Purification and Characterization of Rabbit Ocular

Scheffer C.G. Tseng,*t Andrew J.W. Huang,*t and David Sutler*

Ocular mucin, the major product of conjunctival goblet cells, constitutes the innermost layer of preocular tear film. Ocular mucin is known for its limited amount and inaccessability. Using impres- sion cytology, mucus strands collected from the inferior fornix of either rabbit or eyes were found to contain inflammatory cellular debris. In order to circumvent these difficulties and to isolate native mucin molecule(s), we bathed rabbit eyes in fluid containing isotonic PBS and 5.5 X 10~4 M acetylcholine for 4 or 12 hr. Bathing fluids containing rabbit ocular mucin (ROM), 1 ml per eye, were pooled and combined with 1M guanidine HC1 and protease inhibitors containing EDTA, PMSF, and sodium azide to avoid any possible enzymatic degradation, and then separated under the same condi- tions by Sepharose CL-4B. In parallel, commercial porcine stomach mucin (PSM) was purified and used to compare with ROM. We also developed nitrocellulose-based dot semi-quantitative assays for nucleic acid, , and . PAS-positive fractions monitored by such a dot assay were collected at CL-4B void volume and then separated from nucleic acid contaminants by CsCl-gradient ultracentrifugation. A protein fraction, 65K, poorly-glycosylated, with high contents of Asx, Glx, and Gly was found strongly associated with both ROM and PSM, and was only separable by ultracentri- fugation in 4M guanidine HC1 and CsCl. Purification of the ROM was verified by SDS-polyacryl- amide gel electrophoresis, amino acid analysis, and carbohydrate analysis. These results will allow future exploration of the molecular mechanism by which tear film stability is achieved. Invest Oph- thalmol Vis Sci 28:1473-1482,1987

Mucin, a macromolecular glycoprotein, is the pri- these materials contain polymorphonuclear and mary product of a special type of epithelial cell. mononuclear inflammatory cells as well as desqua- Throughout the body, mucin is the major constituent mating epithelial cellular debris. It is conceivable that of the mucous layer overlying all mucosal surfaces, mucin molecules could have been degraded in these and serves as a lubricative and protective barrier. In materials,8 and therefore may not represent their the eye, mucin is the product of conjunctival goblet functional counterparts. For example, more than one cells and constitutes the innermost layer of the components with molecular weights smaller than 1 preocular tear film.1 It plays an important role in X 106 daltons,4"6 some mucin components with a low maintaining the wettability and spreadability of the carbohydrate/protein ratio,5 and a carbohydrate con- 2 3 tear film ' and ocular surface integrity. In contrast to tent without N-acetyl neuramic acid4 were reported the studies of of other organs in the body, in these earlier reports. In contrast, Kaura and Tif- detailed biochemical analysis has been limited and fany recently analyzed freshly-scraped rabbit ocular incomplete in ocular mucin due to its limited mucus, and preliminarily reported a single compo- amount and relative inaccessability. Several attempts nent, high molecular weight, carbohydrate-rich have been made in the biochemical purification and/ 7 45 67 mucin. In order to explore the molecular interac- or characterization of human and rabbit ocular tions of mucin and ocular surface epithelium as well mucin. The majority of their starting materials were as the controlling mechanism of mucin biosynthesis mucus strands harvested from inferior fornix or the 4 6 and secretion, we sought as the first step to purify and inner caruncle. " As will be shown in this report, characterize rabbit ocular mucin. Precautions were taken with specific modifications of the purification From the *Eye Research Institute of Retina Foundation, Massa- procedures so that the native, undegraded mucin chusetts Eye and Ear Infirmary, and Department of Ophthalmol- macromolecules could be isolated and analyzed. ogy, Harvard Medical School, Boston, Massachusetts, and the fDe- partment of Ophthalmology, Bascom Palmer Eye Institute, Miami, Materials and Methods Florida. Chemical reagents such as cesium chloride (CsCl), Presented in part at the April-May 1986 ARVO meeting, Sara- ethylenediammine tetraacetic acid (EDTA) disodium sota, Florida. salt, guanidine HC1, phenylmethylsufonyl fluoride Supported by NIH grant EY-05656 (SCGT). Submitted for publication: September 23, 1986. (PMSF), sodium azide, and Tris/HCl were obtained Reprint requests: Scheffer C.G. Tseng, MD, PhD, P.O. Box from Sigma Chemical Company (St. Louis, MO). Bi- 016880, Bascom Palmer Eye Institute, Miami, FL 33101. ological standards for nitrocellulose dot assays in-

1473

Downloaded from iovs.arvojournals.org on 09/28/2021 1474 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / September 1987 Vol. 28

eluded calf thymus DNA, bovine serum albumin, degradation. The samples were then stored at -20°C and porcine stomach mucin, as well as rabbit serum before use. Our study conformed to the ARVO Reso- albumin, and were also from Sigma. Bio-dot appa- lution on the Use of Animals in Research. ratus and nitrocellulose acetate paper were products of Bio-Rad Laboratories (Richmond, VA). Electro- Nitrocellulose Dot Assays phoretic reagents such as sodium dodecyl sulfate In order to monitor the subsequent biochemical (SDS), acrylamide, ammonium persulfate, TEMED, fractionations of mucin samples and to minimize the and dithiothreitol (DTT) were also from Bio-Rad. loss of mucin materials through various purification Sepharose CL-4B was from Pharmacia (Uppsala, steps, nitrocellulose dot assays were developed. These Sweden). Acetylcholine chloride (Miochol®) was assays allowed semi-quantitative measurements of from Cooper Vision (Claremont, CA). Xylazine was nucleic acid, protein, and glycoprotein using respec- obtained from Haver-Lockhart (Shawnee, KS), and tive, specific histochemical staining to each compo- ketamine HC1 from Parke-Davis (Morris Plains, NJ). nent. A Bio-Rad dot assay apparatus was used, in which a nitrocellulose acetate filter paper was sand- Impression Cytology wiched between an upper unit containing 96 wells To understand the contents of starting materials, and the lower collecting chamber, to which a powered we decided first to analyze the mucus strands col- vacuum was connected. The filter in each well was lected from inferior fornix of both human and first wet by vacuum suction of 200 /d/well of PBS to rabbit eyes using impression cytology according to remove any trapped bubbles. To each well was added the method we have previously described.9 The re- 50 fi\ of sample containing either serially-diluted sults are shown in Figure 1. Because of potential en- standards of nucleic acid (calf thymus DNA), protein zymatic degradation of mucin molecule(s) using (bovine serum albumin), and glycoprotein (PSM), or such starting materials,8 we decided to take the fol- unknowns obtained from biochemical fractionations. lowing different approaches. We also chose to The test materials were then trapped to the surface of analyze a commercial porcine stomach mucin (PSM) the filter in each well by vacuum suction. The appa- in parallel for a comparison with rabbit ocular ratus was untightened, and the samples were fixed by mucin (ROM). drying the filter paper in a pre-heated oven at 80°C for 5 min. The dried paper was then immersed in either 1% Gill's hematoxylin,9 or 1% Fast Green in Rabbit Ocular Bathing methanol/acetic acid/H2O (20:7.5:72.5), or periodic In order to collect native ocular mucin, rabbit eyes acid-Schiff's reagent (PAS)9 for 1 min for nucleic were bathed with isotonic phosphate buffered saline acid, protein, and glycoprotein assay respectively. (PBS), pH 7.4. In brief, the rabbit was anesthetized The paper was destained with tap water for both with an intramuscular injection of mixtures of 20 mg Gill's hematoxylin and PAS, and with methanol/ace- xylazine and 200 mg ketamine and subsequently tic acid/H2O (20:7.5:72.5) for Fast Green, and was maintained by hourly injections of 100 mg ketamine dried again. The developed coloration was bluish- for a period of 4 hr. On selected occasions, bathing purple for nucleic acid, green for protein, and red for was extended to 10 to 12 hr. One eye at a time was mucin, all of which could be well-preserved on this used. Following initial anesthesia, the rabbit was put dried filter paper for more than a year. The sensitivity on one side and a total of four 4-0 black silk sutures, and specificity of such an assay is illustrated in two on each lid, were placed subcutaneously along Figure 2. both the upper and lower lid margins and one 7-0 proline suture was used to elevate the nictitating membrane. All suture threads were then suspended Gel-Filtration Column Chromatography and secured to four separate stands. By doing so, the Bathing fluid, 1 ml per eye, was then centrifuged at ocular surface formed a basin into which was then 1000 rpm, 4°C for 10 min to remove insoluble 4 added with 1 ml of PBS containing 5.5 X 10" M debris. The supernatant was added to a Sepharose acetylcholine, which as shown will stimulate the se- CL-4B column (1.5 X 50 cm) which was pre-equili- 10 cretion of native mucin by intestinal and conjuncti- brated with 50 mM Tris/HCl, pH 7.4, containing 1 M val" goblet cells. At the end of bathing, the fluid guanidine HC1 with the same concentrations of pro- containing discharged mucin was then collected and tease inhibitors. The column was run at 4°C with a added to a final concentration of 1 M guanidine HC1, flow rate of 10 ml/hr. Fractions of each 1 ml were and 50 mM Tris/HCl, pH 7.4, containing a mixture collected and monitored by a Gilford (Oberlin, OH) of protease inhibitors, 5 mM EDTA, 1 mM PMSF, spectrophotometer at both 260 nm and 280 nm. Ali- and 0.02% (W/V) sodium azide to prevent enzymatic quots of 50 n\ of each fraction were then subjected to

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 9 RABBIT OCULAR MUCIN / Tseng er ol. 1475

Fig. I. Impression cytol- ogy appearance of mucus strands obtained from human (left) and rabbit (right) inferior fornix (origi- nal magnification XI60). Note the inflammatory cells, desquamating epithe- lial cells, and collapsed mucin.

nitrocellulose dot assay for nucleic acid, protein, and Amino Acid Analysis glycoprotein as described above. The lyophilized samples collected during purifica- tion steps of both ROM and PSM were analyzed for CsCl-Density Gradient Ultracentrifugation their amino acid compositions. The samples were The PAS-positive fractions collected at the void first hydrolyzed in 6 N HC1 under nitrogen at 11O°C volume of Sepharose CL-4B column were then for 24 hr. After removal of acid, the sample was ana- pooled and subjected to isocratic CsCl-density gra- lyzed by a Durrum (Sunnyvale, CA) D-500 auto- dient ultracentrifugation according to the procedure mated amino acid analyzer. The analysis was con- of Creeth and Horton.12 Void-volume materials were ducted under a contract by William E. Brown, PhD, pooled from two to three Sepharose CL-4B runs to a at Carnegie-Mellon University, Pittsburgh, Pennsyl- total of 8 to 9 ml, to which guanidine HC1 was added to a final concentration of 4 M and CsCl at a loading density of 1.39 gm/ml, and then centrifuged at Carbohydrate Analysis 40,000 rpm, 4°C, for 64 hr using a Beckman SW 41 The lyophilized samples were also subjected to car- Ti rotor. After ultracentrifugation, tubes were punc- bohydrate analysis using gas-liquid chromatography tured and a total of 16 fractions, each 0.8-0.9 ml, according to the method of Reinhold'5 except that were collected, of which aliquots of 50 jul were re- the methanolysis conditions followed those described moved for dot assays as described above. The specific by Chambers and Clamp.16 The analyses were con- gravity of each fraction was also monitored by weigh- ducted by the Laboratory for Carbohydrate Research, ing 200 jtl of fluid in a microfuge tube. Massachusetts General Hospital, Boston, Massachu- setts, through kind support provided by Roger W. SDS-Polyacrylamide Gel Electrophoresis Jeanloz, PhD. The pooled fractions from either Sepharose CL-4B or CsCl isocratic ultracentrifugation were then dia- Results lyzed and lyophilized. These samples were then sub- Using impression cytology, we studied the mucus jected to SDS-polyacrylamide gel electrophoresis strands collected from inferior fornix of both human with 3% stacking gel and 5% or 10% separating gel and rabbit eyes. As shown in Figure 1, the mucus thus according to the method of Laemmli.13 Samples, obtained actually contained inflammatory cellular each of 10 ng, were reduced by adding 50 mM dithio- debris. Both polymorphonuclear and mononuclear threitol (DTT) and heating at 100°C for 3 min prior cells could be identified in addition to desquamated to electrophoresis. After that, gels were stained for epithelial cells. The mucus materials appeared to be protein with Coomassie Brilliant Blue and for carbo- collapsed and tangled. Because of this finding, it be- hydrate by PAS method.14 came imperative to purify ocular mucin using a dif-

Downloaded from iovs.arvojournals.org on 09/28/2021 1476 INVESTIGATIVE OPHTHALMOLOGY & VI5UAL SCIENCE / September 1987 Vol. 28

the nucleic acid assay by hematoxylin stain was noted to be 0.1-0.2 Mg in 50 pi, 0.2-0.3 Mg in 50 pi for protein, and 1-2 Mg in 50 pi for mucin glycoprotein (Figure 2). oo e w m Immediately after bathing, mucin of the collected N m ^ N IA o T- n o N n ug] ocular surface fluid was solubilized by adding it to a o' o' o o' *- ^ mo final concentration of 1 M guanidine HC1 in 50 mM Tris/HCl, pH 7.4, containing mixtures of protease ONA inhibitors such as sodium azide, EDTA and PMSF. Hematoxylin BSA Since native mucin is one of the largest macromole- cules, with a molecular weight exceeding 1 X 106 dal- PSM tons, one can crudely separate mucin from most other biological contaminants using gel-filtration Fast DNA Sepharose CL-4B column chromatography. A small BSA peak detected at 280 nm was eluted at the void vol- Green ume and separated from most other smaller-molecu- PSM lar-weight components (Figs. 3, 4). The void-volume fractions of ROM were noted to be PAS-positive (Fig. ONA 4, lower). In parallel, commercial PSM was also ana- PAS BSA lyzed by the same column and showed PAS-pbsitive PSM fractions at the void volume similar to those of ROM, and degraded mucin fragments in the included vol- ume (Fig. 4, upper). Since Sepharose CL-4B void-volume fractions also had a high 260 nm absorbance (Fig. 3), indicating possible nucleic acid contaminant, the pooled frac- Fig. 2. Nitrocellulose dot assays for nucleic acid by Gill's modi- tions were then subjected to CsCl-density gradient fied hematoxylin (upper panel), for protein by Fast Green (middle ultracentrifugation for further purification. As shown panel), and for glycoprotein by PAS (lower panels). Consult in Figure 5A, ROM collected after 4 hr bathing had Methods for detailed staining procedures. Calf thymus DNA, bo- two major peaks appearing at high and low specific vine serum albumin (BSA), and porcine stomach mucin (PSM) were used as standards. Note PSM contained minor DNA contami- gravity and one minor in between, according to 260 nant. nm and 280 nm absorbances. The peak at a high CsCl gravity had a high 260 nm/280 nm absorbance ratio and was identified by dot assay as a nucleic acid con- ferent approach with specific precautions to prevent taminant (Fig. 6). The minor peak (labeled as I), any enzymatic degradation if we were to study the noted as a shoulder prior to the last major peak (la- native functional molecule. Using a bathing tech- beled as II), was identified by the dot assay as a glyco- nique, mucin was found to be collectable from the protein^) since it had positive staining with both PAS bathing fluid, ie isotonic PBS, in the presence of 5.5 and Fast Green (Fig. 5A, 6). The last major peak (II) X 10~4 M acetylcholine. The rationale for using ace- with a high 280 nm/260 nm absorbance ratio was tylcholine can be found in previous works published identified as a pure protein fraction by the dot assay by others.10" In a total of 1 ml fluid collected from using Fast Green stain (Fig. 5 A, 6). When 12 hr bath- one eye after 4 hr bathing, the amount of mucin was ing fluid of ROM was analyzed, a similar pattern was estimated by nitrocellulose dot assay to be in the noted (Fig. 5B), except that the protein peak II in- range of 0.5-1.0 mg. Because of such a small quan- creased several-fold (cf Fig. 5A, B). The PSM was tity, we developed nitrocellulose dot assays for semi- then analyzed and compared. It demonstrated two quantitative measurements of nucleic acid, protein, major peaks at mid (labeled as I) and low (labeled as and glycoprotein. These assays allowed us to monitor II) specific gravity (Fig. 5C). There was no detectable the biochemical fractionations using a very small ali- nucleic acid contaminant either by 260 nm/280 nm quot without losing mucin materials. As shown in absorbance ratio or by the dot assay using Hematox- Figure 2, nucleic acid, protein, and glycoprotein ylin (Fig. 6). Peak I was identified as PAS-positive standards could be differentiated by Gill's hematoxy- glycoprotein(s) (Fig. 6), and peak II with a higher 280 lin, Fast Green, and PAS stains, respectively, of nm/260 nm absorbance ratio was identified as a pure which each developed its own special color reaction. protein (Fig. 5C, 6). These results indicated that ul- Serial dilutions of these standards demonstrated a tracentrifugation could effectively separate mucin gradual decrease of color intensity. The sensitivity of glycoprotein from nucleic acid contaminant. Fur-

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 9 RABBIT OCULAR MUCIN / Tseng er ol. 14??

Sepharose CL-4B

Fig. 3. Sepharose CL-4B chromatography of rabbit ocular bathing fluid.Not e a small peak was noted at void volume (indicated by a bar), which was separated from most other compo- nents. Consult Methods for detailed procedure. (Vo = void volume, Vt = total volume).

10 20 30 40 50 60 70 80 90 100 Fraction Number

thermore, a pure protein fraction was also separated from mucin glycoprotein, suggesting the existence of PORCINE STOMACH MUCIN a strong non-covalent binding between these two components. The binding appeared to be strong enough to withstand prior Sepharose CL-4B separa- tion even in the presence 1 M guanidine HC1. In order to evaluate the purity of the mucin com- ponent during purification steps, we analyzed their dialyzed and lyophilized samples by 5% or 10% SDS- polyacrylamide gel electrophoresis. As shown in Fig- ure 7, under reduction by 50 mM DTT, crude PSM 85 » showed a protein doublet with molecular weights of •Id* 64K and 65K daltons, respectively (lane 1). They be- came more intensified in the void-volume fractions RABBIT OCULAR BATHING obtained from Sepharose CL-4B, as compared to their included-volume fractions which consisted of presumed partially-degraded mucin (cf lanes 2 and 3, see also Fig. 4, upper). It is interesting to point out that after subsequent ultracentrifugation, the 64K protein, the lower band of this doublet, was primarily noted after reduction in the PSM mucin Peak I (lane 4). This result indicates that this protein component might be associated with mucin core protein by disul- 85*- fide bond(s). Nevertheless, the 65K protein, the upper std- band of the doublet, was noted in the Peak II (lane 5). Both of them were with low carbohydrate contents, as evidenced by dot assay (Fig. 6), PAS staining of 10% gel (not shown), and subsequent amino acid and carbohydrate analysis (Table 1). Fig. 4. Nitrocellulose dot assay by PAS of CL-4B fractions for Since this 65K protein was separated from mucin porcine stomach mucin (upper) and rabbit ocular bathing fluid molecules only by ultracentrifugation in the presence (lower). Note PAS-positive fractions (#25-30) at void volume on of 4 M guanidine HC1 and CsCl, it suggests that this rabbit ocular bathing fluid.PS M displayed similar fractions at void volume which were continuous with smaller-molecular-weight protein had a strong and probably specific binding mucin fragments.

Downloaded from iovs.arvojournals.org on 09/28/2021 1478 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Seprember 1987 Vol. 28

Rabbit Ocular Mucin Porcine Stomach Mucin T2.5

0.3T

0.2 2.0

0.1

•1.0 -0.1 6 8 10 12 14* 16 0 2 4 6 8 10 12 14 16 Fraction Number Fraction Number Fig. 5C. CsCI-gradient ultracentrifugation of porcine stomach Rabbit Ocular Mucin mucin. Peaks I and II indicate mucin and its associated protein 12hr Bathing fraction respectively. 0.3

0.2

1.5

BSA stc

4 6 8 10 12 14 16 B Fraction Number -16 Fig. 5A, B. CsCI-gradient ultracentrifugation of rabbit ocular mucin after 4 hr (A) and 12 hr (B) bathing. Peaks I and II indicate DNAsh mucin and its associated protein fraction respectively. BSAstt

with mucin molecules. When ROM was analyzed, -16 Sepharose CL-4B indeed effectively removed the ma- jority of small-molecular-weight components (cf lanes 6, 7, and 8). After ultracentrifugation, a faint -16 protein band was noted after reduction of the mucin PSMstd ? fraction I which had a similar mobility to the 64K 1 protein of the PSM doublet (lane 9). Similarly, the BSA sld i protein fraction obtained from peak II at a low spe- cific gravity was found to be a single protein band i with mobility comparable to the 65K protein of PSM Fig. 6. Nitrocellulose dot assays of fractions in Figure 5(1 to 16) doublet (cf lanes 10 and 5). Due to the limited load- for nucleic acid (upper), pure protein (middle), and glycoprotein ing amount, both samples on lanes 7 and 9 showed (lower). Consult Methods for details.

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 9 RABBIT OCULAR, MUCIN / Tseng er al. 1479

PSM ROM

Pig. 7. SDS-polyacryl- amide gel electrophoresis (3% stacking/10% lower gels) of various fractions of porcine stomach mucin (PSM) and rabbit ocular mucin (ROM). For abbrevi- ations, consult Table 1 leg- end.

only faint bands. Throughout all these gel analyses, manner. Fraction 3 had the highest content of carbo- both PSM and ROM were found in the well of upper hydrate (18.6% N-acetyl glucosamine and 26.8% N- 3% stacking gel and not in the 5% lower gel by PAS acetyl galactosamine), and amino acids with high stain, due to their high molecular weights (not contents of serine (22.8%), glutamic acid/glutamine shown). Because of the absence of other small-molec- (16.9%), and glycine (23.8%). These results indicate ular-weight components in the 10% lower gel, their that this fraction might represent the most purified purity was verified (lanes 4 and 9). ROM. From fractions 3 to 1, one begins to see the For further analyses of their purity, the above ly- decreasing trend of the above amino acids, and the ophilized samples were then subjected to analyses of increasing trend of other amino acids such as aspartic amino acid composition and carbohydrate content. acid/asparagine and proline, which were of a high For PSM, amino acid analysis of the purified mucin content in the protein obtained from peak II (Table showed high contents of threonine (28.4%), serine 1). This notion was also supported by the results of (13.4%), and proline (20.7%), as well as high contents the contents of N-acetyl glucosamine and N-acetyl of N-acetyl glucosamine (42.7%) and N-acetyl galac- galactosamine. The results strongly indicate that a tosamine (30.5%) in a total of 1000 amino acids strong non-covalent binding existed between mucin (Table 1). This result was generated after the separa- and this 65K protein. The bound form would result tion of a tightly-associated protein fraction by ultra- in a higher molecular weight, a low carbohydrate centrifugation in the presence of 4 M guanidine HC1 content, and a higher CsCl specific gravity (Fraction and CsCl (Figs. 5, 6, 7). This 65K protein was found 1). In contrast, the unbound purified mucin would be to have a totally different amino acid composition of a lower molecular weight, a high carbohydrate with a relatively high content of aspartic acid/aspara- content, and a lower CsCl specific gravity (Fraction gine, glutamic acid/glutamine, and glycine (cf CL-4B 3). Some similarities were also noted between the Vo fraction, Peak I purified mucin, and Peak II pro- 65K protein of PSM and ROM when their amino tein fraction), and a low carbohydrate content of N- acids and carbohydrates were compared (Columns 3 acetyl glucosamine (2%) and N-acetyl galactosamine and 7, Table 1). It should also be noted that this (<0.6%). For ROM, three consecutive fractions (la- protein was not a serum albumin when amino acids beled as 1, 2, and 3, Fig. 5B) from the mucin fraction composition and carbohydrate content were com- I were analyzed. It is noteworthy that the amino acid pared with those of bovine serum albumin (BSA) pattern changed from fractions 1 to 3 in a gradual (Table 1). The electrophoretic mobility was also dif-

Downloaded from iovs.arvojournals.org on 09/28/2021 1480 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / September 1987 Vol. 28

Table 1. Amino acid composition of mucins (per 1000 A.A.)

PSM* ROM

4B + Vo 4B + Ucf 4B + Ucf 4B+ Ucf 4B + Ucf I II I II (1) (2) (3) BSA

Asx 57 26 88 84 66 61 100 103 Thr 162 289 65 52 47 44 77 60 Ser 170 134 95 127 243 228 86 53 Glx 82 49 119 132 161 169 119 140 Pro 113 207 51 43 21 21 64 49 Gly 111 53 143 132 201 238 78 31 Ala 77 83 79 99 88 87 80 86 Cys — — — — — — — — Val 43 51 50 60 28 25 71 68 Met 5 — — 6 2 — 6 8 He 21 10 25 17 17 14 33 24 Leu 41 34 88 72 20 17 88 114 Tyr 14 <1 <6 15 16 9 26 30 Phe 13 <1 29 33 15 11 43 48 His 50 35 37 39 46 52 33 31 Lys 22 10 81 63 25 24 69 112 Arg 20 20 45 25 6 <1 28 44 N-acetyl glucosamine 254 427 20 100 150 186 10 — N-acetyl galactosamine 187 305 <6 106 268 286 <5 —

* Abbreviations: PSM, porcine stomach mucin; ROM, rabbit ocular fraction collected after CL-4B and CsCI-gradient ultracentrifugation; 1,11,(1) mucin; 4B + Vo, void-volume fractions after Sepharose CL-4B; 4B + Ucf, (2) (3), consult Figure 5 A, B, C.

ferent from that of rabbit serum albumin on a 10% the native functional mucin molecule due to poten- SDS-polyacrylamide gel (not shown). tial proteolytic degradation. To circumvent the The detailed carbohydrate content was measured above problems, we made the following modifica- and is shown in Table 2. The mucin fraction purified tions in the biochemical purification of rabbit ocular by both Sepahrose CL-4B and ultracentrifugation ex- mucin (ROM). hibited high carbohydrate/protein ratio, (87.3% vs We first developed the nitrocellulose dot assay for 12.7%) for PSM and (84.2% vs 15.8%) for ROM. In semi-quantitative measurements of nucleic acid, pro- addition, both mucins had five major sugar compo- tein, and glycoprotein (mucin). This allowed us to nents: N-acetyl galactosamine, N-acetyl glucosamine, monitor fractionations obtained from Sepharose galactose, fucose, and N-acetyl neuramic acid CL-4B as well as CsCI-gradient ultracentrifugation. (NANA), and were devoid of mannose and xylose. Only a small quantity of sample was required so that there was a minimal loss of mucin during these purifi- Discussion cation steps. Herein, we demonstrated the feasibility and efficiency of these assays. Second, we collected The major difficulty in the biochemical studies of ROM by bathing rabbit ocular surface with isotonic ocular mucin is its limited amount and inaccessabi- PBS containing acetylcholine. As shown in this study, lity. In addition, we also demonstrated by impression mucin was recoverable from this bathing fluid. The cytology that mucus strands collected from inferior use of acetylcholine has been helpful in obtaining fornix contained inflammatory cells, epithelial cellu- sufficient amounts of native mucins from intestine10 lar debris, and collapsed mucus. We therefore sus- and conjunctiva." The bathing of isotonic fluid for 4 pected that such starting materials may not contain hr is not physiological and may have included plasma membrane in the crude extracts. Nev- ertheless, the purified mucin has a molecular weight Table 2. Carbohydrate composition 6 of mucins (Wt. %) exceeding 1 X 10 daltons and is devoid of any man- nose, a common monosaccharide present in mem- PSM ROM brane glycoproteins, indicating that the subsequent Carbohydrate content 87.3 84.2 purification steps have effectively eliminated all con- Protein content 12.7 15.8 taminants. Third, it was important to add protease N-acetyl galactosamine 22.7 20.4 inhibitors throughout the biochemical purification N-acetyl glucosamine 24.9 23.0 steps so that any possible enzymatic degradations Galactose 27.2 26.1 Fucose 11.0 12.7 could be minimized. Finally, we confirmed the effec- N-acetyl neuramic acid 1.5 2.0 tiveness of purifying mucin by the use of a combina-

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 9 RADDIT OCULAR MUCIN / Tseng er ol. 1481

tion of Sepharose CL-4B and CsCl-gradient ultracen- N-acetyl glucosamine, N-acetyl galactosamine, galac- trifugation. This point will be further discussed tose, fucose, and low but detectable N-acetyl neura- below. mic acid (NANA) or sialic acid (Table 2). NANA was Mucins have been purified and characterized from also noted in human ocular mucus strands by Chao et various other organ systems including cervix,17"19 al,5 and in rabbit ocular mucus by Kaura and Tiffany bronchus,20 saliva,21 stomach,2223 intestine,24 and in their recent report,7 but not in their earlier report colon.25 These studies indicate that mucins or mucus using human ocular mucus strands.4 These results glycoproteins are macromolecular complex structure indicate that carbohydrate content of mucin can be with molecular weights exceeding 1 X 106 daltons, degraded extracellularly, a mechanism by which except that of saliva,21 and consist of a core protein mucin may be removed from ocular surface under rich in serine, threonine, and proline residues, oligo- normal circumstances. Taken together, these results saccharide chains O-glycosyl-linked through N-acetyl also support the notions that mucin in mucus strands galactosamine residue to either serine or threonine, may not be in its native functional status, and that it and with the possible presence of a carbohydrate-free is important to include protease inhibitors through- portion that may be linked to the glycosylated pep- out purification. tide by covalent disulfide bonds. The amino acid compositions of purified PSM re- In this report, we noted that both ROM and com- vealed a pattern similar to those reported for other mercial porcine stomach mucin (PSM) had a molecu- mucins. There were high contents of threonine, ser- lar weight exceeding 1 X 106 daltons as judged by ine, and proline, three of them comprising 62% of the their void-volume elution from Sepharose CL-4B. total amino acid content (Table 1). A different pat- Using CsCl-gradient ultracentrifugation, ROM was tern was noted in the purified ROM which had high found to have a specific gravity of 1.36 gm/ml, which contents of serine, glutamine/glutamic acid, and gly- was of slightly smaller molecular weight than that of cine, three of them constituting 60-62% of the total PSM, which had a specific gravity of 1.39 gm/ml. The amino acid content (Table 1). Though different from mucin purified by Sepharose CL-4B and CsCl-gra- other mucins, ROM is similar to the human ocular dient ultracentrifugation had a high ratio of carbohy- mucin reported by Chao et al,5 who also showed a drate content vs protein content. For example, the high content of serine, glutamine/glutamic acid, and purified ROM and PSM had a carbohydrate content glycine, constituting 47%-50% of total aminic acid of 84.2% and 87.3% respectively (Table 2). These content. This result indicates that ocular mucin pri- values were comparable to the reported values of marily uses serine to form O-glycosyl linkage with mucins in other organ systems,17"25 which were in the oligosaccharide chains, in contrast to all other re- range of 66% to 81%, except for rat colon mucin.25 ported mucins which form such a linkage through These values were certainly higher than those of both threonine and serine, the former more fre- human ocular mucus reported by Chao et al,5 which quently than the latter.17"25 When amino acid com- had a fluctuating range of carbohydrate contents position was compared among PSM samples before from 57% depending on the sample preparation and and after ultracentrifugation, it also indicated that solubility. Other previous studies on ocular mucins further purification was achieved by dissociating the did not report data on carbohydrate content.4'6-7 The 65K protein (Table 1). Similarly, when sequential ul- discrepancy between our data and those of Chao et tracentrifugation fractions 1 to 3 of ROM are com- al5 may be due to the fact that a 65K protein was still pared, one can also find the gradual change in the tightly associated with mucin after Sepharose CL-4B. proportional amount from mucin-related amino This protein component could only be separated acids such as serine, glutamine/glutamic acid, and from mucin after ultracentrifugation, a procedure not glycine to the 65K protein-related ones such as used by Chao et al.5 This explanation was supported aspartic acid/asparagine and proline (Table 1). Using by our data that carbohydrate content, as viewed by SDS-polyacrylamide gel electrophoresis under a re- the content of N-acetyl glucosamine and N-acetyl ga- duced state, we also observed a faint band constitut- lactosamine per 1000 amino acids, increased from ing part of the doublet together with the 65K protein ROM ultracentrigation fraction 1 to fractions 2 and band in PSM (Figure 7), suggesting a disulfide-bond- 3, where the 65K protein decreased in its association rich domain may be present in PSM, as reported.22'23 with ROM (Table 1). This finding was also seen in Whether this finding is also true in ROM is not clear PSM in which an increase of N-acetyl glucosamine at this time. In summary, we have verified the iden- and N-acetyl galactosamine content per 1000 amino tity and purity of the isolated ROM and PSM. acids was noted when this 65K protein was removed It is worthwhile to further discuss this tightly-asso- by ultracentrifugation (Table 1). When carbohydrate ciated 65K protein. The tight association of this pro- composition was analyzed by gas-liquid chromatog- tein with mucin explains why the results of mucin raphy, both ROM and PSM had high contents of composition reported in the literature are variable,

Downloaded from iovs.arvojournals.org on 09/28/2021 1482 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / September 1987 Vol. 28

depending on whether this protein has been totally and after subsequent trypsin digestion. Biochem J 213:427, removed. We have also investigated the nature of this 1983. protein. The results of amino acid composition, car- 9. Tseng SCG: Staging of conjunctival squamous metaplasia by impression cytology. Ophthalmology 92:728, 1985. bohydrate contents, and electrophoretic mobility in- 10. Specian RD and Neutra MR: Mechanism of rapid mucus se- dicate that this 65K protein is not albumin. The fact cretion in goblet cells stimulated by acetylcholine. J Cell Biol that both PSM and ROM had such a similar protein, 85:626, 1980. ie, a molecular weight of 65K daltons, amino acids 11. Greiner CAM, Peace DG, Allansmith MR, and Greiner JV: rich in aspartic acid/asparagin, glutamic acid/gluta- The effects of acetylcholine on conjunctival goblet cells. ARVO Abstracts. Invest Ophthalmoi Vis Sci 24(Suppl):77, mine, and glycine, and a low carbohydrate content 1983. suggests this protein may be commonly associated 12. Creeth JM and Horton JR: Macromolecular distribution near with mucin molecules. The quantity of this protein the limits of density-gradient columns: Some applications to also increased when ocular bathing was extended the separation and fractionation of glycoproteins. Biochem J from 4 to 12 hr. For these reasons, we speculate that 161:449, 1977. 13. Laemmli UK: Cleavage of structural proteins during the as- this protein may be a putative "mucin binding pro- sembly of the head of bacteriophage T4. Nature (London) tein," a protein present on the apical membrane of 227:680, 1970. epithelial cells with an active role in maintaining 14. Fairbanks G, Steck TL, and Wallach DFH: Electrophoretic mucin on the mucosal surface. Currently, we are analysis of the major polypeptides of the human erythrocyte testing this hypothesis using a monoclonal antibody membrane. Biochemistry 10:2606, 1971. 15. Reinhold VN: Gas-liquid chromatographic analysis of constit- approach. uent carbohydrates in glycoproteins. Methods Enzymol In summary, we have taken the first step in purify- 25:244, 1972. ing and characterizing rabbit ocular mucin. With this 16. Chambers RE and Clamp JR: An assessment of methanolysis information, we hope that we can then explore the and other factors used in the analysis of carbohydrate-contain- ing materials. Biochem J 125:1009, 1971. molecular mechanisms by which the interactions be- 17. Yurewicz EC and Moghissi KS: Purification of human midcy- tween mucin molecule and apical membrane pro- cle cervical mucin and characterization of its oligosaccharides teins or glycoproteins are operated, and by which the with resect to size, composition, and microheterogeneity. J preocular tear film stability is ensured. Biol Chem 256:11895, 1981. 18. Carlstedt I, Lindgren H, Sheehan JK, Ulmsten U, and Win- Key words: conjunctiva, goblet cells, mucin, mucus, rabbit gerup L: Isolation and characterization of human-cervical eyes mucus glycoproteins. Biochem J 211:13, 1983. 19. Hatcher VB, Schwarzmann GOH, Jeanloz RW, and McArthur JW: Purification, properties, and partial structure References elucidation of a high-molecular-weight glycoprotein from cer- vical mucus of the Bonnet Monkey (Macaca radiata). Bio- 1. Nichols BA, Chiappino ML, and Dawson CR: Demonstration chemistry 16:1518, 1977. of the mucous layer of the tear film by electron microscopy. 20. Woodward H, Horsey B, Bhavanandan VP, and Davidson EA: Invest Ophthalmoi Vis Sci 26:464, 1985. Isolation, purification, and properties of respiratory mucus 2. Lemp MA, Holly FJ, Iwata S, and Dohlman CH: The precor- glycoproteins. Biochemistry 21:694, 1982. neal tear film: I. Factors in spreading and maintaining a con- 21. Prakobphol A, Levine MJ, Tabak LA, and Reddy MS: Purifi- tinuous tear film over the corneal surface. Arch Ophthalmoi cation of a low-molecular-weight mucin-type glycoprotein 83:89, 1970. from human submandibular-sublingual saliva. Carbohydrate 3. Holly FJ and Lemp MA: Wettability and wetting of corneal Res 108:111, 1982. epithelium. Exp Eye Res 11:239, 1971. 22. Starkey BJ, Snary D, and Allen A: Characterization of gastric 4. Moore JC and Tiffany JM: Human ocular mucus: Origins and isolated by equilibrium density-gradient centrif- preliminary characterisation. Exp Eye Res 29:291, 1979. ugation in caesium chloride. Biochem J 141:633, 1974. 5. Chao C-C W, Vergnes J-P, and Brown SI: Fractionation and 23. Spee-Brand R, Strous GJAM, and Kramer MF: Isolation and partial characterization of macromolecular components from partial characterization of rat gastric mucous glycoprotein. human ocular mucus. Exp Eye Res 36:139, 1983. Biochim Biophys Acta 621:104, 1980. 6. Iwata S and Kabasawa I: Fractionation and chemical proper- 24. Wesley AW, Forstner JF, and Forstner GG: Structure of intes- ties of tear mucoids. Exp Eye Res 12:360, 1971. tinal-mucus glycoprotein from human post-mortem or surgi- 7. Kaura R and Tiffany JM: Preliminary characterization of the cal tissue: Interferences from correlation analyses of sugar and mucous glycoprotein component of rabbit ocular mucus. Bio- sulfate composition of individual mucins. Carbohydrate Res chem Soc Trans UK 12:651, 1984. 115:151, 1983. 8. Charlstedt I, Lindgren H, and Sheehan JK: The macromolecu- 25. LaMont JT and Ventola AS: Purification and composition of lar structure of human cervical-mucus glycoproteins: Studies colonic epithelial mucin. Biochim Biophys Acta 626:234, on fragments obtained after reduction of disulphide bridges 1980.

Downloaded from iovs.arvojournals.org on 09/28/2021