THE JOURNAL OF COMPARATIVE NEUROLOGY 320257-266 (1992)

Glucose Metabolism in Freshly Isolated Muller Glial Cells From a Mammalian Retina CAROL L. POITRY-YAMATE AND MARCOS TSACOPOULOS Experimental Ophthalmology Laboratory, Department of Oto-neuro-ophthalmology, University of Geneva Medical School, 1211 Geneva 4, Switzerland

ABSTRACT Glucose metabolism was studied in isolated retinal Muller glial cells from the juvenile guinea pig. Cells, once enzymatically isolated and purified, were identified by morphological criteria, positive vimentin immunoreactivity, and histochemical staining for glycogen. Purified suspensions of Muller cells were obtained in quantities sufficient for biochemical analysis ( = 2 x 105/pair of retinas) and light microscopic autoradiography. In bicarbonate-buffered Ringer's medium containing 3H-2-deoxyglucoseand no glucose, 2 80%of the glucose analogue taken up intracellularly by Muller cells was phosphorylated to 'H-2-deoxyglucose-6-phosphate. In autoradiographs, this non-metabolized product provided visual evidence of glucose phosphorylation: the distribution of cell grains mirrored the morphology of individual Muller cells in situ. Exposure to the glycolytic inhibitor iodoacetate (500 KM) caused an 85% decrease in deadenosine triphosphate (ATP) content; concomitantly, 3H-2-deoxyglucose-6-phosphate creased by 90% and paralleled a dramatic decrease of cell labelling in autoradiographs, while levels of 3H-2-deoxyglucose did not change. In the continual absence of glucose, glycogen content decreased with time and this decrease was slowed by 36%in the presence of iodoacetate. This indicated that, in control conditions, glycosyl units from glycogen sustain cellular metabolism, and hence 3H-2-deoxyglucose phosphorylation. 3H-2-deoxyglucose-6-phosphate concentration was 43-fold less than that of ATP in the control conditions so that depletion of ATP during iodoacetic acid (1AA)-blockedglycolysis was not due to hexokinase activity. These results demonstrate that this preparation is adequate for quantitative studies of glucose metabolism at the cellular and molecular level in an important metabolic compartment of the mammalian retinax 1992 Wiley-Liss, Inc. Key words: retinal glia, 3H-2-deoxy-D-glucose,dry autoradiography, glycogen, iodoacetate, hexokinase, guinea pig

The metabolism of glucose through the glycolytic pathway is vital in maintaining retinal function in mammals (Cohen and Noell, '60; Graymore, '70; Krebs, '72; Winkler, '81).When the first comprehensive histochemical studies of glucose metabolism in the vertebrate retina (Kuwabara and Cogan, '60, '61) were undertaken in an effort to localize glucose utilizing cells, Kuwabara and Cogan found that the presence of enzymes of glycolysis and glycogen synthesis were concentrated primarily in Muller cells. These cells are the principal glia in the vertebrate retina. They extend radially through all of the retinal layers containing neurones and photoreceptors, as well as horizontally across two of these layers where neurones make synapses (Dowling, '87). In the juvenile guinea pig retina, we recently combined biochemical and autoradiographic techniques to show that the radiolabeled glucose analog, 3H-2-deoxyglucoseis phos-

o 1992 WILEY-LISS, INC.

phorylated in the absence of exogenous glucose predominantly in Muller glial cells (Poitry-Yamate and Tsacopoulos, '91). This appears to be the first report of such autoradiographic evidence in mammals (c.f. Sperling et al., '82). In addition, this result was obtained after extensive washing of fresh retinas exposed t o 3H-2-deoxyglucose, indicating that there was negligible glucose-6-phosphatase activity in Muller cells (c.f. Goldman, '90). A serious drawback in our previous study was that the localization of label over Muller cells in situ was difficult to show convincingly in autoradiographs, even though, in the juvenile Accepted January 29,1992 Portions of this research have previously appeared in abstract form (Association for Research in Vision and Ophthalmology,1991). Address reprint requests to C.L. Poitry-Yamate,Experimental Ophthalmology Laboratory, 22 rue Alcide-Jentzer, Geneva, Switzerland.

258

C.L. POITRY-YAMATE AND M. TSACOPOULOS

guinea pig retina, the Muller cells are relatively large and exceptionally well recognizable in light micrographs (PoitryYamate and Tsacopoulos, '91). To overcome this drawback, we freshly isolated and purified these cells from the retina of the juvenile guinea pig in order to show that they intensively phosphorylate 3H-2-deoxyglucose(3H-2DG), to 3H-2-deoxyglucose-6-phosphate, (3H-2DG-6P). We predicted that if labeling of individual Muller cells could be shown in autoradiographs, and that if this label corresponded to biochemically determined 3H-2DG-6P,these results would provide strong evidence for their glucose metabolism. We report here this evidence, and we show that the amount of 3H-2DG-6P is greatly decreased by iodoacetic acid (IAA), a potent inhibitor of glycolysis. The primary objective of this paper is to show that this preparation of mammalian Muller cells not only constitutes a powerful model for studying the molecular aspects of glucose metabolism in an important cell of the central nervous system (CNS), but also a model in which to study the role of glial cells in retinal energy metabolism.

MATERIALS AND METHODS Animals Forty-five guinea pigs (Cauiaporcelus cobaya) between 5 and 14 days of age were used in this study. They were kept on a 12 hour light and dark cycle and fed chow and water until experimentation. Before decapitation, animals were lightly anesthetized with an intraperitoneal injection of sodium pentobarbital (30 mgikg body weight).

Reagents 2-deoxy [2,6-3H]glucose (35-42 Ciimrnol) was purchased from Amersham (Zurich, Switzerland). Collagenase (Clostridiopeptidase A) from Clostridium histolyticum (0.18 Uimg), sodium iodoacetate, Aspergillus niger amylo-a-1,4al,6-glucosidase and trypan blue were purchased from Fluka (Buchs, Switzerland). Papain (12 Uimg), hyaluronidase (Type I-S), deoxyribonuclease I (from bovine pancreas), adenosine triphosphate (ATP), and reagents for microdetermination of total protein (kit 690-A) were purchased from Sigma (Munich, West Germany). BSA (fraction V) was purchased from Calbiochem (San Diego, CA). Hexokinase and glucose-6-phosphate dehydrogenase and NADP+ were from Boehringer Mannheim (West Germany). 1243-200ATP Monitoring Reagent for the bioluminescence measurement of ATP was purchased from Pharmacia LKB (Duebendorf, Switzerland).

Preparation of an enriched Muller cell fraction Eyes from guinea pigs were enucleated as previously described (Poitry-Yamate and Tsacopoulos, '91). For any

Abbreviations: 3H-2DG 'H-2DG-6P G6P 1.4'4 ATP PFK GBPDHG HPLC CPm dPm

3H-2-deoxyglucose 3H-2-deoxyglucose-6-phosphate glucose-6-phosphate iodoacetic acid adenosine triphosphate phosphofructokinase glyceraldehyde-3P-dehydrogenase high pressure liquid chromatography counts per minute disinteffrations per minute

one experiment, 2 eyecups were incubated for 60 minutes at room temperature in a circulating standard bicarbonatebuffered Ringer's solution (in mM: NaCl 124; KC1 5; CaCl, 2; KH,PO, 1.25; NaHCO, 20; MgSO, 2; pH, 7.4; 95%,0,5% COJ before retinas were dissected out under dim blue light. Retinal pairs were returned to this solution and incubated further for 60 minutes. We observed that this procedureincreased Muller cell yield. Retinas were subsequently dissociated by 5 sequential operations at 37°C: (1) enzymatic digestion with collagenase (4 mgiml) and hyaluronidase (1mgiml) in Ca-Mg containing Ringer's; (2) exposure to Ca-Mg-free EGTA (2.5 mM) medium acidified to pH 6.0 (with HC1); (3) enzymatic digestion with papain (1mgiml) in EGTA-containing medium acidified to pH 6.5; (4) four 5 minute rinses in Ca-Mg medium (0.05 mM and 1.2 mM, respectively) with the final rinse containing BSA (3%)and DNAase (0.1%); and (5) mechanical disaggregation using a Gilson pipetman and 1 ml pipette tip with an orifice cut to around 3 mm. Up to this step, retinas took on the shape of a folded eye-cup without visible disruption of the tissue surface. Step 1 was performed in standard Ringer's solution; steps 2 through 5 and cell purification in Percoll were carried out in a second Ringer's solution (in mM: NaCll20; KCl 3.1; NaHCO, 5; KH,PO, 0.5; Na,SO, 1.2; HEPES 10; MOPS 10; bicine 5; (D)-glucose 10, pH, 7.3; 95% 0,/5% CO,) with or without CaC1, (0.05 mM) and MgCl, (1.2 mM) as described by Trachtenberg and Packey ('83). With the exception of the cell purification stage, all solutions were stirred by a constant jet of COJO,. The duration of enzyme exposures were each 5 minutes. The supernatant fractions following five successive 5 minute interval triturations were examined microscopically for their cell content. Muller cells were collected in greatest numbers from the top of the second, third, and fourth supernatants and combined. Figure l a shows this combined crude cell suspension. Photoreceptors and inner retinal neurones were also present (Fig. l a , inset). These fractions (total volume = 2.0-2.5 ml) were kept chilled on ice until cell purification. Two equal volumes of the crude suspension were each layered on a Ringer's solution containing 18% stock Percoll, 0 glucose, and 1.2 mM Mg" (total volume of 5 ml). This medium was chilled and oxygenated beforehand. The tube contents were centrifuged for 15 minutes at 700 g and 2°C to form a Percoll gradient in situ. No braking was applied. A single band of cells, enriched in Muller cells and small spherical bodies, was located two-thirds through the resultant gradient. Directly overlying this band was a turbulent layer highly enriched in Muller cells. Photoreceptor segments, medium sized spherical bodies, and grape-like structures were situated in the lightest fractions; fragmented neurones were situated below the band of cells. Lying at the bottom of the gradient was undissociated tissue. Muller cells in the turbulent layer and band of cells were collected and layered on a second identical Percoll gradient to increase Muller cell purity. Muller cells from the second gradient were collected from the turbulent layer until skimming of the band was visible. The gradient medium was removed from Muller cells by a five-volume dilution in standard Ringer's containing 0.05 mM Ca**.The resultant purified preparation of Muller cells is shown in Figure lb. Visual counts of 5 ~1 aliquots of this suspension contained approximately 2 x lo5Muller cellsiretinal pair. This suspension was highly enriched with Muller cells so that any contribution from spherical bodies would be small. On the basis of the values of 380 Km'isphere and 9,800 pm"/

259

GLUCOSE METABOLISM IN MAMMALIAN MULLER GLIA

Fig. 1. a: Crude suspension of pooled cell fractions enriched with Miiller cells after mechanical disaggregation of the retina. Brightfield optics. Bar: 100 p,m. Inset: Boxed area at higher magnification shows photoreceptors, inner retinal neurones, and Muller cells. Bar: 50 pm. b: Darkfield illumination of purified suspension of Muller cells. = 200,000 Muller cellsiretinal pair were obtained in 1 2 hours subsequent to

retinal digestion. Bar: 100 pm. c: Morphological detail of Miiller cell frequently observed in b. EF, endfoot; AR, ascending radial process; DR, descendingradial process; CP, cell perikaryon. Arrows point to side expansions. At scleral (bottom) end of cell, radial strands (*) extend from DR and terminate as angular buttons (arrowhead). Bar: 10 km.

Miiller cell (see Trachtenberg and Packey, ‘83 for similar values), spherical bodies contributed between 5 to 15% of the total cell volume. Approximately 90% of the Miiller cells excluded the vital dye trypan blue (0.1%). The final cell suspension was concentrated in 200 ~1 of standard Ringer’s medium. Figure l c shows the principal characteristics and form of Muller cells most frequently observed in our final cell suspensions (for comparisons see Polyak, ’41 and Reichenbach et al., ’89).

minutes at 37°C. The coverslips were rinsed in PBS (3 x 15 minutes) and dipped briefly in 70% ethanol before mounting on microscope slides in Movial. Slides were viewed under fluorescence epi-illumination on a Zeiss Axiophot microscope and photographed using ILFORD HP5 black and white film. Miiller cells from the juvenile guinea pig were immunoreactive to vimentin but not to GFAP. This observation is consistent with the pattern of vimentinpositive and GFAP-negative immunoreactivity of Miiller cells in the retina of the adult guinea pig (Schnitzer, ’88). Our staining of individual Muller cells was observed along the entire radial cell axis and well delimited within the cell border (results not shown). Free spherical bodies from the purified cell suspension were also labeled positively. However, bodies of comparable diameter adhering to the scleral end of the Miiller cell did not express vimentin immunoreactivity. We suggest that the labeling of spherical bodies belongs to photoreceptors enveloped by Miiller cell cytoplasm. This cluster corresponds closely to Golgi-stained “thin-walled bubbles” and “sockets” of Muller cell cytoplasm described in the ONL (outer nuclear layer) in the rabbit (Reichenbach et al., ’88) and monkey (Polyak, ’41) retina. Upon dissociation and purification, these clusters of these structures were sheared from the distal (scleral) portion of Miiller cells.

Immunofluorescence staining Purified cells were classified as Miiller cells by their positive immunofluorescent staining with antibodies directed against vimentin intermediate fdament protein. Briefly, purified cells were attached to poly-L-lysine (100 pgiml) coated glass coverslips and labeled either with mouse monoclonal or rabbit polyclonal antibodies specific for vimentin (mAb 3B4, Projeu, Heidelberg, Germany or polyclonal rabbit IgG from BioScience Products, Ennentbiirgen, Switzerland) or for the astrocyte marker glial fibrillary acidic protein (GFAP) (mAb 1224, Ready Systems, Zurich, Switzerland or polyclonal rabbit IgG from BioScience Products, Switzerland). Primary antibodies were applied on the cells and the coverslips were inverted onto microscope slides so as to expose all cells to the antibody. This assembly was incubated in a humid chamber at 37°C for 35 minutes. These coverslips were then removed and thoroughly washed (3 x 15 minutes) in PBS (pH 7.4). Secondary antibodies coupled to Texas Red (donkey anti-mouse or anti-rabbit IgG, respectively, Daneva, Hamburg) were applied for 20

Uptake and phosphorylation of “H-2-deoxyglucose Protocol I. Purified suspensions of Miiller cells were incubated in standard Ringer’s carrying = 6-8 FM ‘H-2DG

260

C.L. POITRY-YAMATE AND M. TSACOPOULOS

and no glucose for 55-60 minutes at 37°C under dim white sured by the enzymic fluorometric micro method described light. The uptake and phosphorylation of 'H-2DG was by Nahorski and Rogers ('72). For the histological staining stopped by adding 5 volumes of ice cold Ringer's. The of glycogen, Muller cell pellets were resuspended and fixed contents were immediately centrifuged, the supernatant in ice cold 0.1 M phosphate buffer (pH 7.3) containing removed and the pelleted cells resuspended in 300 ~1 glutaraldehyde (2.5%) for 3 hours. Cells were collected by Ringer's. One or two aliquots were saved for light micro- centrifugation and resuspended in cacodylate buffer (0.05 scopic autoradiography. The remainder was frozen in liquid M, pH 7.3) for 60 minutes and then dehydrated through a nitrogen, lyophylized (15 hours), and prepared for high series of increasing alcohol concentrations (50, 70, 80, 96, pressure liquid chromatography (HPLC) separation of 3H- and loo%), resuspended in propylene oxide, and then 2DG and 3H-2DG-6P as described by Tsacopoulos et al. spread out on clean microscope slides. A mixture of Epon ('88). Radioactivity was eluted for 30 minutes and 60 and propylene oxide (3:1v/v) was dropped onto cells to form fractions were collected. Total radioactivity was calculated thin sheets and left overnight to polymerize at 65°C. by integrating counts per minute (cpm) between 5 and 27 Afterwards, this Epon was partially deplasticized (Mar and minutes, inclusive. Radioactivity in 3H-2DG from Muller Wight, '88). The periodic acid thiosemicarbazyde silver cell homogenates was eluted as a major and a minor peak. proteinate method described by Thi6ry ('63) for the hisAs the minor peak was I 5% of 3H-2DG-6P,only the major tochemical staining of glycogen was used directly on micro3H-2DGpeak is reported in this paper. scopic slides. Control slides underwent incubation in (w-1,4Protocol ZZ. Paired aliquots of Muller cell suspensions a- 1,6-amyloglucosidase to digest glycogen. Stained Muller were simultaneously incubated for 30 minutes in standard cells were photographed with Kodak TMX color film. Ringer's solution carrying = 8-10 FM 3H-2DG and either 500 FM (experimental group) or no (control group) IAA. Light microscopic autoradiography of single Experimental manipulations of both groups were systematMiiller cells ically performed in parallel and as consistently as possible. The method used previously for pieces of retinas (PoitrypH remained at 7.4 during the experiments. After 30 minutes, the cell medium was collected by centrifugation Yamate and Tsacopoulos, '91) was followed except for one (30s). Phosphorylation of 3H-2DG was then stopped as simplification which did not in any way compromise the described in protocol I. Pelleted cells were resuspended in basic principle of the autoradiographic technique (Buchner 300 p,l of standard Ringer's. Aliquots for the determination and Buchner, '801, that of avoiding aqueous phases at all of 3H-2DG, 3H-2DG-6P, protein concentration and ATP stages from fixation to autoradiography. Aliquots of Miiller were immediately frozen in liquid nitrogen and lyophylized cells were whole mounted onto microscope slides and (15-20 hours). Aliquots of Muller cells for light microscopic within 10 minutes freeze-dried (c.f. Saji and Obata, '81).We autoradiography were attached to poly-L-lysine (100 kg/ lost 2 50% of the cells this way by choice, because attaching ml) coated microscope slides. This unit was frozen in all of them to slides, which took 30 minutes or more, led us isopentane cooled with liquid nitrogen and subsequently to question cell membrane integrity. The reproducibility of lyophylized (20 hours). Protein was determined by the cell labeling, and the extent to which the outline of cell label phenol reagent method with bovine albumin as standard. corresponded to the morphology of individual Muller cells, ATP was extracted from cell lysates by the TCA-diluted on the other hand, significantly outweighed this cell loss. method described in BioOrbit application note 200 (MBV After lyophylization, the corners of microscope slides (cellAG, Vevey, Switzerland) and measured with the biolumines- side) were treated with Fixogum (E. Martz GmbH) to cent assay method described in application note 201. All render this surface sticky prior to apposing a nuclear values were normalized with reference to protein content. emulsion coated (Ilford L4) microscope slide (c.f. Saji and Light microscopic observations revealed that the overall Obata, '81). Epon embedding was therefore not necessary. morphology and number of Muller cells exposed to 500 kM Autoradiographic "sandwiches" were exposed for 2 weeks IAA were comparable to that of Muller cells in parallel in the dark at 0°C in an air tight can filled with silica gel. control aliquots for 45 minutes. The experimental protocol Autoradiographs were viewed under oil immersion ( x 63) up to and including the freezing of cells did not exceed this and photographed with ILFORD PAN F or Kodak TMAX film. Unless noted, autoradiographs were not stained. time limit. When methyl bluelborax was used, autoradiographs were Enzymatic assay and histological detection of stained so that silver grains were visible and cell borders were outlined. Muller cell glycogen Glycogen content in Muller cell suspensions was determined at two intervals per experiment: immediately after RESULTS cell purification and after 30 minutes incubation with or Phosphorylationof 3H-2DGin Muller cells without added MA (500 FM). Cell suspensions were then frozen and lyophylized (15 hours) to simulate the autoradioFigure 2 shows a chromatographic profile of 3H-2DG-6P graphic conditions. Thereafter, they were resuspended and that accumulated in Miiller cells following incubation protohomogenized in HC1 (1N). One aliquot was saved for the col I described in Materials and Methods. Visual inspection determination of protein concentration. The remainder was of this major peak and the minor peak, corresponding to deposited on Whatman 3MM filter paper (1 cm2, (Maid- 3H-2DG,indicated that the majority of radioactivity accumustone, England) and this assembly was washed (EtOH 66% lated in Muller cells was in 3H-2DG-6P. Integration of v/v; 3 x 20 minutes), dipped rapidly in acetone (2x1 and cpm/eluent peak for 6 different experiments is shown in then left to dry. Supernatants were collected and frozen Figure 2 (inset).The total average radioactivity taken up by after assemblies were exposed to amylo-(w-1,4-(wl,6-glucosi- Muller cells in 6 different experiments was 507745 5 61121 dase (15 pl/9 ml acetate buffer, 0.1 M, pH = 5.0) for 90 (SEMI cpm. Of this radioactivity, 2 82% was recovered as minutes at 37°C with agitation. Glycogen was then mea- 3H-2DG-6Pand = 11%was recovered as 3H-2DG.

261

GLUCOSE METABOLISM IN MAMMALIAN MULLER GLIA 5

6

l-

a

E Y)

0 r(

0

Q] 2DG-BP

["HI 2DG

0 5

15 Elution time (min)

25

Fig. 2. a: Chromatogram showing the separation of intracellular 'HH-2DG-6P(eluted at 14 minutes) and 3H-2DG(eluted at 20 minutes by high-pressure liquid chromatography (HPLC) from homogenates of purified Miiller cells after 60 minutes incubation in bicarbonate

buffered Ringer's carrying 'H-2DG. Over 8 0 8 of total radioactivlty taken up by cells was intracellularly phosphorylated to 'HH-2DG-6P. Inset: Summary of six different experiments as described in a. Values are means 2 SEM.

Autoradiographs prepared from these experiments are shown in Figure 3. The distribution of cell grains matched the shape and length of Muller cells from two topographical regions of the retina in situ: from the periphery (Fig. 3a), where the retina is thin, and from the centre (Fig. 3b and c) where the retina reaches its maximum thickness (Reichenbach et al., '89). Thus, the overall outline of the cell label at the top of Figure 3a is largely conical and tapers downwards to form a wide but short radial axis. The area indicated by arrows corresponds to the side-expansions of the cell that, in situ, extend laterally across the synaptic layers. In contrast, the overall outline of the cell label at the top of Figures 3b and c is bulbar and tapers to form a slender and long radial axis. The definition and localization of this cell label is best appreciated by visually scanning, from top to bottom and with attention to form, the autoradiograph in Figure 3c placed alongside the freshly purified Muller cell shown in Figure lc. When autoradiographs were stained t o reveal the shape of the cells, as in Figure 3b, cell grains were found to fall abruptly at the cell boundary. When cell grains were focused back and forth through the thickness of the photographic emulsion, the density of cell label was qualitatively uniform from one cell region to another. More than 50 Muller cells in autoradiographs were similarly labeled.

this figure, 'H-2DG-6P decreased by 1 9 5 % when differences in the area under the 3H-2DG-6P peak were calculated between the control (no IAA) and experimental (500 p,M IAA) situation. This difference was not, however, accompanied by a concomitant decrease in 'H-2DG during IAA exposure. In six experiments, the total radioactivity taken up by Muller cells in control experiments amounted to 14,208 ? 2,995 (SEMI cpm/p,g protein; for Muller cells exposed t o IAA, this value amounted to 4,331 862 (SEMI cpm/kg protein, or approximately 70% less than in controls. The average radioactivity in 3H-2DG-6Pand 3H-2DG from these experiments is summarized in Figure 4b. On average, the amount of intracellularly accumulated 3H2DG-6P decreased by approximately 90% in Muller cells exposed to IAA. This decrease was statistically significant (p 5 0.001, n = 6); Differences in 'H-2DG between cell groups were variable but not statistically significant (p 2 0.53, n = 6). The corresponding autoradiographs are shown in Figure 5. Miiller cells from control experiments displayed a similar labeling pattern as shown in Figure 3; that is, cell label extended from the endfoot along the radial axis to the scleral end of the cell. Cell grains were readily distinguishable from background grains when the latter were densely distributed (Fig. 5a) or not (Fig. 5b). Similar labeling was observed in approximately 50 Muller cells. In contrast, no distinction between cell grains and background grains could be made in autoradiographs of IAA-exposed Muller cells. As the total radioactivity was ~ 3 . times 3 less than in the

Inhibition of 3H-2DGphosphorylation by iodoacetate The accumulation of 3H-2DG-6P in Muller cells after exposure to M A (i.e., protocol I1 described in Materials and Methods) is shown in the chromatogram in Figure 4a. In

*

C.L. POITRY-YAMATE AND M. TSACOPOULOS

262

Fig. 3. Light microscopic autoradiographs of isolated Muller cells from the experiments described in Figure 2. Silver grains, corresponding to 'H-2DG-6P, appear as dark spots and are spatially localized over Muller cells. Cell grains extend from endfoot (at top of figure) to scleral end of cell (at bottom). Variations in overall height and breadth of cell grain distributions correspond to the morphology of Muller cells from

different regions of the retina. Autoradiograph in b was stained with methyl blueiborax to reveal Muller cell outline. Degree of similarity between overall cell grain distribution and cell shape, and correspondence of silver grains to "H-2DG-6Pdemonstrate integrity of cell membrane and metabolism of glucose. Muller cells oriented with endfeet at top of figure. Arrowheads in a point to side expansions. Bar: 10 pm.

control situation, it was necessary to stain autoradiographs in order to locate Muller cells. The outline of IAA-exposed Muller cells after staining are shown in Figure 5c and d. When the background grain densities between control and IAA-exposed cells were comparable, as is shown in Figure 5a with 5c and in Figure 5b with 5d, cell grains over IAA-exposedMuller cells diminished significantly. An accumulation of label, if any, was observed over the cell perikarya in approximately 20 IAA-exposed cells in two experiments. Labelling of cells, like those shown in Figure 5c and d were similarly observed in more than 60 Muller cells.

ATP and glycogen content

yellow to orange-brown staining. The distribution of this staining extended along the entire radial length of the cells, in a manner similar to the staining of these cells in situ (Poitry-Yamate and Tsacopoulos, '91). This staining was specific for glycogen since it could be removed by digestion 6-glucosidase (results not shown). with amylo-a-1,4-a1, After cell purification k e . , time = 0) Muller cells contained 171.8 % 17.5 (SEMI nmol of glycosyl unitsimg protein (n = 5). Figure 7 shows that this glycogen decreased with time independent of whether cells were exposed to IAA or not. Differences in glycogen content between cell groups at 30 minutes were statistically significant (paired t-test p I 0.05).

The content of ATP in Muller cells determined from the experiments described in the previous section is shown in Figure 4b. Muller cells in control experiments contained 12.4 5 0.89 (SEM) nmol ATPimg protein (n = 6). This amount was less than in freshly excised intact retina ( = 30 nmol ATP/mg protein). Muller cells exposed to IAA at a concentration of 500 pM contained 1.66 % 0.22 (SEMI nmol ATP/mg protein (n = 61, or approximately 85% less ATP than in the control situation. The results obtained after histochemically staining purified Muller cells for glycogen are shown in Figure 6. Glycogen was present in individual Muller cells as dark

The chief points of interest which emerge from the present work have been: (1) to provide unequivocal evidence in light microscopic autoradiographs of an intense uptake of 3H-2DG and its phosphorylation by hexokinase (Sols and Crane, '54) in freshly purified, individual Muller glial cells, and (2) to biochemically demonstrate that, without exogenous glucose, the Embden-Meyerhof pathway using glycosyl units derived from glycogen operates in these cells to sustain 3H-2DGphosphorylation.

DISCUSSION

GLUCOSE METABOLISM IN MAMMALIAN MULLER GLIA

a

[ 3H]2DG-

I \ 15

25

elution time (min) 15

b 7

m

T

1-

0

Y r(

0

[ 3H]-ZDC

[3H]-2DG-BP

ATP

Fig. 4. a: Chromatogram showing inhibition of WBDG phosphorylation in Miiller glial cells by IAA.'HH-2DG-6Paccumulation (dashed line) decreased by 90% on average compared to that in control conditions (solid line). Note small variation in 'H-ZDG between cell groups. This indicated that 3H-2DG transport into Miiller cells was not substantially affected by IAA. b: Summary of six experiments as in a. '€IH-2DG-6Paccumulation and ATP content, but not "H-BDG, decreased during IAA-blocked glycolysis. Control experiments indicated by striped vertical bars. Values are means 5 SEM: n = 6 for each condition.

Incubations in the absence of glucose with 'H-2DG have similarly been employed in the intact retinae (Sperling et al., '82; Witkovsky and Yang, '82; Tsacopoulos et al., '88; Poitry-Yamate and Tsacopoulos, '91). In these studies, it has been observed that either retinal glial cells (Tsacopou10s et al., '88; Poitry-Yamate and Tsacopoulos, '91) or photoreceptors (Sperling et al., '82; Witkovsky and Yang, '82),but apparently not both are labeled. The explanation for this observation remains to be elucidated.

Fig. 5. Light microscopic autoradiographs of Miiller cells from the experiments described in Figure 4. a and b: Autoradiographs of Miiller cells not exposed to IAA. Cell grains, corresponding to "H-2DG-6P, stand out upon both a high (in a) and low (in b) grain background. In b, the layer of emulsion is not entirely flat and focus is on cell grains over upper half of Muller cell. Unstained. c and d Stained autoradiographs of Miiller cells after exposure to IAA. Cell and background grains are indistinguishable. In d, only the stained outline of the Miiller cell was visible. Compare autoradiographs of comparable background grains. Miiller cell endfeet oriented at top of figures. Bar: 10 +m, and applied to a-d.

263

C.L. POITRY-YAMATE AND M. TSACOPOULOS

264

0

glycogen

IrpHpl G1P

ATP

,

3H-2DG u 3 H - 2 D G - 6 P

Fig. 6. Glycogen staining of wholemounted and Epon-embedded Miiller cells. Staining was dark yellow to orange-brown from EF along AR, CP, and DR to scleral end of the cells. Digestion of glycogen with a-amylase in control slides removed staining in the cells (not shown). For abbreviations, see Figure 1. Bar: 20 pm.

F-1.2 dip

t

C

2 "OI

G6P

Gly,- 3P

T

pyruva t e

C

1

2

3

Fig. 7. Glycogen content of purified Muller cells in the absence of glucose: (11, immediately after purification; (21, after 30 minutes incubation without IAA; (3),after 30 minutes incubation with IAA. Values are means -+ SEM of five experiments.

Cell label in autoradiographs, after incubation of Muller cells in Ringer's medium containing 3H-2DG,corresponded essentially to the non-metabolizable (Nelson et al., '87) product 3H-2DG-6P.This label was observed over the entire (up to 130 Fm) radial length of individual Muller cells from topographically distinct regions of the retina. These results, taken together, substantiate the localization of label previously observed over the endfeet and perikarya of Muller cells in situ (Poitry-Yamate and Tsacopoulos, '91). The phosphorylation of 3H-2DGby Muller cells, a specialized type of brain astrocyte in the retina (Newman, '87), is consistent with the phosphorylation of 3H-2DG by other CNS glial cells (Clarke et al., '84; Tsacopoulos et al., '88; Walker et al., '88). Of particular interest is the finding that the phosphorylation of 3H-2DG by glial cells incubated in the absence of glucose with 3H-2DG from rat brain was 4-5-fold higher than by neurones (Walker et al., '88). Ideally, comparisons of this type between purified populations of Muller and photoreceptor cells would be of general interest towards understanding the relative roles of each cell type to the overall metabolism of glucose in the retina. In contrast to these studies, a gluconeogenic pathway, and by inference glucose-6-phosphatase activity, in cell

Fig. 8. Possible mechanisms during IAA-blocked glycolysis in Miiller cells as described in Figures 4 and 7: W 2 D G phosphorylation in Muller cells is maintained by the glycolysis of glycogen glycosyls (1) when exogenous glucose is unavailable. During IAA-blocked glycolysis (2), ATP production stops (3) and hexokinase activity (4) is inhibited by ATP availability. PFK activity ( 5 ) in turn is activated and this activity not only drains residual ATP but continues to draw upon fructose-6P (6) derived from glycogen. PHP, phosphorylase; Hex, hexokinase; GlP, glucose-1-phosphate; F6P, fructose-6-phosphate; F-1,2 dip, fructose 1,2 diphosphate; Gly-3P, glyceraldehyde-3-phosphate.Asterisk at step 2 indicates IAA.

suspensions enriched with Muller cells was recently reported in the frog retina (Goldman, '90). However, in the retina of the juvenile guinea pig, there is no such evidence (Poitry-Yamate and Tsacopoulos, '91). Furthermore, glucose-6-phosphatase activity is negligible in the mammalian retina in vivo (Blair et al., '89). In the present study, three parameters of Muller cell metabolism decreased when cells were exposed to IAA. These were 3H-2DG-6P accumulation, ATP content and glycogen content. The possible mechanisms bringing about these decreases are discussed below and summarized in Figure 8. Glucose metabolism first involves phosphorylation by hexokinase, and in the mammalian retina, the activity of this enzyme is present in the major layers (Lowry et al., '61; Matchinsky, '70). We have provided biochemical and autoradiographic evidence with 3H-2DGto show that part of this hexokinase activity is localized in Muller cells. The average concentration of 3H-2DG-6P that accumulated intracellularly amounted to: [10,391 cpm/pg protein] x 12.5 dpmi cpm] x [l mmo1/42 Ci] x [ l Ci/2.22 x 10" dpm] = 0.28 nmol/mg protein or = 2 8 pM 3H-2DG-6P. This value is = 4-fold the amount of 3H-2DG in the incubation medium and in the cells, and therefore provided sufficient radioactivity for the visualization of Muller cell label in autoradio-

GLUCOSE METABOLISM IN MAMMALIAN MULLER GLIA graphs (see Tsacopoulos et al., '88). In contrast, when Muller cells were exposed to 500 p,M IAA, 3H-2DG-6P accumulation decreased by 90%.This decrease paralleled a dramatic decrease of cell label in autoradiographs to an extent that cell and background label were indistinguishable. Two mechanisms could explain these changes: (1) inhibition of hexokinase activity, or (2)activation of glucose6-phosphatase. A decrease of ATP below a critical level (substrate availability) or an elevation of glucose-6phosphate (G6P) (product inhibition) would, for the first possibility, render hexokinase inactive (Wilson, '84).Because it is generally accepted that IAA exerts its major inhibitory effect on glyceraldehyde-3P-dehydrogenase (G3PDHG) in the glycolytic pathway (Lehninger, '821, and we are not aware of any studies indicating that IAA activates glucose-6-phosphatase, we believe that inhibition of hexokinase activity is the most likely explanation for the observed decrease of 3H-2DG-6P accumulation in Muller cells. The content of ATP in juvenile guinea pig Muller cells amounted to 12.4 nmolimg protein or ~ 4 . p,mol/g 1 dry weight following a 30 minute incubation in bicarbonatebuffered Ringer's medium with no glucose. This ATP decreased to 1.66 nmolimg protein or ~ 0 . 5 5p,mol/g dry weight with IAA. IAA therefore caused an 87% decrease, on average, in Muller cell ATP content. Using a similar incubation protocol, Winkler ('81) determined that perfused rat retinas contained =3.7 pmol/g dry weight of ATP. This value decreased by 30% in the presence of 2 mM IAA.These differences in ATP decreases raise the question of whether changes in ATP content in the rat retina occur more so in Muller cells than in photoreceptors. This has been suggested by the finding that glycolysis occurs predominantly in honeybee glial cells (Tsacopoulos et al., '88) and ATP content decreased by 70% in honeybee retinal slices exposed to 3 mM IAA (Brazitikos and Tsacopoulos, '91). Winkler ('81) found that retinal ATP content decreased by = 50% when unlabelled 2DG (10 mM) was added to the Ringer's medium containing IAA and pyruvate. He interpreted this result as indicating that hexokinase activity depleted retinal ATP content during IAA-blocked glycolysis. In contrast to his results, the depletion of ATP measured in the present study was not due to hexokinase activity: after a 30 minute incubation, the concentration of intracellularly accumulated 3H-2DG-6P in control experiments amounted to = 28 p M 3H-2DG-6P.Twenty-eight pM 3H-2DG-6Pequals 0.105 p,mol/g dry weight and was therefore = 43-fold less than that of ATP content. Although it is obvious that the content of ATP will decrease if ATP producing steps in glycolysis are blocked by IAA upstream at G3PDHG (Lehninger, '82), this calculation indicates that hexokinase activity, per se, should only cause a negligible decrease of ATP in Muller cells. It appears therefore that most of the 87% decrease in ATP that we measure is due to other mechanisms that consume ATP. In the present study, individual Muller cells stained positively for glycogen along the entire radial axis, similar to the staining pattern of these cells in situ (Poitry-Yamate and Tsacopoulos, '9 1).Glycogen is localized predominantly in Muller cells both in the juvenile and adult guinea pig retina (Kuwabara and Cogan, '61; Eichner and Themann, '62). Therefore, the staining of these cells was a means for their identification. The content of glycogen in purified suspensions of Muller cells amounted to over 175 nmol glycosyl unitsimg protein. Although estimates of glycogen

265

content in Muller cells from other preparations have not previously been reported, we compared our values to those measured in CNS astrocytes because the glycogen content of these cells is reportedly high (Palay and Chan-Palay, '77). We estimated that our Muller cells contained as much as 2-6-fold more glycogen than cultured CNS glial cells (Swanson et al., '90; Sorg and Magestretti, '91). This estimate should, however, be taken with caution, as glycogen content fluctuates with the glucose level in the medium (Passonneau and Crites, '76), as well as with the degree of retinal vascularization (Kuwabara and Cogan, '61). The accepted role of glycogen in the retina is that of a carbohydrate reserve that is used when external glucose concentration falls below need (Ripps and Witkovsky, '85). Because no glucose was provided to Muller cells during 3H-2DGincubations in this study and because 3H-2DGdoes not sustain glycolysis, it was expected that Muller cells would consume a portion of their glycogen in 30 minutes. We found that glycogen decreased by 50% in our glucosefree medium and that this glycogen decrease was not as large in IAA. The decrease in glycogen content in Muller cells during IAA-blocked glycolysis is not in disagreement with IAA exerting its major effect on G3PDHG (Lehninger, '82). When ATP production stops, two mechanisms may be operative: maintenance of phosphofructokinase (PFK) activity or activation of glucose-6-phosphatase. The first mechanism, which we favor, would result from a decrease of ATP (which decreases the K, for fructose-6-phosphate) and from the accumulation of fructose 1,6 diphosphate (Lehninger, '82). Thus, the activity of PFK would be maintained or even increased and draw on fructose-6-phosphate derived from the breakdown of glycogen. Glycosyl units would therefore degrade up to a step between 1,2 fructose diphosphate and glyceraldehyde-3P and this process would continue until ATP is exhausted. Although glucose-6-phosphatase activity cannot presently be excluded, this possibility remains tenuous because, as earlier mentioned, under normal (no glucose) conditions we could not detect significant levels of this activity (Poitry-Yamate and Tsacopoulos, '91) nor have others working on mammalian brain (Nelson et al., '87) and retina (Blair et al., '89) in vivo. Shortly after intravenous or intravitreal injections of IAA into anesthetized rabbits, early morphological changes were found exclusively in Muller cells (Babel and Stangos, '73). The most significant of these changes was a disappearance of P-glycogen granules. On the basis of our cell model, two explanations for this observation are possible: (1) glucose transport into the cell is affected, or as we would like to suggest, (2) hexokinase is inhibited and PFK activity continues to operate. We excluded the first possibility on the basis that, between IAA-exposed and control Muller cells in our study, levels of 3H-2DGwere not statistically different. The most likely possibility is that glucose is no longer phosphorylated while glycolysis still is able to generate glyceraldehyde-3P via glycosyl units derived from glycogen. In conclusion, this is the first report examining the metabolism of glucose in isolated retinal Muller glial cells. Evidence in situ for this metabolism was reported earlier (Poitry-Yamate and Tsacopoulos, '91) and studied here in a population of freshly isolated and purified Muller cells. These cells show an intensive phosphorylation of 3H-2DG that was exceptionally well demonstrated in light microscopic autoradiographs. The effect of IAA on 'H-2DG phosphorylation and the mechanisms for it were explored.

C.L. POITRY-YAMATE AND M. TSACOYOULOS Future experiments will be needed to determine the metabolic fate of glucose, and to identify the glucose transporter system and its location in Muller glial cells. This would be of utmost clinical interest because these cells appear to play an important role in the pathogenesis of diabetic retinopathy (Nork et al., ’87). For neuroscientists, this Muller cell model is well suited to studying the role of glial cells in the exchange of nutritional and metabolic materials with neurones (Selak et al., ’85; Katoh-Semba et al., ’88; Brazitikos and Tsacopoulos, ’91).

ACKNOWLEDGMENTS The authors thank Drs. Elisabeth Rungger and Francoise Assimacopoulos for providing helpful discussions throughout the project and to Drs. F. Assimacopoulos, C. Bader, E. Buchner, M. Dubois-Dauphin, A. Kato, S. Levy, S. Poitry and E. Rungger for critically reading the manuscript. Supported by the American Diabetes Association and the George Kernen Foundation.

LITERATURE CITED Babel, J., and N. Stangos (1973) Essai de correlation entre I’ERG et les ultrastructures de la rbtine. I. Action d u monoiodoadtate. Arch. Opht. (Paris). 33297-312. Blair, N.P., W.E. Shaw, and B. Yue (1989) Glucose-6-phosphatase activity in the retina of the awake rat. Invest. Ophthalmol. Vis. Sci. 30.2268-2271. Brazitikos, P.D., and M. Tsacopoulos (1991) Metabolic signalling between photoreceptors and glial cells in the retina of the drone (Apis mellifera). Brain Res. 567t33-41. Buchner, E., and S. Buchner (1980) Mapping stimulus-induced nervous activity in small brains hy [3H]2-deoxy-D-glucose. Cell Tissue Res. 211:51-64. Clarke, D.W., F.T. Boyd, M.S. Kappy, and M.K. Raizada(1984) Insulin binds to specific receptors and stimulates 2-deoxy-D-glucoseuptake in cultured glial cells from rat brain. J. Biol. Chem. 259:11672-11675. Cohen, L.H., and W.K. Noell (1960) Glucose catabolism of rabbit retina before and after development of visual function. J. Neurochem. 5253-276. Dowling, J.E. (1987) The Retina: An Approachable Part of the Brain. Cambridge, Mass: Harvard University. Eichner, D., and H. Themann (1962) Zur Frage des Netzhautglycogens beim Meerschweinchen. Zeitschrift fur Zell. 56.231-246. Goldman, S.S. (1990) Evidence that the gluconeogenic pathway is confined to an enriched Muller cell fraction derived from the amphibian retina. Exp. Eye Res. 50.913-218. Graymore, C. (1970) Biochemistry of the retina. In C. Graymore (ed): Biochemistry of the Eye. London: Academic Press, Inc., pp. 645-735. Kato-Semha, R., H. Keino, and S. Kashiwamata (1988) A possible contrihution by glial cells to neuronal energy production: Enzyme-histochemical studies in the developing rat cerebellum. Cell Tissue Res. 252:133-139. Krebs, H.A. (1972) The Pasteur effect and the relations between respiration and fermentation. In P.N. Campbell and F. Dickens (eds): Essays in Biochemistry, Biochemical Society, Vol. 8. London, N.Y.: Academic Press, pp. 1-34. Kuwabara, T., and D.G. Cogan (1960) Tetrazolium studies on the retina: 111. Activity of metabolic intermediates and miscellaneous substrates. J. Histochem. Cytochem. 8314-224. Kuwabara, T., and D.G. Cogan (1961) Retinal glycogen. Arch. Ophthalmol. 66:96-104. Lehninger, A.L. (1982) Principles of Biochemistry. New York: Worth Publishers, Inc. Lowry, O.H., N.R. Roberts, D.W. Schulz, J.E. Clow, and J.R. Clark (1961) Quantitative histochemistry of retina. J. Biol. Chem. 2369813-2820. Mar,H., and T.N. Wight (1988)Colloidal gold immunostaining on deplasticized ultra-thin sections. J. Histochem. Cytochem. 36:1389-1395. Matchinsky, F.M. (1970) Energy metabolism of the microscopic structures of the cochlea, the retina, and the cerebellum. In E. Costa and E. Giacohini

(eds): Biochemistry of Simple Neuronal Models. Advances in Biochemical Psychopharmacology. Raven Press, New York: vol2, pp. 217-243. Nahorski, S.R., and K.H. Rogers (1972) An enzymatic fluorometric micromethod for determination of glycogen. Anal. Biochem. 49:492-497. Nelson, T., G.A. Dienel, K. Mori, N.F. Cruz, and L. Sokoloff (1987) Deoxyglucose-6-phosphatestability in vivo and the deoxyglucose method: Response to comments of Hawkins and Miller. J. Neurochem. 49:19491960. Newman, E. (1987) Distribution of potassium conductance in mammalian Muller (glial) cells: A comparative study. J. Neurosci. 7:2423-2432. Nork, T.M., I.H.L. Wallow, S.J. Sramek, and G. Anderson (1987) Muller’s cell involvement in proliferative diabetic retinopathy. Arch. Ophthalmol. 105:1424-1429. Palay, S.L., and V. Chan-Palay (1977) General morphology of neurons and neuroglia. In J.M. Brookhart, V.B. Mountcastle, E.R. Kandel, and S.R. Geiger (eds): Handbook of Physiology. Section I: The Nervous System, Vol. I. Cellular Biology of Neurons, Part I. Bethesda, MD: American Physiological Society, pp. 5-37. Passonneau, J.V., and S.K. Crites (1976) Regulation of glycogen metabolism in astrocytoma and neuroblastoma cells in culture. J. Biol. Chem. 25:2015-2022. Poitry-Yamate, C.L., and M. Tsacopoulos (1991a) Glial (Muller) cells take up and phosphorylate L3HJ2-deoxy-D-glucose in a mammalian retina. Neurosci. Lett. 122241-244. Poitry-Yamate, C.L., and M. Tsacopoulos (1991h) Models for studying carbohydrate metabolism and function of retinal glial cells. Invest. Ophthalmol. Vis. Sci. 32:773. Polyak, S.L. (1941) The Retina. Chicago, Illinois: University of Chicago Press. Reichenbach, A,, H. Schneider, L. Leihnitz, W. Reichelt, P. Schaaf, and R. Schiimann (1989) The structure of rabbit retinal Muller (glial) cells is adapted to the surrounding retinal layers. Anat. Emhryol. 180:71-79. Ripps, H., and P. Witkovsky (1985) Neuron-glia interaction in the brain and retina. Prog. Retinal Res. 4:181-219. Saji, M., and K. Ohata (1981) Stimulus-dependent labeling of cultured ganglionic cell with [“C] 2-deoxyglucose. Brain Res. 212:435-446. Schnitzer, J. (1988) Astrocytes in the guinea pig, horse, and monkey retina: Their occurrence coincides with the presence of blood vessels. Glia l:74-89. Selak, I., S.D. Skaper, and S. Varon (1985) Pyruvate participation in the low-molecular-weight trophic activity for CNS neurons in glia-conditioned media. J. Neurosci. 5:23-28. Sols, A., and R.K. Crane (1954) Substrate specificity of brain hexokinase. J. Biol. Chem. 210:581-595. Sorg, O., and P.J. Magistretti (1991) Glycogen metabolism in primary astrocyte cultures. Experientia 47:A3. Sperling, H.G., E.S. Harcombe, and C. Johnson (1982) Stimulus-controlled labeling of cones in the macaque monkey with ’H-2-D-deoxyglucose. In J.G. Hollyfield and E.A. Vidrio (eds): The Structure of the Eye. New York, Elsevier North Holland, Inc., pp. 55-60. Swanson, R.A., A.C.H. Yu, P.H. Chan, and F.R. Sharp (1990) Glutamate increases glycogen content and reduces glucose utilization in primary astrocyte culture. J. Neurochem. 54:490-496. Thiery, J-P. (1967) Mise en evidence des polysaccharides sur coupes fines en rnicroscopie electronique. J. Microsc. 6:987-10 18. Trachtenherg, M.C., and D.J. Packey (1983) Rapid isolation of mammalian Muller cells. Brain Res. 261:43-52. Tsacopoulos, M.,V. Evequoz-Mercier, P. Perrottet, and E. Buchner (1988) Honeybee retinal glial cells transform glucose and supply the neurons with metabolic suhstate. Proc. Natl. Acad. Sci. U.S.A. 8523727-8731. Walker, P.S., J.A. Donovan, B.G. Van Ness, R.E. Fellows, and J.E. Pessin (1988) Glucose dependent regulation of glucose transport activity, protein, and mRNA in primary cultures of rat brain glial cells. J. Biol. Chem. 263t15594-15601. Wilson, J.E. (1984) Regulation of the mammalian hexokinase activity. Vol. 1. Boca Raton, Florida: CRC Press, pp. 45-85. Winkler, B.S. (1981) Glycolytic and oxidative metabolism in relation to retinal function. J. Gen. Physiol. 77:667-692. Witkovsky, P., and Chen-Yu Yang (1982) Uptake and localization of ‘H-2-deoxy-D-glucose by retinal photoreceptors. J. Comp. Neurol. 204: 105-116.