and Western Blotting. Cell protein lysates were preparedfrom confluent cell cultures as described in ref. 34. Sample protein content wasdetermined by the BCA protein assay (Pierce). Eighty micrograms of totalprotein was loaded per lane for SDS/PAGE (12.5% wt/vol) analysis and thentransferred to PVDF membranes. Membranes were incubated with primaryantibodies and visualized by using a horseradish peroxidase (HRP) secondary

antibody (Cell Signaling) and the HRP Immunstar detection kit (Bio-Rad).Chemiluminescence was recorded with an Image Station 440CF, and resultswere analyzed with the 1D Image Software (Kodak Digital Science).

Confocal Fluorescence Microscopy. IMCD3 cells were grown to confluence on8-well glass slides (177402; NUNC) and fixed with methanol at 20°C for 2 min.Cells were then permeabilized with 0.3% Triton X-100 in PBS and incubatedovernight with Cldn4, MUPP1, LAMP1, 1 Na/K-ATPase, and/or ZO-1 antibodies.Cells were rinsed and incubated with Alexa-Fluor 488 or 568 conjugated second-ary antibodies (Molecular Probes). Samples were covered with an antifadingmounting medium (Vector Laboratories). Preparations were imaged with a 40water-immersion objective by using a laser scanning confocal microscope (modelLSM510; Zeiss). Data were analyzed by using the LSM Image Analyzer postacqui-sitionsoftware(Zeiss). Immunofluorescenceanalysiswasperformedat least threetimes (separate plates) and by evaluation of 10 random fields.

Statistics and Data Analysis. All data are presented as the mean SEM. Datagraphics and statistical analysis were performed by using Instat (version 3.0)and Prism 4 (both from GraphPad Software). Data were analyzed by normalitytests and by the Tukey–Kramer multiple comparison test. Multiple groupcorrections were performed by using the method of Bartlett. In most cases,experiments were performed three times with independent replicates. Totaldata points (n) are identified in figure legends. P values 0.05 were recog-nized as statistically significant.

ACKNOWLEDGMENTS. This work was supported by National Institutes ofHealth Grants DK-19928 and DK-66544 (to T.B.).

1. Burg MB, Kwon ED, Kultz D (1997) Regulation of gene expression by hypertonicity.Annu Rev Physiol 59:437–455.

2. Ferraris JD, Burg MB (2006) Tonicity-dependent regulation of osmoprotective genes inmammalian cells. Contrib Nephrol 152:125–141.

3. Handler JS, Kwon HM (2001) Cell and molecular biology of organic osmolyte accumu-lation in hypertonic renal cells. Nephron 87:106–110.

4. Handler JS, Kwon HM (1996) Regulation of the myo-inositol and betaine cotransport-ers by tonicity. Kidney Int 49:1682–1683.

5. Tian W, Cohen DM (2002) Urea stress is more akin to EGF exposure than to hypertonicstress in renal medullary cells. Am J Physiol 283:F388–F398.

6. Lanaspa MA, et al. (2007) The tight junction protein, MUPP1, is up-regulated byhypertonicity and is important in the osmotic stress response in kidney cells. Proc NatlAcad Sci USA 104:13672–13677.

7. Becamel C, et al. (2001) Interaction of serotonin 5-hydroxytryptamine type 2C receptorswith PDZ10 of the multi-PDZ domain protein MUPP1. J Biol Chem 276:12974–12982.

8. Guillaume JL, et al. (2008) The PDZ protein MUPP1 promotes Gi coupling and signallingof the MT1 melatonin receptor. J Biol Chem 283:16762–16771.

9. Hamazaki Y, Itoh M, Sasaki H, Furuse M, Tsukita S (2002) Multi-PDZ domain protein 1(MUPP1) is concentrated at tight junctions through its possible interaction withclaudin-1 and junctional adhesion molecule. J Biol Chem 277:455–461.

10. Heydecke D, et al. (2006) The multi-PDZ domain protein MUPP1 as a putative scaffold-ing protein for organizing signaling complexes in the acrosome of mammalian sper-matozoa. J Androl 27:390–404.

11. Jeansonne B, Lu Q, Goodenough DA, Chen YH (2003) Claudin-8 interacts with multi-PDZ domain protein 1 (MUPP1) and reduces paracellular conductance in epithelial cells.Cell Mol Biol 49:13–21.

12. Krapivinsky G, Medina I, Krapivinsky L, Gapon S, Clapham DE (2004) SynGAP–MUPP1–CaMKII synaptic complexes regulate p38 MAP kinase activity and NMDA receptor-dependent synaptic AMPA receptor potentiation. Neuron 43:563–574.

13. Niessen CM (2007) Tight junctions/adherens junctions: Basic structure and function.J Invest Dermatol 127:2525–2532.

14. Schneeberger EE, Lynch RD (2004) The tight junction: A multifunctional complex. Am JPhysiol 286:C1213–C1228.

15. Turksen K, Troy TC (2004) Barriers built on claudins. J Cell Sci 117:2435–2447.16. Van Itallie CM, Anderson JM (2006) Claudins and epithelial paracellular transport.

Annu Rev Physiol 68:403–429.17. Balkovetz DF (2006) Claudins at the gate: Determinants of renal epithelial tight

junction paracellular permeability. Am J Physiol 290:F572–F579.18. Kiuchi-Saishin Y, et al. (2002) Differential expression patterns of claudins, tight junc-

tion membrane proteins, in mouse nephron segments. J Am Soc Nephrol 13:875–886.19. Miyakawa H, Woo SK, Dahl SC, Handler JS, Kwon HM (1999) Tonicity-responsive

enhancer-binding protein, a rel-like protein that stimulates transcription in responseto hypertonicity. Proc Natl Acad Sci USA 96:2538–2542.

20. Woo SK, Lee SD, Na KY, Park WK, Kwon HM (2002) TonEBP/NFAT5 stimulates tran-scription of HSP70 in response to hypertonicity. Mol Cell Biol 22:5753–5760.

21. Van Itallie C, Rahner C, Anderson JM (2001) Regulated expression of claudin-4 de-creases paracellular conductance through a selective decrease in sodium permeability.J Clin Invest 107:1319–1327.

22. Van Itallie CM, Fanning AS, Anderson JM (2003) Reversal of charge selectivity in cationor anion-selective epithelial lines by expression of different claudins. Am J Physiol285:F1078–F1084.

23. Enck AH, Berger UV, Yu AS (2001) Claudin-2 is selectively expressed in proximalnephron in mouse kidney. Am J Physiol 281:F966–F974.

24. Furuse M, Furuse K, Sasaki H, Tsukita S (2001) Conversion of zonulae occludentes fromtight to leaky strand type by introducing claudin-2 into Madin–Darby canine kidney Icells. J Cell Biol 153:263–272.

25. Gonzalez-Mariscal L, Betanzos A, Nava P, Jaramillo BE (2003) Tight junction proteins.Prog Biophys Mol Biol 81:1–44.

26. Muller D, et al. (2003) A novel claudin-16 mutation associated with childhood hyper-calciuria abolishes binding to ZO-1 and results in lysosomal mistargeting. Am J HumGenet 73:1293–1301.

27. Muller D, Kausalya PJ, Meij IC, Hunziker W (2006) Familial hypomagnesemia withhypercalciuria and nephrocalcinosis: Blocking endocytosis restores surface expressionof a novel claudin-16 mutant that lacks the entire C-terminal cytosolic tail. Hum MolGenet 15:1049–1058.

28. Latorre IJ, et al. (2005) Viral oncoprotein-induced mislocalization of select PDZ proteinsdisrupts tight junctions and causes polarity defects in epithelial cells. J Cell Sci118:4283–4293.

29. Lee SS, Glaunsinger B, Mantovani F, Banks L, Javier RT (2000) Multi-PDZ domain proteinMUPP1 is a cellular target for both adenovirus E4-ORF1 and high-risk papillomavirustype 18 E6 oncoproteins. J Virol 74:9680–9693.

30. Lee C, Laimins LA (2004) Role of the PDZ domain-binding motif of the oncoprotein E6in the pathogenesis of human papillomavirus type 31. J Virol 78:12366–12377.

31. Massimi P, Gammoh N, Thomas M, Banks L (2004) HPV E6 specifically targets differentcellular pools of its PDZ domain-containing tumour suppressor substrates for protea-some-mediated degradation. Oncogene 23:8033–8039.

32. Rauchman MI, Nigam SK, Delpire E, Gullans SR (1993) An osmotically tolerant innermedullary collecting duct cell line from an SV40 transgenic mouse. Am J Physiol265:F416–F424.

33. Capasso JM, Rivard CJ, Enomoto LM, Berl T (2003) Adaptation of murine inner med-ullary collecting duct (IMCD3) cell cultures to hypertonicity. Ann NY Acad Sci 986:410–415.

34. Capasso JM, Rivard CJ, Berl T (2001) Long-term adaptation of renal cells to hyperto-nicity: Role of MAP kinases and Na-K-ATPase. Am J Physiol 280:F768–F776.

15802 www.pnas.orgcgidoi10.1073pnas.0805761105 Lanaspa et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

17, 2

021