Abstract
Neurotransmitter spillover is a form of communication not readily predicted by anatomic structure. In the cerebellum, glutamate spillover from climbing fibers recruits molecular layer interneurons in the absence of conventional synaptic connections. Spillover-mediated signaling is typically limited by transporters that bind and reuptake glutamate. Here, we show that patterned expression of the excitatory amino acid transporter 4 (EAAT4) in Purkinje cells regulates glutamate spillover to molecular layer interneurons. Using male and female Aldolase C-Venus knock-in mice to visualize zebrin microzones, we find larger climbing fiber-evoked spillover EPSCs in regions with low levels of EAAT4 compared with regions with high EAAT4. This difference is not explained by presynaptic glutamate release properties or postsynaptic receptor density but rather by differences in the glutamate concentration reaching receptors on interneurons. Inhibiting glutamate transport normalizes the differences between microzones, suggesting that heterogeneity in EAAT4 expression is a primary determinant of differential spillover. These results show that neuronal glutamate transporters limit extrasynaptic transmission in a non–cell-autonomous manner and provide new insight into the functional specialization of cerebellar microzones.
SIGNIFICANCE STATEMENT Excitatory amino acid transporters (EAATs) help maintain the fidelity and independence of point-to-point synaptic transmission. Whereas glial transporters are critical to maintain low ambient levels of extracellular glutamate to prevent excitotoxicity, neuronal transporters have more subtle roles in shaping excitatory synaptic transmission. Here we show that the patterned expression of neuronal EAAT4 in cerebellar microzones controls glutamate spillover from cerebellar climbing fibers to nearby interneurons. These results contribute to fundamental understanding of neuronal transporter functions and specialization of cerebellar microzones.
Introduction
Neuronal circuitry can be established through anatomic analysis of synapses, the sites of direct communication between neurons. However, anatomy cannot predict spillover transmission, in which neurotransmitters escape from the synaptic cleft and activate receptors on neighboring cells or synapses (Asztely et al., 1997; Kullmann, 2000). This type of extrasynaptic communication occurs throughout the brain where it augments canonical point-to-point synaptic transmission by recruiting extrasynaptic receptor activation (Scanziani et al., 1996; Asztely et al., 1997; Isaacson, 1999; Scanziani, 2000; Chalifoux and Carter, 2011; Henneberger et al., 2020). Spillover signaling is prominent in the cerebellum where it occurs in the absence of conventional anatomically defined synapses (Nishiyama and Linden, 2007; Szapiro and Barbour, 2007; Mathews et al., 2012; Coddington et al., 2014; Nietz et al., 2017). While these anatomically independent circuit motifs are engaged during sensory processing and contribute to cerebellar computations (Jörntell and Ekerot, 2002; Arlt and Hausser, 2020), the factors controlling its reach and regulation are not fully understood.
Cerebellar climbing fibers (CFs) form powerful synapses with Purkinje cells (PCs) comprised of hundreds of individual synaptic contacts (Palay and Chan-Palay, 1974). Although individual synaptic contacts are nearly completely ensheathed by Bergmann glia that express glutamate transporters excitatory amino acid transporters 1 (EAAT1) and 2 (EAAT2) (Rothstein et al., 1994; Tanaka et al., 1997; Xu-Friedman et al., 2001; Huang and Bergles, 2004; Tsukada et al., 2005; Tsai et al., 2012), glutamate can escape the CF-PC synapse to activate receptors on molecular layer interneurons (MLIs) (Kollo et al., 2006; Szapiro and Barbour, 2007; Mathews et al., 2012; Coddington et al., 2013). One molecular mechanism that may regulate the extent of CF-mediated spillover is the expression levels of EAAT4, a high-affinity glutamate transporter located on PCs (Dehnes et al., 1998; Brasnjo and Otis, 2004; Takayasu et al., 2004, 2005). The expression of EAAT4 follows a parasagittal banding pattern similar to the expression of aldolase C (zebrin), creating microzones of molecularly diverse PCs with high and low levels of EAAT4 (Dehnes et al., 1998; Wadiche and Jahr, 2005; Tsai et al., 2012). In contrast, Dehnes et al. (1998) showed that GLAST, GLT, and EAAC (EAAT1-3, respectively) are uniformly expressed throughout the cerebellum. Although glial EAAT1 and EAAT2 are responsible for the reuptake of the majority of glutamate released at the CF-PC synapse, the perisynaptic location of EAAT4 positions it to fine-tune spillover in a synapse-specific manner (Tanaka et al., 1997; Dehnes et al., 1998). We previously showed that patterned expression of EAAT4 mediates differences in parallel fiber (PF)-mediated activation of PC mGluRs as well as extrasynaptic glutamatergic transmission to adjacent Bergmann glia (Wadiche and Jahr, 2005; Tsai et al., 2012). It is not known, however, whether this physiological difference in transporter expression is sufficient to exert a non–cell-autonomous regulation of glutamatergic signaling from CFs to MLIs.
To test whether patterned expression of EAAT4 controls the amount of glutamate spillover reaching MLIs, we use mice expressing Venus under the Zebrin II/Aldolase C promoter to compare CF-mediated spillover in high and low EAAT4 microzones (Fujita et al., 2014). We assay postsynaptic NMDAR and AMPAR activation in MLIs and use AMPAR low-affinity antagonists to measure the relative differences in glutamate concentration. We conclude that endogenous differences in EAAT4 expression are sufficient to regulate the extent of spillover transmission onto neighboring MLIs despite glial ensheathment of CF synapses (Xu-Friedman et al., 2001) and differential glutamate release between zebrin microzones (Paukert et al., 2010). These results highlight the functional significance of high-affinity neuronal transporters that appear to provide privileged control of synaptically released glutamate beyond the physical and transporter barriers provided by glia (Tzingounis and Wadiche, 2007; Scimemi et al., 2009).
Materials and Methods
Experimental model and subject details
We used male and female knock-in mice with the GFP Venus inserted into exon 2 of the Aldolase C gene on C57BL/6N background (Aldoc-Venus mice) (Fujita et al., 2014). Homozygous mice were maintained in a 12:12 h light:dark cycle and used between postnatal days 12 and 25. Some functional alterations resulting from diminished expression of aldolase C cannot be ignored in experiments using Aldoc-Venus mice. However, several reasons suggest that such alterations may not be significant. First, the expression of aldolase C only starts at postnatal stages (after P9-P14) in PCs (Fujita et al., 2014), around the time of the recordings. Second, the gross shape and size of the brain are indistinguishable among the WT, heterozygote, and homozygote of Aldoc-Venus mice (Fujita et al., 2014). Third, the zebrin pattern is the same among the WT, heterozygote, and homozygote of Aldoc-Venus mice (Fujita et al., 2014). Fourth, general motor performance examined by the rotarod test showed no significance differences among the WT, heterozygote, and homozygote of Aldoc-Venus mice (Luo et al., 2020). All experiments were conducted through protocols approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham under protocol IACUC-08767.
Slice preparation
Mice were decapitated under isoflurane (VetOne), and the cerebellum was rapidly dissected into ice-cold cutting solution containing the following (in mm): 110 choline chloride, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 25 glucose, 11.5 sodium ascorbate, 3 sodium pyruvate, 0.5 CaCl2, and 7.0 MgCl2, bubbled with 95% O2, 5% CO2. The cerebellum was placed in a parasagittal orientation on an agar block, and sections from the vermis (300 µm) were taken using a vibratome (VT1200S, Leica Instruments). Slices were transferred to 35°C ACSF for 30 min and then stored at room temperature for another 30 min. The ACSF contained the following (in mm): 125 NaCl, 2.5 KCl, 1.0 NaH2PO4, 26.2 NaHCO3, 11 glucose, 2.5 CaCl2, and 1.3 MgCl2.
Electrophysiology
Recordings were made from slices maintained at ∼32°C using an inline heater (ALA Scientific Instruments) with constant flow of ACSF at ∼3 ml/min. A 60× water immersion objective on an upright microscope (Olympus) was used to identify PCs and MLIs that were voltage-clamped using a Multiclamp 700B amplifier controlled by pClamp 10.6 software (Molecular Devices). Currents were filtered at 2-5 kHz and digitized at 15-50 kHz using a Digidata 1440A AD converter (Molecular Devices). Patch pipettes were pulled using a P-97 horizontal puller (Sutter Instruments). ACSF contained 100 µm picrotoxin to block GABAA receptors (Abcam).
MLI recordings
Single CFs were stimulated near the PC layer using a theta glass pipette (Sutter Instruments) filled with bath solution. MLIs in the inner and middle thirds of the molecular layer were patched using thick-wall glass pipettes pulled to a resistance of 2.5-5 mΩ. The position and intensity of the stimulating pipette were adjusted until a single CF (all-or-none response) was detected and PF responses were absent.
Only MLIs and PCs with a single CF were included in our analysis. The pipette solution contained the following (in mm): 100 CsMeSO3, 50 CsCl, 10 HEPES, 10 EGTA, 1 MgCl2, 2 MgATP, 0.3 NaGTP, and 5 QX-314, adjusted to pH 7.2 with CsOH. Cells were held at −60 mV unless otherwise noted, and series resistance (Rs) was measured by the instantaneous current response to a −5 mV step. Rs values were 15–20 mΩ. Rs was uncompensated and monitored throughout recordings.
PC recordings
Patch pipettes were pulled from thin-wall borosilicate glass (Sutter Instruments) to a resistance of 0.8-2.0 mΩ and filled with solution containing the following (in mm): 110 CsCl, 35 CsF, 10 HEPES, 10 EGTA, and 5 QX-314 adjusted to pH 7.2 with CsOH. Rs, measured by the instantaneous current response to a −2 mV step with only pipette capacitance canceled, was <5 mΩ and was routinely compensated >80%. Resistance was monitored throughout the recording, and experiments were discarded if substantial changes were observed (>20%). CFs were stimulated with theta glass electrodes (BT-150, Sutter Instruments) filled with 5% NaCl driven by a constant current isolated stimulator (Digitimer North America) and placed in the granule cell layer.
Drug treatments
Drug concentrations are listed in the text or figure legends. Picrotoxin (GABAAR antagonist), NBQX (AMPAR antagonist), QX-314 (Na+-channel blocker), (R)-CPP (NMDA antagonist), and kynurenic acid (KYN, low-affinity antagonist) were purchased from Abcam. DL-TBOA (excitatory-amino-acid-transporter blocker) was purchased from HelloBio.
Quantification and statistical analysis
AxoGraph X (Axograph Scientific) and GraphPad Prism (GraphPad) were used for data analysis. Reported values are ± SEM, and the statistical tests used are stated in the figure legends. Means were compared using paired or unpaired two-tailed t tests or one- or two-way ANOVAs with Holm-Sidak correction for multiple comparison tests. Extra sum-of-squares F tests were used to compare curves fit to different groups within the same experiment. The criterion for statistical significance was p < 0.05, noted by asterisk (*) in figure and exact p-value in text and legend.
Results
Aldolase C Venus fluorescence correlates with EAAT4 expression
We first confirmed that cerebellar microzones defined by zebrin expression exhibit differential EAAT4 expression. We used mice with the GFP Venus inserted into exon 2 of the Aldolase C gene to visualize the well-known zebrin banding pattern (Aldoc-Venus mice) (Fujita et al., 2014), and compared Venus with the pattern of EAAT4 immunostaining (Fig. 1A–C). We found that Venus and EAAT4 expression were correlated (r = 0.81 ± 0.04; n = 6 regions from 3 sections; Fig. 1D), indicating that Venus fluorescence is a good proxy for EAAT4 expression.
Glutamate spillover differs between low and high EAAT4 regions
Although CF-PC synapses are nearly completely ensheathed by Bergmann glia (Xu-Friedman et al., 2001), glutamate receptors on neighboring MLIs are activated following CF stimulation despite no anatomic connection (Jörntell and Ekerot, 2003; Szapiro and Barbour, 2007; Mathews et al., 2012; Coddington et al., 2013). This is likely because of multivesicular release at CF to PC synapses that results in a high synaptic glutamate concentration of ∼10 mm at each release site (Wadiche and Jahr, 2001; Rudolph et al., 2011; Vaden et al., 2019). Guided by Venus fluorescence, we recorded CF-evoked spillover EPSCs in MLIs from low and high EAAT4 microzones (Fig. 1E,F). In the presence of picrotoxin (GABAAR antagonist) and R-CPP (NMDAR antagonist), we adjusted the stimulus intensity to evoke an all-or-none CF EPSC without PF contamination using multiple criteria, including the following: (1) all-or-none recruitment, (2) paired-pulse depression, and (3) slow rise and decay times (Coddington et al., 2013; Nietz et al., 2017). AMPAR-mediated CF EPSCs were larger in low EAAT4 regions, measured either by peak amplitude (low, black circles: 70.1 ± 8.2 pA, n = 19; high, green squares: 39.9 ± 6.1 pA, n = 21; p = 0.005; unpaired t test; Fig. 1G) or charge (low: 323.1 ± 46.1 fC, n = 19; high: 197.8 ± 25.0 fC, n = 20; p = 0.02; unpaired t test). Additionally, we found that EPSCs had faster rise times (low: 0.67 ± 0.05 ms, n = 19; high: 0.87 ± 0.08 ms, n = 21; p = 0.046; unpaired t test) and decay times (low: 3.4 ± 0.3 ms, n = 19; high: 4.2 ± 0.3 ms, n = 19; p = 0.04; unpaired t test) in low EAAT4 regions (Fig. 1G). Since the number of receptors activated and their binding rate are proportional to agonist concentration (Patneau and Mayer, 1991), the larger EPSC amplitude and faster kinetics suggest a greater [glutamate] reaches MLIs in low EAAT4 regions.
These data are consistent with the idea that EAAT4 expression level regulates extrasynaptic [glutamate], but an alternative explanation is that there are differences in AMPAR density between microzones. To address this possibility, we recorded CF-MLI EPSCs at −60 mV and 40 mV and isolated the NMDAR-mediated EPSCs by applying the NMDAR antagonist R-CPP (5 µm) and subtracting the CPP trace from the control trace (Fig. 2A,B). Similar to AMPAR-mediated EPSCs, the amplitude of the NMDAR-mediated response was significantly larger in low EAAT4 regions than in high EAAT4 regions (low: 29.1 ± 6.7 pA, n = 6; high: 5.6 ± 0.7 pA, n = 5; p = 0.01; unpaired t test; Fig. 2A, right). When we normalized the NMDAR current to the AMPAR-mediated current at −60 mV, there was no difference in the NMDA/AMPA ratio between high and low EAAT4 regions (low: n = 7; high: n = 6; p > 0.05; unpaired t test; Fig. 2B, right). Together, these results argue that differences in receptor expression are not responsible for the distinct spillover responses recorded in high- and low-expressing EAAT4 regions, with the unlikely caveat that the expression of both receptor subtypes is similarly regulated.
To further test the idea that differences in [glutamate] between regions underlie the differences in EPSC properties, we compared the sensitivity of CF-MLI EPSCs to the low-affinity AMPAR antagonist KYN. Unlike high-affinity antagonists (i.e., NBQX) that have a slow unbinding rate, KYN unbinds AMPARs quickly and can be replaced by endogenous glutamate. Thus, inhibition by KYN depends on the [glutamate] and is more effective when the [glutamate] is low. Application of KYN (250 µm) differentially blocked spillover transmission across EAAT4 microzones, reducing the EPSC by 49.8 ± 4.8% (n = 12) in low EAAT4 regions and 69.0 ± 2.3% in high EAAT4 regions (n = 10; p = 0.003; unpaired t test; Fig. 3A,B). As a control, we found that a sub-saturating concentration of the high-affinity antagonist NBQX (100 nm) equally inhibited EPSCs in both regions (low: 51.7 ± 2.8%, n = 4; high: 51.0 ± 4.1%, n = 5; p > 0.05; unpaired t test; Fig. 3A,B). We were surprised there was no correlation between the % KYN inhibition and the EPSC amplitude (data not shown), suggesting that the amplitude depends not only on the [glutamate] but also the number of available receptors and other factors, such as desensitization. Thus, we turned to a more sensitive measure of [glutamate], the rise time (Coddington et al., 2013). The spillover EPSC rise time is likely more sensitive than the peak amplitude because it less dependent on the number of receptors available to bind glutamate. Indeed, there was a significant correlation between the % KYN inhibition versus rise time for low and high Venus recordings (Fig. 3C). These data support the idea that the rise time of spillover EPSCs reflects the underlying [glutamate] rather than postsynaptic properties, such as receptor density: slower-rising EPSCs are more sensitive to KYN inhibition in both low and high Venus recordings.
Differences in EPSC amplitude are not because of differences in release probability
Our results suggest that spillover EPSCs in high EAAT4 regions are smaller than EPSCs in low EAAT4 regions because of a lower [glutamate] transient. Differences in [glutamate] could result from differences in CF release properties between microzones (Paukert et al., 2010). To test a potential contribution of presynaptic release properties, we compared the paired-pulse ratio (PPR) in low and high EAAT4 regions. The PPR is commonly used to assay release probability (Pr), which together with the size of the readily releasable pool, dictates the extent of multivesicular release and the synaptic [glutamate] (Rudolph et al., 2011; Vaden et al., 2019). We first recorded from PCs, the primary target of CFs, with paired stimuli at various interstimulus intervals (10 ms to 10 s). Recordings were performed in KYN (250 µm) to reduce receptor saturation, leading to a low PPR that recovered within 10 s, as expected for depletion and recovery of released vesicles (Foster et al., 2002). There was no difference in PPR between high- or low-expressing EAAT4 regions at any interstimulus interval (Fig. 4A; p > 0.05), suggesting that differences in presynaptic properties of CFs do not account for the differences in spillover between the two regions. However, since even low levels of receptor saturation might confound precise measurement of PPRs, we repeated the experiment in MLIs with the idea that the low [glutamate] underlying spillover will assure that receptors are far from saturation. Using this assay, we found a slight but significantly greater paired-pulse depression in high EAAT4 regions (Fig. 4B), consistent with lower Pr reported in zebrin-negative zones using other approaches (Paukert et al., 2010). Together, these results confirm that smaller spillover EPSCs in high EAAT4 zones cannot be explained by presynaptic differences since high EAAT4 zones exhibit higher Pr. Our results further argue that high EAAT4 expression is sufficient to limit extrasynaptic glutamate in zebrin-positive zones despite higher Pr at these CF synapses.
Transporter blockers normalize differences in spillover transmission
To directly test whether glutamate uptake capacity underlies differences in glutamate spillover to MLIs, we compared MLI EPSCs in the presence of TBOA (50 µm), an inhibitor of all EAATs. As expected, TBOA significantly increased the EPSC amplitude in both regions (low: from 52.5 ± 15.4 pA to 76.0 ± 17.5 pA, n = 7, p = 0.004; high: from 20.3 ± 4.2 pA to 56.7 ± 9.4 pA, n = 8, p = 0.0004; paired t test; Fig. 5A,B). Importantly, the effect of TBOA was more pronounced in high EAAT4 regions (increased by 296 ± 8% vs 160 ± 9%, n = 8, n = 7; p = 0.0005; Fig. 5C), and there was no difference in EPSC amplitude between low and high EAAT4 regions when EAATs were inhibited (low: 76.0 ± 17.5 pA, n = 7; high: 56.7 ± 9.4 pA, n = 8; p > 0.05; unpaired t test; Fig. 5D). Since the expression levels of glial transporters are similar across the cerebellum (Dehnes et al., 1998), these results suggest that differences in EAAT4 are responsible for differences in EPSC amplitude. To test whether EAAT inhibition also normalized the glutamate transients underlying spillover EPSCs, we compared the sensitivity of CF-MLI EPSCs in the presence of TBOA to KYN, and NBQX. Indeed, there were no differences in KYN or NBQX inhibition between low and high EAAT4 regions in the presence of TBOA (KYN in TBOA low: 53.8 ± 5.8%, n = 6; high: 65.2 ± 6.1%, n = 6; p > 0.05; NBQX in TBOA low: 57.0 ± 2.2%, n = 5; high: 53.8 ± 3.0%, n = 7; p > 0.05; Fig. 5E). Thus, TBOA normalized both the spillover EPSC amplitude and the underlying extrasynaptic [glutamate] between high and low EAAT4 microzones.
Discussion
The stereotyped cytoarchitecture of the cerebellum contrasts with the molecular patterning of gene expression, exemplified by zebrin bands and expression of EAAT4, that represents distinct microzones. Here we show new functional consequences resulting from the differential expression of EAAT4 across microzones that corroborate its rate-limiting role in extrasynaptic glutamate signaling. We visualized the parasagittal banding pattern using Aldoc-Venus mice, showing that Venus correlates with EAAT4 expression. We found that CF-mediated spillover EPSCs to MLIs are smaller in high EAAT4 zones compared with low EAAT4 zones. This difference is explained by the relative [glutamate] reaching receptors on MLIs rather than differences in postsynaptic receptor density or presynaptic release properties. Indeed, despite evidence that CFs in high EAAT4 zones have higher release probability and greater multivesicular release (Paukert et al., 2010), spillover EPSCs were smaller in MLIs from high EAAT4 regions. As both the spillover EPSC and [glutamate] are normalized by EAAT inhibition, together these results indicate that EAAT4 levels are rate-limiting for controlling extrasynaptic glutamate signaling from CFs in the cerebellar cortex.
Regulation of synaptic transmission by EAATs
EAAT4 is one of five isoforms of sodium-dependent glutamate transporters with the primary function of maintaining low extracellular glutamate concentration throughout the CNS (Tzingounis and Wadiche, 2007). EAAT1 and EAAT2, located on glial cells, are responsible for the reuptake of the majority of the glutamate and prevent glutamate excitotoxicity (Rothstein et al., 1994; Tanaka et al., 1997; Danbolt, 2001; Huang et al., 2004, 2005). EAAT3, EAAT4, and EAAT5 are neuronal glutamate transporters responsible for additional functions along with contributing to maintaining low levels of extracellular glutamate (Brasnjo and Otis, 2001; Veruki et al., 2006; Scimemi et al., 2009). Additionally, EAAT3 is proposed to transport glutamate into GABAergic neurons for GABA synthesis (Sepkuty et al., 2002; Mathews and Diamond, 2003). While transporters can have subtle effects on the [glutamate] at synaptic receptors by rapid buffering (Diamond and Jahr, 1997), extensive evidence suggests that transporters primarily control the spread of transmitter outside and between synapses (Tzingounis and Wadiche, 2007). This explains why transport blockers typically have minimal effects on synaptic currents (Hestrin et al., 1990; Sarantis et al., 1993), or peak synaptic [glutamate] (Wadiche and Jahr, 2001), yet can profoundly alter the properties of compound synaptic currents evoked by simultaneous activation of multiple synapses or synaptic currents at specialized synapses (Isaacson and Nicoll, 1993; Asztely et al., 1997; Arnth-Jensen et al., 2002; Bagnall et al., 2011). Indeed, the contribution of extrasynaptic (or “spillover”) signaling to direct point-to-point transmission at both glutamatergic and GABAergic synapses can be assessed by the relative sensitivity to inhibition of respective transmitter transporters (Isaacson et al., 1993; Scanziani et al., 1997; Overstreet and Westbrook, 2003; Coddington et al., 2013, 2014; Nietz et al., 2017).
It is generally recognized that the majority of glutamate clearance is mediated by diffusion to glial transporters, but our results point to an important role of neuronal EAAT4 in controlling extrasynaptic glutamate signaling at cerebellar synapses. It has been estimated that as much as 20% of the glutamate released from CFs may be taken up by EAAT4 (Otis et al., 1997) curtailing the PC EPSC duration (Takahashi et al., 1996). EAAT4 is unique in that it is a high-affinity neuronal transporter (10-fold higher than GLAST, GLT, and EAAC) (Arriza et al., 1994; Fairman et al., 1995) present exclusively on PCs (Dehnes et al., 1998; Takayasu et al., 2005). Its postsynaptic localization just outside of CF and PF synapses predicts that even subtle differences in its expression level could influence extrasynaptic glutamate signaling (Dehnes et al., 1998; Zheng et al., 2008). Accordingly, the patterned expression of EAAT4 significantly influences the extent of perisynaptic mGluR activation by glutamate released at PF synapses, as well as activation of AMPARs on adjacent Bergmann glia (Wadiche and Jahr, 2005; Tsai et al., 2012). Since EAAT4 influences signaling to glial cell processes that ensheath PF synapses, we reasoned that EAAT4 may also control extrasynaptic signaling to MLIs that respond to glutamate spillover from CF-PC synapses (Szapiro and Barbour, 2007; Coddington et al., 2013). Indeed, we found that a greater [glutamate] reaches MLIs in low EAAT4 zones than in high EAAT4 zones. When ruling out potential presynaptic differences in glutamate release, we found that CFs in high EAAT4 regions may release more glutamate than CFs in low EAAT4 regions. Despite greater release of glutamate from CFs in zebrin-positive (high EAAT4) zones via multivesicular release (Paukert et al., 2010), a lower [glutamate] reaches glutamate receptors on MLIs.
NMDARs and AMPARs on MLIs are differentially distributed at extrasynaptic and synaptic locations, respectively (Clark and Cull-Candy, 2002). Our results showing NMDAR activation by single CF stimuli are thus consistent with extrasynaptic NMDAR localization. The constant AMPAR/NMDAR ratio across zebrin zones is potentially surprising, considering that higher-affinity NMDARs would be expected to operate within a different region of the glutamate dose–response relationship. However, it is likely that NMDARs and AMPARs are not exposed to the same [glutamate], making it difficult to predict their relative responses. Furthermore, extrasynaptic levels of glutamate resulting from spillover are sub-saturating for both receptor subtypes (Dzubay and Jahr, 1999). Additional studies will be required to identify the precise location of MLI receptors activated by CF spillover, as well as quantify the underlying [glutamate] transient at each receptor subtype.
Implications for MLI function and feedforward inhibition
Zebrin microzones comprise anatomically distinct olivocerebellar modules, with PCs in each zone exhibiting specific input and output projections with the inferior olive and cerebellar nuclei, respectively (Sugihara and Quy, 2007; Ruigrok, 2011). Distinct connectivity and computational power likely underlie distinct cerebellar functions, as zebrin-positive modules in the vestibular cerebellum contribute to adaptation of compensatory eye movements, whereas learning during eyeblink conditioning is associated with zebrin-negative modules (Mostofi et al., 2010; Zhou et al., 2014). Furthermore, it was recently shown that PCs in zebrin-positive zones exhibit relatively lower simple spike firing rates than in zebrin-negative zones, in part because of differential activity of transient receptor potential cation channel C3 (TRPC3), since genetic manipulation of TRPC3 selectively alters the firing rate and behavior associated with zebrin-negative zones (Zhou et al., 2014; Wu et al., 2019). Both LTD of PF inputs and feedforward inhibition by MLIs contribute to PC simple spike suppression during Pavlovian eyeblink conditioning (Boele et al., 2018). EAAT4 expression that can regulate both of these processes (see below) may thus regulate cerebellar plasticity in a microzone-specific fashion and play a role in associative and motor learning. Understanding how the molecular diversity of PCs contributes to distinct cellular functionality may provide insight into specialization of modules that mediate heterogeneous network function.
How might EAAT4 contribute to specialization of cerebellar modules? EAAT4 controls the extent of extrasynaptic glutamate signaling, a function previously shown to regulate synaptic plasticity and simple spiking. In low EAAT4 microzones, glutamate released from PFs can overwhelm local transport capacity to activate perisomatic mGluRs, whereas under the same conditions mGluR activation is absent in high EAAT4 zones (Wadiche and Jahr, 2005). There are multiple consequences of mGluR activation, including a depolarizing conductance mediated by TRPC3, endocannabinoid production that can cause presynaptic inhibition, and induction of LTD of PF synaptic transmission (Kim et al., 2003; Brenowitz and Regehr, 2005). Indeed, low EAAT4 zones exhibit robust LTD of PF synaptic transmission that is absent in high EAAT4 regions (Wadiche and Jahr, 2005). EAAT4-mediated control of extrasynaptic mGluR and NMDAR activation is also associated with differential rates of PC simple spiking (Perkins et al., 2018). Thus, EAAT4 expression plays a cell-autonomous role in PC plasticity and spiking.
In addition to the cell-autonomous function, our new data suggest that EAAT4 regulates activation of glutamate receptors on MLIs. CFs do not appear to form anatomically defined synapses with MLIs, but glutamate release from CFs activates AMPARs and NMDARs via spillover (Coddington et al., 2013, 2014). Glutamate spillover from a single CF is sufficient to generate both feedforward inhibition and disinhibition of neighboring PCs, with PCs near the active CF receiving a biphasic sequence of inhibition/disinhibition whereas distant PCs exhibit solely disinhibition (Mathews et al., 2012; Coddington et al., 2013). The ability of CFs to excite and inhibit MLIs depends on the proximity of MLIs to CFs, and may explain the response of MLIs to sensory stimulation (Arlt and Hausser, 2020). Regulation of CF spillover to MLIs by EAAT4 expression suggests an additional layer of organization wherein the functionality of these motifs differs across microzones.
Footnotes
The authors declare no competing financial interests.
This work was supported by NIH F32NS110154 to R.P., NIH R01NS113948 to J.I.W., and NIH R01NS105438 to L.O.-W. We thank Drs. Luke Coddington and Angela Nietz for pilot experiments.
- Correspondence should be addressed to Linda Overstreet-Wadiche at lwadiche{at}uab.edu or Jacques I. Wadiche at jwadiche{at}uab.edu