Annexin VI defines an apical endocytic compartment in rat liver hepatocytes

261Journal of Cell Science 111, 261-269 (1998)Printed in Great Britain © The Company of Biologists Limited 1998JCS9656

Annexin VI defines an apical endocytic compartment in rat liver hepatocytes

David Ortega 1,*, Albert Pol 1,*, Michael Biermer 2, Stefan Jäckle 2 and Carlos Enrich 1,†

1Departamento de Biología Celular, Facultad de Medicina, Universidad de Barcelona, Casanova 143, 08036-Barcelona, Spain2Medizinische Kernklinik und Poliklinik, Universitäts-Krankenhaus Eppendorf, Martinistrasse 52, D-20246 Hamburg, Germany*These two authors contributed equally to this work†Author for correspondence (e-mail: [email protected])

Accepted 5 November 1997: published on WWW 23 December 1997

Annexin VI has been demonstrated previously to be amarker for hepatic endosomes. By western blotting with anaffinity purified anti-annexin VI antibody it was shown thatannexin VI was present in the three morphologically andfunctionally different endosomal fractions from rat liver.We have quantified the gold-labeled endosomes byimmunoelectron microscopy in ultrathin Lowicryl sectionsof rat liver and now demonstrate that 80% of the totallabeling with anti-annexin VI was associated withendocytic structures surrounding the bile canaliculus, theapical domain of hepatocytes, whereas only 20% was foundin the subsinusoidal endosomes. In double immuno-goldlabeling experiments 80% of the Rab5 positive apical

endosomes were also labeled with anti-annexin VIantibodies. However, there was no significant colocalizationwith antibodies to the polymeric immunoglobulin receptor.Finally, we demonstrate that 50% of endosomes containinginternalized gold-labeled transferrin were double labeledwith anti-annexin VI antibodies. Thus, annexin VI becomesthe first known structural protein at the apical ‘early’endocytic compartment of the hepatocyte that may beinvolved in the receptor recycling and transport to lateendocytic/lysosomal compartment pathways.

Key words: Rat liver endosome, Annexin VI, Intracellulartrafficking, Immunocytochemistry

SUMMARY

ed

ind

grjo oflyb).orat

yticerg al.,

teder, VI

ndinticntheere

INTRODUCTION

The hepatocyte is a highly polarized cell with a plasmmembrane organized in discrete functional domains. Tfunctional polarity is originated and maintained by sortinevents in the biosynthetic (trans-Golgi network) and in theendocytic compartments (Drubin and Nelson, 1996). Tdistinct endosomal compartments associated with differplasma membrane domains, basolateral and apendosomes, have been identified in hepatocytes (Limet et1985; Hoppe et al., 1985; Enrich and Evans, 1989; Mulloet al., 1994; Hemery et al., 1996) in MDCK cells (Bomset al., 1989; Parton et al., 1989) or other epithelial ce(Wilson et al., 1987; Speelman et al., 1995; Knight et a1995). The function and distribution of these endosomcompartments is controversial. Basolateral endosomes involved in the uptake and recycling of receptors and liganinvolved in cell maintenance (‘housekeeping endosomeKelly, 1993); in contrast, apical endosomes were thoughbe involved in epithelial cell-type specific processes suchtranscytosis and therefore were specialized for epithecells (Barroso and Sztul, 1994). However, several recstudies suggest that no specialized apical endosocompartment exists (Apodaca, 1994) and in additioindicate that portions of the apical endosomal compartmare unique in composition and function (Hughson aHopkins, 1990; Knight et al., 1995).

In a previous work, a protein of 68 kDa was isolated a

ahisg

woentical al.,ckelllsl.,alaredss’;

t to aslialentmaln,entnd

nd

identified as annexin VI in highly purified endosomes isolatfrom rat liver (Jäckle et al., 1994).

The annexins are cytoskeleton-associated proteins that breversibly to phospholipids at micromolar calciumconcentrations (calcium dependent-phopholipid-bindinproteins; Burgoyne and Geisow, 1989; Moss, 1992; Swaiand Seaton, 1994). They constitute a significant proportiontotal cellular protein of about 1% and have been highconserved throughout evolution (Crompton et al., 1988a,Several studies report a variety of cellular functions fannexins (Moss, 1995), and it is now becoming clear thannexins are also involved in several steps of the endocpathway (Lin et al., 1992; Harder and Gerke, 1993; Gruenband Emmans, 1993; Burgoyne and Clague, 1994; Futter et1993).

The study of intracellular localization of annexin VI byimmunofluoresecence and electron microscopy was attempin rat liver (Tagoe et al., 1994; Weinman et al., 1994); howevsuch studies did not reveal any specific location for annexinin a particular organelle or intracellular structure.

In this research, using immunoelectron microscopy amorphometric quantitation we demonstrate that the malocation of annexin VI in the hepatocyte is the apical endocycompartment. However, this apical endocytic compartmedoes not seem to be involved in the transcytosis of pIgA. Trole of annexin VI in integrated intracellular trafficking and thpossible functions in this apical endocytic compartment adiscussed.

262

s.

7he%

d to of

ith

Mheren

ithperthen

ed

tin atintedre inforal,eld

alnedad

dle-ivetherywas

e, inheen,ed

ateendr

o-y.ighter2%ere

D. Ortega and others

MATERIALS AND METHODS

Reagents Human LDL (low density lipoprotein) (1.025<d<1.050 g/ml) waisolated from blood of normolipidemic human adults by sequencentrifugation and purified by recentrifugation at the upper denslimit (Havel et al., 1955).

Antibodies Rab5A (S-19), affinity purified polyclonal antibody raised againstsynthetic peptide corresponding to amino acids 193-211 mappwithin the carboxy terminal domain of Rab5A of human origin, wfrom Santa Cruz Biotechnology, Inc. A monoclonal antibody agaithe polymeric immunoglobulin receptor (SC-166) was kindprovided by Dr K. E. Mostov (University of California, SanFrancisco). A polyclonal antibody to annexin IV was kindly provideby Dr Volker Gerke (University of Münster, Germany).

Isolation of subcellular fractions Three endosomal fractions, early endosomes (designated CUcompartment of uncoupling of receptors and ligands), late endoso(MVB, multivesicular bodies) and a receptor recycling compartme(RRC, receptor recycling compartment) were isolated from estradtreated rats (Chao et al., 1979) according to the method describeBelcher et al. (1987) and Jäckle et al. (1991). After three days ofα-ethinyl estradiol treatment rats were anaesthetized with dietether, and human LDL (Havel et al., 1955) (5 mg of protein) winjected into the femoral vein. The livers were removed 15 minulater and homogenized in 0.25 M sucrose with protease inhibitormM PMSF, 100 mM leupeptin, 150 mM aprotinin, 1 mM pepstatinThe homogenate was centrifuged for 10 minutes at 2,000 rpm uan SW28 rotor (Beckman) (600 gav). The supernatant was centrifugeat 5,100 rpm for 20 minutes in an SW28 rotor (Beckman) (3,500 gav).Finally, the supernatant was centrifuged at 11,900 rpm for 20 minuin a Type 30 rotor (Beckman) (12,200 gav). 64 ml of the lastsupernatant fraction was diluted with 30 ml of Percoll solution (27 Percoll and 3 ml 2.5 M sucrose at pH 7.4), layered in tubes econtaining a marker density of 1.062 g/ml (Pharmacia, Sweden), centrifuged for 45 minutes at 17,600 rpm in a Type 30 ro(Beckman) (26,700 gav). The fraction over the marker density wacollected, two volumes of 0.9% NaCl were added, and then it wlayered onto 2 ml of 2.5 M sucrose and centrifuged at 11,600 rpm45 minutes in an SW28 rotor (Beckman) (17,800 gav). The whitecrude endosome band was harvested and 2.5 M sucrose was adda final density of 1.15 g/ml. The 1.15 g/ml portion was loaded at bottom of tubes containing a discontinuous sucrose gradient of 1.1.074, 1.11 and 1.139 g/ml densities. The tubes were centrifuged170 minutes at 28,000 rpm in an SW28 rotor (Beckman) (104,0gav). Three distinct populations were obtained: MVB at 1.032/1.07CURL at 1.074/1.110; and RRC at the 1.110/1.130 interfaces. Efraction was collected and ice-cold water was added to make thisotonic. Finally, the isotonic fractions were pelleted by centrifugiat 20,800 rpm for 30 minutes in a Ti 50 rotor (Beckman) (52,000 gav),resuspended in 0.9% NaCl and stored at −80°C.

Preparation of affinity purified annexin VI antibody For the generation of polyclonal antibodies to the 68 kDa protidentified as annexin VI (Jäckle et al., 1994), approximately 1.0 of protein from CURL, MVB and RRC endosome fractions waseparated by SDS-PAGE under non-reducing conditions atransferred to a nitrocellulose filter. After light staining with Poncered the 68 kDa band, identified by its position relative to an interannexin VI marker, was excised. As a modification of the methodOlmsted (1981), the strip of nitrocellulose was rinsed in blockibuffer (PBS containing 3% bovine serum albumin and 0.02% sodiazide) and incubated overnight at 4°C in rabbit serum (diluted

stiality

aing

asnstly

d

RL,mesntiol-d by 17-hylastess (1).

singd

tes

mlachandtorsas

for

ed tothe032, for004;achem

ng

einmgsnd

aunal ofngum in

blocking buffer 1:100) raised against intact endosomal fractionAffinity purified antibodies were eluted with 0.2 M glycine, 1 mMEGTA, pH 2.8, for 20 minutes and immediately neutralized with 0.0volume of 1 M Tris-base. Antibodies were stored at 4°C after taddition of 0.1 volume of 10-fold concentrated PBS and 0.02sodium azide.

Coupling of transferrin with colloidal-gold particles The procedure described by Slot and Geuze (1985) was useprepare human holo-transferrin-colloidal gold complexes of 17 nmdiameter.

Immunoelectron microscopy Livers of young adult Sprague-Dawley rats were perfused in situ w50 ml of PBS containing 2 mM CaCl2 and 2 mM MgCl2 and thenwith 100 ml of 2% paraformaldehyde, 0.1% glutaraldehyde in 0.1 phosphate buffer, pH 7.2, containing 3% polyvinylpyrolidone (Enricand Evans, 1989). Liver pieces were placed in cold fixative, furthdissected and left in fixative overnight at 4°C. The tissue was thwashed in PBS and incubated in 0.5 M NH4Cl in PBS for 60 minutesat 4°C. After washing in PBS the tissue was dehydrated wincreasing concentrations of ethanol in distilled water (30 minutes change) on ice up to 75% ethanol, transferred to 95% ethanol and to absolute ethanol, both changes at −20°C for 2 hours each.Infiltration with resin was with Lowicryl K4M:ethanol (1:2 and then2:1 at −20°C for 1 hour each) and then with daily changes of undilutLowicryl for 1 week at −20°C.

For polymerization, each piece of liver was transferred to a gelacapsule containing Lowicryl and left under u.v. (360 nm) overnight−35°C and for a further 48 hours at room temperature. Ultrathsections were cut using a Reichert Ultracut E microtome and collecon Formvar-coated gold grids. For annexin VI localization, grids weincubated for 1 hour at room temperature with primary antibodiesPBS containing 1% BSA. After washing, the grids were incubated 1 hour at room temperature with Protein A conjugated to colloidgold particles in PBS containing 1% BSA, 0.075% Triton X-1000.075% Tween-20. The double immunolabelling involved thsequential application of two separate antibody-Protein A-goprocedures with intermediate blocking, by fixation, of potentiimmunoreactive sites. After several washes, sections were staiwith saturated ethanolic uranyl acetate, counterstained with lecitrate and examined in an Hitachi HT-600 electron microscope.

As control for single immunostaining, incubation with the seconantibodies (Protein A-gold) only was accomplished. For doublabeling, controls using only one primary antibody and the respectsecond antibodies were performed in order to establish that colocalization was not due to recognition of the same primaantibody. In both cases, the labeling was specific, as no signal obtained (data not shown).

In some experiments rats were treated with gadolinium chloriddissolved in 0.9% NaCl (1 mg per ml) and injected intravenouslya dose of 5 mg per kg body weight, for 24 hours before texperiments begun (Renaud et al., 1989). Rats were thintravenously injected with transferrin-colloidal gold, 17 nm gold(300 mg in 0.9% NaCl) and 5 or 15 minutes later, livers were perfuswith 2% paraformaldehyde, 0.5% glutaraldehyde in 0.1 M cacodylbuffer, pH 7.4. Small pieces of livers were kept overnight in thfixative, then washed and equilibrated in 2.3 M sucrose in buffer afinally frozen in liquid nitrogen. Thin cryosections were used foimmunolabelling with anti-annexin VI antibody and Protein A-10 nmgold was used for the detection.

Finally, isolated endosomes were analyzed by cryimmunoelectron microscopy with the anti-annexin VI antibodSamples, pellets of the three endosomal fractions were fixed overnwith 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4. Aftwashing in the same buffer, the fractions were embbeded in gelatin-phosphate buffer for 10 minutes at 37°C. Samples w

263Annexin VI in apical hepatic endosomes

bytedein,

l,

ediol ofeirtnsIhed91,dll0

icses.ntoot

etted

es

in

cryoprotected by overnight incubation in 2.1 M sucrose contain3% polyvinylpyrolidine and then frozen in liquid N2.

Polyacrylamide gel electrophoresis, western blotting andimmunoprecipitationElectrophoresis of endosomal proteins was carried out in 1polyacrylamide gels as described by Laemmli (1970). Polypeptiwere transferred electrophoretically to nitrocellulose filter paper aprobed with affinity purified antibody (at 1:200 in 1% BSA-PBSAntigens were identified by the ECL system (AmershamImmunoprecipitation was carried out as previously described (Enand Evans, 1989). Protein content was measured according tomethod described by Bradford (1976) using bovine serum albuminstandard. The quantification of annexin VI in the western blots wcarried out by densitometry using a Bio-Image system (Millipore)

RESULTS

Characterization of affinity purified anti-annexin VIantibody: western blotting, immunoprecipitation andimmunocytochemistry on isolated endosomesRat liver endosomes were isolated according to the proceddeveloped by Belcher et al. (1987) and Jäckle et al. (1991) an antibody to annexin VI was prepared and affinity purifias described in Materials and Methods. As shown in Fig. 1,affinity purified anti-annexin VI antibody clearly recognized68 kDa polypeptide in the three rat liver endosome fractionsin a crude endosome fraction. Although characterization ofanti-annexin VI antibody was carried out in our previous wo(Jäckle et al., 1994), now an additional panel of evidenceshown for the affinity purified antibody used in this work.

First of all, it is shown that the treatment of endosomes w3 mM EGTA leads to a complete removal of annexin VI frothe endosomal membranes (Fig. 1a). In addition, annexinwas immunoprecipitated with anti-annexin VI affinity purifieantibody from the supernatant after the EGTA treatment (F1b). Annexin VI, used for the affinity purification, isolated belectroelution (of an excised 68 kDa band of crude endosofraction) was probed against anti-annexin VI and against aannexin IV (kindly provided by Dr V. Gerke, University oMünster, Germany); Fig. 1c shows that only anti-annexin recognises the electroeluted annexin VI. Finally, Fig. 1d, sho

Fig. 1. Characterization of affinity purified anti-annexin VI antibody in rat liver endosomefractions. Endosomal fractions (2 µg protein perlane): MVB, multivesicular bodies; CURL,compartment of uncoupling receptors and ligands;RRC, receptor recycling compartment or crudeendosomes (1 to 7, 40 µg protein per lane) wereisolated from livers of estradiol-treated rats.(a) Endosomal fractions were treated with orwithout 3 mM EGTA by centrifugation and thepellets analysed by western blotting. (b) Lane 1, animmunoblot using the affinity purified anti-annexinVI antiserum of the immunoprecipitate obtainedwhen the supernatant of EGTA treated endosomeswas treated with an anti-annexin VI antibodies.Lane 2, the Coomassie blue stained supernatantafter the EGTA treatment. (c) Purified annexin VI,by electroelution of 68 kDa excised band was probed with anti-anrat albumin (lane 6) were probed with anti-annexin VI; (lane 7), crreducing conditions). In b, c and d the lanes are from separate ge

ing

0%desnd).).

rich the asas

.

ureanded the a or anrk is

ithm VIdig.ymenti-fVIws

that anti-annexin VI does not recognise rat serum albumin western blotting; in fact, when crude endosomes are subjecto SDS-PAGE under non-reducing conditions, it can bobserved that anti-annexin VI does not label the rat albumwhich runs slightly below annexin VI in this gel, labeled withanti-rat serum albumin antibody.

Annexin VI is present in all three isolated endosomafractions but quantification indicates (CURL (57%); RRC(26%); MVB (17%)) an enrichment in the ‘early’ and recyclingcompartments.

Finally, since it has been demonstrated that isolatendosomes retain their in situ properties and that estradtreatment does not affect their morphology, size, enrichmentinternalized ligands, vesicle content, marker enzymes or thlipid and protein composition (Hornick et al., 1984; Chao eal., 1981; Jäckle et al., 1991) we have used isolated fractioto examine the pattern of labeling with anti-annexin Vantibody. Furthermore, it has been shown previously that tconcentration of annexin VI was identical in estradiol-treateand in non-treated isolated endosomes (Jäckle et al., 191994). Cryo-immuno electron microscopy of isolateendosomal fractions (CURL, MVB and RRC) showed that aindividual endocytic structures were decorated with gold (1nm) after incubation with the affinity purified anti-annexin VIantibody and Protein A-gold (Fig. 2). In RRC and CURLfractions most of the labeling was located in the cytoplasmsite of the membrane, but in MVB most of the labeling waassociated within the internal membrane vesicular structurWhen endosomal fractions were submitted to EGTA treatmeand then fixed and immunolabeled with anti-annexin VI, nsignificant labeling to the endosomes was detected (nshown).

Localization of annexin VI in rat liver:immunoelectron microscopyExamination of frozen sections of rat liver treated with thaffinity purified anti-annexin VI antibody and a fluorescensecond antibody showed that fluorescence was concentrapredominantly in the canalicular (apical) region of hepatocyt(data not shown).

However, when thin sections of rat liver embedded

nexin VI (lane 3) or with anti-annexin IV (lane 4); (d) annexin VI (lane 5) andude endosomes were probed with anti-rat serum albumin. (under non-ls.

264

s,he

s thees

neith

ndeas

aytic

ess,larot

dy’lysele-oal

rin-ere

tiched

7

o5icithntrin

ired.

D. Ortega and others

Fig. 2. Cryo-immuno electron microscopy of endosomal fractionswith anti-annexin VI antibody. The same endosomal fractions as inFig. 1 were prepared for immunocytochemistry as described inMaterials and Methods and stained with the affinity-purified anti-annexin VI antibody. (a) CURL (0.3-0.35 µm); (b) MVB (0.4-0.5µm), arrows indicate internal membranes in this endosomal fraction;(c) RRC. Bar, 100 nm.

Lowicryl were stained sequentially with anti-annexin Vantibody and Protein A-gold and were examined under electron microscope (Fig. 3) it was demonstrated that distribution of gold particles in the hepatocyte was associamainly with tubulo-vesicular structures corresponding to tcomponents of the endocytic compartment. These structuwere found to be most abundant at the peri-canalicular areaa morphometric quantification see Table 1A). Very little or ngold labeling was found associated with other organel

Ithethetedheres

(foro

les

identified in the sections, such as mitochondria, nucleuendoplasmic reticulum or peroxisomes. Golgi stacks and tplasma membrane showed very little staining.

In the endocytic vesiculo-tubular structures, gold particlewere precisely located in the periphery, in the dense areas ofvesicular bodies (probably corresponding to internal membrannot clearly defined in the Lowicryl sections) or on the membraappendages. Rat liver Lowicryl sections were also stained wa monoclonal antibody to the polymeric immunoglobulinreceptor (SC-166), as a marker for the transcytotic pathway awith an anti-Rab5 affinity purified antibody, to ascertain th‘class’ of labeled endosomes. Whereas the SC-166 labeling wconcentrated in the microvilli of the sinusoidal plasmmembrane, on the lateral plasma membrane and in endocstructures, close to the apical pole, Rab5 labeled endosomwere distributed evenly in both endocytic compartmentbasolateral and apical, and in vesicular and vesiculo-tubuendosomes (single labeling with anti-SC166 and Rab5, nshown but quantification is in Table 1A).

Despite the predominant apical location, annexin VI labeleendocytic structures most probably correspond to ‘earl(apical) endosomes. In fact, Rab5, a marker for earendosomes, was found to colocalize with annexin VI in thoapical endocytic structures (Fig. 4). On the other hand, doublabeling using anti-annexin VI and anti-pIgR showed nsignificant colocalization in the subsinusoidal or in the apicendocytic compartment (Fig. 5 and Table 1B).

Apical annexin VI labeled endosomes containedinternalized transferrin In the current research, we also studied the transferintracellular pathway, following the uptake of transferrincolloidal gold complexes at different time points. To avoid thrapid uptake of those complexes by the Kupffer cells, rats wetreated with gadolinium chloride to suppress their phagocycapacity (Renaud et al., 1989). At 5 and 15 minutes of ttransferrin injection, livers were fixed by perfusion and preparefor cryo-immunocytochemistry. Cryosections of liverscontaining internalized transferrin-colloidal gold complexes (1nm) were then incubated with anti-annexin VI affinity purifiedantibody and Protein A-gold (10 nm). Fig. 6 shows twendocytic structures containing the transferrin-colloidal gold 1minutes after the transferrin injection; 50% of the endocytstructures containing transferrin-gold were double labeled wannexin VI in the apical compartment. However, no significacolocalization was observed in those livers where transferwas internalized for 5 minutes (not shown).

Intracellular distribution of gold-labeled endocyticstructures The different types of endocytic structures as well as theintracellular distribution in the subsinusoidal (basal) and thpericanalicular (apical) regions in the hepatocyte, was studie

265Annexin VI in apical hepatic endosomes

Table 1. Quantification of immunocytochemistry(A) Simple immunogold labeling quantification

Annexin VI pIgR Rab5

Basal structures Vesicles 8 21 17Tubulo-vesicular 10 10 22Tubules 2 10 10

Total structures 20 41 49

Apical structures Vesicles 20 36 19Tubulo-vesicular 55 10 23Tubules 5 13 9

Total structures 80 59 51

(B) Double immunogold labeling quantification

pIgR-annexin VI Rab5-annexin VI

Simple Simple Simple SimpleDouble annexin VI pIgR Double annexin VI Rab5

Apical structures Vesicles 13 18 18 4 27 2Tubulo-vesicles 10 19 4 32 19 5Tubules 2 12 4 2 7 2

Total structures 25 49 26 38 53 9

Numbers are the percentage of the total number of structures counted = 4211:2042 in the sinusoidal region and 2169 in the apical. Structures were consideredto be in the basolateral region when found within 2 µm of the base of the sinusoidal plasma membrane, and apical endocytic structures when found around thebile canalicular plasma membrane at approx. 2.5 µm. In double-labeled sections we counted the number of endocytic structures with one size of gold label andwith both sizes of gold label. Five livers were used and three Lowicryl blocks of each one. The electron micrographs analyzed have been obtained randomly witha minimum of six grids for each of the antibodies for single and double labeling. In addition, three different types of structures were considered: tubular, vesiculo-tubular (with at least one membranous extension) and vesicular.

Fig. 3. Lowicryl section micrograph inthe apical region of hepatocytes.Ultrathin Lowicryl section labeled withaffinity purified anti-annexin VIantibody. Labeling is concentratedmainly in tubulo-vesicular endocyticstructures (arrowheads). No stainingwas observed in mitochondria (M),peroxisomes (P), Golgi (g) or in themembrane of the canalicular domain(bc). ×34,000.

266

I,stIofedf

e

D. Ortega and others

Fig. 4. Lowicryl sections of rat liverdouble-labeling with anti-Rab5 andanti-annexin VI antibodies. Detail ofthe apical peri-canalicular regions ofthe hepatocyte showing the labeling withanti-Rab5 (15 nm, open triangles) andannexin VI (10 nm). bc, bile canaliculus;M, mitochondria. ×33,000 (a);×58,000 (b). bc

The most abundant endocytic structures were the vesic(58% basal vs 53% apical); the vesiculo-tubular and the tubstructures were similarly distributed (19% basal vs 24% apiand 23% basal vs 23% apical, respectively). Thus the tynumber and distribution of endocytic structures in these tendocytic compartments of the hepatocyte was very simila

Morphometric analysis of gold-labeled endocytic structur

ularularcalpe,wor.es

in these two regions using three antibodies, anti-annexin Vanti-pIgR and anti-Rab5 was carried out (Table 1A). The mosignificant result was the polarized distribution of annexin Vin the apical endocytic compartment of the hepatocyte: 20% the total labeling was detected in the basal region comparwith the 80% found in endosomes at the apical region. Othose, in the apical pole, 55% were vesiculo-tubular. Th

267Annexin VI in apical hepatic endosomes

Fig. 5. Lowicryl sectionof rat liver double-labeled with anti-pIgRand anti-annexin VI.Detail of the apical andthe lateral plasmamembrane regions of anhepatocyte showing thelabeling of endocyticstructures with anti-annexin VI (10 nm)(arrowheads) and withanti-pIgR (15 nm) (opentriangles). M,mitochondria. ×20,000.

Fig. 6. Cryosection of ratliver with 15 minutesinternalized transferrin-17nm gold double-labeledwith anti-annexin VI (10nm). ×71,000. For thequantification of sampleswhere transferrin-goldwas endocytosed, four ratswere used, two for 5minutes and two for 15minutes of internalization.Two samples of each ratwere used for cryoimmunoelectronmicroscopy studies. Atotal of 100 endosomescontaining transferrin-17nm colloidal gold werecounted, in samples from15 minutes. As mentionedin the text, 50% of thoseendosomes containingtransferrin were doublelabeled with annexin VI.N, nucleus.

268

caltot

ilegh

egire

shewndcalt

iseatlyegaled

5VIly’ntd

etat). y

al.,xinotps

n

sc

r,byft toA.

oft

e

D. Ortega and others

amount of labeling with anti-pIgR, associated with endocystructures in the basal and apical regions was found to be similar: 41% basal vs 59% apical. The labeling was detecmainly in vesicular structures. An equal distribution (51% 49%) of Rab5 positive endosomes in the two endocytic regiof the hepatocyte was observed. In this case, the labstructures were mainly vesiculo-tubular in both areas.

Finally, we also studied the double labeling of annexin with pIgR and Rab5 (Table 1B). Owing to the small amouof annexin VI in the subsinusoidal area we only present results corresponding to the apical region. There was significant colocalization of annexin VI and pIgR in the apicregion of hepatocytes. However, 80% of Rab5 labeendosomes were also labeled with anti-annexin VI.

We used the statistical χ2-test for the evaluation of data; wefound χ2=77.32 for annexin VI and Rab5 with P<0.025indicating that colocalization is significant. On the other hanχ2=4.61 for annexin VI and the polymeric IgA receptor, witP>0.025 indicating a non-significant colocalization of thetwo antigens.

DISCUSSION

Annexin VI is a marker for hepatic endosomes and in this stuwe demonstrate that it defines an apical endosomcompartment in the hepatocyte. Despite its specific locationthe apical endocytic compartment, it does not seem toinvolved in the transcytosis of pIgA.

The architecture of the endocytic compartment in epithecells is very complex. Vesiculo-tubular and vesicular structuare the main constituents. These structures undergo a dynfunction in receptor-mediated endocytosis and in the continrecycling of receptors back to the plasma membrane. They participate in the transcytosis of ligands and receptors betwthe two different surfaces: the sinusoidal (basal) acanalicular (apical).

The biochemical composition of different specializepopulations of endosomes is largely unresolved and very proteins have a clearcut location into the basal or to the apendosomes. Furthermore, it has been shown that some prosuch as the small GTP-binding protein Rab5 shares the eapical and basolateral endosomes (Bucci et al., 1994).

In this work, annexin VI has been immunolocalized endocytic structures at the apical pole of hepatocytes. In hepatocyte, it has been shown that this apical region, closthe Golgi and lysosomes, contains endocytic and transcytstructures, identified by different antibodies or by ultrastructufeatures (Hoppe et al., 1985; Enrich and Evans, 1989; Hemet al., 1996; Geuze et al., 1984; Jost-Vu et al., 1986).

In terms of functional role for this apical annexin VI hepatendocytic compartment, one can envisage that at least tdifferent intracellular pathways may converge in this endocycompartment: (i) transcytotic vesicles, on their way fsecretion into the bile (for example pIgA); (ii) the transport the ‘late’ compartment and lysosomes for degradation; and recycling receptors on their way back to the sinusoidal plasmembrane (for example the asialoglycoprotein receptor transferrin receptor) or the constitutive recycling pathway those resident plasma membrane proteins in the canalicplasma membrane (for example the AGp110 or G(glycosylphosphatidylinositol)-anchored proteins; Stamatogl

ticverytedvsonseled

VIntthenoalled

d,hse

dyal

in be

lialresamicousalsoeennd

dfewicalteinsarly

inthee tooticralery

ichreeticorto(iii)maor

forularPIou

et al., 1990; Crawford, 1996). Whereas endocytosis at the apipole of the hepatocyte, against the bile flow, seems difficult proof and most probably physiologically irrelevant (Schmid eal., 1992), recycling of resident membrane proteins in the bcanalicular plasma membrane is plausible and might be throuthe apical endocytic compartment.

In MDCK cells, Apodaca et al. (1994) demonstrated that thapical recycling compartment acts as a ‘sorting’ and recyclinstation returning receptors (transferrin receptor) back to thefunctional surface and is also involved in the final steps of thtranscytosis of pIgA. Although hepatocytes and MDCK cellshare several general aspects of intracellular trafficking, tgeneration of cell polarity and especially the transport of nesynthetized apical membrane proteins is different (Bartles aHubbard, 1988). Perhaps for this reason the hepatic apiendocytic compartment, now defined by annexin VI, is noinvolved in the transcytosis of pIgA although, like in MDCKcells, it contains transferrin. Further evidence indicating that thapical annexin VI positive compartment is not involved in thtranscytosis of pIgA is that when HRP, a fluid phase marker thgoes into the bile, was injected into the liver there were mainvesicular structures filled with HRP around the apical pole of thhepatocyte (Renston et al., 1980; Hoppe et al., 1985; D. Orteand C. Enrich, unpublished), whereas 55% of structures labewith anti-annexin VI were vesiculo-tubular (see Table 1A).

The finding of an ‘early’ endosomal marker such as Rablocated in structures around the bile canaliculus, annexin positive, strongly suggests the existence of an apical ‘earcompartment in the hepatocyte similar to the compartmedemonstrated to exist in MDCK (Apodaca et al., 1994) ancalled the ‘apical recycling compartment’, or the ‘commonendosomal compartment’ described in Caco2 cells by Knightal. (1995), or the ‘apical early endosomal compartment’ in rintestinal epithelial cells described by Speelman et al. (1995

The functional role for annexin VI has been questioned bothers and controversial data have been published (Lin et 1992; Smythe et al., 1994). Whereas it seems clear that anneVI is not involved in the early steps of endocytosis one cannrule out a possible role in the transport towards the late steof the endocytic route. Since Ca2+ is important for endosomefusion one may speculate that cellular levels of annexin VI cabe important for Ca2+ homeostasis and therefore for theregulation of late fusions events (Moss, 1997). Annexin VI ithe first structural protein identified in the apical endocyticompartment of the hepatocyte.

This work was carried out at the Departamento de Biologia CelulaFacultad de Medicina, University of Barcelona, and was supported FISS 95/0479 grant to C.E. and Deutsche ForschungsgemeinschaS.J and M.B. The authors are grateful to Dra. N. Cortadellas, Rivera and A. Garcia from ‘Serveis Cientifico-Tècnics de laUniversitat de Barcelona’ for expert assistance in the preparationLowicryl and cryo-sections for electron microscopy. A.P. is a recipienof a fellowship from FISS (Spain).

REFERENCES

Apodaca, G., Katz, L. A. and Mostov, K. E. (1994). Receptor-mediatedtranscytosis of IgA in MDCK cells is via apical recycling endosomes. J. CellBiol. 125, 67-86.

Barroso, M. and Sztul, E. S. (1994). Basolateral to apical transcytosis inpolarized cells is indirect and involves BFA and trimeric G protein sensitivpassage through the apical endosome. J. Cell Biol. 124, 83-100.

269Annexin VI in apical hepatic endosomes

of

alth

of

of

um

.

tem

in-

r

rand

t

n

Bartles, J. R. and Hubbard, A. L. (1988). Plasma membrane protein sorting iepithelial cells: Do secretory pathways hold the key ? Trends Biochem. Sci.13, 181-184.

Belcher, J. D., Hamilton, R. L., Brady, S. E., Hornick, C. A., Jäckle, S.,Schneider, W. J. and Havel, R. J. (1987). Isolation and characterization othree endosomal fractions from the liver of estradiol-treated rats. Proc. Nat.Acad. Sci. USA84, 6785-6789.

Bomsel, M., Priydz, K., Parton, R. G., Gruenberg, J. and Simons, K.(1989). Endocytosis in filter-grown Madin-Darby canine kidney cells. J. CellBiol. 109, 3243-3258.

Bradford, M. M. (1976). A rapid and sensitive method for the quantitation microgram quantities of protein. Anal. Biochem. 72, 248-254.

Bucci, C., Wandinger-Ness, A., Lütcke, A., Chiariello, M., Bruni, C. B. andZerial, M. (1994). Rab5a is a common component of the apical abasolateral endocytic machinery in polarized epithelial cells. Proc. Nat.Acad. Sci. USA91, 5061-5065.

Burgoyne, R. D. and Geisow, M. J. (1989). The annexin family of calcium-binding proteins. Cell Calc. 10, 1-10.

Burgoyne, R. D. and Clague, M. J. (1994). Annexins in the endocyticpathway. Trends Biochem Sci. 19, 231-232.

Chao, Y.-S., Windler, E. E., Chen, G. C. and Havel, R. J. (1979). Hepaticcatabolism of rat and human lipoproteins in rats treated with 17-α-ethinylestradiol. J. Biol. Chem. 254, 11360-11366.

Chao, Y.-S., Jones, A. L., Hradek, G. T., Windler, E. E. T. and Havel, R. J.(1981). Autoradiographic localization of the sites of uptake, cellultransport, and catabolism of low density lipoproteins in the liver of normand estrogen-treated rats. Proc. Nat. Acad. Sci. USA78, 597-601.

Crawford, J. M. (1996). Role of vesicle-mediated transport pathways hepatocellular bile secretion. Semin. Liver Dis. 16, 169-189.

Crompton, M. R., Moss, S. E. and Crumpton, M. J. (1988a). Diversity in thelipocortin/calpactin family. Cell 55, 1-3.

Crompton, M. R., Owens, R. J., Totty, N. F., Moss, S. E., Waterfield, M. D.and Crumpton, M. (1988b). Primary structure of the human, membranassociated calcium-bindig protein p68: a novel member of a protein famEMBO J. 7, 21-27.

Drubin, D. G. and Nelson, W. J. (1996). Origins of cell polarity. Cell 84, 335-344.

Enrich, C. and Evans, W. H. (1989). Antibodies to hepatic endosomesIdentification of two endosome antigens. Eur. J. Cell Biol. 48, 344-352.

Futter, C. E., Felder, S., Schlessinger, J., Ullrich, A. and Hopkins, C. R.(1993). Annexin I is phosphorylated in the multivesicular body during tprocessing of the epidermal growth factor receptor. J. Cell Biol. 120, 77-83.

Geuze, H. J., Slot, J. W., Strous, G. J. A. M., Peppard, J., von Figura, K.,Hasilik, A. and Schwartz, A. L. (1984). Intracellular receptor sorting duringendocytosis: comparative immunoelectron microscopy of multiple receptin rat liver. Cell 37, 195-204.

Gruenberg, J. and Emmans, N. (1993). Annexins in membrane traffic. TrendsCell Biol. 3, 224-227.

Harder, T. and Gerke, V. (1993). The subcellular distribution of earlyendosomes is affected by the annexin II2p112 complex. J. Cell Biol. 123,1119-1132.

Havel, R. J., Eder, H. A. and Bragdon, J. H. (1955). The distribution andchemical composition of ultracentrifugally separated lipoproteins in humserum. J. Clin. Invest. 34, 1345-1353.

Hemery, I., Durand-Schneider, A. M., Feldmann, G., Vaerman, J. P. andMaurice, M. (1996). The transcytotic pathway of an apical plasmmembrane protein (B10) in hepatocytes is similar to that of IgA and occvia a tubular pericentriolar compartment. J. Cell Sci. 109, 1215-1227.

Hoppe, C. A., Connolly, T. P. and Hubbard, A. L. (1985). Transcellulartransport of polymeric IgA in the rat hepatocyte: biochemical anmorphological characterization of the transport pathway. J. Cell Biol. 101,2113-2123.

Hornick, C. A., Jones, A. L., Renaud, G., Hradek, G. and Havel, R. J.(1984). Effect of chloroquine on low-density lipoprotein catabolic pathwin rat hepatocytes. Am. J. Physiol. 246, G187-G194.

Hughson, E. J. and Hopkins, C. R. (1990). Endocytic pathways in CaCo-2cells: identification of an endosomal compartment accessible from bapical and basolateral surfaces. J. Cell Biol. 110, 337-348.

Jäckle, S., Runquist, E., Brady, S., Hamilton, R. L. and Havel, R. J. (1991).

n

f

of

nd

aral

in

e-ily.

.

he

ors

an

aurs

d

ay

oth

Isolation and characterization of three endosomal fractions from the livernormal rats after lipoprotein loading. J. Lipid Res. 32, 485-498.

Jäckle, S., Beisiegel, U., Rinninger, F., Buck, F., Grigoleit, A., Block, A.,Gröger, I., Greten, H. and Windler, E. (1994). Annexin VI, a markerprotein of hepatocytic endosomes. J. Biol. Chem. 269, 1026-1032.

Jost-Vu, E., Hamilton, R. L., Hornick, C. A., Belcher, J, D. and Havel, R. J.(1986). Multivesicular bodies isolated from rat hepatocytes. Cytochemicevidence for transformation into secondary lysosomes by fusion wiprimary lysosomes. Histochemistry85, 457-466.

Kelly, R. B. (1993). A question of endosomes. Nature 364, 487-488. Knight, A., Hughson, E., Hopkins, C. R. and Cutler, D. F. (1995). Membrane

protein trafficking through the common apical endosome compartment polarized Caco-2 cells. Mol. Biol. Cell6, 597-610.

Laemmli, U. K. (1970). Cleavege of structural proteins during the assembly the head of bacteriophage T4. Nature227, 680-685.

Limet, J. N., Quintart, J., Schneider, Y.-J. and Courtoy, P. J. (1985).Receptor-mediated endocytosis of polymeric IgA and galactosylated seralbumin in rat liver. Eur. J. Biochem. 146, 539-548.

Lin, H. C., Südhof, T. C. and Anderson, R. G. W. (1992). Annexin VI isrequired for budding of clathrin-coated pits. Cell 70, 283-291.

Moss, S. E. (1992). The Annexins. Portland Press Research MonographLondon.

Moss, S. E. (1995). Annexins taken to task. Nature378, 446-447. Moss, S. E. (1997). Annexins. Trends Cell Biol. 7, 87-89. Mullock, B. M., Perez, J. H., Kuwana, T., Gray, S. R. and Luzio, J. P. (1994).

Lysosomes can fuse with late endosomal compartment in a cell-free sysfrom rat liver. J. Cell Biol. 126, 1173-1182.

Olmsted, J. B. (1981). Affinity purification of antibodies from diazotizedpaper blots of heterogeneous protein samples. J. Biol. Chem. 256, 11955-11957.

Parton, R. G., Prydz, K., Bomsel, M., Simons, K. and Griffiths, G. (1989).Meeting of the apical and basolateral endocytic pathways of the MadDarby canine kidney cell in late endosomes. J. Cell Biol. 109, 3259-3272.

Renaud, G., Hamilton, R. L. and Havel, R. J. (1989). Hepatic metabolism ofcolloidal gold-low-density lipoprotein complexes in the rat: Evidence fobulk excretion of lysosomal contents into bile. Hepatology9, 380-392.

Renston, R. H., Maloney, D. G., Jones, A. L., Hradek, G. T., Wong, K. Y.and Goldfine, I. D. (1980). Bile secretory apparatus: evidence for a vesiculatransport mechanism for proteins in the rat, using horseradish peroxidase [125I]insulin. Gastroenterology78, 1373-1388.

Schmid, J., Klapper, H. and Fuchs, R. (1992). Characterization of basolateraland apical endocytic pathways in rat hepatocytes. In Hepatic Endocytosis ofLipids and Proteins(ed. E. Windler and H. Greten), pp. 31-36. ZuckschwerdVerlag.

Slot, L. W. and Geuze, H. J. (1985). A new method of preparing gold probesfor multiple-labeling cytochemistry. Eur. J. Cell Biol. 38, 87-93.

Smythe, E., Smith, P. D., Jacob, S. M., Theobald, J. and Moss, S. E. (1994).Endocytosis occurs independently of annexin VI in human A431 cells. J. CellBiol. 124, 301-306.

Speelman, B. A., Allen, K., Grounds, T. L., Neutra, M. R., Kirchhausen, T.and Wilson, J. M. (1995). Molecular characterization of an apical earlyendosome glycoprotein from developing rat intestinal epithelial cells. J. Biol.Chem. 270, 1583-1588.

Stamatoglou, S. C., Rui-Chang, G., Mills, G., Butters, T. D., Zaidi, F. andHughes, R. C. (1990). Identification of a novel glycoprotein (AGp110)involved in interactions of rat liver parenchymal cells with fibronectin. J. CellBiol. 111, 2117-2127.

Swairjo, M. A. and Seaton, B. A. (1994). Annexin structure and membraneinteractions: A molecular perspective. Annu. Rev. Biophys. Biomol. Struct.23, 193-213.

Tagoe, C. E., Boustead, C. M., Higgins, S. J. and Walker, J. H. (1994).Characterization and immunolocalization of rat liver annexin VI. Biochim.Biophys. Acta1192, 272-280.

Weinman, J. S., Feinberg, J. M., Rainteau, D. P., Gaspera, B. D. andWeinman, S. J. (1994). Annexins in rat enterocyte and hepatocyte aimmunogold electron-microscope study. Cell Tissue Res. 278, 389-397.

Wilson, J. M., Whitney, J. A. and Neutra, M. R. (1987). Identification of anendosomal antigen specific to absorptive cells of suckling rat ileum. J. CellBiol. 105, 691-703.