Functional glutamate transport in rodent optic nerve axons and glia

Recent findings suggest that synaptic‐type glutamate signaling operates between axons and their supporting glial cells. Glutamate reuptake will be a necessary component of such a system. Evidence for glutamate‐mediated damage of oligodendroglia somata and processes in white matter suggests that glutamate regulation in white matter structures is also of clinical importance. The expression of glutamate transporters was examined in postnatal Day 14–17 (P14–17) mouse and in mature mouse and rat optic nerve using immuno‐histochemistry and immuno‐electron microscopy. EAAC1 was the major glutamate transporter detected in oligodendroglia cell membranes in both developing and mature optic nerve, while GLT‐1 was the most heavily expressed transporter in the membranes of astrocytes. Both EAAC1 and GLAST were also seen in adult astrocytes, but there was little membrane expression of either at P14–17. GLAST, EAAC1, and GLT‐1 were expressed in P14–17 axons with marked GLT‐1 expression in the axolemma, while in mature axons EAAC1 was abundant at the node of Ranvier. Functional glutamate transport was probed in P14–17 mouse optic nerve revealing Na+‐dependent, TBOA‐blockable uptake of D‐aspartate in astrocytes, axons, and oligodendrocytes. The data show that in addition to oligodendroglia and astrocytes, axons represent a potential source for extracellular glutamate in white matter during ischaemic conditions, and have the capacity for Na+‐dependent glutamate uptake. The findings support the possibility of functional synaptic‐type glutamate release from central axons, an event that will require axonal glutamate reuptake. © 2008 Wiley‐Liss, Inc.


INTRODUCTION
High-affinity, Na 1 -dependent glutamate transporters are responsible for clearing synaptically released glutamate from the extracellular space of the CNS (Anderson and Swanson, 2000;Danbolt, 2001;Gegelashvili and Schousboe, 1998). Recent findings suggest that synaptictype glutamate release is present in CNS white matter (Kukley et al., 2007;Ziskin et al., 2007). Effective glutamate reuptake mechanisms must exist in white matter if this is the case. Of the five glutamate transporters that have been cloned, three are widely distributed in the mature CNS (Danbolt, 2001;Furuta et al., 1997b;Gegelashvili and Schousboe, 1998;Rothstein et al., 1994).
Glutamate-aspartate transporter (GLAST or EAAT1) and glutamate transporter 1 (GLT-1 or EAAT2) are commonly known as glial glutamate transporters, whereas excitatory amino acid carrier 1 (EAAC1 or EAAT3) is regarded as the principal transporter in neurons.
In addition to the physiological importance of white matter glutamate reuptake, the voltage and Na 1 -dependence of these glutamate transporters can result in the reversal of transport under ischaemic conditions, which may act as one source of the extracellular glutamate that is responsible for excitotoxic cell injury during disorders such as stroke (Anderson and Swanson, 2000;Attwell et al., 1993). In addition to injury in gray matter regions of the brain, stroke affects white matter regions that are made up of axons, astrocytes, and oligodendrocytes (Stys, 2004). Excitotoxic injury has also been implicated in other forms of white matter disease such as spinal cord injury and multiple sclerosis (Li et al., 1999;Matute et al., 2001). Oligodendrocyte damage is central to loss of function in ischaemic white matter injury and is thought to result from glutamate-mediated excitotoxicity (Dewar et al., 2003;Pantoni et al., 1996;Tekkok and Goldberg, 2001).
In addition to adult disease, white matter injury is a particularly important component of the lesions that underlie neuro-developmental disorders such as cerebral palsy, which in many cases appear to be ischaemic in origin (Kinney and Back, 1998;Volpe, 2001Volpe, , 2003. Ischaemia evokes significant rises in extracellular glutamate concentration in white matter (Chiu and Kriegler, 1994;Shimada et al., 1993), and recent immuno-electron microscopy studies suggest that glutamate is released from both astrocytes and axons in ischaemic developing white matter . There is little information about the type of transporters mediating glutamate uptake in white matter tracts of the central nervous system. Prior studies have shown EAAC1, GLAST, and GLT-1 expression in mature white matter (Choi and Chiu, 1997;Furuta et al., 1997b;Kugler and Beyer, 2003), although the cellular distribution of these transporters is controversial, and little is known regarding their cellular distribution in developing white matter. Several studies have shown GLT-1 expression in neonatal central axons using light microscopy (Furuta et al., 1997b), but the close apposition of axonal and glial membrane casts the axonal expression of this ''glial'' transporter into doubt. However, membrane expression of GLT-1 in embryonic spinal cord axons before the appearance of astrocytes has been shown using immuno-electron microscopy (Yamada et al., 1998). Recent studies have shown that interruption of glutamate transport in white matter leads to excitotoxic injury of axons and glia (Domercq et al., 2005), highlighting the importance of glutamate clearance in these structures.
Here we use transgenic mice, where EGFP expression is under the control of glial cell-type specific promoters, coupled with standard and electron immuno-histochemistry to examine glutamate transporter expression in the P14-17 and mature optic nerve. Staining for the uptake of exogenously perfused D-aspartate and for the localization of glutamate was used to confirm that transporter expression was functional, revealing high levels of transporter expression in glia and in particular within axons.

MATERIALS AND METHODS Ethical Approval
All animal procedures were approved by local ethical review and conformed to UK home office regulations.

General
Transgenic mice (FVB/N) carrying the EGFP coding sequence under the control of CNP promoters 1 and 2 (Yuan et al., 2002) were kindly donated by the laboratory of Vittorio Gallo (CNMC Research, Washington DC). Heterozygous males were mated with wild-type females and transgenic littermates identified. A tendency for epileptic seizures in the colony was largely removed by out-breading with wild-type animals and subsequent back-crossing. Transgenic mice (FVB/N) with GFP under control of the GFAP promoter were obtained from The Jackson Laboratory (Bar Harbour, Maine). GLT-1 and GLAST knockout mice were a generous gift from David Attwell's laboratory (University College London), together with wild-type littermates from the GLT-1 mice. All animals were maintained in accordance with local ethical guidelines.

Immuno-Histochemistry
For immuno-histochemistry, wild-type optic nerves from P14-17 mice were dissected in 0.1 M PBS and fixed in 4% paraformaldehyde for 30 min. A minimum of three nerves from three different animals were analyzed for each age/antibody. Nerves from >P30 animals are termed ''adult'' throughout. The optic nerves were subsequently incubated in 0.1 M PBS plus 20% sucrose w/v for 5 min prior to freeze-sectioning and subsequent blocking for 60 min in 0.1 M PBS 10% fetal goat serum plus 0.5% Triton-X 100 and incubated in this solution plus primary antibody at 4°C overnight. Affinity-purified rabbit polyclonal antibodies against GLT-1, GLAST, and EAAC1 together with specific blocking peptides were obtained from Alpha Diagnostics (San Antonio, TX) and were used at a 1:100 dilution. The amino acid sequences used for production of all the transporter antibodies used in this study are shown in Table 1 (GLT-1 (1) and GLAST (1) are the above-mentioned antibodies). Mouse monoclonal antibodies against neurofilament-70 (NF-70, 1:200) were obtained from Chemicon Europe (Southampton, UK). Affinity-purified rabbit polyclonal antibody raised against D-aspartate (1:2,000) was obtained from Cell Sciences (Canton, MA) and has a cross-reactivity ratio of 1:10,000 for L-aspartate determined by ELISA (information provided by Cell Science). Rabbit polyclonal anti CNPase was obtained from Sigma UK (1:100). The appropriate Alexa-conjugated secondary antibody (Cambridge Bioscience, UK) was applied for 120 min following washing, and single-plane fluorescent sections were imaged at 360 on an Olympus scanning confocal microscope. Slow scanning and image averaging was required to resolve individual axons, which range from 0.1 to 1 lm in diameter at this age. The resulting degree of photo-bleaching confounded attempts to collect image stacks. Alexafluor-568 conjugated secondary antibodies were employed when labeling transgenic mice to allow spectra discrimination between the label and EGFP (emission maxima at 520 nm). In all cases, each staining protocol was performed on a minimum of three optic nerves from three separate animals.

D-Aspartate Uptake
For D-aspartate uptake studies, optic nerves from wild type and transgenic mice were dissected and placed in artificial cerebrospinal fluid (aCSF), composition (in mM): NaCl, 126; KCl, 3; NaH 2 PO 4 , 2; MgSO 4 , 2; CaCl 2 , 2; NaHCO 3 , 26; glucose, 10; pH, 7.45, bubbled with 5% CO 2 /95% O 2 and maintained at 37°C. This solution contained 500 lM D-aspartate and trials indicated that a 60 min incubation period produced good cell loading. Shorter periods of incubation failed to produce reliable D-aspartate staining and the relatively prolonged incubation period presumably reflects the slow penetration of D-aspartate into this whole-mount preparation. A very similar protocol has been previously used in the isolated lamprey spinal cord, a preparation that is physically similar to the isolate rodent optic nerve (Gundersen et al., 1995). Given the relatively long incubation, it is possible that D-aspartate will be partially metabolized in the optic nerve, although the amino acid is thought to be largely nonmetabolizable (Bender et al., 1997). Following the incubation, optic nerves were washed for 5 min in aCSF and fixed in 3% glutaraldehyde prior to processing for immuno-histochemistry (see above). Zero-Na 1 aCSF had NaCl replaced with N-methyl-D-glucamine, NaH 2 PO 4 replaced with KH 2 PO 4 , KCl reduced to 1 mM and NaHCO 3 replaced with choline-HCO 3 (pH 7.45). All chemicals were obtained from Sigma (UK).

Electron Microscopic Immuno-Histochemistry
For pre-embedded glutamate transporter electron microscopic immuno-histochemistry in adult white matter, we used five Sprague-Dawley rats. Deeply anesthetized animals were transcardially perfused with 500 mL of fixative containing 4% paraformaldehyde, 0.1% glutaraldehyde, and 0.2% picric acid in 0.1 M phosphate buffer (pH 7.4). Both longitudinal and transversal vibratome floating sections were preincubated in 10% goat serum in PBS (blocking solution) for 1 h. Sections were then incubated with the anti-GLAST (5 lg/mL), anti-GLT-1 (0.4 lg/mL), or anti-EAAC1 (1.14 lg/mL) antisera diluted in blocking solution for 3 days at 4°C with continuous shaking. Antibodies are listed in Table 1 for use with rats (marked ''2''). Since the epitopes targeted in these experiments are intracellular, the antibodies are likely to stain only transected cells. In negative control sections, the primary antibody was omitted. After sev-eral washes in PBS, tissue sections were incubated in 1.4 nm gold-labeled goat anti-rabbit IgG (Fab' fragment, 1:100 in blocking solution; Nanoprobes) for 4 h. Following washes, gold particles were silver-intensified with an HQ Silver Enhancement kit (Nanoprobes) for about 10 min. Tissue sections were postfixed with 1% osmium tetroxide, dehydrated, and embedded in epoxy. Ultrathin sections were counter-stained with uranyl acetate and lead citrate and examined using a PHILIPS EM208S electron microscope. To estimate the density of silverintensified particles present in each cellular element (astrocytes, oligodendrocytes, myelin, axon tracts, and nodes of Ranvier), electron microphotographs were taken. Scion Image software (NIH, Frederick, MD) was used to measure the area and the number of immunoparticles within cells (excluding those in mitochondria and nuclei). For membrane counts, only gold particles within 30 nm of the membrane were included. Particles located within mitochondria and/or nucleus were assumed to reflect the level of background staining and the density of staining in these structures was subtracted from the membrane counts.
For postembedded glutamate, glutamate transporter and D-aspartate (nerves incubated at 500 lM for 60 min) electron microscopic immuno-histochemistry optic nerves from P14-17 wild-type mice were postfixed in 3% glutaraldehyde/Sorenson's buffer. Nerves were postfixed with 2% osmium tetroxide and dehydrated prior to infiltration in epoxy. Sections were counterstained with uranyl acetate and lead citrate and examined with a Jeol 100CX electron microscope, see  for further details. Antibodies and concentrations were the same as listed above for mice light immuno-histochemistry at this age (marked ''1'' in Table 1), with the exception that the anti-D-aspartate was used at a concentration of 1:1,000. Secondary antibodies were goat antirabbit conjugated to 30-nm gold particles (British Biocell).

Cell Type Identification
Astrocytes in the developing optic nerve can vary widely in somata size and may not contain the typical glial filaments, but can generally be identified by the presence of a wide-bore endoplasmic reticulum, glycogen granules, a relatively dark cytoplasm, and an irregularshaped nucleus with chromatin clumped under the envelope. Astrocytes also extend processes to form the glial limitans and the glial end feet of capillaries. Oligodendroglia generally have larger somata and may have myelinating axons embedded in the somata or processes. Other features used to identify cells of this lineage include a relatively light cytoplasm, numerous large mitochondria, often at one pole of the somata, large nuclei of an oval shape containing evenly dispersed chromatin, the presence of two nucleoli, narrow bore endoplasmic reticulum, and the absence of astrocyte features such as glycogen granules. Axons can be reliably identified by the presence of neurofilaments and microtubules. For further details and examples of cell identification in developing optic nerve using similar fixing and staining protocols, see Wilke et al. (2004) and Thomas et al. (2004).

Statistics
Statistical differences were assessed by Student's ttest or ANOVA test as appropriate. Particle count data are presented as mean 6 SEM.

High Affinity Glutamate Transport Expression in
P14-17 Mouse Optic Nerve EAAC1, GLAST, and GLT-1 were all detected in wildtype P14-17 optic nerve by antibody staining (Fig. 1, left panels). In each case, the staining was ablated by incubation with a specific blocking peptide ( Fig. 1, right panels). Antibody specificity was further checked by performing staining on brain sections from P14-17 GLAST and GLT-1 knockout mice (see Fig. 2). Two different anti-GLT-1 antibodies are used for different protocols in this study (see Table 1) and no detectible reactivity was seen for either in the GLT-1 knockout animals, while cell staining was apparent in both wild-type littermates (to the GLT-1 mice) and in the GLAST knockouts ( Fig. 2, top two rows). In contrast, detectible GLAST staining was not found in the brains of GLAST knockout, but was seen in the GLT-1 knockouts and the wildtype controls (Fig. 2, bottom row). In wild-type optic nerve, EAAC1, GLAST, and GLT-1 staining was apparent in numerous cell bodies (Fig. 1, arrows). Staining of somata will correspond to glial cell bodies in this neuron somata-free preparation. Expression of all three proteins was not restricted to somata and staining was seen throughout the nerve presumably due to staining of glial processes and/or axons. Such staining is similar to that described previously (Li et al., 1999), and was at a somewhat lower level of intensity to that seen in the somata. In the current investigation, we initially imaged control sections to determine the imaging parameters required to eliminate background fluorescence prior to using the same parameters for tests sections, thus ensuring that our images were not biased toward bright objects at the expense of dim objects.

D-Aspartate Uptake Via High Affinity Glutamate
Transport in P14-17 Mouse Optic Nerve Functional high affinity Na 1 -dependent glutamate transporters will transport exogenous D-aspartate into cells that can subsequently be assessed by antibody staining (Gundersen et al., 1993(Gundersen et al., , 1995. Perfusion with 500 lM D-aspartate for 60 min produced extensive D-aspartate staining in somata and non-somatal regions of the P14-   (1),'' and that used for pre-embedded immuno-electron microscopy ''GLT-1 (2)'' and for the GLAST antibody. Staining was performed on mice brain sections from GLAST knockout animals, GLT-1 knockout animals, and wild-type littermates of the GLT-1 knockout. Note that no GLT-1 immuno-reactivity was detected using either GLT-1 antibody in the GLT-1 knockout brain and no GLAST reactivity was detected in the GLAST knockout brain. Scale bar 5 20 lm.
17 optic nerve (Fig. 3a). Preabsorbing the D-aspartate antibody for 30 min in 500 lM D-aspartate prior to staining ablated the immuno-staining (Fig. 3b). D-Aspartate uptake was largely abolished by zero-Na 1 conditions (Fig. 3c) and by the specific transport inhibitor DL-threob-benzyloxyaspartate (TBOA, 10 lM; Fig. 3d), confirming the expression of functional high affinity Na 1 -dependent glutamate transport in somata and non-somatal regions of the optic nerve. This relatively low TBOA concentration was used to minimize hetero-exchange of TBOA for intracellular D-aspartate (Anderson and Swanson, 2000;Montiel et al., 2005), but will result in only partial inhibition of GLAST. This may explain why D-aspartate staining appears to be lower under zero-Na 1 conditions, which will block all the glutamate transporters.

Functional Glutamate Transporter Expression in Axons in P14-17 Mouse Optic Nerve
The expression pattern of GLT-1, GLAST, and EAAC1 and the pattern of uptake of D-aspartate into areas surrounding somata may indicate axonal expression of glutamate transporters in developing optic nerve. Double labeling for neurofilament suggested co-localization of this axonal marker with GLT-1, EAAC1, and GLAST ( Fig. 4a-c). Immuno-gold labeling for EAAC1 revealed that while this transporter is expressed within oligodendrocyte processes that are closely apposed to axons, and to a limited extent within the axoplasm, no staining was apparent within the axolemma itself (Fig. 4d,e). Postembedded immuno-staining for GLAST was not successful, but immuno-gold GLT-1 labeling revealed extensive expression within the axolemma both of premyelinated and myelinated axons, in addition to staining of near-by astrocyte processes (Fig. 4f,g).
Light microscopy suggested that D-aspartate uptake occurs into axons within nerves previously incubated with this amino acid (Fig. 5a,b, arrows), in addition to uptake into glial processes running parallel to axons (Fig. 5c,d, arrows). Although axonal D-aspartate uptake was somewhat punctate in many regions, areas of diffuse staining along the center of axons were apparent and presumably represent axoplasmic accumulation. At the ultrastructural level, postembedded immuno-gold labeling for glutamate in control nerves showed the presence of the neurotransmitter in astrocytes, axons, and oligodendrocytes (Fig. 6a). The presence of D-aspartate in incubated nerves confirmed the loading of the amino acid into axons and oligodendrocytes, with highly variable loading into astrocytes (Fig. 6b). No staining was apparent in control nerves (Fig. 6c). Postembedded staining was used in the P14-17 tissue since in our hands it allows a greater level of tissue detail to be observed, which is required to examine subcellular expression in small diameter P14-17 axons.
Blinded counting of gold particles showed that astrocytes accumulate significantly more D-aspartate than either axons or oligodendrocytes within incubated optic nerves, although between them axons and oligodendrocytes have a greater total capacity for D-aspartate uptake (axon 1 oligodendrocyte uptake) than do astrocyte alone (Fig. 6d). The population distribution of goldparticle density within astrocytes was found to be highly variable, with a population of astrocytes that accumulated little D-aspartate and other cells that accumulated much more (Fig. 6e). Particle counts were collected from 2 optic nerves from separate animals, with 16 astrocytes, oligodendrocytes, and axons analyzed from a total of 8 sections (2 of each from each section).

Functional Glutamate Transporter Expression in
Astrocytes in P14-17 Mouse Optic Nerve Transgenic mice, in which EGFP expression is under the control of glial cell-specific promoters, were used to probe the expression of glutamate transporters and the uptake of D-aspartate into astrocytes and oligodendrocytes. Optic nerves from P14-17 EGFP-GFAP mice retained EGFP fluorescence following fixation, although not all astrocytes in this mutant express detectable levels of fluorescent protein (Nolte et al., 2001;Zhuo et al., 1997). Immuno-staining revealed an absence of EAAC1, and a relatively high level of GLT-1 expression in these cells (Fig. 7a,b). Only diffuse staining for GLAST was found in identified glial cells in P14-17 mouse optic nerve, which could not be discriminated from background levels of staining (not shown). The occasional GLAST (1) somata seen in Fig. 1b are presumably therefore neither astrocytes nor oligodendrocytes, and may possibly be precursor cells. At the ultrastructural level, gold particles for GLT-1 were present at high den-  are closely apposed to axons (Ax). Light staining was found in the axoplasm of axons (arrow head), but not in the axolemma. Note on the right a rare gold particle in the cell membrane of an astrocyte process (ap) aligned next to an axon (arrow). (f, g) Immuno-gold reactivity for GLT-1. Note the heavy gold-particle labeling within the axolemma in both panels (arrows), in addition to staining within the axoplasm and within the membrane of astrocyte processes (ap) (arrows). The areas within the white boes are shown at higher gain below. Scale bar 5 10 lm in (a-c) and 1 lm in (d-g). The cell types are identified using standard criteria for this preparation . sity in the cell membranes of astrocyte processes (Fig.  7c, arrows). Assessed at the light microscopic level, the capacity of astrocytes to take up exogenous D-aspartate was found to be highly variable with some cells having undetectable levels of D-aspartate and others having very high levels (Fig. 7d,e). This tendency was not related to the position of the cells within the nerve and was unlikely to arise from preferential loading of Daspartate into more peripherally located astrocytes.

Functional Glutamate Transporter Expression in Oligodendroglia in P14-17 Mouse Optic Nerve
At the light microscopic level, the majority of oligodendrocytes in P14-17 EGFP-CNP optic nerve expressed EAAC1, with only diffuse low levels of staining observed for GLT-1 (Fig. 8a-d). A population of EAAC1 (2) oligodendrocytes were also observed (not shown). D-Aspartate uptake into oligodendrocytes was at a somewhat higher level than that seen in surrounding tissue (Fig. 8e-h) and D-aspartate (2) cells were not seen. Uptake was often particularly marked within processes of EGFP-CNP oligodendrocytes (Fig. 8g), but the EGFP fluorescence was retained only poorly in the more distal processes that are beginning to ensheath axons at this point in development. In separate experiments, oligodendrocytes in preloaded wild-type mice were antibody-stained against CNPase to show these fine processes in greater detail, revealing D-aspartate uptake into these structures (Fig.  8e,f, arrows). This may explain how D-aspartate is present in the EAAC1 (2) somata, if the amino acid is taken up in the processes before diffusing throughout the cytoplasm. Alternatively, D-aspartate into these somata may result from some form of low affinity uptake. This series of stainings also revealed CNPase (2) axon-like structures loaded with D-aspartate (Fig. 8e,f, arrow heads), which correspond to the NF-70 (1) axons seen in Fig. 5.

Glutamate Transporter Expression in Mature
Optic Nerve Glutamate transporter expression was also examined in mature mouse and rat optic nerve by electron microscopy, revealing a similar pattern in both (we will show the data from the rat). Mature optic nerve astrocytes were found to mainly express GLT-1 in their somata and processes (Fig. 9a), in agreement with the idea that this is the major glutamate transporter present in these cells. GLT-1 immuno-particles are almost exclusively found in plasma membranes within astroglia. As in P14-17 optic nerves, GLT-1 was also expressed in oligodendrocytes but in the adult cells it was found mostly in plasma membranes (Fig. 9b). No GLT-1 expression was found in axonal tracts of the adult optic nerve (Fig. 9a,c,d).
GLAST was also expressed in both astroglial somata and processes of the adult optic nerve (see Fig. 10), but the density of immuno-particles was lower than for GLT-1. As for GLT-1, GLAST was mostly associated with plasma membranes of astrocytes. GLAST was also found in oligodendrocytes, especially in their myelin sheaths (Fig. 10a,c, arrows). GLAST was not detected in axonal tracts including the nodes of Ranvier (Fig. 10c,d).
Most oligodendrocytes and astrocytes in the adult optic nerve expressed EAAC1 (see Fig. 11). In addition, EAAC1-immunolabeling was found in axons and at the nodes of Ranvier (Fig. 11a,c,d). Particles were located in the axoplasm of nodes and also at the nodal membrane; however, in internodal regions there was scattered immuno-labeling. EAAC1 was also present in the cytoplasm of astrocytes and oligodendrocytes in the optic nerve, a feature which could be due to the rapid trafficking and recycling of this transporter between the cytoplasm and the plasma membrane (Danbolt, 2001). The expression of glutamate transporters at the nodes of Ranvier in adult rat optic nerve was also examined by immunofluorescence using an antibody against Na 1 channels, a marker of this structure. Consistent with our EM findings, we observed the presence of EAAC1, but not of GLT1 and GLAST in this structure (see Fig. 12).
A quantitative analysis of the distribution of glutamate transporters revealed that GLT-1, GLAST,and Fig. 6. Ultrastructural localization of excitatory amino acids in P14-17 mouse optic nerve. (a) Immuno-gold localization of glutamate reactivity in control optic nerve showing gold particles within an oligodendrocyte (Oli), an astrocyte (Ast), and within axons (arrows). The box shows absence of staining when the antibody was ablated by prior incubation with glutamate. (b) D-aspartate reactivity in an optic nerve subject to the D-aspartate incubation protocol. Note the gold particles within an oligodendrocyte and within axons (arrows) and the absence of staining within the astrocyte. (c) D-aspartate reactivity is absent in control optic nerve. Scale bars 5 1 lm. (d) Blinded counting of D-aspartate gold particles. Astrocytes accumulate significantly more D-aspartate than either axons or oligodendrocytes within the optic nerve. (e) The population distribution of gold-particle density within astrocytes was highly variable, with a population of astrocytes that accumulated little D-aspartate. *** 5 P > 0.001. EAAC1 are all expressed at high levels in astrocytes, with GLT-1 the most heavily expressed (Fig. 13a). In addition, EAAC1 was abundant in oligodendrocyte somata and GLT-1 was present in these cells to a lesser degree, with GLAST present in the myelin sheath (Fig.  13a). EAAC1 was found at high levels in axons, particularly, at the nodes of Ranvier. Particles were counted in 5 optic nerves from separate animals with a total of 482 astrocytes, 183 oligodendrocytes, 786 myelin sheaths, 173 nodes of Ranvier, and 800 axonal tracts analyzed. The density of gold particles present in cell membranes was also assessed (Fig. 13b), which correlated well with the whole cell counts. Membrane counts were performed on 3 optic nerves from different animals with 156 astrocytes, 76 oligodendrocytes, and 87 nodes of Ranvier analyzed.

DISCUSSION
We have shown a high capacity for Na 1 -dependent, TBOA-blockable, D-aspartate uptake into P14-17 mouse optic nerve astrocytes, axons, and oligodendrocytes and have documented high-affinity glutamate transporter expression in white matter of two rodent species at two developmental points. Axons, astrocytes, and oligoden-droglia are also shown to contain cytoplasmic glutamate, a prerequisite of ongoing glutamate uptake although other sources of cytoplasmic glutamate exist. These findings indicate that glutamate regulation is important in central white matter tracts such as the optic nerve and that it is achieved by glutamate transporter expression in axons and oligodendroglia in addition to astrocytes. Glutamate transporters were expressed in optic nerve oligodendrocytes, axons, and astrocytes both during development and in the adult. This implies that the extracellular concentration of glutamate is carefully regulated by these proteins in white matter, just as it is in gray matter.
The current preparation, the rodent optic nerve, has the advantage that glial and axon maturation is relatively coherent and has been well characterized. For example, in the P14-17 rodent optic nerve 10-25% of axons have received at least a first layer of myelin, with 10-15% of axons being contacted by oligodendrocyte processes but not yet receiving a myelin layer and 65-80% of axons being premyelinated (all axons are myelinated in the adult) (Hildebrand and Waxman, 1984). The P14-17 rodent optic nerve is, therefore, at a similar developmental point to the term central white matter that is most commonly subject to hypoxic-ischaemic injury at birth (Brody et al., 1987;Craig et al., 2003;Inder and Fig. 7. Glutamate transporter expression in P14-17 optic nerve astrocytes. (a, b) GFP expression in GFP (GFAP) astrocytes (green, left) and labeling for either GLT-1 (a, red, middle) or EAAC1 (b, red, middle). The images are overlaid on the right, showing expression of GLT-1 in astrocytes with no co-expression of EAAC1. (c) Immuno-gold labeling for GLT-1 showing expression in the membrane of astrocyte processes (''ap'', arrows) and staining only in the cytoplasm of a neigh-boring oligodendrocyte (''Oli'', arrow-heads). (d) GFP expression in astrocytes. (e) Co-staining for D-aspartate in the same incubated nerve. The images are overlaid below, showing D-aspartate uptake into one of the astrocytes (yellow) with several D-aspartate (2) astrocytes present. Note the presence of D-aspartate (1) somata (left top corner) that are GFP (2), presumably oligodendrocytes. Scale bar 5 10 lm except in ''c'' where it is 1 lm. Volpe, 2000). In the rodent optic nerve, astrocytes develop before oligodendrocytes and many astrocytes have a mature phenotype by P14-17 while the oligodendrocytes will form a diverse developmental group ranging from precursor to fully mature cells (Hildebrand and Waxman, 1984;Skoff, 1990;Skoff et al., 1976;Vaughn, 1969).

Glutamate Transporter Expression in White Matter
White matter glutamate transporter expression has previously been examined mainly in the adult rat optic nerve, a preparation that contains protein and mRNA for GLT-1, EAAC1, and GLAST (Choi and Chiu, 1997;. Two groups have published different cell distributions of these transporters in this preparation.  reported GLAST expression in interfascicular oligodendrocytes, EAAC1 in a population of unidentified glia and GLT-1 in astrocytes. In contrast, Kugler and Beyer (2003) found GLAST expression in most astrocytes, a population of EAAC1 (1) astrocytes and GLT-1 expression in oligodendrocytes and axons. Kugler and Beyer (2003) used b-tubulin as an oligodendrocyte marker, which may also be expressed in astrocytes, for example, (Medrano and Steward, 2001), and it is possible that some cells identified in their study as oligodendrocytes were in fact astrocytes. In addition to optic nerve, high levels of GLAST, GLT-1, and EAAC1 expression have been reported in other white matter structures such as subcortical white matter, in particular during postnatal development (Furuta et al., 1997b).
The current study used pre-embedded immuno-electron microscopy to demonstrate GLT-1, GLAST, and EAAC1 expression within the cell membrane of mature astrocytes and EAAC1 expression within the cell membrane of oligodendrocytes. There has been no previous examination of the cellular distribution of glutamate transporter expression in developing white matter. The current study employed EGFP expression linked to cellspecific promoters to identify glial cell types in P14-17 mouse white matter, removing any possibility of antibody interactions during double labeling, and used postembedded immuno-electron microscopy to examine the subcellular distribution of the proteins. At P14-17, membrane expression of GLT-1 was found in astrocyte processes, with only cytoplasmic expression seen in oligodendrocytes. EAAC1 was expressed in oligodendrocytes and not in astrocytes, while staining for glial GLAST expression was not easily discriminated from background levels. The major findings are summarized in Figure 14. The EGFP-expression approach was also used to examine mature mouse optic nerve, producing results similar to those described with pre-embedded immuno-electron microscopy. Uptake of exogenous Daspartate was also employed to prove that developing optic nerve glial glutamate transporters were functional in situ.

Glutamate Transporter Expression in Axons
The current study revealed intense membrane expression of GLT-1 in axons in P14-17 optic nerve. In contrast, mature axons did not express GLT-1 or GLAST while strong EAAC1 expression was found largely localized to the node of Ranvier. Previous studies have shown GLT-1 expression in immature mouse spinal cord axons up to the first postnatal week (Yamada et al., 1998), and in fetal ovine central axons (Northington et al., 1998(Northington et al., , 1999, while EAAC1 is present in mature cerebellar Purkinje cell axons (Furuta et al., 1997a). While neuronal GLT-1 mRNA has been reported in several studies (Berger and Hediger, 1998;Schmitt et al., 1996), GLT-1 protein expression is generally absent from mature neurons (see Danbolt, 2001;Rothstein et al., 1994), although expression has been reported in axon terminals of hippocampal neurons (Chen et al., 2002(Chen et al., , 2004Schmitt et al., 2002). Both GLT-1 and GLAST are expressed within the retina (Euler and Wassle, 1995;Rauen and Kanner, 1994;Rauen et al., 1996), although expression of neither protein has been found in the developing retinal gan-   glion cells that extend the optic nerve axons (Reye et al., 2002). The high level of D-aspartate uptake and the localization of glutamate within axons confirmed the presence of functional transport, which in terms of Daspartate uptake capacity was comparable to that seen in glial cells and occupied a larger proportion of the nerve. Axonal uptake was blocked by either Na 1 removal or by the specific transport inhibitor TBOA, and this is the first direct proof of functional glutamate transport expression in axons. The high level of expres- Fig. 11. Ultrastructural localization of EAAC1 in adult rat optic nerve. (a, c) Transverse and longitudinal sections showing EAAC1 expression in both axons (ax) and astrocyte processes (ap). (b) EAAC1 is also found in oligodendrocytes (oli) as well as in astrocyte processes (ap) in their vicinity. (d) EAAC1 reactivity is also found at the nodes of Ranvier (R). Scale bars 5 300 nm. Arrowheads point to examples of gold particles in plasma membranes. sion of EAAC1 at the nodes of Ranvier indicates a high level of glutamate regulation in the node in mature axons. EAAC1 immuno-particles are often present in the cytoplasm, a feature which could be interpreted as protein in transit to the plasma membrane (Danbolt, 2001).

Glutamate Transporters and White Matter Injury
At P14-17, oligodendrocytes accumulated high levels of D-aspartate during exogenous perfusion, which is consistent with the elevated levels of glutamate transporter expression in these cells at this point in development, in particular, EAAC1. Average D-aspartate uptake into astrocytes was higher than into any other cell compartment, although the degree of uptake varied considerably from cell to cell. It is unclear why some astrocytes accumulated low levels of D-aspartate, a phenomenon that did not appear to relate to cell position in the nerve or the presence of a population of astrocytes with low glutamate transport levels, which were not seen. One possibility is that some of the astrocytes are not yet physiologically competent, since it is known that many optic nerve astrocytes lack the morphological features of mature astrocytes at this age (Skoff et al., 1976). Recent studies have shown rapid astrocyte swelling leading to necrosis in ischaemic developing white matter, which will liberate astrocyte glutamate into the extracellular space in a glutamate transporter-independent fashion . In addition, the presence of functional GLT-1 demonstrates the potential for transportmediated glutamate release from these cells. Thomas et al. (2004) also demonstrated a high resting intracellular glutamate levels in oligodendrocytes that was elevated following ischaemia, suggesting both effective glutamate uptake under normal conditions and the absence of uptake reversal during ischaemia, although ischaemic glutamate release from these cells is also possible (Fern, 2000;Back et al., 2006). It is not known how [Na 1 ] i , pH i , and membrane potential are affected by ischaemia in these cells, but it would appear that the conditions are not met for significant reversal of glutamate uptake in this preparation (but see Back et al., 2006). The current findings suggest that glutamate release from axons via reverse transport is a potentially important factor in excitotoxic (Dewar et al., 2003;Wilke et al., 2004) and nonexcitotoxic (Oka et al., 1993) white matter damage during ischaemia.

White Matter Glutamate Transporter Function
Glutamate receptors are expressed in developing glia, and may regulate cell development and fate (Belachew and Gallo, 2004). Expression of AMPA/kainate receptors is particularly marked in immature cells of the oligodendrocyte lineage (Itoh et al., 2002), which in the de-  Oligodendrocytes express lower levels of GLT-1 and high levels of EAAC1, and myelin has GLAST. The internodal segment of axons and the nodes of Ranvier have moderate and high levels of EAAC1, respectively. * P < 0.05, *** P < 0.001, as compared to negative control sections (primary antibody omitted in the procedure). (b) Quantitative analysis of the distribution of glutamate transporters in cell membranes of adult optic nerve astrocytes, oligodendrocytes, and nodes of Ranvier. The pattern of distribution is similar than for entire tissue profiles. Background levels were subtracted from the counting. veloping optic nerve are in the process of forming an exquisitely close morphological arrangement with neighboring axons (Butt and Ransom, 1993). We and others have recently demonstrated the presence of functional NMDA-type glutamate receptors in oligodendrocyte processes (K arad ottir et al., 2005;Micu et al., 2006;Salter and Fern, 2005). The activation of these receptors is likely to be important in a variety of diseases and must have significant developmental consequences.
We have now shown high levels of glutamate transporter expression in both developing axons and glia, which is consistent with a tight maintenance of extracellular glutamate in the zones where axons and glia meet. Conjunctions between developing axons and their ensheathing oligodendrocytes and surrounding astrocytes may therefore be analogous to neuronal synapses, where release and clearance of extracellular glutamate is essential for cell-cell interaction. Two recent studies have reported vesicular-type glutamate release from axons onto neighboring glial cells (Kukley et al., 2007;Ziskin et al., 2007). The current study confirms for the first time that functional glutamate reuptake mechanism exists within a central white matter tract, adding further to the weight of evidence for synaptic-type glutamatergic signaling between axons and glia.