Postsynaptic Glutamate Transport at the Climbing Fiber-Purkinje Cell Synapse

See allHide authors and affiliations

Science  05 Sep 1997:
Vol. 277, Issue 5331, pp. 1515-1518
DOI: 10.1126/science.277.5331.1515


The role of postsynaptic, neuronal glutamate transporters in terminating signals at central excitatory synapses is not known. Stimulation of a climbing fiber input to cerebellar Purkinje cells was shown to generate an anionic current mediated by glutamate transporters. The kinetics of transporter currents were resolved by pulses of glutamate to outside-out membrane patches from Purkinje cells. Comparison of synaptic transporter currents to transporter currents expressed in Xenopus oocytes suggests that postsynaptic uptake at the climbing fiber synapse removes at least 22 percent of released glutamate. These neuronal transporter currents arise from synchronous activation of transporters that greatly outnumber activated AMPA receptors.

The glutamate transporters EAAT3 and EAAT4 are expressed at high levels in cerebellar Purkinje cells (PCs) (1). Transporters on PCs exhibit substrate-induced anion currents (2) similar to those associated with other native (3, 4) and cloned glutamate transporters (5-7). Because the conductance associated with the glutamate transporter has a high permeability to NO3 and SCN(4, 6), they were used as the major intracellular anions in PC recordings. Extracellular stimulation in the granule cell layer of cerebellar slices elicited all-or-none excitatory postsynaptic currents (EPSCs), suggesting activation by single climbing fiber (CF) afferents (Fig. 1A) (8). These currents were mediated in part by AMPA (α-amino3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptors, because 10 to 25 μM NBQX blocked more than 95% of the current (Fig. 1A) and addition of the specific AMPA receptor antagonist GYKI 52466 (15 to 25 μM, n = 3) or theN-methyl-d-aspartate (NMDA) receptor antagonist (±)-3-(2-carboxy piperazin-4-yl)-propyl-1-phosphonic acid (CPP) (2.5 μM) caused no further inhibition. After blockade of AMPA and NMDA receptors, stimulation still activated inward currents with NO3 (Fig. 1B) (n = 25) or SCN (n = 6) but not with Cl- (Fig. 1B, inset) (n = 3) or gluconate-based (n = 3) pipette solutions. With Cl in the extracellular solution, these responses displayed current-voltage (I-V) relations with the reversal potential >+50 (Fig. 1B) indicating a high permeability to NO3 and SCN. AMPA receptor EPSCs in the same recordings reversed at −6 ± 2 mV (n = 6). Both the AMPA receptor EPSCs and the synaptic, anionic currents were all-or-none with identical stimulation thresholds, indicating that they resulted from activation of the same CF (Fig. 1C). Anionic currents were antagonized by glutamate transporter antagonists l-trans-2,4-pyrrolidine dicarboxcylic acid (PDC, 300 μM),d,l-threo-β-hydroxyaspartic acid (THA, 300 μM), and the EAAT4-preferring substrate L-α-amino-adipate (L-AA, 500 μM), but were unaffected by the EAAT2-specific antagonist dihydrokainic acid (DHK, 300 μM) (Fig. 1D) (5,9, 10). As expected from their activities as transported substrates, application of PDC, THA, and L-AA activated large inward currents (Fig. 1D). Consistent with its immunolocalization in PCs (1), these results indicate that CF stimulation activates a postsynaptic glutamate transporter current mediated at least in part by the glutamate transporter EAAT4.

Figure 1

Climbing fiber stimulation elicits anion currents associated with glutamate transport. (A) With 130 mM NO3 in the pipette and 100 μM picrotoxin in the bath, CF EPSCs are strongly diminished by 25 μM NBQX and 5 μM CPP. (Insets) Means of three successive responses before and after blockade of AMPA and NMDA receptors. V m = −25 mV. (B)I-V relation for the response in NBQX and CPP (n = 10 PCs). (Insets) Responses at differentV m's (−80 to +20 mV, ΔV = 20 mV) recorded from a PC filled with NO3 (top) and from a different PC filled with Cl (bottom). (C) The threshold for the CF EPSC was determined (mean: 50 μs) by changing stimulus duration in 10-μs increments; stimulus intensity is expressed as the change relative to the threshold intensity. At each intensity the average peak response was determined for two to four trials; the mean of failures and successes accounts for the intermediate amplitude level at 0 μs. After blocking ionotropic receptors, stimuli were delivered over the same range of intensities. Mean peak amplitudes before (○) and after (•) NBQX/CPP are plotted versus relative stimulus intensity for three PCs. (D) Transported antagonists (THA,n = 8; PDC, n = 6; and L-α-AA, n= 7) that inhibit (filled bars) the NBQX-insensitive response also elicit inward currents (open bars), whereas a glial-specific antagonist, DHK (n = 2), is without effect.

The density and localization of functional glutamate transporters in the synaptic cleft is not known, but will affect their ability to regulate the glutamate concentration transient in the cleft and could prevent glutamate from “spilling over” to neighboring synapses. If transporters are distant from release sites, then their initial activation may be delayed and the rise time of the PC synaptic transporter current (STC) may be slowed by diffusion. The initial points of rise from the baseline noise of CF EPSCs and STCs were not different (0.1 ± 0.2 ms, n = 10,P > 0.3), suggesting that some transporters are located very near release sites. However, the STC rises more slowly than the EPSC (Fig. 2A) (20 to 80% rise time: 2.63 ± 0.31 ms versus 0.37 ± 0.03 ms) and decays with two exponential components, both slower than the two components of the EPSC decay (τFAST = 28 ± 4.8 ms versus 3.8 ± 0.23 ms; τSLOW = 94 ± 16 ms versus 17 ± 0.9 ms: fraction of fast component = 76 ± 5% versus 70 ± 7% for STCs and EPSCs, respectively; n = 9).

Figure 2

Kinetics of transporter currents evoked by synaptic release and by glutamate pulses to outside-out patches. (A) Superimposition of a normalized CF EPSC (smooth trace) and an STC (noisy trace). V m = −36 mV. (B) Outside-out patch responses (below) to pulses of 2 mMl-glutamate (1.3- and 52-ms duration) before, during 300 μM THA application, and after recovery of the response (52-ms pulse only). The open-tip currents (above) illustrate the time course of solution exchange. V m = −80 mV.

To determine whether the rising and decaying kinetics of the STC were slower than those of the EPSC because of diffusional delays or slower intrinsic kinetic properties of the transporter conductance, we measured the kinetics of the transporter current activated in outside-out patches by rapid applications of l-glutamate in the presence of AMPA and NMDA receptor antagonists (11). Brief pulses (0.8 to 2 ms) of 2 mM l-glutamate gave transient responses that were blocked by THA (Fig. 2B) (300 μM,n = 7) and antagonized by PDC (300 μM: 9 ± 1% remaining, n = 3). Longer applications (>2 ms) resulted in the appearance of a sustained component (Fig. 2B). In addition, six of six patches responded to pulses of 10 mMd-aspartate, an amino acid that elicits transporter current (5, 6, 10). Rise times (20 to 80%) of the transient current were 0.75 ± 0.08 ms (n = 6) in response to brief pulses of 2 mM glutamate. The decay of patch responses elicited by 1-ms glutamate pulses (2 mM) were well fitted by a single exponential time course (τ = 7.8 ± 0.8 ms; n = 6). The slower rise and decay of the synaptic responses to glutamate may be the result of asynchronous release, dendritic filtering, diffusion to some distantly located transporters, and a slow phase of clearance at the synapse (12-14) that cannot be reproduced in the patch experiments. Together these data indicate that postsynaptic transporters are exposed to a rapid transient of glutamate after release by CF terminals.

Transmitter removal at many synapses in the central nervous system may require transporters on postsynaptic neurons as well as on surrounding glial cells (1). STCs can be used to estimate the number of glutamate molecules transported into the PC in response to CF activation. Charge movement during transport results from the sum of coupled ionic movements (1 K+ countertransported, 3 Na+ and 1 H+ cotransported with 1 glutamate) and the flux of anions that are not stoichiometrically coupled to substrate movement (5, 6, 15). If the total amount of charge flux per cycle is known, the number of glutamate molecules transported into a PC can be estimated. Simultaneous measurements of radiolabeled glutamate and charge flux were made from oocytes expressing either EAAT4 or EAAT3 and containing 30 to 35 mM intracellular NO3 (Fig.3) (16). Specific uptake in EAAT4-expressing oocytes was 35 ± 7 fmol/s and the ratio of charge to glutamate flux was 33 ± 6 elementary charges (e) per glutamate molecule (n = 7); for EAAT3-expressing cells, the values were 547 ± 120 fmol/s and 2.6 ± 0.1 e per glutamate molecule (n = 8). These results are consistent with a higher expression level or turnover rate of EAAT3 relative to EAAT4 (or both), and the report that most of the EAAT4 current is due to anion flux (5). CF elicited EPSCs (Fig.4A), and STCs (Fig. 4B) were recorded with gradients of permeant anions identical to those used for the oocyte flux measurements (17). The normalized I-V curve for STCs recorded with this pipette solution (Fig. 3, filled circles; n = 4 to 8) matches the I-V curves from oocytes expressing EAAT4 (Fig. 3A, open circles; n = 8) and EAAT3 (Fig.3B, open circles, n = 7). STCs were integrated to yield a mean charge of 1.5 ± 0.2 pC (n = 10). The charge/flux measurements obtained for EAAT3 and EAAT4 permit upper and lower estimates, respectively, of the number of glutamate molecules transported during the STC. If the STC is mediated solely by EAAT4, then 2.8 × 105 ± 0.3 × 105glutamate molecules are transported. If EAAT3 is the sole transporter type, then 3.7 × 106 ± 0.4 × 106glutamates are transported (n = 10). Although these experiments cannot determine the relative contributions of any transporter subtype precisely, the estimate of charge transfer/glutamate flux for EAAT4 places a lower limit on the number of glutamate molecules transported into the PC during CF synaptic transmission.

Figure 3

Determination of the number of charges per glutamate molecule transported by EAAT4 or EAAT3. (A) Normalized I-V plots for EAAT4-injected oocytes incubated in 96 mM NO3 . (▵) I-Vrelation in the high-[NO3 ]outincubation solution. Radiolabeled flux measurements were made with a solution containing permeant anion concentrations identical to those used in the brain slice experiments, and theI-V curve for EAAT4-expressing oocytes in this solution (○, n = 8) matches theI-V curve for the peak of the STC (•,n = 4 to 8). (Inset) Membrane current recorded from an EAAT4-injected Xenopus oocyte voltage-clamped at −55 mV in response to 25 μM [3H]l-glutamate. (B)I-V curves for EAAT3-expressing oocytes and the peak STC as in (A) (n = 7). (Inset) Membrane current in an EAAT3-injected Xenopus oocyte in response to 25 μM [3H]l-glutamate.

Figure 4

Estimation of CF quantal content and uptake by the PC. (A) CF EPSC recorded at −25 mV with 35 mM NO3 in the recording pipette. (B). The STC recorded in the same cell at −60 mV in 25 μM NBQX, 25 μM GYKI 52466, and 2.5 μM CPP ± 300 μM THA. (C) Three traces showing asynchronous release of unitary EPSCs (aEPSCs) in response to CF stimulation in Sr2+. (D) Average of 188 aEPSCs collected from 100 stimuli delivered to the CF synapse in (C). Integration yielded 242 pC.V m = −85 mV.

The number of glutamate molecules released was estimated by measuring the quantal content of the CF EPSC corresponding to each STC. Charge movement associated with a quantum of transmitter was measured by recording asynchronous quantal events (aEPSCs) after CF EPSCs evoked in a solution containing 0 CaCl2, 0.5 mM SrCl2, 3.3 mM MgCl2 (n = 3 PCs) (Fig. 4, C and D) (18). Quantal content of CF EPSCs was estimated to be 385 ± 57 (n = 10). Assuming that 4000 molecules of transmitter are contained in each vesicle (19), these data predict that postsynaptic transport by EAAT4 would be expected to remove 880 molecules per active release site or 22 ± 4% (n = 10) of glutamate released at the CF synapse.

Previous studies have shown that inhibition of transporters prolongs the decay of CF EPSCs (14) and EPSCs in other preparations (12, 20). We have found that transporters located on Purkinje neurons remove a substantial fraction of the glutamate released by CF contacts. Other pathways for glutamate removal include diffusion, transport into Bergmann glia (21), and possibly CF terminals. The transport-associated currents observed in response to CF stimulation and in patches have surprisingly rapid kinetics, given the slow turnover rates of glutamate transporters (22,23). The fast rise of these currents may result from localization of transporters near release sites allowing rapid, nearly synchronous binding of glutamate to the large fraction of the transporter population that at negative membrane potentials is already bound by Na+ (23). Similar synchronous activation of transporters has been reported at serotonergic synapses in the leech (24). The rapid activation of these currents also indicates that the anion conductance associated with glutamate transporters opens soon after glutamate binding.

Assuming the pulse of glutamate in the cleft is too short to result in multiple transport cycles, at least 880 EAAT4 transporters are required to be bound per exocytotic event. Because EAAT3, which has a much lower anion flux per transport cycle than EAAT4, is also expressed in PCs (1), the number of transporters activated by each quantal event may be even higher. Given AMPA receptor channel conductances (25) and mEPSC amplitudes from this study, ∼50 AMPA receptors are activated by a vesicle of transmitter. Thus, transporters activated by the contents of a single vesicle outnumber activated receptors by a ratio of greater than 15 to 1. Such a large population of transporters located near release sites will both speed the clearance and help restrict the spread of glutamate, first by binding and then transporting the transmitter (26).


View Abstract

Stay Connected to Science

Navigate This Article