Metabolic Interactions between Neurons and Glial Cells


The vasculature of both the retina and brain can autoregulate, meaning that blood flow is altered in response to neuronal activity. This tight coupling between neuronal activity and blood flow, or neurovascular coupling, was first described in the brain more than a century ago by Roy & Sherrington. Retinal glia are not just passive supportive cells but rather they play an active role in directly changing neuronal activity. Glial cells sense neuronal activity and alter blood flow accordingly by directly communicating with the inner retinal vasculature. Thus, glia play a pivotal role not only in maintaining normal neuronal function but also in ensuring adequate retinal blood flow. , Even though glia are essential for normal retinal function, for the most part their roles are in response to neuronal function. Stimulation of retinal whole-mounts by light or direct glial stimulation led to either vasoconstriction or vasodilation of inner retinal blood vessels. In particular, these vascular caliber changes were linked to increases in intracellular calcium within retinal glia. Therefore, like their counterparts in the brain, retinal glia induce caliber changes in capillaries in response to neuronal activity.

Recently, with the advent of ion-sensitive fluorescent indicators and calcium imaging, it has emerged that glial cells can generate active responses and play a profound role in directly modulating both the activity of neurons and the vasculature. , Glia communicate with one another via increases in intracellular calcium in the form of a calcium wave that propagates from one astrocyte to another via gap junctions or by the release of bursts of extracellular ATP. ,

Studies in cell culture, brain slices and more recently, retinal whole-mounts have demonstrated that transmitters released from neurons induce transient elevations in intracellular calcium in glia. , Mechanical, chemical and light stimulation can evoke increases in intracellular calcium in both astrocytes and Müller cells that propagate to neighboring glia as a “wave”. It is well established that extracellular ATP evokes large increases in intracellular calcium within both astrocytes and Müller cells and that glia release ATP extracellularly in response to stimulation. The source of the calcium elevations in retinal glia is thought to be primarily from intracellular stores, although there is also a range of calcium-permeable channels, pumps and exchangers that could mediate calcium influx into glia from the environment.

The functional significance of the elevation in intracellular calcium within retinal glia is two-fold: direct modulation of neurons and alteration of vessel caliber. Studies in cortical glial cultures have shown that glia can release chemical transmitters such as ATP, glutamate and the NMDA receptor co-agonist d -serine in a calcium-dependent manner. , Moreover, these released “gliotransmitters” can directly alter neuronal function. For example, many types of retinal neurons are known to express receptors to ATP (called P2 receptors), including photoreceptors, amacrine cells and ganglion cells. Moreover, photoreceptor function is modulated by extracellular ATP. Therefore, it is possible that the release of ATP extracellularly by glia could in turn modulate a variety of neuron cell types from photoreceptors to ganglion cells. In regards to a glial-dependent modulation of vessel caliber, increases in intracellular calcium within cortical astrocyte endfeet are linked to marked vasodilation in the adjacent arteriole, suggesting a relationship between astrocytes and the vasculature within different regions of the central nervous system , in response to an increase in neural activity.

Energy for excitatory glutamatergic synaptic transmission in the mammalian retina, as elsewhere in the central nervous system (CNS), is provided by the metabolism of blood-borne glucose. Indeed, retinal preparations become synaptically silent as a result of glucose depletion. From the ultrastructural point of view, the remarkable concentration of mitochondria in synaptic terminals of axons indicates that glutamatergic synapses have high capacity for oxygen consumption and are major users of metabolic energy.

Although electrophysiological evidence shows that neurotransmission through the inner retina is supported by glycolysis there is presently no experimental evidence showing that synaptic activity of pre- and post-synaptic retinal neurons is directly sustained by glucose. Indirect evidence, however, is suggested from the classical work of Lowry and co-workers on the distribution of enzymes of glucose metabolism determined from pure samples of each retinal layer from monkey and rabbit. A brief introductory overview of this particular paper is given below as it offers invaluable insight to the contribution of Müller glia to overall retinal function and energy metabolism.

All enzymes of glycolysis are in the cytoplasm rather than in the mitochondria. To initiate glycolysis, hexokinase irreversibly phosphorylates glucose to glucose-6-phosphate (G6P). The distribution of hexokinase was confined to the layer containing inner segments of photoreceptors, the inner synaptic layer, and to the inner-most retinal layer bordering the vitreous. The second step of glycolysis is the conversion of G6P to fructose-6P by glucose-phosphate isomerase. This enzyme’s distribution was largely confined to the inner- and outer-synaptic layers. Phosphofructokinase, the third enzyme in glycolysis, irreversibly phosphorylates fructose-6P to fructose-1,6diP and its distribution was confined to both synaptic layers and the innermost retinal layer. The ninth enzyme in glycolysis, phosphoglyceromutase converts 3-phosphoglycerate to 2-phosphoglycerate and its distribution was similar to the distribution of phosphofructokinase.

In tissues with adequate oxygen supply, pyruvate formation is the 11th and last step of glycolysis. The metabolism of pyruvate for energy production consumes oxygen and completes the breakdown of glucose to CO 2 and water through the process of oxidative metabolism. Upon careful study, the distribution of the aforementioned glycolytic enzymes corresponds to the morphological position of Müller glial cells in situ. Müller cells extend radially through all of the retinal layers from the photoreceptor inner segments to the inner limiting membrane bordering the vitreous; and they extend fine filaments laterally in both synaptic layers. They also form an additional physical and functional cell layer to the diffusion of substances from the blood to neurons. Indeed, Kuwabara & Cogan , undertook the first comprehensive histochemical study to identify Müller cells as the primary glucose utilizing cells in the retina.

Together, these landmark studies paved the way to our present day understanding of cellular coupling in maintaining retinal function and metabolism, and which forms the major theme of this chapter. As a concluding introductory note, one is reminded that regardless of how the energy consumption of the retina (or other parts of the CNS) is altered locally to meet changing demands, it is of major importance to know from a neurophysiological viewpoint just which cell types and what cellular events are associated with local changes in blood flow, metabolism and tissue oxygenation.

Retinal oxygen distribution and consumption

Under normal conditions, O 2 is the limiting factor in retinal metabolism. Oxygen is used by mitochondria, and their distribution is important in understanding locations of high O 2 demand. The O 2 consumption (QO 2 ) of the retinal pigment epithelial (RPE) cells is about 20% of that of the retina per mg protein. Mitochondria are densely observed in the inner segments (IS) of photoreceptors. Cones have more mitochondria than rods. , Mitochondria are also found in each rod spherule and cone pedicle. In the inner retina, the inner plexiform layer has a larger amount of mitochondria than the nuclear layers.

Inner retina

The distribution of oxygen tension (P o 2 ) close to the vitreo–retinal interface is heterogeneous, being higher close to the arteriolar wall. Preretinal and transretinal P o 2 profiles indicate that O 2 diffusion from the arterioles affects the P o 2 in the juxta-arteriolar areas ( Fig 12.1 ). , O 2 reaches the vitreous by diffusion from the retinal circulation. In contrast, far from the vessels, the preretinal P o 2 remains constant and the average preretinal P o 2 from the vitreal side is similar to that measured in the inner retina. In the inner retina, P o 2 averages about 20 mmHg, but up to 60 mmHg close to the arteriolar wall.

Figure 12.1

Transretinal oxygen partial pressure (P o 2 ) profiles recorded in intravascular ( A ) and juxta-arteriolar ( B ) retinal areas in mini pigs. The intraretinal values indicate a progressive decrease of the tissue P o 2 from both the vitreoretinal interface (ILM) and the pigment epithelium (PE) toward the middle of the retina, with the minimum mean value recorded at 40% and 50% retinal depth. At the vitreoretinal interface, the higher P o 2 suggests that oxygen diffusing from the larger vessels reaches the inner retina. Each point is the mean ± SE of 13 measurements. The drawings indicate the pathway of the microelectrode through the retina. ILM: internal limiting membrane, RPE: retinal pigment epithelium.

(Modified from Pournaras CJ, Retina . 1995;15(4):332–47.)

Dark and light O 2 consumption

Inner retinal oxygen consumption was the same in light and darkness, indicating no influence of light adaptation as there is in the outer retina. In the inner retina, ganglion cells have much higher firing rates if a stimulus is presented repeatedly than if the same amount of light is delivered as a steady background. Consequently, one would expect the inner retina to use more energy when a stimulus is flickering. Indeed, there is, in response to a flickering stimulus, a higher lactate production in inner retina of rabbit than during darkness or steady illumination and a higher deoxy- d -glucose uptake in monkey retina.

Outer retina

Trans-retinal P o 2 measurements have provided data about the O 2 supply to the photoreceptors and their QO 2 . P o 2 profiles made in cat, , pig and monkey indicate that oxygen diffuses from the inner retina and from the choroid towards the middle of the retina, i.e. the outer plexiform layer (OPL). The choroidal circulation supplies about 90% of the photoreceptor’s O 2 use.

Photoreceptor QO 2 in darkness

In the dark-adapted retina, photoreceptor QO 2 (Q OR ) depends strongly on choriocapillary P o 2 in cat and monkey. , In the inner segments of photoreceptors (IS) the local value of QO 2 is about five times higher than in the outer segments (OS). A similar range for QO 2 in the outer retina was obtained in rat, rabbit and pig retinas. In both cat and monkey, the average value of P o 2 in the choriocapillaries is about 50 mmHg, and the corresponding average value of QO 2 in the outer retina is 4–5 mL O 2 /100g −1 /min −1 . ,

The average P o 2 value in cat was 5 mmHg, the minimum was frequently indistinguishable from zero in the dark as predicted by Dollery et al. The amount of oxygen consumed is an indirect measure of ATP synthesis and thus of ATP utilization. The ATP produced in the dark fuels many cellular processes, mainly the Na + /K + -ATPase in the IS, which extrudes a large amount of sodium that enters through the light-dependent channels in the OS. An additional process is the turnover of cGMP that holds these channels open. ,

As noted above, there is no evidence about whether individual rods and cones use different amounts of O 2 . Cones in the primate fovea appear to use slightly less O 2 than the parafoveal photoreceptors. ,

Photoreceptor QO 2 in light

QO 2 in the outer retina is lower in steady light than in darkness in various animals investigated. The activity of the Na + /K + -ATPase decreases in light, but the turnover of cGMP increases, , so the decrease in Q OR is not as great as the decrease in the pump rate. The maximum size of the overall change appears to be species-dependent.

Role of glycolysis underlying retinal function: from whole retina to its parts

Visualization of dynamic functional activity in the retina is an indirect measurement of neuronal activity. It is important to appreciate that any dynamic “functional imaging” of the central nervous system, e.g. fMRI or PET imaging, measures local changes in brain metabolism and physiology that are associated with neuronal activity. Therefore, examining and evaluating retinal function necessarily involves understanding the energy metabolism of this tissue.

The significance of glucose and its metabolism down the glycolytic pathway in mammalian retina is attested to by the measured high rate of aerobic glycolysis in vitro (high capacity for oxygen consumption), , its susceptibility to iodoacetate, a strong Pasteur effect (inhibition of glucose utilization) and the aerobic and anaerobic production of lactate.

The adult retina, as is the case for every CNS region, depends on an uninterrupted supply of blood-borne glucose. Under normal conditions, glucose is virtually the sole substrate supporting the intense energy metabolism required to maintain retinal function, e.g. normal electrical responsiveness to light and neurotransmission. Lactate generated from either anaerobic glycolysis or glycogenolysis within the retina has recently been proposed as another important energy source during synaptic transmission, but its uptake and metabolism by retinal cells has yet to be demonstrated in vivo. A critical evaluation of lactate use in the CNS has been presented.

The distribution of key enzymes in glucose metabolism through the individual neuronal and synaptic retinal layers and the dehydrogenases for several of the intermediate stages of glucose degradation in retina have been documented in a series of pioneering histochemical work and set the stage early for later metabolic studies of the glycolytic and oxidative capacity of this tissue. Kuwabara and Cogan’s studies suggested for the first time that the Müller glial cells may serve a significant metabolic role, in stark contrast to the general view at that time as having only a structural function. Indeed, of the two major cell types in retina – the Müller glia and photoreceptors – metabolic studies have demonstrated that Müller cells both in situ and when acutely isolated, preferentially and massively take up and phosphorylate glucose, part of which is stored as glycogen. , Further evidence to confirm this finding has been performed in vivo.

More recent supporting evidence comes from the use of iodoacetate (IAA), a well-known glycolytic poison that exerts its effect when the transformation of glucose is committed to proceed through glycolysis. In the 1950s, it was shown that intravenous delivery of sodium iodoacetate in rabbit, monkey and cat abolished the electrical response of the visual pathway to illumination within minutes, with a resulting histological picture similar to that presented in human retinitis pigmentosa. This led to the speculation that the initial effect must be on the visual cells, although Warburg suggested that the different cell types in the retina may not contribute equally to the general biochemical picture. Indeed, this suggestion has been unambiguously supported by the differential suppression of the component waves of the extracellular electroretinogram (ERG). , However, beyond this evidence, the identity of the retinal cell types taking up IAA remains unknown. Recently this was explored using synchrotron-based x-ray fluorescence of IAA at the cellular level in situ ( Fig. 12.2 ). The fluorescence map ( Fig. 12.A ) generated from the dark-adapted retina ( Fig. 12.2B ) showed that IAA was taken up specifically by Müller glia and not by retinal neurons, including photoreceptors, indicating that the effect of IAA on neurons is not direct but secondary to inhibiting glycolysis in glia (Poitry-Yamate 2009, unpublished results). Together, these results suggest a key role played by Müller glia in transporting glucose from the blood into the retina.

Figure 12.2

The glycolytic poison iodoacetate is localized to Müller glial cells and not to retinal neurons: indirect evidence of which cell type depends directly on glucose metabolism.

( A ) Synchrotron radiation-based hard x-ray fluorescence map (90 µm height × 20 µm width) of the period table element iodine from iodoacetate (orange) subsequent to scanning the retina from the inner (top) to outer (bottom) layers. Iodoacetate was localized in the retina to Region 1 comprising the endfeet of Muller glia. The metabolic experiment consisted of dark-adapting the tissue prior to a 50 min exposure ex vivo to bicarbonate buffered Ringer’s physiological solution carrying iodoaceate and d -glucose. Note that iodoacetate inactivates the glyceraldehyde-3-phosphate dehydrogenase reaction and therefore inhibits the 6th step in glycolysis. ( B ) Phase contrast image of the retina from which the fluorescence map shown in A was obtained. Region 1, Müller glial endfeet; Region 2a, inner plexiform layer; Region 2b, inner nuclear layer; Region 3a, photoreceptor layer; Region 3b, photoreceptor segments. ( C ) Reference, methyl-blue stained coronal section used for identifying individual retinal layers. ( D ) X-ray fluorescence energy emission spectra from Region 1. An La energy emission at 3.9 keV is specific to iodine from iodoacetate (blue trace). The y-axis expresses the number of fluorescence photons emitted from iodine subsequent to exposing the preparation to synchrotron ligtt.

Acutely isolated mammalian photoreceptors produce 14 CO 2 from 14 C(U)-glucose while photoreceptor outer segments produce both lactate from glucose and 14 CO 2 from 14 C(U)-glucose. These authors interpreted their results as indicating that both glycolysis and the pentose phosphate pathway contribute to maintaining photoreceptor function. Considering that only one in six carbons from 14 C(U)-glucose is converted to 14 CO 2 through the pentose phosphate pathway, and the abundance of photoreceptor mitochondria, it is likely that the 14 CO 2 reflects the rate of mitochondrial respiration. Photoreceptors of some species do not express Gpi 1, the enzyme catalyzing the isomerization of glucose-6-phosphate to fructose-6-phosphate, leaving the phosphate pentose pathway as the only possible downstream path, with a gain of 2 NADPH molecules potentially serving as reducing agent for the reduction of retinaldehyde.

A number of studies of glucose metabolism have been undertaken in intact retinal tissue or with acutely isolated cell models in vertebrate/mammalian retina. Of these, one is unique for studying not only glucose metabolism but also metabolic compartmentation: the cell model of acutely isolated Müller cells still attached to photoreceptors (termed “the cell complex”) shown in Figure 12.3 and further discussed in sections 4, 6 and 8. This study confirmed not only the previous work by Kuwabara and Cogan, but showed for the first time in a mammalian preparation of the CNS tissue that glial cells transform rather than simply transfer the primary energy substrate glucose and supply neurons with a glucose-derived metabolite.

Figure 12.3

Acutely isolated cell models for exploring retinal function, metabolism and the trafficking of metabolites between glial cells and photoreceptor-neurons.

( A ) Honeybee drone, illustrating its crystalline-like structure of 6 photoreceptors that form a rosette surrounded by glia. The outline of the extracellular space between glia is seen as web-like. The compartmentation of glycolysis to glia and of oxidative metabolism to photoreceptors renders this CNS model unique for studying neuron-glial interactions

(modified from Tsacopoulos et al, Proc Natl Acad Sci USA 1998; 85(22): 8727–31).

Cell culture models of transformed rat Müller cells, human RPE and transformed mouse photoreceptor cells and ganglion cells were all found to produce lactate, aerobically and anaerobically in the presence of 5 mM glucose. This may not be surprising as culturing techniques influence cell metabolism and function. The composition of culturing medium may be a key underlying factor to explain why cells are predominantly glycolytic, irrespective of cell type. However, the production of lactate by these cells did not significantly differ with the addition of 10 mM lactate. In general, lactate production and its release into the extracellular space as a lactate anion plus a proton (H + ) creates an extracellular pH gradient, i.e. increased proton concentration and consequently pH values become less than 7.4. Depending on the magnitude, direction and time course of the extracellular pH gradient, lactate may accumulate extracellularly, or alternatively, lactate may be taken up on a proton-linked monocarboxylate transporter. In this context, transformed, cultured neurons did not transport or metabolize exogenous lactate, consistent with a lactate-containing solution adjusted to a pH of 7.4 rather than a pH of <7.4.

Biochemical specialization of glial cells

The major glial cell type in vertebrate retina is the radial Müller glial cell (also termed radial fibers or sustentacular cells of Heinrich Müller). Structurally, they are elongated, possess a prominent specialized region called endfeet at the inner limiting membrane, and are vertically oriented with respect to the retinal layers ( Fig. 12.4 ). Since Müller cells extend through the synaptic and nuclear layers of the retina from the inner to outer limiting membranes they are in intimate apposition to every neuron cell type. Müller glia also serve as an additional physical and functional cell layer to the diffusion of substances into and out of the extracellular space, the vitreous, the subretinal space and retinal vascular supply. Histochemical evidence has shown that glycogen synthesis, glycogenolysis and anaerobic glycolysis are localized to Müller glial cells in situ. This was confirmed and quantitated in living intact, dark-adapted retina using biochemical and autoradiographic methodologies , and provided strong experimental evidence for the working hypothesis of all retinal cell types, Müller glia play a major role beyond the blood–retinal barriers in transporting glucose from the blood into the neural retina. However, once in the neural retina, it remains to be shown along the Müller cell’s entire radial length whether the distribution of transporters related to energy substrate uptake and release are tailored to this cell’s own metabolic needs, but yet adapted to the function and metabolic needs of their immediate neuronal environment.

Figure 12.4

Müller glia in vertebrate retina.

( A ) Methyl-blue stained retinal section highlighting the large endfeet and radial structure of Müller glia (Mü) through the thickness of the retina. Laminar organization of this tissue allows for clear identification of nuclear and synaptic layers. G, ganglion cell; AC, amacrine cell; BC, bipolar cell; HC, horizontal cell; IS and OS, inner and outer segments of photoreceptors. Arrow at top indicates direction of light hitting the retina; *, synaptic layers. Müller endfeet form the vitreoretinal interface. ( B ) Müller cell after acute isolation and purification; approximately to scale with Müller cells shown in (A). The descending radial process (Z), radial strands (y) and terminal angular buttons (x) of the Müller cell are landmarks of that part of the Müller cell in contact with the outer synaptic layer, and the cell body and inner segments of photoreceptor, respectively

(modified from Poitry-Yamate et al, J Neurosci 1995;15(7 Pt 2): 5179–91).

Two functional and biochemical specializations unique to Müller glia are their capacity to inactivate the excitatory neurotransmitter glutamate , and inhibitory neurotransmitters GABA and glycine. They are the exclusive and/or predominant cellular site of:

  • (1)

    glutamine synthetase activity for the synthesis and release of glutamine, a precursor for photoreceptor neurotransmitter resynthesis; and

  • (2)

    carbonic anhydrase for the conversion of water and CO 2 of neuron origin to bicarbonate, an enzymatic activity implicated in the regulation of intracellular and extracellular pH and volume.

Role of glycogen

As an endogenous source of glucose-6-phosphate, glycogen, as well as glycogen phosphorylase, are exclusively localized to the cytoplasm of Müller glial cells in situ in a variety of mammalian species. This important study suggests that Müller cells can effectively mobilize this energy store but it remains unclear whether it is for the purpose of meeting their own energy needs, or alternatively partly those of the surrounding neurons in the form of glycogenolytically generated lactate one isoform of glycogen phosphorylase is expressed in cone photoreceptors (brain type), while another isoform (muscle type) is expressed in the inner plexiform synaptic layers of primate retina. So, although glycogen metabolism in Müller glia is established (e.g. see Ripps & Witkovsky for a review), the function of Müller cell glycogen is still far from clear. The prevailing view is that glycogen plays the role of an emergency retinal carbohydrate reservoir supporting neuron function when glucose delivery is compromised such as during retinal ischemia. This is in contrast to another view that Müller cell glycogen is mobilized as an immediate and accessible energy store under normal physiological conditions ( Fig. 12.5 ), such as changes in illumination, i.e. light and darkness, and which are linked to the direct effects of altering photoreceptor function. , In addition, glycogenolysis can be stimulated in the retina by neurotransmitters increasing cAMP, such as vasoactive intestinal peptide, which are contained in and released from amacrine cells. Glycogen is also localized in neurons of the cat retina, particularly of the rod-driven pathway, whose rod-driven components are selectively sensitive to prolonged hypoglycemia, , but whether this glycogen is mobilized is not presently known.

Figure 12.5

Lactate released by Müller glial cells is formed from exogenous radiolabeled glucose in the light-adapted cell complex but from glycogen in the dark-adapted cell complex.

The preparation shown in Fig. 12.3B was maintained either in darkness or light before exposure to uniformly radiolabeled glucose ( 14 C(U) glucose) with the aim to determine the contribution of glucose and glycogen to the production of lactate when modulating photoreceptor metabolism and neurotransmitter release. It is known that: (1) photoreceptors release neurotransmitter glutamate and that their metabolism is increased in darkness; and (2) glutamate release stimulates glycolysis and production of lactate. The results shown are simplified and drawn as a pie to reflect changes both inside the cells (top) and in the surrounding bath (bottom). Pie size represents the total pool size of lactate, i.e. radiolabeled + non-radiolabeled lactate. The size of the wedge represents the specific activity of lactate, so the larger the wedge, the larger is the contribution of exogenous glucose to the formation of lactate. The direction of the solid arrow in the left panel indicates that the amount of radiolabeled lactate (pink) in the bathing solution was much greater than that inside the preparation. This is only possible when lactate is produced from 14 C-glucose and is released earlier than non-radiolabeled lactate formed from glycogen. In the right panel, the direction of the solid arrow indicates that the amount of non-radiolabeled lactate (blue) in the surrounding bath greatly increased. This is only possible when glycogen (gly) in glia is mobilized to produce additional, unlabeled lactate.

Functional neuronal activity and division of metabolic labor

A continuous supply of blood-borne glucose is vital to maintaining retinal function, but does not necessarily dictate that all retinal cells (both neurons and glia) take up and metabolize this energy substrate as is generally believed. Indeed, the conventional view of glucose metabolism in brain is that glucose is the principal substrate for oxidative metabolism in both cell types, i.e. neurons and glia. The experimental evidence described in the previous sections is a major departure from this view and raises three highly controversial and still unresolved questions about the brain and about the generality of the findings in retina to other parts of the central nervous system:

  • (1)

    is glucose, the major brain energy substrate, taken up in a cell-type specific manner?

  • (2)

    does glycolysis predominate in one particular cell type, physiological condition or brain region and oxidative metabolism of glucose in another? and

  • (3)

    does coupling of metabolic and physiological changes to changes in neuronal activity involve the transformation of blood glucose by glial cells and do they supply neurons with glucose-derived metabolite(s)?

Underlying these issues is the idea of a predominant division of metabolic labor between neurons and glia. In other words, energy substrate production and substrate use is partitioned in a relatively cell-specific manner. This working hypothesis, summarized in Figure 12.6 , was developed and tested in retina in the mid-1990s, in insect and mammal, and has seen a revival in recent years, particularly with regards to the recognition that the coordinated action of glial cells and neurons extends to energy metabolism throughout the central nervous system.

Figure 12.6

Proposed scheme of retinal function, from bee to mammals, highlighting metabolite trafficking between glial cells and photoreceptors.

The honeybee drone model is shown in ( A ) and the mammalian model is shown in ( B ). Both models display relative metabolic compartmentation: glycolysis to glia and oxidative metabolism to photoreceptor-neurons. Commonalities of the models include production of a glycolytic product, i.e. lactate or alanine, their release extracellularly and uptake by photoreceptors; maintenance of the redox potential in glia, i.e. NADH/NAD ratio; and photoreceptor release of glutamate that acts to stimulate glial glycolysis. Glutamate is thus a chemical signal that turns the nourishing of neurons by glia into a function, instead of a passive process. In (B), the cytosol is indicated as C ; mitochondria are indicated as M .

As will be developed in the next sections, the major theme is that glial cells transform rather than simply transfer the primary energy substrate glucose, and supply activated neurons with a glucose-derived metabolite, i.e. lactate and/or alanine. In this regards, the vertebrate retina has proven to be a CNS model of choice because its laminar organization of metabolism and blood flow lends itself to the study of compartmentation by virtue of its structure. It should, however, be recognized that the high specialization of this nervous tissue, e.g. phototransduction and high energy metabolism, makes it different from other regions of the CNS and any comparisons must be made with caution. Moreover, the retina has two barriers, the retinal capillary endothelial cells of the inner blood–retinal barrier (BRB), and the retinal pigment epithelium (RPE) comprising the outer BRB which is located between the photoreceptors and choroid. Thus, their expression and distribution of glucose transporters and glycolysis will inevitably determine whether one, or alternatively both, of these barriers is the rate-limiting step for glucose delivery to the neuroretina.

Cellular compartmentation of energy substrates other than glucose

One distinctive property of the mammalian retina is its large production of endogenous lactate in the presence and absence of oxygen. The production of lactate is a normal function of retinal tissue and can effectively replace glucose in oxygenated Ringer’s solution in maintaining retinal oxidative metabolism and photoreceptor function. It was only in the mid-1990s that this phenomenon was explored at the cellular level and when the idea was launched that energy metabolism underlying retinal function in mammals was relatively compartmentalized, and moreover orchestrated between glial cells and neurons.

The use of high-resolution light microscopic autoradiography of 3 H-2DG in the dark-adapted retina in situ, coupled to HPLC for the identification of silver grains in autoradiograms showed that glucose is not taken up by the majority of retinal cells but preferentially taken up and phosphorylated by the Müller glia ( Fig. 12.4C,D ), leaving the question as to the identification of the energy substrate used by neurons. Using a cell model of acutely isolated Müller cells, the metabolic fate of glucose-6-phosphate and the identity of metabolites released by these cells was first assessed by Poitry-Yamate and Tsacopoulos. The experimental results of that study raised the hypothesis that lactate, synthesized and released in situ by a retinal cell type having high glycolytic capacity, may be transferred to and metabolized by photoreceptors that possess a high respiratory capacity. This working hypothesis was tested in the acutely isolated retinal model of the cell complex (see Fig. 12.2B ) comprising Müller cells still attached to photoreceptors, and where the effect of altering illumination on glucose metabolism was quantitated. A major finding of that study (see section 8 below) showed that lactate formation versus lactate use was cell-type dependent. As already discussed in section 4, sources of this glial lactate were from exogenous glucose, or alternatively, from endogenous glycogen ( Fig. 12.5 ). The functional role for the release into the extracellular space of lactate and H + by Müller glial cells during either glycolysis or glycogenolysis is expected to prevent intracellular acidification and maintain the regeneration of NAD + for glycolysis to proceed. Lactate and other glucose-derived metabolites released from the Müller glia would participate in the regeneration of the excitatory neurotransmitter glutamate, released by photoreceptors in darkness.

Within a larger framework, the current debate in the neuroscientific community is whether in other parts of the CNS, a net production and extracellular release by glial cells of glucose/glycogen-derived lactate provides sufficient fuel for activated neurons in vivo. NMR studies on brain function and energy metabolism have in recent years provided indirect support of a lactate flux from glia to neurons as providing the coupling mechanism between increased fluxes of the glucose carbon into the glycolytic pathway in glia, and into the oxidative pathway in neurons, in the form of lactate, during glutamatergic/excitatory synaptic transmission. Unraveling the current debate surrounding the glial-neuron lactate shuttle hypothesis (ANLSH) , in the CNS at large is a key step towards: (1) understanding the regulation of retinal and cerebral energy metabolism during neuronal activity; and (2) interpreting 18 FDG PET and fMRI images in both clinical medicine and fundamental neuroscience, and for which there is accumulating evidence for the central importance of glial cells in brain imaging signals.

Experimental models used to study the interaction between photoreceptors and glial Müller cells

In vitro studies of the retina of the honeybee drone

The most comprehensive and quantitative experimental tests in the literature demonstrating that glial cells supply neurons with a metabolic substrate to sustain neuron function were first formulated and established in the retina of the honeybee drone. This elegant crystalline-like CNS preparation is characterized by mitochondrial-rich, glycogen-poor photoreceptors grouped in rosette-shaped clusters and their surrounding mitochondrial-poor and glycogen-rich glial cells (see Fig. 12.3 ), indicating that glycolysis/glycogenolysis is largely confined to glia and that oxidative metabolism is largely confined to photoreceptors. What follows is a historical and chronological overview of the major experimental questions, experimental results and interpretations.

If there are no conventional synapses in drone retina and only the photoreceptors are directly excitable by light, what is the evidence that photoreceptors depend on surrounding glia for their metabolic needs?

In retinal drone slices, three lines of evidence argue that photoreceptors depend on the surrounding glial cells for their metabolic needs:

  • (1)

    Anoxia rapidly abolishes the receptor potential of photoreceptors stimulated every 5s by a light flash ( Fig. 12.7A ). However, simultaneous monitoring of the pH in the extracellular space changed by less than 0.02 units ( Fig. 12.7B ), in contrast with that of many other tissues where hypoxia leads to anaerobic glycolysis and acidification. These results indicated that anoxia does not lead to anaerobic glycolysis in drone retinal slices and that photoreceptor energy metabolism is obligatorily aerobic.

    Figure 12.7

    Characterization of photoreceptor metabolism and function in honeybee drone: evidence that photoreceptors metabolically depend on surrounding glia.

    ( A ) The effect of anoxia on abolishing receptor potential in photoreceptors but as shown in ( B ). The absence of an extracellular acidification as would be expected if anaerobic glycolysis is activated, indicates that photoreceptor energy metabolism is obligatorily aerobic. ( C ) The effect of blocking mitochondrial respiration (noted as 4) that uses a carbohydrate substrate reversibly suppressed the light-induced consumption of oxygen by photoreceptors. C and R are control and reversibility responses, respectively. This result indicated that the substrate of photoreceptor metabolism is obligatorily a carbohydrate. ( D ) The glucose analog, 2-deoxy- d -glucose (2DG) leads to a decrease in the partial pressure of oxygen (P o 2 ), so had no effect on the light-induced increase in oxygen consumption in photoreceptors. This result indicated that the continued function of photoreceptors does not depend on their direct uptake of glucose. The effect of the glycolytic poison, iodoacetate (IAA) on oxygen consumption and thereby of the partial pressure of oxygen in bee photoreceptors is not direct as glycolysis takes place only in the glia in this preparation (see Fig. 12.7A ).

  • (2)

    Amobarbital (or amytal) a specific blocker of mitochondrial respiration that uses carbohydrate substrate reversibly suppresses the light-induced change of oxygen consumption by photoreceptors ( Fig. 12.7C ), indicating that the substrate of photoreceptor metabolism is a carbohydrate.

  • (3)

    Superfusing slices of drone retina with the glucose analog 2-deoxy- d -glucose (2DG) had no effect on the light-induced increase of QO 2 , as monitored by decreases in P o 2 ( Fig. 12.7D ). As 2DG is transported into cells, phosphorylated, and little of it is metabolized further, 2DG6P accumulates intracellularly. This indicates that the continued function of photoreceptors does not depend directly on their uptake of glucose.

When bee retinal slices are exposed to the glycolytic poison IAA, the light-induced change in QO 2 is gradually abolished. Is this modulation of QO 2 a direct effect of IAA in photoreceptors?

IAA exerts its effect in cells where glucose uptake and metabolism are committed to the glycolytic pathway. Identifying whether the cellular site of glucose phosphorylation is cell-type specific is a first step towards this answer. The distribution and density of black grains biochemically identified as 3 H-2DG-6-phosphate ( H-2DG6P) in light microscopic autoradiographs ( Fig. 12.8 ) of bee retinal slices exposed to tritiated 2DG provides this answer. Transport and phosphorylation of this sugar analog are localized in situ to the glia and absent in photoreceptors. Hence, IAA has obligatorily exerted its direct effect on the metabolism of glucose in glial cells, which in turn lead to downstream effects on the oxygen consumption in photoreceptors.

Jan 23, 2019 | Posted by in OPHTHALMOLOGY | Comments Off on Metabolic Interactions between Neurons and Glial Cells

Full access? Get Clinical Tree

Get Clinical Tree app for offline access