Mucin-like Domain of Enteropeptidase Directs Apical Targeting in Madin-Darby Canine Kidney Cells*

Enteropeptidase, a type II transmembrane protein of the enterocyte brush border, is sorted directly to the apical membrane of Madin-Darby canine kidney II cells. Apical targeting appears to be mediated by an N-termi-nal segment that contains a 27-amino acid residue O glycosylated mucin-like domain consisting of two short mucin-like repeats, A and B. Targeting signals within these repeats were characterized by using green fluorescent protein (GFP) as a reporter. Constructs with a cleavable signal peptide and both repeats A and B were secreted apically. Similar constructs lacking mucin repeats were secreted randomly. Either repeat A or B was sufficient to direct apical targeting of GFP. O -linked oligosaccharides alone were not sufficient for targeting because fusion to a different O -glycosylated motif did not alter the random secretion of GFP, and several constructs with mutations in either repeat A or B were O -glycosylated and secreted randomly. In addition, repeat B appears to contain an apical targeting signal that functions in the absence of glycosylation. Density gradient centrifugation indicated that, unlike several other apically targeted membrane and soluble proteins, apical sorting of mucin-GFP chimeric proteins does not appear to utilize lipid rafts. Enteropeptidase, a serine protease localized to the brush border of duodenal enterocytes (1), is targeted directly to the apical surface of Madin-Darby canine kidney II (MDCK) 1 cells (2). fluoride, pepstatin A, trans-epoxysuccinyl- L -leucyl-amido(4-guanidino)butane, bestatin, leupeptin, and aprotinin, from Sigma) was added. Cell debris was removed by centrifugation, and the supernatants were concentrated 10-fold by ultrafiltration (Centricon- 10; Amicon, Beverly, MA) and stored at (cid:2) 20 °C. Samples were analyzed by SDS-PAGE on 5–15% Tris-glycine gels (Bio-Rad) under reducing conditions and transferred by electroblotting onto nitrocellulose membranes (pore size 0.45 (cid:2) m, Bio-Rad). Membranes were incubated with monoclonal anti-GFP antibody (1:10,000 dilution in 2% nonfat milk in Tris-HCl, pH 7.5, 150 m M NaCl, 0.05% Tween 20) at room temperature overnight. After washing, the membrane was incubated with peroxi-dase-conjugated rabbit anti-mouse IgG (1:10,000 dilution in 2% nonfat dry milk in Tris-HCl, pH 7.5, 150 m M NaCl, 0.05% Tween 20) (Dako Corp., Carpinteria, CA) at room temperature for 2 h. Bound antibody was detected with the chemiluminescent ECL detection system (Amer-sham Biosciences) and XAR5 film (Eastman Kodak Co.). Films were exposed for several different times to obtain density signals within the linear response range. After optical scanning, signals were quantitated with NIH Image 1.6.1 (developed at the National Institutes of Health and available on the Internet at rsb.info.nih.gov/nih-image/). Glycosidase Digestions— Glycosidases were obtained from Oxford Glycosciences (Bedford, MA). Samples of concentrated conditioned me- dium (40 (cid:2) l) were heated to 100 °C for 5 min in 20 m M sodium citrate phosphate, pH 5.5, 0.2%

Enteropeptidase, a serine protease localized to the brush border of duodenal enterocytes (1), is targeted directly to the apical surface of Madin-Darby canine kidney II (MDCK) 1 cells (2). Apical sorting of enteropeptidase may involve at least two distinct signals. One is located in the C-terminal serine protease domain and depends on N-glycosylation; another is proposed to be located in an N-terminal segment that includes an O-glycosylated mucin-like domain and three potential N-glycosylation sites (2). The apical targeting signals within the Nterminal region of enteropeptidase have not been characterized structurally.
Several distinct classes of apical sorting signals have been identified, suggesting the existence of several apical targeting mechanisms. For example, apical sorting can be mediated by transmembrane domains (3)(4)(5), by glycosylphosphatidylinosi-tol anchors (6,7), by PDZ-interacting domains (8,9), or by N-linked oligosaccharides (10,11). In some proteins, such as sucrase-isomaltase (12) and dipeptidyl peptidase IV (13,14), O-linked oligosaccharides also appear capable of mediating apical sorting. The juxtamembrane segment of the neurotrophin receptor p75 contains clustered O-linked oligosaccharides and is required for apical targeting (15,16). The O-linked glycan-dependent apical sorting of sucrase-isomaltase also is accompanied by association with glycosphingolipid and cholesterol-rich membrane microdomains or lipid rafts (13,14). Inhibition of terminal ␣2,3-sialylation with benzyl-2-acetamido-2deoxy-␣-D-galactopyranoside (GalNAc-␣-O-benzyl) blocks the apical transport of several brush border-associated glycoproteins, including dipeptidylpeptidase IV (14) and the mucin MUC1 (17), suggesting that the terminal structures of certain O-linked oligosaccharides or mucin-like domains may direct apical targeting in some polarized cells. In general, these studies of O-glycosylation-dependent targeting have depended on the abolition of targeting by mutagenesis or metabolic inhibitors, and none has directly identified a signal that is sufficient for apical targeting.
To characterize apical sorting determinants within the Nterminal segment of enteropeptidase, we employed a modified green fluorescent protein (GFP) as a reporter (18) in transfected MDCK cell lines expressing various chimeric enteropeptidase-GFP proteins. The results demonstrate that either of two short O-glycosylated mucin-like repeats of enteropeptidase can confer apical targeting on a heterologous protein. However, O-linked oligosaccharides are not sufficient for apical targeting, and some apical targeting activity is retained by a peptide that is not glycosylated. The apical targeting function also does not appear to involve stable interaction with lipid rafts.

EXPERIMENTAL PROCEDURES
Construction of Deletion Mutants-Plasmids pSM AB (His) 6 -GFP and pS(His) 6 -gGFP were derived from pdLin-BEK, which encodes the human prothrombin signal peptide and His 6 tag from pHL-BEK (19) linked to the codon for the Ala-166 of plasmid pBEK (2). A fragment (SM AB ) encoding the human prothrombin signal peptide (S) plus His 6 tag linked to two mucin-like repeats (M AB ) (amino acids 166 -192) of bovine enteropeptidase (20) was made by PCR (GeneAmp reagents, PerkinElmer Life Sciences) with N-terminal primer (5Ј-gacagctcgagatggcgccacgtc-3Ј (XhoI site underlined) and C-terminal primer (5Ј-ccgtggatcccgtggggttgccag-3Ј (BamHI site underlined). Fragment S(His) 6 encoding the prothrombin signal peptide and His 6 tag without mucin-like repeats was generated by PCR with the same N-terminal primer and a different C-terminal primer (5Ј-cggtggatcccggctgctgtgatgatg-3Ј (BamHI site underlined). The PCR fragments SM AB (His) 6 and S(His) 6 were digested with XhoI and BamHI and ligated into the same sites of vector pEGFPN1 (GenBank TM accession number U55762, CLONTECH Laboratories, Palo Alto, CA) to generate pSM AB (His) 6 -GFP and pS(His) 6 -gGFP, respectively. A fragment encoding the prothrombin signal peptide (S) was generated from template pS(His) 6 -gGFP with primers 5Ј-gacagctcgagatggggtcaaag-3Ј (XhoI site underlined) and 5Ј-ca-ggatcctgcctgtgcacaaggc-3Ј (BamHI site underlined), digested with XhoI and BamHI, and ligated into pEGFPN1 to generate pS-GFP. A fragment encoding mucin repeats (M AB) was made similarly from template pSM AB (His) 6 -GFP with primers 5Ј-tagggatccagcctctttggagaat-3Ј (BamHI site underlined) and 5Ј-gctcctcgcccttgctcacca-3Ј. The fragment (M AB ) was digested with BamHI and ligated into the BamHI site of pS-GFP to create plasmid pSM AB -GFP.
Complementary pairs of oligonucleotides were synthesized encoding mucin-like repeat A (M A ) and repeat B (M B ). The M A oligonucleotides were 5Ј-ttccgggatccagcctctttggagaatttctctacgataagtcctgcaacaacgtcacgggatccaccg-3Ј and 5Ј-cggtggatcccgtgacgttgttgcaggacttatcgtagagaaattctccaaagaggctggatcccggaa-3Ј (BamHI sites underlined). Oligonucleotides encoding M B were 5Ј-ttccgggatccagaaaagctaacaaccagcattcctctggcaaccccacgggatccaccg-3Ј and 5Ј-cggtggatcccgtggggttgccagaggaatgctggttgttagcttttctggatcccggaa-3Ј (BamHI sites underlined). After annealing, the double-stranded fragments were digested with BamHI and ligated into the BamHI site of pS-GFP to give pSM A -GFP and pSM B -GFP, respectively.
Mutagenesis-Point or clustered mutations were generated with the QuikChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions, using pSM A -GFP or pSM B -GFP as the template and an oligonucleotide primer that contains the desired mutation. All product plasmids were sequenced (BigDye cycle DNA sequencing kit, PE Applied Science, Foster, CA) to confirm the accuracy of the construction.
Transfection-MDCK II cells (American Type Culture Collection) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen) as described (2). Cells growing in 6-well tissue culture plates were washed with phosphate-buffered saline and incubated with 5 g of plasmid DNAs premixed (1:6 w/v) with 30 l of PerFect lipid (pfx-2, Invitrogen) in serum-free Dulbecco's modified Eagle's medium. After 5 h, fetal bovine serum was added to 10%. After an additional 18 h, cultures were split 1:50 and cultured in 48-well plates for selection in 0.5 mg/ml geneticin (Invitrogen). Positive clones were identified by fluorescence microscopy and Western blotting with anti-GFP monoclonal antibody (MMS-118P, Berkeley Antibody Co., Richmond, CA).
Analysis of the Polarity of Protein Secretion-MDCK cells expressing various chimeric proteins were cultured on Transwell filters (pore size 0.4 m; Corning Costar Corp., Cambridge, MA) until a tight monolayer was formed according to transmembrane resistance (2). Cells were washed three times on both sides of the membrane with 2 ml of phosphate-buffered saline and cultured with 2 ml of serum-free medium (OPTI-MEM, Invitrogen) added to both apical and basolateral compartments. After 24 h, medium was collected from the apical and basolateral chambers, and 0.1% (v/v) protease inhibitor mixture (4-(2-aminoethyl)benzenesulfonyl fluoride, pepstatin A, trans-epoxysuccinyl-L-leucylamido(4-guanidino)butane, bestatin, leupeptin, and aprotinin, from Sigma) was added. Cell debris was removed by centrifugation, and the supernatants were concentrated 10-fold by ultrafiltration (Centricon-10; Amicon, Beverly, MA) and stored at Ϫ20°C. Samples were analyzed by SDS-PAGE on 5-15% Tris-glycine gels (Bio-Rad) under reducing conditions and transferred by electroblotting onto nitrocellulose membranes (pore size 0.45 m, Bio-Rad). Membranes were incubated with monoclonal anti-GFP antibody (1:10,000 dilution in 2% nonfat milk in Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20) at room temperature overnight. After washing, the membrane was incubated with peroxidase-conjugated rabbit anti-mouse IgG (1:10,000 dilution in 2% nonfat dry milk in Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20) (Dako Corp., Carpinteria, CA) at room temperature for 2 h. Bound antibody was detected with the chemiluminescent ECL detection system (Amersham Biosciences) and XAR5 film (Eastman Kodak Co.). Films were exposed for several different times to obtain density signals within the linear response range. After optical scanning, signals were quantitated with NIH Image 1.6.1 (developed at the National Institutes of Health and available on the Internet at rsb.info.nih.gov/nih-image/).
Glycosidase Digestions-Glycosidases were obtained from Oxford Glycosciences (Bedford, MA). Samples of concentrated conditioned medium (40 l) were heated to 100°C for 5 min in 20 mM sodium citrate phosphate, pH 5.5, 0.2% SDS, and 5% ␤-mercaptoethanol and cooled to room temperature. Detergent Nonidet P-40 (Roche Molecular Biochemicals) was added to a final concentration of 2%, and the samples were incubated at 37°C for 18 h without or with 1 milliunit/microliter Streptomyces plicatus endoglycosidase H. Reactions with peptide-N-glycosidase F (PNGase F) were prepared similarly except that the buffer was 20 mM sodium phosphate, pH 7.5, 50 mM EDTA, 5% ␤-mercaptoethanol, and 0.2% SDS, and the amount of added enzyme was 0.5 unit. To analyze proteins for O-glycosylation, samples of concentrated condi-tioned media (40 l) were digested with Arthrobacter ureafaciens neuraminidase (0.5 milliunit/microliter) at 37°C for 1 h in 100 mM sodium acetate, pH 5.0, followed by digestion with O-glycanase (0.05 milliunit/ microliter) in 100 mM sodium citrate phosphate, pH 6.0, and 100 g/ml bovine serum albumin at 37°C for 16 h. Digested samples were analyzed by SDS-PAGE and Western blotting with monoclonal anti-GFP IgG as described above.
Analysis of Raft Association-Detergent-insoluble glycosphingolipidenriched raft domains were prepared by cell lysis in ice-cold Triton X-100 (for 10 min without agitation) and then sucrose density gradient centrifugation as described previously (2). Proteins in the fractions were precipitated with 10% trichloroacetic acid on ice for 30 min, and pellets were resuspended in 200 l of 0.2 N NaOH. Samples (20 l) of trichloroacetic acid-concentrated fractions were analyzed by SDS-PAGE and immunoblotting with either monoclonal anti-GFP IgG (1: 10,000) or monoclonal anti-caveolin IgG (1:10,000; C37120, Transduction Laboratories, Lexington, KY) as described above.

Targeting Activity of the Enteropeptidase Mucin Domain-
Previous studies suggested that apical sorting of enteropeptidase in MDCK cells depends on a signal near the N terminus, between amino acid residues 50 -197 (2). This juxtamembrane region contains four potential N-glycosylation sites and a mucin domain consisting of two tandem O-glycosylated Ser/Thrrich repeats (Fig. 1A) (20,21). Potential targeting signals within these mucin repeats were characterized by using green fluorescent protein (GFP) as a reporter (Fig. 1B) (18). Control constructs (GFP and His 6 -gGFP) were secreted randomly ( Fig.  2 and Table I). Constructs with both mucin repeats (M AB -GFP and His 6 -M AB GFP) were secreted apically ( Fig. 2A and Table  I), indicating that the enteropeptidase mucin domain can direct apical targeting of a heterologous protein.
Digestions with neuraminidase and O-glycanase showed that the mucin repeats were extensively O-glycosylated (Fig.  2B). Secreted M AB -GFP and His 6 M AB -GFP had apparent masses of 53 and 56 kDa, respectively (Fig. 2B, lanes 5 and 7). Enzymatic removal of O-linked oligosaccharides reduced the apparent mass of each protein to 34 and 36 kDa (Fig. 2B, lanes  6 and 8), and these values are similar to the calculated masses of 34.5 and 36.6 kDa, respectively, for the polypeptides alone. The specificity of these glycosidases suggests that the oligosaccharide structures consist almost exclusively of monosialylated or disialylated structures, which are related to Sia␣2-3Gal␤1-3(Sia␣2-6)GalNAc-O-Ser/Thr (22). Although mucin repeat A contains a potential N-glycosylation site (Asn-Phe-Ser), further digestion with N-glycanase or endoglycosidase H did not affect the mass of M AB -GFP and His 6 -M AB GFP, indicating that this site is not utilized (data not shown). As expected, GFP (27.5 kDa) was not glycosylated (Fig. 2B, lanes 1 and 2). However, His 6 -gGFP was O-glycosylated, presumably on one or more of the Ser/Thr residues flanking the His 6 tag sequence. The removal of O-linked oligosaccharides reduced the apparent mass from 33-37 to 30 kDa (Fig. 2B, lanes 3 and 4), which is similar to the calculated mass of 29.6 kDa for the polypeptide alone. Despite this O-glycosylation, His 6 -gGFP was secreted randomly (Table I).
The number of mucin repeats in enteropeptidase varies considerably, with one repeat in human, two in bovine and porcine, three in rat, and four in mouse enteropeptidase (21). Such heterogeneity suggests that targeting signals could reside in single mucin repeats. Therefore, the behavior of chimeric proteins containing mucin repeat A or B was examined. When stably expressed in MDCK cells, constructs containing either mucin repeat were secreted apically ( Fig. 3A and Table I), indicating that both repeats possess functional targeting signals. Glycosidase digestions demonstrated that each mucin repeat was O-glycosylated (Fig. 3B) and not N-glycosylated (data not shown).
Glycosylation and Targeting-The enteropeptidase mucin repeats contain 11 Ser/Thr residues that could be O-glycosylated (Fig. 1A) (20). Replacement of all Ser/Thr residues by alanine abolished the targeting activity of mucin repeat A, whether present in one or three copies ( Fig. 4 and Table I). Mucin repeat B had reduced but significant apical targeting activity after all Ser/Thr residues were mutated, suggesting that features of the amino acid sequence may contribute to targeting ( Fig. 4 and Table I). Glycosidase digestions confirmed that none of these constructs had N-linked or O-linked oligosaccharides (data not shown).
The role of O-linked glycosylation was investigated further by restoring selected Ser/Thr residues, singly or in clusters. For mucin repeat A, all constructs that contained any Ser/Thr residues were O-glycosylated and constructs M A S172A/T173A/ S175A-GFP and M A T173A-GFP lost apical targeting activity ( Fig. 5 and Table I). These data suggest that O-glycosylation is not sufficient for apical targeting and that a central cluster of Ser/Thr residues may be important for apical targeting of M A -GFP. As observed for mucin repeat A, all mucin repeat B constructs that contained Ser/Thr residues were O-glycosylated, including M B -T184A/T185A/S186A-GFP, which has only one Thr residue (Fig. 6). Mutation of Thr-185 alone randomized the targeting of M B -T185A-GFP, whereas all other mutants tested retained apical targeting similar to that of M B -GFP ( Fig.  6 and Table I). Therefore, O-glycans appear not to be required, but the modification of selected Ser/Thr residues can inhibit or potentiate the apical targeting of M B -GFP.
M AB -GFP Does Not Associate with Lipid Rafts-Sphingolipid and cholesterol-rich membrane microdomains, or lipid rafts, have been proposed to participate in the delivery of many apically sorted proteins (23), but enteropeptidase appears to be an exception. Bovine enteropeptidase, a type II transmembrane protein, was targeted to the apical membrane of transfected MDCK cells, but association with lipid rafts could not be demonstrated (2). Similar results were obtained for secreted M AB -GFP (Fig. 7), which contains a subset of the apical targeting signals found in full-length enteropeptidase. Triton X-100 lysates were prepared from MDCK cells expressing GFP and M AB -GFP, and the extracts were fractionated by sucrose gradient centrifugation (2). Caveolin, a marker for lipid rafts, was recovered in low-density fractions 4 -5 near the top of the gradient (Fig. 7). In contrast, GFP (Fig. 7A) Table I. B, the indicated proteins were treated with (ϩ) or without (Ϫ) neuraminidase and O-glycanase and analyzed by gel electrophoresis and Western blotting. gradient. Therefore, M AB -GFP is secreted apically but does not appear to associate intracellularly with lipid rafts.

DISCUSSION
The results presented here show that small mucin-like peptides of 12-15 amino acid residues, derived from bovine enteropeptidase, can direct the apical secretion of GFP. O-glycosylation contributes to the efficiency of apical targeting, but no specific O-glycan appears to be required. For mucin repeat B, some apical targeting activity persists after all glycosylation sites are eliminated by mutagenesis (Table I). Several O-glycosylated peptides have no targeting activity in this system, indicating that glycosylation alone is not sufficient. Some apical membrane and secreted proteins associate with lipid rafts, but this is not the case for transmembrane or soluble variants of enteropeptidase (2) Apically secreted mucin-GFP chimeric proteins also did not associate with rafts (Fig. 7). Therefore, enteropeptidase mucin-like domains contain relatively compact apical targeting signals that depend on both O-glycosylation and amino acid sequence context but appear to function independent of lipid rafts. This appears to be the first example of an O-glycosylated peptide that can confer apical targeting on a randomly sorted heterologous protein.
Previous studies have implicated O-glycans in the apical targeting of several proteins, although the evidence is indirect  Fig. 1) expressing the indicated amino acid sequences inserted between the signal peptide and GFP, and the polarity of protein secretion was assessed in transfected MDCK cells. Ala residues replacing Ser or Thr in wild type mucin-like repeats are in boldface. Values represent the mean Ϯ S.E. of at least three independent experiments.  Table I for the indicated constructs. The polarity of protein secretion (Panel A) and sensitivity to digestion with neuraminidase and O-glycanase (Panel B) were assessed as described under Fig. 2. and mixed. Sucrase-isomaltase (12) and the neurotrophin receptor (16) are membrane proteins with O-glycosylated stalk regions adjacent to their transmembrane domains. Both are targeted to the apical surface of MDCK cells, and deletion of their stalk domains abolishes apical targeting (12,16). These data suggest that O-glycosylation mediates apical targeting, but other studies suggest that sorting signals reside elsewhere. Replacement of the N-terminal transmembrane domain with a cleaved signal peptide causes the random secretion of sucraseisomaltase despite the presence of the O-glycosylated stalk, suggesting that membrane association contributes to targeting and the stalk region is not sufficient (12). The replacement of Gln-117 by Arg, at a location near the N terminus of the isomaltase domain, causes random delivery to the apical and basolateral membranes of MDCK cells and suggests that features of sucrase-isomaltase distinct from its stalk region are necessary for targeting (24). Furthermore, O-glycosylation does not correlate with the apical targeting of aminopeptidase N and lactase-phlorizin hydrolase, which are apical membrane proteins with O-glycosylated stalk regions near their transmembrane domains. Deletion of these stalk regions does not affect their apical targeting in MDCK cells (25,26).

EKLTTSIPLAAP
These various studies indicate that O-glycosylation can contribute to apical targeting but is not sufficient and sometimes is not necessary. Enteropeptidase mucin-like domains exhibit similar properties, in that their apical targeting activity is diminished by mutagenesis of only certain O-glycosylated residues (e.g. Thr-173 or Thr-185, Table I). Such residues might bear oligosaccharides that are particularly potent targeting signals, possibly disialylated species. Clustered O-linked oligosaccharides might be required for targeting, and disruption of a central member of the cluster could be sufficient to impair targeting. Although the accumulating data do not exclude a direct targeting function for O-glycans, other models appear to be equally plausible. For example, O-glycans could play an indirect role, supporting a primary targeting signal that resides elsewhere in the protein (27). Alternatively, features of protein and oligosaccharide structure could collaborate to form a complete targeting signal that contains both carbohydrate and peptide determinants, in which case these structures could be spatially close together. Any of these models would be compatible with the observation that short O-glycosylated peptides can target GFP for apical secretion.
The association of membrane proteins with lipid rafts often correlates with their delivery to apical cell surfaces but, as noted for the proposed relationship between O-glycosylation and targeting, many exceptions are known. Sucrose-isomaltase, aminopeptidase N, aminopeptidase A, and dipeptidyl peptidase IV associate mainly with lipid rafts (12,24,28), whereas lactase-phlorizin hydrolase (29), maltase-glucoamylase (28), and (full-length) enteropeptidase (2) do not, yet all are apical membrane enzymes of the intestinal brush border. Similar variability has been reported for apically secreted proteins, natural or engineered. Thyroglogulin (30) and soluble prohormone convertase 2 (31) have been recovered bound to lipid rafts, whereas clusterin (32) and the ectodomains of placental alkaline phosphatase and the neurotrophin receptor have not (33). Similarly, the mucin-like domains of enteropeptidase direct apical secretion of GFP in the absence of a demonstrable association with lipid rafts (Fig. 7). These results indicate that stable binding to lipid rafts is not a requirement for apical protein sorting. Weaker associations that are not preserved during the isolation of rafts might contribute to targeting, and different experimental approaches would be required to identify them.
Apical targeting can occur with or without stable raft association, glycosylation, or membrane anchoring. However, the apical targeting of specific proteins has been found to depend on one or more of these particular features. This heterogeneity may reflect the existence of multiple apical sorting pathways that employ distinct signals. Alternatively, the apparent diversity of signals could reflect an indirect role of various protein FIG. 6. Mutation of selected Ser/Thr residues in mucin repeat B. The amino acid sequence inserted before GFP is given in Table I for the indicated constructs. The polarity of protein secretion (Panel A) and sensitivity to digestion with neuraminidase and O-glycanase (Panel B) were assessed as described in the Fig. 2 legend.   FIG. 7. Relationship of chimeric M AB -GFP to lipid rafts. MDCK cells expressing GFP (Panel A) or M AB -GFP (Panel B) were lysed with ice-cold 1% Triton X-100, and the lysate was subjected to sucrose density gradient centrifugation. Fractions were collected from the top (fraction 1) to the bottom (fraction 10) of each gradient and analyzed by Western blotting for GFP or caveolin as described under "Experimental Procedures." structures, such as oligosaccharides, in the stabilization of a single class of sorting determinant that could be protein-based (27). These classes of mechanism need not be mutually exclusive. Studies of the relatively simple targeting signals present in enteropeptidase mucin-like domains may facilitate the evaluation of these models.