Polarized Apical Targeting Directed by the Signal/Anchor Region of Simian Virus 5 Hemagglutinin-Neuraminidase*

To examine the possibility of independent cytoplasmic/transmembrane domain-based apical sorting, we have investigated paramyxovirus SV5 hemagglutinin-neuraminidase (HN), a type II membrane protein with a small N-terminal signal/anchor region. In SV5-infected Madin-Darby canine kidney (MDCK) cells, >90% of HN is found on the apical surface. We have expressed chimeric proteins in which the N terminus of HN, including its signal/anchor region, is attached to a (normally cytosolic) reporter pyruvate kinase (PK). PK itself expressed immediately downstream from a cleavable signal peptide was converted to a 58-kDa N-linked glycosylated form, which was secreted predominantly (80%) to the basolateral surface of MDCK cells. By contrast, stably expressed PK chimeras, now anchored as type II membrane proteins with either the first 48 or 72 amino acids of HN, received similar N-linked glycosylation, yet exhibited polarized transport with a preferentially (75%) apical distribution. These results suggest that the N-terminal signal/anchor region of HN contains independent sorting information for apical specific targeting in MDCK cells.

Enveloped viruses have provided an extremely useful paradigm to understand polarized protein traffic in epithelial cells (1). Viral membrane glycoproteins, even when expressed without other viral constituents, contain information sufficient for selective transport to either the apical or basolateral surface (2)(3)(4)(5)(6)(7)(8)(9)(10). Indeed, analyses of both viral and endogenous glycoproteins expressed in epithelial cells, using recombinant DNA approaches, have shown that the information required for polarized transport depends on specific structural determinants contained in the newly synthesized glycoproteins (11); from these analyses, several common themes have emerged.
Specifically, attachment of a glycosylphosphatidylinositol anchor to protein moieties that are otherwise luminal is sufficient for polarized surface delivery (12,13) that is typically (but not in every case) (14) directed apically. In addition, studies using type I membrane glycoprotein chimeras as well as truncation mutants have suggested that the ectodomains of these membrane proteins possess signals for apical transport (9,(15)(16)(17)(18). Most recently, this idea has been extended to suggest that both secretory glycoproteins and the ectodomains of membrane glycoproteins may use N-linked carbohydrates either indirectly (19) or directly (20) for polarized sorting. However, Roth et al. (21) and Green et al. (22) found that complete inhibition of glycosylation did not affect polarized viral glycoprotein expression. On the other hand, information for the basolateral transport of the vesicular stomatitis virus G-protein is contained within its cytoplasmic/transmembrane extension (23), and furthermore, specific motifs in the cytoplasmic tails of several mammalian membrane protein receptors and lysosomal membrane glycoprotein lgp120 are now recognized to be essential for the targeting of these proteins to the basolateral surface (24 -28). Of course, it should be noted that most of these studies have employed single-spanning membrane proteins with a type I topology (29), and it has not been excluded that specific basolateral targeting information can also exist in protein luminal domains (9, 23, 30 -32).
SV5, which is released by budding from the apical surface of infected epithelial cells (2), possesses two glycoproteins, the hemagglutinin-neuraminidase (HN) 1 and fusion protein. HN is a type II membrane protein with a small N-terminal signal/ anchor region and a large C-terminal ectodomain (33,34). In recent years, chicken muscle pyruvate kinase (PK) has been a favored reporter for protein targeting, as chimeric genes containing specific sorting information can relocate the normally cytosolic PK moiety to the nucleus (35,36), the membrane of the endoplasmic reticulum (37), or the mitochondria (38). With this in mind, a chimera including the signal/anchor portion of HN was found to redirect PK to the cell surface, suggesting that specific protein targeting information may reside in this region of HN (39). However, it has been unclear if this segment could also account for polarized apical delivery of HN. In this study, we have utilized HN-PK chimeras to examine this question in stably transfected MDCK cells.

EXPERIMENTAL PROCEDURES
Construction of Chimeric HN-PK and Secretory PK cDNAs-Construction of the HN 48 PK chimera via an in-frame EcoRI linker (5Ј-CGGAAT-TCC-3Ј, encoding the hydrophilic linker Arg-Asn-Ser) in pSV63 has been described (previously called APK) (39). Intact HN 48 PK was excised from pSV63 with XhoI and subcloned first into pGEM7Z and then, using HindIII and XbaI sites, into pRC/CMV (Invitrogen). This produced the plasmid pRC/HN 48 PK, encoding a protein joining HN residues 1-48 to PK residues 17-529 by a 3-amino acid linker (encoding Arg-Asn-Ser), driven by the cytomegalovirus promoter. To construct HN 72 PK, an oligonucleotide (5Ј-TTGGATCCAAGCTTATGGTTGCAGAAGATGCCCCT-3Ј) containing a HindIII site followed by HN nucleotides 320 -340 was used as an upstream primer, and an oligonucleotide (5Ј-GTCGGAATTCCTGCGA-CAGAGAGAATATT-3Ј) containing HN nucleotides 521-538 followed by an EcoRI linker was used as a downstream primer to amplify the Nterminal fragment 1-72 of the HN gene by polymerase chain reaction as described previously (40). The amplified DNA fragment was isolated after digestion with HindIII and EcoRI, and along with the PK moiety isolated from pRC/HN 48 PK by digestion with EcoRI and XbaI, a three-way ligation was performed using pRC/CMV vector that had been digested with Hin-dIII and XbaI. Thus, the hybrid protein expressed by the recombinant plasmid pRC/HN 72 PK encodes HN residues 1-72 joined to PK residues 17-529 by a 3-amino acid linker (encoding Arg-Asn-Ser), driven by the cytomegalovirus promoter. To construct secretory PK, an oligonucleotide (5Ј-ACTGAAGCTTATGGACAGCAAAGGT-3Ј) containing a HindIII site followed by bovine preprolactin nucleotides 68 -83 was used as an upstream primer, and an oligonucleotide (5Ј-ATCGGAATTCCGGTG-GAGACCACACC-3Ј) containing the sequence of preprolactin nucleotides 145-160 followed by an EcoRI linker was used as a downstream primer to amplify by polymerase chain reaction the preprolactin signal peptide provided in plasmid form (gift of Dr. V. Lingappa, University of California, San Francisco). The amplified DNA fragment was then isolated after digestion with HindIII and EcoRI, and along with the same PK cDNA moiety (isolated from pRC/HN 48 PK by digestion with EcoRI and XbaI), a three-way ligation was performed using pRC/CMV vector that had been digested with HindIII and XbaI. Thus, the hybrid protein expressed by this recombinant plasmid encodes the N-terminal cleavable prolactin signal peptide residues 1-31 joined to PK residues 17-529 by a 3-amino acid linker (encoding Arg-Asn-Ser), driven by the cytomegalovirus promoter.
Stable Expression in MDCK Cells-MDCK cells (strain II) were transfected with the above recombinant plasmids by the Lipofectin method as described previously (40). Colonies were selected and maintained in the presence of G418 (500 g/ml; Life Technologies, Inc.) in Dulbecco's modified Eagle's medium supplemented with 10% newborn calf serum. Subclones of transfected MDCK cells were induced with 10 mM sodium butyrate for 16 h (26) and then labeled for 2 h with 100 Ci of [ 35 S]methionine/cysteine in methionine/cysteine-deficient medium. The labeled cells were lysed in 1 ml of TNT lysis buffer (50 mM Tris, 150 mM NaCl, and 1% Triton X-100, pH 7.4) plus 0.1% SDS, and the clarified lysates were incubated overnight at 4°C with rabbit anti-PK bound to protein A-agarose and analyzed by 8.5% SDS-polyacrylamide gel electrophoresis (SDS-PAGE). At least two colonies expressing the desired protein from each cDNA construct were randomly selected for further analysis.
Selective Cell-surface Biotinylation-Costar Transwell polycarbonate filter units with a pore size of 0.4 m were used. Cells (2 ϫ 10 6 ) were plated and grown for 4 days, and the formation of electrically tight cell monolayers was confirmed by measuring transepithelial resistance with chopstick electrodes (World Precision Instruments, Inc., New Haven, CT). All clones exhibited resistance between 100 and 125 ⍀⅐cm 2 prior to their use in experiments. Selective cell-surface biotinylation was performed according to published procedures (41). For measurement of polarized protein distribution at steady state, biotinylated proteins isolated with streptavidin-agarose were resolved by SDS-PAGE and then analyzed by specific immunoblotting.
Pulse-Chase Experiments-Filter-grown cells were pulse-labeled with 400 Ci of [ 35 S]methionine/cysteine in methionine/cysteine-deficient medium. After different periods of chase in complete Dulbecco's modified Eagle's medium containing excess methionine and cysteine, the apical and basolateral surfaces were selectively biotinylated as described above. The cells were then lysed in TNT lysis buffer, clarified by centrifugation, and immunoprecipitated with antibodies bound to protein A-Sepharose. Immunoprecipitates were denatured in 1% SDS, diluted 10-fold into TNT lysis buffer, and then re-precipitated after incubation with 10 l of 4% streptavidin-agarose (Sigma). The radiolabeled biotinylated proteins were then analyzed by SDS-PAGE, and radioactivity was quantified using a PhosphorImager.
Analysis of Labeled Protein Secretion-For assaying labeled secretion of PK, the transfected MDCK cell lines were either pulse-labeled as described above (see Fig. 4B) or continuously labeled for 8 h with 100 Ci of [ 35 S]methionine/cysteine (see Fig. 4A). Media bathing the labeled cells were collected from either the apical or basolateral chamber, and cellular debris was removed by centrifugation. The clarified media were examined by immunoprecipitation, SDS-PAGE, and fluorography.
Infection with SV5-Polarized MDCK cells grown on Transwell units were incubated with SV5 at 20 plaque-forming units/cell in Dulbecco's modified Eagle's medium containing 0.5% bovine serum albumin for 1 h at 37°C. At this time, the cells were washed, and the medium was replaced with fresh Dulbecco's modified Eagle's medium supplemented with 2% newborn calf sera. Infected cells were maintained at 37°C for 14 h for further analysis.
Immunoblot Analysis and Densitometric Quantitation-For immunoblotting, biotinylated samples were analyzed by reducing 8.5% SDS-PAGE. The proteins were transferred to nitrocellulose filters (0.2-m pore size; Schleicher & Schuell) and incubated for 1 h at room temperature in 5% nonfat dry milk in phosphate-buffered saline and then for 2 h with primary antibodies diluted in the same buffer. Protein-bound antibodies were detected with horseradish peroxidase-conjugated specific secondary antibodies using enhanced chemiluminescence (ECL, Amersham Corp.). Immunoblots were quantified by scanning densitometry. The method was validated by using serial dilutions of protein samples to establish a linear relationship of ECL densities in the range of protein concentrations studied.
Antibodies-Monoclonal antibodies to SV5 HN were kindly provided by Dr. R. E. Randall (42). An antiserum to chicken PK was prepared in rabbits, which received a total of 250 g of antigen (five ϫ 50 g). The first intramuscular injection of PK was emulsified in Freund's complete adjuvant, followed by three boosts at weekly intervals emulsified in Freund's incomplete adjuvant. Finally, a single intravenous dose of PK (without adjuvant) was administered 5 days before bleeding.

Expression and Polarized Exocytotic Discharge of Secretory
PK-Previous studies have established that for chimeras that position the normally cytosolic PK moiety in the lumen of the secretory pathway, a major fraction of the expressed chimera undergoes N-linked glycosylation and intracellular transport to the cell surface (39). Since our goal was to attach putative signals for polarized protein targeting to the PK reporter, we first set out to examine the fate of a chimeric protein expressed after ligating a cDNA fragment encoding the N-terminal 31amino acid signal peptide of preprolactin directly to PK itself  48 PK, HN 72 PK, and secretory PK proteins were constructed by linking the HN N-terminal 48 or 72 amino acids or the bovine preprolactin signal sequence (prePRL Signal Seq.), respectively, via an EcoRI linker (which adds Arg-Asn-Ser) to PK. The construction of HN 48 PK, HN 72 PK, and secretory PK, expressed in pRC/ CMV plasmids, is described under "Experimental Procedures." Amino acid numbers are indicated above each construct.
("secretory PK") ( Fig. 1). This signal peptide contains the information necessary for normal cleavage by signal peptidase (43,44), resulting in a threonine residue as the new N terminus of the mature protein. After G418 selection of stably transfected MDCK cells, clones expressing this form of PK were identified by immunoprecipitation and SDS-PAGE analysis.
Unglycosylated secretory PK with an uncleaved signal is predicted to be a polypeptide of 547 amino acids with a molecular mass of ϳ60 kDa, whereas unglycosylated secretory PK after cleavage of the signal peptide is 517 amino acids with a predicted molecular mass of ϳ56 kDa. Indeed, when we expressed a leaderless cytosolic form of PK in COS cells, this unglycosylated polypeptide lacking a signal peptide migrated on SDS-PAGE at the predicted size of ϳ56 kDa (data not shown). For expression in the lumen of the secretory pathway, the PK cDNA sequence contains one potential N-linked glycosylation site (45). With this in mind, we observed that in MDCK cells expressing secretory PK, intracellular immunoreactive forms of PK included different closely migrating species: the majority ran as an ϳ58-kDa species, whereas a smaller portion ran at ϳ56 kDa ( Fig. 2A, second lane). 2 To see if the larger form occurred as a consequence of N-linked glycosylation, the secretory PK forms expressed in MDCK cells were treated with PNGase F. In this case, all forms collapsed into a single species of ϳ56 kDa ( Fig. 2A, third lane). Because PNGase F removes N-linked glycans that are added in the lumen of the secretory pathway, these molecular mass data strongly suggest that secretory PK had indeed been translocated into the endoplasmic reticulum lumen and underwent removal of the prolactin signal peptide and that most of the secretory PK received N-glycosylation as has been reported for the membrane-bound form (39). Such N-linked glycosylation has been proposed to have important benefits in favoring protein stability and export through the secretory pathway (46 -49).
We therefore examined the stability and export of newly synthesized secretory PK in MDCK cells over a 12-h chase. As expected at the zero chase time, secretory PK was quantitatively intracellular (Fig. 3A). There was intracellular loss of a portion (ϳ50%) of labeled PK during the first 2 h of chase (Fig.  3B) that is not understood; loss was seen for both the 58-and 56-kDa forms (Fig. 3A). Following this, however, between 2 and 12 h of chase, the remaining immunoprecipitable 58-kDa PK was progressively released from the cells to the medium, with t1 ⁄2 Ϸ 4 h and with an efficiency of ϳ90% over this time course (Fig. 3C). As expected, PNGase F treatment converted PK in the medium back to an ϳ56-kDa species (Fig. 2B). The release of ϳ58-kDa PK was not due to nonspecific cell leakage or lysis, as the 56-kDa form of PK was retained quantitatively within the MDCK cells (Fig. 3A). Moreover, secretory PK recovered free in the medium was not sedimentable in a 1-h spin at high speed (data not shown). Thus, the data in Figs. 1-3 indicate that the prolactin signal-PK chimera enters the lumen of the secretory pathway and undergoes N-linked glycosylation. While a portion of the PK protein is apparently degraded intracellularly, the remaining glycosylated 58-kDa form is competent for intracellular transport and behaves as a valid secretory protein marker.
Next, we investigated the polarity of PK secretion from filtergrown MDCK cells. Stable MDCK transfectants expressing secretory PK were continuously labeled for 8 h, and apical and basolateral media were collected and analyzed by immunoprecipitation, SDS-PAGE, and fluorography (Fig. 4A, upper panel). Interestingly, only 20% of secreted PK (ϳ58 kDa) was found on the apical side of transfected cells, with 80% recovered in the basolateral medium (Fig. 4A, lower panel). By contrast, the endogenous MDCK secretory protein gp80/clusterin (50) was released with apical predominance (data not shown). To exclude possible rapid apical secretion of PK followed by reinternalization and transcytosis to the basolateral side, polar-2 Previous expression of the PK moiety in the lumen of the secretory pathway has established that a minor fraction of PK can undergo a second distinct N-glycosylation event at a non-consensus site (39). We found that in all MDCK clones expressing secretory PK within the first 10 passages, besides ϳ70% of PK that is singly glycosylated and migrates at ϳ58 kDa, an additional minor form of intracellular and released PK could be detected at ϳ60 kDa. We established that this minor form represented PNGase F-and endo-␤-N-acetylglucosaminidase H-sensitive glycosylation at a non-consensus site, in addition to the glycosylation that occurs at the consensus asparagine, identical to that which has been reported for a fraction of HN 48 PK (39). For secretory PK, there is a hint of this doubly glycosylated form secreted into the medium of clonal MDCK cells from early passage in the second lane of Fig. 2B. However, as a function of passage number, expression of this additionally glycosylated form diminished to below the limits of detection in our pulse-labeling experiments. We should note that the presence or absence of this additional glycosylation event had no detectable effect on the routing of secretory PK to the basolateral surface. A, the cells were lysed in TNT lysis buffer with 0.1% SDS, immunoprecipitated with anti-PK serum, and then analyzed by 8.5% SDS-PAGE. Mock, cells transfected with empty vector; PK, cells transfected to express recombinant secretory PK; P-F, secretory PK after digestion with PNGase F. B, the labeling medium was collected, clarified by centrifugation, and then immunoprecipitated with anti-PK serum. Mock, medium derived from cells transfected with empty vector; PK, medium bathing cells expressing recombinant secretory PK; P-F, secreted PK digested with PNGase F. The second and third lanes were collected from clonal MDCK cells after early passage (Ͻ10), whereas the fourth and fifth lanes represent clonal MDCK cells after later passage (Ͼ25), indicating stability of expression and secretion of 58-kDa PK. FIG. 3. Export of secretory PK. MDCK cells stably expressing secretory PK were pulse-labeled with [ 35 S]Met/Cys for 5 min and chased for the indicated times. From a set of cells at each chase time, the bathing medium (M) was collected, the cells (C) were lysed, and both were clarified by centrifugation. A, aliquots of each were taken for immunoprecipitation with anti-PK serum and analysis by SDS-PAGE. B, total PK recovery relative to that obtained at the zero chase time was quantified by densitometric film scanning of 58-and 56-kDa PK bands. C, up to 90% of newly synthesized PK recovered at each chase time was secreted into the medium over the 10 -12-h time course. arb. units, arbitrary units.
ized cell monolayers were pulse-labeled with [ 35 S]methionine/ cysteine for 5 min, and the kinetics of initial release of 58-kDa PK was followed as a function of chase time (Fig. 4B, left  panels). Evidently, from the moment that PK secretion is first detected, the protein is directly delivered with a fixed polarized ratio leading to ϳ80% in the basolateral medium (Fig. 4B, right  panel). Clearly, the glycosylated PK molecule exhibits no sorting signal for apical specific transport in MDCK cells.
Expression and Polarized Distribution of the HN 48 PK Protein in MDCK Cells-We next examined the expression of the HN 48 PK chimera ( Fig. 1), in which the HN N terminus including its signal/anchor is linked to the luminally oriented PK protein (39). Immunoprecipitates with anti-PK serum from cells transfected with pRC/HN 48 PK revealed two closely migrating species, a major band at ϳ67 kDa and a slightly less intense band at ϳ65 kDa (Fig. 5A). When treated with PNGase F to remove N-linked oligosaccharides, the two different species migrated with the same mobility as a nonglycosylated form at ϳ63 kDa (Fig. 5B), in agreement with previous reports (39). Thus, the luminal domain of HN 48 PK underwent N-linked glycosylation in MDCK cells, as is the case for secretory PK (Fig. 2) and for HN 48 PK expression in CV1 cells, where the chimera underwent glycosylation en route to residence at the cell surface (39).
To examine the polarized surface distribution of the HN 48 PK chimera in filter-grown MDCK cells, we utilized cell-surface biotinylation. As a control, in SV5-infected MDCK cells, ϳ95% of blottable HN was found on the apical surface (Fig. 6, left). Interestingly, linking the N-terminal region containing the signal/anchor of HN to PK (Fig. 1) caused a change in the targeting of the glycosylated PK reporter (Fig. 5) from a basolateral (Fig. 4) to an apical surface distribution. Specifically, ϳ75% of HN 48 PK was now present on the apical surface of transfected cells (Fig. 6, right). The data indicate that the N-terminal region of HN, even when transplanted onto a ba-solaterally targeted luminal domain, by itself is sufficient for preferential apical distribution, although this steady-state distribution is less completely apical than that found for the full-length HN protein in the setting of expression with other viral components.
To determine whether the apical distribution of HN 48 PK was accomplished by direct polarized delivery from the trans-Golgi network, we examined the kinetics of first appearance of the newly synthesized protein on the apical and basolateral surfaces. Filter-grown monolayers were pulse-labeled with [ 35 S]methionine/cysteine for 20 min and chased for various times. At each chase interval, replicate cell monolayers were biotinylated on either the apical or basolateral side. As a control, virally infected MDCK cells were examined, and it was found that initial delivery of intact HN was Ն90% to the apical surface (Fig. 7A), whereas the HN 48 PK protein arrived at 90 and 180 min with an apical preference in the 80 -85% range (Fig. 7B) (and this apical delivery was already detectable at 45 min (data not shown)). The ratio of polarized initial delivery of HN 48 PK (apical/basolateral Ն 4:1) to the cell surface was slightly higher than the ratio of protein distribution at steady state (apical/basolateral Х 3:1), just opposite of what would be  48 PK-transfected MDCK cell line was labeled, lysed, and immunoprecipitated as described above. HN 48 PK immunoprecipitates were dissociated from the agarose beads and then analyzed either undigested (U) or after digestion with PNGase F (P-F).
FIG. 6. Polarized surface distribution of intact HN and HN 48 PK chimera. Monolayers of filter-polarized MDCK cells were either infected with SV5 or stably transfected with the HN 48 PK gene (see Fig. 1). The cell-surface proteins were tagged using surface-selective biotinylation, precipitated with streptavidin-agarose, and then analyzed by reducing 8% SDS-PAGE. The proteins were transferred to nitrocellulose filters and analyzed by immunoblotting with a monoclonal antibody to HN (upper left panel) or polyclonal anti-PK (upper right panel). Immunoblots developed by ECL were quantified by densitometry, and the results are expressed as the percent apical (Ap; black bars) and basolateral (Bl; hatched bars) surface expression for each protein (lower panel). The quantitative data represent the means Ϯ S.D. from two independent experiments. expected if this protein distribution were derived from basolateral to apical transcytosis. Thus, these results indicate that the polarized sorting of HN 48 PK occurs largely at the trans-Golgi network.
Although the apical polarity of newly delivered HN 48 PK was impressive, by 6 h after synthesis, a substantial decline in the level of newly synthesized HN 48 PK on the cell surface was observed (Fig. 8, dashed line). Since most of the delivery of HN 48 PK to the cell surface was apical, the subsequent loss of most of the HN 48 PK from the cell surface also occurred on the apical side (Fig. 7B). No HN 48 PK was recovered in the bathing medium, indicating that the protein was not shed from transfected cells. 3 However, clathrin-coated internalization of HN from the cell surface and its subsequent degradation in the endocytic pathway have been described (51); the kinetics of turnover of newly synthesized HN 48 PK protein in MDCK cells (Fig. 8, solid line) suggested the possibility of a similar fate for the HN signal/anchor-containing chimera. Although the ordi-nates for the two graphs plotted on Fig. 8 are not the same, we note that Ն75% of newly synthesized HN 48 PK has been reported to reside on the plasma membrane at the 3-h chase time (39). Taken together, the data suggest that despite cellular events that transpire after surface externalization (which may diminish the apparent apical protein distribution), attachment of the HN signal/anchor causes newly synthesized, glycosylated PK protein to be redirected from the basolateral to the apical surface.
Expression and Polarized Surface Distribution of HN 72 PK-Few if any apical sorting signals for membrane proteins have been previously described to exist in cytoplasmic/transmembrane domains, as there has been greater emphasis on the possibility that such signals are contained in the ectoplasmic domain (29). Thus, we constructed HN 72 PK (Fig. 1) to investigate whether including an increased portion of the HN N terminus, containing more of the juxtamembrane ectoplasmic region, could enhance the efficiency of apical transport. To determine the expression of HN 72 PK in stable MDCK transfectants, radiolabeled cells were immunoprecipitated with anti-PK serum and analyzed by SDS-PAGE and fluorography. As shown in Fig. 9A, HN 72 PK was expressed as a predominant band of ϳ71 kDa, with a minor band at ϳ69 kDa. When the same sample was treated with PNGase F, the two species migrated with the same mobility as a nonglycosylated form at ϳ67 kDa, indicating that the HN 72 PK chimera undergoes Nlinked glycosylation. We therefore proceeded to examine the polarized surface distribution of HN 72 PK in transfected MDCK cells by selective cell-surface biotinylation (Fig. 9B). Quantitatively, ϳ75% of the HN 72 PK chimera was apically distributed, with ϳ25% of the surface protein found basolaterally (Fig. 9C). Evidently, the HN 72 PK protein behaved similarly to the HN 48 PK chimera (Fig. 6). Thus, as for the nerve growth factor receptor (16), inclusion or exclusion of luminal juxtamembrane residues has no specific effect on apical protein distribution.
Of note, using the same pRC/CMV expression vector to produce G418-resistant MDCK cells, several attempts to obtain clones stably expressing the full-length HN protein were unsuccessful, possibly because the hemagglutinin-neuraminidase activities produce a selective growth disadvantage in comparison with MDCK cells that do not express this protein. Thus, 3  we cannot exclude that additional information directing apical targeting may reside in more distal portions of the ectoplasmic domain of the HN protein.

DISCUSSION
In SV5-infected MDCK cells, the viral HN protein resides almost exclusively on the apical surface ( Fig. 6) (2). Since viral membrane proteins contain autonomous information regulating transport to either the apical or basolateral surface (52), there is reason to think that the HN protein may also contain such signals. In recent years, the cytoplasmic tails of certain membrane proteins have been the focus of attention as sites for basolateral sorting signals (24 -28), whereas the ectodomains of membrane proteins have been the focus of attention as sites for apical sorting signals (9,(15)(16)(17)(18). Nevertheless, in MDCK cells, truncation of the cytoplasmic tail of the vesicular stomatitis virus G-protein to only 1 amino acid still leads to a predominantly basolaterally targeted protein (23), whereas addition of the cytoplasmic tail with or without the transmembrane domain of influenza hemagglutinin to the truncated vesicular stomatitis virus G-protein may (53) or may not (9, 30) disrupt the basolateral targeting. Moreover, the "rules" that suggest that cytoplasmic/transmembrane domains are relevant only for basolateral targeting may not apply to the polarized sorting for type II membrane glycoproteins since a chimera involving the cytoplasmic/transmembrane domain of influenza virus neuraminidase (normally an apically expressed protein (5)) redirects the ectodomain of the human transferrin receptor (normally a basolaterally expressed protein) to the apical surface (54).
To identify possible polarity signals contained in the Nterminal region of the HN protein including the signal/anchor, we used chimeric constructions with PK, which has been commonly employed by others as a successful reporter for protein targeting studies (35)(36)(37)(38)(39). Remarkably, when secretory PK was expressed after attaching a cleavable prolactin signal peptide to its N terminus, in which the "consensus" information for signal peptide cleavage is conserved, the PK protein was secreted from the basolateral surface of transfected MDCK cells with high fidelity, approaching or equal to that of secreted proteins with defined basolateral sorting signals (31,32). This finding is all the more remarkable in light of the fact that the basolateral secretion of this "neutral reporter" was exclusively restricted to those forms that had undergone N-linked glycosylation, which has recently been suggested to play an indirect (19) or direct (20) role in apical specific trafficking. Thus, considering the earlier studies indicating that complete inhibition of glycosylation did not affect polarized glycoprotein expression (21,22), it appears that a role for N-glycans in apical specific targeting in MDCK cells exists only in the context of some proteins and not others (55). The 58-kDa form of secretory PK, although glycosylated, exhibits an apparently dominant basolateral sorting signal. With this in mind, the luminal PK domain serves as a useful recipient for testing potential apical targeting information in the N-terminal region of HN.
We expressed two HN-PK chimeras that contained the Nterminal signal/anchor of HN plus increasing segments of the HN ectodomain (Fig. 1). The results with both chimeras were similar: the apical distribution of the proteins was Х75% (Figs. 6 and 9). Although one can never exclude the possibility that the conformation of basolaterally secreted PK (i.e. after cleavage from the signal peptide) is subtly different from that of PK tethered to the signal/anchor of HN, there is no evidence to indicate this, suggesting that the N-terminal domain of HN contains signals for apical sorting. We note that, in general, expression of the membrane protein chimeras was not as polarized as that of their endogenous counterparts, which could reflect competing apical and basolateral signals in different protein regions, and this might also be influenced by internalization of the chimeras after delivery to the apical cell surface (Figs. 7B and 8). Thus, we hypothesize that in this type II membrane protein, a dominant apical signal exists in the Nterminal domain containing the signal/anchor, which can account for most if not all of the polarized targeting of the HN protein to the apical surface. Further studies are needed to define more precisely the nature of apical specific targeting signals in this region of HN. The finding of apical specific targeting signals in the cytoplasmic domain would be of special interest in light of recent developments in the identification of new adaptor proteins that may play roles in molecular sorting of apical membrane proteins at the level of the trans-Golgi network (56).