|Home | About | Journals | Submit | Contact Us | Français|
Glutamate receptors, which play a major role in the physiology and pathology of CNS gray matter, are also involved in the pathophysiology of white matter. However the cellular and molecular mechanisms responsible for excitotoxic damage to white matter elements are not fully understood. We explored the roles of AMPA and GluR5 kainate receptors in axonal Ca2+ deregulation.
Dorsal column axons were loaded with a Ca2+ indicator and imaged in vitro using confocal microscopy.
Both AMPA and a GluR5 kainate receptor agonists increased intra-axonal Ca2+ in myelinated rat dorsal column fibers. These responses were inhibited by selective antagonists of these glutamate receptors. The GluR5-mediated Ca2+ rise was mediated by both canonical (i.e. ionotropic) and non-canonical (metabotropic) signalling, dependent on a pertussis toxin-sensitive G protein and a phospholipase C-dependent pathway, promoting Ca2+ release from IP3-dependent stores. Additionally, the GluR5 response was significantly reduced by intra-axonal NO scavengers. In contrast, GluR4 AMPA receptors operated via Ca2+ induced Ca2+ release, dependent on ryanodine receptors, and unaffected by NO scavengers. Neither pathway depended on L-type Ca2+ channels, in contrast to GlurR6 kainate receptor action 1. Immunohistochemistry confirmed the presence of GluR4 and GluR5 clustered at the surface of myelinated axons; GluR5 co-immunoprecipitated with nNOS and often co-localized with nNOS clusters on the internodal axon.
Central myelinated axons express functional AMPA and GluR5 kainate receptors, and can directly respond to glutamate receptor agonists. These glutamate receptor-dependent signalling pathways promote an increase in intra-axonal Ca2+ levels potentially contributing to axonal degeneration.
The precise mechanisms of glutamate-mediated toxicity in white matter are not completely established. This transmitter likely causes damage to glia given that both astrocytes and oligodendrocytes express a variety of glutamate receptors 2–8, with oligodendrocytes being particularly vulnerable to excitotoxic cell death 9–12. Whether glutamatergic signalling is directly involved in irreversible axonal injury in disorders such as stroke, multiple sclerosis and neurotrauma is not known, though a role for glutamate-dependent excitotoxicity is suspected given the protective effects of AMPA/kainate antagonists in models of spinal cord injury, stroke and experimental autoimmune encephalomyelitis 2,4,13–18. The beneficial effect of glutamate antagonism was hypothesized to be due to sparing of glia and myelin, but the observed axonal protection remains unexplained. To date, no conclusive data exist showing expression of functional glutamate receptors on central myelinated axons. Here we show that myelinated dorsal column axons express GluR4 AMPA receptors as well as GluR5 kainate receptors; the GluR5 effect is mediated in large part via a non-canonical mechanism through activation of G protein, phospholipase C and release of Ca2+ from intracellular stores by activation of IP3 receptors. GluR4 AMPA receptors on the other hand appear to participate in Ca2+-induced Ca2+ release through ryanodine-dependent Ca2+ stores.
Experiments were performed on spinal cord dorsal columns in vitro from adult Long Evans male rats. Thoracic spinal cord was removed and placed in cold oxygenated zero-Ca2+ solution containing (in mM): NaCl 126, KCl 3, MgSO4 2, NaHCO3 26, NaH2PO4 1.25, MgCl2 2, dextrose 10 and EGTA 0.5, oxygenated with 95% O2-5% CO2. Freshly excised dorsal columns were loaded for 2 hours with Ca2+-insensitive reference dye (red dextran-conjugated Alexa 594, 250 μM) to allow identification of axon profiles, and the dextran-conjugated Ca2+ indicator Oregon Green BAPTA-1 (250 μM) (both from Molecular Probes) using a suction electrode applied to the cut end. The final dye concentration in the axons was estimated at ≈ 2 μM. Tissue was transferred to a custom-built chamber on a Nikon C1 confocal microscope and imaged every 60 sec at 37°C with a 60× 1.0 NA water immersion lens warmed to 37°C. Green signal was ratioed against the Ca2+-insensitive red channel, and then percent change during exposure to various agents compared to control was calculated. PTX was first activated by adding ATP (1mM) and glutathione (2mM) and incubated at 37°C overnight. Final PTX concentration in the loading pipette was 5 μM.
For light microscopy, deeply anesthetized rats were perfused with saline then 4% paraformaldehyde in 0.1 M phosphate buffer. Dorsal columns were excised, post-fixed, and immersed in 20% sucrose overnight. 40 μm sections were cut with a freezing microtome and washed with Tris buffer containing 1% Triton X-100. After 1 hr blocked In 10% NGS in Triton X-100, primary antibodies against GluR5 (Chemicon;1:50), GluR4 (Chemicon;1:50), nNOS (Abcam; 1:100) and NF160 (Sigma; 1:1000) were applied for 24 hrs at 4°C. Secondary antibodies (Texas red anti-rabbit or anti-mouse, Cedarlane) were applied at a 1:100 dilution, and Alexa 488 anti-goat and anti-rabbit (Molecular Probes) at 1:500 for 1hr at room temperature. Neurofilament 160 directly conjugated to Alexa 660 was used at 1:100 dilution. Slides were imaged on a Nikon C1 confocal with a 60× 1.4 NA oil immersion objective.
Means are shown with standard deviations. Statistical differences were computed using ANOVA with Tukey’s test for multiple comparisons using Igor (Wavemetrics. Lake Oswego, OR) unless otherwise stated. N’s are numbers of individual axons unless otherwise noted.
We measured relative [Ca2+] changes in live adult rat spinal cord dorsal column myelinated axons using Oregon Green-488 BAPTA-1 fluorescence (Fig. 1). GluR5-containing kainate receptors (hereafter referred to as GluR5’s) were activated by the selective agonist (RS)-2-amino-3-(3-hydroxy-5-tert-butylisoxazol-4-yl)propanoic acid (ATPA 10 μM) 20, while the relatively selective agonist AMPA (alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid, 100 μM), was used to activate AMPA receptors (Fig. 1). Both agents induced an increase in axoplasmic Ca2+-dependent fluorescence (FCa.ax) (mean increase after 30 min: 130 ± 63%, n = 53 axons for ATPA and 83 ± 50%, n = 56 for AMPA) which was largely blocked by the combined AMPA/kainate receptor antagonist 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione (NBQX 50 μM, FCa.ax increase at 30 min: 17 ± 14%, n = 32 with ATPA and 18 ± 20%, n = 45, with AMPA; P = 10−5 vs. agonist alone, both groups). In contrast, the GluR5 antagonist (S)-1-(2-Amino-2-carboxyethyl)-3-(2-carboxybenzyl)pyrimidine-2,4-dione (UPB302 20 μM) 21 selectively reduced the ATPA-induced response (FCa.ax increase 38 ± 35%, n = 48, P =10−5 vs. ATPA alone) but had no effect on the AMPA-induced Ca2+ rise (85 ± 19%, n = 27, P ≈ 1 vs. AMPA alone). Comparing the Ca2+ responses elicited by AMPA vs. ATPA in the presence of UPB302, also yielded a highly significant difference (P = 6 × 10−4), further supporting the relatively selective block of GluR5 over AMPA receptors by this antagonist. Conversely, the AMPA receptor antagonist 1-naphthyl acetyl spermine, did not reduce the ATPA-induced FCa.ax increase (133 ± 73%, n = 44, P = 1) but strongly reduced the response elicited by AMPA (11 ± 24%, n = 32, P = 10−5) (Fig. 1). Taken together, these findings strongly suggest that ATPA preferentially activated GluR5’s whereas AMPA acted as a selective agonist of AMPA receptors in our tissue.
To further characterize the source of Ca2+ mobilized by activation of GluR5 or AMPA receptors, experiments were performed in Ca2+-free perfusate, with the addition of 0.5 mM EGTA (Fig. 2). Under these conditions, the ATPA-induced Ca2+ rise was significantly reduced but not abolished (FCa.ax increase: 43 ± 31% n = 36, P = 10−5 vs. ATPA in normal Ca2+), suggesting that part of the GluR5-dependent effect was mediated by Ca2+ release from intracellular stores independent of extracellular Ca2+. In contrast, the AMPA response was virtually abolished in 0Ca2+ perfusate (FCa.ax increase: 6 ± 23%, n = 27, P = 10−5 vs. AMPA in normal Ca2+). We previously showed that activation of axonal voltage-gated Ca2+ channels leads to Ca2+ release from axonal Ca2+ stores, and is an important mechanism contributing to Ca2+ overload in damaged central axons 22. We therefore examined the effect of nimodipine (10 μM) to investigate any involvement of L-type Ca2+ channels. This blocker did not reduce the ATPA- nor the AMPA-induced FCa.ax increase (Fig. 2), indicating that these receptors induced axonal Ca2+ rises independently of L-type Ca2+ channel activation.
Kainate receptor signalling may occur either ionotropically or metabotropically via receptor coupling to G proteins 23,24. For instance, in cultured dorsal root ganglion cells that express GluR5 subunits, activation of GluR5-containing kainate receptors induces a metabotropic action with release of Ca2+ from intracellular stores through activation of G-proteins and phospholipase C 24. We loaded spinal axons with pertussis toxin (PTX, see Methods) designed to inhibit G-protein signalling. Under these conditions, the axonal Ca2+ rise in response to ATPA application was greatly reduced (FCa.ax increase 17 ± 28% n = 38, P = 10−5 vs. no PTX). Moreover, either inhibition of phospholipase C by U73122 (10 μM) or antagonism of IP3 receptors by 2-aminoethoxydiphenyl borate (2-APB, 100 μM) reduced ATPA-induced Ca2+ rise to a degree comparable to PTX (FCa.ax increase in U73122: 11 ± 17%, n = 33; 2-APB: 16 ± 20%, n = 27, P = 10−7 vs. no drug). Taken together, these data suggest that while axonal ATPA-induced GluR5 responses exhibit some dependence on extracellular Ca2+, the bulk of the resulting axonal Ca2+ rise in response to activation of these receptors was mediated mainly by non-canonical signalling through G-proteins, activation of PLC and release of Ca2+ from intra-axonal IP3-dependent Ca2+ stores.
In contrast, the near-complete abolition of axonal Ca2+ increase in response to AMPA in zero-external Ca2+ perfusate suggests that AMPA receptors permeate Ca2+ directly into myelinated spinal axons; however, this extracellular Ca2+ dependence does not exclude an intracellular Ca2+ source, through a Ca2+-induced Ca2+-release mechanism for instance. Indeed, ryanodine (50 μM), a selective antagonist of ryanodine receptors present on endoplasmic reticulum, significantly reduced AMPA receptor-mediated axonal Ca2+ rise (16 ± 34% n = 26, P = 10−7 vs. no ryanodine). Thus, the strong dependence of AMPA responses on external Ca2+, coupled with the significant inhibitory effects of ryanodine, indicates that while AMPA receptors do permeate Ca2+ directly, the majority of the resulting axonal Ca2+ signal is due to Ca2+-induced Ca2+ release (CICR) from ryanodine-dependent intracellular stores, similar to CICR classically described in cardiac muscle cells 25. However, unlike cardiac myocytes, the initial Ca2+ “trigger” appears to be delivered by Ca2+-permeable AMPA receptors; this is consistent with the strong inhibitory effect of 1-naphthyl acetyl spermine, a selective inhibitor of GluR2-lacking, Ca2+ permeable AMPA receptors, on the AMPA-induced Ca2+ rise (Fig. 1). Together with our previous data from spinal axons showing that depolarization promotes Ca2+ release from ryanodine-sensitive Ca2+ stores 22, the present results indicate that axons exhibit both a depolarization-induced (skeletal muscle type, dependent on RyR1) and Ca2+-induced (cardiac muscle type, RyR2) Ca2+ release. Indeed, both RyR1 and RyR2 are expressed in these axons 22; whether AMPAr-mediated Ca2+ influx indicated by the present study acts selectively on axonal RyR2 is currently unknown however. Ca2+ release via IP3 receptors is also known to be potentiated by increased cytosolic-Ca2+ in a biphasic manner 26. The significant though incomplete reduction of Ca2+ rise by 0Ca2+ perfusion in response to ATPA is consistent with a dual action of GluR5’s, reflecting a small component of Ca2+ entry (see next section) coupled with a strong metabotropic action dependent on G-proteins and IP3 synthesis.
The protective effect of AMPA/kainate receptor antagonists in different white matter injury paradigms was suggested to be secondary to sparing of glial elements, possibly leading to a reduction of diffusible messengers (e.g. NO and free radicals) that may potentially damage axons (for review see 9). To investigate a potential role of NO in the AMPA or ATPA-induced Ca2+ responses, we used the NO scavenger myoglobin 27. Selective intracellular loading of axons with myoglobin, which will scavenge NO originating both from within and from outside the axon, significantly reduced axoplasmic Ca2+ rise induced by ATPA (61 ± 34% n = 45, P = 10−8 vs. no myoglobin) but did not reduce the AMPA response (92 ± 36% n = 29, P = 0.95 vs. no myoglobin), suggesting that only the GluR5-mediated ATPA response depended (at least partially) on NO production (Fig. 3). To demonstrate an intra-axonal action of NO in response to GluR5 activation, the ATPA experiment was repeated with bath-applied myoglobin, which failed to blunt the axonal Ca2+ response (FCa.ax increase: 144 ± 39%, n = 24, P = 0.99 vs. no bath-applied myoglobin). Together, the above results suggest a close functional relationship between GluR5 kainate receptors and NO production in myelinated axons, likely synthesized by neuronal nitric oxide synthase (nNOS). Indeed, the partial reduction of the GluR5-mediated Ca2+ response by removal of bath Ca2+ is remarkably similar to that seen with intra-axonal myoglobin (Fig. 3) (P = 0.93). Moreover, combined removal of bath Ca2+ and intra-axonal myoglobin is not additive (FCa.ax increase: 38 ± 36%, n = 33, P ≈ 1 vs. ATPA+ 0Ca2+/EGTA; P = 0.66 vs. ATPA+intra-axonal myoglobin). nNOS is a Ca2+-activated enzyme 28, therefore it is quite plausible that the small quantity of Ca2+ admitted across the axolemma by GluR5 serves to promote NO synthesis, which in turn may amplify the “metabotropic” arm of the GluR5-dependent pathway, ultimately leading to Ca2+ release from IP3-dependent stores. This Ca2+ release can further enhance nNOS function (e.g. 29) thus providing Ca2+-dependent positive feedback in an environment (i.e. the periaxonal space) where sources of Ca2+ are limited. Further evidence linking GluR5 with NO synthesis by nNOS is provided by immunochemical methods.
Immunohistochemical characterization of GluR5 and nNOS in dorsal columns revealed frequent punctate staining at the surfaces of neurofilament-positive axon cylinders (Fig. 4). Approximately 39% of GluR5 clusters present at the axon surface were co-localized with nNOS (Fig 4F), conversely, 58% of nNOS clusters co-localized with punctate GluR5 staining. This indicates that these signalling molecules may be distributed for other roles, such as nNOS association with the axonal GluR6 kainate receptor subtype 1. Moreover, immuno-electron microscopy confirmed that many GluR5-positive clusters were localized to the axolemma, rather than overlying myelin (Fig. 4G). In addition immunoprecipitation of dorsal column lysate with a GluR5 antibody resulted in a strong band immunoreactive for nNOS (Fig. 4H), indicating that these proteins are physically associated in this tissue. Taken together, these data are consistent with expression of GluR5 kainate receptors on the internodal axolemma, and suggest that these receptors activate nNOS to generate intra-axonal NO. A similar punctate staining pattern was observed for GluR4 (Fig 4A), indicating that AMPA receptors are also expressed on the axolemma under the internodal myelin sheath, further supported by the immunogold staining in Fig. 4C.
Glutamate-mediated toxicity of white matter has been described in several studies, many of which have implicated glutamate receptors 2,4,14–16,18; as well, glutathione depletion by glutamate-cystine exchange in oligodendroctyes, also plays a role in glutamate-dependent damage in cultured oligodendrocytes 30. These glia, which have been shown in vivo and in vitro to be particularly sensitive to excitotoxic insults, were considered the main cellular target of glutamate action 9, whereas astrocytes appear far more resistant to this neurotransmitter 30. Here we show that in addition to a glial effect of glutamate, GluR5-containing kainate receptors and GluR4-containing AMPA receptors are expressed on the internodal axolemma in central myelinated fibers, and directly participate in mediating axonal Ca2+ increases in responses to agonist. These receptors are well poised to mediate deleterious Ca2+ deregulation in axons; indeed, white matter vulnerability to ischemia increases with age 31, and it is quite conceivable that these axonal receptors are directly responsible for promoting axonal degeneration in a variety of disease states.
The numerous colocalized and physically associated GluR5 receptors and nNOS clusters distributed along axon cylinders are similar to previously reported Ca2+ channel -ryanodine receptor clusters 22, and GluR6 - nNOS “nanocomplexes” 1. It appears therefore that the internodal axolemma exhibits an even richer collection of signalling molecules than was previously thought, now including GluR4, GluR5 and GluR6, the latter two physically and functionally associated with nNOS. The very focal clustering of the above receptors and downstream effector molecules is curious, and suggests that the internodal axolemma is a very inhomogeneous structure exhibiting highly specialized and spatially restricted functions. Indeed, in this study GluR4 AMPA receptors appear to admit small amounts of extracellular Ca2+ which is then amplified by CICR, dependent on ryanodine receptors. Unlike depolarization-induced Ca2+ release 22, or GluR6-mediated Ca2+ rise 1, the AMPA receptor-dependent responses are independent of L-type Ca2+ channels, but instead appear dependent on Ca2+ ions admitted through the Ca2+-permeable receptor itself. CICR can be triggered by Ca2+ permeable channels other than voltage-gated channels; in dendritic spines for instance, NMDA receptor-mediated Ca2+ influx triggers Ca2+ release from ryanodine-sensitive stores 32. The purpose of CICR in axons is unknown, but may parallel other cells where internal Ca2+ provides the majority of a biochemical signal. In cardiac myocytes for instance, 80% of the Ca2+ responsible for contraction originates from the sarcoplasmic reticulum, mobilized via CICR 33. It is likely that myelination heavily restricts the availability of extracellular Ca2+ for signalling in the internodal periaxonal space, therefore axons may have adopted a strategy for amplifying these Ca2+ signals by instead relying on internal sources, with a small influx of this ion across the axolemma being a sufficient trigger. In contrast to the AMPA-mediated responses, the GluR5-mediated Ca2+ rise occurs by both canonical (i.e. ionotropic) and non-canonical (metabotropic) signalling, whereby a small inflow of Ca2+ is amplified through a G-protein, phospholipase C-dependent cascade ultimately leading to Ca2+ release from IP3-dependent Ca2+ stores. In contrast to the AMPA-dependent effect, this GluR5-mediated mechanism is upregulated by NO, likely synthesized by nNOS within the axon; the mechanism by which NO stimulates IP3-dependent Ca2+ release is unknown, but may involve direct or indirect (via activation of soluble guanylate cyclase) activation of IP3 receptor activity34 (see Fig. 6 in the accompanying paper 1).
Axonal pathology from many different insults such as ischemia, trauma and immune attack shares remarkable similarities, with focal swelling and often irreversible transection of the fiber 35. While the pathophysiological mechanisms of axonal swelling and transection are unknown, it is interesting to speculate that the highly localized clusters of signaling molecules on the internodal axolemma, each leading to localized Ca2+ release, may induce focal destruction of axonal cytoarchitecture, in turn possibly leading to the localized axonal pathology observed in many disorders. If so, this raises interesting therapeutic possibilities to reduce axonal damage in many disorders such neurotrauma, multiple sclerosis, and stroke. Our observations further underscore the parallels between neurons and myelinated axons with respect to the “source specificity” of Ca2+ loading, whereby the locus of Ca2+ entry (via channels)/release (from intracellular Ca2+ stores) may be more important than the total cellular Ca2+ load 36,37.
NINDS, CIHR, Heart and Stroke Foundation of Ontario Center for Stroke Recovery, HSFO Career Investigator and AHFMR Scientist Awards, and the generosity of private donors to PKS. GWZ is a CIHR Investigator and an AHFMR Scientist. Additionally supported by CIHR (GWZ) and NINDS (BDT).