Membrane-Associated Mucins in Normal Human Conjunctiva

Abstract

purpose. To examine the presence of specific membrane-associated mucins in normal human conjunctiva.

methods. Glycoconjugates were extracted from membranes with two detergents: octylglucoside and Triton X114. Mucins were separated by cesium chloride density gradient centrifugation. Size was assessed by gel filtration on Sepharose CL2B and charge by ion-exchange chromatography on MonoQ. Cross reaction with antibodies against mucin gene products was assessed in blots of electrophoresis gels.

results. Extraction of total tissue membranes yielded material with a buoyant density typical of mucins. Gel filtration showed material reacting with antimucin antibodies in a range of molecular sizes. Agarose electrophoresis confirmed the presence of MUC1 and MUC4 and the absence of MUC2 or MUC5AC. Isolation of membrane mucins by sequential, exhaustive extraction with octylglucoside followed by Triton X114 suggested the existence of mucins in different membrane environments. Reagents to carbohydrate epitopes revealed high mobility material, comigrating with MUC1 and MUC4. Low mobility membrane-bound mucins did not cross-react with any antibodies to mucin genes known to be expressed in human conjunctiva.

conclusions. Membrane-associated mucins are distinct from secreted mucins in normal human conjunctiva. MUC1 and MUC4 mature products decorate the membranes of conjunctival epithelial cells. Their segregation between octyl glucoside and the detergent and aqueous phases of Triton X114 suggests a variety of membrane anchoring modes.

Conjunctival and corneal mucins participate in the preocular gel resting on the apical cell membranes. ^{1 2} In common with all surface mucosal epithelia, cornea and conjunctiva express MUC1, a well-documented membrane associated mucin. Conjunctiva also expresses MUC5AC, ³ low levels of MUC2, ⁴ and MUC4 ³ a mucin with both membrane and secretory potential. Unique characteristics of normal human secreted ocular mucins are their wide size range and short oligosaccharide side chains, in particular GalNAcα-O-Ser/Thr (Tn) and NeuAcα2-6GalNAcα-O-Ser/Thr (sialyl-Tn). ⁵  

Current knowledge and understanding of the nature, occurrence and function of membrane mucins is limited. Interactions between membrane mucins and the preocular mucus gel have been proposed. ^{1 3} Direct demonstration of the presence of mucins in cellular membranes is required to support these claims, which is the rationale for this study. 

We have studied the membrane-associated mucins in total cellular membranes from homogenates of normal human conjunctival tissue, and which float at the top of CsCl density gradients. Extraction with detergents that are known to solubilize membrane glycoconjugates has been used to prepare membrane associated material. Mucins have been isolated by centrifugation in CsCl density gradients. As both size and charge affect inter-molecular interactions the mucins were further fractionated by gel filtration to obtain a molecular size profile, and by ion-exchange chromatography to ascertain charge. Detection of purified material with anti-mucin antibodies has allowed the tissue specific characterization of membrane associated mucins, which provides a reference for disease-associated changes. 

Materials and Methods

A flow chart of procedures of isolation, purification and fractionation is shown in Figure 1 , to clarify the origin and history of each of the subpopulations analyzed. 

Samples

Each collection of tissue comprised eight large pieces of both bulbar and tarsal conjunctiva, and small fragments of limbal conjunctiva. The eyes, donated with permission for transplantation and research, were collected within 24 hours of death, and maintained refrigerated until dissection. The tissue was treated as previously described. ⁵ Membrane rafts, comprising the total tissue membranes, which floated at the top of cesium chloride density gradients, were separated, pooled and stored at −20°C, in 4 M guanidine hydrochloride (4 M GuHCl; Sigma, Poole, UK) until use. Two separate membrane pools were analyzed. One was extracted with 1-O-n-octyl-β-d-glucopyranoside (octylglucoside; Sigma) only and is referred to as the whole sample. The second was extracted with octylglucoside and subsequently subjected to phase separation with Triton X114 (purified for membrane research; Boehringer-Mannheim, Lewes, UK) to eliminate“ contaminating” secreted material. The phase separation yielded an aqueous phase and a detergent phase (Fig. 1) . 

Extraction

Total cellular membranes were extensively washed with 10 mM phosphate-buffered saline, pH 7.4, and then homogenized on ice with 60 mM octylglucoside in 10 mM HEPES, pH 7.4. The supernatant was centrifuged at 14,000g for 10 minutes. This procedure was repeated until no more material could be extracted. Depletion was checked by cross-reactivity with antibody anti-M1against MUC5AC peptide core (Table 1) and with wheat germ agglutinin (WGA; Vector, Burlingame, CA, Table 1 ). Octyl glucoside solubilizes all glycoconjugates with one transmembrane domain or with a membrane anchor. ⁶  

The second pool of membranes was extracted as above with octylglucoside and then further extracted with 1% Triton X114, on ice, in 0.1 M Tris-HCl, 0.15 M NaCl, pH 7.4, containing protease inhibitors. ⁵ The extract was warmed at 37°C for 10 minutes and then centrifuged at 12,000g for 20 minutes at room temperature to effect phase separation. Each phase was reextracted twice ⁶ to enrich it with its own soluble material. The collected samples were the detergent and aqueous phases. They were further processed identically, but separately (Fig. 1) . 

After purification by density gradient centrifugation, mucinlike material was fractionated by mass and charge. Agarose electrophoresis was used as a further analytical tool, to highlight the dispersity within each, otherwise homogenous, population. 

Purification by Density Gradient Centrifugation

This procedure separates mucins from other glycoconjugates with lower glycosylation, ⁷ and was performed as previously detailed. ^{1 5} The presence of mucins and glycosylated material in individual fractions was detected using dot blots on PVDF membranes (Immobilon; Millipore, Bedford, MA) using anti-MUC5AC antibodies and WGA. For further analyses gradients were divided in three pools: 1.2 to 1.3 g/ml (DG1), 1.3 to 1.4 g/ml (DG2), and 1.4 to 1.5 g/ml (DG3). Mucins in each pool were further fractionated by molecular size. 

Fractionation by Gel Filtration and Ion Exchange Chromatography

Gel filtration on Sepharose CL2B and MonoQ ion exchange were carried out as detailed before. ⁵ Mucins too large to be fractionated by Sepharose, V ₀ (M _r > 4–5 × 10⁶, fractions 9–16), mucins within the resolution of the column, V _i (fractions 17–25), and those too small to be fractionated, V _t (fractions 26–30) were pooled. Ion exchange chromatography on MonoQ was used to determine the charge distribution of V ₀, V _i , and V _t within DG1, DG2, and DG3, respectively, in the aqueous and detergent phases (Fig. 1) . The presence of mucins was probed in 5-μl dot blots on PVDF membranes. Mucins in each size range were subjected to electrophoresis to assess gene product composition. 

Electrophoresis

Mucins are known to be separated optimally by agarose electrophoresis. However, because small molecules may elude detection, polyacrylamide gels were used to detect small mucins or link moieties. 

Agarose electrophoresis on 1% agarose gels and vacuum blotting were carried out as described before. ⁵ Each gel contained V ₀ and V _i fractions of purified secreted mucins (1.3–1.4 g/ml buoyant density) and a sample of crude conjunctival extract as references for antibody cross reactions. 

Vacuum blots of agarose gels were probed with the following reagents: wheat germ agglutinin and Molucella laevis lectins, antibodies BC2 to MUC1 and CT1 against the cytoplasmic tail of MUC1, antibodies M4.171 and M4.275 to MUC4, and antibodies anti-M1 and LUM 5–1 against MUC5AC. The specificity of antibodies and lectins used in this study is described in Table 1 . 

8.5% Acrylamide/Bis (BioRad, Hector, CA)-resolving gels in 0.4 M Tris-HCl, pH 8.8, with a stacking gel of 4.0% Acrylamide/Bis in 0.1 M Tris-HCl, pH 6.8, were run on a BioRad Mini-Protean II (BioRad). Sample buffer containing 0.1 M Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 5% mercaptoethanol, and 0.05% bromophenol blue was added and mixed without heating. Electrophoresis was performed in 25 mM Tris, pH 8.3, with 0.2 M glycine and 1% SDS, for 45 minutes at 200 V. Transfer to PVDF membranes was achieved with a Trans-Blot SD semi-dry electrophoretic transfer cell (BioRad) in buffer containing 44.5 mM Tris borate and 1 mM EDTA at a constant 150 mA for 1 hour. Kaleidoscope high-molecular-weight markers (Bio-Rad) were used for calibration. Blots were probed with WGA to detect glycosylated species, and gels were stained with Amido black to assess transfer quality. 

Dot Blots

Dot blots were made onto PVDF membrane using a 96-well format vacuum dot blotter (BioRad), as previously detailed. ⁵ Densitometry of stained blots was performed using a flatbed scanner (Scantouch 2000; Nikon Corporation, Tokyo, Japan) and ScanAnalysis 2.50 (Biosoft, Cambridge, UK). Reactivity profiles are shown in arbitrary gray units, which are a measure of the intensity of cross reaction. 

Results

Conjunctival Cell Membranes

The total cellular membranes collected from conjunctival tissue homogenates were harvested from the top of the cesium chloride density gradients. The remainder of the secreted mucins and other cellular material fractionate throughout these gradients and were removed from the membrane fraction. 

The Whole Sample

Using the anti-MUC5AC antibody anti-M1, mucins were detected between 1.2 and 1.5 g/ml, which is a typical buoyant density range for mucins (not shown). Sepharose CL 2B chromatography showed mucinlike material especially in the size range fractionated by the column (Fig. 2A ). Anti-MUC5AC mirrored WGA cross-reactivity. The antibody TKH2-sialyl-Tn showed a wider cross reaction, suggesting the presence of sialyl-Tn–containing, non-MUC5AC mucins. 

In electrophoretic profiles, WGA (Fig. 2B) and M. laevis lectin (not shown) detected low- and high-mobility bands in all subpopulations. The former had similar mobility to secreted mucins, and the latter migrated beyond secreted mucins. The antibody BC2 to MUC1-VNTR, cross-reacted strongly with the fast band, weakly with the slow band of membrane-associated (Fig. 2C) , and not with secreted mucins. MUC1 cross reaction was observed in both V ₀, which was WGA-negative, and in the WGA-positive V _i of DG1. The MUC4 gene product was detected with a mixture of two antibodies, M4.171 and M4.275, which cross-reacted only with the fast band of membrane-derived mucins (Fig. 2D) . The anti-MUC1– and anti-MUC4–positive bands had different mobilities (ScanAnalysis 2.50). Cross reaction with LUM 5–1 against MUC5AC gene product was weak in membrane-derived but strong in secreted mucins. 

Aqueous and Detergent Phases

Octyl-glucoside–depleted membranes were further extracted with Triton X114. After purification on density gradient centrifugation, mucinlike material was found between 1.25 and 1.45 g/ml in each phase. After gel filtration, the profile of cross reaction with WGA in each purified phase was similar to the profile shown in Figure 2A . MonoQ ion exchange chromatography (Amersham Pharmacia Biotech, Little Chalfont, UK), indicated a preponderance of relatively low charge (Fig. 3A ), lower in the detergent than in the aqueous phase. The most extended charge distribution was observed in the included fraction of the detergent phase. 

For detection on western blot analysis, samples were pooled as indicated in Figure 3B . Detergent phase mucins could not be detected on agarose. On polyacrylamide blots, however, most of the WGA-positive material migrated, with an apparent M _r above 220 kDa, a size range associated with mucins. ⁸ The mobilities of bands in V ₀ aqueous phase were not different from those observed in the whole sample, though the more mobile band seemed under-represented. In contrast, V _i fractions (Fig. 3B) uniformly lacked the least mobile band. 

Discussion

We have shown that extraction of total membranes from human conjunctival tissue yields material whose behavior on density gradient centrifugation is typical of mucins. Gel filtration indicated that extracted material cross-reacts with antimucin antibodies and subtends a range of molecular sizes. Agarose electrophoresis confirmed the presence of MUC1 and MUC4 within the largest membrane mucins and the absence of MUC2 and MUC5AC from this subpopulation. Differences in mobility between secreted and membrane mucins are indicative of the presence of different species of mucins in each population. Secreted and membrane-associated mucins share, however, Tn and sialyl-Tn carbohydrate epitopes, as confirmed by cross-reactivity with MLL and TKH2. 

The mucins isolated from total conjunctival cell membrane rafts are derived from material that will have contained cell surface and organelle membranes and physiologically or artificially lipid-complexed mucins. Mucins were extracted from membranes using two detergents selected for their ability to extract proteins with membrane anchors. Triton TX114 indeed extracted further material from the octyl glucoside-depleted membranes, highlighting the presence of molecules with different extraction properties with respect to the two detergents. This procedure enables the detection of mucins in different environments, improving our ability to follow their synthesis and turnover. Because the fractionation of membrane extracts into aqueous and detergent phase does not guarantee a single origin, we compared these isolates with secreted mucins. Any mucin present in both isolates can be presumed nonmembranal. 

There is a clear difference in the behavior of the membrane-derived and secreted mucins. Secreted mucins are MUC1- and MUC4-negative and strongly cross-reactive with MUC5AC antibodies, whereas membrane-associated mucins are strongly MUC1- and MUC4-positive and show only a weak reaction with MUC5AC. We can conclude, therefore, that extraction with octylglucoside yielded a population of mucins different from secreted mucins that had been extracted in guanidine hydrochloride. ⁵ Both anti-MUC1 and anti-MUC4 antibodies cross-react with high- and low-glycosylated mucins, suggesting a heterogenous population of gene products, some of which may be precursors or retrafficked mucins. Our results suggest that in the conjunctiva MUC4 is membrane-associated, as shown recently by Price–Schiavi et al. ⁹  

The presence of MUC5AC in membrane preparations needs explanation. The mucin may have been associated with endoplasmic reticulum membranes before export to secretory granules, but not attached to apical cell membranes. Another source of MUC5AC may be lipid complexation during extraction or as part of mucin turnover. 

We have shown that MUC1 and MUC4 are among the membrane-associated mucins in the conjunctiva. We have shown that cross-reactivity with these antibodies is more restricted than the range of mucinlike glycosylated material on agarose blots. No other MUC message has been demonstrated in either cornea or conjunctiva; therefore there must be products of yet unknown genes that decorate the apical membranes of human conjunctival epithelium. 

 Supported by The Guide Dogs for the Blind Association and The Wellcome Trust.

 Submitted for publication June 8, 1998; revised December 21, 1998, June 1, 1999, and July 23, 1999; accepted August 17, 1999.

 Commercial relationships policy: N.

 Presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May, 1997.

 

Figure 1.

 

Purification scheme. The scheme shows the treatment of the two populations of total cellular membranes described in the text.

Table 1.

 

Specificity of Anti-Mucin Reagents

Reagent	Specificity	Reference
WGA	GlcNAc, sialic acids	10
Molucella laevis	Tn, sialylTn	11,12
anti-M1	MUC5AC peptide	13
BC2	Synthetic peptide of MUC1 VNTR	14
CT1	Cytoplasmic tail of MUC1	15,16
M4.171 and M4.275	Synthetic peptides of MUC4 VNTR	17
LUM 5-1	Synthetic peptide between VNTR and C-terminal of MUC5AC:RNQDQQGFPKMC	18

 VNTR, variable number tandem repeats. This region is highly glycosylated in mature mucins.

Figure 2.

 

Analysis of the whole sample. (A) Gel filtration on Sepharose CL2B. Cross reaction with WGA, indicating the presence of sialic acids and N-acetylglucosamine (GlcNAc). Traces represent DG1, 1.2 to 1.3 g/ml (•); DG2, 1.3 to 1.4 g/ml (□); and DG3, 1.4 to 1.5 g/ml (▵). (B through D) Vacuum blots of agarose electrophoresis of whole sample mucins after gel filtration. Two purified secreted mucin fractions and a sample of crude tissue homogenate have been included in each gel for cross reaction and mobility comparisons. (B) Cross reaction with WGA. DG1-membrane–derived material is WGA-negative. High migration bands(≤220 kDa molecular weight marker) appear in membrane-derived material, but not in secreted mucin lanes. (C) Cross reaction with antibody BC2 against MUC1 gene product. This antibody cross-reacted strongly both with the low or nonglycosylated (WGA-negative) material in DG1 V ₀ and with the glycosylated (WGA-positive) high-mobility bands in the less buoyant membrane mucins. No cross reaction was observed in the secreted mucin lanes. (D) Cross reaction with antibodies against MUC4 gene product. The cross reaction with antibodies M4.171 and M4.275 against synthetic MUC4 VNTR peptides is confined to the high mobility band in membrane lanes, in both low and high glycosylation material. The mobilities of the MUC4-positive bands are different from the MUC1-positive bands (Fig. 2 C). (V ₀: fractions 9–16; V _i : fractions 17–25; V _t : fractions 26–35).

Figure 3.

 

Analysis of aqueous and detergent phases. (A) Charge fractionation by ion exchange chromatography on MonoQ. Charge distributions of aqueous (upper panel) and detergent phase (lower panel) mucins. The traces represent: V ₀ (○); V _i (•); V _t (♦). (V ₀: fractions 9–16; V _i : fractions 17–25; V _t : fractions 26–35). Cross reactivity with WGA was followed in dot blots, and expressed in arbitrary gray units. To compare peak positions, individual column runs, as well as the two panels, are aligned by conductivity, which was calculated from the measured resistance of the eluate. Departures from strictly monotonous increases in conductivity are seen as loops in the cross-reactivity traces. (B) Electrophoretic profiles of charge-fractionated mucins. WGA cross reaction of monoQ profile of aqueous phase V _i and its electrophoretic behavior on agarose (upper blot) and polyacrylamide (lower blot). Electrophoretic lanes are aligned with the parts of the charge profile they represent. The arrowheads mark the stacking gel, the dashes the positions of 220, 66, and 14 kDa molecular weight markers.

The authors thank Jacques Bara (Hospital St-Antoine, Paris, France), Joy Burchell (ICRF, London, UK), Ingemar Carlstedt (University of Lund, Sweden), Michael McGuckin (University of Queensland, Australia) and Nathan Sharon (Weizman Institute, Israel) for the gifts of antibodies and lectins. This work would not have been possible without the help of the Bristol Eye Bank.