Visualization of glutamate as a volume transmitter

Abstract

Glutamate is the major excitatory neurotransmitter in the central nervous system. Although glutamate mediates synaptically confined point-to-point transmission, it has been suggested that under certain conditions glutamate may escape from the synaptic cleft (glutamate spillover), accumulate in the extrasynaptic space, and mediate volume transmission to regulate important brain functions. However, the inability to directly measure glutamate dynamics around active synapses has limited our understanding of glutamatergic volume transmission. The recent development of a family of fluorescent glutamate indicators has enabled the visualization of extrasynaptic glutamate dynamics in brain tissues. In this topical review, we examine glutamate as a volume transmitter based on novel results of glutamate imaging in the brain.

Yohei Okubo received his PhD from the Department of Pharmacology, Graduate School of Medicine, the University of Tokyo in 2005. After a short-time training as a post-doctoral fellow, he became a Research Associate at the University in 2006. He has studied functions and the spatiotemporal dynamics of calcium, inositol 1,4,5-trisphosphate and glutamate in neurons and glial cells using the fluorescence imaging technique. Masamitsu Iino has been the head of the present department since 1995. He has studied the basic principles of spatiotemporal regulation of Ca²⁺ signalling, and showed the importance of the regenerative property of intracellular Ca²⁺ release in the shaping of Ca²⁺ signals. To characterize the spatiotemporal features of signalling molecules upstream and downstream of Ca²⁺ signals such as IP₃, nitric oxide and glutamate, his laboratory imaged these signalling molecules, developing their new indicators. The results shed new light on the spatiotemporal signalling mechanism of these important molecules. Based on these results, his laboratory is now searching for new cell functions that are regulated by Ca²⁺ signals in the brain.

Glutamate is the major excitatory synaptic transmitter in the mammalian central nervous system. Upon synaptic glutamate release, glutamate concentration within the narrow synaptic cleft between the pre- and postsynaptic membranes rapidly reaches millimolar concentrations, and then decays within a few milliseconds by diffusion and reuptake mechanisms (Clements et al. 1992; Rothstein et al. 1996; Asztely et al. 1997; Rusakov & Kullmann, 1998; Barbour, 2001; Mashimo et al. 2010). These rapid and spatially confined glutamate dynamics ensure precise point-to-point transmission at excitatory synapses. However, spatiotemporally dense synaptic activities may increase glutamate concentration in the extrasynaptic space, which is referred to as ‘glutamate spillover.’ Glutamate spillover is expected to activate extrasynaptic glutamate receptors distributed around the active synapse to mediate volume transmission (Barbour & Hausser, 1997). It has been proposed that glutamatergic volume transmission regulates a variety of important neural and glial functions (Sem’yanov, 2005; Haydon & Carmignoto, 2006).

The spatiotemporal dynamics of neurotransmitter in the extrasynaptic space is the key determinant of volume transmission. However, the details of extrasynaptic spatiotemporal glutamate dynamics have remained elusive; they have been estimated by numerical modelling studies or inferred indirectly using electrophysiological and other measurements. The notion that synaptically released glutamate functions as a volume transmitter had not been confirmed by direct measurements of extrasynaptic glutamate dynamics. Recently, several groups have developed new fluorescent indicators of glutamate (Okumoto et al. 2005; Namiki et al. 2007; Hires et al. 2008). One of these indicators has been found to have a sufficient sensitivity and dynamic range for the measurement of glutamate dynamics in brain tissues during physiologically relevant neural inputs (Namiki et al. 2007; Okubo et al. 2010). In this topical review, we first summarise the physiological functions that involve glutamatergic volume transmission. We then compare the properties of glutamatergic volume transmission estimated by previous studies with those revealed by new glutamate imaging techniques.

Neural and glial functions mediated by glutamatergic volume transmission

Activation of neuronal extrasynaptic glutamate receptors by glutamate spillover underlies many physiological functions (Fig. 1). Homosynaptic and/or heterosynaptic activation of presynaptic group II metabotropic glutamate receptors (mGluRs) regulates presynaptic release probability at excitatory and inhibitory synapses (Scanziani et al. 1997; Mitchell & Silver, 2000). Activation of postsynaptic group I mGluRs localised at perisynaptic locations, on the other hand, regulates several physiological functions, including synaptic plasticity and synaptic maintenance (Furutani et al. 2006; Luscher & Huber, 2010). Activation of axonal kainate receptors by glutamate spillover mediates biphasic regulation of the excitability of presynaptic fibres (Schmitz et al. 2001). N-methyl-d-aspartate receptors (NMDARs) containing the NR2B subunit localise in extrasynaptic structures, and they may play a role distinct from that of the synaptic NMDARs, e.g. the extrasynaptic NMDARs, but not the synaptic NMDARs, are involved in neuronal survival (Hardingham et al. 2002) and Huntington's disease (Milnerwood et al. 2010). Neuronal extrasynaptic glutamate receptors are activated not only via glutamate spillover but also by glutamate released from glial cells (see below).

Figure 1. Neural and glial functions mediated by glutamatergic volume transmission.

Left, neuronal functions mediated by glutamatergic volume transmission. AMPAR, NMDAR, kainate receptor and mGluR localised to various extrasynaptic locations or within neighbouring synapses are activated by glutamate spillover to mediate physiological and pathophysiological events. Right, astrocytic functions mediated by glutamatergic volume transmission. Astrocytic mGluR is activated by glutamate spillover to induce gliotransmitter release and haemodynamic response.

Glutamate spillover may be involved in excitatory transmission activating ionotropic glutamate receptors, including α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs), NMDARs and kainite receptors. Synaptic crosstalk, i.e. activation of ionotropic glutamate receptors in neighbouring synapses by glutamate spillover, has been found in several types of synapse (Carter & Regehr, 2000; Arnth-Jensen et al. 2002; DiGregorio et al. 2002; DeVries et al. 2006). Excitatory current induced by synaptic crosstalk contributes significantly to the transmission at cerebellar glomeruli (DiGregorio et al. 2002) and retinal cone photoreceptor synapses (DeVries et al. 2006). Furthermore, glutamate spillover can mediate transmission between neurons that have no direct synaptic contact. Such neurotransmission, mediated purely by glutamate spillover, is observed in the olfactory bulb (Isaacson, 1999) and the cerebellum (Szapiro & Barbour, 2007).

Glutamate spillover may mediate not only interactions between neurons but also signal transmission from neurons to glial cells (Fig. 1). Although glial cells do not usually receive direct synaptic contacts from neurons, glutamate released from neurons induces intracellular Ca²⁺ responses in astrocytes, the major type of glial cells in the brain, through activation of glutamate receptors (Cornell-Bell et al. 1990; Porter & McCarthy, 1996; Wang et al. 2006). The intracellular Ca²⁺ increase in astrocytes is thought to regulate various astrocytic functions. One of such functions is the release of gliotransmitters, such as ATP (Guthrie et al. 1999), d-serine (Schell et al. 1995) and glutamate (Parpura et al. 1994). Astrocytic-derived ATP and its hydrolysed product, adenosine, suppress adjacent glutamatergic synapses to mediate negative feedback regulation (Koizumi et al. 2003; Zhang et al. 2003; Pascual et al. 2005). d-Serine functions as an endogenous co-agonist of NMDARs (Schell et al. 1995), and d-serine released by astrocytes is required for the NMDAR-dependent classical hippocampal long-term potentiation (Henneberger et al. 2010). Glutamate released by astrocytes preferentially activates extrasynaptic NMDARs in multiple adjacent neurons, which enhances neuronal excitability (Parri et al. 2001; Fellin et al. 2004). Furthermore, group I mGluRs in neurons are activated by astrocytic-derived glutamate to potentiate presynaptic release probability (Fiacco & McCarthy, 2004; Perea & Araque, 2007). Thus, glutamate spillover plays an important role regulating gliotransmitter release from astrocytes, although controversy remains regarding the physiological relevance of astrocytic gliotransmitter release (Agulhon et al. 2010; Hamilton & Attwell, 2010). Another important astrocytic function triggered by glutamate spillover is the regulation of blood flow in the brain (Fig. 1). Glutamate spillover-triggered Ca²⁺ responses in astrocytes induce dilatation or constriction of adjacent blood vessels to induce haemodynamic responses (Zonta et al. 2003; Mulligan & MacVicar, 2004; Gordon et al. 2008).

Imaging extrasynaptic glutamate dynamics in the brain

There have been efforts to directly measure glutamate concentration in the brain. Microdialysis and amperometric measurements have been carried out to measure glutamate concentration changes within the brain tissue (Baker et al. 2002; Dash et al. 2009). However, these methods have limitations with regard to spatiotemporal resolution. To gain greater spatiotemporal resolution, fluorescent indicators of glutamate have been developed to directly visualise glutamate dynamics (Okumoto et al. 2005; Namiki et al. 2007; Hires et al. 2008). FLIPE (Okumoto et al. 2005) and SuperGluSnFR (Hires et al. 2008) are genetically encoded, fluorescence resonance energy transfer (FRET)-based indicators, in which cyan fluorescent protein (CFP) and a variant of yellow fluorescent protein (YFP) are fused to the N- and C-termini of a bacterial glutamate binding protein ybeJ, respectively. The FRET intensity between CFP and the variant YFP changes with the concentration of glutamate, with an apparent dissociation constant (K_d) of 630 nm for FLIPE and 2.5 μm for SuperGluSnFR. These indicators have a maximum FRET ratio change (ΔR_max) of 11% for FLIPE and 44% for SuperGluSnFR, and have been expressed on the plasma membrane to image glutamate concentration changes on the extracellular surface of neurons in culture (Okumoto et al. 2005; Hires et al. 2008). However, these indicators have not been successfully used to image glutamate dynamics in brain tissues with intact structures. FLIPE has been modified so that CFP is inserted within the ybeJ sequence to yield FLII⁸¹PE-1μ, which has a K_d of 1 μm and ΔR_max of 48% (Deuschle et al. 2005). Purified FLII⁸¹PE-1μ protein was introduced into the extracellular space of cerebral slice preparations by diffusion, and was used to image glutamate dynamics before the indicator was washed out. Although a nerve stimulation-induced FRET intensity change of FLII⁸¹PE-1μ was imaged under a wide-field fluorescence microscope with sampling frequencies of 50 Hz or lower (Dulla et al. 2008), a cell-level spatial resolution was not attained.

Another type of fluorescent glutamate indicator has been generated (Namiki et al. 2007) and successfully used to image fine extrasynaptic glutamate dynamics in brain tissue preparations (Okubo et al. 2010). The new hybrid-type fluorescent glutamate indicator consists of the glutamate-binding domain of AMPAR subunit GluR2 and a fluorescent small molecule (Oregon Green 488) conjugated near the glutamate-binding pocket (Fig. 2A). This indicator, the glutamate (E) optical sensor (EOS), shows high affinity (K_d of 148 nm) and selectivity for glutamate, and achieves a large dynamic range (maximum fluorescence intensity change, ΔF_max of 21% measured on the cell surface) (Namiki et al. 2007; Okubo et al. 2010). EOS can be evenly and densely labelled on the cell surface through a tight biotin–streptavidin interaction, and can be used for high-resolution imaging of glutamate dynamics around neurons in culture (Namiki et al. 2007). Variants of EOS have been generated with a different K_d and a greater dynamic range: K716A-EOS has a K_d of 174 nm and ΔF_max of 29%, and L401C-EOS has a K_d of 1.57 μm and ΔF_max of 48%. The sister indicators (collectively called EOS), in combination with two-photon microscopy, opened the door to visualizing extrasynaptic glutamate dynamics in preparations with intact brain structures (Namiki et al. 2007; Okubo et al. 2010). EOS labelling in the extracellular space of acute cerebellar slice preparations through biotin–streptavidin interaction resulted in roughly even and bright staining with silhouettes of the cellular components (Fig. 2C). When the parallel fibres (PFs) are stimulated with a physiologically relevant pulse train (5 pulses at 100 Hz; Chadderton et al. 2004), the EOS fluorescence intensity under a two-photon microscope increases within a circular area on a sagittal plane (Fig. 2D). The observed EOS signal predominantly reports extrasynaptic glutamate concentration changes, because a negligible amount of EOS is immobilised within the synaptic cleft (Namiki et al. 2007; Okubo et al. 2010). Thus, the EOS imaging method allows direct measurement of extrasynaptic glutamate dynamics adjacent to excitatory synapses.

Figure 2. Imaging of extrasynaptic glutamate dynamics by EOS.

A, a schematic of the EOS structure. EOS is composed of the recombinant S1S2 domain of GluR2 (in magenta) and a conjugated fluorescent dye (in green). B, dose–response relationship between glutamate concentration and fractional change in fluorescence intensity (ΔF/F₀) of EOS sister indicators. Estimated K_d values of K716A-EOS and L401C-EOS were 174 nm and 1.57 μm, respectively. C, EOS immobilised in the molecular layer of cerebellar slices. The magnified image (lower panel) shows a roughly even distribution of EOS in the extracellular space with silhouettes of cellular components. D, imaging of EOS signals in response to PF stimulation (5 pulses at 100 Hz). ΔF/F₀ is colour coded as indicated. Modified from Okubo et al. (2010), copyright (2010) National Academy of Sciences, USA.

Spatiotemporal summation of glutamate spillover

Modelling studies predicted that glutamate released from a single exocytotic vesicle is immediately diluted by passive diffusion and active reuptake via glutamate transporters, which confines the activation of glutamate receptors within the synaptic cleft (Barbour, 2001). On the other hand, model-based analyses also showed that glutamate released within a narrow time window from a single synapse in succession and/or from neighbouring synapses may be summed in the extrasynaptic space (Barbour & Hausser, 1997). Thus, spatially and temporally packed synaptic activity would be expected to induce extrasynaptic glutamate dynamics that may be sufficient for volume transmission. Indeed, electrophysiological measurements suggested that both the density of stimulated nerve fibres and the number of repetitive stimulation pulses were critical for the activation of extrasynaptic glutamate receptors (Carter & Regehr, 2000; Arnth-Jensen et al. 2002; Marcaggi & Attwell, 2005). However, the extrasynaptic glutamate dynamics had not been directly measured.

EOS-based glutamate imaging of the PF–Purkinje cell synapse in the cerebellar cortex has been used to address this issue. While a single PF stimulation pulse generated only a faint EOS signal, two to five pulses at 100 Hz induced a significant EOS response (Fig. 3A, upper panel). Thus, the extrasynaptic glutamate dynamics around PF–Purkinje cell synapses had been finally caught in action, and the results indicated that the extrasynaptic glutamate dynamics indeed undergo temporal summation. Another important advancement made with EOS-based measurement was the ability to directly estimate the extrasynaptic glutamate concentration. Due to a relatively low rate constant of glutamate–EOS dissociation, EOS is a leaky integrator of the glutamate concentration. Therefore, the glutamate concentration during synaptic inputs can be estimated by the numerical deconvolution of EOS signals. The time course of the estimated glutamate transients reaches micromolar concentrations for tens of milliseconds in response to two to five pulses of stimulation (Fig. 3A, lower panel). Such extrasynaptic glutamate dynamics are sufficient to activate high-affinity glutamate receptors, including NMDAR (EC₅₀≈ 2 μm; Clements & Westbrook, 1991) and mGluR (EC₅₀≈ 10 μm; Hayashi et al. 1993). This is consistent with previous findings that at least two pulses of repetitive PF input are required to activate perisynaptic mGluRs in Purkinje cells (Finch & Augustine, 1998; Takechi et al. 1998; Okubo et al. 2004; Marcaggi & Attwell, 2005). The EOS measurements were subsequently performed in hippocampal and neocortical slice preparations, with essentially the same results (Okubo et al. 2010).

Figure 3. Direct evaluation of extrasynaptic glutamate dynamics by EOS imaging.

A, temporal summation in extrasynaptic glutamate dynamics upon repetitive PF stimulation in the cerebellar cortex and estimation of corresponding glutamate concentration. Upper traces show the time course of the EOS signal and the lower traces show the corresponding time course of glutamate concentration estimated by the deconvolution method (Okubo et al. 2010). Magenta traces indicate averaged data. B, spatial summation in extrasynaptic glutamate dynamics. The simultaneous measurement of EOS signals and EPSCs upon PF stimulation (5 pulses at 100 Hz) shows a correlation between extrasynaptic glutamate dynamics and PF input density. C, EOS signal upon 5, 10 and 20 pulses of PF stimulation at 100 Hz. Increasing the number of stimulation pulses increased the spatial width of the EOS signal. Peripheral signals emerged after 10 and 20 pulses of stimulation that showed significant latency. D, extrasynaptic glutamate dynamics upon sensory input from the hind paw. EOS was immobilised in the rat sensory cortex corresponding to the hind paw (left and middle panels). Peak EOS signal induced by 200 ms tactile stimulation to the hind paw is shown (right panel). Modified from Okubo et al. (2010), copyright (2010) National Academy of Sciences, USA.

Spatial summation of extrasynaptic glutamate dynamics has also been analysed in the vicinity of small bundles of activated PFs. EOS imaging was carried out simultaneously with PF-induced excitatory postsynaptic current (EPSC) measurements from Purkinje cells as an indicator of the number of active synapses. The EOS signal increased with EPSC, showing the dependence of extrasynaptic glutamate dynamics on the density of stimulated PFs (Fig. 3B). The summation of glutamate released from adjacent PF synapses is consistent with the dependence of mGluR activation on the PF input density (Marcaggi & Attwell, 2005).

Limiting concentration of glutamate spillover

EOS imaging methods provide direct evidence that synaptic activity generates an increase in the local glutamate concentration that exceeds micromolar levels for tens of milliseconds in the extrasynaptic space. Although such extrasynaptic glutamate dynamics are sufficient to activate high-affinity glutamate receptors including NMDAR and mGluR as discussed above, they are insufficient to activate AMPAR (EC₅₀≈ 500 μm; Jonas & Sakmann, 1992). Thus, the activation of extrasynaptic AMPAR must take place only in very close proximity to synaptic clefts or ectopic glutamate release sites, where glutamate concentrations may approach millimolar concentrations (DiGregorio et al. 2002; Matsui & Jahr, 2003).

Spatial spreading distance of glutamate spillover

Prior to the introduction of glutamate imaging techniques, the spatial spreading of glutamate in the extrasynaptic space, i.e. the working distance of glutamatergic volume transmission, had not been directly measured. EOS-based glutamate imaging can provide a direct measurement of spreading distance of glutamate spillover. In response to a physiologically relevant stimulation (5 pulses of PF stimulation at 100 Hz; Fig. 3C, left panel), there was an immediate increase in the EOS signal, the width of which is almost constant during the imaging period. This observation indicated that glutamate concentration increases only within a region that contains active synapses, and that the glutamate signal is effectively contained within that region (core region). However, when prolonged stimulation with 10–20 pulses was applied, additional EOS signals with latency appear on both sides of the core region (Fig. 3C, middle and right panels). This suggested that under such conditions glutamate cannot be sequestered within the core region and overflows laterally for a distance of a few micrometres. However, even in such an extreme condition, the distance of glutamate spreading is no greater than a few micrometres. Thus, glutamatergic volume transmission between components separated by over several micrometres is not expected to occur in physiologically relevant conditions. These data set the upper limit of the effective range of glutamate spillover.

Glutamate spillover imaging in vivo

To further evaluate extrasynaptic glutamate dynamics under physiological conditions, in vivo glutamate imaging has been carried out in the brain (Fig. 3D). In these studies, EOS was immobilised in the rat sensory cortex through the biotin–streptavidin interaction as in slice preparations. In response to a tactile stimulation of the hind paw, significant EOS signals were observed at the appropriate sensory area through a cranial window under a wide-field fluorescence microscope (Fig. 3D, right panel). This indicated that glutamatergic volume transmission takes place under physiological conditions. Together with the finding that sensory input from whiskers induces mGluR-mediated astrocytic Ca²⁺ responses at the corresponding cortical sensory area in vivo (Wang et al. 2006), these results showed that physiological inputs are sufficient to induce volume transmission via high-affinity glutamate receptors in the brain in vivo.

Conclusion and perspectives

In conclusion, the research summarised here indicates that the extrasynaptic glutamate concentration reaches sufficient levels to activate high-affinity glutamate receptors in response to spatiotemporally summable synaptic activities. Thus, glutamate is a physiological volume transmitter and may regulate important neuronal, glial and vascular functions. The EOS-based glutamate imaging method may be used to obtain further information regarding glutamatergic volume transmission. One of the most promising future applications of EOS is the direct evaluation of glutamate release by astrocytes as a gliotransmitter. As described above, controversy exists regarding Ca²⁺-triggered glutamate release under the physiologically relevant conditions. Since the extracellular glutamate concentration upon agonist activation of astrocytes may reach 1–6 μm (Lee et al. 2007), glutamate release from astrocytes may be detected by EOS. Furthermore, in vivo glutamate imaging may be used to locate active synapses during physiological responses of the brain, providing a new modality for functional brain imaging.

Glossary

Abbreviations

AMPAR

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor

CFP

cyan fluorescent protein

EOS

glutamate (E) optical sensor

EPSC

excitatory postsynaptic current

FRET

fluorescence resonance energy transfer

mGluR

metabotropic glutamate receptor

NMDAR

N-methyl-d-aspartate receptors

PF

parallel fibre

YFP

yellow fluorescent protein