Mapping of two new epitopes on the apomucin encoded by MUC5AC gene: Expression in normal GI tract and colon tumors

Authors


Abstract

Three hybridomas secreting monoclonal antibodies (MAbs) against human (62M MAb) or rat (463M and 589M MAbs) gastric mucins were isolated. These MAbs immunoreacted against a human recombinant protein encoded by the 3′ region of the MUC5AC gene. We have mapped 2 new gastric mucin epitopes and the M1-f epitope previously characterized by the 19/21M1 MAbs on MUC5AC-encoded apomucin. The M1-f, 463/589M and 62M epitopes are located in the MUC11p15/von Willebrand factor (vWF)-A3uD4 domain, in the D4-(vWF)-like domain and in the C- and CK-vWF-like domains of MUC5AC, respectively. The 463/589M and 62M MAbs stained the surface epithelium of human gastric mucosae, but not the normal colon mucosae (except 463/589M MAbs, which immunoreacted with 5 of 49 cases). All hyperplastic polyps are stained strongly with the 463/589M MAbs and faintly with the 62M MAb. In addition, 463/589M epitope was detected in 64% of the adenomas and in 93% of the mucosae adjacent to adenocarcinomas; in contrast, only 9% of the adenomas and 29% of the mucosae adjacent to adenocarcinomas expressed the 62M epitope. The expression pattern of the 463/589M epitope in colonic carcinogenesis is different from that of the 19/21M1 epitope, although the 2 epitopes are encoded by MUC5AC gene. © 2002 Wiley-Liss, Inc.

Gastric M1 mucin has been defined by several monoclonal antibodies (MAbs) as an early marker of human1,2 and rat3 colonic carcinogenesis. The gastric M1 mucin and MUC5AC gene have been detected in the human precancerous colonic47 and pancreatic8–10 epithelium using immunohistology and in situ hybridization, respectively. In pancreatic fluids removed using fine-needle aspiration under computerized scan control, an immunoradiometric assay (IRMA) of M1 mucin11 improves the diagnosis of mucinous cystadenomas, a precancerous lesion for which surgery was suggested.10 On the other hand, the early expression of this M1 mucin during colonic carcinogenesis could give excellent tools (i) for a better understanding of the early events associated with precancerous stages, (ii) for detecting patients with high a risk of developing colonic and pancreatic cancers and (iii) for studying the protective effect of drugs on the progression of colonic carcinogenesis.12 Consequently, a better understanding of the biochemical nature of the M1 epitopes will be of invaluable help in this process.

Eight anti-gastric M1 mucin MAbs were selected using immunohistology on gastrointestinal mucosae fixed with ethanol. These MAbs immunoreact against mucus cells of normal surface gastric mucosae, colonic adenomas and macroscopically normal colonic mucosae adjacent to adenocarcinomas but not against normal colon.4, 5 Among the 8 MAbs against gastric M1 mucin described up to now, 2 of them 19M1 and 21M1, which recognized the same or neighboring epitopes,5 immunoreacted against a recombinant protein encoded by the 3′ region of the MUC5AC gene13 when expressed in a baculovirus/insect cell system. During the selection of hybridoma cells secreting anti-gastric mucin MAbs, we have also isolated 3 other MAbs (463M, 589M and 62M)5,14 staining mucins from mucus cells of surface gastric epithelium and goblet cells of colonic adenomas but with patterns different from those observed using the other anti-M1 MAbs. Consequently, these 3 MAbs were not classified as “anti-M1 antibodies.” In our study, using transfection of small fragments of the 3′ region of the MUC5AC cDNA into COS-7 cells, we have mapped 2 new mucin epitopes within the already known M1-f epitope (MAbs 19/21M1)5 on the C-terminal region encoded by the MUC5AC gene. We determined the smallest cDNA fragment that encoded a peptide corresponding to each epitope. We then compared the immunohistologic pattern in normal gastrointestinal tract and colonic tumors (hyperplastic polyps, adenomas and mucosae adjacent to adenocarcinomas) of the 2 new mucin epitopes with that of the M1-f epitope.

MATERIAL AND METHODS

Mucin Antigens

Human Gastric M1 Mucin Preparation.

Mucin was isolated from mucinous ovarian cyst liquids or by scraping human gastric mucosa and purified using CsCl gradient and Sepharose chromatography, as already described.11

Recombinant MUC5AC Mucins.

Two fusion proteins containing the secretory alkaline phosphatase (SeAP) at its N-terminal end and MUC5AC cDNAs 3′-coding sequence at its C-terminal portion (L31 plasmid isolated by Lesuffleur: EMBL data bank accession number Z48314)15 were obtained as already described:13 M1-A encoded by a cDNA containing nt 21 to nt 1515 of L31 plasmid and M1-B encoded by a cDNA containing nt 21 to nt 3310 of L31 plasmid.

Anti-Mucin Monoclonal Antibodies

Cell Fusion.

A BALB/c mouse was immunized with mucin preparation isolated by scraping human or rat gastric mucosa. Human mucins were isolated as already described4 from normal gastric mucosa belonging to an OLe(a−b+) individual. Rat mucins (isolated by scraping gastric mucosae) were purified by chromatography on Sepharose 4B and Ultrogel A4 as already described.14 Supernatants of hybridomas were screened using the immunoperoxidase method on human pyloric-duodenal mucosa. Only the clones secreting MAbs that stained the mucus cells of surface gastric epithelium, but not the goblet cells of the normal colon, were selected for immunoreactivity on adenomas and/or mucosa adjacent to adenocarcinomas. The antibody isotype was determined by double immunodiffusion with class- and subclass-specific antisera (Nordic, Tilburg, The Netherlands).

Other Anti-mucin MAbs.

MAbs against the peptide core of gastric mucins were also used. These are termed anti-M1 MAbs: 1-13M1, 2-11M1, 2-12M1, 9-13M1, 58M1, 19/21M1 and 45M1, and they characterize the M1-a, -b, -c, -d, -e, -f and -g epitopes, respectively,4, 5 were also used. The 19M1 and 21M1 MAbs recognize the same or neighboring M1 epitope (M1-f).5 Another MAb raised against rat gastric mucin denoted 660 MAb, which recognized an oncofetal marker of rat colonic carcinogenesis, was also used.14

Immunohistology

Human Tissues.

Human gastrointestinal mucosae were obtained from kidney donors.4 Tissues were surgically removed within 5 min after kidney removal. Most patients were in their fourth decade of life (mean age 33 years) with ages ranging from 17 to 54 years. Patients had different ABO and Lewis phenotypes: ALe(a+b−), ALe(a−b+), ALe(a−b−), BLe(a+b−), BLe(a−b+), OLe(a+b−) and OLe(a−b+). Colonic adenocarcinomas including strips of mucosa adjacent to the tumor, adenomas and hyperplastic polyps were obtained from the Centre Médico-Chirurgical de la Porte de Choisy (Paris, France). They were surgically resected from the distal colon.

Immunoperoxidase.

Deparaffinized sections were preincubated for 3 min in PBS containing 0.1% Tween-20 and then incubated for 30 min with the undiluted hybridoma supernatants. After 3 rinses in PBS/Tween-20, the sections were incubated for 30 min with peroxidase-labeled anti-mouse IgG (H+L, Santa Cruz, CA) diluted 1:200 in PBS/Tween-20. After 3 rinses in PBS/Tween-20, peroxidase activity was revealed with amino-ethylcarbazol (Sigma, St. Louis, MO). Cell nuclei were stained with 1% hematein (2 min). Specificity of the immunoreactivity was controlled by inhibition of staining after absorption of 1 ml of hybridoma supernatant with 10 mg (dry weight) of lyophilized ovarian mucinous cyst fluids containing gastric M1 mucin.

Immunoradiofixation Assay

This method uses a solid phase system (polypropylene stars) to adsorb purified 21M1 IgG1 (first layer), as already described.5 After BSA saturation and 3 washes with PBS containing 0.1% Tween-20, the coated stars were incubated with 300 μl of antigen solution (M1-A or M1-B recombinant mucin: 200 μg/ml) overnight at 37°C. After 3 rinses with PBS/Tween, 300 μl of radiolabeled anti-mucin MAb (100,000 cpm) were incubated overnight at 37°C. After 3 washes with PBS/Tween, the radioactivity of each star was estimated in a gamma counter.

Competition of Immunoradiofixation.

This method was used to determine whether each of the anti-mucin MAbs characterized a different epitope. The first layer consisted of purified IgG1 from 1-13M1 MAb; the second layer was composed of M1 mucin preparation (1μg of carbohydrate) isolated from an ovarian mucinous cyst. The third layer was a solution of MAbs containing 100 μl of radiolabeled antibody (100,000 cpm in 100 μl) and 200 μl of the cold competitive MAb at 10 μg/ml. When inhibition was observed, different dilutions (from 1:1–1:1,000) of this cold antibody solution were used to indicate progressive inhibition. A control was performed where the cold MAbs were omitted in order to determine the amount of cpm corresponding to 100% of fixation of radiolabeled MAbs. Reverse inhibition experiments were performed for each pair of cold and radiolabeled MAbs.

Mapping of the Epitopes

Construction of Recombinant Plasmids.

A pKS-L31 plasmid15 containing the 3′ end cDNA of MUC5AC was digested with restriction enzymes (BamHI, XhoI, PstI and ApaI). After isolation by preparative agarose gel electrophoresis, the fragments obtained were purified using a QIAquick gel extraction kit (Qiagen, Hilden, Germany) and ligated into the pSecTag2 B vector (Invitrogen, Leek, The Netherlands). Transformation of Escherichia coli hosts was performed using a TOP10F' one shot™ kit (Invitrogen). Thus, 3 recombinant expression plasmids (Bam/Xho 1.8 kb, Bam/Pst 1.6 kb and Bam/Apa 1 kb) were obtained. Seven recombinant plasmids were obtained by amplification of fragments of interest by PCR (Fig. 1). Oligonucleotide primers used in amplification experiments were synthesized by Genset Oligos (Paris, France) (Table I). PCR products were inserted by TA Cloning® into the pCR®2.1 vector (Invitrogen). They were subcloned into the pSecTag2 B vector using restriction sites added to the primers. The recombinant plasmids obtained were designated with the name of the primers used (JES). The JES4/JES11 plasmid (732 bp) was digested with HindIII (at the 5′ end of JES4) and BamHI (nt 270), to obtain the JES4/Bam plasmid containing the first 270 nucleotides of L31 plasmid (Fig. 1). Constructs were sequenced by Génome Express (Montreuil sous Bois, France) with the T7 forward primer to confirm that the cDNAs were correctly fused to the Ig kappa chain leader sequence at the 5′ end.

Figure 1.

Figure 1

Diagram summarizing the results observed after transfection of different recombinant plasmids into COS-7 cells and incubation with the different anti-gastric mucin MAbs. Top: Schematic diagram of the c-terminal region encoded by the MUC5AC gene. Epitopes characterized by the different anti-gastric mucin MAbs are designated as follows: circles, 19M1 and 21M1; squares, 463M and 589M; triangles, 62M.

Table I. Sequence and location of primers used
Primer designationPrimer sequence (5′ to 3′)1Position2Orientation3
  • 1

    Added;1> enzyme restriction sites are underlined.

  • 2

    The position of nucleotides in this table and in the text corresponds to the L31 sequence.153Strand orientation: sense (S), antisense (AS).

JES3ATG CGG CCG CGT GCA TGG GGG ACA CTG GGA CGC CTC T2826–2851AS
JES4CGC AAG CTT AAC CAG GAC CAG CAG GGA CCC T3–24S
JES8CGC AAG CTT CAG CCC GGC GCC GTG GTC TCC2061–2081S
JES9CGC GAA TTC CAT ATG GGA GGT TCT CCT TGG2879–2899AS
JES11CAC CAG CAC GTA CGT TCA GTT GTC CAG GAA G704–734AS
JES14CGC AAG CTT TTC CTG GAC AAC TGC ACG TAC705–725S
JES18CGC AAG CTT GCT TAC TCC ACC CAA ACC TGC216–236S
JES25CGC GAA TTC TAC ACA GGG CAT AAT AGC AAT G306–327AS
JES26CGC GAA TTC ATG GCA GGC ACG GAG CAG GCG1300–1320AS
JES27CGC GAA TTC TAT GGT ACA CAG CAC AGG TCG2479–2499AS
JES29CGC AAG CTT TGT GCT GTG TAC CAT AGG AGC2484–2504S
JES31CGC GAA TTC TGG GTG TGG CCC AGG TCT CAC475–495AS

Transfection of Recombinant Plasmids.

COS-7 cells were transfected with pSecTag2, carrying fragments of the 3′ end cDNA of MUC5AC. pSecTag2 vector expresses and secretes the recombinant peptide fused to the c-myc epitope, which is used as a positive control. Transfections were performed in 35 mm dishes using 2 μg plasmid DNA and 10 μl SuperFect™ Transfection Reagent (Qiagen) per dish delivered in a volume of 700 μl. SuperFect Reagent was added to plasmid DNA diluted in 100 μl DMEM, and the mixture was mixed and incubated at room temperature for 10 min. DMEM supplemented with 10% FBS was added, and the solution was mixed by pipetting, transferred onto PBS-washed COS-7 cells (700 μl/well), incubated at 37°C and 5% CO2 for 3 hr and removed. Cells were gently washed once with sterile PBS, and DMEM supplemented with 10% FBS was added. Cells were incubated for 48 hr.

Immunocytology

Cells were fixed for 1 hr at −20°C with methanol. Dishes were incubated 3 times in PBS/Tween-20 for 5 min before incubation with anti-gastric mucin MAbs for 30 min. After 3 washes with PBS/Tween-20, peroxidase-labeled anti-mouse Ig (diluted 1:200; Santa Cruz Biotechnology, Santa Cruz, CA) was incubated for 30 min. Immunoperoxidase reactions were performed using aminoethylcarbazol. Before microscopic examination, cell nuclei were stained with 1% hematein.

Immunochemical Characterization on Artificial Peptides

One hundred sixty-one overlapping 15-mer peptides frameshifted by 5 residues, representing the complete peptides encoded by the JES4/JES11, Bam/Apa 1kb and JES8/JES9 plasmids (Fig. 1), were synthesized on a cellulose membrane (Abimed, Langenfield, Germany) by the Spot technique.16 The synthesis was performed by using an ASP 412 spotter (Abimed), as previously described.17 The set of membrane-bound peptides was probed by incubation with anti-gastric mucin MAbs (diluted 1:100). The binding was revealed by alkaline phosphatase-conjugated anti-mouse antibody (Sigma; diluted 1:1000), as described.16 The membrane was reused when necessary after stripping the membrane with regeneration buffers containing urea and mercaptoethanol so as to remove precipitated dye and bound antibodies.

RESULTS

463M, 589M and 62M MAbs: Characterization and Isotype

The 62M and 463/589M MAbs were obtained after cell fusion of SP2O cells with spleen cells from mouse immunized with human and rat gastric mucin, respectively. They were selected for their reactivities on mucus cells of surface epithelium of the human pyloric mucosa, on the colon adenomas and on the mucosae adjacent to adenocarcinomas. They were also selected for their absence of immunoreactivity on the normal human colon. Hybridoma cells were cloned twice. Immunodiffusion with specific class antisera indicated that the 62M, 463M and 589M MAbs were IgG1.

Mapping of the Epitopes

Immunoreactivity of 463/589M and 62M MAbs on Recombinant M1A and M1B Mucins.

The IRMA curves show that the 463M MAb immunoreacted on M1-A and M1-B recombinant mucins as the 21M1 MAb, whereas the 62M MAb immunoreacted on M1-B exclusively (Fig. 2). On the other hand, the competition of immunoradiofixation shows that the inhibition of radiolabeled MAb by unlabeled MAb occurred only when both MAbs were identical. However, cross-inhibitions were observed with 463M and 589M MAbs, thus demonstrating that these MAbs recognized an identical epitope or neighboring epitope (Fig. 3).

Figure 2.

(a) IRMA using M1-A recombinant mucin. (b) IRMA using M1-B recombinant mucin. Circles, 21M1 MAb; squares, 463M MAb; triangles, 62M MAb.

Figure 3.

Competition of immunoradiometric fixation. Quantity of cpm obtained by incubation of 100 μl of 125I-463M IgG (a) or 125I-589M IgG (b) containing 100,000 cpm with increasing concentrations of unlabeled anti-gastric mucin. Solid squares, 463M MAB; open squares, 589M MAb. The solid phase was coated by using IgG1 of 1-13M1 MAb. The second layer was composed of an M1 antigen preparation (1 μg of carbohydrates).

Construction of Recombinant Plasmids by Direct Digestion of PKS-L31 Plasmid.

First we obtained 3 cDNA fragments cut with restriction enzymes that permit insertion into the expression vector. These recombinant expression plasmids (Bam/Xho 1.8 kb, Bam/Pst 1.6 kb and Bam/Apa 1kb; Fig. 1) contain fragments with the same 5′ end but of different lengths. The recombinant plasmid Bam/Xho 1.8 kb lacks the first 268 nucleotides of recombinant M1-A but is 566 nucleotides longer on the 3′ end. The Bam/Xho 1.8 kb plasmid was transfected into COS-7 cells. Then cells were incubated with different anti-mucin MAbs. It appeared that neither 19/21M1 MAbs nor 62M MAb reacted with the fusion protein encoded by this fragment, whereas 463/589M MAbs reacted with the recombinant protein. In a subsequent experience, the COS-7 cells were transfected with Bam/Pst 1.6 kb or Bam/Apa 1 kb plasmids. The same results were obtained. These first results allowed us to localize 3 portions of MUC5AC cDNA-encoding epitopes recognized by 19/21M1, 463/589M and 62M MAbs. To define these epitopes further, other recombinant plasmids were obtained by PCR amplification.

Characterization of the 19/21M1 Epitope.

19/21M1 MAbs did not react with any of the 3 recombinant proteins encoded by the plasmids mentioned above. However, these MAbs recognized the recombinant protein encoded by the JES4/JES11 plasmid, which contains the first 734 bp of L31. These findings suggested that the M1-f epitope, which is recognized by the 19/21M1 MAbs, could be encoded by the first 270 bp. Surprisingly, however, no reactivity was observed with the recombinant protein encoded by the JES4/Bam plasmid containing the first 270 nucleotides of L31. To characterize the M1-f epitope further, the overlapping fragments JES18/JES31 (nt 216 to nt 495) and JES4/JES25 (nt 3 to nt 327) were included into the pSecTag2 vector after amplification by PCR (Fig. 1). The protein encoded by the JES18/JES31 cDNA fragment was recognized by 19/21M1 MAbs, but the one encoded by the JES4/JES25 cDNA fragment was not, although the recombinant protein was correctly expressed, as demonstrated using an anti-myc antibody.

Characterization of the 463/589M Epitope.

The 463/589M MAbs reacted with the recombinant proteins encoded by the plasmids mentioned above (Bam/Xho, Bam/Pst and Bam/Apa). The 463M MAb reacted with more cells than the 589M MAb. Moreover, the immunoperoxidase reaction was also more intense. No immunostaining was obtained on the COS-7 cells after transfection with the JES4/JES11 (nt 3 to nt 734) cDNA fragment. Thus, 463/589M MAbs might react with an epitope encoded by the fragment of the 3′ end MUC5AC cDNA situated between nt 734 and nt 1232. The JES14/JES26 plasmid (nt 705 to nt 1320) (Fig. 1) was transfected to verify this hypothesis. The protein encoded by this fragment was recognized by 463/589M MAbs.

Characterization of the 62M Epitope.

The 62M MAb did not react with any of the recombinant proteins encoded by the plasmids mentioned above. However, this MAb recognized the recombinant protein encoded by JES8/JES9 plasmid (Fig. 1), which contains 840 bp encoding the last 264 amino acids of MUC5AC. The fragments JES8/JES27 (nt 2061 to nt 2499) and JES29/JES3 (nt 2484 to nt 2851) were obtained by PCR and included in the expression vector, but the 62M MAb did not immunoreact with COS-7 cells transfected by these cDNA fragments.

Identification of Peptide Epitopes by the Spot Method.

The amino acid sequences of JES4/JES11, Bam/Apa 1kb and JES8/JES9 plasmids (Fig. 1) were presented in the form of an array of 161 overlapping 15-mer peptides (5-residue overlap) synthesized on a cellulose membrane. The proteins encoded by these plasmids are recognized by the 19/21M1, 463/589M and 62M MAbs, respectively. The 5 different anti-gastric mucin MAbs were evaluated individually for their capacity to react with the set of immobilized peptides. No positive result was observed.

Immunoperoxidase Localization

Staining of Normal Gastrointestinal Tissues.

Both the 463M and 589M MAbs stained strongly the mucous cells of surface gastric epithelium and a few goblet cells of the duodenum near the pylorus (Table II). The duodenum, as well as the jejunum, ileum, cecum and right and left colon, were unstained. However, in 5 of 49 cases, small intestinal and colonic goblet cells were strongly stained. This staining was not related to any ABO or Lewis phenotype of individuals.

Table II. Immunostaining of anti-M1 MAbs in normal gastrointestinal mucosae and tumoral colonic mucosae1
MucosaeMAbs
19/21M12463/589M62M362M4
  • 1

    Scoring: −, no reactivity; +/− very low reactivity, +, low reactivity, ++, moderate reactivity, +++, strong reactivity; +++++, very strong reactivity; nd, not done.

  • 2

    See refs. 5 and 13.

  • 3

    After 1 hr of incubation.

  • 4

    After overnight incubation.

Normal
 Pylorus+++++(30/30)+++++(30/30)++(14/16)+++(16/16)
 Duodenum+/−(30/30)+/−(25/30)(16/16)+/−(16/16)
+++(5/30)
 Jejunum(5/5)(5/10)(5/5)ndnd
+(5/10)
 Left colon(49/49)(44/49)(49/49)ndnd
+++(5/49)
Tumoral
 Hyperplastic polyps+++++(8/8)+++(8/8)+(8/8)+++(8/8)
 Adenomas+++(21/25)++(16/25)+(1/11)+++(7/11)
 Transitional mucosae+(25/26)+++(26/28)+/−(8/28)+(7/8)
 Adenocarcinomas+++(1/27)+(1/27)+(1/27)++(1/6)

After standard processing, the 62M MAb stained mucus cells of surface gastric epithelium, but the staining was weaker and less homogenous (absent in patches) than that of the 463/589M or 19/21M1 MAbs. The 62M MAb did not stain the small and large intestine. Duodenal goblet cells near the pylorus-duodenal junctions were negative. However, after overnight incubations of 62M antibody at room temperature, the 62M immunoreactivity pattern was identical to the 19/21M1 immunoreactivity throughout the normal gastrointestinal tract.

Tumoral Human Mucosae.

The 463/589M and 62M MAbs stained the goblet cells of all hyperplastic polyps (8 of 8), as did the 19/21M1 MAbs. However, in addition, the 463/589M MAbs stained goblet cells in the mucosae adjacent to hyperplastic polyps. The 62M MAb staining was very weak. The same phenomenon was observed with the adenomas. The 62M MAb stained only 1 of 11 adenomas versus 16 of 25 and 21 of 25 with 463/589M and 19/21M1 MAbs, respectively. In the transitional mucosae (mucosae macroscopically normal adjacent to adenocarcinomas), 463/589M MAbs stained most of the goblet cells. However, using serial sections, the glands that were strongly stained with 19/21M1 MAbs were also more strongly positive with 463/589M MAbs (Fig. 4, arrow). Serial sections showed that the 19/21M1 MAbs stained the same areas as the 6 other anti-M1 Mabs (Fig. 4b). The 62M MAb stained this transitional mucosae faintly, and there were less goblet cells positive with 62M Mab than with 19/21M1 MAbs. However, after an overnight incubation, using serial sections, the 62M immunoreactivity pattern was identical to the 19/21M1 immunoreactivity in the transitional mucosae (Fig. 4c) as well as in the hyperplastic polyps and the adenomas.

Figure 4.

Immunoperoxidase staining of serial sections of histologic normal colonic mucosa adjacent to adenocarcinoma (×100) stained with 463M MAb (a), 21M1 MAb at 1 hr incubation (b) and 62M at overnight incubation (c). In the 3 cases, the same glands are strongly stained by the 3 MAbs (arrow).

DISCUSSION

In a previous study, we have shown that the M1-f epitope, immunoreacting with 2 anti-M1 MAbs (19M1 and 21M1), is encoded by the MUC5AC gene.13 This was accomplished by using a baculovirus/insect cell system to express the 3′ end of human MUC5AC cDNA, which was subsequently recognized by 2 MAbs (19M1 and 21M1) raised against M1 mucin, suggesting that M1 mucin may well be a product of the MUC5AC gene. The MUC5AC gene product is a secreted gel-forming mucin encoded by a gene contained in a cluster on chromosome band 11p15.5 with 3 other gel-forming mucin genes: MUC2,MUC5B and MUC6.18 The C-terminal regions of MUC2-, MUC5AC- and MUC5B-encoded apomucins exhibit striking sequence similarities with the D4, B, C and CK domains of the human von Willebrand factor.19, 20, 21 The C-terminal region of the MUC6-encoded apomucin has been reported to contain only the CK-vWF-like domain.22

In our study, using a molecular biological approach, we have mapped the M1-f epitope to the C-terminal region of the MUC5AC-encoded protein. This epitope is located within a peptide (encoded by nt 216 to nt 495 of cDNA of the L31 plasmid) containing the MUC11p15-type and the beginning of the A3uD4-vWF-like domains20,21 (Fig. 5). The amino acids building the M1-f epitope appear to be located in a peptide encoded by nt 216 to nt 269, but they themselves are not sufficient. In fact, some amino acids located downstream (encoded by a fragment beginning at nt 327) seem necessary for the recognition of the M1-f epitope by 19/21M1 MAbs. The other anti-gastric M1 mucin MAbs did not recognize the C-terminal region of the MUC5AC-encoded apomucin. However, surprisingly, 3 new anti-gastric mucin MAbs, showing a different immunohistologic pattern on colonic adenomas as well as mucosae adjacent to adenomas than M1 MAbs, immunoreacted with recombinant proteins encoded by MUC5AC cDNA. The 463/589M epitope is located in the D4-vWF-like domain (peptide encoded by nt 705 to nt 1320). The 62M epitope is located in a peptide containing the C- and CK-vWF-like domains (encoded by nt 2061 to nt 2899). The C-terminal region of the MUC5AC-encoded apomucin is implicated in the oligomerization of this mucin23–25 and in the binding to pS2/TFF1.26 Consequently, these MAbs could be useful for the future study of these interactions.

Figure 5.

(a) Partial restriction map of the L31 plasmid (3310 bp). (b) Apomucin region (1042 amino acids) corresponding to the partial restriction map of the L31 plasmid. Epitopes characterized by the different anti-gastric mucin MAbs are designated as follows: circles, M1-f; squares, 463/589M; triangles, 62M. Numbers correspond to the nucleotide or amino acid numbers of the L31 sequence.15 (c) Schematic diagram of the C-terminal region encoded by the MUC5AC-encoded gene.20,21

To determine the peptide sequence that builds the anti-gastric mucin epitopes, we performed the Spot method using sets of overlapping 15-mer peptides corresponding to the sequences found by molecular biology. Unfortunately, no positive reactions were obtained. Therefore, the epitope recognition of these anti-gastric mucin antibodies would appear to be dependent on the tertiary structure of the protein. These results suggest that these epitopes are built by amino acids that are not sequential but discontinuously located throughout the immunoreactive peptide core. These data could explain the destruction of these epitopes with reducing agents. Moreover, this is in agreement with the observation that most MAbs to globular proteins show a conformation-dependent recognition27 and therefore do not react with short peptides.

The 3 gastric mucin epitopes have been localized to domains within the C-terminal region of the MUC5AC gene product. This region exhibits striking sequence similarities with domains of the C-terminal regions of MUC2-, MUC5B- and even MUC6-encoded apomucins (with regard to the CK-vWF-like domain). Therefore, a cross-immunoreactivity of these anti-gastric mucin MAbs with another mucin other than the MUC5AC gene product could be suspected. Nevertheless, among the genes encoding mucins, only the MUC5AC gene was found to have the same expression pattern (using in situ hybridization) as the M1 mucin epitopes (using immunochemistry). The M1 mucin is restricted to columnar mucus cells of gastric surface epithelium, goblet cells of the tracheal surface epithelium and colonic adenomas consistent with the distribution of MUC5AC gene by in situ hybridization.6, 8, 28, 29 Thus, all these results suggest that the 19/21M1 MAbs are specific to the MUC5AC gene product.

Until now, only a few MAbs against the MUC5AC gene product have been reported.30, 31 They were obtained against synthetic peptides (TTSTTSAP) encoded by the tandem repeat domains of the MUC5AC gene (CLH2)30 or against deglycosylated mucin (SOMU-1)31 but not against native gastric mucin, as were the anti-M1 antibodies.13 The CLH2 MAb30 immunoreacted strongly with the deglycosylated forms of gastric mucin and weakly with the native form. In immunohistochemical studies, this MAb stained supranuclear areas but not the mucin vesicles in the cytoplasm, as did the anti-M1 MAbs. On the other hand, the SO-MU1 MAb31 immunoreacted strongly with native gastric mucin, as did anti-M1 MAbs (unpublished data), and although the SO-MU1 MAb was obtained against the deglycosylated form, it stained the cytoplasm of columnar cells of surface gastric epithelium. However, to our knowledge, the localization of the DNA fragment coding the SO-MU1 epitope on the MUC5AC gene has not been precisely reported, but may be associated with a small cystein domain located between 2 tandem repeat regions.31

The staining patterns of 463/589M and the 62M MAbs on polyps and transitional mucosae were different from those of typical anti-M1 MAbs. Indeed, the 463/589M MAbs stained 5 of 49 normal colons. This strong immunoreactivity on the normal intestinal mucosae is not very well understood because the MUC5AC gene was never expressed in the goblet cells of intestinal mucosae, as assessed by in situ hybridization.28 On the other hand, the 463/589M MAbs stained more colonic goblet cells than the anti-M1 MAbs in the histologic normal colonic mucosae adjacent to tumoral mucosae (Fig. 4). In the case of the 62M MAb, a weaker immunoreactivity than the anti-M1 MAbs was found with the standard immunoperoxidase method. However, after an overnight incubation at room temperature, the 62M immunoreactivity pattern was similar to the M1 pattern in the polyps as well as the mucosae adjacent to adenocarcinomas. This phenomenon could be explained by a lower antibody concentration or a weaker affinity than the other anti-M1 MAbs. Finally, this 62M MAb could be regarded as a genuine anti-M1 MAb.

In conclusion, we mapped on the 3′ region of the MUC5AC gene, 3 cDNA fragments coding 3 conformational epitopes. Their corresponding MAbs immunoreacted strongly with the native mucin of the surface gastric epithelium and more or less strongly on the hyperplastic polyps, adenomas and the mucosae adjacent to adenocarcinomas. The precise determination of epitopes specifically expressed during colonic and pancreatic carcinogenesis will provide useful tools for the early detection of these severe malignancies.

Acknowledgements

The authors are grateful to Drs. C. Granier and S. Villard (CNRS UMR 5094, Faculté de Pharmacie, Montpellier, France) for their able assistance in performing the Spot method. They thank Dr. T. Lesuffleur for providing pBS-L31 plasmid, Dr. N. Maurin and C. Guernier for technical help and D. Catala for efficient contribution to the cell culture.

Ancillary