Cellular Localization of Glutamate and Glutamine Metabolism and Transport Pathways in the Rat Ciliary Epithelium

Abstract

Purpose.: To investigate how glutamate and glutamine levels are established in the aqueous humor by identifying the transporters and metabolism pathways that contribute to the differential accumulation of glutamate and glutamine between the distinct epithelial cell layers that constitute the ciliary body.

Methods.: Postembedding immunohistochemistry and silver intensification were used to quantify the relative distributions of glutamate, glutamine, and related amino acids (aspartate, alanine, GABA, and glycine) in the pigmented (PE) and nonpigmented (NPE) epithelial cells of the ciliary body. Fluorescent immunocytochemistry was used to localize Na⁺-dependent glutamate transporters (EAAT1–5), glutamine transporters (LAT1, LAT2, and b^0,+AT), and the enzyme glutamine synthetase (GS) in the ciliary epithelium. Intravitreal injection of the GS inhibitor methionine sulfoximine (MSO) or the EAAT functional probe D-aspartate was used to modulate GS activity and indirectly monitor glutamate uptake from the aqueous, respectively.

Results.: Although glutamate, glutamine, and alanine were preferentially accumulated in NPE relative to PE cells, no such differential distribution of aspartate, GABA, or glycine was observed. This differential distribution of amino acids was abolished by a single injection of MSO that caused a decrease in glutamine and an increase in glutamate levels in NPE compared with PE cells. This amino acid distribution plus an observed strong labeling of EAAT3 in the interface between the PE and the NPE cell layers indicate that EAAT3 mediates the uptake of glutamate from the blood. Weaker EAAT3 labeling of the basolateral membranes of NPE cells, coupled with the accumulation of injected D-aspartate by the ciliary epithelium, indicates that NPE cells also mediate glutamate uptake directly from the aqueous. In contrast, the basolateral localization of LAT1 and b^0,+AT in NPE cells suggest that these transporters may mediate glutamine efflux from the NPE cells into the aqueous.

Conclusions.: The basolateral membrane localization of EAAT3 and LAT1/b^0,+AT in NPE cells indicates that the low glutamate and high glutamine levels observed in the aqueous are determined by glutamate uptake and glutamine efflux, respectively. Furthermore, the concentration gradient for glutamine efflux appears to be generated by the active accumulation of glutamate by EAAT3, located in the apical membrane of NPE cells and the subsequent conversion of the accumulated glutamate to glutamine by GS in NPE cells. This suggests that in contrast to fluid transport, which uses both the PE and the NPE cell layers, the transepithelial transport of glutamine occurs primarily in the NPE cell layer.

A major function of the ciliary body is to secrete the aqueous humor, a clear fluid that inflates the globe and nourishes the avascular tissues that make up the anterior segment of the eye. ^1,2 Evidence has accumulated to suggest that the ciliary body also acts as an endocrine system that secretes a variety of peptides that may modulate aqueous humor dynamics. ^3,4 The secretory activity of the ciliary body is accomplished by a double-layered epithelium that outlines the ciliary processes. This tissue consists of a nonpigmented epithelial (NPE) cell layer that faces the aqueous humor and a pigmented epithelial (PE) cell layer that interfaces with the vascular stroma. ⁵ Tight junctions located between the NPE cells form the blood-aqueous barrier that restricts the passage of large molecules into the eye. ⁶ Because the PE and NPE cells are coupled by gap junctions, ^7,8 it is believed that the PE-NPE couplet represents the functional secretory unit that drives transepithelial transport of ions and water. Given the fact that reducing the secretory rate of aqueous humor is one of the major strategies to lower intraocular pressure to protect against the progression of glaucoma, ⁹ the transport of ions and water by the functionally coupled epithelial cell layers has been extensively studied. Although these studies have revealed that different transport proteins are used in different species, it is largely agreed that aqueous humor production by the ciliary epithelium is driven by a net Cl⁻ secretion. ^10,11 This movement of Cl⁻ occurs in three steps: the uptake of Cl⁻ from the stroma by PE cells expressing a variety of species-specific electroneutral Cl⁻ transporters, diffusion through gap junctions into the NPE cells, and Cl⁻ exit from the NPE cell into the aqueous by way of Cl⁻ channels. 

In contrast to ion and water transport, far less is known about the transport of nutrients such as amino acids by the ciliary epithelium. Using traditional biochemical methods, it has been shown that many amino acids are more abundant in the ciliary body than the stroma. ^12,13 Studies that used an isolated ciliary body-iris preparation established that amino acids are actively accumulated by the ciliary epithelium. ^{14 –16} In addition, by systemically administering non-metabolizable amino acid tracers and monitoring their appearance in the aqueous humor, it was shown that active secretion of amino acids from the ciliary body into the aqueous also occurs. ^17,18 Three transport systems were thought to be involved, one each for basic, acid, and neutral amino acid. ^17,19 However, the subsequent molecular identification of these specific transporters has received only limited attention. The GABA transporter (GAT2) was localized to the NPE cells in the rat, ^20,21 whereas a taurine transporter was found in the mouse NPE cells by in situ hybridization. ²² Although not molecularly localized, a glutamate transporter was detected in the rabbit ciliary epithelium using D-aspartate uptake as a functional probe of glutamate transporter activity. ²³ PCR screening of a human ciliary body library indicates that the excitatory amino acid transporter (EAAT) isoforms 1 and 2 and the vesicular glutamate transporter 1 are also expressed. ³  

More recently, Marc and Cameron ²⁴ have used a sensitive postembedding immunohistochemical technique to map the relative abundance of amino acids in the zebrafish ciliary epithelium. This technique has the advantage of quantifying amino acid levels in the two cell layers, allowing amino acid distributions to be correlated with amino acid transporter expression enabling the relative contribution of PE and NPE cell layers to the transepithelial transport of amino acids to be determined. Their results showed a preferential accumulation of glutamine in the NPE cells relative to the PE cells in the zebrafish. ²⁴ Interestingly, the NPE layer of the rat ciliary epithelium is known to exclusively express the enzyme glutamine synthetase (GS), which catalyzes the conversion of glutamate to glutamine. ²⁵ Most recently, Lanford et al. ²⁶ have studied the distribution of glutamate, the glutamate transporters (EAAT1), and GS in the human and monkey ciliary body and found them to be concentrated in NPE cells. Taken together, these observations obtained in different species show that the molecular machinery for the uptake and metabolism of glutamate and glutamine is concentrated in the NPE cell layer and suggests that the PE-NPE cell couplet may not be required for amino acid transport across the ciliary epithelium. 

To investigate this further, we used postembedding immunohistochemistry to quantify amino acid levels in the presence and absence of the GS inhibitor methionine sulfoximine (MSO), coupled with immunolocalization of glutamate and glutamine transporters to investigate the relative role played by the PE and NPE cell layers in transepithelial transport and metabolism of glutamate/glutamine in the rat ciliary epithelium. We show that the distribution of glutamate and glutamine was higher in the NPE cell layer than in the PE cell layer, and this asymmetric distribution of the two amino acids could be abolished by inhibiting GS activity. Although the sodium-dependent glutamate transporter EAAT3 was localized predominantly to the PE-NPE interface, the facilitative glutamine transporters LAT1 and b^0,+AT were both found to be expressed on the basolateral membrane of NPE cells. Taken together, our data suggest that unlike ion and fluid transport, in which the functional unit is the PE-NPE cell couplet, NPE cells appear to be the principal site of glutamate uptake and glutamine efflux in the ciliary epithelium. 

Materials and Methods

Tissue Collection and Preparation

All animals were treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Albino Sprague-Dawley rats were used to avoid the possibility of high melanin levels in the PE cells affecting quantification of amino acid signal levels. ^27,28 Rats (P26-P28) were deeply anesthetized by intramuscular injection of 75 mg/kg ketamine (Parnell Laboratories, Auckland, New Zealand) and 10 mg/kg xylazine (Troy Laboratories, Smithfield, New South Wales, Australia). After tissue collection, animals were killed by an overdose of anesthetic injected into the heart. For fluorescence immunocytochemistry, whole eyes were rapidly excised, and several cuts were introduced in the cornea and a small part of the retina close to the optic nerve to increase fixative penetration. After 24 hours of fixation in a mixture of 0.75% paraformaldehyde and 0.01% glutaraldehyde (pH 7.4), whole eyes were cryoprotected and cryosectioned (temperature, −18°C; thickness, 16 μm). For postembedding immunocytochemistry, whole eyes were fixed in a mixture of 1% paraformaldehyde and 2.5% glutaraldehyde (pH 7.4) for 1 hour. Further dissection was performed to isolate the retina and ciliary processes. Next, tissues were dehydrated through a graded series of cold methanol/distilled water (80%, 90%, 100% vol/vol) and embedded in epoxy resin (ProSciTech, Queensland, Australia). Semithin (500 nm) sections were obtained using an ultramicrotome (Ultrascut S; Reichert-Jung, Wetzlar, Germany) and mounted onto 12-well coated slides (Teflon; Cell-line, Newfield, NJ). 

Fluorescence Immunocytochemistry

Cryosections were incubated in blocking solution (6% normal goat serum, 1% bovine serum albumin, 0.5% Triton-X, 0.05% thimerosal in phosphate-buffered saline) for 1 hour and then incubated overnight in the presence of the following primary antibodies: EAAT1 (1:100; Alpha Diagnostic, San Antonio, TX), EAAT2 (1:100; Alpha Diagnostic), EAAT3 (1:200; Alpha Diagnostic), EAAT4 (1:500; Alpha Diagnostic), EAAT5 (1:100; Alpha Diagnostic), LAT1 (1:10, CosmoBio, Carlsbad, CA), LAT2 (1:5; Sigma Aldrich, St. Louis, MO), b^0,+AT (1:10, CosmoBio), d-aspartate (1:5000; kind gift by Robert E. Marc, University of Utah, Salt Lake City, UT), GS (1:2000; BD Biosciences, Franklin Lakes, NJ), Jacalin-FITC (1:100; EY Laboratories, San Mateo, CA). To visualize immunolabeling, tissues were incubated in the dark with secondary goat anti-rabbit Alexa 488 (1:400; Molecular Probes, Eugene, OR) or goat anti-mouse Alexa 594 (1:400; Molecular Probes) for 4 hours and washed, and coverslips were mounted using an anti-fading medium (AF1; Citifluor, Leicester, UK). 

Postembedding Immunohistochemistry

The postembedding immunohistochemical procedure used here has been extensively described. ^29,30 Briefly, resin sections were etched in a 1:5 solution of sodium ethoxide/ethanol and rehydrated through a graded series of methanol/distilled water (100%, 60%, 30% vol/vol). Subsequently, tissues were placed in 1% sodium borohydride for 30 minutes, washed, and blocked for 1 hour (6% goat serum in PBS) before overnight incubation with primary rabbit polyclonal antibodies (kind gifts by Robert E. Marc, University of Utah; all are commercially available through Signature Immunologics, Salt Lake City, UT): glutamate (1:5000), glutamine (1:5000), aspartate (1:10,000), alanine (1:1000), GABA (1:5000), and glycine (1:5000). The amino acid antibodies were tested for cross-reactivity using dot immunoassays to confirm specificity to the relevant amino acid–coupled antigen (see also antibody data sheets from the commercial sources; Signature Immunologics at http://www.immunologics.com/products.html). The primary IgGs were detected with secondary goat anti-rabbit IgG coated with a 1 nm gold particle (1:100; British BioCell International, Cardiff, UK) and visualized by the reduction of silver ions (1% silver nitrate) onto the gold particles by hydroquinone in citrate buffer. ^29,30  

Intravitreal Injection Experiments

To inhibit glutamine synthetase activity, MSO (ICN Biomedicals Inc., Aurora, OH) was injected into the vitreous of 12 anesthetized Sprague-Dawley rats (P26-P28). MSO irreversibly inhibits glutamine synthetase by binding to the active site of glutamate synthetase as methionine sulfoximide phosphate. ^{31 –33} A 2-μL injection of MSO (150 mM) was delivered into the vitreous, which gave a final concentration of 30 mM based on an approximate vitreous volume of 7.9 μL in 4-week-old rats. ³⁴ Six rats were kept for 12 hours and six were kept for 24 hours before tissue collection, with normal rat chow and water available ad libitum. In another six Sprague-Dawley rats (P26–28), the functionality of EAATs was determined secondary to intravitreal injection of d-aspartate (12.5 mM final vitreal concentration). Tissue was harvested after 12 minutes. For both experiments, control saline injection (0.9% NaCl, pH 7.4) was performed in the contralateral eye. Eyes were examined after injection and before dissection using an ophthalmoscope. If the lens was damaged, the data from these animals were not used. 

Image Capture and Quantitative Analysis

Fluorescent labeling was visualized using a confocal laser scanning microscope (SP2; Leica, Wetzlar, Germany) fitted with an argon/krypton mixed ion laser and appropriate filter sets. For postembedding silver immunocytochemistry, antibody labeling was viewed on a Leica DMR light microscope (Leica Microsystems Ltd., Heidelburg, Germany) and was photographed under constant light with a fixed camera gain and gamma using an attached digital camera (DS-5Mc; Nikon, Tokyo, Japan). This gives a log-linear pixel value scale over more than a 2-log unit range, where each pixel value equals 9.407 × 10⁻³ log units. ^29,35 Marc et al. ^29,35 provided the theoretical framework and experimental evidence outlining the conversion of pixel value to millimolar amino acid concentration. We also calculated the change in immunoreactivity (Δ immunoreactivity) in ciliary epithelium secondary to MSO inhibition. We adopted the method outlined by Napper et al. ³⁶ to depict the fold increase or decrease in immunoreactivity. 

To ensure appropriate quantification, minimal image processing was performed (Photoshop, version 8; Adobe Systems Inc., Mountain View, CA). Images were inverted with a logical NOT operation before measurement, such that increasing pixel intensity was indicative of higher immunoreactivity. ^29,35 Five measurements were obtained from each ciliary process, and approximately three to five processes were studied from each sample tissue to calculate the average pixel value. To control for potential variation in camera exposures, a zero value for pixel density was assigned to an area of resin adjacent to the section and was deducted from the average pixel value obtained. The final mean pixel value was calculated from six independent experiments. Difference in pixel value between the control and MSO-treated eyes was expressed as the mean (± SEM). Statistical significance was determined by performing Student's t-test (two-tailed), with P < 0.01 considered statistically significant. 

Results

Cellular Distribution of Glutamate and Related Amino Acids in the Rat Ciliary Epithelium

The postembedding immunohistochemical technique provides the ability to quantify the distribution of a range of small metabolites with single-cell resolution. ^29,35 This methodology is ideally suited to determining the relative distribution of glutamate and its related amino acids between the PE and NPE cell layers. Using this approach we examined the distribution of glutamate, glutamine, and GABA because they are both immediate precursors or metabolites of glutamate (Fig. 1). Similarly, alanine and aspartate distributions were examined because they are linked to glutamate through transamination (Fig. 1). The distribution of glycine was also studied as a control because it is not directly associated with glutamate metabolism. 

Figure 1.

 

Schematic of the glutamate metabolism pathways. Conversion of glutamate (Glu) to glutamine (Gln) is catalyzed by the enzyme GS. GABA is also derived from glutamate by the enzyme glutamic acid decarboxylase (GAD₆₅ and GAD₆₇). Transamination reaction between glutamate and alanine (Ala) or aspartate (Asp) is catalyzed by alanine aminotransferase (ALT) and aspartate aminotransferase (AAT), respectively. Glutamate dehydrogenase also catalyzes the interconversion of α-ketoglutarate (α-KG) and glutamate, thereby providing multiple pathways for the glutamate carbon skeleton to enter the tricarboxylic acid cycle.

Labeling for all these amino acids was evident in the three distinct regions of the ciliary body process: the inner stroma filled with blood vessels, the PE layer that surrounds the stroma, and the NPE layer that faces the aqueous humor (Fig. 2). However, the labeling intensity for glutamate (Fig. 2A), glutamine (Fig. 2B), and alanine (Fig. 2C) was considerably stronger than that observed for aspartate (Fig. 2D), GABA (Fig. 2E), or glycine (Fig. 2F). Furthermore, in the case of glutamate (Fig. 2A) and glutamine (Fig. 2B), a difference in the labeling intensity between the two layers of the ciliary epithelium was evident, with labeling stronger in the NPE cells than in the PE cells by a factor of 2.011 and 2.056 for glutamate and glutamine, respectively. It is intriguing to find higher levels of glutamate and related amino acids in the NPE layer relative to the PE layer, especially because the NPE and PE cells layers are known to be coupled by an extensive network of gap junctions. ^7,8 This specific asymmetry in the metabolically linked amino acids suggests that unique metabolic or uptake pathways for glutamate and glutamine may exist in the NPE cells. 

Figure 2.

 

Amino acid distributions in the rat ciliary epithelium. Images (left) and associated intensity plots (right) of the relative distributions of amino acids in the normal ciliary epithelium revealed by postembedding immunocytochemistry using antibodies against glutamate (A), glutamine (B), alanine (C), aspartate (D), GABA (E), and glycine (F). Amino acid labeling intensity is plotted as the log relative concentration (mM) ± SEM deduced from the pixel value of the labeling intensity in the NPE and pigmented PE epithelial cell layers. S, stroma. Scale bars, 50 μm.

Glutamine Synthetase and Amino Acid Metabolism in the Ciliary Epithelium

With regard to metabolic pathways, previous studies have localized GS, the enzyme responsible for the conversion of glutamate to glutamine, specifically in the NPE cells of the rat ciliary epithelium. ^25,26 In this present study, a similar GS labeling pattern was observed that consisted of strong labeling in the NPE cells and the vascular endothelium of the stroma, and an absence of labeling in the PE cell layer produced a clear immunonegative gap between the NPE layer and the stroma (Fig. 3A). To determine the functional significance of GS localization in the NPE cell layer, we injected the GS inhibitor MSO into the vitreous. MSO irreversibly inhibits GS activity by binding to the active site of the enzyme as methionine sulfoximide phosphate, leading to protein degradation. ^{31 –33} Twelve hours after the initial MSO injection, GS immunolabeling in the NPE cells was almost completely abolished, an observation consistent with its degradation (Fig. 3B). Twenty-four hours after MSO injection, immunoreactivity for GS protein reappeared in the NPE cells, suggesting that the new GS protein had been synthesized (Fig. 3C). This inhibition of GS had reciprocal effects on glutamate and glutamine labeling intensity in the ciliary epithelium (Figs. 4, 5). Glutamate levels were increased in both NPE and PE cells (Fig. 4A). However, the change in immunoreactivity was greater in the PE cells than in the NPE cells (Fig. 5A), resulting in a relatively uniform glutamate labeling in the two cell layers not seen before GS inhibition (Fig. 5B). 

Figure 3.

 

Localization of GS in the rat ciliary epithelium. Confocal images of GS labeling in the absence (A) and presence (B, C) of the GS inhibitor MSO. (A) In the absence of MSO, GS labeling was localized solely in the NPE cells, although some labeling in the stroma (S) was also observed. (B) Twelve hours after the intravitreal injection of MSO, though GS immunolabeling was not detected in the non-pigmented epithelial cells, GS labeling in the stroma was still visible. (C) Twenty-four hours after MSO injection, faint labeling of GS was again evident in the NPE cells. Scale bars, 50 μm.

Figure 4.

 

Amino acid distributions in the rat ciliary epithelium after the inhibition of GS. Images (left) and associated intensity plots (right) of the relative distributions of glutamate (A), glutamine (B), alanine (C), aspartate (D), GABA (E), and glycine (F) in the ciliary epithelium 12 hours after intravitreal injection of the GS inhibitor MSO revealed by postembedding immunocytochemistry. Amino acid labeling intensity is plotted as the log relative concentration (mM) ± SEM deduced from the pixel value of the labeling intensity in the NPE and PE epithelial cell layers. S, stroma. Scale bars, 50 μm.

Figure 5.

 

Inhibiting GS changes the relative distribution of amino acids in the rat ciliary epithelium. (A) Plot showing the mean change in labeling intensity observed in NPE and PE cells caused by inhibiting GS activity for 12 hours with MSO. This analysis indicated that GS inhibition significantly increased glutamate levels in both the NPE and PE cell layers but decreased levels of glutamine and alanine. (B) Plot showing the ratio between NPE and PE cells for amino acids in control versus MSO-treated eyes. A ratio close to 1 indicated amino acid levels were comparable between NPE and PE cells. MSO injection tended to abolish the asymmetry in amino acid distribution. *P < 0.01, two-tailed t-test.

In contrast, MSO inhibition of GS significantly diminished glutamine labeling intensity in the NPE cells but had only a minor, albeit significant, effect on glutamine levels in PE cells (Figs. 4B, 5A). The level of glutamine levels in the two cell layers was similar after GS inhibition (Fig. 5B). In addition, significant changes in labeling intensity were detected for the glutamate-related amino acid alanine (Figs. 4C, 5) but not for aspartate (Fig. 4D) or GABA (Fig. 4E). Small changes in glycine immunoreactivity between the two cell layers were evident, but they were not significantly different (Fig. 5). Thus the reciprocal changes in glutamate and glutamine levels observed after MSO injection suggested that the ciliary epithelium actively accumulated glutamate, which was subsequently converted to glutamine by the GS specifically expressed in the NPE cell layer. 

Glutamate Uptake in NPE Cells

To determine the site of this glutamate uptake, isoform-specific antibodies raised against the EAAT family of Na⁺-dependent glutamate transporters ^{37 –42} were used to determine which of the five known EAAT isoforms are expressed in the rat ciliary epithelium. The panel of commercial antibodies used in this study were first tested in the retina and yielded the expected labeling patterns for EAAT1 to EAAT4. ^{43 –46} EAAT5 has also been localized in the retina, ⁴⁷ but in our hands the commercially available antibody did not yield specific labeling in either the retina or the ciliary epithelium (data not shown). Although our panel of antibodies detected EAAT1 to EAAT4 in the retina (data not shown), only EAAT3 labeling was detected in the ciliary epithelium (Fig. 6). EAAT3 labeling in the double-layered epithelium was strongest at the apical-apical junction between NPE and PE cells (Fig. 6A). This labeling was preferentially associated with the NPE cells because there was also a punctate labeling of NPE cells, which, at high magnification, vaguely outlined the shape of NPE cells (Fig. 6A, inset). In an attempt to further characterize EAAT3 labeling in NPE cells, ciliary processes were double-labeled with EAAT3 and Jacalin, a lectin marker that preferentially labels NPE cells. ⁴⁸ Jacalin labeling outlined the membrane of NPE cells and colocalized with EAAT3 at the apical interface between the NPE and PE cells (Fig. 6B). This colabeling with Jacalin indicated that EAAT3 associated with both the apical and the basolateral membranes of NPE cells. EAAT3 labeling appeared to be specific because preabsorption of the primary antibody with its corresponding antigenic peptides showed no significant labeling (Fig. 6C). 

Figure 6.

 

Localization of glutamate uptake pathways in the rat ciliary epithelium. Confocal images showing the cellular localization of the glutamate transporter EAAT3 in the ciliary epithelium. (A) Intense EAAT3 labeling was strongly localized at the apical interface between the NPE and PE cells. (inset) High-magnification image showing EAAT3 labeling can also be found on the basolateral membranes of NPE cells. (B) Double labeling of the ciliary epithelium with Jacalin, a lectin that specifically labels NPE cells, and EAAT3 highlights the predominant location of EAAT3 at the apical-apical interface between the NPE and PE cells. (inset) High-magnification image. (C) Preabsorption of the EAAT antibody with its corresponding antigenic peptide abolishes all EAAT3 labeling in the NPE cells, indicating the high specificity of the antibody labeling. A weak signal, however, was observed in the stroma (S). Scale bar, 50 μm.

Because the apical and basolateral membrane domains of the NPE cells are physically separated by the tight junctions that form the blood-aqueous barrier, this localization suggests that EAAT3 can mediate glutamate uptake from both the blood and the aqueous. To confirm this assignment of EAAT3 to the basolateral membrane of NPE cells and, hence, the functional ability of EAAT3 to accumulate glutamate from the aqueous humor, we injected d-aspartate into the vitreous to determine glutamate transporter activity. d-Aspartate is an exogenous, non-metabolizable amino acid that is accumulated by cells that express functional EAAT transporters, and its presence can be detected by immunocytochemistry using a specific antibody. ³⁵ d-Aspartate has been used previously as a reporter of glutamate uptake in specific cell types in the retina, ^27,35,44,49 and similar results were obtained in the present study in control experiments performed on the rat retina (Fig. 7A). In the ciliary epithelium, tissue collected for immunohistochemistry approximately 12 minutes after an initial injection of d-aspartate showed strong uptake of d-aspartate in the NPE cell layer with a slightly lower accumulation in PE cells (Fig. 7B). The absence of d-aspartate labeling in the stroma indicates not only that d-aspartate is not normally present in the ciliary body but also that the blood-aqueous barrier was not affected by the injection procedure. The ability of NPE cells to uptake d-aspartate introduced by intravitreal injection indicates that the basolateral membrane contains functionally active EAAT3 transporters that are capable of mediating glutamate uptake from the aqueous humor. 

Figure 7.

 

EAAT3 transporters expressed on the basolateral membranes of NPE cells are functional. Images showing the localization of d-aspartate, a functional probe for glutamate transporters in the control tissue retina (A), and the ciliary epithelium (B) 12 minutes after intravitreal injection of d-aspartate. (A) Retinal labeling was confined to the somata of Müller cells (arrows) and their processes traversing the outer and inner plexiform layers. Labeling was strongest in the inner limiting membrane, suggesting that Müller cells are the initial site of d-aspartate uptake. (B) d-Aspartate labeling in the ciliary epithelium was colocalized in both cell layers of the ciliary epithelium. However, labeling intensity was slightly higher in the NPE cells, indicating that the NPE cells are the initial site of d-aspartate uptake. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bars, 50 μm.

Glutamine Efflux in NPE Cells

Having shown that NPE cells are capable of glutamate uptake, we next wanted to determine whether the ciliary epithelium expresses transporters known to mediate glutamine efflux. In other tissues such as the kidney and the small intestine, facilitative glutamine transport is mediated by systems L (LAT1 and LAT2) and b^0,+ (b^0,+AT). ⁵⁰ To determine which of these glutamine transporters were expressed in NPE cells, ciliary processes were labeled with isoform-specific antibodies raised against LAT1, LAT2, or b^0,+AT (Fig. 8). Both LAT1 and b^0,+AT labeling were evident in the basolateral membranes of NPE cells (Figs. 8A, 8C), whereas LAT2 labeling was cytoplasmic and localized in both NPE and PE cells (Fig. 8B). These results indicate that of the three possible glutamine transporter candidates, LAT1 and b^0,+AT are most likely to mediate glutamine efflux from the NPE cells into the aqueous. 

Figure 8.

 

Localization of glutamine transporters in the rat ciliary epithelium. Confocal images of ciliary body processes showing the localization of the glutamine transporters LAT1 (A), LAT2 (B), and b^O,+AT (C) in the ciliary epithelium. Although LAT1 and b^O,+AT label the basolateral membrane of NPE cells, LAT2 labeling is diffuse and cytoplasmic within both cell layers. Scale bar, 50 μm.

Discussion

In this study we have used immunolocalization techniques before and after the manipulation of glutamine synthetase activity to show that the NPE cells of the rat ciliary epithelium contain pathways for the uptake of glutamate, its conversion to glutamine, and the subsequent efflux of the accumulated glutamine into the aqueous humor (Fig. 9). In this model, NPE cells accumulate glutamate from the blood through the sodium-dependent glutamate transporter EAAT3, which is strongly localized to the apical membranes of NPE cells (Fig. 6). More diffuse labeling of EAAT3 on the basolateral membranes of NPE cells was also evident; it appeared to be responsible for the uptake of glutamate from the aqueous as intravitreal injection of the non-metabolizable glutamate analog d-aspartate caused it to be accumulated in both cell layers of the ciliary epithelium (Fig. 7B). We propose that this rapid conversion of glutamate to glutamine creates a concentration gradient that can then be used by transporters to facilitate the diffusion of glutamine across the basolateral membrane of NPE into the aqueous humor. Consistent with this hypothesis, we localized the glutamine transporters LAT1 and b^O,+AT to the basolateral membrane of NPE cells (Fig. 8). 

Figure 9.

 

Molecular model of glutamate uptake and glutamine efflux pathways in the ciliary epithelium. Levels of glutamine and glutamate in the aqueous are determined by glutamate uptake, glutamate metabolism are glutamine efflux pathways that are localized in the NPE cells.

In monkey lenses, Langford et al. ²⁶ showed a similar distribution for EAAT1 suggesting that although species-specific differences in isoform expression exist, the mode of glutamate uptake is essentially similar in rodents and primates. Of all the EAAT isoforms screened in the rat, only EAAT3 was localized to the ciliary epithelium. In neurons, EAAT3 mediates active high-affinity accumulation of glutamate by the cotransport of two to three sodium ions, a proton, and the countertransport of a potassium ion. ⁵¹ In addition, EAAT3 has been implicated as a major route of cysteine uptake for neurons in the brain, ^{51 –53} suggesting that EAAT3 may have physiological roles other than glutamate uptake in neurons. Whether this holds true in the ciliary epithelium is yet to be determined. However, the enzyme γ-glutamyltranspeptidase, which hydrolyzes glutathione to the dipeptide cysteine-glycine, has been localized in the basal membrane of the NPE cells, ⁵⁴ and the cysteine-glycine dipeptide is the precursor of extracellular cysteine. Hence, EAAT3 may represent the previously unidentified mechanism for cysteine uptake in the ciliary epithelium. Further studies are required to corroborate this speculation. 

The observed change in amino acid labeling intensity after GS inhibition (Figs. 4, 5) shows not only that the glutamate accumulated by EAAT3 is then converted to glutamine in NPE cells but also that the differential distribution of glutamate and glutamine observed between the NPE and PE cells is the result of metabolic pathways specifically located in the NPE cells. Consistent with this view, the observed increase in glutamate and reduction in alanine immunoreactivity imply that glutamate is being fed into the tricarboxylic acid cycle rather than being used to synthesize alanine (Fig. 1). Furthermore, because NPE and PE cell layers are highly coupled by gap junctions at their apical-apical interface, ^7,8 we would have expected the two cell layers to have similar levels of amino acids. However, the localization of GS activity and EAAT3 solely in the NPE cells enables these cells to establish and maintain concentrations for glutamate and glutamine that are higher than those observed in PE cells. When GS activity is inhibited, glutamate is no longer converted to glutamine, and the high degree of PE-NPE cell coupling by way of gap junctions results in an equalization of the glutamate and glutamine concentrations between the two cell layers (Fig. 5B). Finally, the localization of GS activity specifically in NPE cells is also consistent with the observation that NPE cells have a higher metabolic capacity than PE cells, ⁵⁵ which may contribute to amino acid uptake and metabolism. 

The uptake of glutamate by EAAT3 and its subsequent conversion to glutamine creates a concentration gradient that favors the facilitated diffusion of glutamine into the aqueous humor. We have shown that two Na⁺-independent facilitative glutamine transporters, LAT1 and b^0,+AT, are localized to the basolateral membrane of NPE cells (Fig. 8), suggesting that these transporters are likely candidates for mediating glutamine efflux from NPE cells into the aqueous. These transporters are members of a large heteromeric amino acid exchanger family, the members of which function as obligatory exchangers with other amino acids (1:1 stoichiometry) and which require coexpression with a heavy subunit (4F2hC) and a corresponding light subunit (LAT1/b^0,+) for functional cell surface expression. ⁵⁰ Although LAT1 and b^0,+AT exchange a broad range of neutral and neutral/dibasic amino acids, b^0,+AT (K _m ∼83 μM) appears to have a higher affinity for glutamine than LAT1 (K _m ∼2.2 mM). ⁵⁰  

If our model (Fig. 9) is correct, then the balance between the efflux of glutamine and the influx of glutamate across the basolateral membranes of NPE cells will together establish the high and low concentrations of glutamine and glutamate, respectively, that have been measured in the aqueous. ²⁶ The aqueous humor delivers nutrients to the avascular tissues in the front of the eye, and previous studies have shown that the lens is dependent on glutamine as a source of glutamate. Glutamine is actively transported to the lens from the aqueous, ⁵⁶ where it subsequently undergoes deamidation through glutaminase to form glutamate within the lens. ⁵⁷ Thus it appears that the ciliary epithelium and the lens express metabolic pathways that ensure low resting levels of glutamate in the aqueous humor. 

This maintenance of low resting levels of glutamate in the aqueous humor appears to be important for overall ocular health. The amino acid glutamate also functions as a neurotransmitter and could be toxic to the neural retina if present at inappropriate concentrations in the posterior chamber. ^{58 –60} Indeed, vitreal glutamate levels have been shown to be elevated in a number of diseases, including retinal detachment, ⁶¹ inherited retinal dystrophy, ⁶² and diabetes. ⁶³ Changes in glutamate/glutamine metabolism and transport by the ciliary epithelium would be expected to directly impact on glutamate/glutamine homeostasis in the avascular lens epithelium. In this regard, GS is frequently a target of metabolic insult because of its dependence on ATP as a substrate. In a rat diabetic model, the conversion of glutamate to glutamine declines in the retina, ^64,65 despite the preservation of glutamate uptake systems. ⁶⁶ Others have also reported lower glutamine levels in the rat retina and the vitreous in the diabetic human eye. ^67,68 Although it remains to be established whether GS function is affected in the ciliary epithelium during metabolic diseases, a parallel reduction of amino acids in the aqueous humor and the lens during diabetes have been noted. ^69,70 However, having demonstrated that experimental inhibition of GS activity in the ciliary epithelium is detrimental to the accumulation of glutamine in NPE cells, we are now in a position to determine whether glutamate levels are also impacted in the lens. Exploring further the potential for metabolic cross-talk between the ciliary epithelium and the lens will therefore be a focus for future work. 

Footnotes

 Disclosure: R. Hu, None; J.C. Lim, None; M. Kalloniatis, None; P.J. Donaldson, None