Transcytosis of aminopeptidase N in Caco-2 cells is mediated by a non-cytoplasmic signal.

In Caco-2 cells, aminopeptidase N is transported to the apical membrane from the trans Golgi network by both the direct and the indirect pathway (Matter, K., Brauchbar, M., Bucher, K., and Hauri, H.-P. (1990) Cell 60, 429-437). The aim of this study was to determine the importance of the transmembrane or cytoplasmic domain of aminopeptidase N for transport of aminopeptidase N by the indirect pathway by analysis of mutated forms of aminopeptidase N recombinantly expressed in Caco-2 cells. A tail-less and two secretory forms of aminopeptidase N, all deprived of the cytoplasmic tail, were transported to the basolateral plasma membrane in proportions equivalent to the wild type enzyme. This shows that no cytoplasmic basolateral sorting signal is involved in directing aminopeptidase N to the basolateral plasma membrane. Both the wild type and the tail-less aminopeptidase N were transcytosed from the basolateral to the apical plasma membrane, whereas no transcytosis of two secretory forms could be detected, showing that the transmembrane domain is important for efficient transcytosis to take place. A significant difference in transcytosis kinetics of the human and the porcine wild type aminopeptidase N was observed. This indicates that transcytosis of aminopeptidase N from the basolateral to the apical membrane does not occur by default transport but involves an active sorting mechanism.

In Caco-2 cells, aminopeptidase N is transported to the apical membrane from the trans Golgi network by both the direct and the indirect pathway (Matter, K., Brauchbar, M., Bucher, K., and Hauri, H.-P. (1990) Cell 60, 429 -437). The aim of this study was to determine the importance of the transmembrane or cytoplasmic domain of aminopeptidase N for transport of aminopeptidase N by the indirect pathway by analysis of mutated forms of aminopeptidase N recombinantly expressed in Caco-2 cells. A tail-less and two secretory forms of aminopeptidase N, all deprived of the cytoplasmic tail, were transported to the basolateral plasma membrane in proportions equivalent to the wild type enzyme. This shows that no cytoplasmic basolateral sorting signal is involved in directing aminopeptidase N to the basolateral plasma membrane. Both the wild type and the tail-less aminopeptidase N were transcytosed from the basolateral to the apical plasma membrane, whereas no transcytosis of two secretory forms could be detected, showing that the transmembrane domain is important for efficient transcytosis to take place. A significant difference in transcytosis kinetics of the human and the porcine wild type aminopeptidase N was observed. This indicates that transcytosis of aminopeptidase N from the basolateral to the apical membrane does not occur by default transport but involves an active sorting mechanism.
The plasma membrane of epithelial cells is divided into an apical and a basolateral domain by tight junctions. Polarized epithelial cells are able to selectively direct membrane-bound proteins to either of these membrane domains. It has been shown that there are at least two pathways to the apical membrane: a direct pathway from the trans Golgi network and an indirect pathway from the trans Golgi network to the basolateral membrane followed by transcytosis to the apical domain (for a review, see Rodriguez-Boulan and Powell (1992)). Proteins transported to the apical and basolateral plasma membrane in polarized cells are generally believed to be guided by sorting signals. The best studied example of a protein using the indirect pathway is the polymeric immunoglobulin receptor in Madin-Darby canine kidney (MDCK) 1 cells. A cytoplasmic ba-solateral sorting signal directs the polymeric immunoglobulin receptor to the basolateral domain (Casanova et al., 1991a;Aroeti et al., 1993). Then, two cytoplasmic endocytosis signals ensure rapid endocytosis (Okamoto et al., 1992) to endosomes where the receptor may recircle to the basolateral membrane domain again guided by the basolateral sorting signal or transcytose to the apical membrane domain (Aroeti and Mostov, 1994). Transcytosis is promoted by ligand binding (Hirt et al., 1993;Song et al., 1994) and phosphorylation of Ser-664 (Hirt et al., 1993;Casanova et al., 1990). The transport to the apical plasma membrane seems to be signal mediated by ill-defined part(s) of the extracellular domain (Mostov et al., 1987).
Another protein, dipeptidyl peptidase IV, can also use the indirect pathway to achieve an apical expression in both MDCK cells (Casanova et al., 1991b) and Caco-2 cells (Matter et al., 1990). It has been reported that dipeptidyl peptidase IV in MDCK cells carries a basolateral sorting signal on the cytoplasmic domain and an apical sorting signal on the extracellular domain (Weisz et al., 1992). In both MDCK cells (Weisz et al., 1992) and Caco-2 cells (Matter et al., 1990), only part of the newly synthesized dipeptidyl peptidase IV molecules use the indirect pathway, whereas the rest uses the direct pathway to the apical membrane.
We have previously shown that aminopeptidase N (APN) carries an apical sorting signal on the extracellular domain directing the protein to the apical side in MDCK cells (Vogel et al., 1992a(Vogel et al., , 1992b, and thus, in this respect, APN resembles both dipeptidyl peptidase IV and the polymeric immunoglobulin receptor. In MDCK cells, it has been shown that the majority of newly synthesized APN molecules are transported via the direct pathway (Wessels et al., 1990). In Caco-2 cells, APN is transported to the apical side via both the direct and the indirect pathways (Matter et al., 1990;Le Bivic et al., 1990). We have set out to investigate the sorting signals directing the indirect transport of APN in Caco-2 cells with special attention to possible sorting signals in the cytoplasmic tail.
In the present study, we found that the transport of APN to the basolateral membrane is independent of sorting signals in the cytoplasmic domain. However, the transmembrane domain is necessary for transcytosis of APN from the basolateral to the apical plasma membrane. In addition, we show that the Caco-2 cell line, which is of human origin, is able to transcytose human APN more efficiently than porcine APN.

MATERIALS AND METHODS
Cell Culture and DNA Constructions-Caco-2 cells (a gift from M. Spiess, Biocenter, Basel) were grown in minimal essential medium supplemented with 2 mM L-glutamine, 10% fetal bovine serum, 1% non-essential amino acids, 100 units/ml penicillin, and 100 g/ml strep-tomycin. Subconfluent Caco-2 cells were transfected by the calcium phosphate transfection procedure (Graham and Van der Eb, 1973) with the modifications for adherent epithelial cells described in (Sambrook et al., 1989). The transfected cells were selected using 0.65 mg/ml G418, and stable clones were isolated using cloning cylinders. For most experiments, 10 6 cells per well were seeded into 24-mm Transwell chambers (Costar Europe Ltd. Badhoevedorp, The Netherlands), allowing separate access to the apical and basolateral membrane. The medium was changed every 2 or 3 days. Filters were used for experiments 3-6 days after confluence as indicated by tightness of the filter. Tightness of filter-grown cells was assayed by filling the inner chamber to the brim and allowing it to equilibrate overnight.
The construction of the anchor-minus and the anchor/stalk-minus human APN cDNA was previously described (Vogel et al., 1992a(Vogel et al., , 1992b. The construction of the cDNA for the tail-less and Tyr-6 porcine APN was introduced by polymerase chain reaction and sequenced to verify the construction. All cDNAs were introduced into the eukaryote expression vector pTEJ-4 (Johansen et al., 1990). Cells were transfected with 20 g of pTEJ-4 containing the relevant cDNA and 2 g of pSV2neo (Southern and Berg, 1982).
Antibodies-The monoclonal antibody G43 (a kind gift from Dr. B. Delmas, Institut National de la Recherche Agronomique, Jouy-en-josas France) was raised against porcine APN and does not cross-react with human APN (Delmas et al., 1994). The monoclonal antibody 43E6 (a kind gift from Dr. R. A. Ashmun, St. Jude Children's Research Hospital, Memphis, TN) reacts with human APN but not porcine APN.
Enzymatic Assay-The apical and basolateral media were collected and assayed for their aminopeptidase activity as described for microvillus aminopeptidase (Sjöström et al., 1978) using the substrate L-alanine-p-nitroanilide.
Antibody Binding Assay-The assay was performed on tight filtergrown monolayers of cells. All washing and antibody binding steps were performed at 4°C. Cells were washed with phosphate-buffered saline, 0.1 mM Ca 2ϩ , 1 mM Mg 2ϩ (PBS/CM) containing 0.5% bovine serum albumin (BSA) and incubated for 2 h with a saturating amount of antiserum G43 (diluted 1:250 in PBS/CM, 0.5% BSA). After rinsing the cells three times in PBS/CM containing 0.5% BSA, they were incubated for 30 min with sheep anti-mouse Ig, 125 I-labeled F(abЈ) 2 fragment antibody (10 7 cpm/ml) (Amersham) diluted in PBS/CM containing 0.5% BSA. Cells were then washed three times with PBS/CM, 0.5% BSA, twice with PBS/CM, lysed in 1 ml of SDS-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (60 mM Tris (pH 6.8), 2% SDS, 100 mM dithiothreitol, and 0.01% bromphenol blue), and counted on a ␥ counter (a minimum of 10 6 counts/sample). The background was determined as the counts adhering to non-transfected cells under the same conditions and was deducted from each sample. The assay was performed in triplicate.
Metabolic Labeling and Immunoprecipitations-For labeling, tight filter-grown monolayers of cells were washed once in minimal essential medium lacking methionine but containing 10% dialyzed fetal bovine serum (methionine-free medium) and then incubated 30 min at 37°C in methionine-free medium (1 ml/well). 0.3 mCi of [ 35 S]methionine in 1 ml of methionine-free medium was then added to each well on the basolateral side, and the cells were incubated at 37°C for the indicated pulse time. The media were then changed to complete media containing 10 times the normal concentration of methionine, and the incubation was continued at 37°C for the designated chase period. The cells were lysed in PBS containing 1% Triton X-100, 5 mM EDTA, and 0.5% deoxycholate for 30 min at 4°C. The extracts were centrifuged 20,000 ϫ g for 60 min, and the supernatant was incubated 60 min with 100 l of a 50% protein A-Sepharose slurry. For immunoprecipitation, 2 l of monoclonal mouse anti-porcine APN antibody (G43) or 2 l of monoclonal mouse anti-human APN antibody (43E6) and in both cases 20 l of 50% protein A-Sepharose were added to the supernatant, and the mixture was incubated 2 h at 4°C. The immunoprecipitates were washed five times in lysis buffer and analyzed by electrophoresis on 10% SDS-PAGE gels. The gels were dried, exposed to x-ray films, and processed for densitometry or analyzed on a phosphorimager (Molecular Dynamics PhosphorImager SI TM ).
Analyses of the Arrival at the Membranes-Tight filter-grown monolayers of cells were labeled with [ 35 S]methionine for 20 min and chased for the times indicated. Filters were transferred to 4°C and kept in PBS/CM until biotinylation. For each time point, one filter was biotinylated from the apical side and one from the basolateral side. Before biotinylation, the filters were washed four times in ice-cold PBS/CM, incubated 20 min in PBS/CM at 4°C, and then with s-NHS-biotin (0.5 mg/ml in PBS/CM), 2 ϫ 20 min. Unreacted s-NHS-biotin was then blocked with 50 mM NH 4 Cl in PBS/CM. The cells were extracted and immunoprecipitated as described above. The immunoprecipitate was boiled in 80 l of SDS-PAGE sample buffer without dithiothreitol and diluted in 1 ml of 25 mM Tris-HCl, pH 7.8, 250 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% BSA, and 10% fetal bovine serum and spun 3000 ϫ g for 5 min. The supernatant was then incubated with streptavidinagarose (50 l of a 50% slurry) at 4°C overnight. The streptavidin beads were washed four times with 25 mM Tris-HCl, pH 7.8, 500 mM NaCl, and 0.5% Triton X-100 and then three times in 10 mM Tris-HCl, pH 7.8, and 150 mM NaCl. The streptavidin precipitate was analyzed on 10% SDS-PAGE gels. The gels were dried, exposed to x-ray films, or analyzed on a phosphorimager.
Assay for Transcytosis-Confluent monolayers were labeled with [ 35 S]methionine for 20 min and chased for 1 h as described above. Thereafter, the cells were washed four times with PBS/CM prewarmed to 37°C and biotinylated from the basolateral side with 0.5 mg/ml s-NHS-SS-biotin dissolved in prewarmed PBS/CM for 20 min at 37°C. The cells were then chased again for 0 or 4 h as described above. If indicated the cells were reduced in freshly prepared 50 mM 2-mercaptoethanesulfonic acid (MESNA) in 50 mM Tris-HCl, pH 8.7, 100 mM NaCl, 2.5 mM CaCl 2 for 3 ϫ 20 min. The cells were washed in PBS/CM containing 20 mM iodoacetamide and subsequently dissolved in lysis buffer. The cell lysate was finally immunoprecipitated; The immunoprecipitate was boiled for 5 min in the presence of 20 l of 10% SDS and diluted with 1 ml of buffer A (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, and 0.2% BSA). The sample was spun 3000 ϫ g for 3 min, and the supernatant was incubated 12 h with streptavidin-agarose (50 l of a 50% slurry). The beads were washed once in buffer A, three times with buffer B (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 0.5% Triton X-100, 0.1% SDS, and 0.2% BSA), three times in buffer C (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 0.5% Triton X-100, and 0.2% BSA), and once in buffer D (50 mM Tris-HCl, pH 8.0). The immunoprecipitate was analyzed on 10% SDS-PAGE gels and quantitated as described above. The assay was performed in duplicate.

Secretory Forms of APN Are Secreted from Both Sides of
Caco-2 Cells-We have previously investigated the recombinant expression and secretion of two secretory forms of human APN (anchor-minus and anchor/stalk-minus) in MDCK cells (Vogel et al., 1992a(Vogel et al., , 1992b. To investigate the way in which these forms of APN are secreted from Caco-2 cells, eukaryote expression vectors carrying the cDNA for anchor-minus and anchor/stalk-minus APN (Fig. 1) were transfected into Caco-2 cells, and stable clones were isolated. The media of the clones were assayed for aminopeptidase enzymatic activity. 24-h media from non-transfected Caco-2 cells contained low (0 -3 milliunits/ml) concentrations of aminopeptidase activity, whereas 24-h media from transfected clones chosen for this study contained at least 60 milliunits/ml. 24-h media from transfectants grown on polycarbonate filters were collected from the apical and basolateral sides and assayed for aminopeptidase activity. For each construct, two clones were investigated with essentially the same result. The anchor-minus APN is secreted 36 Ϯ 5% to the apical side and 64 Ϯ 5% to the basolateral side, and anchor/stalk-minus is secreted 48 Ϯ 6% to the apical side and 52 Ϯ 6% to the basolateral side (Table I). The distribution of the secretory forms of APN between the apical and basolateral domains is thus indeed different from the endogenous wild type membrane-bound human APN, which is mainly present on the apical membrane (Matter et al., 1990;Le Bivic et al., 1990). We were unable to detect apical to basolateral or basolateral to apical transcytosis when the secretory forms of APN were applied at either side of filter-grown non-transfected Caco-2 cells.
It has previously been shown that in Caco-2 cells about 40% of the endogenous human APN is transported by the indirect pathway (Matter et al., 1990), leaving about 60% to the direct pathway using cell culture conditions (5-13 days post-confluent) similar to ours (3-6 days post-confluent). Using 15-20 days post-confluent cells, 60 -70% of the endogenous APN was transported by the indirect pathway (Le Bivic et al., 1990). The present study thus shows that Caco-2 cells transport the secre-tory forms of APN to the plasma membrane with a distribution between the apical and the basolateral side very similar to what has been reported for the endogenous membrane-bound enzyme. The cells are, however, unable to transcytose the basolateral pool of the secretory forms to the apical side. This is in contrast to what has been reported for the endogenous membrane-bound enzyme (Matter et al., 1990;Le Bivic et al., 1990). This points to an important role for the cytoplasmic and/or the transmembrane domain for efficient transcytosis of APN in Caco-2 cells.

The Cytoplasmic Domain Is of No Importance for the Steady State Cell Surface Distribution of APN-
To investigate the role of the cytoplasmic tail, two mutations in the APN cDNA were made. To distinguish the recombinant protein from the endogenous human APN in Caco-2 cells, we used the porcine APN cDNA and analyzed the protein with a monoclonal antibody (G43) specific to porcine APN (Delmas et al., 1994). A cDNA coding for a form of the porcine APN where the 9-amino acid N-terminal cytoplasmic sequence is exchanged for a Met-Ala-Arg sequence (tail-less) was constructed (Fig. 1). In the other construct, the only tyrosine in the cytoplasmic tail of the porcine APN cDNA was mutated to alanine (Tyr-6).
The tail-less, Tyr-6, and wild type porcine APN cDNAs were transfected into Caco-2 cells, and stable clones were isolated. The clones were screened using [ 35 S]methionine labeling followed by immunoprecipitation, and clones expressing tail-less, Tyr-6, and wild type APN were identified. Pulse-chase experiments showed that the tail-less and the Tyr-6 APN were converted from the high mannose to the complex glycosylated form with kinetics very similar to that of the wild type porcine APN (data not shown).
To determine the polarized distribution of the recombinant proteins, filter-grown transfected Caco-2 cells were incubated at 4°C with saturating concentrations of antiserum G43 either on the apical or the basolateral side, followed by incubation with 125 I-labeled Fab fragments. For each construct, two clones were investigated with essentially the same result. Total cellular extracts were counted on a ␥ counter. Quantitation of two to five experiments showed ( Table 1) that about 70% of the cell surface expressed wild type, tail-less, and Tyr-6 porcine APN located on the apical side. These results show that the cytoplasmic tail has no influence on the steady state distribution of the porcine APN between the apical and the basolateral membrane.
Human Wild Type APN Is More Efficiently Polarized Than Porcine Wild Type APN-It has earlier been demonstrated that at steady state about 98% of the endogenous human APN resides on the apical membrane (Le Bivic et al., 1990). This is a noticeably better polarization than the polarization of the recombinantly expressed wild type porcine APN reported in

TABLE I Apical expression of aminopeptidase N in transfected Caco-2 cells
In the antibody binding assay, filter-grown cells were incubated at 4°C with antiserum either on the apical or the basolateral side followed by 125 I-labeled F(abЈ) fragments. Total cellular extracts were counted on a ␥ counter. Apical expression as percentage of the total cell surface expression is shown. In the biotinylation assay, filter-grown cells were labeled with a short pulse of [ 35 S]methionine and chased overnight. The cells were then biotinylated from either the apical or the basolateral side and lysed, and APN was immunoprecipitated. Biotinylated APN was recovered from the immunoprecipitate by adsorption to streptavidin-agarose, analyzed by SDS-PAGE, and quantitated. Apically expressed aminopeptidase N is shown as percentage of total cell surface aminopeptidase N. Apical and basolateral 24-h media from tight filtergrown cells were collected, and the aminopeptidase activity was measured. Apically secreted aminopeptidase activity as percent of total secreted aminopeptidase activity is shown. The mean Ϯ S.D. and the number of experiments (n) are shown. this study. To investigate this difference in polarity, cells were transfected with an expression vector carrying the cDNA for wild type human APN, and a stable clone expressing aminopeptidase activity in amounts comparable to the clones expressing wild type porcine APN was isolated. After 3 days of confluence, the aminopeptidase N activity that could be ascribed to the endogenous human aminopeptidase N was maximally 20% of the total aminopeptidase activity in all the clones used in this paper. The antibodies available to us against human APN did not function well in the antibody-binding assay to determine the steady state distribution on the two membrane domains. Instead, a clone expressing recombinant wild type human APN was labeled with [ 35 S]methionine for a short pulse, chased overnight, and biotinylated from either the apical or the basolateral side. Thereafter, the cells were extracted and immunoprecipitated with a monoclonal antibody against human APN (43E6), streptavidin precipitated, analyzed on 10% SDS-PAGE gels, and quantitated. The analysis showed that Caco-2 cells transfected with an expression vector containing the wild type human APN cDNA had 94 Ϯ 7% of the surface human APN on the apical side (Table I). Non-transfected Caco-2 cells had 95 Ϯ 12% of the surface human APN on the apical side. In a similar assay, transfected Caco-2 had 77 Ϯ 13% of the surface wild type porcine APN on the apical side and 73 Ϯ 15% of the surface tail-less APN on the apical domain.
These results indicate that the difference in sorting efficiency is not dependent on expression levels since Caco-2 cells polarize the human APN with the same efficiency regardless of whether the cells also express recombinant human APN or not. This difference between the human and the porcine APN was further investigated by studying the appearance of newly synthesized proteins on the two cell surface domains. Cells were pulse labeled with [ 35 S]methionine for 20 min, and after various chase times the plasma membrane proteins present in either the apical or basolateral domain were biotinylated. Biotinylated APN was purified by immunoprecipitation using protein A-agarose, released from the beads by boiling, and reprecipitated with streptavidin-agarose. The resulting purified antigens were then analyzed on SDS-PAGE gels and quantitated. Caco-2 cells expressing recombinant wild type human APN were investigated, and the enzyme was found to be transported initially to both the apical and the basolateral membrane with a clear transient pool at the basolateral membrane (Fig. 2, panel A). The basolateral fraction disappeared after 3 h of chase, which resembles what has been reported for the endogenous human APN (Matter et al., 1990;Le Bivic et al., 1990). The porcine APN was also transported to the two membranes simultaneously (Fig. 2, panel B), and a transient pool of APN was observed at the basolateral side, albeit its disappearance from the basolateral side is slower and less effective than that of the human APN. The basolateral pool is still clearly visible after an overnight chase (data not shown). Caco-2 cells are thus able to polarize the human APN more efficiently than the porcine APN.
Both Wild Type and Tail-less Porcine APN Are Transcytosed-The appearance of newly synthesized tail-less APN on cell surface domains was also investigated as described above. The tail-less APN also appeared on both membranes, with a transient basolateral pool resembling that of the wild type porcine APN. No significant difference could be seen between the wild type porcine APN and the tail-less form.
To assess directly whether the newly synthesized wild type and tail-less porcine APN that appear transiently on the basolateral membrane are routed to the apical cell surface, we used the following protocol. Wild type or tail-less porcine APN expressing cells, grown on filters, were pulse labeled with [ 35 S]methionine for 20 min, chased for 60 min, and then biotinylated on the basolateral surface with the cleavable reagent s-NHS-SS-biotin at 37°C for 20 min followed by a second chase for 0 or 4 h. At the indicated times, the cells were reduced twice with MESNA from the apical side or from the basolateral side. These experiments show that both wild type and tail-less APN, biotinylated from the basolateral side, can be partly reduced from the apical side (Fig. 3) after a 4-h chase. A control experiment showed that without a second chase, the biotin could be reduced from the basolateral side but not from the apical side. Thus, the wild type and tail-less porcine APN molecules are transcytosed from the basolateral to the apical side to approximately the same degree.

DISCUSSION
It is well known that both porcine and human APN form homodimers (Danielsen, 1994, Jascur et al., 1991. Thus, it is possible that the recombinant and endogenous APN expressed together in Caco-2 cells in this study form heterodimers. However, the anchor-minus and anchor/stalk-minus APN secreted from Caco-2 cells are unlikely to represent molecules that pass through the cell in a heterodimer state (between soluble and membrane-bound APN) since we have previously experienced, during purification of the enzyme, that the dimer is very stable (Sjöström and Norén, 1982). When analyzing the recombinantexpressed porcine aminopeptidase N, it is possible that maximally 1 ⁄4 of the molecules (estimated from the expression level of aminopeptidase enzymatic activity of non-transfected and transfected cells) are in a heterodimer state and thus, if present, constitute a minor fraction.
Transport from the trans-Golgi network to the Plasma Membrane-It has been demonstrated for several proteins that the signal for basolateral sorting resides in the cytoplasmic domain (for a review, see Matter and Mellman (1994)). To our knowledge, no basolateral sorting signal has been identified so far on the ectodomain of a membrane protein. The secretory and tail-less forms of APN do not carry a cytoplasmic basolateral sorting signal since they do not have a cytoplasmic tail. They are, however, still routed to the basolateral domain in proportions equivalent to the wild type enzymes. This strongly indicates that no cytoplasmic basolateral sorting signal is involved in directing APN to the basolateral membrane in Caco-2 cells. Thus, the basolateral transport is either guided by a new class of basolateral sorting signals located on the ectodomain or the transport to the basolateral domain in this cell type occurs by default.
Transcytosis from the Basolateral to the Apical Plasma Membrane-In this study, we found that the basolateral pool of wild type human APN disappears faster than that of the wild type porcine aminopeptidase N and that the wild type human APN obtains and maintains a more polarized distribution after 2 h of chase than that of wild type porcine APN. This indicates that the human APN is transcytosed more efficiently than the porcine APN. The registered rate of transcytosis of the porcine forms could be due to dimer formation between the recombinant porcine form and the endogenous human APN. The porcine forms might thus be unable to transcytose on their own, making the difference in kinetics even bigger.
It is not known whether the transcytotic transport from the basolateral to the apical domain is a signal-mediated process, but the finding in this study that the porcine and the human APN are transcytosed with different kinetics provides evidence that this is not a default pathway, and hence active sorting is involved. The signal involved is not located in the cytoplasmic part since the human and the porcine APN transcytose with significantly different kinetics despite a 100% identity in the cytoplasmic tail. However, the transmembrane domain is necessary for the ability to transcytose efficiently, as seen by the lack of transcytosis of the soluble forms. Whether a transcytosis signal is located in the transmembrane domain or it is merely the membrane attachment that is important to ensure efficient endocytosis is not yet known.
In this study, we found that the Caco-2 cell line that is derived from a human colon adenocarcinoma is able to sort human APN more efficiently than porcine APN. When the porcine APN cDNA was expressed in MDCK cells, 92 Ϯ 5% of the total surface-expressed APN was located on the apical domain, 2 suggesting that the difference in polarization in Caco-2 cells is not due to accidentally introduced mutation(s) in the cloned porcine cDNA. It could be hypothesized that polarization is an old and well conserved trait, but the present study indicates that the mechanisms leading to 2 L. K. Vogel, H. Sjöström, and O. Norén, unpublished observation.

FIG. 3. Transcytosis of wild type and tail-less porcine APN in
Caco-2 cells. Cells were pulsed for 20 min and chased for 60 min. The cells were thereafter biotinylated from the basolateral side with s-NHS-SS-biotin, a cleavable analogue of biotin, at 37°C and further chased for 0 or 4 h. After this second chase, cells were reduced with MESNA from the apical (A) or basolateral (B) side or reduction was omitted (none). The biotinylated aminopeptidase N was purified by the double immunoprecipitation and streptavidin precipitation protocol described in the legend to Fig. 2, analyzed by SDS-PAGE gels, and quantitated. The means of at least two independent experiments are shown. polarization have undergone some modification during evolution. The differences between the efficiency in sorting of the human and porcine APN may provide an opportunity to map the sorting signal operating on the transcytotic pathway in Caco-2 cells. There is 79% identity between the human and the porcine APN amino acid sequences (Delmas et al., 1992). A range of chimeras between the human and porcine APN has already been analyzed in MDCK cells to map a virus binding side on APN (Delmas et al., 1994), and they were found to be well expressed on the cell surface indicating a correct three-dimensional structure of the chimeric molecules that allows cell surface expression.