BT
RF
Wa*
BcmhrbpiemtabosmoFvsapmttwtccsimot
HbtspA(
BA
SIC–LIV
ER,
PA
NCREA
S,A
ND
BILIA
RY
TRA
CT
GASTROENTEROLOGY 2006;131:878 – 884
ASIC–LIVER, PANCREAS, AND BILIARYRACT
adixin Is Required to Maintain Apical Canalicular Membrane Structure andunction in Rat Hepatocytes
EI WANG,* CAROL J. SOROKA,* ALBERT MENNONE,* CHRISTOPH RAHNER,‡ KATHY HARRY,* MARC PYPAERT,‡
nd JAMES L. BOYER*
Liver Center and the ‡Department of Cell Biology, Yale University School of Medicine, New Haven, ConnecticuttdeDs
thatfmttrcbmqdbitBcl
bd(cs
scemtOb
ackground & Aims: Ezrin-radixin-moesin proteins areross-linkers between the plasma membrane and actin fila-ents. Radixin, the dominant ezrin-radixin-moesin protein in
epatocytes, has been reported to selectively tether multidrug-esistance–associated protein 2 to the apical canalicular mem-rane. However, it remains to be determined if this is itsrimary function. Methods: An adenovirus-mediated short
nterfering RNA (siRNA) was used to down-regulate radixinxpression in collagen sandwich– cultured rat hepatocytes andorphologic and functional changes were characterized quan-
itatively. Results: In control cultures, an extensive bile can-licular network developed with properly localized apical andasolateral transporters that provided for functional excretionf fluorescent cholephiles into the bile canalicular lumina.iRNA-induced suppression of radixin was associated with a
arked reduction in the canalicular membrane structure asbserved by differential interference contrast microscopy and-actin staining, in contrast to control cells exposed to adeno-irus encoding scrambled siRNA. Indirect immunofluorescencehowed that apical transporters (multidrug-resistance–associ-ted protein 2, bile salt export pump, and multidrug-resistancerotein 1) dissociated from their normal location at the apicalembrane and were found largely associated with Rab11-con-
aining endosomes. Localization of the basolateral membraneransporter, organic anion transporting polypeptide 2 (Oatp2),as not affected. Consistent with this dislocation of apical
ransporters, the biliary excretion of glutathione-methylfluores-ein and cholylglycylamido-fluorescein was decreased signifi-antly in the radixin-deficient cells, but not in the controliRNA cells. Conclusions: Radixin is essential for maintain-ng the polarized targeting and/or retaining of canalicular
embrane transporters and is a critical determinant of theverall structure and function of the apical membrane of hepa-ocytes.
epatocytes are highly polarized epithelial cells whose api-cal canalicular domain is designed for the production of
ile. This secretory process depends on a group of membraneransporters at this apical pole that are members of the ABCuperfamily of export pumps. These include the bile salt exportump (Bsep, Abcb11), the multidrug-resistance protein (Mdr1,bcb1), and the multidrug-resistance–associated protein 2
Mrp2, Abcc2) among others. Under normal physiologic condi-
ions, the transport of bile salts into bile generates bile salt–ependent bile flow whereas bile salt–independent flow is gen-rated in large part by the excretion of glutathione via Mrp2.isorders that impair these transport proteins result in chole-
tatic liver injury.1,2
Although the maintenance of secretory polarity of the hepa-ocyte is critical for its normal function, little is known aboutow these cells establish and maintain this functionally distinctpical domain.3 The ERM (ezrin-radixin-moesin) family of pro-eins plays an important role in regulating the structure andunction of specific domains of the cell cortex by cross-linking
embrane and actin cytoskeletons.4 The dominant ERM pro-ein in hepatocytes is radixin,5 which is localized primarily athe canalicular membrane of hepatocytes.5,6 At 4 weeks of ageadixin knock-out mice show a selective loss of Mrp2 from theanalicular membrane and begin to develop conjugated hyper-ilirubinemia, reminiscent of Dubin–Johnson syndrome in hu-an beings.7 These findings suggest that radixin may be re-
uired for the tethering of Mrp2 to the apical canalicularomain. Radixin also is reduced and associated with redistri-ution of MRP2 within intracellular structures of hepatocytes
n patients with primary biliary cirrhosis.8 However, in contrasto radixin-deficient mice, P-glycoproteins (MDR1, MDR3, andSEP) also are redistributed to intracellular structures andolocalize with MRP2 in these patients with chronic cholestaticiver disease.
To clarify the role of radixin in the canalicular localization ofile transporters and the integrity of the apical canalicularomain, we used adenovirus-mediated short interfering RNAs
siRNAs) to suppress radixin expression in collagen sandwich–ultured rat hepatocytes. This culture method has been de-cribed previously9,10 and sustains the expression of hepatocyte-
Abbreviations used in this paper: Ad-siRadixin, adenovirus-mediatedhort interfering RNA for radixin; Bsep, bile salt export pump; CGamF,holylglycylamido-fluorescein; CMFDA, 5-chloromethylfluorescein diac-tate; ERM, ezrin-radixin-moesin; MOI, multiplicity of infection; Mrp2,ultidrug-resistance–associated protein 2; Mrp3, multidrug-resis-
ance–associated protein 3; Mdr1, multidrug-resistance protein 1;atp2, organic anion transporting polypeptide 2; PBS, phosphate-uffered saline; siRDX, siRadixin; siRNA, short interfering RNA.© 2006 by the American Gastroenterological Association (AGA) Institute
0016-5085/06/$32.00
doi:10.1053/j.gastro.2006.06.013
sscsaiactl
BtFfrC(SCLi(apMsdi
t1cAra
pwWm11iss
8sf
apc
tnEci
cwpcpflft
ahrdtm(O
1g7mwa
w0aNbtwwIwEflt
w
BA
SIC–L
IVER
,PA
NCREA
S,A
ND
BIL
IARY
TRA
CT
September 2006 RADIXIN MAINTAINS BILE CANALICULI 879
pecific proteins and the maintenance of bile canaliculartructure and function. Our studies show that radixin defi-iency results in a profound reduction in canalicular membranetructures and a dissociation of bile transporters from thepical canalicular membrane. This in turn leads to a functionalmpairment in the canalicular excretion of substrates for Mrp2nd Bsep. These results provide clear evidence that radixin is aritical requirement not just for the tethering of Mrp2 but forhe normal maintenance of the canalicular membrane and theocalization and function of its transport proteins.
Materials and MethodsReagentsBD Adeno-X Expression Systems 2 was purchased from
D Biosciences (Bedford, MA). Alexa-conjugated secondary an-ibodies, TO-PRO 3, 5-chloromethylfluorescein diacetate (CM-DA), and Alexa 594 – conjugated phalloidin were purchasedrom Molecular Probes (Eugene, OR). Cholylglycylamido-fluo-escein (CGamF) was a gift from Alan Hofmann (San Diego,A). The following antibodies were used: mouse anti-Mrp2
Alexis Biochemicals, San Diego, CA), rabbit anti-radixin (Cellignaling Technology, Beverly, MA), goat anti-radixin (Santaruz Biotechnology, Santa Cruz, CA), mouse anti-MDR (Signetaboratories, Dedham, MA), rabbit anti-Bsep (Kamiya Biomed-
cal Co, Seattle, WA), mouse anti-ZO-1 and rabbit anti-Rab11Zymed Laboratories, San Francisco, CA), and mouse anti-�-ctin (Sigma, St. Louis, MO). Rabbit anti-Oatp2 was kindlyrovided by Bruno Stieger (Zurich, Switzerland). Rabbit anti-rp3 was raised in our laboratory. Western blotting showed its
pecificity for Mrp3 as opposed to Mrp2 by its presence in Mrp2eficient (transport minus, TR�) rat liver and its up-regulation
n cholestasis.
Construction of Recombinant siRNA-Expressing AdenovirusThe siRNA sequences target rat radixin gene at posi-
ions 106 –125 (si-radixin [siRDX]A), 627– 646 (siRDXB), 1011–030 (siRDXC), and 1219 –1238 (siRDXD), relative to the startodon. A scrambled sequence (5=-GAC TCC GAA CAT GTACG T-3=) was used for the control siRNA (siControl). The
ecombinant adenovirus encoding siRNA was constructed andmplified according to the instructions of BD Biosciences.
Cell CultureHepatocytes were isolated from rat liver by collagenase
erfusion as described previously from this laboratory.11 Cellsere cultured on collagen-coated dishes or glass coverslips inilliams E medium with the addition of 5% fetal calf serum, 10mol/L HEPES buffer, 2 mmol/L L-glutamine, 1.8 g/L glucose,�mol/L dexamethasone, 4 mg/L insulin, 100 U/mL penicillin,
00 �g/mL streptomycin, and 10 mg/L gentamicin. Cells werenfected with adenovirus twice: at 1 hour and at 16 hours aftereeding and were overlaid with gelled collagen 24 hours aftereeding.
ImmunoblotCells were lysed in RIPA buffer (50 mmol/L Tris, pH
.0, 150 mmol/L NaCl, 1% NP-40, 0.5% deoxycholate, 0.1%odium dodecyl sulfate, and protease inhibitors) and centri-
uged at 1000g for 10 minutes. The supernatants were fraction- vted on sodium dodecyl sulfate–polyacrylamide gel electro-horesis and analyzed by Western blot and enhancedhemiluminescence method.
Light MicroscopyDifferential interference contrast microscopy was used
o quantitatively assess the length of the bile canalicular chan-el using the software Openlab 3.1.5 (Improvision, Coventry,ngland). Images were obtained randomly from 10 fields peroverslip (objective, 20�) and analyzed from 4 separate exper-ments.
Separate coverslips were fixed in 4% paraformaldehyde or inold methanol for immunofluorescence. Primary antibodiesere diluted in blocking buffer (1% bovine serum albumin inhosphate-buffered saline [PBS]– 0.05% Triton X-100) and in-ubated on cell layers for 2 hours at room temperature. Appro-riate secondary antibodies were incubated for 1 hour anduorescence labeling was observed with a Zeiss LSM 510 con-
ocal microscope (Thornwood, NJ) using appropriate filters andhe same gain levels for all conditions.
Electron MicroscopyAfter 5 days of culture, coverslips were coded (by W.W.)
nd hepatocyte monolayers were fixed with 2% paraformalde-yde/2.5% glutaraldehyde and postfixed with 1% osmium tet-oxide. The cultures were block-stained with 2% uranyl acetate,ehydrated in acetone series, and epon embedded. Thin sec-ions were stained with lead citrate and uranyl acetate. Electron
icrographs were acquired at 14,000 � primary magnificationby C.R.) on a Philips 410 electron microscope (FEI, Hillsboro,
R) and evaluated blindly by J.L.B. and A.M.
CMFDA and CGamF Excretion by RatHepatocytesHepatocytes were incubated with 5 �mol/L CMFDA or
�mol/L CGamF diluted in HEPES buffer (0.35 g/L KCl, 0.25/L MgSO4, 0.18 g/L CaCl2, 0.16 g/L KH2PO4, 4.8 g/L HEPES,.9 g/L NaCl, and 0.9 g/L glucose, pH 7.4) at 37°C for 15inutes. Cultures were imaged immediately (for CGamF) orere incubated for an additional 15 minutes before being im-ged (for CMFDA).
For quantitative assessment, hepatocytes were pretreatedith HEPES buffer or Ca2�-free HEPES buffer (0.35 g/L KCl,.16 g/L KH2PO4, 4.8 g/L HEPES, 7.9 g/L NaCl, 0.9 g/L glucose,nd 10 mmol/L ethylene glycol-bis(�-aminoethyl ether)-,N,N=,N=-tetraacetic acid, pH 7.4) at 37°C for 10 minutesefore incubation with 2 �mol/L CMFDA for 10 minutes. Cellshen were incubated in HEPES buffer at 37°C for 30 minutes,ashed, and lysed in 1% Triton X-100 in PBS. The cell lysatesere measured by a Synergy HT microplate reader (Bio-Tek
nstruments, Winooski, VT) (�ex � 485 nm, �em � 530 nm,here ex means excitation and em means emission). The Biliaryxcretion Index � (fluorescence Standard- fluorescence Ca2�-free)/uorescence Standard � 100%, where the fluorescence representshe fluorescence units per mg of cell lysates.12
Data AnalysisData represent the mean � SD. Statistical significance
as tested with a paired t test, assuming significance with a P
alue of less than .05.
pcmastlcsA
rdrfclbat
tLPs2
FchwPao
FecoTsastttbbl
BA
SIC–LIV
ER,
PA
NCREA
S,A
ND
BILIA
RY
TRA
CT
880 WANG ET AL GASTROENTEROLOGY Vol. 131, No. 3
ResultsAdenovirus-mediated siRNA EffectivelyInhibits Radixin Expression in Sandwich-Cultured Rat HepatocytesTo address the requirement for radixin in normal he-
atocyte function, we attempted to inhibit radixin expression inollagen sandwich– cultured rat hepatocytes using adenovirus-ediated siRNA. We first tested the 4 siRNA target sequences innormal rat liver cell line, clone 9, to evaluate their ability to
uppress radixin. Immunoblot analysis in Figure 1A indicateshat all 4 siRNAs individually and in combination reduced theevel of radixin expression to 19%–34% of Ad-siControl–treatedells. The most potent siRNA, siRDXC, was selected for theubsequent studies in rat hepatocytes and is referred to as
igure 1. Adenovirus-mediated siRNA effectively inhibited radixinxpression in collagen sandwich–cultured rat hepatocytes. (A) Clone 9ells were treated with PBS, Ad-siControl, or the 4 siRNAs individuallyr in various combinations at a virus dose of multiplicity of infection 10.he cells were harvested 3 days after treatment and cell lysates wereeparated by polyacrylamide gel electrophoresis, and blotted for radixinnd normalized to �-actin. (B) Rat hepatocytes treated with PBS, Ad-iControl, or Ad-siRDXC were harvested 1, 3, and 5 days after thereatment and cell lysates were blotted for radixin and �-actin. (C) Hepa-ocytes were treated with PBS, Ad-siControl, or Ad-siRDXC with mul-iplicity of infection 25 or 50, harvested on day 5, and cell lysates werelotted for radixin and �-actin. (D) The hepatocyte lysates also werelotted for Mrp2, Bsep, Mrp3, and Mdr1. A representative result of at
east 3 experiments is shown.
d-siRadixin. Figure 1B shows that Ad-siRadixin progressively �
educed radixin expression and generated significant knock-own 5 days after the initial infection. This down-regulation ofadixin expression was dose-dependent and multiplicity of in-ection 50 resulted in reduction to 18% of Ad-siControl–treatedells (Figure 1C). No alteration was observed in the expressionevels of �-actin (Figure 1C), Mrp2, Bsep, Mdr1, and Mrp3 (aasolateral transporter) (Figure 1D). Thus, Ad-siRadixin wasble to potently and specifically suppress radixin expression inhis sandwich-cultured hepatocyte model.
Cell viability of 5-day cultures was examined using the lac-ate dehydrogenase assay (Thermo Electron, Louisville, CO).actate dehydrogenase activity in the media of 5-day cultures ofBS-, Ad-siControl–, and Ad-siRadixin–treated cells was notignificantly different (12.20 � 4.09, 13.48 � 2.89, and 16.40 �.30 U/L, respectively) (mean � SD of 5 experiments).
igure 2. Ad-siRadixin reduced canalicular structures in sandwich-ultured hepatocytes. (A) Differential interference contrast images ofepatocytes that were treated with PBS, Ad-siControl, or Ad-siRadixinere acquired on days 1, 3, and 5. Five-day cultures of cells treated withBS, Ad-siControl, or Ad-siRadixin were labeled with (B) anti-radixinntibody or (C) Alexa 594 phalloidin for actin filaments. Both show lossf apical canalicular structures after Ad-siRadixin treatment. Bar, 10
m.
antdfanBhcl(tgesccnwts2rahdopts
hpcvcbttesi
icltsoai(Z
sctea
FHctmtab
BA
SIC–L
IVER
,PA
NCREA
S,A
ND
BIL
IARY
TRA
CT
September 2006 RADIXIN MAINTAINS BILE CANALICULI 881
Ad-siRadixin Reduced the CanalicularStructures of Collagen Sandwich–CulturedHepatocytesTo assess the role of radixin in the structure of the
pical canalicular domain, the formation of the bile canalicularetwork was examined by differential interference contrast op-ics over a 5-day period after administration of the Ad-siRa-ixin. As shown in Figure 2A, 1 day after seeding, hepatocytesormed uniform monolayers and there was little or no sign ofpical domain formation. By day 3, a continuous, anastomoticetwork was established in all the cultures along cell borders.y day 5, the canalicular lumina assumed a more refined andomogeneous appearance in PBS- and Ad-siControl–treatedells. Quantitative assessment of the lengths of these canalicu-ar channels was 439 � 26 and 457 � 25 �m/field, respectivelymean � SD of 4 experiments). However, in striking contrast,he canalicular structures in the Ad-siRadixin–treated cells de-enerated by 56% to 195 � 25 �m/field (mean � SD of 4xperiments, P � .0001). In some regions, only a few canaliculartructures could be observed. Notably, this degeneration ofanalicular structures occurred in parallel over the same timeourse as the down-regulation of radixin as shown by immu-oblot (Figure 1B). Immunofluorescence showed that radixinas concentrated at the apical domains of PBS- or Ad-siCon-
rol–treated cells, whereas the signal was diminished in Ad-iRadixin–treated cells (Figure 2B). Actin localization (FigureC) in PBS- and Ad-siControl–treated cells displayed a normalailroad-track alignment of actin filaments bordering the can-licular membrane. Often the lumen appeared to be moreighly dilated in the Ad-siControl–treated cells. In radixin-eficient cells, actin staining appeared as single lines with littler no separation (Figure 2C). In some areas the pattern wasunctate and disrupted. Localization of the tight junction pro-eins, Zonula occludens-1 (ZO-1), and claudin 3 showed aimilar pattern of labeling (data not shown).
Ultrastructural analysis of PBS- and Ad-siControl–treatedepatocytes revealed the presence of dilated bile canaliculi asreviously described in collagen sandwich– cultured hepato-ytes (Figure 3A and B).10 The significant dilation of the sealedacuoles resulted in loss of microvilli, as commonly seen inholestasis. After Ad-siRadixin treatment the lumina of manyile canaliculi were largely collapsed (Figure 3C). Tight junc-ions appeared normal and the presence of microvilli suggestshat the collapse of the lumen may be a result of deficientxcretion of osmotically active bile components into theseealed spaces. Blind evaluation (J.L.B. and A.M.) consistentlydentified the Ad-siRadixin micrographs.
Ad-siRadixin Alters the Distribution of ApicalMembrane Transporters in HepatocytesLocalization of membrane transporters was evaluated
n 5-day cultures of hepatocytes by indirect immunofluores-ence. As shown in Figure 4, Mrp2, Bsep, and Mdr1 wereocalized primarily on apical membranes in PBS- and Ad-siCon-rol–treated cells. Remarkably, in Ad-siRadixin–treated cells,trong intracellular accumulation of the apical transporters wasbserved primarily in a perinuclear distribution and only inreas that lacked intact bile canaliculi. Residual stubby canal-culi also were labeled. All apical membrane proteins testedMrp2, Bsep, Mdr1, dipeptidyl peptidase IV), but not actin or
O1, were found in this intracellular compartment. In contrast, otaining for a basolateral transporter, Oatp2, was unchangedompared with control cells (Figure 4). These findings indicatehat radixin is selectively required for apical, but not basolat-ral, targeting/retaining of transport proteins in hepatocytesnd that radixin deficiency results in a generalized impairment
igure 3. Electron micrographs of sandwich-cultured hepatocytes.epatocytes were treated with PBS, Ad-siControl, or Ad-siRadixin andultured for 5 days. Hepatocytes treated with (A) PBS or (B) Ad-siCon-rol developed dilated bile canaliculi. Tight junctions are intact and fewicrovilli are observed. (C) The bile canalicular lumina of Ad-siRadixin–
reated cells were largely collapsed. The tight junctions remain intactnd the loss of severe dilation results in the appearance of microvilli. BC,ile canaliculi. Bar, 2 �m.
f the localization of apical membrane proteins.
pcpRRcpcrtmwrMt
d
rmCm(l
hbtbnarPs5Tn
rcaEbiw
FRb(cIbcimc
Fptw(msA
BA
SIC–LIV
ER,
PA
NCREA
S,A
ND
BILIA
RY
TRA
CT
882 WANG ET AL GASTROENTEROLOGY Vol. 131, No. 3
Intracellular Fraction of Mrp2 ColocalizesWith Rab11Because the intracellular localization of apical trans-
orters in radixin knock-down cells was reminiscent of Rab11-ontaining endosomes described for Bsep in WIF-B9 cells,13 weerformed double-labeling immunofluorescence for Mrp2 andab11 in the collagen-sandwiched hepatocytes. In control cells,ab11 was localized to a subapical compartment and a perinu-lear compartment that were both distinct from the Mrp2-ositive canaliculi (Figure 5). However, in Ad-siRadixin–treatedells Mrp2 largely colocalized with Rab11 in an expanded pe-inuclear compartment in the knock-down areas (as defined byhe lack of dilated canaliculi) (Figure 5). This perinuclear accu-
ulation of apical membrane transporters did not colocalizeith mannosidase II, a Golgi marker (data not shown). Thus,
adixin suppression appears to increase the accumulation ofrp2 in recycling endosomes coincident with its apical reduc-
ion.
Ad-siRadixin Decreases the Rate of BiliaryExcretion of Glutathione-Methylfluoresceinand CGamFThe excretory function of Mrp2 and Bsep in radixin-
igure 4. Ad-siRadixin altered the localization of apical bile trans-orters in hepatocytes. Hepatocytes were treated with PBS, Ad-siCon-rol, or Ad-siRadixin and cultured for 5 days. Cells were double-labeledith anti-Mrp2 and anti-Bsep antibodies (top 2 rows), with anti-Mdr1
third row), or anti-Oatp2 antibodies (bottom row). Note that all 3 apicalembrane transporters, but not the basolateral membrane protein,
howed significant intracellular localization in knock-down areas afterd-siRadixin treatment. Arrow shows collapsed canaliculi. Bar, 10 �m.
eficient hepatocytes was assessed using CMFDA and CGamF, (
espectively. In PBS- or Ad-siControl–treated cells, glutathione-ethylfluorescein (Mrp2 substrate derived from CMFDA) andGamF (Bsep substrate) were efficiently transported and accu-ulated in canalicular lumina with little retention in cytosol
Figure 6A). In contrast, both fluorescent substrates accumu-ated intracellularly in Ad-siRadixin–treated cells.
To quantitate these differences in fluorescent dye excretion,epatocytes were treated with standard or Ca2�-free HEPESuffer before the bile excretion assay. Ca2�-free treatment dis-urbed the integrity of the tight junction barrier surroundingile canaliculi and thus permitted the release of bile compo-ents as described.12 The Biliary Excretion Index was measureds the percentage of glutathione-methylfluorescein fluorescenceeleased into canalicular lumina. The Biliary Excretion Index ofBS- and Ad-siControl–treated cells was 39.2% and 48.6%, re-pectively (P � .05), whereas the value was reduced by nearly0% to 21.3% in Ad-siRadixin–treated cells (P � .05) (Figure 6B).hese results indicate that radixin is required to maintainormal bile excretory function in hepatocytes.
DiscussionThe principal finding in this study is that radixin is
equired for the maintenance of the structure of the apical bileanalicular membrane and the localization and function ofpical canalicular membrane transport proteins. When thisRM protein is suppressed in cell culture, the once-establishedile canalicular domain is disrupted markedly. This impairment
n the normal configuration of the apical canalicular membraneas confirmed by the altered distribution of actin filaments.
igure 5. The intracellular fraction of Mrp2 was colocalized withab11. In Ad-siControl cells Mrp2 (green) was localized to the dilatedile canaliculi. Occasional weak staining was seen intracellularly. Rab11
red) was found in a subapical and perinuclear compartment in theseells. The merged image shows the separation of these compartments.
n contrast, in Ad-siRadixin–treated cells, knock-down areas (as definedy the lack of intact, dilated bile canaliculi) revealed extensive perinu-lear colocalization of both Mrp2 and Rab11. Occasional, stubby canal-
culi were seen labeled for Mrp2 (arrow). Bar, 10 �m (top 2 rows). Highagnification of the knock-down area showed that Mrp2 was largely
olocalized with Rab11 in the intracellular compartment. Bar, 5 �m
bottom row).
Amilacitpsccdt
zsrDasscw
iaasdf
rapfMtsdmeupMgcwloitctslst
MredpbmbmctcaivWieweh
tmuo
FgwCcdBmrwP
BA
SIC–L
IVER
,PA
NCREA
S,A
ND
BIL
IARY
TRA
CT
September 2006 RADIXIN MAINTAINS BILE CANALICULI 883
ctin filaments have long been known to be required for theaintenance of the canalicular domain. Microfilament-perturb-
ng agents result in loss of the normal microvilli in the cana-icular membrane.14 Radixin, a cross-linker of actin filamentsnd plasma membrane, is a known component of the hepato-ellular microvilli.6 Loss of microvilli in radixin-deficient mice7
s consistent with our study and indicates that radixin is essen-ial for maintaining normal hepatic canalicular membrane mor-hology. The specificity of our findings for radixin is empha-ized by the lack of significant disturbance of the bileanalicular structure and secretory function in the PBS- andontrol siRNA–treated cells and by the absence of significantifferences in lactate dehydrogenase released into the media inhese cell culture systems.
The requirement of ERM proteins for normal apical organi-ation appears to be a common property in epithelial cells. Theingle ERM protein in Drosophila, Dmoesin, is enriched in thehabdomere base of the photoreceptor. Down-regulation ofmoesin profoundly disrupts the cortical cytoskeleton and
pical membrane organization.15 Radixin-deficient mice alsouffer from deafness because of the loss of stereocilia, thepecially developed giant microvilli in cochlear sensory hairells.16 Ezrin is critical for the assembly of the apical terminal
igure 6. Ad-siRadixin decreased the rate of biliary excretion oflutathione-methylfluorescein and CGamF. Hepatocytes were treatedith PBS, Ad-siControl, or Ad-siRadixin and cultured for 5 days. (A)ells were incubated with CMFDA or CGamF and visualized by confo-al microscopy. Bar, 10 �m. (B) Hepatocytes were treated with stan-ard or Ca2�-free HEPES buffer before the bile excretion assay. Theiliary Excretion Index was measured as the percentage of glutathione-ethylfluorescein fluorescence released into canalicular lumina. Data
epresent the means � SD of 5 experiments. P � .05, PBS–comparedith Ad-siControl–treated cells. *P � .05, Ad-siRadixin–compared withBS- or Ad-siControl–treated cells.
eb region of developing mouse intestine, which is particularly b
mportant for the propagation of polarity during the formationnd expansion of secondary lumina.17 Ezrin also is expressedbundantly at the actin cortical layers of parietal cells in mousetomach.18 Ezrin knock-down mice suffer from severe achlorhy-ria. The loss of gastric acid secretion is caused by defects in theormation/expansion of gastric canalicular apical membranes.
A second important finding in the present study is thatadixin is required not only for the localization of Mrp2 to thepical membrane, but also for other canalicular membraneroteins including Bsep and Mdr1. When Mrp2 is retrievedrom the plasma membrane after radixin suppression, Bsep and
dr1 also relocate to the intracellular pool. Consistent withhis morphologic finding, the excretion of both Mrp2 and Bsepubstrates into the bile canalicular lumen is reduced by theeficiency of radixin. Phalloidin-induced changes in actin fila-ent organization also alter the localization of canalicular
xport pumps,19 supporting the importance of normal canalic-lar organization for the polarized distribution of apical trans-orters. Our findings also help explain the observation thatDR1 colocalizes with abnormally redistributed MRP2 in re-
ions where radixin is largely reduced in livers of primary biliaryirrhosis patients.8 In cholestasis it is yet to be determinedhether a loss of radixin precedes the loss of microvilli, cana-
icular lumen dilation, and dissociation of apical transporters,r whether these all occur simultaneously. However, it is of
nterest to note that in radixin knock-out mice, these choles-atic changes also occurred in older animals.7 Perhaps otherompensatory factors are present in younger mice, preventinghe demonstration of cholestasis. Nevertheless, the currenttudy emphasizes the critical importance of radixin in theocalization and function of canalicular transporters andtrongly suggests that deficient radixin function must be part ofhe pathophysiology of cholestasis.
Our finding that the increased intracellular fraction ofrp2 colocalizes with Rab11 guanosine triphosphatase, a
ecycling endosome marker, suggests that Mrp2 undergoesndocytosis on the disruption of canaliculi induced by ra-ixin suppression. Because the redistributed apical trans-orters do not colocalize with Golgi markers, they must note produced solely by de novo protein synthesis. Instead, it isost likely that they are retrieved from the canalicular mem-
rane. Apical membrane recycling involving Rab11 endoso-al compartments has been described with Bsep in WIF-B9
ells.13 Mrp2, as well as Bsep and Mdr1, all have been iden-ified in intracellular vesicular structures in addition to theanalicular membrane in rat liver.20,21 Furthermore, Mrp2nd Bsep are regulated rapidly by retrieval from and/ornsertion into the canalicular membrane via intracellularesicles.21–25 In support of our finding, a recent study in
IF-B9 cells colocalized apical ABC transporters and Rab11n polarized cells and restricted these transporters to Rab11ndosomes in nonpolarized cells where no canalicular poleas formed.26 It remains to be determined if the Rab11
ndosome mediates an Mrp2 trafficking pathway in normalepatocytes in vivo.
In summary, the present study indicates that down-regula-ion of radixin in sandwich-cultured hepatocytes by adenovirus-
ediated siRNA disorganizes the structure of the bile canalic-lus, which leads to impairment in bile excretory functionwing to both a reduction in the total amount of apical mem-
rane and retrieval of bile acid and organic solute transporters
fTnhRgtcasce
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
YSj
BA
SIC–LIV
ER,
PA
NCREA
S,A
ND
BILIA
RY
TRA
CT
884 WANG ET AL GASTROENTEROLOGY Vol. 131, No. 3
rom the canalicular membrane to intracellular compartments.hese observations indicate that radixin is a critical determi-ant of the structure and function of the apical membrane ofepatocytes and is not just a tethering protein for Mrp2.ather, radixin is also essential for maintaining polarized tar-eting and/or retaining of other critical bile export pumps onhe canalicular membrane. These findings prompt revision ofurrent concepts regarding the role of radixin in the hepatocytend provide new insights into the molecular events in cholesta-is. Further studies on the alteration of radixin’s function inholestasis should broaden our understanding of the pathogen-sis of cholestatic syndromes.
References
1. Trauner M, Boyer JL. Cholestatic syndromes. Curr Opin Gastro-enterol 2004;20:220–230.
2. Pauli-Magnus C, Meier PJ. Hepatocellular transporters and cho-lestasis. J Clin Gastroenterol 2005;39:S103–S110.
3. Wang L, Boyer JL. The maintenance and generation of membranepolarity in hepatocytes. Hepatology 2004;39:892–899.
4. Bretscher A, Edwards K, Fehon R. ERM proteins and merlin:integrators at the cell cortex. Nat Rev 2002;3:586–599.
5. Fouassier L, Duan CY, Feranchak AP, Yun C, Sutherland E, SimonF, Fitz JG, Doctor B. Ezrin-radixin-moesin-binding phosphoprotein50 is expressed at the apical membrane of rat liver epithelia.Hepatology 2001;33:166–176.
6. Amieva MR, Wilgenbus KK, Furthmayr H. Radixin is a componentof hepatocyte microvilli in situ. Exp Cell Res 1994;210:140–144.
7. Kikuchi S, Hata M, Fukumoto K, Yamane Y, Matsui T, Tamura A,Yonemura S, Yamagishi H, Keppler D, Tsukita S, Tsukita S. Radixindeficiency causes conjugated hyperbilirubinemia with loss of Mrp2from bile canalicular membranes. Nat Genet 2002;31:320–325.
8. Kojima H, Nies AT, König J, Hagmann W, Spring H, Uemura M,Fukui H, Keppler D. Changes in the expression and localization ofhepatocellular transporters and radixin in primary biliary cirrhosis.J Hepatol 2003;39:693–702.
9. Chandra P, Lecluyse EL, Brouwer KL. Optimization of cultureconditions for determining hepatobiliary disposition of tauro-cholate in sandwich-cultured rat hepatocytes. In Vitro Cell DevBiol Anim 2001;37:380–385.
0. Talamini MA, Kappus B, Hubbard A. Repolarization of hepato-cytes in culture. Hepatology 1997;25:167–172.
1. Boyer JL, Phillips JM, Graf J. Preparation and specific applicationsof isolated hepatocyte couplets. Methods Enzymol 1990;192:501–516.
2. Liu X, Chism JP, LeCluyse EL, Brouwer KR, Brouwer KLR. Corre-lation of biliary excretion in sandwich-cultured rat hepatocytesand in vivo in rats. Drug Metab Dispos 1999;27:637–644.
3. Wakabayashi Y, Lippincott-Schwartz J, Arias IM. Intracellular traf-ficking of bile salt export pump (ABCB11) in polarized hepaticcells: constitutive cycling between the canalicular membrane andrab11-positive endosomes. Mol Biol Cell 2004;15:3485–3496.
4. Elias EZ, Hruban JB, Boyer JL. Phalloidin-induced cholestasis: amicrofilament-mediated change in junctional complex permeabil-
ity. Proc Natl Acad Sci U S A 1980;77:2229–2233. P5. Karagiosis SA, Ready DF. Moesin contributes an essential struc-tural role in Drosophila photoreceptor morphogenesis. Develop-ment 2004;131:725–732.
6. Kitajiri S, Fukumoto K, Hata M, Sasaki H, Katsuno T, NakagawaT, Ito J, Tsukita S, Tsukita S. Radixin deficiency causes deafnessassociated with progressive degeneration of cochlear stereocilia.J Cell Biol 2004;166:559–570.
7. Saotome I, Curto M, McClatchey AI. Ezrin is essential for epithe-lial organization and villus morphogenesis in the developing in-testine. Dev Cell 2004;6:855–864.
8. Tamura A, Kikuchi S, Hata M, Katsuno T, Matsui T, Hayashi H,Suzuki Y, Noda T, Tsukita S, Tsukita S. Achlorhydria by ezrinknockdown: defects in the formation/expansion of apical canal-iculi in gastric parietal cells. J Cell Biol 2005;169:21–28.
9. Rost D, Kartenbeck J, Keppler D. Changes in the localization ofthe rat canalicular conjugate export pump mrp2 in phalloidin-induced cholestasis. Hepatology 1999;29:814–821.
0. Soroka CJ, Pate MK, Boyer JL. Canalicular export pumps trafficwith polymeric immunoglobulin A receptor on the same microtu-bule-associated vesicle in rat liver. J Biol Chem 1999;274:26416–26424.
1. Kipp H, Pichetshote N, Arias IM. Transporters on demand. Intra-hepatic pools of canalicular ATP binding cassette transporters inrat liver. J Biol Chem 2001;276:7218–7224.
2. Schmitt M, Kubitz R, Lizun S, Wettstein M, Häussinger D. Regu-lation of the dynamic localization of the rat Bsep gene-encodedbile salt export pump by anisoosmolarity. Hepatology 2001;33:509–518.
3. Beuers U, Bilzer M, Chittattu A, Kullak-Ublick G, Keppler D,Paumgartner G, Dombrowski F. Tauroursodeoxycholic acid in-serts the apical conjugate export pump, Mrp2, into canalicularmembranes and stimulates organic anion secretion by proteinkinase C-dependent mechanisms in cholestatic rat liver. Hepa-tology 2001;33:1206–1216.
4. Misra S, Varticovski L, Arias IM. Mechanisms by which cAMPincreases bile acid secretion in rat liver and canalicular mem-brane vesicles. Am J Physiol 2003;285:G316–G324.
5. Mottino AD, Crocenzi FA, Pozzi EJS, Veggi LM, Roma MG, Vore M.Role of microtubules in estradiol-17�-D-glucuronide-induced al-teration of canalicular Mrp2 localization and activity. Am J Physiol2005;288:G327–G336.
6. Wakabayashi Y, Dutt P, Lippincott-Schwartz J, Arias IM. Rab11aand myosin Vb are required for bile canalicular formation inWIF-B9 cells. Proc Natl Acad Sci U S A 2005;102:15087–15092.
Received October 17, 2005. Accepted June 2, 2006.Address requests for reprints to: James L. Boyer, MD, Liver Center,
ale University School of Medicine, PO Box 208019, 333 Cedartreet, 1080 LMP, New Haven, Connecticut 06520-8019. e-mail:
[email protected]; fax: (203) 785-7273.W.W. and C.J.S. contributed equally to this work.Supported by National Institutes of Health grants DK 25636 and
30-34989 (J.L.B.).
