Altered fine structures of corneal and skeletal keratan sulfate and chondroitin/dermatan sulfate in macular corneal dystrophy.

The content and fine structure of keratan and chondroitin/dermatan sulfate in normal human corneas and corneas affected by macular corneal dystrophies (MCD) types I and II were examined by fluorophore-assisted carbohydrate electrophoresis. Normal tissues (n = 11) contained 15 microg of keratan sulfate and 8 microg of chondroitin/dermatan sulfate per mg dry weight. Keratan sulfates consisted of approximately 4% unsulfated, 42% monosulfated, and 54% disulfated disaccharides with number of average chain lengths of approximately 14 disaccharides. Chondroitin/dermatan sulfates were significantly longer, approximately 40 disaccharides per chain, and consisted of approximately 64% unsulfated, 28% 4-sulfated, and 8% 6-sulfated disaccharides. The fine structural parameters were altered in all diseased tissues. Keratan sulfate chain size was reduced to 3-4 disaccharides; chain sulfation was absent in MCD type I corneas and cartilages, and sulfation of both GlcNAc and Gal was significantly reduced in MCD type II. Chondroitin/dermatan sulfate chain sizes were also decreased in all diseased corneas to approximately 15 disaccharides, and the contents of 4- and 6-sulfated disaccharides were proportionally increased. Tissue concentrations (nanomole of chains per mg dry weight) of all glycosaminoglycan types were affected in the disease types. Keratan sulfate chain concentrations were reduced by approximately 24 and approximately 75% in type I corneas and cartilages, respectively, and by approximately 50% in type II corneas. Conversely, chondroitin/dermatan sulfate chain concentrations were increased by 60-70% in types I and II corneas. Such changes imply a modified tissue content of individual proteoglycans and/or an altered efficiency of chain substitution on the core proteins. Together with the finding that hyaluronan, not normally present in healthy adult corneas, was also detected in both disease subtypes, the data support the conclusion that a wide range of keratocyte-specific proteoglycan and glycosaminoglycan remodeling processes are activated during degeneration of the stromal matrix in the macular corneal dystrophies.

The macular corneal dystrophies (MCD) 1 are a group of autosomal inherited diseases (1-3) with a severe effect on the structural integrity and function of the corneal stroma (4 -6). The pathological alterations are usually diagnosable during puberty, with the appearance of focal opacities in both corneas that gradually develop into more widespread areas of cloudiness that lead to loss of vision. The disease phenotype has been biochemically and genetically linked to defects in keratan sulfate metabolism by stromal fibroblasts (keratocytes) (7)(8)(9), and at least three biochemical subtypes (types I, IA, and II) of the disease have been identified. Such classifications are derived from the observations that immunoreactive sulfated keratan sulfate is completely absent from the stroma and the sera (type I), absent from the stroma and sera, but detectable inside the keratocyte (type IA) (10 -12) or present at much reduced levels in the stroma and the sera (type II) (13)(14)(15).
In a number of patients identified with MCDs, the defective gene localizes to chromosome 16 and was shown to encode the N-acetylglucosamine-6-sulfotransferase (C-GlcNAc6ST) (14, 16 -18), the enzyme that initiates sulfation of keratan sulfate chains on proteoglycans. So far, all mutations associated with subtype I have been localized to the coding region of the gene. Cell extracts prepared from MCD type I corneas are unable to sulfate GlcNAc residues on exogenously added oligosaccharide acceptors (19), unlike the normal corneal extracts, suggesting that these types of mutations result in the production of nonfunctional enzyme. Furthermore, comparisons of biosynthetic radiolabeling of normal and MCD I corneas in short term explant cultures demonstrated that only cells in the normal but not diseased tissue secreted keratan sulfate substituted proteoglycan core proteins (7,9,20). Mutations in the sulfotransferase gene in subtype II patients are deletions and/or replacements in the untranslated regions of the gene, where regulatory elements may be localized. These defects may therefore lead to a reduced transcription of the enzyme and possibly result in an enzyme deficiency limited to the corneas, thus explaining the published studies that reported normal levels of sulfated keratan sulfate in the serum of the subtype II patients.
Despite the extensive genetic and diagnostic studies of the MCDs, there is currently no persuasive mechanistic link between the insufficient sulfation of keratan sulfate of corneal proteoglycan core proteins and the progressive opacity in the adult cornea without an apparent effect on corneal clarity in utero and in early childhood in these diseases. First, there is no reported chemical analyses of keratan sulfate chain fine struc-ture, accumulated in vivo by cells harboring the MCD types I and II mutations in C-GlcNAc6ST activity, nor has there been any examination of the effect of impaired keratan sulfate syn-
FIG. 2. Quantitation of digestion products from normal and MCD corneal keratan sulfate. Products were quantitated after FACE separation, using Scion image analyses software (see "Materials and Methods"). Data are expressed as nanomoles of un-, mono-, and disulfated disaccharide per mg dry weight of corneal tissue. The values for the normal samples represent the mean Ϯ S.D. of duplicate analyses of 11 separate normal adult human corneas. The values for each of the diseased corneas (samples [12][13][14][15][16][17][18][19] represents the mean Ϯ S.D. of duplicate analyses of each tissue sample. The relative molar abundances of the products were further used to calculate the proportional sulfation of GlcNAc and Gal residues (see Table II). ϫ designates the unclassified diseased tissue. thesis on the metabolic fate of other glycosaminoglycans in the corneal stroma. Thus, changes in dermatan sulfate, chondroitin sulfate, and hyaluronan are also be associated with corneal clouding in post-injury or post-operative scarring, keratoconus (21,22), bullous keratopathy, and Fuch's dystrophy (23,24) but alterations in these glycosaminoglycans as potential players in the disorganization of the corneal stroma in MCDs has not been considered.
Following our recent development of fluorophore-assisted carbohydrate electrophoresis (FACE) technology for glycosaminoglycan quantitation and sulfation analyses (25)(26)(27)(28)(29), we now report a detailed study of the fine structural characteristics of both keratan sulfate and chondroitin/dermatan sulfate chains in cornea and cartilage from individuals affected by MCD type I and in corneas of individuals affected by MCD type II. The study illustrates the high conservation of both glycosaminoglycan contents and fine structures in normal corneas and revealed distinct changes in these parameters in the diseased tissues. These findings are discussed with respect to the likelihood that mutations in a single glycosaminoglycan sulfotransferase have profound and wide ranging effects on metabolic pathways for other tissue proteoglycans and glycosaminoglycans.

MATERIALS AND METHODS
Tissue Specimen-Portions of MCD corneas, removed at time of keratoplasty, were immediately frozen for biochemical procedures. Additional portions of corneas were fixed in formalin and embedded in paraffin and either stained with hematoxylin/eosin/Hale's colloidal iron or examined immunohistochemically with monoclonal antibody 1/20 5D4 (from ICN Biomedical Inc., Costa Mesa, CA). Serum concentrations of antigenic keratan sulfate for each individual were determined by enzyme-linked immunosorbent assay (11) using the same antibody, in a masked fashion, without knowledge about the individual subjects prior to the assay. After obtaining an informed consent, biopsies of the external ear cartilage were obtained from random subjects with histopathologically confirmed MCD type I. Normal ear and nasal cartilages were obtained from the Tissue and Organ Donor Bank of Illinois. The details of the disease typing are summarized in Table I.
Glycosaminoglycan Preparation, Enzymatic Depolymerization, and FACE Analyses-Portions of the frozen tissues (0.5-1.0 mg dry weight) were digested at 60°C for 24 h with 50 g of proteinase K (cornea) or papain (cartilages) in 300 l of ammonium acetate buffer, pH 7.0.
Proteinases were inactivated at 100°C for 5 min, and undigested tissue debris was removed by centrifugation through 0.45-m Ultrafree Microfiltration units (10,000 ϫ g, 10 min at room temperature). Portions of the clarified protease digests (containing products from 0.5 mg dry weight of tissue) were centrifuged through Microcon 3 ultrafiltration devices to remove buffer salts, amino acids, and small peptides. The retentates were recovered in 75 l of 0.1 M ammonium acetate, pH 6.0, and digested with either 2.5 milliunits of keratanase II or 2.5 milliunits of endo-␤-galactosidase (Escherichia freundii) for 22 h at 37°C. The enzymes were inactivated at 100°C for 5 min, the digests spun through Microcon 3 filters, and the keratan sulfate hydrolase products recovered in the filtrate. The retentates were washed once with 70 l of water and then recovered in 75 l of 0.1 M ammonium acetate, pH 7.3, and digested for 22 h at 37°C with 5 milliunits of each chondroitinase ABC and ACII. All digests were dried by speedvac evaporation before product fluorotagging and FACE separation (25)(26)(27)(28)(29).
To assess the quantitative recovery of keratan sulfate-, polylactosamine-, and chondroitin/dermatan sulfate-substituted glycopeptides after ultrafiltration through Microcon 3, portions of one normal cornea (sample 2), one MCD type I cornea (sample 13), and one MCD type II cornea (sample 18) were digested with proteinase K in 0.1 M ammonium acetate, pH 7.0, enzyme-inactivated, samples divided into two equal portions, and buffer evaporated by speedvac lyophilization. One set of digests was resuspended in 0.1 M ammonium acetate, pH 7.0, and incubated with 2.5 milliunits each of chondroitinase ABC and ACII, and the other set was suspended in 0.1 M ammonium acetate, pH 6.0, and incubated with 2.5 milliunits each of keratanase II and endo-␤-galactosidase. Product yields were determined by FACE separation and showed that both the fine structural compositions and the tissue contents (calculated per mg dry weight of tissue) of all glycosaminoglycan types in each sample were essentially identical to those obtained with glycopeptides isolated from that sample after Microcon 3 ultrafiltration (see Table II).
Keratan Sulfate, Polylactosamine, and Chondroitin/Dermatan Sulfate Chain Length-Portions of purified glycopeptides (from 0.1 to 0.5 mg dry weight of tissue) from five normal corneas, all diseased corneas, and from two normal and three MCD type I cartilages were dissolved in 100 l of 0.1 M ammonium acetate, pH 6.0, and incubated for 18 h in the absence or presence of 2.5 milliunits of endo-␤-galactosidase. Buffer was removed by speedvac lyophilization, and samples were resuspended in 50 l of 0.1 M ammonium acetate, pH 5.0, for additional digestion with 0.1 milliunit of ␤-N-acetylglucosaminidase for 24 h. The molar amounts of non-reducing terminal GlcNAc released during the exo-glucosaminidase digestions were determined by FACE. Keratan sulfate and polylactosamine chain lengths were then calculated from the molar ratios of endo-␤-galactosidase-generated non-reducing terminal GlcNAc and total keratanase II and endo-␤-galactosidase products (see Table III); both data sets were obtained separately with glycopeptides prepared from equivalent dry weight portions of tissue. Average chain lengths for skeletal keratan sulfates and polylactosamines were also calculated from the molar ratio of non-reducing terminal NeuA and the total keratanase II and endo-␤-galactosidase-generated products (29), and these values were found to be essentially identical to those calculated by the above method.
Values for the number of average disaccharides per chain for chondroitin/dermatan sulfate (Table IV) were determined after FACE separation of chondroitinase ABC/ACII digestions, as described (25,28), and represent the molar ratio of non-reducing terminal GalNA4S (as the only detectable terminal residues in chondroitinase digests) to the sum of all ⌬disaccharide products (⌬di0S, ⌬di4S, and ⌬di6S). a Based on enzyme-linked immunosorbent assay using monoclonal antibody 1/20 5-D-4 (see "Materials and Methods").
c Images of FACE gels of KS and CS/DS digestion products are shown only for samples 1-3. d Values for immunoreactive KS in normal adults has been shown to be ϳ200 -300 ng/ml as determined by enzyme-linked immunosorbent assay with monoclonal antibody 1/20 5-D-4 (see "Materials and Methods").

FACE Analyses of KS Sulfation in Normal and MCD Corne-
as-Glycopeptides prepared from normal and diseased corneas were digested with keratanase II and endo-␤-galactosidase under conditions previously shown to completely depolymerize unsulfated, mono-, or disulfated and fucosylated regions of KS chains (29), and all digestion products were separated by FACE. Typical gel images are shown in Fig. 1, and quantitative data derived thereof are summarized in Fig. 2. In all 11 normal corneas (samples 1-11, only samples 1-3 are shown in Fig. 1), mono-and disulfated disaccharides (Gal␤1,4GlcNAc6S and Gal6S␤1,4GlcNA6S, bands 3 and 8) were generated with kera-tanase II, with a large proportion of the monosulfated disaccharides further hydrolyzed to the monosaccharides, Gal and GlcNAc6S (Fig. 1, panel A, left-hand margin), by an exo-galactosidase activity in the keratanase II enzyme (29). Endo-␤galactosidase digestion of aliquots of the normal samples generated unsulfated disaccharides (GlcNAc␤1,3Gal) and the monosulfated disaccharide (GlcNAc6S␤1,3Gal) (Fig. 1, panel B, bands 1 and 4, respectively). A set of higher molecular weight oligosaccharides composed of multiple consecutive repeats of disulfated disaccharides that confer immunoreactivity with several monoclonal antibodies (30) were also produced by this enzyme (ODS, Fig. 1, panels B and C). No NeuA-terminating  oligosaccharides were detected in either enzyme digest indicating that such non-reducing terminal caps are absent or of very low abundance on keratan sulfate in normal adult human corneas. When glycopeptides from the eight diseased corneas were digested with either keratanase II or endo-␤-galactosidase, distinctly different spectra of products were obtained (Fig. 1, panels A-C, samples [12][13][14][15][16][17][18][19]. First, digests from samples 12 and 14, of patients diagnosed by immunoassay as carrying the MCD type I mutation, were devoid of any of the sulfated mono-, di-, or oligosaccharide products that were abundantly generated in all normal samples. Instead, an increased amount of the unsulfated disaccharide (GlcNAc␤1,3Gal, band 1) was generated by endo-␤-galactosidase ( Figs. 1 and 2, samples 12 and 14), in keeping with previous observations that glycoproteins secreted by MCD type I keratocytes are substituted with polylactosamine instead of keratan sulfate (20). Keratanase II and endo-␤-galactosidase digestion of glycopeptides from corneas of patients identified with the MCD type II mutations (samples 16 -19) were also distinctly abnormal. These contained unsulfated disaccharides (Fig. 1, panel B), and notably they also contained mono-and disulfated disaccharides. In all four MCD type II samples, Gal sulfation, required for generation of disulfated disaccharides (Gal6S␤1,4GlcNAc6S), was proportionally lower than in the normal samples (Table III), indicating that the activity of the Gal-6-sulfotransferase may additionally be compromised in the MCD type II cells or that keratocytes from such patients produce less core protein acceptors than the normal cells. Moreover, the disulfated disaccharides that were present in the MCD II type keratan sulfate were not organized in multiple repeats, as ODS-oligosaccharides were not detected after the endo-␤-galactosidase digestion (Fig. 1, panel C). The values for each of the diseased corneas (samples [12][13][14][15][16][17][18][19] represents the mean Ϯ S.D. of duplicate analyses of each tissue sample. The relative molar abundances of the products were further used to calculate the proportional sulfation of GlcNAc and Gal residues (see Table II). ϫ designates the unclassified diseased tissue.
For samples 12 and 14 (MCD type I) and samples 16 -19 (MCD type II), both immunochemical (Table I) and biochemical fine structure analyses (Table II) of MCD keratan sulfate types were in good agreement. However, for sample 13, classified by immunoassay as MCD type I, the corneal keratan sulfate showed unexpectedly high levels of sulfated products in our biochemical FACE assays (Table II). Mono-and disulfated disaccharides were present, and the latter were further organized into oversulfated oligosaccharides (Fig. 1, panels A and  B). The apparent lack of immunoreactivity of this sample may be explained if the keratan sulfate was atypically distributed in the matrix, such that epitope concentrations or accessibility remained below detection, and it is clear that this patient should be reclassified as an MCD type II.
Subtype classification by immunochemistry (Table I) was not possible for sample 15, but the biochemical keratan sulfate analyses were very similar to the two typical MCD type I samples (samples 12 and 14), with regards to the elevated levels of polylactosamine (Fig. 1, panel B). In addition, small amounts of both mono-and disulfated disaccharides but no oversulfated oligosaccharide sequences (ODS) were detected in endo-␤-galactosidase digests (Fig. 1, panels A and C, respectively), typical of that seen in the MCD type II corneas (Fig. 2,  lower panel). It is likely that this patient may have an additional defect in the Gal-6-sulfotransferase but should be classified as an MCD type I variant.
Quantitation of glycosaminoglycans from the enzyme digestion products after FACE separation (Table III) clearly showed that all MCD mutations lead to a significant reduction in keratan sulfate/polylactosamine contents per dry weight of tissue (Table III). Thus, the total keratan sulfate (g per mg dry weight, as sum of all enzyme digestion products) was ϳ15 g for normal and ϳ1.8 g for MCD type I, between 0.5 and 2.1 g for MCD type II, and 2.2 or 4.2 g for the unclassified samples 13 or 15, respectively. It should be noted that the decreased keratan sulfate contents was attributable to both decreased chain sizes (3-4 in diseased and 14 disaccharides in normals, Fig. 3 and Table III) and fewer chains per mg of tissue (0.5-1.8 nmol in diseased and 2.2 in normals, Table III). The latter results are consistent with a reduction of keratan sulfate-substituted lumican or keratocan core proteins in the stromal extracellular matrix.

FACE Analyses of Chondroitin/Dermatan Sulfate and Hyaluronan from Normal and MCD
Corneas-In addition to lumican and keratocan, the corneal stroma contains chondroitin/ dermatan sulfate on decorin, and tissues were therefore analyzed for sulfation and contents of these glycosaminoglycan types (31)(32)(33)(34)(35). The FACE analyses of chondroitin lyase digests ( Fig. 4 and Fig. 5) showed that normal corneas contained significant amounts of unsulfated chondroitin (detected as ⌬di0S), as well as dermatan sulfate as shown by generation of ⌬di4S by chondroitinase ABC but not by chondroitinase ACII. Chondroitin 6-sulfate, detected as the ⌬di6S product after chondroitinase ABC or ACII digestion, was a relatively minor component in normal tissues.
There were obvious differences in both chondroitin/dermatan sulfate composition and contents between normals and diseased corneas (Fig. 5 and Table IV). There was a marked decrease in the content of unsulfated chondroitin from 4.5 g per mg dry weight of tissue in normals (n ϭ 11) to 0.3-2.6 g per mg dry weight in the eight diseased tissues. With the exception of samples 13 and 15, all diseased corneas had elevated levels of chondroitin 4-and 6-sulfate. Chondroitin 4-sulfate had increased from 2.5 g per mg dry weight in the normal corneas to ϳ3.6 -10.9 g per mg dry weight and chondroitin 6-sulfate from 0.7 g per mg dry weight to 1.8 -7.9 g per mg dry weight. No correlation was found, however, between the extent of loss of unsulfated polymer and the changes in sulfation patterns of chondroitin/dermatan sulfate and disease subtypes for the samples analyzed here. The calculated number of average disaccharides per chain for chondroitin/dermatan sulfate was also smaller in the diseased than the normal corneas. By using the assumption that both unsulfated and sulfated regions are co-localized within the same chain, normal tissues  Table IV). Furthermore, in contrast to the decreased keratan sulfate chain concentrations (Table III), those for chondroitin/dermatan sulfate chains were increased by ϳ70% in MCD corneas (Table IV), consistent with an accumulation of proteoglycan core proteins substituted with such chains. Corneas from patients 13, 14, and 17 also showed elevated levels of hyaluronan (Fig. 4), a matrix component that is essentially absent from normal corneas. Interestingly, both hyaluronan and chondroitin/dermatan sulfate proteoglycan accumulations have been reported in a number of corneal diseases (36) and suggested to reflect a more generalized response of the keratocytes to wounding.

FACE Analyses of Keratan Sulfate from Normal and MCD I Cartilages-
The recent molecular genetics studies (18) suggested that the GlcNAc-6-sulfotransferase gene affected in MCD has a restricted expression in corneal keratocytes. This conclusion appears to conflict with other published data (11) in which serum and cartilages of patients with MCD type I lacked the immunoreactive keratan sulfate. Therefore, we also examined the fine structure of auricular cartilage keratan sulfate from MCD type I patients (Figs. 6-8).
Cartilages from patients with MCD type I were totally devoid of sulfated keratan sulfate, as glycopeptides derived from this tissue gave no products after digestion by keratanase II (Fig. 6, panel B, samples 22-24). However, endo-␤-galactosidase digestion generated an abundance of unsulfated disaccharides and the unsulfated NeuA capped tetrasaccharides (Fig. 6,  panel B). Keratan sulfate chain elongation was reduced in this tissue, with polylactosamine chains in MCD type I cartilages composed of 3 Ϯ 1 disaccharides, compared with keratan sulfate chains in normal cartilages which were composed of 7 Ϯ 2 disaccharides per chains. Chondroitin sulfate and hyaluronan contents (g per mg dry weight of tissue) were significantly decreased in the diseased cartilages, relative to the normal (Table IV), but chondroitin sulfate fine structures (sulfation and chain lengths) were essentially identical in normal and MCD I cartilages (data not shown). Therefore, these data com- pletely support earlier immunochemical data showing that the MCD type I defect in the GlcNAc-ST alters the matrix composition of both cartilages and corneas.

DISCUSSION
The FACE analyses of keratan and chondroitin/dermatan sulfates in normal and diseased corneas reported here provide new insights into structural aspects of these components in stromal function. The sulfation patterns and the contents per dry weight of both glycosaminoglycan types were remarkably similar for all 11 normal samples. This tight control of proteoglycan composition and content undoubtedly underlines the necessity of these molecular components for maintenance of tissue matrix organization and corneal transparency. On the other hand, sulfation, chain length, and contents of each glycosaminoglycan type were altered in each one of the eight diseased corneas analyzed here, and such alterations are likely involved in the consequential disruption of matrix organization and the impaired corneal transparency (32,38).
The accurate biochemical determination of contents and fine structures of chondroitin/dermatan and keratan sulfate in very small pieces of diseased tissues provided considerably more insight into the classifications of the keratan sulfate abnormalities in MCD types than the previously available immunological criteria, and in two cases provided a more accurate designation of the disease subtype. Traditionally, the keratan sulfate deposited in diseased corneas had been quantitated and characterized by immunoassays with keratan sulfate-specific monoclonal antibodies such as 5D4. It is important to note that high affinity binding by these commonly used antibodies requires multiple adjacent repeats of disulfated disaccharides (Gal6S␤1,4GlcNAc6S) (30). Therefore, lack of immunoreactivity cannot be simply interpreted as an absence of sulfated keratan sulfate, as only disulfation of the polymer may be decreased, creating a low epitope density. As an alternative measure of keratan sulfate, biochemical analyses of metabolically labeled keratan sulfate produced in vitro by corneal explants had been described (8,9,20,39). Despite the more detailed structural information obtained by this approach, the data have been difficult to interpret, since the keratan sulfate synthetic pathways in cultured tissues and cells are labile (40,41), and radiolabeling may therefore not reflect actual amounts or composition of glycosaminoglycans resident in the tissue at time of removal from the patients. Nonetheless, the FACE analyses data obtained for corneas of MCD type I patients 4 and 6 generally support the interpretation of both immunochemical data (summarized in Table I) and previous radiolabeling experiments with corneal tissues from these two indi-viduals (20). Hence, an MCD I type mutation in the coding region of the sulfotransferase must result in synthesis of inactive forms of the enzyme, leading to the synthesis and secretion of proteoglycans substituted with polylactosamine instead of keratan sulfate.
Distinct structural abnormalities were detected for keratan sulfates deposited in corneas of MCD type II patients (Figs. 1 and 2). Sulfation of GlcNAc and Gal residues was detected, although both were at proportionally lower levels compared with those seen in keratan sulfates from normal corneas (Table  II). Such a finding would be consistent with confinement of MCD type II mutations to the regulatory region of the sulfotransferase gene, leading to decreased induction of enzyme expression and possibly only in keratocytes. An unexpected finding was that FACE analyses of endo-␤-galactosidase products from MCD II samples did not show the typical pattern of oversulfated oligosaccharides, seen in abundance in digests of keratan sulfates from normal corneas (ODS, Fig. 1, panels B and C). The absence of such oligosaccharide sequences might be expected to decrease 5D4 reactivity greatly, but these corneas reacted with the keratan sulfate antibody (see Table I). A diagnostic positive immunohistochemical staining of tissue sections may occur even at the very low concentrations of core protein-bound disulfated keratan sulfate chain regions or, alternatively, indicate that the abnormally elongated and sulfated keratan sulfate may contain other structural characteristics that produce a positive reactivity with this type of antibody.
The detailed structural description of abnormal keratan sulfate produced by keratocytes and chondrocytes harboring MCD mutations also provided for additional insights into the mechanisms of sulfation and chain elongation during biosynthesis of this glycosaminoglycan. First, the total absence (in MCD I) or relative decrease (in MCD type II) in Gal sulfation (Table II) confirms previous observations that Gal sulfation may be dependent on preceding or concurrent sulfation of GlcNAc residues by GlcNAc6ST (42,43). In addition, the absence of stretches of adjacent disulfated disaccharides in MCD type II keratan sulfate (Fig. 1, panel B) suggests that a critical minimum polymer size is necessary for synthesis of oversulfated regions in the chain interior. Conversely, as suggested by earlier in vitro studies (44,45), keratan sulfate chain elongation may only proceed in conjunction with efficient sulfation, and the undersulfated chains in both types of MCD, and in both tissues examined, were significantly shorter than those isolated from normal tissues. Thus, premature chain termination in the absence of appropriate sulfation may occur if sulfation FIG. 8. Identification of the non-reducing terminal tri-and tetrasaccharides in human auricular and nasal cartilages. Glycopeptides from MCD type I auricular cartilage (sample 20) were digested with endo-␤-galactosidase alone (lane a), endo-␤-galactosidase and neuraminidase (lane b), or endo-␤-galactosidase, neuraminidase, and ␤-galactosidase (lane c). Products were fluorotagged and separated by FACE (using an extended (2.5 h) electrophoresis time). The migrations of disaccharide glcNAc␤1, 3gal, the trisaccharide gal␤1,4glcNAc ␤1,3gal, and neuA were determined by coelectrophoresis of standards as reported (29) (data not shown).
per se stabilizes the conformation of the non-reducing terminal sugar for recognition by the appropriate glycosyltransferase during active polymerization. Moreover the C-GlcNAc6ST enzyme may be critical to the stabilization of the proposed multienzyme complexes containing the glycosyl-and sulfotransferases that act collectively during the process of keratan sulfate synthesis (45). Indeed, both impaired chain elongation and decreased contents of substituted lumican or keratocan core proteins are the reason for significantly lower concentrations of sulfated and/or unsulfated (Gal␤1,4GlcNAc) polymer in all diseased corneas (Table III).
It is also significant that all diseased corneas examined here showed a substantial increase in sulfated chondroitin/dermatan sulfate and, in some cases, hyaluronan. Accumulation of these glycosaminoglycans in the corneal stroma has been reported for other degenerative diseases (36,46), where their elevated production by keratocytes likely represents a metabolic response of the cells to an abnormal (for example keratan sulfate-deficient) extracellular matrix. It is therefore entirely possible that the progressive development of the clouding in MCD corneas results not simply from the lack of keratan sulfate but also from the deposition of these additional glycosaminoglycans, which may themselves interfere with normal spacing between collagen fibrils. MCD type I mutation also affects keratan sulfate synthesis by chondrocytes, but in addition, the auricular cartilages from these patients had significantly lower contents of 4-and 6-sulfated chondroitin sulfate and hyaluronan per tissue dry weight (Table IV). The mechanism underlying this compositional difference is unclear, but it is possible that the biological control of these processes such as chondrogenesis or cartilage turnover in postnatal growth requires the presence of proteoglycans with sulfated keratan sulfate chains (47).
We have shown in this study that glycosidase depolymerization of glycosaminoglycans coupled with FACE analyses allowed accurate determination of their tissue contents and sulfation, representing a powerful tool for relating base mutations in genes encoding glycosaminoglycan-specific post-translational modification enzymes, to structural and functional abnormalities of secreted proteoglycans. Since analyses of this type can be readily performed on very small amounts of tissue, such as a few milligram, a broader application for specimen analyses obtained at keratoplasty is likely to provide novel insight into abnormalities in proteoglycan contents and structures in a range of other corneal diseases.