Expression Patterns for Glucose Transporters GLUT1 and GLUT3 in the Normal Rat Lens and in Models of Diabetic Cataract

Abstract

purpose. To determine whether the expression levels and cellular distribution of the facilitative glucose transporters GLUT1 and -3 undergo changes in the hyperglycemic lens.

methods. Hyperglycemia was induced in vivo by injecting rats with streptozotocin or in vitro by culturing lenses in the presence of 50 mM glucose. Northern blot analysis and quantitative RT-PCR were used to detect changes in GLUT1 and -3 transcript levels, and Western blot analysis was used to monitor changes in GLUT3 protein expression levels in diabetic rats. Immunocytochemistry was used to map the cellular distribution of GLUT3 in normal and hyperglycemic lenses.

results. GLUT1 and -3 were found to be differentially expressed in the epithelial and fiber cells, respectively. In the fiber cells, the distribution of GLUT3 protein changed as a function of fiber cell differentiation. In young differentiating fiber cells, GLUT3 was mainly found in the cytoplasm, but with increasing depth into the lens became inserted into the narrow sides of older fiber cells, before becoming completely dispersed around the entire membrane of the oldest fiber cells. Hyperglycemia had similar effects on tissue damage and transporter expression in both the in vitro and in vivo models. Tissue damage was characterized by an initial local cell swelling that with prolonged insult gradually spread and resulted in the creation of large areas of tissue liquefaction. Northern blot analysis and quantitative RT-PCR showed that transcript for GLUT3 but not GLUT1 was upregulated under hyperglycemic conditions. This increase in GLUT3 expression was confirmed at the protein level by both Western blot analysis and immunocytochemistry. In hyperglycemic lenses, GLUT3 antibody labeling was localized to the region of tissue liquefaction.

conclusions. GLUT3 in the lens exhibits dynamic changes in expression levels and cellular localization as a function of fiber cell differentiation and hyperglycemia. In the lens cortex, regions of GLUT3 overexpression and hyperglycemic tissue damage overlap, suggesting a functional relationship.

The energy to maintain lens transparency is primarily derived from the uptake of extracellular glucose and its subsequent metabolism. ¹ Recently, we have shown that in neonatal rats there is a differential expression of two facilitative glucose transporter isoforms with GLUT1 being located to the epithelial cells and GLUT3 to the cortical fiber cells. ² Our identification of GLUT3 in the fiber cells suggests that these cells directly extract glucose from the extracellular space. This notion supports the postulated existence of an internal microcirculation system that is generated by a circulating flux of ions that transports water and nutrients by convection and solvent drag, respectively. ^{3 4} Furthermore, although GLUT1 and -3 are both classified as high-affinity glucose transporters, GLUT3 has the higher affinity. ⁵ Thus, the differential expression of these two transporters in the lens creates an affinity gradient for glucose uptake with the higher affinity transporter, GLUT3, extracting glucose from the tortuous extracellular space, where its concentration may normally be rate-limiting. 

In the diabetic lens elevated levels of extracellular glucose are associated with the accumulation of sorbitol, ⁶ a product of glucose metabolism, suggesting that a significant increase in glucose uptake occurs. This increase in sorbitol induces osmotic and oxidative stresses that have been postulated to overwhelm the ability of the circulation system to control lens hydration. ⁴ The earliest tissue damage associated with diabetic cataract is a localized zone of cortical fiber cell swelling. ⁷ This initial cell swelling is then followed by more extensive tissue breakdown that results in the formation of fluid lakes and cortical opacities. The localized nature of the tissue damage observed in response to hyperglycemia suggests that the regional uptake of glucose may contribute to this pattern of damage. Our finding that GLUT3 is the predominant glucose transporter in fiber cells indicates that this transporter is the most likely candidate to mediate the accumulation of intracellular glucose in these damaged fiber cells. However, being a high-affinity glucose transporter, ⁵ GLUT3 would be expected to be near saturation at physiological plasma glucose concentrations. Thus, if GLUT3 is responsible for the increase in glucose uptake, we conclude that the number of glucose transporters in the fiber cells must be upregulated in response to hyperglycemia. 

In other tissues, the effect of elevated glucose on transporter expression is controversial ^{8 9 10} and is associated with a differential regulation of both GLUT1 and -3 in a concentration- and time-dependent manner. ¹¹ Thus, to investigate these questions in the lens, we have used two models of sugar cataract that produce identical damage phenotypes. Because both models require the use of adult rats, we performed a detailed mapping of GLUT1 and -3 expression in the normal adult rat lens. Although this verified the differential expression of GLUT1 and -3 observed in neonatal rats, ² it further revealed that GLUT3 exists as a cytoplasmic pool of transporters that undergo a differentiation-dependent insertion into the membranes of the fiber cells. In addition, we determined that exposure to hyperglycemia induced an increase in the expression of GLUT3, but not of GLUT1, at transcript and protein levels. Because this increase in GLUT3 expression was localized to the zone of cortical fiber cell damage, it suggests that GLUT3 is the transporter responsible for the increased glucose uptake that occurs in the diabetic rat lens. 

Materials and Methods

Hyperglycemic Models of Lens Cataract

All animals were treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Two models of hyperglycemic insult were used. An in vivo model used 28-day-old female rats injected with 60 mg/kg streptozotocin (Sigma-Aldrich, St. Louis, MO) in 0.9% NaCl. Age-matched control female rats were injected with 0.9% NaCl. Blood glucose levels were monitored weekly, before lens tissue collection. Streptozotocin-induced diabetic animals with a blood glucose level lower than 11.5 mmol/mL were not used for further study. Blood glucose levels were also measured in age-matched control animals (average, ∼6.2 mmol/mL) and any abnormal animals were discarded from further study. After 1, 2, 3, or 4 weeks after injection, whole lenses were extracted from the rat eyes in sterile RNase free (dimethyldicarbonate [DMDC]-treated) phosphate-buffered saline (PBS; Sigma, St. Louis, MO). The second in vitro model used cultured lenses. Lenses were removed from 28-day-old female rats and transferred into sterile modified M199 (Sigma-Aldrich) with a curved glass rod. Lenses were incubated for 24 hours at 37°C in a CO₂ incubator. Lenses that were damaged during the extraction process became cloudy after this time and were discarded. Typically, approximately 50% of the extracted lenses were discarded, leaving a very small number of viable cultured lenses for further analysis. Transparent lenses were transferred into individual wells of a 24-well culture tray containing 2 mL of either modified M199 or modified M199 plus 50 mM glucose. Lenses were incubated in the appropriate medium for up to 8 days. Culture media were replaced daily with sterile, prewarmed media of the same experimental composition. 

Morphological Analysis

The transparency of lenses from both models was initially monitored by digital dark-field microscopy. Lenses were then fixed in 25% Karnovsky’s solution (50 mM Na cacodylate, 1% paraformaldehyde, 1.25% glutaraldehyde) in PBS (pH 7.4; osmolality 300 mOsmol/kg) for 4 hours at room temperature for morphologic analysis. Fixed lenses were superglued to the plate of a vibratome (Vibratome 1000; Technical Products International, Inc., St. Louis, MO). Equatorial or axial sections (170 μm thick) were cut. Sections were incubated in FITC-conjugated wheat germ agglutinin (WGA; 1 μg/mL in PBS) overnight in the dark at room temperature. Sections were then given four 10 minute washes in PBS. Labeled sections were mounted in a medium that reduces fading (Citifluor; Agar Scientific, Stansted, UK), and examined by confocal microscopy. 

Northern Blot Analysis

Total RNA was isolated from lens tissue with a kit (High Pure; Roche Diagnostics, Mannheim, Germany) according to standard manufacturer’s protocol. RNA was isolated from three preparations: whole lenses, epithelial cells, and fiber cells. To separate epithelial and fiber cells, lenses were decapsulated with a sharpened pair of forceps. ¹² Because the epithelial cells stay largely attached to the capsule, the lens could be separated into epithelial (material adhering to the capsule) and fiber cell portions which were then processed separately. Total RNA was electrophoresed for approximately 5 hours through a 1%-agarose formaldehyde gel in 3-(N-morpholino)propanesulfonic acid (MOPS) buffer at 4°C. RNA was transferred overnight by capillary action to a nylon membrane. Hybridization was performed overnight at 50°C according to the manufacturer’s instructions (Roche Diagnostics) with 10 ng/mL of digoxigenin random-labeled DNA probes. The sequence-verified cDNAs were obtained with RT-PCR products derived as in Merriman-Smith et al. ² GLUT1-, GLUT3-, and Cx46-specific transcripts were detected with anti-DIG antibodies conjugated to alkaline phosphatase (1:20,000: Roche Diagnostics) in buffer (50 mM maleic acid, 75 mM NaCl pH 7.5) for 30 minutes, followed by chemiluminescence (CDP-star; Roche Diagnostics) and exposure onto autoradiograph film (Hyperfilm ECL; Amersham, Arlington Heights, IL). 

Quantitative PCR

Quantitative PCR real-time analysis was performed with a fluorescein PCR detection system (LightCycler) and DNA master mix (SYBR Green I kit; both from Roche Diagnostics). cDNA was transcribed from total lens RNA extracted from age-matched diabetic and nondiabetic animals with the a cDNA synthesis system (Expand; Roche Diagnostics) according to the manufacturers standard protocol. Before cDNA synthesis, RNA tertiary structures were removed by a 10-minute incubation of the RNA (0.5 μg) and random hexamer (5 pM) at 65°C. Appropriate PCR thermal cycling conditions, such as amplification efficiency, fluorescence acquisition temperature, and optimal cDNA dilution, were determined for all primer sets. Five dilutions from a known concentration of sequenced confirmed cDNA, was prepared for every experiment to give a standard curve allowing for optimal quantification of gene concentration. To compensate for variations in the RT’s efficacy, the target gene was normalized to a reference gene. Primers and cDNA sequences are published in Merriman-Smith et al. ² PCR reactions were performed in 20-μL reaction volumes with final concentrations of 1× DNA master mix (LightCycler-DNA Master SYBR Green I; Roche Diagnostics), 0.5 μM sense primer, 0.5 μM antisense primer, 2 mM MgCl₂, and 2 μL template. Template consisted of either the cDNA samples (3–300 pg), sequenced DNA template (standard dilution curve), or the false amplification control (water only). Before thermocycling, the Taq polymerase was activated by a 30-second incubation at 95°C. Cycling conditions were: for the GLUT1 primer set, melting at 95°C for 0 seconds, annealing at 50°C for 5 seconds, and extension at 72°C for 10 seconds, for 45 cycles; for the GLUT3 primer set melting at 95°C for 0 seconds, annealing at 55°C for 5 seconds, and extension at 72°C for 10 seconds, for 40 cycles; and for the Cx46 primer set, melting at 95°C for 0 seconds, annealing at 55°C for 5 seconds and extension at 72°C for 15 seconds, for 40 cycles. All reactions were performed in duplicate, and a maximum difference of 0.5 cycles between the threshold cycle (C_T; the exponential phase of amplification, at least 10 times above baseline emission) was accepted. Nontemplate controls (no RT and water only) were included for each primer set to measure levels of contaminants. These samples were consistently low, with a difference in C_T of at least 15 to 20 cycles. Fluorescence for the real-time PCR reactions were recorded and analyzed with the software (LightCyler; Roche Diagnostics) supplied with the thermocycler. Transcript concentrations were determined for each of the samples by extrapolation from the standard curve. 

Immunocytochemistry

Control and experimental lenses were fixed in 2% paraformaldehyde in PBS (pH 7.4; osmolality 300 mOsmol/kg) for 4 hours at room temperature. Fixed lenses were sectioned at 180 μm with a vibratome or at 16 μm with a cryostat (CM3050; Leica Lasertechnik, Heidelberg, Germany). Peptide-specific antibodies directed against the cytoplasmic tails of GLUT1 and -3 (Research Diagnostics, Flanders, NJ) were used to label lens sections that were first permeabilized with a 30-minute incubation in 0.1% Triton X-100 (Sigma-Aldrich). After three 15-minute washes in PBS, the lens sections were incubated with either primary antibody diluted to 0.5 μg in PBS for 2 hours at room temperature. For each antibody, appropriate controls were performed that used either no primary antibody or the primary antibody preincubated in the presence of its specific antigenic peptide. After three 5-minute washes in PBS, the sections were incubated with the appropriate secondary antibody for 1 hour. To detect GLUT3 an anti-rabbit IgG rhodamine red (Molecular Probes, Eugene OR) diluted 1:120 was used and an anti-goat FITC (Santa Cruz Biotechnology, Santa Cruz, CA) diluted 1:120 was used to detect anti-GLUT1. When appropriate, tissue architecture was visualized with FITC-conjugated WGA, as described earlier. Sections were washed three times for 5 minutes in PBS before being mounted onto slides and examined by confocal microscopy (model TCS 4D; Leica Lasertechnik). 

Western Blot Analysis

Rat lenses were homogenized in 10 mL of Tris-buffered saline (TBS; 10 mM Tris-HCl [pH 7.4], 5 mM EDTA, and 5 mM EGTA). The homogenate was centrifuged at 12,000g for 15 minutes at 4°C and resuspended in 1 mL of 4 M urea and 5 mM Tris (pH 9.5). This process was performed three times, and the final pellet was resuspended in 100 μL of 5 mM Tris-HCl (pH 8.0), 2 mM EDTA, 2 mM EGTA, and 100 mM NaCl. The concentration of lens protein was determined with the bicinchoninic acid (BCA) protein detection kit (Pierce, Rockford, IL), according to the manufacturer’s protocol. Proteins were separated on a 10% SDS polyacrylamide gel and transferred onto a nitrocellulose membrane by electrophoresis for 90 minutes at 170 mA. Membranes were incubated overnight at room temperature in a blocking solution (1% BSA and 0.1% Tween 20 in 1× TBS (2 mM Tris-HCl, 140 mM NaCl [pH 7.6]) and subsequently incubated for 2 hours in with either rabbit anti-GLUT3 or rabbit anti-Cx46 (Alpha Diagnostic International, San Antonio, TX) antibodies diluted 1:1000 in 1× TBS. Membranes were then exposed to biotinylated anti-rabbit IgG secondary antibody (Amersham Biosciences Corp., Piscataway, NJ) diluted 1:1000 for 1 hour, followed by streptavidin horseradish peroxidase (Amersham Biosciences Corp.) diluted 1:1000. After each incubation, membranes were rinsed three times with water and washed three times for 15 minutes in 1× TBS. The presence of GLUT3 or Cx46 protein was detected by chemiluminescence and exposed on autoradiograph film (Hyperfilm, ECL recycling kit; Amersham Biosciences Corp.). Western blots were stripped with a Western blot recycling kit (Alpha Diagnostic International, San Antonio, TX) according to the manufacturer’s instructions. Band intensities were measured using image processing software (Image ver. 4.1; Scion, Frederick, MD). 

Results

Expression of GLUT1 and -3 in the Normal Adult Rat Lens

To characterize transcript levels in the normal adult rat lens, Northern blot analysis was performed on RNA separately extracted from epithelial cells and fiber cells (Fig. 1A) . GLUT1 transcript was abundantly expressed in the epithelial cell fraction but was not found in the fiber cells (Fig. 1A) . GLUT3, however, was predominantly expressed in the fiber cells with little expression detected in the epithelial cells. The expression of GLUTs in the lens was further investigated at the protein level by immunocytochemistry. In axial sections, GLUT1 labeling was localized to the basolateral membrane of the epithelial cell layer (Fig. 1B) . In contrast, GLUT3 labeling was minimal in the epithelium and predominantly localized to the cortical fiber cells (Fig. 1B) . 

Newly elongated fiber cells near the equator did not appear to express GLUT3, suggesting that the expression of GLUT3 in the fiber cells is a differentiation-dependent phenomenon (Fig. 1B) . To investigate this possibility further, high-resolution mapping of GLUT3 in the lens was conducted in equatorial sections, in which the cell membranes were labeled with WGA-FITC (Fig. 2) . In the overview, GLUT3 initially appeared cytoplasmic before localizing to fiber cell membranes in the deeper lens (Fig. 2A) . Neither the cytoplasmic nor the membrane labeling was observed in sections labeled with GLUT3 antibodies preincubated with antigenic control peptide (Fig. 2A , inset). At higher resolution, GLUT3 protein was detected intracellularly in a cortical zone approximately 100 μm in from the capsule (Fig. 2B) . At a depth of approximately 200 to 400 μm from the capsule, GLUT3 labeling became increasingly associated with the membranes and was confined mostly to the narrow sides of the hexagonal fiber cells (Fig. 2C) . Fiber cells deeper in the lens (400–600 μm) exhibited a more uniform membrane labeling pattern, with both the broad and narrow sides labeled (Fig. 2D) . 

Effects of Elevation of Glucose Concentration on Tissue Architecture

In the rat, long-term diabetes is associated with a reduction in GLUT1 mRNA expression in the kidney ⁸ but with an increased expression of GLUT3 in the placenta ^{9 10} and in the hippocampus. ^{13 14} To assess what effect elevated glucose levels have on the expression of glucose transporters in the lens, two model systems were used. The first involved the chronic elevation of glucose. This was achieved in vivo by a bolus injection of streptozotocin that destroyed the beta islet cells, rendering the injected rats diabetic. ¹⁵ The second model involved an acute in vitro elevation of glucose achieved by culturing lenses in the presence of 50 mM glucose. ¹⁶ Examination of lens opacification and fiber cell architecture by dark-field and confocal microscopy, respectively, revealed that the elevation of glucose in the two models induced essentially identical damage phenotypes but with different time courses. Both models were characterized by cortical opacities caused by a discrete zone of tissue liquefaction. Because the changes in lens morphology responsible for this opacification have been documented in streptozotocin-injected rats ⁷ (and are also evident in Fig. 5 ), herein we show only the changes in lens transparency and fiber cell morphology induced in vitro by culturing lenses in 50 mM glucose (Fig. 3) . Lenses cultured in the presence of normal glucose levels for 8 days were transparent and exhibited a typically well-ordered array of cortical fiber cells (Fig. 3A) . In contrast, lenses exposed to 50 mM glucose showed increasing degrees of opacification (Fig 3 , insets) and fiber cell damage over the 8 days of exposure. After 2 days of exposure, an inner zone of swollen fiber cells that was bounded on both sides by apparently normal fiber cells was already apparent (Fig. 3B) . At 4 days of exposure to high glucose, fiber cell swellings became more abundant (Fig. 3C) until, after 8 days of exposure, the cells burst, creating areas of tissue liquefaction (Fig. 3D) . Thus, it appears that the cortical opacities observed in our in vivo and in vitro models of hyperglycemia were both initiated by fiber cell swellings that originated in a discrete zone, some 100 to 200 μm in from the capsule. 

Changes in GLUT1 and -3 Transcript Levels in Response to Hyperglycemia

Because chronic hyperglycemia is known to affect gene expression in a number of tissues, ^{17 18} we determined whether GLUT1 and -3 transcripts are upregulated in our two lens models of hyperglycemia. Initially, Northern blot analysis was performed with RNA extracted from the whole lenses (epithelium plus fiber cells) obtained from the two models of hyperglycemia and their respective controls. Blots were hybridized with probes specific for GLUT1, GLUT3, and the reference gene Cx46. ¹⁹ Cx46 was chosen as a reference gene because we found that its expression levels were unaffected by elevated glucose (Figs. 4A 4B) . More traditional housekeeping genes often used as markers, such as β-actin and GAPDH, were considered inappropriate for this work, because both had been documented to be affected either directly or indirectly by hyperglycemia in other tissues (John M, et al. IOVS 2001;42:ARVO Abstract 546). ²⁰ No signal was detected for GLUT1 (data not shown). This is not surprising, because GLUT1 transcript was detectable by Northern blot analysis only when RNA was extracted exclusively from epithelial cells collected from over 15 lenses (Fig. 1A) . In contrast, GLUT3 transcript was readily detected in both control and experimental lenses as a prominent 4-kb band by Northern blot analysis, thereby allowing the effects of hyperglycemia on GLUT3 expression to be studied. Although no change in the level of GLUT3 expression occurred in normal lenses as a function of time (Figs. 4A 4B ; top left), GLUT3 expression increased in the second and third weeks after streptozotocin injection (Fig. 4A , top right) or after 2 and 4 days in culture with 50 mM glucose (Fig. 4B , top right). In both systems, this upregulation was followed by a reduction in the GLUT3 transcript levels at 4 weeks after streptozotocin injection or 8 days in culture. In both models of hyperglycemia, no change was detected in the expression levels of our reference gene Cx46. 

Because Northern blot analysis was not a sensitive enough tool to monitor GLUT1 expression levels in our models of hyperglycemia, a quantitative PCR approach was adopted to analyze GLUT expression levels in lenses from streptozotocin-injected rats. To compensate for variations in the amount of starting template, the concentrations of the GLUT transcripts were again standardized to our reference gene, Cx46. Using this approach, no notable up- or downregulation in the expression level of GLUT1 was found in lenses of normal or streptozotocin-injected rats (Fig. 4C) . In contrast, the expression level of GLUT3 in lenses from streptozotocin-injected rats was upregulated by a factor of 3.5 at 3 weeks after injection, before returning to a level similar to that seen in normal rat lenses at 4 weeks after injection. Thus, the results achieved with quantitative PCR indicate that GLUT1 expression in the lens epithelium is not regulated in response to hyperglycemia. Furthermore, the results obtained for GLUT3 by quantitative PCR and those obtained by Northern blot analysis are in remarkable agreement. 

Localization of GLUT3 Protein in the Diabetic Lens

To determine how the increase in GLUT3 protein relates to the tissue damage observed in the hyperglycemic lenses, GLUT3 protein was localized in equatorial sections taken from streptozotocin-injected rats (Fig. 5) . In an equatorial section taken from a saline-injected rat most of the GLUT3 labeling was detected intracellularly, with a small amount of GLUT3 in the membrane (Fig. 5A) . An increase in GLUT3 membrane labeling was detected in a section from a rat 1 week after injection of streptozotocin (Fig. 5B) . At 2 weeks after injection, an increase in both membrane and cytoplasmic GLUT labeling was detected (Fig. 5C) . In weeks 3 and 4 after injection, an increase in intracellular labeling of GLUT3 was detected (Figs. 5D 5E) . Higher-resolution imaging demonstrated that a significant fraction of GLUT3 was localized as intracellular pools of GLUT3 in this damaged area (Fig. 5F) . Thus, the observed increase in GLUT3 membrane labeling at weeks 1 and 2 after injection was localized to the region in the lens where the fiber cells were destined to sustain extensive tissue damage. With continued diabetic insult, GLUT3 tended to accumulate in the cytoplasm, and this accumulation may play a role in exacerbating tissue damage. 

Upregulation of GLUT3 Protein in the Diabetic Lens

Western blot analysis was performed on urea-stripped lens fiber cell membranes to confirm whether the amount of GLUT3 protein inserted into the membrane increases after diabetic insult (Fig. 6) . A progressive increase in a 47-kDa product was detected with the GLUT3 antibody from weeks 1 to 3 after injection of streptozotocin (Fig. 6A) . At 4 weeks after injection, the level of the 47-kDa product decreased, and it was replaced by lower molecular mass bands indicative of GLUT3 protein degradation. ^{21 22} Such degradation is a common feature in cataractogenesis, because cellular breakdown activates calcium-dependent proteases in the lens. The blot was stripped, and Western blot analysis was performed with a Cx46 antibody to detect any difference in protein loading (Fig. 6B) . No difference was obvious, as shown by the line plot. To investigate further the upregulation of GLUT3, the ratio of GLUT3 to Cx46 band intensities was normalized and plotted against weeks elapsed since injection (Fig. 6C) . A significant increase in GLUT3 expression was detected that peaked at 3 weeks before declining, presumably because of increased protein degradation. 

Discussion

Glucose provides the energy requirements for lens clarity and homeostasis, yet excess glucose or, hyperglycemia can lead to lens opacification. Because the uptake of glucose is mediated by members of the facilitative transporter family (GLUT1 and -3), we investigated how these transporters respond to hyperglycemic insults that ultimately lead to cataract. To induce the hyperglycemic insult we compared in vivo and in vitro models of sugar cataract. Not only did we find differential expression of these transporters in the normal lens, we also found that hyperglycemia differentially upregulated the fiber cell specific GLUT3 in the cortical zone of tissue damage. The implications of our findings in the normal and hyperglycemic lens are discussed. 

Glucose Transporters in the Normal Lens

We have shown in a previous study in neonatal rat lenses that two members of the glucose transporter family, GLUT1 and -3, are differentially expressed in the epithelium and fiber cells, respectively. ² In other tissues, the relative expression levels of these two isoforms appear to change as a function of development and growth. ²³ Therefore, we first investigated the expression patterns of GLUT1 and -3 in the normal adult rat lens. Although no differences in the distribution patterns of the two isoforms were observed in the adult lens relative to the neonatal lens, the use of a superior sectioning protocol ²⁴ allowed the distribution of GLUT3 to be more extensively investigated. We found that subcellular GLUT3 labeling changed as a function of fiber cell differentiation. Initially, GLUT3 labeling was cytoplasmic, but as the fiber cells elongated labeling became increasingly associated with the membrane, indicating that GLUT3 is inserted into the membrane from a cytoplasmic pool. The cytoplasmic staining observed in this study has also been reported in other tissues. ²⁵ In these tissues, GLUT3 is thought to cycle between the plasma membrane and intracellular pool through clathrin-mediated pathways, ²⁶ through a translocation process similar to that of the related isoform GLUT4. ^{26 27} In the lens, clathrin-coated pits have been observed, ²⁸ but whether they are involved in the insertion of GLUT3 remains to be determined. 

A similar differentiation-dependent insertion into the membrane from a cytoplasmic pool of protein has been observed for the major fiber cell membrane protein MP20. ²⁹ For MP20, membrane insertion occurred some 400 μm from the capsule, whereas GLUT3 insertion occurred earlier at approximately 200 μm. In both cases it appears that the fiber cells produce membrane proteins while they still have intact protein synthesis capacity. These proteins are stored in the cytoplasm until they receive the appropriate message to signal trafficking to the membrane. Once in the membrane, subtle differences in the subcellular distribution of GLUT3 occurred as a function of fiber cell differentiation. Initially, GLUT3 was found in the narrow sides of the hexagonal fiber cells, but with increasing depth into the lens, the protein became more uniformly dispersed around the entire cell membrane. We have observed similar changes in the subcellular distribution of gap junctions as a function of fiber cell differentiation. ^{24 30} However, in contrast to GLUT3, the gap junctions were initially restricted, not to the narrow but to the broad sides of the fiber cells before becoming dispersed around the entire cell membrane. This implies that some form of adhesion protein or cytoskeletal anchor, responsible for maintaining subcellular domains, is lost during the course of fiber cell differentiation, thereby promoting membrane protein dispersion. 

Glucose Transporters in the Diabetic Lens

In our study, elevated glucose in the lens, in both models of hyperglycemia, was associated with complex changes in GLUT3 mRNA levels that differed between the acute and chronic phases of the disease. Because GLUT1 mRNA remained unchanged in both the saline and streptozotocin-injected rats, upregulation of the transcript for this glucose transporter cannot be implicated in the progression of the disease in the lens. However, the 3.5-fold increase in GLUT3 mRNA in the later weeks of insult and the localized increase in GLUT3 protein suggests that GLUT3 is responsible for the enhanced glucose uptake observed in diabetic rats. ⁶  

A focal increase in GLUT3 protein was localized to the zone of damage induced by hyperglycemia. In the initial week after insult, a significant increase in membrane labeling was detected. This is important, because models of osmotic cataract induced by hyperglycemia are believed to involve the accumulation of glucose and its conversion to the osmolyte sorbitol. The amount of GLUT3 protein increased in the membrane before an increase in GLUT3 mRNA was detected, supporting the idea that GLUT3 is inserted into the membrane from a cytoplasmic pool. In the later weeks of diabetic insult, there appeared to be a massive increase in the amount of intracellular GLUT3. This increase could be due to increased translation of GLUT3 protein, inhibition of the insertion of GLUT3 into the membrane, or an increase in the removal of GLUT3 from the membrane adding to the cytoplasmic pool. Regardless of the actual mechanisms, our data suggest that in the lens GLUT3 recycling occurs and furthermore that the transcript and protein levels are independently regulated, a phenomena found for other tissues. ^{31 32}  

The mechanisms involved in the regulation of GLUT3 transcription in the lens have not been addressed in the present study. However, others have found that glucose itself can activate signaling systems that alter physiological and pathologic processes ³³ and is associated with upregulation of genes involved in transcriptional regulation, such as c-fos, c-jun, Sp1, Sp3, and Oct-3. ^{33 34} In the lens the manipulation of these transcription factors affects aspects of cell proliferation, differentiation, conformation, and viability, ³⁵ all of which affect normal lens function. The overexpression of mitogen-activated protein kinase kinase (MEK), an upstream kinase in the extracellular signal-regulated kinase (ERK)-1 and -2 signaling pathway, in the fiber cells of the mouse lens leads to an increase in glucose uptake ³⁶ and is later associated with cortical damage, similar to but not identical with that in the diabetic rat lens. ⁷ The elevated intracellular glucose levels observed in these mouse lenses were not attributed to GLUT3, but to the inappropriate additional expression of GLUT1 in the fiber cells. 

In summary, these results explain the increase in glucose-derived metabolites during diabetes and show that the glucose transport system is dynamic. GLUT3 is locally upregulated in the diabetic lens, accounting for the high lenticular glucose levels in diabetic animals. An increase in GLUT3 protein observed in a region overlapping the cortical zone of disrupted cell structure may contribute to the osmotic damage in the cortex. Hence, the localization of GLUT3 to the swollen fiber cell membranes suggests that GLUT3 mediates the observed increased uptake of glucose and is therefore a potential target for anti-cataract therapies. 

 Supported by the Health Research Council of New Zealand, the Lottery Grants Board, and the University of Auckland Research Committee.

 Submitted for publication December 2, 2002; revised March 12 and April 2, 2003; accepted April 3, 2003.

 Disclosure: B.R. Merriman-Smith, None; A. Krushinsky, None; J. Kistler, None; P.J. Donaldson, 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.

 

Differential distribution of GLUT1 and -3 in the adult rat lens. (A) Northern blot of total RNA extracted from the lens epithelium and fiber cells hybridized with GLUT1 (left) and GLUT3 (right). (B) Axial section from a lens showing GLUT1 labeling (red) is predominantly localized to the basolateral membrane of the epithelium (arrow), whereas the GLUT3 protein (green) is localized to the fiber cells. Cp, capsule; F, fiber; E, epithelium.

Figure 2.

 

The distribution of GLUT3 as a function of fiber cell differentiation. An equatorial section double labeled with GLUT3 (green) and the general membrane label WGA (red). (A) Low-power overview image of GLUT3 labeling in the lens; inset: shows no labeling when the GLUT3 antibody was preabsorbed using its antigenic peptide. (B–D) Representative high-power images taken from the area indicated in (A); (B) 50 μm from the capsule; (C) 250 μm from the capsule; (D) 600 μm the capsule. Cp, capsule; Nu, fiber cell nuclei.

Figure 5.

 

Distribution of GLUT3 protein in the lenses of streptozotocin-injected rats. Equatorial sections double labeled with the GLUT3 antibody (green) and the membrane stain WGA (red) from lenses of rats injected with either saline (A) or streptozotocin (B–F). (A) One week after saline injection and (B) 1 (C), 2 (D), 3 (E), and 4 weeks after injection with streptozotocin. (F) High-magnification image from the region indicated in (E) with the symbol ⋆. Punctate intracellular GLUT3 labeling. Cp, capsule.

Figure 3.

 

Effects of hyperglycemia on lens transparency and tissue morphology. Confocal micrographs of FITC-conjugated WGA-labeled equatorial sections showing tissue morphology of lenses cultured for 8 days in 5 mM glucose (A) and for 2 (B), 4 (C), and 8 (D) days in the presence of 50 mM glucose. Dark-field images (insets) showing lens transparency are displayed for each time point. Cp, capsule.

Figure 4.

 

Transcript levels in response to hyperglycemia. (A) Northern blot analysis of total RNA extracted from six whole lenses per time point taken from saline-injected (left) and streptozotocin-injected (right) rats hybridized with GLUT3 (top) and Cx46 (bottom) probes. The number of weeks after injection is indicated. (B) Northern blot analysis of total RNA extracted from two whole lenses per time point taken from lenses cultured in (left) 5 and (right) 50 mM glucose hybridized with GLUT3 (top) and Cx46 (bottom) probes. The length of time in culture is indicated. (C) Relative proportions of GLUT1 and -3 transcripts quantified by real-time PCR for saline- and streptozotocin-injected rats over the course of 4 weeks. Transcript levels were normalized to Cx46 levels. The data at each time point represents the mean and SEM of results in three separate experiments. In each experimental group and time point, RNA was pooled from four lenses extracted from four different animals. For each experiment reactions were run in duplicate and repeated three times. (– – –) hyperglycemic; (—) normal; (•, ▾) GLUT1/Cx46, (▪, ▴) GLUT3/Cx46.

Figure 6.

 

Upregulation of GLUT3 in lens membranes. Lens membranes were purified from streptozotocin-injected rats 1 to 4 weeks after injection and probed with antibodies against GLUT3 and Cx46. (A) GLUT3 was detected as a 47-kDa band (top). A line scan of pixel intensity (bottom) highlights that GLUT3 expression initially increased before declining, presumably because of degradation. (B) The blot in (A) was striped and reprobed with Cx46 antibodies (top). A line scan through the 46-kDa band indicates that equal amounts of Cx46 protein were loaded (bottom). (C) Plot showing the ratio of GLUT3/Cx46 expression as a function of weeks after injection. Values were normalized to the GLUT3/Cx46 ratio at 0 weeks after injection. This ratio was calculated from other Western blots (data not shown) loaded with a higher concentration of lens membrane proteins to facilitate detection of GLUT3.

The authors thank Colin Green for his tuition in the use of the confocal microscope, and to Roche Diagnostics, especially John McKay and Alison Weaver, for supplying the use of the LightCycler and for technical instructions.