Release of carcinoembryonic antigen from human colon cancer cells by phosphatidylinositol-specific phospholipase C.

Carcinoembryonic antigen (CEA) is released from colon cancer cells into the circulation where it is monitored clinically as an indicator of the recurrence or progression of cancer. We have studied the mechanism of CEA membrane attachment and release using the human colonic adenocarcinoma cell line LS-174T, specimens of human colon cancers, and serum from colon cancer patients. CEA release by cells in vitro and in vivo is associated with the conversion of CEA from a membrane-bound, hydrophobic molecule to a soluble, hydrophilic form with no apparent decrease in molecular mass. When LS-174T cell membranes were incubated with various buffers, proteases, and phospholipases, the only agents that released CEA and converted it to the hydrophilic form were preparations of phosphatidylinositol-specific phospholipase C (PI-PLC). Both [3H]ethanolamine and [3H]palmitate could be incorporated metabolically into CEA but only palmitate was released by treatment with PI-PLC, consistent with the presence of a glycosyl-phosphatidylinositol linkage. PI-PLC treatment also release significant quantities of CEA from living monolayers and from seven human colon cancer specimens. These experiments suggest that cellular CEA is anchored to membranes by a covalent linkage to a membrane phosphatidylinositol molecule, and that an endogenous phospholipase may be important for releasing CEA in vitro and in vivo.

Human bronchus and intestine express the same mucin gene.

The amino acid and sugar composition of mucins from various organs is similar but not identical. This could arise by one or more of the following: organ-specific processing of a single core protein, organ-specific splicing of a single mucin mRNA, or organ-specific expression of various mucin genes. To begin to investigate the source of this variability, we examined (a) immunological cross-reactivity and (b) cDNA cross-hybridization, among several mucin-secreting organs of the human body. Peptide-directed antibodies raised against both nondeglycosylated (LS) and deglycosylated (HFB) intestinal mucin strongly stained mucous cells in the bronchial epithelium and submucosal glands, indicating homology between mucins of the bronchus and intestine at the peptide level. By screening a bronchus cDNA library with an intestinal mucin cDNA, SMUC-41, we isolated a bronchus mucin cDNA, HAM-1. This cDNA is 96% homologous to the first repeat of SMUC-41. HAM-1 hybridized to restriction fragments of human genomic DNA identical to those hybridizing to SMUC-41 on Southern blots. SMUC-41 also hybridized to polydisperse transcripts in the bronchus, cervix, gall bladder, and mammary gland, indicating mucin homology among all these organs at the RNA level. We conclude that the bronchus and intestine express a common mucin gene, which is likely co-expressed by at least several other mucin-secreting organs.

MUC-2 human small intestinal mucin gene structure. Repeated arrays and polymorphism.

MUC-2, the first described intestinal mucin gene, has become important as a prototype for secreted mucins in several organ systems. However, little is known about its protein backbone structure and hence its role in diseases such as colon cancer, ulcerative colitis, and cystic fibrosis, which are known to have mucin abnormalities. Studies in this manuscript show that MUC-2 contains two distinct regions with a high degree of internal homology, but the two regions bear no significant homology to each other. Region 1 consists mostly of 48-bp repeats which are interrupted in places by 21-24-bp segments. Several of these interrupting sequences show similarity to each other, creating larger composite repeat units. Region 1 has no length polymorphisms. Region 2 is composed of 69-bp tandem repeats arranged in an uninterrupted array of up to 115 individual units. Southern analysis of genomic DNA samples using TaqI and HinfI reveals both length and sequence polymorphisms which occur within region 2. The sequence polymorphisms have different ethnic distributions, while the length polymorphisms are due to variable numbers of tandem repeats.

Biochemical and immunological characterization of mutant L-M cells with altered levels of dibutyryl cyclic AMP-inducible alkaline phosphatase.

Phosphotyrosine-Sepharose 4B was synthesized and used to purify L-cell alkaline phosphatase. Antibodies to this enzyme interacted with the alkaline phosphatase of strains A-1-2 and A-3-3, mutants that express the enzyme constitutively. This and thermal stability studies suggest that these mutants contain the same alkaline phosphatase isozyme as their parent strain.

Transcriptional regulation of the human placental-like alkaline phosphatase gene and mechanisms involved in its induction by sodium butyrate.

The human alkaline phosphatases constitute a multigene family with at least four members. Placental-like alkaline phosphatase (PLAP) is of particular interest because it is frequently present in tumors, where it serves as a marker of malignant transformation. Moreover, its expression is highly inducible by differentiating agents such as sodium butyrate. In the present study we have examined the PLAP gene promoter in order to better understand the mechanisms involved in its expression and induction. The PLAP promoters from four colon cancer cell lines with widely varied butyrate-inducible alkaline phosphatase activity were thermally amplified and sequenced. The overall sequence similarity of this region was found to be 99% between cell lines; thus, sequence variation of the promoter does not appear to account for the differential expression of this marker. We therefore analyzed the activity of the LS174T cell PLAP promoter using transient transfection experiments. Here, the 5'-flanking region of the gene was found to have positive regulatory elements in nucleotides -1 to -170 and -363 to -512 (relative to the start of transcription). A negative control element was also found to be present in the region between nucleotides -170 and -363. Mobility shift electrophoresis indicated that a nuclear factor bound to the promoter between bases -182 and -341. Furthermore, the activity of the PLAP promoter was found to be inducible by sodium butyrate. In contrast, the closely related placental alkaline phosphatase gene promoter exhibited almost no response to this agent. These results confirm that the activity of the PLAP promoter is stimulated by sodium butyrate and delineate regions that control this induction process.

Molecular cloning of complementary DNAs encoding alkaline phosphatase in human colon cancer cells.

We have isolated a set of complementary DNA (cDNA) clones that together encode the alkaline phosphatase of human colon cancer LS174T cells. These clones include two cDNAs isolated from a conventionally prepared oligodeoxythymidylate-primed lambda ZAP cDNA library and three cDNA clones prepared by using the polymerase chain reaction. The deduced amino acid sequence of the alkaline phosphatase primary transcript contains 532 amino acids. This enzyme is similar to, but not identical with, placental alkaline phosphatase (PLAP); it exhibits 12-19 amino acid substitutions when compared to the various alleles of PLAP. Also, it is similar to PLAP in that it is apparently attached to the cell membrane by a phosphatidylinositol-containing anchor as judged by the ability of phosphatidylinositol-specific phospholipase C to release it from membranes. It is different from PLAP however, in terms of its signal sequence which only contains 19 amino acids as compared to 22 for PLAP. Moreover, the 3'-untranslated region of the LS174T cell alkaline phosphatase message diverges considerably from the PLAP message. The LS174T cell alkaline phosphatase cDNAs are actually much more similar to the "germ cell" alkaline phosphatase gene than they are to PLAP. Only 7 amino acid substitutions exist between the LS174T cell enzyme and the alkaline phosphatase encoded by the germ cell alkaline phosphatase genomic DNA clone isolated by Millan and Manes (Proc. Natl. Acad. Sci. USA, 85: 3024-3028, 1988). Furthermore, the 3'-untranslated region of the LS174T cell alkaline phosphatase message is very similar to the sequence immediately downstream of the coding region of the germ cell alkaline phosphatase genomic DNA clone. Thus, these results indicate that this colon cancer cell alkaline phosphatase is likely to represent an allelic variant encoded at the germ cell alkaline phosphatase locus.

Heterogeneity in the induction and expression of carcinoembryonic antigen-related antigens in human colon cancer cell lines.

Sodium butyrate induces morphological and biochemical changes consistent with a more differentiated phenotype in some colon cancer cell lines. These changes include increased expression of carcinoembryonic antigen (CEA) and other oncodevelopmental markers. We utilized domain-specific probes and polyclonal antibodies against CEA-related antigens to study sodium butyrate-induced expression of the CEA gene family in a villous adenoma-derived cell line, which is nontumorigenic in nude mice (VACO 235), and two colonic carcinoma cell lines known to respond to sodium butyrate exposure by phenotypic differentiation (HT-29 and LS 174T). The induction begins as quickly as 24 h after exposure and occurs primarily at a transcriptional level, although some translational control is also evident. No evidence was found for gene amplification, rearrangement, or methylation to account for the mechanism of this transcriptional control. [35S]Cysteine pulse-labeled cell lysate immunoblots and polyadenylated RNA blot hybridization suggest that increases in mRNA transcript and CEA-related glycoprotein levels are primarily due to increased synthesis rather than decreased degradation. A considerable amount of heterogeneity is seen in the biosynthesis of the CEA-related glycoproteins, with each cell line showing a distinct pattern of CEA-related antigen expression from a limited number of mRNA transcripts.

Mouse intestinal goblet cells expressing SV40 T antigen directed by the MUC2 mucin gene promoter undergo apoptosis upon migration to the villi.

Mucinous colorectal cancers exhibit a characteristic set of molecular genetic alterations and may be derived from progenitor cells committed to the goblet cell lineage. Previously, we demonstrated that the MUC2 mucin gene promoter drives transgene reporter expression with high specificity in small intestinal goblet cells of transgenic mice. On the basis of these experiments, we reasoned that the MUC2 promoter could be used to drive SV40 T antigen (Tag) expression in the same cell type, decoupling them from their normal antiproliferative controls. A line of mice was established (MUCTag6) that expressed Tag in intestinal goblet cells as determined by RNA blot and immunohistochemical analysis. These goblet cells were markedly involuted however, most notably in the villi. Endogenous intestinal MUC2 message levels were reduced to about one third the normal level in these mice. However, absorptive cell lineage markers were comparable with nontransgenics. Bromodeoxyuridine-positive S-phase cells are limited to crypts in nontransgenic intestine but are present in both crypts and villi in MUCTag6. In contrast, mitotic cells were not present in the villi, indicating that MUCTag6 villi goblet cells do not progress into M phase. Apoptotic cells positive for terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling were increased more than fourfold in MUCTag6 villi (P < 0.0001), and apoptotic goblet cells were evident. Electron microscopic examination of MUCTag6 intestinal villi revealed the presence of degraded cell remnants containing mucin goblets together with other cell debris, further indicating apoptosis of the goblet cell lineage. Thus, the expression of Tag in intestinal goblet cells releases them from normal antiproliferative controls, causing their inappropriate entry into S phase even after they transverse the crypt/villus junction. They do not, however, progress to M phase. Instead, they undergo apoptosis with a high degree of efficiency in S or G(2) phase. These experiments demonstrate that apoptosis effectively blocks inappropriate goblet cell proliferation in the intestine, supporting its proposed role as an antineoplastic mechanism.

Analysis of dipeptidyl peptidase IV gene regulation in transgenic mice: DNA elements sufficient for promoter activity in the kidney, but not the intestine, reside on the proximal portion of the gene 5'-flanking region.

The dipeptidyl peptidase IV (DPPIV) gene encodes a brush border membrane exopeptidase that is expressed in a tissue-restricted fashion. To examine the regulation of DPPIV transcription in various tissues in vivo, we examined the expression of DPPIV 5'-flanking region (promoter)-human growth hormone reporter constructs in transgenic mice. These mice exhibited cell-type specific reporter expression in kidney. Surprisingly, however, only very low to non-detectable levels of reporter were found in small intestine. These results indicate that DNA elements sufficient for DPPIV expression in kidney, but not intestine, reside in the 5'-flanking region of the gene.

Alteration in mucin gene expression and biological properties of HT29 colon cancer cell subpopulations.

Previous studies from our laboratory have shown that HT29 cells selected by adaptation to methotrexate (HT29-MTX) express mature mucins that differ in their immunoreactivity to antibodies against gastric mucin and in the level of one of two major gastric mucin MUC5AC (MUC5) mRNA compared with parental HT29 cells. In this study, we examined the expression of another major gastric mucin, MUC6 mRNA, as well as that of MUC2, -3 and -5 mRNAs in HT29-MTX cells. We also examined their relationship to mucin-related antigen expression and biological properties of the cells such as adhesion to matrigel and E-selectin and in vitro invasiveness, liver colonising activity and degree of differentiation of nude mouse xenograft. Slot blot and Northern analysis revealed markedly increased levels of MUC5 mRNA but no change in MUC6 mRNA level in HT29-MTX cells compared with parental HT29 cells which express barely detectable levels of MUC6 mRNA. A nuclear run-on study showed that MUC5 mRNA was up-regulated at the transcriptional level. The marked increase in MUC5 mRNA was associated with a significant increase in the expression of human gastric mucin and apomucin antigens in HT29-MTX cells. When the adhesive capacity of two cell lines was compared, HT29-MTX cells showed significantly lower adhesion to E-selectin consistent with their lower expression of sialyl Le(x) and sialyl Le(a) antigens compared with HT29 cells. HT29-MTX cells also showed lower adhesive capacity to matrigel than HT29 cells. Interestingly, HT29-MTX cells exhibited significantly decreased liver colonisation capacity in nude mice following splenic vein injection. Furthermore, nude mouse xenograft tumours produced by HT29-MTX cells exhibited a significantly greater degree of differentiation, consisting of mucin-secreting glands than those produced by HT29 cells. In conclusion, these results indicate a shift of predominantly colonic-type mucins to the gastric type, specifically the surface epithelial cell type (MUC5) but not the mucous neck cell or antral gland type (MUC6) in HT29-MTX cells and strongly suggest that altered regulation of mucin genes and the degree of differentiation in cancer cells may be responsible for the altered biological behaviour of these cells.

Effects of prostaglandins, derivatives of cyclic 3':5'-AMP, theophylline, cholinergic agents and colchicine on clot retraction in dilute platelet-rich plasma and gel-separated platelet test systems.

In dilute suspensions of platelet-rich plasma (PRP) or gel-separated platelets (GSP), dibutyryl-cAMP (DBcAMP) and monobutyryl-cAMP inhibited platelet-mediated fibrin clot retraction in concentrations of 2--3 X 10(-6) M, with complete inhibition at 1--3 X 10(-4) M. Prostaglandin E1 (PGE1), which inhibited fibrin clot retraction in concentrations greater than 1.5--3 X 10(-8) M, was a more effective inhibitor than either PGE2 or PGF2 alpha. In the presence of theophylline (10-4 M), concentrations of DBcAMP, PGE1, PGE2 and PGF2 alpha necessary to inhibit fibrin clot retraction were reduced 50-fold for DBcAMP and 2.5 to 20-fold for the prostaglandins. In dilute PRP or GSP, inhibition of fibrin clot retraction does not result from inhibition of thrombin-induced platelet aggregation. Thus, compounds which increase platelet cAMP levels result in the inhibition of platelet-mediated fibrin clot retraction, and this inhibitory effect may be mediated, at least in part, through suppression of platelet contractility. Cyclic GMP, dibutyryl-cGMP and carbamylcholine-Cl (which stimulate guanylate cyclase) did not influence fibrin clot retraction, and did not prevent inhibition of fibrin clot retraction by DBcAMP and PGE1. Colchicine, in concentrations known to disrupt platelet microtubules (2.5 X 10(-6) M to 2.5 X 10(-3) M), had little inhibitory effect on either fibrin clot retraction or platelet (3H)-serotonin release.

Mucin genes and the proteins they encode: structure, diversity, and regulation.

Mucins are the structural components of the mucus gels that protect the respiratory, gastrointestinal, and reproductive tracts. These polydisperse glycoproteins (250,000 to 20,000,000 D) are approximately 80% carbohydrate on a mass basis and have a high intrinsic viscosity due to their large size and extreme hydrophilicity. Mucin oligosaccharides, the structures responsible for this hydrophilicity, are heterogeneous in size and structure but are chiefly O-linked, i.e., they initiate from N-acetylgalactosamine residues attached to threonine and serine residues of the polypeptide backbone. Our understanding of the structure of mucins has advanced rapidly in the last few years with the isolation and sequencing of cDNA clones that encode mucin polypeptide backbones. All currently well-characterized mucins have been found to contain extended arrays of tandemly repeated peptides rich in potential O-glycosylation sites. Less is known about the unique sequences that flank the tandem repeat arrays of secretory mucins, but currently available information indicates that these flanking regions contain cysteine-rich stretches that participate in mucin oligomer formation. Thus, secretory mucins appear to consist of oligomers containing heavily glycosylated domains flanked by unique sequences required for polymerization. Progress has also been made in characterizing the genes that encode mucins. At least four human mucin genes are known at present, although many others may remain to be discovered. Moreover, much work remains before we gain an understanding of the mechanisms involved in the expression of mucin genes and their tissue-specific regulation.

Human mucin glycoproteins: varied structures predict diverse properties and specific functions.

It has been clear for some time now that mucin glycoproteins have extensive, highly glycosylated tandem repeat domains. What is becoming increasingly apparent, however, is the diversity of unique sequences present on different mucins. Variations in mucin-unique sequences clearly result in major differences in physical and biological properties. MUC1-unique sequences are needed for membrane binding, while MUC2-unique sequences apparently mediate polymerization. As the sequences of other mucins are determined, undoubtedly additional structures will be determined that confer specific properties and functions upon this interesting class of glycoconjugates.

Isolation of the membrane-binding peptide of NADPH-cytochrome P-450 reductase. Characterization of the peptide and its role in the interaction of reductase with cytochrome P-450.

A peptide identified as the membrane-associated segment of NADPH-cytochrome P-450 reductase has been generated by steapsin protease treatment of vesicle-incorporated reductase and isolated by preparative gel electrophoresis. This peptide remains associated with vesicles when steapsin protease digests of vesicle-incorporated reductase were fractionated by Sepharose 4B chromatography, confirming its identity as the membrane-binding peptide. The molecular weight of the membrane-binding peptide was 6400 as determined by gel filtration in 8 M guanidine hydrochloride, and its amino acid content was not especially hydrophobic. The activity of reconstituted hydroxylation systems consisting of reductase, cytochrome P-446, and dilauroyl phosphatidylcholine was not inhibited by large molar excesses of purified membrane-binding peptide. Moreover, when purified reductase and cytochrome P-446 were added to liposomes which contained the membrane-binding peptide, it was determined that mixed function oxidase activity was reconstituted as effectively as when vesicles without the membrane-binding peptides were used. Similar results were obtained with reductase, cytochrome P-450, and detergent-solubilized liposomes (with or without the membrane-binding peptide). Thus, the membrane-binding peptide does not appear to interact with either of these two forms of the hemoprotein in a site-specific manner to prevent reconstitution of hydroxylation activity.

Dibutyryl cAMP-inducible alkaline phosphatase in animal cell plasma membranes: fluorescence detection of mutant clones on polyester cloth.

We have developed a rapid screening assay that allows us to estimate the alkaline phosphatase content of mouse L-M cell colonies immobilized on polyester cloth. This permitted the identification and isolation of two mutant clones with increased constitutive alkaline phosphatase activity and six clones that fail to express this activity when treated with dibutyryl cyclic AMP. Both of the strains with increased constitutive activity have basal enzymatic activities that are 6- to 7-fold higher than the activity of the parental strain. The extents to which the cyclic nucleotide further induces alkaline phosphatase in these two strains are different, however, indicating that they represent two classes of mutants. Studies using amino acids and synthetic peptides as alkaline phosphatase inhibitors suggest that only one alkaline phosphatase isoenzyme predominates, in both the parental and the mutant cell lines, with or without induction by cyclic nucleotide. Comparison to mouse tissues indicates that our cell lines express an isozyme resembling that found in kidney and bone. The six clones that fail to express alkaline phosphatase activity when treated with dibutyryl cyclic AMP also have extremely low basal levels of the enzyme. All of these mutant strains continue to synthesize protein when treated with dibutyryl cyclic AMP and undergo growth cessation and morphological changes in the presence of this agent. Thus, the mutations all appear to affect factors specific to the expression of alkaline phosphatase activity rather than factors that affect general cellular responsiveness or permeability to dibutyryl cyclic AMP. The characterization of these strains may help elucidate mechanisms of eukaryotic membrane protein biogenesis, enzyme induction, and regulation of gene expression by cyclic nucleotides.

Cyclic-AMP-stimulated synthesis and release of carcinoembryonic antigen by pancreatic cancer cells.

The effect of cyclic adenosine 3':5'-monophosphate (cAMP) upon the synthesis and release of carcinoembryonic antigen (CEA) was studied in the human pancreatic ductal cancer cell line, SW-1990. Incubation for up to 24 h with forskolin, an activator of adenylate cyclase, or isobutylmethyl xanthine, a theophylline analog, increased cellular cAMP levels by over 100-fold and significantly increased CEA release and cellular CEA content. Whereas cAMP levels were augmented within 10 min of exposure to these agents, CEA release and CEA cell content were not increased until 90 min and 24 h, respectively. Similar results were obtained using dibutyryl-cAMP, a cAMP analog, but not using sodium butyrate, a metabolite of dibutyryl-cAMP. Cells were incubated with 35S-cysteine and 3H-glucosamine in the presence or absence of forskolin in order to compare the effects of high cAMP levels upon the synthesis and release of total proteins, total glycoproteins, and immunoprecipitable CEA. Both CEA synthesis and release were enhanced by forskolin, but these effects were not specific to CEA since the release of labeled proteins and glycoproteins also increased. In addition, altered CEA expression caused by forskolin was consistently associated with a cessation of cell division, an effect which was reversible upon removing the agent. There was no effect upon cell morphology or viability. The data indicate that increased levels of cellular cAMP in pancreatic cancer cells is associated with decreased cell proliferation and increased expression of CEA and other glycoproteins.

Identification and characterization of the MUC2 (human intestinal mucin) gene 5'-flanking region: promoter activity in cultured cells.

The initiation point for MUC2 gene transcription is located within a 7000-base GC-rich region of the mucin gene cluster found on chromosome 11p15.5. The promoter activity of the 5'-flanking region of the MUC2 gene was examined following its cloning into the luciferase-producing pGL2-Basic reporter vector. A short segment comprising bases -91 to -73 relative to the start of transcription was found to be important for basal promoter activity in all cell lines tested. Electrophoretic mobility shift assays demonstrated nuclear protein binding to this region, which contains the consensus CACCC motif (5'-GCCACACCC). This element has been shown to be functionally important in several promoters that are active in diverse cell types. Competition experiments using an Sp1 oligonucleotide and antibody supershift experiments indicated that both Sp1 and other Sp1 family members bind to this element. Inclusion of the region between bases -228 and -171 in pGL2-Basic constructs increased normalized luciferase reporter activity by almost 3-fold in C1a cells, which produce relatively high levels of MUC2 mRNA. Significantly lower levels of normalized luciferase activity resulted when the same construct was transfected into cultured cell lines that express low or undetectable levels of MUC2, suggesting a possible role for this region in conferring cell-type specificity of expression. We also demonstrate, using actinomycin D, that the MUC2 mRNA is long-lived, at least in cultured cells. Moreover, no evidence was found that the MUC2 mRNA turned over more rapidly in LS174T cells, which produce relatively low levels of MUC2 mRNA, as compared with C1a cells, which produce high levels of mRNA. Thus a long mRNA half-life appears to be an important mechanism involved in achieving elevated levels of MUC2 mRNA.

Biosynthesis of two distinct types of mucin in HM3 human colon cancer cells.

Mucins, high-M(r) glycoproteins with a large amount of O-glycosidically linked carbohydrate, protect the colonic epithelial surface and are altered in ulcerative colitis and colon cancer. At least two mucin genes, MUC2 and MUC3, are expressed at high levels in the human intestine. As an experimental model for studying the biosynthesis of human intestinal mucins, we used HM3 colon cancer cells. When mature mucins labelled with [3H]glucosamine or [3H]threonine were analysed by gel filtration, it was found that secreted mucins (M(r) > 10(8) were larger than soluble cellular mucins (M(r) approx. 5 x 10(6)). Only secreted mucin was sensitive to reduction. Both MUC2 and MUC3 proteins, identified by labelling with [3H]threonine or [35S]cysteine and immunoprecipitation with antibodies to synthetic mucin peptides, were already of large size (M(r) > 180,000) by the earliest labelling time (5 min). The MUC3 precursor was completely degraded by trypsin, but the MUC2 precursor had a trypsin-resistant fragment of M(r) approx. 240,000 containing threonine and cysteine. The trypsin-resistant MUC2 fragment contained N-linked carbohydrate, as indicated by a decrease in size as a result of peptidyl N-glycosidase digestion or tunicamycin treatment of HM3 cells. These results show that HM3 colon cancer cells produce at least two distinct human intestinal mucins. They also indicate that the mechanisms of biosynthesis of intestinal mucins differ from those of other mucin-like glycoproteins that have been studied.

Molecular cloning of human intestinal mucin (MUC2) cDNA. Identification of the amino terminus and overall sequence similarity to prepro-von Willebrand factor.

Secretory mucins consist of a protein backbone that is catenated by disulfide bonds, heavily O-glycosylated, and packaged into storage granules prior to release from cells. In this paper, we identify and sequence cDNAs that encode the amino terminus of the MUC2 protein, a major form of mucin found in human intestines and airways. The protein sequence was found to contain a repetitive element of approximately 350 amino acids with considerable sequence similarity to the D domains of prepro-von Willebrand factor. A total of four of these D domains were found to be present in the MUC2 protein, three in the amino-terminal region, and one in the carboxyl-terminal region. Prepro-von Willebrand factor contains four D domains itself, and the overall positioning of the D domains in the two proteins is similar. Prepro-von Willebrand factor contains a 741 residue pro-protein that has been implicated in both the disulfide-linked oligomerization of von Willebrand factor and its packaging into secretory vacuoles. A similar region is present in the MUC2 amino terminus sequence, suggesting that the mechanisms involved in the polymerization and packaging of MUC2 may parallel those already described for von Willebrand factor.