Elevated Expression of O-GlcNAc–Modified Proteins and O-GlcNAc Transferase in Corneas of Diabetic Goto-Kakizaki Rats

Abstract

purpose. The hexosamine biosynthetic pathway is one of the possible mechanisms involved in diabetic keratopathy. The purpose of this study was to examine the role of O-glycoside–linked N-acetylglucosamine (O-GlcNAc) modification of proteins in the pathogenesis of diabetic keratopathy in the Goto-Kakizaki (GK) rat, which has spontaneous development of diabetes.

methods. An anti-O-GlcNAc antibody, an anti-O-GlcNAc transferase antibody, and digoxigenin (DIG)-labeled cRNA probes were used to examine the localization of O-GlcNAc–modified proteins, O-GlcNAc transferase protein and mRNA in the corneas of diabetic GK rats and in those of nondiabetic Wistar rats. The corneas from Wistar rats were organ cultured in control medium or in medium containing 100 μM O-(2-acetamide-2-deoxy-d-glucopyranosylidene) amino-N-phenyl-carbamate (PUGNAc), an inhibitor of O-GlcNAcase, the enzyme that removes O-GlcNAc from proteins. The morphologic changes were examined by electron microscopy.

results. In normal corneas, immunoreactive O-GlcNAc and O-GlcNAc transferase were observed in the epithelial, endothelial, and stromal cells. In the diabetic corneas, their immunoreactivities in the epithelium were increased in intensity. Morphologically, the number of hemidesmosomes in the epithelial basal cells was lower than that in those cells from the nondiabetic rats. In some areas, the basement membrane had detached from the epithelial basal cells. The PUGNAc treatment of Wistar rat corneas increased the level of O-GlcNAc immunoreactivity and caused a decrease in the number of hemidesmosomes and the detachment of corneal epithelial cells from the basement membrane.

conclusions. The elevated expression of O-GlcNAc–modified proteins and O-GlcNAc transferase may play a causative role in the corneal epithelial disorders of diabetic GK rats.

Keratopathy has been reported to occur in the corneas of diabetic patients. ^{1 2} Various degrees of epithelial disturbance such as superficial punctate keratopathy, recurrent corneal erosion, or persistent epithelial defects take place in the diabetic cornea. Diabetic keratopathy is based on many factors, such as decreased corneal sensitivity, hypolacrimation, aberration of basal cell adhesion, suppressed cell division, and abnormality of the epithelial basement membrane. The involvement of the activation of the polyol pathway, in which aldose reductase is the key enzyme, and the accumulation of advanced glycation end products (AGEs) have been implicated in the pathogenesis of diabetic keratopathy. ^{2 3 4 5} However, the pathogenic mechanisms remain to be elucidated. 

Recently, other than the activation of the polyol pathway and the accumulation of AGEs, the hexosamine pathway has also been suggested as one of the possible pathogenic mechanisms underlying diabetic complications. ^{6 7} Excess glucose increases the glucose flux through the hexosamine pathway. Uridine diphosphate N-acetylglucosamine (UDP-GlcNAc) is the end product of the hexosamine pathway and is a donor for both N-linked and O-linked protein glycosylation. O-linked N-acetylglucosamine moieties (termed O-GlcNAc) are attached to serine or threonine residues of many nuclear and cytoplasmic proteins by a UDP-N-acetylglucosamine/peptide N-acetylglucosaminyltransferase (EC 2-4-1, O-GlcNAc transferase). ^{8 9 10} Because O-GlcNAc transferase utilizes this UDP-GlcNAc in catalyzing the attachment of O-GlcNAc to proteins, the increased level of extracellular glucose may lead to elevated intracellular O-GlcNAc modification of proteins. Our studies have demonstrated that the concentration of O-GlcNAc–modified proteins increases in the pancreas of rats with streptozotocin (STZ)-induced diabetes. ¹¹  

Many nuclear and cytosolic proteins are glycosylated on serine or threonine residues by O-linked β-N-acetylglucosamine (O-GlcNAc). Modification by O-linked GlcNAc is a regulatory posttranslational process that has a complex dynamic interplay with O-phosphorylation. O-GlcNAc and O-phosphate alternatively occupy the same or an adjacent site. A reciprocal relationship between O-GlcNAc- and phosphorylation-mediated modification has been observed. ¹² O-GlcNAc modification may regulate phosphorylation and be involved in signal transduction, ¹³ and it has been suggested that a high level of O-GlcNAc modification induces insulin resistance by inhibiting the insulin signaling cascade. ¹⁴  

O-GlcNAc transferase is a unique nuclear and cytosolic glycosyltransferase that contains multiple tetratricopeptide repeats. ^{9 15} The liver enzyme contains 2 immunologically related subunits of M_r 110 kDa (α-subunit) and 78 kDa (β-subunit). Other tissues, such as brain, contain only the α-subunit, which contains the active site. The α-subunit forms homo- and heterotrimers that may have different binding affinities for UDP-GlcNAc over the entire physiological range. ¹⁶ The cDNA encoding the O-GlcNAc transferase α-subunit has been cloned from rats, Caenorhabditis elegans, and humans. ^{15 17} O-GlcNAc transferase is highly conserved from C. elegans to humans, and maps to the locus for X-linked Parkinsonia dystonia. ¹⁸ O-GlcNAc transferase is unlike any glycosyltransferase previously described. ^{19 20} It has been found in all tissues examined and is especially abundant in the pancreas and brain. ^{15 17} In the pancreas, O-GlcNAc transferase is abundant in the islet cells, ²¹ and in the brain it is found in the neurons. ²² However, the localization of O-GlcNAc transferase protein in the cornea has not been determined yet. 

The spontaneously diabetic Goto-Kakizaki (GK) rat is a nonobese model of non–insulin-dependant diabetes mellitus (NIDDM) that was developed by the selective breeding of glucose-intolerant Wistar rats. ^{23 24 25} To elucidate the role of O-GlcNAc modification in the pathogenesis of diabetic keratopathy, we studied the localization of O-GlcNAc transferase proteins, transcripts, and O-GlcNAc–modified proteins and examined whether the expression of O-GlcNAc–modified proteins and O-GlcNAc transferase are altered in diabetic corneas. 

Methods

Animals and Tissues

Animal studies were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. All experimental procedures using laboratory animals were approved by the Animal Care and Use Committee of Kyorin University School of Medicine. The corneas of 15-, 33-, and 62-week-old male (n = 6 for each age) Goto-Kakizaki rats and Wistar rats (as controls), obtained from Kurea (Tokyo, Japan), were used in the present study. Serum glucose levels (mean ± SEM) in Wistar and GK rats were 158.0 ± 12.0 and 375.9 ± 11.6 mg/dL at 15 weeks, 118.5 ± 10.5 and 333.8 ± 22.4 mg/dL at 33 weeks, and 167.0 ± 28.6 and 314.9 ± 48.7 mg/dL at 62 weeks, respectively. As has been reported, ^{24 26 27} serum insulin levels were approximately two times higher in GK rats. 

Antibodies

Rabbit polyclonal anti-O-GlcNAc transferase antibody (AL-25, purified IgG) was kindly provided by Gerald W. Hart (Johns Hopkins University, Baltimore, MD). ¹⁵ AL-25 recognizes both 110- and 78-kDa subunits of O-GlcNAc transferase. ^{15 21 28} The mouse monoclonal anti-O-GlcNAc antibody (RL2) was obtained from Affinity BioReagents (Golden, CO). RL2 specifically recognizes O-GlcNAc in β-O-glycoside linkage to either serine or threonine. ²⁹ Horseradish peroxidase (HRP)–conjugated donkey anti-rabbit IgG antibody, HRP-conjugated goat anti-mouse IgG antibody, normal rabbit IgG, and normal mouse IgG were obtained from Jackson ImmunoResearch (West Grove, PA). Fluorescence-conjugated (Alexa₄₈₈) donkey anti-mouse IgG and a nucleic acid stain (TO-PRO-3) were obtained from Molecular Probes (Eugene, OR). 

Preparation of cRNA Probes

Antisense and sense RNA probes were prepared by in vitro transcription from a fragment of O-GlcNAc transferase cDNA (nucleotides 1193-1942) prepared based on the cDNA sequence of rat O-GlcNAc transferase in the GenBankTM/EBI Data Bank (accession number U76557; http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD) by use of T3 or T7 RNA polymerase in the presence of digoxigenin (DIG)-linked UTP in the reaction mixture, as described previously. ²¹  

In Situ Hybridization

In situ hybridization was performed with digoxigenin (DIG)-cRNA on cryosections, as described earlier. ²¹ Cryosections of unfixed rat cornea were cut at a 4-μm thickness, thaw mounted onto silane-coated slides, allowed to air dry, fixed in 4% paraformaldehyde-PBS for 1 hour, incubated twice in 0.1% (wt/vol) diethyl pyrocarbonate (DEPC)-PBS (15 minutes each time), and rinsed with DEPC–treated 4× SSC. The sections were then prehybridized in prehybridization buffer (50% formamide, 1× Denhardt solution, 0.6 M NaCl, 10 mM Tris-HCl, 1 mM EDTA, 100 mg/mL salmon sperm DNA, and 100 mg/mL yeast tRNA) at 45°C for 3 hours, and then hybridized in hybridization buffer (50% [vol/vol] formamide, 1× Denhardt solution, 0.6 M NaCl, 10 mM Tris-HCl [pH 7.4], 1 mM EDTA, 100 mg/mL salmon sperm DNA, and 0.4 mg/mL DIG-labeled probe RNA) at 50°C for 15 hours. The sections were next sequentially rinsed with DEPC-treated 4× SSC at 45°C for 15 minutes and with 50% formamide and 2× SSC at 45°C for 15 minutes, and then incubated in 1 mg/mL RNase A solution at 37°C for 15 minutes. Next, they were rinsed with DEPC-treated 2× SSC at 45°C for 15 minutes and then twice with DEPC-treated 0.5× SSC at 45°C for 15 minutes. The sections were thereafter incubated in a solution of polyclonal sheep anti-DIG Fab antibody conjugated to alkaline phosphatase (1:500 diluted in DIG buffer containing 0.1 M Tris-HCl [pH 9.5], which in turn contained 0.1 M NaCl and 50 mM MgCl₂). The signal was detected with nitroblue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3-indolylphosphate (BCIP; Roche Diagnostics, Indianapolis, IN). 

Immunohistochemical Localization of O-GlcNAc and O-GlcNAc Transferase

Corneas from diabetic and nondiabetic rats were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 1 hour at 4°C. The specimens were then embedded in paraffin and cut into 4-μm-thick sections. The sections were deparaffinized in xylene and rehydrated through graded concentrations of ethanol. They were then treated with 3% hydrogen peroxide in PBS for 15 minutes and subsequently rinsed with PBS. To increase the immunoreactivity, we placed the sections in 0.01 M citric acid-buffered solution (pH 7.0) and microwaved them (500 W) for 5 minutes. After having been thoroughly washed, the sections were incubated with 5% normal bovine serum albumin (BSA) in PBS for 20 minutes at room temperature. They were then incubated with rabbit anti-O-GlcNAc transferase polyclonal antibody (AL25, 1:200), or with mouse anti-O-GlcNAc monoclonal antibody (RL2, 1:200), in 0.1% BSA in PBS for 1 hour at room temperature. After another wash in PBS, the sections were incubated with HRP-conjugated secondary antibodies (1:200) for 1 hour at room temperature, washed again with PBS, and immersed in 3,3′-diaminobenzidine-4HCl (DAB, 0.2 mg/mL)-H₂O₂ (0.005%) for 10 minutes at room temperature. For a control experiment, the specimens were incubated with normal rabbit or mouse IgG or with 0.1% BSA-PBS alone instead of the primary antibodies. No positive staining was observed in the control experiment (data not shown). 

Immunofluorescence observation for O-GlcNAc localization was performed as described earlier. ^{28 30} Corneas were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.3) for 1 hour at 4°C. After having been washed with PBS, the specimens were embedded in optimal cutting temperature (OCT) compound (Miles, Elkhart, IN). Frozen sections (4 μm thick) were cut, washed with PBS, and treated for 10 minutes with 5% BSA in PBS. The sections were then incubated with RL2 (1:200) for 1 hour at room temperature, washed with PBS, and subsequently incubated with fluorescence-conjugated donkey anti-mouse IgG antibody (1:200, Alexa₄₈₈; Molecular Probes). Nuclei were stained with a nucleic acid stain (TO-PRO-3; 1:500, Molecular Probes). After a final wash with PBS, the specimens were mounted in 90% glycerol and 0.1M Tris-HCl buffer (pH 8.5) containing 0.5 mM p-phenylene diamine, and observed under a laser scanning confocal microscope (MRC-1024; Biorad, Richmond, CA). A control experiment was performed in the same manner. Immunostaining intensity in nuclear and cytoplasmic regions was quantified on a computer (Macintosh; Apple Computer, Cupertino, CA) using the public domain NIH Image program (available by ftp at zippy.nimh.nih.gov/ or at http://rsb.info.nih.gov/nih-image; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD) or ImageJ (a Java image processing program similar to NIH image and also developed by Wayne Rasband, which runs in Windows [Microsoft, Redmond, WA] and is available from http://rsb.info.nih.gov/ij/). The results (mean ± SEM) represent duplicate measurements in six separate experiments. The terms increase and decrease are used only when the results are statistically significant (Student’s t-test, P < 0.05). 

Electron Microscopy

Corneas were fixed in phosphate-buffered 2.5% glutaraldehyde (pH 7.4), postosmicated, and dehydrated in graded concentrations of alcohol. After immersion in propylene oxide, the specimens were embedded in Epon resin (Embed 812, Electron Microscopy Sciences, Fort Washington, PA). Ultrathin sections were cut perpendicular to the epithelium, doubly stained with uranyl acetate and lead citrate, and examined with a transmission electron microscope (TEM-1010C; JEOL, Tokyo, Japan). 

Tissue Culture

Corneas of Wistar rats were removed, pinned to a rounded paraffin post, and incubated for 4 days in a completely defined medium ³¹ in the absence or presence of PUGNAc (100 μM; Carbogen, Aarau, Switzerland) at 35°C in a 5% CO₂ humidified atmosphere. The medium was changed every day. The concentration of 100 μM PUGNAc used in this study was chosen on the basis of preliminary experiments. At 10 μM, PUGNAc did not affect corneal morphology, and at 200 μM the results were the same as those at 100 μM. 

Results

Changes in Localization of O-GlcNAc in the Diabetic Cornea

The localization of O-GlcNAc was examined immunohistochemically in the Wistar and GK rat corneas in animals 15, 33, and 62 weeks of age. In all the nondiabetic corneas examined at the different ages, O-GlcNAc immunoreactivity was observed mainly in the nuclei of the epithelial cells, endothelial cells, and stromal cells (Fig. 1) . In the diabetic corneas, no change in O-GlcNAc immunoreactivity was observed at 15 weeks of age (Figs. 1A 1B 1C 1D) ; however, by 33 weeks the immunoreactivity had increased in intensity (Figs. 1E 1F) . Especially in the epithelial basal cells, intense immunoreactivity was observed not only in the nucleus but also in the cytoplasm (Fig. 1F) . In the 62-week-old diabetic cornea, intense staining was observed in both the intermediate and basal layers of the epithelium (Fig. 1J) . No change in immunoreactivity was observed in the endothelium of the diabetic cornea (Figs. 1C 1D 1G 1H 1K 1L) . 

Changes in Localization of O-GlcNAc Transferase in the Diabetic Cornea

The localization of O-GlcNAc transferase was examined immunohistochemically in the Wistar and GK rat corneas obtained in animals 15, 33, and 62 weeks of age. Almost the same localization as found for O-GlcNAc was observed. In all the nondiabetic corneas examined at different ages, O-GlcNAc transferase immunoreactivity was observed mainly in the nuclei of the epithelial, endothelial, and stromal cells (Fig. 2) . At 15 weeks no change in O-GlcNAc transferase immunoreactivity was observed in the diabetic corneas (Fig. 2A 2B 2C 2D) . In the 33-week diabetic corneas, immunoreactivity was observed not only in the nuclei but also in the cytoplasm of the epithelial basal cells (Fig. 2F) . In the 62-week-old diabetic cornea, intense immunoreactivity was observed in both the intermediate and basal layers of the epithelium (Fig. 2J) . No change in immunoreactivity was observed in the diabetic corneal endothelium (Figs. 2C 2D 2G 2H 2K 2L) . 

In Situ Hybridization Analysis

The localization of O-GlcNAc transferase mRNA in the 62-week-old corneal epithelium of normal and diabetic corneas was examined by in situ hybridization histochemistry with a DIG-labeled antisense RNA probe. The hybridization signals were observed in all layers of the corneal epithelium of either kind of rat (Fig. 3A 3B) . Although O-GlcNAc transferase immunoreactivity was highest in the basal cells of the normal corneal epithelium (Fig. 2I) , hybridization signals were higher in the suprabasal cells (Fig. 3A) . The hybridization signals in the diabetic corneal epithelium (Fig. 3B) were more intense than those in the normal one (Fig. 3A) . When an equal concentration of the sense RNA probe was used instead of the antisense riboprobe, no positive signals were observed in the normal corneal epithelium (Fig. 3C) . These results indicate that O-GlcNAc transferase mRNA is expressed in corneal epithelial cells and is expressed more in the diabetic ones than in the nondiabetic ones. 

Morphologic Changes in the Diabetic Cornea Observed by Electron Microscopy

The GK rat cornea was examined for possible morphologic changes. At 15 weeks, no morphologic change was observed (data not shown). At 33 and 62 weeks, the epithelial basement membrane in some areas had detached from the epithelial basal cell surfaces, although no change in the thickness of the epithelial basement membrane was noted (Fig. 4) . Cell processes of epithelial basal cells were observed between the basal surface of basal cells and the basement membrane. In the diabetic cornea, there were fewer hemidesmosomes in the epithelial basal cells than in the normal cornea (Figs. 4 5) . Anchoring fibrils were poorly developed (Fig. 4) . 

Effect of Elevated Expression of O-GlcNAc

To examine whether the elevation of O-GlcNAc modification may cause a change in the attachment between the corneal epithelium and stroma, corneas were cultured in the presence of 100 μM PUGNAc, which is an inhibitor of O-GlcNAcase, the enzyme that catalyzes removal of O-GlcNAc. For this analysis, an immunofluorescence–histochemical method was used to quantify O-GlcNAc immunoreactivity. In the Wistar rat corneas cultured in the absence of PUGNAc, intense immunoreactivity was observed in the nucleus of the epithelial cells, whereas the cytoplasm of the cells showed weak immunoreactivity (Fig. 6) . In the corneal epithelium cultured in the presence of 100 μM PUGNAc, the immunoreactivity of O-GlcNAc increased in both the nucleus and cytoplasm of the epithelial cells, but the localization of O-GlcNAc did not change (Fig. 6) . These results indicate that PUGNAc treatment increased the level of O-GlcNAc–modified proteins in the corneal epithelium. Figure 7 shows electron micrographs of Wistar corneas cultured in the presence or absence of PUGNAc. Whereas hemidesmosomes were well developed in the cornea cultured in the absence of PUGNAc, they were poorly developed in the PUGNAc-treated cornea. Moreover, intercellular spaces between epithelial cells were enlarged, and the basement membrane had detached from the basal surface of epithelial basal cells in some areas of the PUGNAc-treated corneas. These results indicate that the elevated expression of O-GlcNAc leads to reduction in cell–cell adhesion and attachment of the corneal epithelium to the stroma in PUGNAc-treated corneas. 

Discussion

The present immunohistochemical results showed for the first time that O-GlcNAc transferase mRNA and protein are expressed in corneal epithelial cells, endothelial cells, and stroma cells; localization of O-GlcNAc–modified proteins is consistent with that of O-GlcNAc transferase; and O-GlcNAc transferase expression at both mRNA and protein levels and O-GlcNAc modification are elevated in the diabetic cornea. The localization of O-GlcNAc transferase was almost the same as that of aldose reductase, a key enzyme in the polyol pathway that is one of the mechanisms of hyperglycemia-induced diabetic complications. ⁴ These results therefore suggest that O-GlcNAc transferase may also be involved in diabetic keratopathy. 

In the normal cornea the nucleus, where most O-GlcNAc–modified proteins are found, was stained more intensely than the cytoplasm. It is now clear that dynamic O-GlcNAc modification of nuclear and cytosolic proteins is as abundant and widespread as their modification by phosphorylation. ^{13 32} To date, many nuclear O-GlcNAc–modified proteins have been identified, including RNA polymerase II, its associated transcription factors, chromatin-associated proteins, nuclear oncogene, and tumor suppressor proteins, steroid receptors, and nucleoporins (p62, p58, and p54). ^{33 34 35 36 37} O-GlcNAc modification of nuclear proteins is involved in various aspects of gene expression. ³⁵ We showed earlier that O-GlcNAc transferase was predominantly localized in the euchromatin in pancreatic cells and aortic smooth muscle cells, ^{21 28} and is involved in the regulation of transcription. ^{12 35} The present immunohistochemical results indicate that O-GlcNAc modification may also correlate with the transcription of genes in the corneal cells. 

Nucleocytoplasmic O-GlcNAc modification is dynamic and abundant, exhibiting characteristics more like those of phosphorylation than those of typical N- and O-linked glycosylation. ^{38 39} O-GlcNAc transferase is the enzyme for attachment of GlcNAc to proteins and exists in both the nucleus and cytoplasm. UDP-GlcNAc, which is the end product of the hexosamine pathway, is the substrate of O-GlcNAc transferase. The activity of O-GlcNAc transferase is exquisitely responsive to intracellular UDP-GlcNAc and UDP. Many data show that aberrant O-GlcNAc modification is correlated with diabetes. ^{14 40} Our present results indicate that an increase in extracellular glucose leads to elevated intracellular O-GlcNAc modification of proteins. 

Present results indicate that the major effect of modifying elevated O-GlcNAc concentrations appears to be on cell adhesion. The corneal epithelium adheres to the stroma through attachment apparatuses that include hemidesmosomes, keratin filaments that are linked to the hemidesmosomes, anchoring filaments, anchoring plaques, and anchoring fibrils. ⁴¹ Our electron microscopic observations showed that the numbers of hemidesmosomes and anchoring fibrils were decreased in the diabetic cornea and that the basement membrane had detached from the epithelial basal cells in some areas. This observation is compatible with the clinical observation that in the corneas of diabetic patients, the corneal epithelium often easily detaches from the stroma (Kenyon K, et al. IOVS 1978;17:ARVO Abstract 1, S245), ⁴² and with the report that the penetration of anchoring fibrils into the diabetic stroma decreases. ⁴³ Phosphorylation of proteins is necessary for the complete formation of hemidesmosomes. ⁴⁴ Many cytoskeletal and membrane proteins, such as talin, vinculin, keratins, MAPS, and tau, are known to be both phosphorylated and O-GlcNAc–modified. ^{45 46 47 48 49} It also has been reported that integrins are necessary for epithelial adhesion and are altered in diabetic retinopathy. ^{50 51} Excess phosphorylation of α6β4 integrins reduces its hemidesmosome localization and decreases keratinocyte attachment to laminin, which is one of the major components of the basement membrane. ⁵² Although so far we do not know whether integrins are O-GlcNAc–modified, we speculate that the combinational effect of phosphorylation and O-GlcNAc modification of adhesion proteins and cytoskeletal proteins may be necessary for hemidesmosome formation and epithelial cell adhesion. Therefore, not only excess phosphorylation but also excess O-GlcNAc modification of these proteins may cause a decrease in cell adhesion and detachment of the basement membrane from the epithelial basal cells. 

To examine this speculation, we cultured the normal cornea with PUGNAc. PUGNAc treatment increased the level of O-GlcNAc–modified proteins, decreased the number of hemidesmosomes and caused detachment of the basement membrane from the epithelial basal cells as seen in the diabetic cornea (Figs. 6 7) . Furthermore, PUGNAc-treatment decreased cell–cell junctions (Fig. 7B) . We think these results support our speculation that elevation of O-GlcNAc modification of proteins may cause a decrease in cell adhesion—not only in the hemidesmosomes but also the desmosomes, which are major cell–cell adhesive junction link keratin filaments. Keratin filaments are formed by water-insoluble keratin proteins. Of approximately 30 different keratin molecules, keratin-8, -13, and -18 have been identified to be O-GlcNAc modified. ^{46 47} Several potential functions of O-GlcNAc modification of keratin-8 and -18 has been proposed. ⁵³ These include roles in filament assembly, subcellular localization and degradation protection. ⁵³ PUGNAc treatment decreased the keratin filaments that linked to hemidesmosomes (Fig. 7D) and desmosomes (data not shown). We speculate that O-GlcNAc modification of keratins may play a role in cell adhesion by regulating keratin filament assembly or link of keratin filaments to hemidesmosomes and desmosomes. The function of O-GlcNAc modification of keratins remains to be defined. 

In the endothelium of cornea showing diabetic keratopathy, morphologic aberration of hexagonal cells, as judged from specular microscopy, and an abnormal coefficient of variation value have been found. ⁵⁴ In the present study, intense immunoreactivity of both O-GlcNAc and O-GlcNAc transferase was also found in endothelial cells of both diabetic and nondiabetic corneas. However, no difference in staining intensity between normal and diabetic endothelium was observed. Further study is necessary to elucidate the reason why this immunoreactivity did not change in the diabetic endothelium. 

In summary, the results of our present study indicate that O-GlcNAc transferase is localized in corneal epithelial, endothelial, and stromal cells and that O-GlcNAc transferase expression and O-GlcNAc modification are increased in the diabetic cornea. These findings suggest that this elevated expression of O-GlcNAc transferase and increase in the level of O-GlcNAc–modified proteins may be linked to diabetic keratopathy. The development of O-GlcNAc transferase inhibitors as another type of antidiabetic drug appears to be a promising approach to treat diabetic keratopathy. Although, unlike the human cornea, the rat cornea lacks Bowman’s membrane, the rat corneal culture system will be a good model for analyzing the relationship between O-GlcNAc and diabetic keratopathy and for assessing the efficacy of antidiabetic drugs toward diabetic keratopathy. 

 Supported in part by Grant-in-Aid for Scientific Research 10877004 (YA) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

 Submitted for publication March 5, 2003; revised April 28, 2003; accepted May 12, 2003.

 Disclosure: Y. Akimoto, None; H. Kawakami, None; K. Yamamoto, Taisho Pharmaceutical Co., Ltd. (E); E. Munetomo, Taisho Pharmaceutical Co., Ltd. (E); T. Hida, None; H. Hirano, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Immunohistochemical localization of O-GlcNAc–modified proteins in nondiabetic (A, C, E, G, I, K) and diabetic (B, D, F, H, J, L) rats at three time points. O-GlcNAc immunoreactivity was observed in the nuclei of the epithelial, endothelial, and stromal cells of both kinds of cornea. At 15 weeks (A–D) there was no change in O-GlcNAc immunoreactivity in the diabetic corneas. At 33 weeks (E–H) the immunoreactivity in the diabetic epithelial basal cells was increased in its intensity and was observed in both the cytoplasm and nucleus. At 62 weeks (I–L) in the diabetic epithelium, strong immunoreactivity was observed in both the cytoplasm and nucleus of the winged cells and basal cells. The immunoreactivity of the endothelium in both Wistar and GK rat corneas was the same. En, corneal endothelium; Ep, corneal epithelium; D, Descemet’s membrane; S, stroma. Scale bar, 10 μm.

Figure 2.

 

Immunohistochemical localization of O-GlcNAc transferase in nondiabetic (A, C, E, G, I, K) and diabetic (B, D, F, H, J, L) corneas at three time points. O-GlcNAc transferase immunoreactivity was observed in the nuclei of the epithelial, endothelial, and stromal cells of both kinds of cornea. At 15 weeks (A–D) there was no change in the O-GlcNAc transferase immunoreactivity of the diabetic corneas. At 33 weeks (E–H) in the diabetic epithelial basal cells, the immunoreactivity was increased and observed in both the cytoplasm and nucleus. At 62 weeks (I–L), in the diabetic epithelium, a strong immunoreaction was observed in both the cytoplasm and nucleus of the winged cells and basal cells. In contrast, immunoreactivity in the endothelium had not changed. Abbreviations are as in Figure 1 . Scale bar, 10 μm.

Figure 3.

 

Distribution of O-GlcNAc transferase mRNA in the 62-week-old rat cornea epithelium detected by in situ hybridization histochemistry. (A) Nondiabetic corneal epithelium showed hybridization signals with the DIG-labeled antisense cRNA probe. (B) The diabetic corneal epithelium showed more intense signals than the nondiabetic one. (C) No hybridization signals were observed in the nondiabetic corneal epithelium when the DIG-labeled sense cRNA probe was used. Abbreviations are as in Figure 1 . Scale bar, 10 μm.

Figure 4.

 

Electron micrographs of diabetic and nondiabetic corneas of 62-week-old rats, showing the interface between the corneal epithelium and stroma. (A, B) Low- and (C, D) high-magnification images; (A, C) nondiabetic and (B, D) diabetic corneas. Hemidesmosomes and anchoring fibrils (arrows) were well developed in the nondiabetic cornea. The basement membrane in some areas had detached from the basal surface of the epithelial basal cells (B). Many cell processes of the basal cells were observed between the basement membrane and the basal surface of the epithelial basal cells (D). Hemidesmosomes and anchoring fibrils (arrows) were poorly developed in the diabetic cornea (B, D). EB, epithelial basal cell; S, stroma. Scale bar, 1 μm.

Figure 5.

 

Number of hemidesmosomes per micrometer of basal cell membrane in nondiabetic (▪) and diabetic (□) rat corneas. Data represent the mean ± SEM (n = 6). *P < 0.001.

Figure 6.

 

(A) Distribution of immunofluorescence with anti-O-GlcNAc in the Wistar rat corneal epithelium cultured in the absence (a, c) or presence (b, d) of 100 μM PUGNAc. Corneas were fixed with 4% paraformaldehyde. Frozen sections (10 μm in thickness) were reacted with anti-O-GlcNAc antibody (RL2) and then with fluorescence-conjugated goat anti-mouse IgG. Nuclei were stained with nucleic acid stain and the sections were observed with a laser scanning confocal microscope. (a, b) Green indicates the distribution of anti-O-GlcNAc reactivity. (c, d) Nucleic acid staining images (blue), which are the same optical field as in a and b, respectively. Ep, corneal epithelium; S, stroma. Scale bar, 10 μm. (B) Quantification of anti-O-GlcNAc immunoreactivity. The intensity of anti-O-GlcNAc immunoreactivity in both the nucleus and cytoplasm of the corneal epithelial cells cultured in the presence of PUGNAc increased significantly compared with that in the corneal epithelium cultured in the absence of PUGNAc. Data represent the mean ± SEM (n = 6). *P < 0.001.

Figure 7.

 

Electron micrographs of the Wistar rat cornea cultured for 4 days. (A, B) Corneal epithelium and stroma; (C, D) interface between corneal epithelium and stroma. (A, C) Cornea cultured in the absence of PUGNAc. Hemidesmosomes were well developed (C). (B, D) Cornea cultured in the presence of 100 μM PUGNAc. Intercellular spaces were enlarged in the corneal epithelium (B). The basement membrane in some areas detached from the basal surface of epithelial basal cells and hemidesmosomes were poorly developed (D). Ep, corneal epithelium; EB, epithelial basal cell; S, stroma. Scale bar (A, B) 10 μm; (C, D) 1 μm.

The authors thank Gerald W. Hart (Johns Hopkins University) for providing antibody and invaluable discussions, Minoru Fukuda, Sachie Matsubara, Miki Kanai, and Tomoko Shibata (Laboratory for Electron Microscopy and Department of Anatomy, Kyorin University School of Medicine) for technical assistance, and Akihiko Kudo and Masami Kanai (Department of Anatomy, Kyorin University School of Medicine) for helpful discussions.