Receptor-mediated Transcytosis of IgA in MDCK Cells Is via Apical Recycling Endosomes Gerard Apodaca, Leonid A. Katz, and Keith E. Mostov Department of Anatomy, Department of Biochemistry and Biophysics, and Cardiovascular Research Institute, University of California, San Francisco, California 94143 Abstract. Classically, the polymeric immunoglobulin estimate that approximately 80 % of basolaterally inter- receptor and its ligand, IgA, are thought to be sorted nalized IgA resides in the apical endosomal compart- from basolateral early endosomes into transcytotic ment. In addition, approximately 50% of basolaterally vesicles that directly fuse with the apical plasma mem- internalized transferrin, a basolateral recycling protein, brane. In contrast, we have found that in MDCK cells has access to this apical endosomal compartment and IgA is delivered from basolateral endosomes to apical is efficiently recycled back to the basolateral surface. endosomes and only then to the apical cell surface. Microtubules are required for the organization of the Do w When internalized from the basolateral surface of apical endosomal compartment and it is dispersed in nlo a d MDCK cells IgA is found to accumulate under the nocodazole-treated cells. Moreover, this compartment e d apical plasma membrane in a compartment that is ac- is largely inaccessible to fluid-phase markers added to fro m cessible to two apically added membrane markers: either pole of the cell, and therefore seems analogous http anti-secretory component Fab fragments, and avidin to the recycling endosome described in nonpolarized ://ru p internalized from the biotinylated apical pole of the cells. We propose a model in which transcytosis is not re cell. This accumulation occurs in the presence of api- a specialized pathway that uses unique transcytotic ss.org cal trypsin, which prevents internalization of the vesicles, but rather combines portions of pathways /jcb /a ligand from the apical cell surface. Using a modifica- used by non-transcytosing molecules. rticle tion of the diaminobenzidine density-shift assay, we -p d f/1 2 5 /1 /6 7 /1 2 6 Primary function of epithelial cells is to regulate the ex- two surfaces meets (Bomsel et al., 1989; Parton et al., 1989; 13 6 change of macromolecules between their external Bomsel et al., 1990; Fujita et al., 1990). In addition, some 0/6 7 environment and the underlying tissue. This selec- proteins are sorted in early endosomes into transcytotic vesi- .pd tive exchange is made possible by the impermeability of the cles and delivered to the opposite plasma membrane surface. f by g monolayer to both large and small molecules and specialized Transcytosis is a key membrane trafficking process in ue endocytotic pathways that allow transit of macromolecules polarized cells because it allows for the exchange of macro- st o n across these cells (Mostov et al., 1992). Epithelia charac- molecules from one cell surface to the opposite one (Apo- 23 F teristically have discrete plasma membrane domains at the daca et al., 1991; Mostov et al., 1992). In addition, this is eb apical and basolateral poles of the cell, that are separated by the only pathway for delivery of newly synthesized mem- rua tight junctions (Simons and Wandinger-Ness, 1990). Each brane proteins to the apical cell surface that is universally ry 20 2 domain has a distinct protein and lipid composition which found in all epithelial cells examined, and in some epithelial 3 is maintained despite an enormous flux of membrane traffic cells, such as hepatocytes, it is virtually the only pathway for to and from each surface. In MDCK cells, for instance, an membrane proteins to this surface (Bartles et al., 1987). area of plasma membrane equivalent to 40 % of the cell sur- Transcytosis is also central to the establishment of cell polar- face is endocytosed per hour (von Bonsdorff et al., 1985). ity (Zurzolo et al., 1992). Fluid-phase material endocytosed from the apical and Much of our understanding of the transcytotic pathway basolateral surfaces of these cells is delivered to correspond- comes from studies of the polymeric immunoglobulin recep- ingly distinct apical and basolateral early endosomes that are tor (pIgR) ~ and its ligand dimeric IgA. The generally ac- not thought to interchange their contents (Bomsel et al., cepted model of pIgR traffic (Fig. )1 is based on studies in 1989; Parton et al., 1989). Although much of the endocy- rat liver and MDCK cells transfected with the pIgR cDNA tosed fluid, and many membrane components are recycled (Geuze et al., 1984; Hoppe et al., 1985; Apodaca et al., to the original plasma membrane surface, some are sent in 1991). The pIgR and its ligand are internalized, along with a microtubule dependent process to a common late endo- the other receptors and fluid phase markers, in coated pits some or prelysosome, where material endocytosed from the Address all correspondence to Gerard Apodaca, Ph.D., Department of .1 Abbreviations used in this paper: DAB, diaminobanzidine; FSG, fish skin Anatomy, Box 0452, University of California, San Francisco, CA 94143- gelatin; MTOC, microtubule organizing center; pIgR, polymeric-immuno- 0452. globulin receptor; SC, secretory component, Tf, transferrin. © The Rockefeller University Press, 0021-9525194/04/67/20 $2.00 The Journal of Cell Biology, Volume 125, Number ,1 April 1994 67-86 67 at the basolateral (sinusoidal) surface (step 1). Subsequently, clonal antibody ascites to ZO-I, a protein associated with tight junctions, the receptor is delivered to a tubulo-vesicular early en- was obtained from Chemicon (Temecula, CA) and was used at 1:100 dilu- tion. Guinea pig anti-SC Fab fragments (derived from affinity purified dosomal compartment shared with other receptors. Here the guinea pig serum) were used at 10-20 #g/ml and were prepared as described pIgR and its ligand are segregated from recycling receptors (Breitfeld et al., 1989b). Fab-HRP was prepared by conjugating reduced into microdomains of endosomal tubules and packaged into YlaF to sulfosuccinimidyl 4-(N-maleimidomethyl) cyclohexane-l-cau'boxylate structures, which at least in thin sections have the appear- (Pierce Chemical Co., Piscataway, NJ) derivatized HRP according to the manufacturer's protocol and was used at 25 /zg/ml. Avidin-TRITC and ance of vesicles (step 2) (Geuze et al., 1984). These "trans- avidin-HRP were obtained from Vector Laboratories (Burlingame, CA) and cytotic vesicles" are thought to contain a variety of trans- used at 25 #g/ml. ExtrAvidin-10 nm gold was from Signm (St. Louis, MO) cytosing molecules (Barr and Hubbard, 1993), but not and was washed with PBS and diluted 1:5. Lysine-fixable dextran-FITC recycling molecules (Sztul et al., 1991). In both hepatocytes (10,000 tool wt) was from Molecular Probes (Eugene, OR) and used at and pIgR-expressing MDCK cells transcytotic vesicles are 10-15 mg/ml. Rabbit antiserum to the tmns-Golgi network resident protein TGN-38 (kindly provided by G. Banting, University of Bristol, Bristol, En- found accumulated near and in continuity with the apical gland) has been described (Wilde et al., 1992) and was used at 1:100 dilu- (bile canalicular) cell surface (Geuze et al., 1984; Hoppe et tion. Canine apo-transferrin (Sigma) was loaded with iron as described al., 1985; Hunziker et al., 1990). Microtubules are required (Podbilewicz and Mellman, 1990) and used at 25/zg/hal. In addition this for this accumulation of transcytotic vesicles to occur. In protein was coupled to keyhole limpet hemocyanin and the conjugate used to produce rabbit antibodies against this protein (coupling and immuniza- nocodazole-treated MDCK cells IgA is internalized into tion of rabbits was performed by Caltag, South San Francisco, CA). Anti-Tf basolateral endosomes but translocation of IgA from the antibodies were affinity purified on a canine-Tf sepharose coltmm and used basolateral to apical pole of the cells is prevented (Hunziker at 10-20 #g/ml. The mouse monoclonal antibody, H68.4, was kindly et al., 1990). Upon reaching the apical surface, the pIgR is provided by I. Trowbridge (Scripps Research Institute, La Jolla, CA) and cleaved to SC, which is released with IgA into secretions recognizes the Tf receptor from a number of different species (White et al., 1992). Expression of the Tf receptor in pIgR-expressing MDCK was (step 3). A vesicular fraction highly enriched in pIgR and D confirmed by immunoprecipitating the 95,000-Mr protein from radiola- ow IgA has been isolated from rat liver and termed transcytotic beled cell lysates. For immunofluorescence studies the ascites was diluted nlo carrier vesicles (Sztul et al., 1991). These vesicles have been 1:100. AC17 is a mouse monoclonal antibody (kindly provided by E. Rod- ad e a cyshtoownt ic cell-free to vesicles fuse fusion with with assay isolated the that apical reconstitutes bile surface canalicular (Sztul the fusion membranes et al., of 1993). trans- in 9t5ri,io0g0nu0 e-zMr- Boulain, (Nabi et lysosomal al., Cornell 1991). membrane UniTvheres ity mouse glycoprotein Medical monoclonal College, and antibody was New used DMYork, 5~ at 1:100 NY) specific dilu- to a d from http We were interested in investigating the nature of the sub- faonrt iserum cc-tubulin, specific was obtainfeodr 7-tubulin from (kindly Sigma and provided used at by 1:250 T. Stearns, dilution. University A rabbit ://rup aapnicda l in IgA-containing pIgR-expressing vesicles/tubules MDCK cells. found Do theiyn hepatocytes represent oanfd California, was used San at a Francisco, 1:100 dilution. CA) has FITC- been descroirbe d Texas (Streeardn-sc onjugated et al., 1991) goat ress.org transcytotic vesicles that are waiting to fuse with the apical anti-human IgA, goat anti-rat IgG, goat anti-mouse IgG, goat anti-rabbit /jcb IgG antibodies were obtained from Jackson ImmunoResearch Laboratories /a s(uGrefuazcee , et as al., has 1984; been Hoppe suggested et al., by 1985; several Hunziker investigators et al., (WestCn'ove, anti-rabbit antibodies PA) and used had at minimal 10-20 t~g/ml. cross-reactivity The anti-rat, with anti-mouseaec,h other aanndd rticle-p d 1990)? Alternatively does the apical accumulation of IgA with human serum proteins. f/1 2 represent ligand that has been internalized from the apical Cell erutluC 5/1/6 cell surface into apical endosomes, or IgA delivered from 7/1 2 basolateral endosomes directly to apical endosomes? In ad- MDCK strain II cells expressing the wild-type rabbit pIgR have been de- 61 3 dition, we wished to determine at what point in the trans- scribed (Breitfeld et al., 1989a). Although the results presented in this pa- 60 cytotic pathway IgA is segregated away from macro- per are from one clone of MDCK strain II cells transfected with the plgR /67 molecules that are either delivered to prelysosomes, (e.g., eclDoNnAes, of we pIgR-expressing have obtained vMeDryC K similar cells resul(tsn ot shown). using twoC ells other were independemnati ntained .pdf b fluid-phase markers), or recycled back to the basolateral cell in Minimal Essential Medium (MEM; obtained from the UCSF Cell Cul- y gu e surface (e.g., transferrin). oT address these questions we ture Facility) supplemented with 10% FBS (Hyclone, Logan, UT), 100 st o have analyzed the transcytotic pathway of IgA in pIgR- U/ml penicillin, and 100/~g/ml streptomycin in 5% CO2/95 % air. In order n 2 expressing MDCK cells. Using these cells it is possible to to maintain a high level of receptor expression, new cells were thawed every 3 F follow the distribution of IgA in relationship to ligands and 3--4 cells wk, were and cultured were split on 12-I:10 ram and diam, pessnged 0.4-t~m once pore weekly. size FoTrra nswells all experiments, (Costar; ebrua fluid-phase markers that define the major endocytotic path- Cambridge, MA) as described (Breitfeld et al., 1989a). The cells were fed ry 2 ways, and to analyze events occurring at both the apical and everyday and used 3-4-d postculture. 02 3 basolateral poles of the cells. In contrast to the model for noitazilanretnI of sdnagiL dna esahp-diulF transcytosis described above, we find that in MDCK cells ,srekraM elozadocoN ,tnemtaerT gnippirtS of IgA is delivered from basolateral endosomes to apical endo- Cell-surface Ligands, and Biotinylation somes and only then to the apical cell surface. In addition, we find that transferrin (Tf), a basolateral recycling protein, Ligands and fluid phase markers were internalized from the apical and/or has access to this apical endosomal compartment, where it basolateral surface of filter-grown MDCK cells. Prior to "If internalization, is recycled back to the basolateral surface. it was necessary to incubate the cells for a minimum of 4 h at 37°C in MEM/BSA (MEM, Hanks Balanced Salts, 0.6% wt/vol BSA, 20 mM Hepes, pH 7.4) to deplete intraccllular stores of Tf. Otherwise, a Tf- dependent signal could not be obtained in our immunottuorescenco analy- Materials and Methods sis. All incubations in MEM/BSA were performed in a circulating water bath. It was also possible to deplete the ceils of'IT by incubating them over- Antibodies and Proteins night in MEM, Earle's balanced salts, 0.6% BSA, 20 mM Hepes, pH 7.4 in the tissue culture incubator. No obvious harmful effects were observed Purified human dimeric IgA was kindly provided by J.-P. Vaerman (Catholic in serum-starved cells incubated in either fashion, as the distribution of University of Louvaln, Brussels, Belgium) and was used at a concentration pIgR and Tf-receptor were unaltered and the appropriate ligands were faith- of 50/zg/ml. IgA-HRP was custom prepared from human dimeric IgA and fully recycled or transcytosed. For basolateral uptake of ligands or fluid- HRP by Zymed (South San Francisco, CA) and used at 25/~g/ml. Rat mono- phase markers the cells were rinsed with MEM/BSA at either 81 ° or 37°C The Journal of Cell Biology, Volume 125, 1994 68 and the edge of the filter on the side opposite the ceils was carefully blotted a capped syringe at -20°C. The slides were stored at -20°C until viewing to remove excess medium. The Transweil unit was placed on a 1~-52 drop with the conf~al microscope. of MEM/BSA containing the ligand or fluid-phase marker. rOF apical up- The specificity of the immunoflnorescence staining was confirmed as take, the cells were rinsed with MEM/BSA at the appropriate temperature, follows: excess fluid was aspirated from the cell-side of the Transwell and 051 gl /gA. No IgA signal was detected in cells in which this ligand was not ofligend or fluid-phase marker, diluted in MEM/BSA, was added. All incu- internalized, or in MDCK cells not expressing the pIgR. The anti-rat, bations were performed in a humid chamber. At the end of the experiment anti-mouse, and anti-rabbit secondary antibodies did not cross-react with cells were typically washed two to three times with MEM/BSA equilibrated the IgA, and the anti-human IgA secondary antibody did not cross-react at the appropriate temperature, and either rapidly cooled down to 4°C or with any of the other ligands, fluid-pbase markers, antibodies, or secondary if appropriate fixed immediately. antibodies. Nocodazole (Calbiochem-Behring Corp., San Diego, CA) was dissolved ZO-I, ~-TfR, TGN-38, ~-tubuUn, and "~-tubuUn. The specificity of in DMSO at 33 mM and stored at -20°C. In all experiments in which this staining for these molecules was based on their established distribution in drug was used, cells were pretreated 60 rain at 4°C in the presence of 33 these and other cells. None of these antibodies cross-reacted with IgA, and Mt~ nocodazole. The drug was included in subsequent incubations. when omitted from the staining protocol no signal was detected. The distri- In many of the experiments ceil-surface receptors and their ligands were bution of these proteins was identical whether or not IgA was internalized. stripped from the cell surface as follows: ceils were treated for 30 min at .fT No signal for Tfwas detected in starved cells in which the ligand was C04 with 50-100 im/gt~ of TPCK-treated trypsin or 50/~g/rni of proteinase not internalized or if "If was added to the apical surface of the cells for 30 K diluted in MEM/BSA. Subsequently, the ceils were washed twice with rain at 37°C. In addition, in cells that had internalized Tfbut were incubated ice cold MEM/BSA, one time 01 rain with either 125/~g/ml soybean trypsin with preimmune serum no labeling was detected. Labeling was not detected inhibitor or 2 mM phenylmethylsuifonyl fluoride dissolved in MEM/BSA. in cells in which the affinity purified anti-Tf antibodies were omitted and For morpholngicai analysis the cells were subsequently rinsed with PBS these antibodies did not cross-react with IgA. The distribution of Tf was containing 0.5 mM MgCI2 and 0.9 mM 21CaC (PBS+), and immediately identical whether or not IgA was internalized. fixed. Before internaliTation of avidin-TRITC, avidin-gold, or avidin-HRP it gninnacS Laser lacofnoC sisylanA of yltnecseroulF was necessary to biotinylate the apical surface of the cells. All manipula- delebaL slleC D tions were performed at 4°C. Cells were washed three times with HBSS o w containing treated twice Ca for +2 and 51 rain Mg +2with salts 0.5 and mg/ml 20 mM of NHS-LC-biotin Hepes, pH 7.4, (Pierce, SSBI-I( )+Rock- and The samples were analyzed using a krypton-argon laser coupled with a Bio- nload Pad MRC600 confocal head, attached to an Optiphot H Nikon microscope e ford, cquickly ence IL) analysis and dissolved one NHS-SS-biotin time in for the 01same rain was buffer. with used, MEM/BSA. The and cells following were In the then apical immunofluores- washed internaliza- twice with simultaneously a Plan Ape for 60X F1TC 4.1 and NA Texas objective red (orlens. propidium The samples iodide were or scanned TRITC) d from h tavidin igintathione osolution. n of avidin-TRITC, complexes This in 0.9reagent ml by ofincubating the was 38 apical mM prepared surface NaCI. the cells by Immediately of dissolving the twice cells for was pri 20 o5r5 1 stripped rain mg to use in of of reducing reduced 1 biotin- ml of emission dide, shows = 0.3 or or FITC TRITC 3.5, using fluorescence 0.5 thfluorescence. e s/scan, 1K and 5 while frames/imnge, K2 Collection filter the right blocks. parameters Kaiman shows In each Texas filter, were figure red, motor as propidium follows: the step left size zopanel om io- -- ttp://rupress.o 5.0 ,mt~ diaphragm closed or set at 3/1 open. The data was analyzed using rg calf serum and 60 1~ of 50% (wt/vol) NaOH was added. Ceils were washed Comos sofh~tre and regions of colocaiization were identified using the /jcb wFixation ith PBS + and and immediately Fluorescent fixed. Labeling of sUeC tmerge format oshop side program (TIFF) function. and (Adobe the The contrast Co., images Mountain were levels converted oView, f the CA) images to on tagged-information-file- a adjusted Macintosh in thlel ci Pho- (Ap- /article-pd Samples were fixed with paraformaidehyde using a pH-shifl protocol (Bom- ple, Cupertino, CA). The contrast-corrected images were imported into f/125 sel et al., 1989) or with glutaraidehyde. In the pH-shift protocol, cells erew. Pngemaker (Aldns Corporation, Seattle, )AW and printed from an Agfa /1/6 9,800 imngesetter at 2,400 dots per inch, using a line screen of 051 7 fixed 5 rain at room temperature with 4% paraformaldehyde, 80 mM /1 lines/inch. 2 Pipes/KOH, pH 6.5, 5 mM EGTA, 2.0 mM MgCl2, and then transferred 61 3 to 4% paraformatdehyde dissolved in 001 mM NaBorate, pH 1L0, and in- 6 0 cubated 01 rain at room temperature. Cells were washed 2 x 3 rain with Ultrastructural Analysis of lgA-HRP sisotycsnarT /67 PBS, pH 8.0, and nonmacted paraformaidehyde was quenched 01 min at .pd rsolved oom temperature in PBS, pH with 8.0). 57 Following mM ,1Ch-IN two 20 5-rain mM washes Glycine, with pH PBS, 0.8 (both pH dis- ,0.8 Following washed three internalization times quickly of with IgA-HRP MEM/BSA and avidin-gold, and one time the with cells ice were cold f by gu e nonspecific sites were blocked with PBS, 0.7% (wt/vol) fish skin gelatin PBS .+ The ceils were immediately fixed by adding ice-cold 0.5 % (vol/vol) st o (FSG), and 0.025% (wt/vol) saponin (Sigma). In experiments in which the glutaraldehyde in 200 mM Na cncodylate, pH 7.4, 1 mM CaCI2, 5.0 mM n 2 nucleus was stained with propidium iodide 001 lm/g~t of boiled RNAse A MgCI2, and incubating the cells for 30 rain at room texture. Cells 3 F was added to the blocking solution. In some experiments coils were fixed were rinsed three times with 200 mM Na cacudylate buffer, pH 7.4, and eb with 0.2 % (vol/vol) glutaraldehyde dissolved in 80 mM Pipes/KOH, pH ,8.6 %1.0 (wt/vol) diaminobenzidine (DAB), dissolved in 200 mM cacodylate rua 5 mM EGTA, 2.0 mM MgCI2, for 01 rain at room temperature. Non- buffer, was added for 2 rain at room temperature. The BAD solution was ry 2 reacted glutaraldehyde was quenched by incubating the cell three times 51 aspirated and replaced with fresh BAD solution containing %10.0 (vol/vol) 02 3 rain with freshly prepared 1 mg/mi 4I-IBaN dissolved in PBS, pH .0.8 The H202 and incubated 30 rain at room temperature in the dark. Samples were cells were washed twice with PBS and blocked in PBS-FSG-sapouln as de- rinsed with 200 mM Na cacodylate, pH 7.4, and osmicated with %1 4OsO scribed above. (wt/vol), 200 mM Na cacodylate, pH 7.4, %1 (wt/vol) IGFe(CN)6, for 90 The fixed cells were incubated with the appropriate primary antibodies, rain at 4°C. After several rinses with 0241 the samples were block-stained diluted in PBS-FSG-saponin, for 45 min at 37°C in a humid chamber, and overnight with 0.5% (wt/vol) uranyl acetate in H20. Filters were de- then washed 3 x 3 rain with PBS-FSG-sapouin, 1 x 3 rain with PBS- hydrated in a graded series of ethanol, embedded in the epoxy resin 211-XL sapouln, and 1 × 3 rain with PBS-FSG-saponin. Subsequently, the cells (Ladd Res. Inds., Inc., Burlington, VT), and sectioned with a diamond were incubated 45 mill at 37°C with the appropriate combination of fluores- knife (Diatome, Fort Washington, PA). Sections, 200-225-nm thick (as de- cently labeled secondary antibodies and propidium iodide (2/~g/ml) diluted termined by their interference colors) were mounted on butvar-coated nickel in PBS-FSG-sapouin. The cells were washed 3 × 3 rain with PBS-FSG- grids and viewed at 80 Vk in a Zeiss (West Germany) EM-10 electron micro- saponin, 1 × 3 rain with PBS-saponin, and 2 × 3 rain with PBS, pH ,0.8 scope without further contrasting. alone, 5 rain with 0.1% Triton X-100, dissolved in PBS, pH 8.0, followed oT confirm that IgA-HRP was efficiently transcytosed, this ligend was by a 5-rain wash in PBS, pH ,0.S alone. Cells were posttixed in 4% parafor- basolaterally internalized for 01 rain at 37°C by pIgR-expressing ceils. Fol- maldehyde dissolved in 001 mM Na-cucodylate, pH 7.4, for 30 rain at room lowing a 2-h chase at 37°C approximately %08 of IgA-HRP was found in tempera~ue, washed twice with PBS, and mounted inp-phenylene diamine the apical medium (l~'mcytosed), 5 % was released basolaterally, and the (Sigma) (Johnson, )1891 which was prepared as follows: 2 ml of 200 mM remainder was cell associated, IgA-HRP was not significantly intemulized Tris, pH 8.2, was added to 20 mi of glycerol and mixed by bubbling nitrogen in the presence of 100-fold excess unlabeled IgA or by MDCK cells not ex- gas through the solution for 20 min at room temperature. 200 milligrams pressing the plgR. Similarly, no BAD reaction was detected at the EM level of p-phenylene diamine was added and the mixing continued for two addi- if an excess of IgA (2.5 mg/ml) was added during the internalization of the tional hours. The mounting medium was stable for 7-10 d when stored in IgA-HRP to the cell surface, if the IgA-HRP was omitted during the inter- Apodaca et .la Transcytosis of IgA Is via Apical Endosomes 69 nalization step, or if IgA-HRP was internalized by non-transfected MDCK cally these values were less than 5 % of the total counts). Finally these values cells. were normalized to those obtained when AgI152~ and Avidin-HRP or Fab-HRP were cointernalized for 01 rain at 37°C or 120 rain at 18°C. sisylanA of fTIW gnilcyceR Results Iron-saturated "If was iodinated to a specific activity of 5.0-9.0 × 601 cpm/~tg using ICI as described (Breitfeld et al., 1989a). The cells were depleted of endogenous "If by incubating for 4 h in MEM/BSA and fTI521 noitalumuccA of lgA in a Subapical Compartment of (5 ~tg/mi) was internalized from the basolateral surface of the cells for 2 h gnisserpxe-Rglp MDCK Cells at 18°C. The cells ware washed two times quickly and one time 3 rain with MEM/BSA at 18°C before incubation at 37°C in the presence of 50 ~tg/ml In Fig. 2 we have analyzed the time and temperature require- of cold "Ft. At the appointed times the cells were rapidly cooled on ice and ments for basolaterally internalized IgA to accumulate at the the apical and basolateral media were collected. 12SlTf was stripped from apical pole of pIgR-expressing MDCK cells. Individual sec- the cell surface by incubating the cells for 60 rain at 4°C with 750 mM Gly- tions, obtained with a confocal microscope, are shown from cine, pH 2.5, diluted 5:1 with PBS +. The Transwells were rinsed with PBS + and the filters were cut out of their holders. The total fTJ-1521 ini- the basal portion of the cell below the nucleus (level 4), from tially bound to the cells includes ligand recycled to the basolateral surface, the lateral surface of the cell at the level of the nucleus (level ligand transcytosed into the apical medium, ligand stripped from the cell 3), from the apical region of the cell above the nucleus surface with acid, and cell associated ligand not sensitive to glycine strip- (level 2), and from the apical pole of the cell at or above the ping (endocytosed), and was quantitated in a gamma counter (Beckman In- struments, Palo Alto, CA). 125ITf uptake was inhibited >95% when the level of the tight junctions (level 1 ). radioactive llgand was internalized in the presence of a 100-fold excess of Following a 5-rain pulse at 37°C basolaterally internalized cold ligand. IgA was found in discrete vesicles below, at the level of, and above the nucleus (Fig. 2 A). At the level of the tight junc- Analysis of p3IIgA sisotycodnE and tions the majority of the IgA remained at the cell margins, Do w citotycodnetsoP etaF however, some IgA was also found in apical elements at the nlo a apex of the cell. When the 5-rain pulse was followed by a 10- de AgI1521 was iodinated using the ICI method to a specific activity of d 1.0-2.0 x 701 cpm//~g. Endocytosis of AgI1521 was measured as de- rain chase, IgA was found throughout the cell including a fro m scribed (Breitfeld et al., 1989b). When specified 25/~g/mi of trypsin was large accumulation of ligand apically in both clusters and as h AigII5n21cluded was in the stripped apical from medium. the cell At surface the conclusion and the results of the quantitated experiment as stributed mall individual and radiated vesicles towards that the were margins typically of the centrally cell (Fig. dis- 2 ttp://rup dfeastec ribed of a praebionvte ernalized for the analysis cohort of of f'1'I521 IzsIIgA recycling. was analyzed The postendocytotic as described B). Following a 25-rain chase little IgA could be found in ress.o (iBnr eitfeld the apical et al., medium. 1989b). When specified 25/tg/mi of trypsin was included tthhee bIagsAa l was and largely lateral distributed portions of in the vesicular ceil (Fig. elements 2 C). Instead, present rg/jcb/a in the apical cytoplasm above the nucleus and at the apex of rticle BAD tfihs-ytisneD yassA the cell. Following the 25-rain chase the intensity of IgA -p d staining had decreased and by 90 rain of chase little IgA f/12 Iinn ternalization the original DABo f markers density-shift (one of assay, which cells was were conjugated homogenized to HRP following and the could be detected in the cell (Fig. 2 D). These results demon- 5/1/6 7 other radiolabeled) and a fraction rich in endosomal markers was purified strate that IgA moves rapidly from peripheral basolateral en- /12 by centrifugation (Courtoy et al., 1984). During homogenization there was dosomes and subsequently accumulates at the apical pole of 61 3 always some breakage of endosomes. This resulted in loss of radiolabeled the cell as it is being transcytosed. The distribution of the 60 lsiingkna l vesicles and the when release these of HfRraPct ions conjugate were reacted that could with non-specificallyD AB and H202. cross- Fol- pIgR was almost identical to its ligand; it was found through- /67.pd lwoew ing have the modified advice othfiJs . originKaalp lan (pUnriovteorsciotly , by of omitting Utah, Salt the Lake homogenization City, UT), oouft the the cell. cell but was especially concentrated at the apical pole f by gu e step and performing the DAB reaction on whole cells. After internalization The distribution of IgA in cells incubated at 18°C was also st o of fT1521 (5 /~g/mi), 125IIgA (5 #g/mi), Fab--HRP (25 #g/ml), or assessed. At this temperature IgA can be recycled basolater- n 2 aravdiiodliabne-lHedR P (25/~g/ml) ligands stripped the cells from were the washed cell surface with ice-coldw ith 100 MEM/BSA ~tg/ml trypsin and ally, however, release of the ligand from the apical pole of 3 Feb (in the case of 125IIgA) or with 50/zg/ml proteinase K (in the case of the cell is inhibited (Hunziker et al., 1990). Following a 30- rua )fTI521 for 3 x 10 rain at 4°C. The cells were then washed twice with rain incubation at 18°C the IgA was found in a peripheral en- ry 2 0 ice-cold HBSS +. DAB reaction buffer (0.5 ml) was added to both apical dosomal compartment that lies close to the basolateral cell 23 and basal compartments of the Transwell. This reagent was prepared by add- surface (Fig. 2 e). Although similar to the distribution seen ing 3.3 ml of 3 mg/mi of DAB (dissolved in HBSS +, pH adjusted to 7.4 after 5 rain at 37°C, at 18°C there was little IgA located with NaOH, and filtered), and 20/zl of 30% (vol/vol) H202 to 20 mi of HBSS +. In control reactions H202 was omitted from the DAB reaction above the nucleus. After 2 h of internalization, IgA was also buffer. Following a 45-rain incubation at 4°C the cells were washed twice found not only in the peripheral endosomes but also through- with HBSS +, the filters were carefully excised from their holders, boiled out the cytoplasm (Fig. 2f). At the level of the tight junctions for 2 min in 0.4 mi of SDS lysis buffer (0.5% w/v SDS, 100 mM some IgA was also found concentrated in a centralized spot triethanolamine, pH 8.6, 5 mM EDTA, 0.02% wt/vol NAN3), and vortex shaken for 51 rain at 4°C. Under these conditions <5 % of the total counts of bright fluorescence. This cluster of IgA was often more were associated with the filter. The supernatants were then centrifuged at compact and condensed than observed at 37°C and may 100,000 g in a Sorvall TAOTFR rotor for 25 rain at 20°C. Radioactivity was reflect alterations in the interaction of this compartment with quantitated in a gamma counter. the cytoskeleton at 18°C. This distribution is similar to that In this assay radiolabeled ligand present in the avidin-HRP or Fab-HRP reported by Hunziker et al. (Hunziker et al., 1990). The dis- filled apical endosomal compartment is cross-linked by the DAB reaction into a dense, detergent-insoluble complex that is recovered in the pellet fol- tribution of IgA at 2 h was not altered if the incubation period lowing centrifugation. The absolute value for the amount of ligand present was extended to 4 h (not shown). Like at 37°C, IgA internal- in the DAB cross-linked endosomes was obtained by dividing the amount ized for 30 rain at 18°C was translocated to the apical pole of ligand present in the pellet by the total amount of ligand present in both of the cell when the cells were chased in ligand free medium the pellet and supernatant. The percent of ligand that pelleted when H202 was omitw, d from the DAB reaction was subtracted from this value (typi- for 90 rain at 18°C (not shown). The Journal of Cell Biology, Volume 125, 1994 70 Transcytosing IgA Accumulates in Apical Endosomes accessible to apically internalized ligands. A prediction of the classical model for transcytosis presented in Fig. 1 is that We weren ext interested in determining the nature of the api- transcytotic vesicles wouldb e inaccessible to apically inter- cal compartment in which the IgA was accumulating. We nalized ligands. If, however, the apically distributed IgA was considered three possibilities: the apically distributed IgA accessible to apically internalized ligands it might suggest could represent transcytotic vesicles that had not yet fused that one of the latter possibilities was true. with the apical plasma membrane, or IgA that had reached In the first part of our analysis we have labeled apical endo- the apical surface and had been re-internalized into apical somes with monovalent anti-SC Fab fragments (derived endosomes, or IgA that was delivered directly from baso- from affinity purified anti-SC antibodies). When added api- lateral to apical endosomes. To distinguish among these cally the anti-SC Fab fragments act as a pseudoligand for the three possibilities we determined if the basolaterally inter- pIgR and are efficiently internalized and recycled from the nalized IgA that accumulates in the apical compartment is apical cell surface (>95 % of this ligand is recycled apically) (Breitfeld et al., 1989b). Little of this ligand is transcytosed in the apical-basolateral direction. As demonstrated in Fig. 3 a there was a significant degree of colocalization of basolaterally internalized IgA with Fab fragments internal- ized apically for 10 min at 37°C. Colocalization of the two ligands was also apparent in endosomal structures that lay below the level of the tight junctions and above the nucleus (not shown). Of course at the light level it is not possible to rule out that IgA and Fab fragmenatrse segregated into sepa- D o rate subdomains of an apically distributed compartment. wn lo However, using a density-shift assay (described below) we ad e estimate that approximately % 75 of the IgA is present in this d fro Fab-labeled apical endosomal compartment (see Fig. 6 b for m h quantitation). In this and many of the subsequent experi- ttp ments IgA was continuously internalized for 30 rain at 37°C ://ru p tbroig htly. completely The distribultaiboenl all of tohfe IgA IgA-accessible internalized compartmenftosr 30 min at ress.org 37°C is like that showinn Fig. 2 b, with the majority of IgA /jcb /a pperreispehnetr al in the basolateral structures. apical region of the Sicmeillla r and results some present were ob- in rticle-p d tained if cells were pulsel abeled with IgA for 5 min and sub- f/1 2 sequently chased prior to the addition of apical ligand. 5/1 /6 Identical results were observed if Fab fragments were in- 7 /1 ternalized from the basolateral surface and IgA was internal- 26 1 3 ized apically (not shown). Similarly, the meeting of apically 6 0 added Fab fragments with basolaterally internalized IgA oc- /67 .p curred in cells incubated at 18°C (Fig. 3 b), a temperature df b at which transcytosis is inhibited. The significant degree of y g u colocalization observed at 37 ° and 18°C suggests that the est o apically accumulated IgA did not represent unfused trans- n 2 cytotic vesicles, but rather an endosomal compartment that 3 F e was accessible to both basolaterally internalized IgA and bru a apically internalized ligands. ry 2 0 2 3 lgA Is Delivered to Apical Endosomes Under Conditions Where Apical Endocytosis of the Ligand Is Prevented The apical accumulation of IgwAe observed coulbde the re- erugiF .1 model Classical for transcytosis of the plgR in MDCK sult of ligand that was first delivered to the apical plasma cells. In the classical model for transcytosis AgI is internalized, membrane and then internalized into an apical endosomal along with and molecules, recycling fluid-phase into a common compartment. It is known that cleavage of the pIgR to SC is ). 1 (step early endosome basolateral Within com- endosomal this slow relative to apical endocytosis; consequently, the pIgR de- are markers Fluid-phase occur. events sorting several partment and its bound ligand can be endocytosed from the apical cell livered to (e.g., receptors many while endosomes, late )rotpecer-fT surface (Breitfeld et al., 1989b). If this is the case it should is IgA pack- membrane. back plasma basolateral the are recycled be possible to preventI gA from being internalized from the with fuse ultimately 2), which (step vesicles transcytotic into aged apical cell surface by including trypsin in the apical medium. proteinase a surface cell apical the At membrane. plasma apical the sevaelc the receptor to ,CS releasing it into secre- IgA bound and We have previously used trypsin to cleave pIgR molecules tions (step .)3 BEE, basolateral early LE, endosome; late endo- as theya rrive at the basolateral surface of the cell, thereby, some/prelysosome; ,JT junction. tight preventing their subsequent endocytosis and transcytosis Apodaea et al. Transcytosis of lsA Is via Apical Endosomes 17 Figure .2 Time and tempera- ture dependence of IgA accu- mulation in a subapical com- D partment of pIgR-expressing ow n MDCK cells. IgA was inter- loa d nalized from the basolateral e d surface of MDCK cells ex- fro m pressing the pIgR for 5 rain at h pu37l°Cs e (a). of IFgoAl lowing the cells the 5-rain were ttp://rup trhae pidly absence washed of andl igand chased for 01in ress.org (b), 25 (c), or 90 min at 370C /jcb b(ads)o.la terally IgA was also for internalized30 (e) or /article -p 021 rain at 18°C (f). At the d f/1 end of the internalization 2 5 period the cells were washed /1/6 7 and ligand bound to cell sur- /1 2 face receptors stripped with 61 3 trypsin at 4°C. The cells were 60 /6 fixed with paraformaldehyde 7 .p and stained for IgA, for the df b tight junction protein ZO-1, y g u and for the nucleus. Individ- e ual sections, obtained with a st on scanning laser confocal mi- 23 F croscope, are shown from the e b basal portion of the cell below rua the nucleus (level 4), from the ry 2 0 lateral surface of the cell at the 23 level of the nucleus (level 3), from the apical region of the cell above the nucleus (level 2), and from the apical pole of the cell at or above the level of the tight junctions (level 1 ). The samples were scanned simultaneously for FITC and Texas red (or propidium io- dide) emission which are dis- played in the left and right halves of each panel, re- spectively. All images are at the same magnification. Bar, I0/zm. ehT lanruoJ of Biology, Cell emuloV 125, 4991 27 (Casanova et al., 1991; Aroeti et al., 1993). In contrast, if lateral surface of the cell while at the same time trypsin was IgA is delivered directly from basolateral to apical endo- included in the apical medium. As shown in Fig. 4 (d and somes then addition of apical trypsin should have littleef fect e) the delivery of IgA to the apical compartment of these on the ability of the IgA to enter the apical endosomal com- trypsin-treated cells occurred largely undiminished at either partment. 37* or 18°C, respectively. These results suggested that under To distinguish between the two possibilities described these conditions much of the IgA was being delivered abovwee first needed to confirm that trypsin treatment could directly to this apical compartment. We have also performed be used to prevent internalization of IgA from the apical cell the analysis described in Fig. 2 (a-d) in the presence of api- surface. [l~I]IgA was prebound to pIgR molecules at the cal trypsin. Similar results were obtained, although as ex- apical cell surface at 40C, and then the cells were warmed pected there was a more rapid decrease in IgA signal (data up for 1-5 min at 37°C in the absence or presence of 25 not shown). As an additional control for the effectiveness of #g/ml of trypsin. As shown in Fig. 4 a, in the absence of the apical trypsin treatment in preventing internalization of trypsin, the [t2sI]IgA was internalized, albeit inefficiently, plgR present at the apical plasma membrane, Fab fragments during the 37°C incubation. However, in the presence of tryp- were added in conjunction with the trypsin to the apical sur- sin no internalized AgI]I521[ was detected during the same face of the cellss howinn Fig. 4 (d and e). Under these condi- 5-min internalization period (Fig. 4 b). In addition, the vast tions internalization of Fab fragments was completely majority of IgA (>95 %) was rapidly released into the apical prevented. Compare untreated cells in the left panel of Fig. medium even after a 1-min incubation period in the presence 3 a with trypsin-treated cells in the left panel of Fig. 4 d. of the proteinase. These observations demonstrate that tryp- We have performed several additional controls to verify sin is able to rapidlyr emove IgA bound to plgR from the cell the effectiveness of the apical trypsin in preventing IgA-pIgR surface before it has a chance tob e internalized, and that the complexes from being internalized from the apical plasma D o binding of IgA to the pIgRd oes not render the plgR insensi- membrane. First, identical results were obtained using as lit- wn lo tive to trypsin treatment. tle as 5 t~g/ml trypsin, indicating that we were working in ad e In addition, we have analyzed the effect of apical trypsin considerable excess of the trypsin needed to cleave the pIgR. d fro on the fate of a 5-min pulse of basolaterally internalized Second, Fab fragments bind to pIgR-IgA complexes (Le- m h [mI]IgA. As is shown in Fig. 4 c the addition of trypsin in- maitre-Coelho et al., 1981; and Bomsel, M., and K. E. ttp creases the initial rate of ligand transcytosis over that ob- Mostov, unpublished results), so if pIgR-IgA complexes ://rup gsesetsr ved that in the in absence the absence of proteinase. of trypsin This a observatfiorna ction of sug- the tirzaantsiione,n tly the appFeaabr ed fragments at the could apical still surface bind. befoTrhei rd, reinwtheernna l- ana- ress.org transcytosing IgA is being reinternalized from the apicalc ell lyzed by PAGE, trypsin-treated baF]I~21[ fragments re- /jcb /a csaulr facea nd secretions presumably of the cell. recycling It also pointwsh ile in out the transit importancet o the api- amdadiintieodn , intact if soybean and trypsiwne re not inhibitor degraded was added (not shotwno). the tryp- In rticle-p d of using apical trypsin to prevent reinternalization of the sin-Fab medium after the incubation at 37 ° or 180C, the Fab f/1 2 pIgR from the apical cell surface. fragments were rapidly internalized to the same extent as 5/1 /6 Having confirmed that trypsinc ould be effectively used to non-trypsin-treated Fabs (not shown). This indicatetsh at the 7/1 prevent ligand internalization we determined the effect of effect of trypsin is to inactivate the ability of the plgR to bind 26 1 3 apical trypsin on IgA accumulation in the apical compart- Fab fragments, and not to inactivate the Fab fragments. 60 /6 ment. IgA was internalized continuously from the baso- To confirm that IgA was being delivered to an apical en- 7.p d f b y g u e st o n 2 3 F e b ru a ry 2 0 2 3 Figure .3 AgI delivered is to an compartment. endosomal apical AgI saw internalized from the basolateral surface of the cells for 30 min at and 370C anti-SC last during the apically internalized simultaneously were fragments Fab 01 rain of period incubation the (a). Alterna- tively, AgI saw basolaterally and internalized baF apically fragments for 021 rain at (b). the end 18°C At of experiment the the were cells and washed cell surface ligand stripped from the apical cell surface with trypsin. Cells with were paraformaldehyde, stained fixed with the appropriate antibodies, and scanned simultaneously for red emission which FITC and Texas are in displayed the and left right halves of each panel, Sections are respectively. at or the above level of junctions. the tight Images are at the same magnification. Bar, 10/zm. acadopA te .la sisotycsnarT of AgI Is aiv Apical semosodnE 37 dosomal compartment under conditions where apical inter- al., 1989). Under these conditions there was significant nalization of the ligand is prevented, we have performed the colocalization of the two ligands in the apical region of the following experiments. IgA was internalized basolaterally cell, both at the level of the tight junctions (Fig. 5 a), and for 20 min at 370C. The apical surface of the cell was bi- in the apical cytoplasm below the level of the tight junctions otinylated and avidin-TRITC was added apically while IgA (Fig. 5 b). Avidin-TRITC was not detected below this level was internalized basolaterally for an additional 10 min. (data not shown). In addition, colocalization of the two Trypsin was included in the apical medium throughout the markers was observed if the period of internalization of the 30-rain internalization period. The internalized avidin-TRITC two markers was limited to just 10 min at 37°C (data not serves as a general membrane marker for the apical en- shown), confirming that a large fraction of the IgA was rap- dosomal compartment and unlike Fab fragments internaliza- idly being delivered directly to an apical early endosomal tion of avidin-TRITC is not prevented by the action of tryp- compartment. sin, as the pool of biotinylated proteins (and lipids) is largely To demonstrate that colocalization of IgA and avidin was trypsin resistant. The 10-min internalization period is long occurring at the ultrastructural level, IgA-HRP was internal- enough to fill the apical early endosomal compartment but ized for a total period of 30 min at 37°C, while avidin-gold short enough to prevent entry of the avidin-TRITC into the was internalized for the final 10 min from the biotinylated prelysosomal compartment (Bomsel et al., 1989; Parton et apical pole of the cell. Apical trypsin was included through- D o w n lo a d e d fro m h ttp ://ru p re ss.o rg /jcb /a rticle -p d f/1 2 5 /1 /6 7 /1 2 6 1 3 6 0 /6 7 .p d f b y g u e st o n 2 3 F e b ru a ry 2 0 2 3 Figure .4 Trypsin prevents internalization of IgA from the apical cell surface. In a and b [125I]IgA was bound to the apical surface of cells for 90 min at 4°C, the cells were washed, and then reincubated at 37°C in the absence (a) or presence (b) of 25 #g/rrd of trypsin added to the apical chamber of the Transwell. At the indicated times the cells were rapidly cooled on ice and the apical media was collected. [nsI]IgA was stripped from the cell surface by using acid treatment at 4°C and the filters were cut out of their holders. The total AgI]Is2~[ initially bound to the cells included ligand released into the apical medium, ligand stripped from the cell surface with acid, and cell- associated ligand not sensitive to stripping (endocytosed), and was quantitated in a gamma counter. Values are from duplicate filters and varied <10%. At time 0, prior to warming the cells up to 37°C, virtually 100% of the ligand was at the cell surface. In c AgI]I52~[ was internalized from the basolateral surface of the cells for 5 rain at 37°C, the cells were washed, and then chased for 120 rain in the absence (-) or presence of (+) apical trypsin. The percent of total ligand released apically (tmnscytosed) and basolaterally (recycled) are shown. The remainder of the counts were intracellular. Values are from duplicate filters and varied <10 %. In d-e IgA was internalized basolaterally for 30 rain at 37°C or 120 rain at 18°C, respectively, while Fab fragments plus 25 #g/ml of trypsin were added apically. Cells were fixed with paraformaldehyde, stained with the appropriate antibodies, and scanned simultaneously for F1TC and Texas red emission which are displayed in the left and right halves of each panel, respectively, d and e are projections (sums) of four sections from the apical pole of the cell. Use of projections gives one an overview of the distribution of a particular marker in a single image. Images are at the same magnification. Bar, 01 #m. ehT lanruoJ of Cell ,ygoloiB emuloV 125, 4991 47 D o w n lo a d e d fro m h ttp ://ru p re ss.o rg /jcb /a rticle -p d f/1 2 5 /1 /6 7 /1 2 6 1 3 6 0 /6 7 .p d f b y g u e st o n 2 3 F e b ru a ry 2 0 2 3 Figure .5 IgA is delivered to an apical endosomal compartment under conditions that prevent internalization of the ligand from the apical cell surface. In a and b IgA was internalized basolaterally for 30 min at 37°C while avidin-TRYIC was added during the last 10 min of the incubation periotdo the biotinylated apical pole of the cell. Trypsin was included in the apical pole of the cell throughout the internaliza- tion period. At the end of the experiment cell surface avidin-SS-biotin complexes were stripped from the cell surface with reduced glutathione. Cells were fixed with paraformaldehyde, stained with the appropriate antibodies, and scanned simultaneously for FITC and TRITC emission which are displayed in the left and fight halves of each panel, respectively, a is a section at the level of the tight junctions. b is a section above the nucleus, approximately 2-gm below the level of the tight junctions. Arrows are intended as landmarks to guide the reader in identifying regions of colocalization. All images are at the same magnification. In c IgA-HRP was internalized from the basolateral cell surface for 30 rain at 37°C. Avidin-gold, bound to the biotinylated apical pole of the cell, was internalized apically during the last 01 rain. Trypsin was included in the apical medium throughout the internalization period. Meeting was observed in tubulovesicular elements close to the apical cell surface. Cells were processed for electron microscopy as described in the Materials and Methods. Solid arrows, examples of structures in which IgA-HRP and avidin-gold colocalize; open arrows, examples of structures in which only IgA-HRP is found; arrowheads, examples of areas where avidin-gold is found in the absence of IgA-HRP. Bars: (a and b) 01 #m; (c) 0.5 #m. acatxxlA et al. sisotycsnarT of lgA Is aiv Apical semosodnE 57 out the 30-rain internalization period. The IgA-HRP was expect maximum colocalization of the two markers, and in- detected by DAB cytochemistry. IgA-HRP was found in deed observe a high degree of colocalization of the two mark- vesicular and tubular elements distributed throughout the ers in apical endosomes (Fig. 6 a). Our maximal levels for cytoplasm but concentrated in the apical region of the ceils colocalization, as determined by the HRP cross-linking as- when semi-thick sections (200-225 urn) of epon-embedded say, are comparable to or even better than those obtained in cells were examined. Particularly in the apical region of the many similar studies (Ajioka, 1986; Ward et al., 1990). ceil, just under the apical plasma membrane, there were When we normalize to the maximum amount of mIIgA large accumulations of short tubules that contained both found in the apical endosomal compartment, we estimate avidin-gold and IgA-HRP (Fig. 5 c, solid arrows). This ob- that 80.5 + 2.39% of AgII~zp internalized for 30 min at servation confirms the IgA-HRP is entering an apical early 37"C resides in an apical endosomal compartment (Fig. 6 b). endosomal compartment. Colocalization also occurred in This result is consistent with our morphological analysis in more vesicular organeiles close to the plasma membrane which the majority of transcytosing IgA resides in the apical (Fig. 5 c) and occasionally in perinuelear multivesicular region of the cell in a compartment that overlaps with apical bodies. endosomes. Moreover, we obtain a very similar result when Although it was possible to find regions of IgA-HRP that AgIIs2~ is internalized basolateraliy and avidin-HRP api- had no gold (Fig. 5 c, open arrows), it was less common to cally for only 01 rain at 37"C (>80%). In addition, AgIq~l find avidin-gold in the absence of the other marker (Fig. 5 was also delivered to apical endosomes following a 2-h inter- ,c arrowheads). We believe this reflects the inefficient inter- nalization period at 180C. Under these particular conditions nalization of the avidin-gold particles. In addition, some of we estimate that approximately 34.4 + 2.0% of the ligand the IgA-HRP structures could represent tubules or vesicles had been delivered to the apical endosomes (Fig. 6 b). This derived from basolateral early endosomes not yet fused with observation is consistent with our morphological evidence D o the apical endosomal compartment, recycling vesicles de- (see Fig. 2 f) that although IgA is delivered to apical endo- wn lo rived from basolateral endosomes, or some other uniden- somes at this reduced temperature, a large fraction of the ad e tiffed compartment. In these examples >90% of the avidin- ligand remains in basolateral compartments that are presum- d fro gold particles are found to colocalize with the IgA-HRP. ably inaccessible to apicaily internalized ligand. The lower m h Avidin-gold was not detected in the basal portions of the level of colocalization at 18"C may also reflect a less efficient ttp call. Similar results were obtained if the two markers were internalization of ligand that occurs at 18"C. ://ru p internalized tion, we observe for 120 significant min at 18"C colocalization (data not shown). between In addi- IgA- formed As a control both a for morphological the validity of analysis our assay of we IgA have transcytosis also per- ress.org HRP and apically internalized ricin-gold (data not shown) and the HRP density-shift assay in nocodazole-treated cells. /jcb /a winternalized hich is a non-specific and recycled membrane at the apical marker cell surface that is (van efficiently Deurs ternalized As demonstrated IgA and in apically Fig. 6 (c-d) internalized meeting Fab of basolaterally fragments was in- rticle-p d et al., 1990). prevented in nocodazole-treated ceils. As expected, the IgA f/1 2 remained along the basolateral cell surface (Fig. 6 c) and was 5/1 not delivered to the cytoplasm at the apical pole of the cell /67 Biochemical Estimation of the Amount of lgA (Fig. 6 d). The Fab fragments were found dispersed across /126 Delivered to the Apical Endosomai Compartment the apical region of the cell in small vesicles (Fig. 6 d, left). 136 0 oT quantitate the amount of IgA delivered directly to apical These results confirm previous observations that the translo- /67 .p endosomes we have used a modification of the DAB density- cation of IgA from the basolateral to apical pole of the cell df b shift protocol (Courtoy et al., 1984). In our assay p2qIgA requires microtubules, and is largely inhibited in nocoda- y g u was internalized basolaterally for a total period of 30 rain at zole-treated ceils (Breitfeld et al., 1990; Hunziker et al., est o 37"C to completely fill all of the IgA-accessible compart- 1990). Similar results were observed if the experiments were n 2 ments. Avidin-HRP was internalized during the last 01 rain performed at 37"C (data not shown). 3 F e of the incubation period from the biotinylated apical pole of As predicted, entry of pzqIgA into apical endosomes bru a the cell. Apical trypsin was included throughout the inter- was inhibited sevenfold at both *73 and 18"C when the DAB- ry 2 nalization period. At the end of the experiment, ligand was density shift assay was performed in nocodazole-treated cells 02 3 removed from the ceil surface, and the cells were treated (Fig. 6 b). Again, these estimates were normalized to the with DAB and .2021-1 When ceils are treated in this manner, maximum amount of mIIgA found in the apical en- tzqIgA present in the avidin-HRP filled apical endosomal dosomal compartment. This observation confirms that the compartment is cross-linked by the DAB reaction into a majority of IgA was being delivered from basolateral endo- dense, detergent-insoluble complex. When the cells are solu- somes to apical endosomes, and that the process was bilized in SDS the p~IIgA present in DAB cross-linked microtubule dependent. In addition, the sevenfold inhibition apical endosomes is recovered by centrifugation. AgIqz~ of meeting in drug-treated cells confirms that tzqIgA was present in other compartments remains soluble and is found only density shifted when both the pzqIgA and avidin- in the supernatant following centrifugation. The percent of HRP were in the same compartment. Note that endocytosis IgA in the pellet represents that fraction of IgA present in the from both poles of the cell is largely unaffected by drug treat- apical endosomal compartment. It is well established that ment and the amount of t~IIgA internalized by the DAB cross-linking is not 100% efficient, even for markers nocedozole-treated ceils actually increased (data not shown) that are completely colocalized (Courtoy, 1984; Ajioka, (Breitfeld et al., 1990; Hunziker et al., 1990). 1986; Ward et al., 1990). As such our values were normal- As additional controls for the specificity of the DAB ized to reactions in which IgA and avidin-HRP were co- density-shift assay we report that there was no shift when internalized from the biotinylated apical pole of the cell (see avidin-tLRP was not internalized, and when t2qIgA and legend to Fig. 6 for these values). Under these conditions we avidin-HRP were bound to opposite poles of the ceil at 4"C, ehT lanruoJ of lleC ,ygoloiB emuloV ,521 4991 67