Transcytotic vesicular carriers for polymeric iga receptors accumulate in rat hepatocytes after bile duct ligation

In rat hepatocytes, the polymeric IgA receptor (plgA-R) mediates the transcytosis of polymeric IgA (plgA) from the blood to the bile (Brown and Kloppel, 1989). Polymeric IgA receptors expressed on the sinusoidal plasma membrane (the equivalent of the basolateral plasmalemmal domain in other epithelial cells) bind to circulating plgA. Receptor-ligand complexes thus formed as well as vacant receptors are apparently constitutively internalized via coated pits (Courtoy et al. 1983). Following a sorting event in an endosomal compartment (Geuze et al. 1984), plgA-R and receptor-ligand complexes are rapidly transported across the cell to the bile canalicular plasma membrane (the equivalent of the apical domain in other epithelial cells), via smooth-surfaced transcytotic carriers as demonstrated by a variety of techniques (Hoppe et al. 1985; Renston et al. 1980, 1988; Takahashi et al. 1982). At some point along the transcytotic pathway, most probably upon arrival at the bile canalicular plasma membrane (Musil and Baenziger, 1987), the receptor is cleaved and its 80×10³Af_r ectodomain fragment (known as secretory component) is secreted into the bile either unoccupied or complexed to plgA.

The plgA-R form found in rabbit hepatocytes and mammary gland cells is a member of the immunoglobulin superfamily (Hunkapiller and Hood, 1986; Williams, 1984). It has extensive similarity to the heavy chain of IgG. It consists of a large ectodomain with 629 amino acids, that binds plgA, a short (23aa) hydrophobic transmembrane sequence, and a 103 amino acid endodomain (Mostov et al. 1984; Mostov et al. 1980). In vitro studies have shown that the plgA-R is synthesized as a transmembrane glycoprotein, which is the precursor to the 80×10³M_r secretory component found in external secretions (Mostov et al. 1980; Mostov and Blobel, 1982).

Consistent with these in vitro studies, experiments carried out in the intact rat have identified three transmembrane forms of the hepatic plgA-R having molecular masses of 105×10³M_r, 116×10³M_r and 120×10³M_r (Sztul et al. 1985a,b). According to present evidence, the 105×10³M_r polypeptide is the newly synthesized, core-glycosylated form of the plgA-R. This precursor is transported from the rough endoplasmic reticulum to the Golgi complex and is expected to provide a marker for vesicular carriers operating between these two organelles. In the Golgi complex, the 105×10³M_r form is terminally glycosylated to the 116×10³M_r intermediate. Presumably, distinct vesicular carriers transport the 116×10³M_r form to the sinusoidal plasmalemma where it is found together with the 120×10³M_r form. The latter is kinetically defined as the mature receptor and the precursor to the 80×10³M_r biliary receptor.

We have previously shown that the 120 × 10³M_r polypeptide is the phosphorylated derivative of the 116×10³M_r intermediate and that the phosphorylated residues are serine(s) located in the cytoplasmic tail (Larkin et al. 1986). More recently, it has been shown that, in transfected MDCK cells expressing an exogenous plgA-R in which the phosphorylatable serine 664 is replaced by alanine, the rate of transcytosis is slowed and receptors recycle to the basolateral domain. This latter study identifies the phosphorylated serine 664 as the signal responsible for targeting the receptor into the transcytotic pathway (Casanova et al. 1990). Taken together, these findings validate the 120 × 10³M_r form of the plgA-R as the transcytotic itinerant and, therefore, as a suitable membrane marker for transcytotic vesicular carriers.

In the present study, we have used cell fractionation to validate further this assumption and bile duct ligation to manipulate the in vivo transcytotic pathway in rat hepatocytes for the purpose of accumulating transcytotic vesicular carriers. Our biochemical and morphological findings indicate that bile duct ligation results in a substantial accumulation of this class of vesicular carriers in the pericanalicular cytoplasm of hepatocytes. We have provided initial characterization of these transcytotic carriers with regard to their ultrastructural morphology and to two passenger proteins, plgA-R and plgA.

Male Sprague-Dawley rats (120–150 g) were purchased from Carnrn Laboratory Animals (Wayne, NJ) and Charles River Laboratories, Inc. (Wilmington, MA). All animals used in biochemical studies were allowed free access to food and water. Animals used for morphological studies were deprived of chow for 12–18 h prior to sacrifice to reduce the glycogen content of hepatocytes and thereby improve the results obtained in specimen preparation procedures.

Reagents and supplies were obtained from the following sources: leupeptin, phenylmethylsulfonyl fluoride (PMSF), hemoglobin, and protein A-Sepharose beads from Sigma Chemical Co. (St Louis, MO); ¹²⁵I-labeled staphylococcal protein A (2–10 μCi μg⁻¹ (immunological grade)] from New England Nuclear (Boston, MA); nitrocellulose filters (BA 85, 0.45 gm pores) from Schleicher & Schuell (Keene, NH); rabbit anti-rat IgA IgG from Calbiochem (La Jolla, CA); rhodamine-coryugated goat anti-rabbit IgG (H & L) from Tago, Inc. (Burlingame, CA); goat anti-rabbit IgG-Au₁₀ (GAR₁₀) from Janssen Life Sciences Products (Piscataway, NJ); alkaline phosphatase-conjugated goat anti-rabbit IgG (H chain specific) from Promega (Madison, WI); the phosphatase substrate system BCIP/PBT from Kirkegaard & Perry Laboratories, Inc. (Gaithersburg, MD); and SDS-PAGE reagents from Bio-Rad Laboratories (Richmond, CA)

Cholestasis was obtained by bile duct ligation as described by Easter et al. (1983). Under Nembutal anesthesia (30 mg kg⁻¹ i.p.) and through a longitudinal mid-line incision in the abdominal wall, the liver was exposed and the stomach and intestines were repositioned to gain access to the common bile duct which was: (1) dissected free of the portal vein and connective tissue; (2) ligated in two places along its length with silk thread; and (3) transected between the ligatures. After replacing the viscera in the abdominal cavity, the incision was closed with two sets of Michel clips, one applied to the muscle layer and the other to the skin. Cholestasis was verified by two criteria: (1) monitoring the change in urine color from clear to dark yellow/brown, indicating the appearance of biliary pigments therein; and (2) post-mortem examination that showed that the ligatures remained in place and the proximal stump of the duct was distended by accumulated bile. Sham-operated rats were subjected to all the steps of the procedure except ligation and transection of the bile duct. Pilot experiments showed that untreated and sham-operated rats were equivalent controls having the same receptor distribution in hepatic membranes, serum and bile (data not shown). Thus, these two types of controls were used interchangeably in experiments. All surgical procedures followed aseptic, not sterile, technique. Rats subjected to surgery were allowed to survive for one to three days before being killed to collect specimens for biochemical or morphological studies.

Livers were removed from untreated control, sham-operated and cholestatic rats, placed in ice-cold buffer A (0.1M Tris-HCl, pH 7.4, 0.25 M sucrose supplemented with 0.1 mM leupeptin and 1 mM PMSF), and minced with scissors. The mince, washed free of blood, was homogenized in buffer A to give a 30% (w/v) homogenate, which was centrifuged (7000g’, 10 min, 4 °C) to pellet large particles. The recovered supernatant was centrifuged at 105 000 g for 90 min at 4 °C to yield a total microsome (TM) pellet, which contains all cellular compartments involved in receptor synthesis, processing and transport (i.e. smooth and rough endoplasmic reticulum, Golgi complex, vesicular carriers and plasma membrane-derived vesicles). The TM pellet was resuspended in buffer A by gentle homogenization (Larkin et al. 1986). Cell fractions (i.e. Golgi light and Golgi heavy, residual microsomes, and a crude vesicular carrier fraction (collected in a 1.15 M sucrose layer)) were isolated from resuspended TM by a previously reported modification (Howell and Palade, 1982) of the procedure of Ehrenreich et al. (1973).

Bile was collected via a cannula (PE-50 tubing, Clay Adams, Parsipanny, NJ) inserted into the common bile ducts of Nembutal-anesthetized, untreated control and sham-operated rats. In the case of cholestatic animals, bile (<50 μl/rat) was collected from the distended stumps of ligated bile ducts using a 1cm³ syringe with a 26 gauge needle. Blood (∼0.5μl/rat) was withdrawn (under ether anesthesia) from the portal vein prior to liver removal. The serum was separated from the clot by lowspeed centrifugation.

Cell fractions were solubilized by boiling for 2 min in 0.4 % SDS in water (final concentration). After cooling, each lysate was subjected to a low-speed centrifugation and the decanted supernatant received an addition of a Triton X-100 mixture (2 % Triton X-100, final concentration) (Sztul et al. 1985b) to sequester the SDS. Polymeric IgA-R were immunoprecipitated from the lysates using a 1:30 antiserum dilution to generate antigen-antibody complexes, which were absorbed onto protein A-Sepharose beads. The beads were collected by low-speed centrifugation, washed and boiled for 2 min in sample buffer (62.5 mM Tris-HCl, pH 6.8, 2.7% SDS, 100 mM dithiothreitol, 10% sucrose (final concentrations)) to release antigen-antibody complexes, which were then subjected to SDS-PAGE.

Samples of hepatic TM, cell fractions, serum and bile were boiled for 2 min in sample buffer. The proteins of each sample were separated by SDS-PAGE in 5% to 10% (22 cm × 23 cm × 1.5 mm) or 5% to 15% (20cmxl5cmx0.8mm) gradient gels as described by Maisel (1971). In the larger gels, the loads were electrophoresed at a constant current of 25 mA for 12–15 h, whereas in the smaller gels they were electrophoresed at 50 V through the stacking gel and then at a constant current of 25 mA through the resolving gel for ∼2.5 h.

Proteins separated by SDS-PAGE were transferred to nitrocellulose filters. In the experiments in which antigens were detected with ¹²⁵I-labeled protein A followed by autoradiography, the transfers were quenched with hemoglobin and processed as described (Gershoni and Palade, 1982). When antigens were visualized by the enzymatic detection of alkaline phosphatase-tagged antibodies, transfers were quenched with Blotto and processed as described (Johnson et al. 1984). For all immunoblotting studies, anti-pIgA-R antiserum, used at the dilution stated in the figure legends, was prepared as described by Papermaster et al. (1976) and by Larkin et al. (1986). Quantitative densitometrie analysis of the autoradiograms of immunoblots were performed on a Bio Rad model 620 Video Densitometer.

Under ether anesthesia, a laparotomy was performed and a cannula, connected to a constant pressure perfusion apparatus, was inserted into the portal vein. The caudal vena cava was cut above the diaphragm to allow drainage (in the thoracic cavity) during the rest of the procedure. The blood was removed from the liver by perfusing the organ through the portal vein at a pressure of ∼40mmHg (1 mmHg=0.133kPa) with PBS, pre-warmed to 37 °C. The liver was then perfused for 2 min with 2% formaldehyde, 0.75 M lysine, 0.01M sodium periodate in phosphate buffer prepared as described by McClean and Nakane (1974). Small selected parts of the organ were removed, cut into ∼1mm pieces, further fixed by immersion for a total of 2h and then cryoprotected by infiltration (1 h) with 2.3 M sucrose in phosphate buffer containing 50 % polyvinylpyrolidone (Schnabel et al. 1989; Tokuyasu, 1986). Infiltrated tissue specimens were mounted on aluminum nails, frozen in liquid nitrogen, and sectioned on a Reichert Ultracut E Ultramicrotome equipped with an FC-4E cryoattachment.

Semithin frozen sections (0.5 gm) were transferred to glass slides coated with poly-L-lysine (M_r 40500; lmgml⁻¹), then incubated overnight at 4°C (in a humidified chamber), either with rabbit anti-rat plgA-R antibody or rabbit anti-rat plgA antibody (1:1000 dilution) in PBS/1% albumin (pH7.4) followed by incubation for 1 h at room temperature with rhodamine-conjugated goat anti-rabbit serum (1:50 dilution). After washing with PBS, the sections were mounted in 90% glycerol in PBS containing 0.1 % p-phenylenediamine, and examined by epifluorescence using a Zeiss Photomicroscope II and Kodak Tri-X-Pan (ASA 400) film.

The liver was fixed in situ by perfusion with 3% glutaraldehyde in 0.1M cacodylate buffer, pH 7.4, after which selected samples were excised and cut into small blocks, which were immersed in the same fixative for 6 h. As described by Milici et al. (1987), the tissue was processed sequentially through 1 % osmium tetroxide, 2% uranyl acetate, standard dehydration in ethanol and embedding in Epon. Thin sections were stained with uranyl acetate and lead citrate prior to being examined and micrographed in a Phillips 301 electron microscope.

The liver was fixed in situ by perfusion with 3% formaldehyde and 0.05% glutaraldehyde in 0.1M cacodylate buffer, pH 7.4, and processed through cryoprotection and mounting on aluminum nails as described under Immunofluorescence studies, below. As described by Schnabel et al. (1989) and Tokuyasu (1986), thin (60–90 run) frozen sections were prepared and incubated at room temperature with anti-rat plgA-R antibody (1:100) for lh followed by incubation for an additional 1 h with colloidal gold (10 nm)-conjugated goat anti-rabbit IgG (1:50). After post-fixation for 10 min with 2% glutaraldehyde, the sections were stained with 2 % osmium tetroxide and 2 % uranyl acetate (15 min each) and ‘absorption-stained’ with 0.002% lead citrate in 2.2% polyvinyl alcohol for 5 min.

In preliminary attempts to separate different classes of vesicular carriers using different plgA-R forms as membrane markers, we used well-characterized cell fractionation techniques (Ehrenreich et al. 1973; Howell and Palade, 1982) to prepare from hepatic total microsomes (TM) a fraction that met our criteria for a crude vesicular carrier fraction (CVCF). This fraction was obtained by centrifuging resuspended TM in a discontinuous sucrose density gradient (G-l). Vesicles that floated out of the TM load into a 1.15 M sucrose layer were collected. They accounted for only 5 % of the protein of the TM load but contained all three forms of the receptor and were enriched in the mature 120×10³M_r form (Fig. 1, lane 2). (The remaining TM protein was distributed among other fractions of the G-l gradient (data not shown).) Moreover, biosynthetically labeled 120×10³M_r was found in this fraction 60 min after intravenous injection of [³⁵S]cysteine (Sztul et al. 1985a). These data are consistent with the 1.15 M crude vesicular carrier fraction being enriched in transcytotic carriers. We then subjected the 1.15 M CVCF to further subfractionation on shallow sucrose density gradients (0.85 M-1.22 M; G-2) in an attempt to separate transcytotic carriers (identified by the 120×10³M_r form) from exocytic vesicular carriers (identified by the 105 and 116×10³M_r forms) counting on possible differences in their buoyant densities. As shown in Fig. 1, the enrichment of the 120 × 10³M_r receptor in the more buoyant G-2 subfractions (lanes 3–5) demonstrates the feasibility of separating transcytotic carriers from mixed exocytic carriers (lanes 6–9). The results of these experiments also showed that in the starting preparation, each class of vesicular carriers was present in amounts small enough to make its biochemical analysis difficult.

We used the cholestatic rat liver as a model system in which to accumulate transcytotic vesicular carriers, since available data indicate that cholestasis causes a reduction in bile flow (Oelberg and Lester, 1986; Phillips et al. 1986). We expected that, under these conditions, the 120 ×10³M_r receptor form (the transcytotic carrier marker) would accumulate within hepatocytes. To test this hypothesis directly, hepatic TM were prepared from untreated control and cholestatic rats killed 1, 2 and 3 days after bile duct ligation. TM were solubilized, their proteins were separated by SDS-PAGE and the receptor forms present therein were detected by immunoblotting. The qualitative results are presented in Fig. 2A and their quantitation in Fig. 2B. In hepatic TM prepared from control liver, the receptor triplet (120, 116 and 105×10³M_r) was clearly resolved, whereas the 80 × 10³M_r biliary truncated form of the receptor was not detectable (Fig. 2A, lane C). In contrast, TM prepared from cholestatic livers 1, 2 and 3 days after duct ligation showed a progressive accumulation of the 116×10³M_r form and especially of the 120×10³M_r receptor in addition to still unidentified (but immunologically related) proteins (cholestatic antigens) with apparent molecular masses of 124, 90 and 80 × 10³M_r (Fig. 2A, ligated). Other immunoreactive bands that migrated in the gel immediately ahead of the 105× 10³M_r form (Fig. 2A, ligated, 3d) are most probably receptor precursors at various stages of core glycosylation. Similar variant forms, less than the 105 × 10³M_r precursor in mass, were observed in control TM preparations heavily labeled biosynthetically with [³⁵S]cysteine (data not shown).

Quantitative densitometrie analysis of the blot in Fig. 2A showed that, in TM prepared from 2- and 3-day cholestatic livers, the 120×10³M_r receptor form accumulated to levels 2.5- and 3-fold higher than in controls, respectively. The novel 124×10^sM_r cholestatic antigen, which was not detectable in control TM, accumulated in cholestatic TM to a level less than that of the 120 and 116×10³M_r forms (Fig. 2B)

The accumulation of the 120×10³M_r form suggested that receptor proteolysis and secretion into bile might be arrested or drastically curtailed in cholestatic hepatocytes. Thus, we compared the protein profiles of bile collected from control and 3-day cholestatic rats. Proteins, acetone precipitated from bile samples, were separated by SDS-PAGE, and visualized by Amido Black staining (Fig. 3, lane 2) or by immunoblotting (Fig. 3, lanes 3–8). Control bile contained the 80×10³M_r biliary form of the receptor, IgA (heavy and light chains) and albumin (Fig. 3, lane 2), each identified by Western blotting with relevant antibodies (Fig. 3, lanes 3, 5, 7). In contrast, in the bile collected from the dilated stumps of ligated ducts the concentration of the 80×10³M_r biliary receptor was markedly reduced (Fig. 3, lane 4), a result consistent with the demonstrated intracellular accumulation of its precursor, the 120×10³M_r receptor (Fig. 2). IgA chains were also reduced in concentration (Fig. 3, lane 6); however, albumin continued to be found in the same bile at concentrations comparable to or higher than in controls (Fig. 3, lane 8). The same applied to transferrin (data not shown).

Bile duct ligation caused a reduction in the secretion of the 80×10³M_r form into the bile, but the accumulation of the 120×10³M_r receptor in cholestatic hepatocytes was not as high as expected, assuming continued receptor synthesis over 2–3 days. These results suggested that the 120 × 10³M_r receptor is proteolytically processed in cholestatic hepatocytes but that the product of proteolysis (secretory component) ultimately comes to reside in a liver compartment (sinusoids?) other than the bile canaliculi. To check this assumption, we examined the blood from cholestatic rats, since: (1) hepatocytes are in direct contact with blood plasma in the spaces of Disse; and (2) a rise in secretory IgA in the blood is diagnostic for cholestasis (Lemaitre-Coelho et al. 1978; Jones et al. 1984). Bile and blood were collected from control and cholestatic rats killed 1, 2 and 3 days after duct ligation. Biliary and serum proteins were separated by SDS-PAGE and receptors were detected by immunoblotting. The results are shown in Fig. 4. By comparison to nonligated controls (Fig. 4, lanes 5 and 1), duct ligation caused the expected decrease in the secretion of the 80 × 10³M_r receptor form into the bile (Fig 4, lanes 6–8) and a concomitant rise in an 80×10³M_r immunoreactive protein in the serum (Fig. 4, lanes 2–4) Presumably, this novel serum protein is the 80×10³M_r biliary form of the receptor as judged by its immunoreactivity with anti-receptor antibody and by its electrophoretic comigration with the 80× 10³M_r biliary standard. The changes in the concentration of the 80 × 10³M_r receptor in either bile or blood were established within the first 24 h after duct ligation and remained at their respective levels, low in bile and high in the blood, throughout the time course of the experiment.

The accumulation of the 120×10³M_r receptor in TM prepared from cholestatic livers suggested that transcytotic vesicular carriers accumulate in cholestatic hepatocytes. To investigate this possibility, we performed indirect immunofluorescence tests to localize the plgA-R on frozen, semi-thin liver sections. In sham-operated controls, plgA-R positive sites were found along the sinusoidal plasma membrane and in cytoplasmic punctate clusters, presumably representing transcytotic vesicular carriers associated with the bile canaliculi. The canaliculi themselves were not resolved clearly (Fig. 5A). Lighter, punctate staining scattered throughout the cytoplasm might represent vesicular carriers shuttling receptors between other intracellular compartments. In the hepatocytes of 1-day cholestatic liver, labeling along the sinusoidal plasmalemma persisted, but the number of receptor-positive sites in the pericanalicular cytoplasm increased markedly (Fig. 5B). The label was clearly localized to punctate cytoplasmic sites rather than to dilated bile canaliculi; moreover, the lumina of many of the latter appeared to be devoid of label (Fig. 5, C-l). The pericanalicular distribution of the receptor-positive sites in 3-day cholestatic hepatocytes (Fig. 5, C-3) was identical to that seen in 1-day specimens.

In control hepatocytes, plgA was detected as fine dots scattered through the cytoplasm, and as particles clustered in the pericanalicular cytoplasm (Fig. 5D), most probably representing transcytotic vesicular carriers. The ligand was not detected on the sinusoidal plasmalemma as clearly as was the receptor, suggesting that most of the local copies of the latter were unoccupied. One day after duct ligation, plgA was localized to fine dots scattered throughout the cytoplasm but also accumulated close to bile canaliculi (Fig. 5E). In addition, the antigen was detected in sinusoid-associated cells (Fig. 5E), which were generally plgA-negative in all control specimens examined. By 3 days after duct ligation, however, plgA distribution had shifted dramatically from pericanalicular clusters of fine dots in hepatocytes to larger bodies and apparently diffuse cytoplasmic staining in sinusoid-associated cells (Fig. 5F), These findings suggested that, with advancing cholestasis, plgA either does not bind to, or dissociates from, hepatocytic plgA-R and is preferentially taken up by other cell types, i.e. sinusoidal endothelia.

The accumulation of plgA in the sinusoidal endothelia of cholestatic rats detected by immunofluorescence correlated with the appearance of large inclusions of varied density detected by electron microscopy in the same type of cells. Fig. 6A and B compares two representative areas from sham-operated and 3-day cholestatic livers. Similar inclusions are found in large numbers in Kupffer cells (data not shown).

We assumed that the punctate clusters of receptor-positive sites localised by immunofluorescence experiments to the pericanalicular cytoplasm consisted of transcytotic vesicular carriers. To confirm the vesicular nature of these plgA-R labeled structures and to extend the studies of Rank and Wilson (1983), we performed bile duct ligations and sham operations on rats, fixed the livers in situ, and processed them for either conventional transmission electron microscopy or immunogold cytochemistry to localize receptors. As assessed by transmission electron microscopy, the integrity of hepatocytes was maintained in the cholestatic condition. The cells showed clearly defined sinusoidal and biliary plasmalemmal domains separated by apparently intact tight junctions similar to those observed in sham-operated controls (Fig. 7A). In 2-day cholestatic livers, however, the bile canaliculi were either markedly dilated (micrograph not shown) or collapsed and provided with fewer microvilli (Fig. 7B) than in controls (Fig. 7A). These specimens also showed increased numbers of small pleiomorphic vesicles in the cytoplasm surrounding the bile canaliculi (Fig. 7B) as compared to controls (Fig. 7A). Our findings are generally consistent with the large amount of data on cholestatic liver morphology reported by other investigators (see Jones et al. 1976; Popper and Schaffner, 1970). However, in the time frame of our experiments, we observed neither vesicles nor deposits within the bile canaliculi (Renston et al. 1983), nor the modifications of the spaces of Disse (Collier et al. 1986) previously reported by other investigators.

Immunogold localization of plgA-R on frozen, thin sections showed that in the hepatocytes of 3-day sham-operated animals, the antigen was localized in small vesicles around bile canaliculi (Fig. 8A). In agreement with the immunofluorescence findings, the receptor was also detected on the sinusoidal plasmalemma and associated coated pits and vesicles and at low concentrations in endoplasmic reticulum (ER) and Golgi elements (data not shown). In 3-day cholestatic specimens, plgA-R were detected in a larger number of pleiomorphic vesicles, both spherical and tubular, located along and around bile canaliculi (Fig. 8B). The diameter of these vesicles was generally larger than those in control hepatocytes. Moreover, the number of gold particles (i.e. detectable antigenic sites) per vesicle was markedly increased by comparison with the controls. These findings validated the assumption that the fine dots seen by immunofluorescence represent vesicular structures, specifically marked by anti-pIgA-R antibody. The receptor was also detected on the biliary plasmalemma, in bile capillaries (Fig. 8B), and, as in controls, on the sinusoidal plasmalemma and on ER and Golgi membranes (data not shown). But there was no indication of plgA-R accumulation at any other intracellular site besides the pericanalicular vesicular carriers.

We have used bile duct ligation to manipulate the in vivo transcytotic pathway in rat hepatocytes for the purpose of identifying and accumulating transcytotic vesicular carriers. This first study should facilitate subsequent isolation and characterization of these important transport organelles. The 120×10³M_r form of the plgA-R was used as a membrane probe to identify directly transcytotic carriers, thereby avoiding inferences based on plgA alone or on non-specific fluid phase markers. This distinction is important, given the reported dissociation of plgA from plgA-R transport in hepatocytes (Kloppel et al. 1987; Fig. 5, present study) and in transfected MDCK cells expressing an exogenous plgA-R (Breitfeld et al. 1989).

In the present study we have demonstrated the accumulation of plgA-R-carrying vesicles in the pericanalicular cytoplasm of rat hepatocytes after ligation of the common bile duct. Several lines of evidence indicate that these accumulated, receptor-positive vesicles are transcytotic vesicular carriers en route from the sinusoidal plasmalemma to the bile canalicular plasmalemma. First, by immunoblotting we have shown that the concentration of the 120×10³M_r receptor form increased in TM prepared from cholestatic liver to levels ∼threefold higher than in controls, whereas the accumulation of other receptor forms was considerably less pronounced (Fig. 2). Significantly, the 120×10³M_r receptor form has been validated as the transcytotic itinerant by cell fractionation (Fig. 1), phosphorylation (Casanova et al. 1990; Larkin et al. 1986), and kinetic (Sztul et al. 1985b) studies. Second, by immunofluorescence tests we have shown that the increase in liver plgA-R content occurred concomitantly with the pericanalicular accumulation of sites positive for plgA-R (Fig. 5A-C, Fig. 8) and plgA (Fig. 5, D-E). These results are in general agreement with those of Wilson et al. (1980) and Rank and Wilson (1983). Furthermore, our findings are similar to those of Barr and Hubbard (1989), who reported the accumulation of vesicles positive for not only newly synthesized apical plasmalemmal markers but also plgA-R in the pericanalicular cytoplasm of rat hepatocytes after bile duct ligation. Third, by immunogold cytochemistry we have confirmed the vesicular nature of the plgA-R-carrying structures, which were generally larger and contained more antigenic sites than those in control cells, suggesting that fusion of transcytotic carriers may occur in response to cholestasis. Fourth, our morphological findings localize plgA-R along the entire transcytotic pathway; that is, at the sinusoidal plasmalemma (Fig. 5), in vesicular carriers, at the bile canalicular plasmalemma, and in bile canaliculi (Fig. 8). We assume that in the case of the latter the signal comes from the 80 × 10³M_r secretory component crosslinked by the fixative to the biliary plasmalemma. These findings suggest that after bile duct ligation, the transcytotic pathway is functionally intact, results consistent with an earlier report using horseradish peroxidase as a transcytotic probe (Renston et al. 1983). But the rate of trafficking must be low enough to permit accumulation of transcytotic carriers in the pericanalicular cytoplasm.

Recently, Breitfeld et al. (1989) have demonstrated biochemically the presence of uncleaved exogenous plgA-R on the apical plasma membrane of transfected MOCK cells and the apical endocytosis of plgA. On account of these findings, we can not rule out entirely the possibility that the receptor-positive vesicles visualized in Fig. 8 are involved in endocytosis from the bile canalicular plasmalemma. However, results obtained on transfected MDCK cells in vitro are not necessarily applicable to hepatocytes in situ. In the latter’s case, endocytosis -if extant -would involve the 80 × 10³M_r cleaved ectodomain rather than the entire plgA-R molecule (see below).

Finally, our biochemical and morphological findings indicate that the accumulation of transcytotic carriers in the cholestatic liver is a generalized phenomenon, since: (1) changes in receptor distribution were observed throughout all liver specimens examined by immunocytochemistry; and (2) the accumulation of the 120×10³M_r receptor was found in each TM preparation examined; that is, in samples that were large enough to render unlikely local variations that could affect the interpretation of our findings.

By immunoblotting, we have shown that the intracellular accumulation of the 120×10³M_r receptor was accompanied by drastically reduced secretion of the 80×10³M_r form into bile and by its appearance in detectable amounts in serum, a result consistent with the findings of Lemaitre-Coelho et al. (1978). Under the same experimental conditions, however, the biliary concentrations of albumin and transferrin remained similar to those in control livers. The underlying mechanisms responsible for this differential delivery of proteins to the bile warrant further study. One possible explanation is that plgA-R biogenesis is altered in response to cholestasis. Our preliminary biosynthetic studies indicate that the synthesis and processing of the plgA-R occur in cholestatic liver with kinetics similar to the control condition. Furthermore, newly synthesized 80×10³M_r secretory component appears in both bile and plasma of cholestatic rats (data not shown). However, the kinetics of these latter events remains to be clarified.

Another possible explanation is that cholestasis-induced changes in the structure (Anderson et al. 1989; Easter et al. 1983) or permeability (Boyer, 1983; Metz et al. 1977) of tight junctions may allow paracellular exchange of serum and biliary proteins. Alternatively, functional changes in tight junctions may cause hepatocytes to lose their plasmalemmal polarity by allowing lateral diffusion of domain-specific plasmalemmal proteins as suggested for bile salt carriers (Fricker et al. 1987) and plasma membrane antigens (Durand-Schneider et al. 1987). Redistribution of putative plasma membrane ‘docking proteins’ may alter intracellular targeting of vesicular carriers.

In addition, the redistribution of the 80 × 10³M_r receptor from bile to blood (Fig. 4) as well as the leveling off of plgA-R content in hepatocytes in cholestatic liver (Fig. 2) could be explained by ‘reverse transcytosis’; that is, apical to basolateral movement of the receptor. Apical endocytosis of plgA has been demonstrated in transfected MDCK cells (Breitfeld et al. 1989). Furthermore, the apical to basolateral transport of horseradish peroxidase in hepatocytes (Jones et al. 1984) and in MDCK cells (Parton et al. 1989) and of IgG complexes in rat neonatal intestinal epithelial cells (Abrahamson and Rodewald, 1981; Rodewald et al. 1983) and in transfected MDCK cells (Hunziker and Mellman, 1989) have been documented. A similar transport for plgA-R in hepatocytes remains to be established. Characterization of the cholestatic antigens found only in total microsomal fractions prepared from rat liver after bile duct ligation is one of our immediate objectives. The 90 and 80×10³Af_r antigens are similar in molecular weight to plgA-R forms reported as cleavage products in the intestinal epithelium (Ahnen et al. 1986). If this also obtains in the cholestatic liver, their presence would strengthen the case for apical endocytosis and reverse transcytosis, assuming that receptor cleavage occurs apically, as indicated by the continued presence (albeit in reduced amounts) of the 80 × 10³M_r receptor in the bile (Fig. 4).

As was the case with the receptor, bile duct ligation also caused a reduced secretion of plgA into bile (Fig. 3), a result in agreement with published data (Kloppel et al. 1987; Renston et al. 1980). Comparative tests showed that in controls and at 1 day after duct ligation, plgA-positive sites accumulated in the pericanalicular cytoplasm (Fig. 5E), a distribution similar, but not identical, to that of the receptor (Fig. 5B). We assume that the discrepancy reflects incomplete occupancy of the receptor by its ligand. However, with advancing cholestasis, the hepatocyte-associated plgA decreased while sinusoid-associated endothelial and Kupffer cells became intensely ligand positive (Fig. 5F). A similar redistribution of the plgA-R was not observed (Fig. 5C-1 and C-3). Our data on plgA distribution differ from an earlier report by Rank and Wilson (1983), who showed persistent, pericanalicular accumulation of plgA at 3 days after duct ligation. This apparent discrepancy remains to be resolved. However, our findings may be relevant to results published by Kloppel et al. (1987), who reported that after relief from short-term cholestasis, the liver regained its ability to secrete nearnormal levels of the 80×10³M_r receptor (secretory component) into the bile whereas plgA secretion remained depressed, suggesting that the majority of receptors processed for secretion into the bile was vacant under these experimental conditions.

At present, we do not have a satisfactory explanation for the dissociation of plgA from plgA-R transport. It may reflect restricted accessibility of plgA to the receptor because of the collapse of the spaces of Disse (Collier et al. 1986) or deposition of obstructive material therein (Ren-ston et al. 1983), although in our experiments such changes were not found consistently. Alternatively, it may reflect modifications of the ligand or activation of nonspecific receptors in the sinusoidal endothelium induced by cholestasis. Since the 80×10³M_r secretory component appears in the plasma as documented, ligandreceptor complexes may form in the circulating blood from which they may be cleared by macrophages and endothelial cells. This latter hypothesis, however, raises the question of why the ligand alone, and not the receptor fragment, is detected in sinusoidal endothelial cells by immunofluorescence. This issue remains to be elucidated; it may reflect a change in conformation that renders receptor epitopes inaccessible to their antibodies.

In conclusion, our morphological and biochemical data have confirmed the 120×10³M_r form of the rat hepatic plgA-R as the transcytotic itinerant and have shown that bile duct ligation is a useful technique for the intracellular accumulation of transcytotic carriers. These findings should facilitate the isolation of these important transport organelles. Chemical characterization of their membrane components is needed for identifying the molecules and understanding the mechanisms involved in sorting specific cargo proteins and targeting vesicular carriers to the membranes of their proper destination.

We thank Hans Stukenbrok of the Immunocytochemistry core facility (Yale) for his invaluable help with the immunogold studies, Patricia Sanchez (UCHC) and Linda Ketchman (Yale) for their technical assistance, Anne Curley-Whitehouse (Yale) for photography, Dale Stradley (Yale) for secretarial assistance, and Drs Jennifer Stow and Richard G. W. Anderson for critical reading of the manuscript. Special acknowledgment goes to Dr Marilyn Farquhar for her support and many contributions to this project.

This research was supported by a grant from the National Institutes of Health (GM 27303) to George E. Palade and by a Research Initiation Grant from UCHC to Janet M. Larkin.