Abstract
Mucins are high molecular weight glycoproteins with a variety of postulated biological functions, including physicochemical protection from toxins and mutagens, adhesion modulation, signal transduction, and regulation of cell growth. Mucins are widely and differentially expressed in the gastrointestinal tract. To date, studies of cellular expression have relied on light microscopy using in situ hybridization and immunohistochemistry. Although informative, it has been difficult with these techniques to ascertain exactly which cell types are producing a given mucin. We studied expression of MUC1, MUC2, and MUC4 apomucins in a series of normal colon biopsies using a combination of immunoelectron microscopy and light microscopy. MUC1 mucin was localized to both goblet and columnar cells, where it was seen in secretory vesicles, microvilli, and in cytoplasmic remnants in goblet cell thecae. MUC2 expression was restricted to goblet cells, in which reactivity was concentrated in the rough endoplasmic reticulum (RER). MUC4 expression was seen in both columnar and goblet cells, localized to the RER. The inability to detect MUC2 and MUC4 apomucins in the Golgi complex and the mature mucous gel probably represents masking of peptide epitopes following O-glycosylation. This study has helped clarify lineage-specific mucin synthesis in the normal colon. (J Histochem Cytochem 47:1063–1074, 1999)
M
MUC1 is a transmembrane molecule with an extensive extracellular mucin domain (Gendler et al. 1991). MUC1 is a multifunctional molecule, with roles in modulation of cellular adhesion, maintenance of cellular polarity, signal transduction, and cellular immunology (reviewed in Gendler and Spicer 1995; Agrawal et al. 1998).
Located on chromosome 11p15.5 is a cluster of four mucin genes, MUC2, MUC5AC, MUC5B, and MUC6, which encode classical gel-forming mucins found predominantly in the gastrointestinal, reproductive, and respiratory tracts. These mucins are characterized by the presence of domains containing many cysteine residues that form intermolecular disulfide bonds, resulting in oligomerization critical for gel formation (Gum et al. 1992; McCool et al. 1994; van Klinken et al. 1998).
The gene encoding MUC4 is localized to chromosome 3q29 (Porchet et al. 1991; Gross et al. 1992). Recently, Nollet et al. (1998) have reported the MUC4 N-terminus sequence, which shows sequence homology with the rat sialomucin ASGP-1 (Wu et al. 1994). Sequence analysis also reveals putative cleavage sites, implying that MUC4 may be a secreted mucin, and the C-terminus contains two extracellular EGF-like domains that may bind c-erbB-2 like the rat homologue ASGP-2 (McNeer et al. 1997; Moniaux et al. 1999).
In the colon, the precise localization of cells producing the various mucin gene products is frequently difficult to ascertain with absolute certainty owing to section thickness at the light microscopic level, as well as difficulties in detecting fully glycosylated and oligomerized mucins. MUC1 apomucin has been localized to the crypt base by Carrato et al. (1994) and Cao et al. (1997). Synthesis of MUC2 has been attributed to the goblet cells of the gastrointestinal tract, in which anti-MUC2 VNTR antibodies produce intense cytoplasmic staining but fail to stain the mature mucin droplets or adherent mucous gel (Devine et al. 1993; Ho et al. 1993; Carrato et al. 1994). Secretory MUC2 can be demonstrated after removal of oligosaccharides (Ajioka et al. 1996). MUC4 is strongly expressed in the crypts of normal colon (Xing et al. 1997), where goblet cell staining was most conspicuous.
Very little work has been done on ultrastructural localization of mucin gene products. MUC1 has been demonstrated in breast carcinoma cells using the anti-MUC1 MAbs H23 (Rulong et al. 1995), Ca1, and HMFG2 (Beesley 1993). Using the MAb LDQ10, Gambús et al. (1993) demonstrated localization of MUC2 apomucin in goblet cells of the gastrointestinal tract (Gambús et al. 1993). To the best of our knowledge, there has been no report of ultrastructural localization of MUC4 apomucin. This article reports the results of investigations into the subcellular localization of MUC1, MUC2, and MUC4 apomucins in normal colon.
Materials and Methods
Monoclonal Antibodies
The MAbs used in this study are as follows: BC2 and MUSE11 against MUC1 (Ban et al. 1989; Xing et al. 1989); 3A2 against MUC2 (Devine et al. 1993); and M4.275 against MUC4 (Xing et al. 1997). All of these antibodies recognize peptide epitopes in the VNTR domains of their respective apomucins. The MAb 401.21, directed against α-gliaden, (Hill and Skerritt 1989), is an IgG1 (the same isotype as BC2, MUSE11, 3A2, and M4.275), nonreactive with human tissues, and was used as a negative control antibody to exclude nonspecific binding of mouse immunoglobulins to tissue sections.
Colon Tissue Samples
Cup biopsy specimens of normal sigmoid colon mucosa (n = 16) were obtained, with informed consent, from nine patients undergoing routine colonoscopy at the Royal Brisbane Hospital. The patients (two men and seven women) ranged in age from 33 to 88 years. Reasons for colonoscopy included abdominal pain (one), rectal bleeding (one), mild inflammatory bowel disease (one), surveillance after resection of a colonic adenocarcinoma involving the hepatic flexure (one), and follow-up of previous colorectal polyps (five). In all cases, the colon mucosa was macroscopically and histologically normal.
Samples were divided immediately for ultrastructural studies and for paraffin embedding for light microscopy and immunohistochemical analysis. Samples for light microscopy were fixed for 4 hr in 10% neutral buffered formalin before routine paraffin embedding, and representative samples of colon mucosa (approximately 0.5 mm3) were fixed for 2 hr at room temperature (RT) in freshly prepared 4% paraformaldehyde in 0.1 M Sörensen's phosphate buffer, pH 7.2, containing either 0%, 0.05%, or 0.5% glutaraldehyde for ultra-structural examination. After fixation, samples were washed three times with phosphate buffer and stored at 4C before final processing.
For ultrastructural studies, three techniques were used for dehydration and embedding of tissues: (a) dehydration and embedding in LR White resin (London Resin; Reading, UK) at RT, followed by heat (50C) polymerization (Newman and Hobot 1993a; (b) progressively lowered temperature (PLT) 0C to −45C dehydration and embedding in Lowicryl HM20 (Polysciences; Eppelheim, Germany), followed by UV polymerization (Armbruster et al. 1982; Robertson et al. 1992); and (c) freeze-substitution at −86C, followed by embedding in Lowicryl HM20 and UV polymerization at −45C (Schwarz and Humbel 1989; Newman and Hobot 1993b; Maunsbach 1994).
(A) MUC1 (BC2 MAb) expression is localized most conspicuously to the apical cell membranes. Staining of flocculent material within goblet cells is also apparent. Occasional goblet cells also demonstrate intense cytoplasmic reactivity with this antibody. Paraffin section. Bar = 20 μm. (B) MUC1 (MUSE11 MAb) staining confined to the apical cell membranes in the crypt base. There is relatively little cytoplasmic or goblet cell thecal staining. Paraffin section. Bar = 40 μm. (C) In one instance, intense MUC1 reactivity in the goblet cell contents was seen in an isolated crypt base. The crypt base at lower right shows similar staining to that seen in B. Paraffin section. Bar = 40 μm. (D) MUC2 (3A2 MAb) immunoreactivity restricted to the cytoplasm of goblet cells. Note absence of staining in the thecae. Paraffin section. Bar = 20 μm. (E) Goblet and columnar epithelial cells react strongly with M4.275 (MUC4) in the crypt base, with expression diminishing to infrequent cells on the surface epithelium. Paraffin section. Bar = 40 μm.
Dehydration and Embedding
Tissue samples for LR White resin embedding were dehydrated at RT through graded ethanol solutions, 20 min each in 30%, 50%, 75%, and 95%, then two changes each of 45 min in 100% ethanol. Resin infiltration was carried out in mixtures of ethanol:LR White resin (hard grade), 30 min each in 1:1 and 1:3, followed by three changes of 100% resin over a 12-hr period. Tissue was embedded in fully filled gelatin capsules and polymerized for 48 hr at 50C.
Samples embedded in HM20 resin with PLT were dehydrated through a series of aqueous ethanol solutions at progressively lowered temperature. Incubation times were 60 min for each step as follows: 30% ethanol at 0C, 50% at −20C, 75%, 95%, and 3 × 100% at −45C. Infiltration with Lowicryl HM20 was carried out at −45C in mixtures of ethanol:HM20 resin, 60 min each in 3:1, 1:1, 1:3, then pure resin, then pure resin overnight, and finally two changes of pure resin, each of 60 min duration. Final embedding was done in fully filled gelatin capsules. Resin polymerization was carried out under indirect 360-nm UV light (Philips TL 20W 05) at −45C for 48 hr, followed by a further 48 hr at 0C.
For freeze-substitution embedding, phosphate ions in the tissue were removed by rinsing three times in 0.1 M Hepes buffer, pH 7.2, over a period of 30 min. Samples were cryoprotected by immersion in 2.3 M sucrose in 0.1 M Hepes buffer for 12 hr at 4C. They were then plunge-frozen in liquid nitrogen, quickly transferred to prechilled −86C methanol containing 0.5% uranyl acetate (UA) (Schwarz and Humbel 1989), and stored in a −86C freezer (Forma Scientific; Marietta, OH) for 40 hr. The methanol:UA solution was then changed three times at 24-hr intervals with prechilled −86C methanol (no added uranyl acetate) before being transferred to a Revco −45C freezer (Revco Scientific; Ashville, NC). Infiltration and embedding with Lowicryl HM20 resin were carried out as described above for PLT, but substituting methanol for ethanol.
Sectioning
Blocks were sectioned with a Leica Ultracut UCT ultramicrotome (Leica; Vienna, Austria) using diamond and glass knives. Ultrathin sections 50–80 nm thick were collected on uncoated 300-mesh nickel grids. Semithin sections 0.5 μm thick were mounted on silanized glass slides.
Paraffin Section Immunohistochemistry
The immunohistochemical techniques have been described elsewhere (Biemer-Hüttmann et al. 1999). Briefly, paraffin sections (3–4 μm) were affixed to Superfrost Plus adhesive slides (Menzel-Gläser; Braunschweig, Germany) and air-dried overnight at 37C. After dewaxing in xylol and rehydration in Tris-buffered saline (0.05 M Tris, 0.15 M NaCl; TBS), pH 7.2-7.4, the sections were incubated in 1% periodic acid in distilled H2O for 30 min (MUC1) or in 0.1% porcine trypsin (ICN Biomedicals Australasia; Sydney, Australia) with 0.1% CaCl2 in PBS for 30 min (MUC2), or transferred to 0.01 M citric acid buffer, pH 6, and boiled twice for 5 min each and then transferred to TBS (MUC4) (Shi et al. 1991).
Endogenous peroxidase activity was quenched by incubating the sections in 1.0% H2O2, 0.1% NaN3 in TBS. Nonspecific antibody binding was inhibited by incubating the sections in 4% skim milk powder in TBS for 15 min, followed by a brief wash in TBS. The sections were then placed in a humidified chamber and incubated with 10% normal (nonimmune) goat serum (Zymed; San Francisco, CA) for 20 min. Excess normal serum was decanted from the sections and the primary antibody applied overnight at RT, except for BC2 and MUSE11 (MUC1), which were applied for 60 min. MAbs were used at the following concentrations: BC2 (1.5 μg˙ml-1), MUSE11 as neat tissue culture supernatant (TCSN), 3A2 (4 μg˙ml-1) against MUC2, and M4.275 (5 μg˙ml-1) against MUC4.
Sections were washed in three changes of TBS for 5 min each [the first buffer change contained 0.5% (v/v) Triton X-100] and then incubated with biotinylated goat anti-mouse immunoglobulins (Zymed) for 30 min. Sections were washed again in three changes of TBS for 5 min each [the first wash contained 0.1% (v/v) Triton X-100] incubated with streptavidin-horseradish peroxidase conjugate (Zymed) for 15 min, and washed in three changes of TBS for 5 min each. Color was developed in 3,3′-diaminobenzidine (Sigma Chemical; St Louis, MO) with H2O2 as substrate for 5 min. Then sections were washed in running tapwater, lightly counterstained in Mayer's hematoxylin, dehydrated through ascending graded alcohols, cleared in xylene, and mounted using DePeX (BDH Gurr; Poole, UK).
As negative controls, serial sections were stained as above but substituting 401.21 (anti-α-gliaden) at 5 μg˙ml-1 overnight in place of mucin antibodies, as well as staining sections as detailed above but incubating the sections with TBS alone, omitting the primary antibody.
The sections were scored by two observers (JRJ and MDW) using a teaching microscope, and a consensus was achieved for each section. Note was made of the architectural distribution of positively staining cells (crypt base or surface epithelium) as well as cellular localization (cytoplasmic, apical membrane, intracytoplasmic vacuoles), and staining intensity was scored subjectively.
Resin Section Immunocytochemistry
Pretreatment. Antibodies MUSE11 and BC2 (MUC1) and 3A2 (MUC2) required sections to be pretreated with 1.0% aqueous periodic acid for 10 min before washing and immunolabeling. M4.275 (MUC4) required sections to be heat-retrieved in 100C 0.01 M citrate buffer, pH 6.0, for 10 min, after which they were allowed to cool for 15 min before washing in deionized water and proceeding with the immunolabeling procedure. Heat retrieval of freeze-substituted tissues labeled for MUC1 and MUC2 was assessed but not found to be of benefit, so was omitted for these mucins.
(A) Semithin resin section stained with BC2 MAb (MUC1), showing intense apical membrane reactivity with increased label in the goblet cell mucin. (B) Semithin resin section showing MUSE11 (MUC1) membrane reactivity but sparse labeling of goblet cell contents. (C) MUC2 immunoreactivity in goblet cell cytoplasm. (D) Columnar and goblet cells stained intensely with MUC4 MAb. (A-C) PLT/HM20 resin, 4% paraformaldehyde, 0.5% glutaraldehyde fixation. (D) PLT/HM20 resin, 4% paraformaldehyde. Bars = 20 μm.
Immunogold Staining: Semithin Sections. In a moist chamber, semithin sections were covered with a blocking buffer of 4% normal (nonimmune) goat serum, 1% BSA, 0.05% Tween-20, and 20 mM NaN3 in 0.05 M TBS, pH 7.2, for 30 min at RT. After blotting off excess buffer, sections were incubated in primary antibodies of MUSE11 (neat TCSN), BC2 (10 μg˙ml-1), 3A2 (10 μg˙ml-1) or M4.275 (200 μg˙ml-1) diluted with blocking buffer for a period of 24 hr at 4C. Sections were washed six times for 5 min with TBS containing 0.05% Triton X-100 (TBS-TX), then incubated in 5-nm goat anti-mouse gold conjugate (British BioCell International; Cardiff, UK) diluted with 0.05 M TBS, pH 8.2, containing 1% BSA, 20 mM NaN3, and 0.05% Tween-20 to a final concentration of 1:150 for 3 hr at RT. Sections were washed twice for 5 min with TBS-TX buffer, then twice for 5 min in deionized water before being postfixed for 5 min in 2% aqueous glutaraldehyde. After washing four times for 5 min in deionized water, the sections were silver-enhanced using a SEKL15 silver enhancing kit (BBI) according to the manufacturer's instructions, the progress being periodically monitored in a light microscope and terminated by washing in deionized water when enhancement had reached a satisfactory level. Sections were then lightly counterstained with hematoxylin and eosin, washed in several changes of distilled water, dried, and mounted with BIOMOUNT mounting medium (British Biocell). Results were observed and photographed with a Nikon E800 photomicroscope (Nikon; Tokyo, Japan).
Immunogold Staining: Ultrathin Sections. In a moist chamber, grids were immersed in drops of blocking buffer for 30 min at RT. After blotting away excess buffer, grids were incubated in primary antibodies of MUSE11 (neat TCSN), BC2 (8 μg˙ml-1), 3A2 (6 μg˙ml-1), or M4.275 (160 μg˙ml-1) diluted with blocking buffer for a period of 24 hr at 4C. After this, grids were washed six times for 5 min with 0.05 M TBS, pH 7.2, containing 1% BSA, 20 mM NaN3, and 0.05% Tween-20 (wash buffer). Grids were then incubated for 2 hr at RT in goat anti-mouse-gold conjugate diluted with 0.05 M TBS, pH 8.2, containing 1% BSA, 20 mM NaN3, and 0.05% Tween-20 to a final concentration of 1:200 (5 nm; British Biocell), 1:40 (10 nm; Chemicon International, Temecula, CA) or 1:100 (15 nm; British Biocell). After gold labeling, grids were rinsed once in wash buffer, washed four times for 5 min with deionized water, then postfixed for 5 min with 2% aqueous glutaraldehyde. Sections were then thoroughly washed with deionized water and optionally silver-enhanced using a SEKL15 silver-enhancing kit (British Biocell) for 2–4 min. Enhancement was terminated by several rinses with deionized water. Sections were then counterstained in 5% aqueous uranyl acetate for 7 min and Reynold's lead citrate for 2 min, and washed with deionized water after each step. Grids were observed and photographed in a Jeol 1200 EXII electron microscope (Jeol; Tokyo, Japan).
Negative controls included substituting the first antibody with either similarly diluted normal goat serum or 401.21 (anti-α gliaden), and replacing the secondary antibody-gold conjugate for one that has an affinity for a species other than that of the first antibody.
Results
Immunohistochemical results for the three apomucins were essentially consistent with those previously described by Devine et al. (1993), Carrato et al. (1994), Ajioka et al. (1996), Xing et al. (1997), and Cao et al. (1997).
MUC1 Mucin (BC2 and MUSE11 MAbs)
In paraffin and semithin resin sections, MUC1 apomucin was detected in the cytoplasm of both columnar and goblet cells but was expressed most strongly on the apical membrane surface of the lowest part of the crypts. There was also evidence of staining of flocculent material within the goblet cell thecae in 11/16 biopsies (Figures 1A and 2A). In one instance, there was intense goblet cell thecal reactivity in the base of an isolated crypt (Figure 1C).
At the ultrastructural level, gold label was most concentrated on the apical surface (microvilli, glycocalyx) of both columnar and goblet cells in the crypts, label density being roughly equal (Figures 2A, 2B, and 3A–3C). A small amount of label was associated with small mucin vesicles in the upper sector of crypt columnar cells (Figure 3C). In some but not all biopsies, heavy MUC1 label was seen on cytoplasmic remnants and at the peripheral edges of mucin thecae in goblet cells (Figures 3A and 3D). Comparable with the results seen in paraffin sections, mature columnar and goblet cells on the surface epithelium did not label for MUC1. One biopsy did show atypical distribution of MUC1 when stained with BC2 and MUSE11. BC2 labeled areas of goblet cell microvilli only, whereas MUSE11 showed a more typical labeling pattern. In all biopsies there was also no evidence of MUC1 in endocrine cells or cells of the lamina propria.
MUC2 Mucin (3A2 MAb)
At the light microscopic level, MUC2 expression appeared restricted to cells clearly identifiable as goblet cells. The staining was seen as intense reactivity in the cytoplasm, particularly in the perinuclear region (Figures 1D and 2C). Only approximately 20–40% of goblet cells were reactive with MAb 3A2.
LR White resin, 4% paraformaldehyde, 0.5% glutaraldehyde fixation. (A) MUC1 (BC2)-labeled ultrathin section. Label is restricted to apical microvilli, cytoplasmic remnants (small arrows), and marginal edges of the theca (large arrows). 10-nm gold. Bar = 0.5 μm. (B) Columnar cell microvilli labeled with BC2 MAb (MUC1). 15-nm gold. Bar = 0.2 μm. (C) MUSE11 (MUC1)-labeled columnar cell microvilli. Note label on columnar cell secretory vesicles (arrows). Silver-enhanced 5-nm gold. Bar = 0.5 μm. (D) MUSE11 (MUC1)-labeled cytoplasmic remnants in goblet cell mucin. 10-nm gold. Bar = 0.25 μm.
Immunogold staining with MAb 3A2 localized MUC2 to the RER in all goblet cells (Figures 4A and 4B). The Golgi complex remained unlabeled. Unlike MUC1 and MUC4, for which label intensity was reduced in mature cells, goblet cells in the upper part of crypts and in the surface epithelium reacted as strongly to MUC2 labeling as those in the lower crypt. In no instances did columnar cells, endocrine cells, cells of the lamina propria, or secreted mucus label for MUC2.
MUC4 Mucin (M4.275 MAb)
MUC4 staining at the light microscopic level was most pronounced in the crypt base, with columnar and goblet cells being strongly labeled (Figures 1E and 2D). In all specimens there was a tendency for overall staining intensity to diminish with progressive cell maturity, such that the surface epithelium was frequently only sporadically and weakly stained. In these cases, the most intensely stained cells were goblet cells, whereas lower in the crypts the columnar cells were most intensely stained.
Ultrastructural immunocytochemistry localized the majority of MUC4 to the RER and, to a lesser extent, in transport vesicles in both columnar and goblet cells in normal colon epithelium (Figures 4C and 4D). The Golgi complex in each cell type failed to label for MUC4. Label was restricted mostly to the lower three quarters of the cells, with no label apparent on the luminal surface of either cell type. Mature columnar and goblet cells in the surface epithelium had little or no visible MUC4 reactivity. MUC4 label intensity was higher in columnar cells than in goblet cells. No MUC4 reactivity was seen in endocrine cells, cells in the lamina propria, or secreted mucus.
Technical Observations
Tissues fixed without glutaraldehyde were best processed by freeze-substitution to achieve the best preservation of cell organelles and overall architecture. Tissues fixed in the presence of 0.5% glutaraldehyde could be processed successfully by any of the techniques used and provided good ultrastructural morphology.
The addition of glutaraldehyde (up to 0.5%) to the fixative solution did not appear to be deleterious to MUC1 or MUC2 labeling with the MAbs used in this study. However, MUC4 antigenicity was severely affected by glutaraldehyde in the fixative; 0.5% glutaraldehyde almost completely masked MUC4 in colon tissue and 0.05% considerably reduced label density. Antigenicity was not retrievable. Consequently, only tissues fixed in 4% paraformaldehyde were used for detection of MUC4. However, even tissues fixed in 4% paraformaldehyde required heat retrieval in citrate buffer before useful levels of labeling were achieved.
There were differences between the two MUC1 MAbs in terms of the benefit produced by partially deglycosylating the semithin and ultrathin sections before immunostaining. Although BC2 reacted without prior section deglycosylation, staining was enhanced by periodate treatment. MUSE11, on the other hand, was unreactive without oxidization, although the maximal benefit appeared to be achieved by pretreatment with 1% periodic acid for 10 min, with no added staining seen with longer deglycosylation protocols. Apical membrane staining with MUSE11 was frequently stronger after periodate treatment than in similarly pretreated sections with BC2. MUC2 failed to label with 3A2 without prior deglycosylation in periodic acid.
The addition of uranyl acetate to methanol during freeze-substitution processing appeared to diminish MUC1 and MUC2 staining, although this effect could be partially reversed by the use of heat antigen retrieval in citric acid buffer. Uranyl acetate fixation during freeze-substitution reduced MUC4 labeling. Higher levels of label were observed in both LR White- and PLT-embedded tissues (which had no UA treatment). However, ultrastructural preservation was superior in tissues fixed with 4% paraformaldehyde when they were processed using freeze-substitution methods.
(A,B) LR White resin, 4% paraformaldehyde, 0.5% glutaraldehyde fixation; (C) PLT/HM20 resin, 4% paraformaldehyde fixation; (D) freeze-substitution/HM20 resin, 4% paraformaldehyde fixation. (A) Ultrathin resin section stained with 3A2 MAb (MUC2). Label is confined to goblet cell cytoplasm, with no evidence of staining in the adjacent columnar cell (C) or goblet cell Golgi complex (arrows). Bar = 1.0 μm. (B) MUC2 staining with 3A2 is restricted to the RER in goblet cells. Bar = 0.5 μm. (C) MUC4 labels both columnar (C) and goblet cells (G), with heavier label present in the columnar cells. Bar = 2.0 μm. (D) MUC4 staining with M4.274 MAb is restricted to the RER (arrows), with no evidence of membrane-associated mucin. Bar = 0.2 μm.
Discussion
This study has shown the benefits of both heat antigen retrieval and periodate deglycosylation in immunoelectron microscopic examination of mucin epitopes. Deglycosylation before immunohistochemical protocols is a well-established technique for exposing pep-tide epitopes that might otherwise be obscured by attached sugar chains. Periodate oxidation of tissues to partially cleave carbohydrate moieties to improve the demonstration of mucin antigens was first described by Bara et al. (1992) and has been successfully applied to immunoelectron microscopy without seriously compromising tissue morphology (Stirling and Graff 1995). Although most glycosylation occurs in the Golgi complex, limited O- and N-glycosylation of mucins has been detected in the RER (Egea et al. 1993). It may be necessary to ensure that there is no interfering carbohydrate present to adequately detect apomucins even in the RER. Alternatively, periodate pretreatment may be beneficial by an additional, as yet unrecognized, mechanism.
To the best of our knowledge, this study is the first to illustrate the reversibility of uranyl acetate fixation using heat antigen retrieval. Little is known about the mechanisms by which uranyl acetate binds to tissues, although Carlemalm et al. (1982) demonstrated binding of UA specifically to hydrophilic protein domains. The antibodies used in this study were all raised against hydrophilic sequences in the respective mucin VNTR regions, so heating may result in re-exposure of epitopes concealed by UA binding. Alternatively, UA may induce conformational changes in the structure of proteins rendering them less amenable to antibody detection. Langford (1983) described the induction of a globular form in normally linear microtubule-associated proteins on treatment with UA.
The biological functions of many of the mucins secreted by colon epithelium remain uncertain. On the basis of this study and others, it appears that MUC1 synthesis is greatest in the lower crypts and diminishes with cell maturation (Carrato et al. 1994; Ajioka et al. 1996). MUC1, which projects well above most other membrane-associated molecules, such as integrins, is believed to have anti-adhesive and protective properties. It is possible that the thickness and/or complexity of the mucous gel in the lower crypts is somewhat less than higher up the crypts and on the surface epithelium, and that MUC1 fulfills some of the protective functions normally associated with the mucous gel, which is predominantly composed of MUC2. It is recognized that the glycosylation of mucins produced in the crypt base is generally less complex than that of mucins produced by more mature cells (Jass 1997). MUC1 also has been implicated in the maintenance of cellular polarity, an important feature of epithelial cells, but why the need for this should decrease with progressive maturation is also unclear. The finding of MUC1 associated with cytoplasmic remnants trapped between goblet cell secretory granules is interesting because it appears to indicate that MUC1 and MUC2 are separately compartmentalized. It is difficult to assess in functional terms the contributions MUC1 might make to the final mucous gel layer.
MUC2 comprises the major proportion of colon mucous gels in biochemical assays (Tytgat et al. 1994), and its production by gut goblet cells has been recognized for some time. The inability to detect MUC2 core protein at the light microscopic level in a significant proportion of goblet cells was not supported by the ultrastructural studies, in which all cells identifiable as goblet cells had at least some gold labeling. This almost certainly represents differences between cells in MUC2 synthetic activity, such that cells producing low levels of protein are not detectable at the light microscopic level. It may well represent a cyclical pattern of activity and quiescence related to secretory status, with MUC2 synthesis being switched off in cells that have a full theca before the discharge of the contents, at which time synthesis resumes. Specian and Oliver (1991) reported that this process may take as little as 60 to 120 min. Alternatively, some goblet cells may have only one attempt at MUC2 synthesis, but mature at differing rates, leading to the scattered distribution of immature MUC2 staining. There is no readily apparent difference in the proportion of cells expressing MUC2 core protein between the lower crypt and surface epithelium, suggesting that the rates of production of the mucous gel are roughly equivalent among various architectural zones in the colon.
To date, no unequivocal roles for MUC4 mucin have been ascribed, although recent reports demonstrating homology between MUC4 and the rat sialomucin ASGP-1/ASGP-2 imply that this molecule may be involved in growth regulation. Like ASGP-2, MUC4 has two EGF domains capable of binding members of the erb-B family (McNeer et al. 1997; Moniaux et al. 1999), and the gradual loss of expression in mature surface epithelial cells may indicate differing cell growth dynamics between crypt base and surface epithelium. Although the amino acid sequence suggests that MUC4 is a transmembrane molecule, we have been unable to demonstrate any localization to the cell membranes with the antibody M4.275. However, M4.275 almost certainly is unable to detect mature glycosylated mucin, and translation to the membrane, if it occurs, must be a postglycosylation event.
It is well recognized that the mucins expressed by colorectal neoplasms differ from those of normal epithelium. We have interpreted the present samples as being normal despite the varied clinical indications for colonoscopy. There were no consistent differences among any of the samples. The illustrated material was derived from patients with abdominal pain and rectal bleeding who were found to be normal on colonoscopic examination.
Apart from alterations in glycosylation, two mechanisms may account for altered patterns of mucin expression in colorectal neoplasms: metaplasia and lineage preference, with columnar cells dominating over goblet cells or vice versa (Ajioka et al. 1997). A subset of colorectal cancers expresses MUC1 but little or no MUC2 (Blank et al. 1994; Ajioka et al. 1996, 1997). It has been suggested that the immature columnar cells of the cancerous epithelium might be related histogenetically to normal crypt base columnar cells (Ajioka et al. 1996). Consistent with this suggestion is the recapitulation of the mucinous phenotype of the crypt base by MUC1+ adenocarcinomas (Ajioka et al. 1997). In addition, these cancers include MUC1+ intracytoplasmic lumina, which are understood to be formed by an invagination of apical columnar cell membranes bearing microvilli. The present findings show, however, that goblet cells as well as columnar cells express MUC1. The finding of MUC1 in normal colorectal goblet cells does not preclude the concept of columnar lineage dominance in colorectal cancers, in turn providing an explanation for the mucinous phenotype MUC2−/MUC1+. A separate study has demonstrated the converse dominance by cells of goblet lineage in serrated neoplasms (Biemer-Hüttmann et al. 1999).
In conclusion, this study has demonstrated, at both the light microscopic and the ultrastructural level, variations in mucin core protein synthesis by goblet cells and columnar cells at differing stages of maturation. To the best of our knowledge, this is the first study demonstrating the separate compartmentalization of MUC1 and MUC2 mucins in goblet cell thecae. It is hoped that in linking the expression of apomucins MUC1, MUC2, and MUC4 to particular intestinal cell lineages, it may be possible to objectify patterns of differentiation in colorectal neoplasia. An approach to classification that relates neoplastic cells to normal counterparts is not novel and, in the case of the lymphoid and hemopoietic systems, for example, has achieved a high level of sophistication.
Footnotes
Acknowledgments
We wish to thank gastroenterologists Drs Graham Radford-Smith and Michael Ward for kindly providing clinical specimens used in this study, and Dr Yuji Hinoda and Dr Pei-Xiang Xing for their kind donations of MUSE11 and M4.275 MAbs, respectively. We also thank Dr Michael McGuckin for critical comments on the manuscript during its preparation.