Corresponding author D. Ogden: National Institute for Medical Research, Mill Hill, London NW7 1AA, UK. Email: email@example.com
1Tetanic stimulation of parallel fibres (PFs) produces a slow EPSP (sEPSP) or slow EPSC (sEPSC) in Purkinje neurones (PNs), mediated by type 1 metabotropic glutamate receptors (mGluR1). The conductance change underlying the sEPSP was investigated with rapid photolytic release of L-glutamate from nitroindolinyl (NI)-caged glutamate with ionotropic glutamate receptors blocked, and showed a slow mGluR1-activated cation channel.
2In cerebellar slices rapid photolytic release (t1/2 < 0.7 ms) of 7-70 μM L-glutamate on PNs voltage clamped at −65 mV activated first a transient inward current, peaking in 8 ms, followed by a slow inward current with time course similar to the PF sEPSP, peaking at −1 nA in 700 ms.
3The initial current was inhibited by 300 μM threo-hydroxyaspartate (THA) and did not reverse as the potential was made positive up to +50 mV, suggesting activation of electrogenic glutamate uptake.
4The slow current was inhibited reversibly by 1 mM (R,S)-MCPG or the non-competitive mGluR1 antagonist CPCCOEt (20 μM), indicating activation of metabotropic type 1 glutamate receptors. The mGluR current was associated with increases of input conductance and membrane current noise, and reversed close to 0 mV, indicating activation of channels permeant to Na+ and K+.
5The sEPSC was not blocked by Cd2+, Co2+, Mg2+ or Gd3+ ions, by the inhibitor of hyperpolarisation-activated current (IH) ZD7288, or by the purinoceptor inhibitor PPADS. Activation was not affected by inhibitors of phospholipase C (PLC) or protein kinase C (PKC), nor mimicked by photorelease of InsP3 or Ca2+. The results show that mGluR1 in PNs produces a slow activation of cation-permeable ion channels which is not mediated by PLC activation, Ca2+ release from stores, or via the activation of PKC.
PFs in the molecular layer of the cerebellar cortex are the principal excitatory input to PNs, driving PNs at frequencies up to 50 Hz. Single PF stimuli evoke fast postsynaptic potentials, acting through glutamate AMPA-type receptors. However, brief tetanic stimulation of PFs produces a slow EPSP seen in the presence of AMPA receptor antagonists (Batchelor & Garthwaite, 1993). The sEPSP is blocked by metabotropic glutamate receptor antagonist MCPG (Batchelor et al. 1994, 1997; Hirono et al. 1998). Application of selective mGluR type 1 agonists also produces an excitation of PNs, resulting in sustained activation of fast spikes, or, in the presence of TTx, slow spiking of dendritic origin (Tempia et al. 1998; Watkins & Ogden, 1999). Approximately 175 000 excitatory synapses are made on postsynaptic spines in the dendritic tree of each PN (Napper & Harvey, 1988). Immunogold labelling has shown mGluR1α receptors located postsynaptically on PN spines at the periphery of the PF synapse where they will be exposed to l-glutamate released from PF terminals (Baude et al. 1993).
A role of mGluR1 receptors in motor coordination is suggested by the observation that mGluR1α-deficient mice are ataxic, with lesions in the cerebellar molecular layer and developmental abnormalities in the innervation of PNs (Aiba et al. 1994; Conquet et al. 1994; Ichise et al. 2000). Also, neoplastic cerebellar ataxia, in which there is a deficit in motor coordination, has been shown to be associated with autoantibodies generated against mGluR1 (Sillevis-Smitt et al. 2000). Thus mGluR1 and the PF/PN sEPSP may have a role in motor coordination. However, the ionic mechanism of the sEPSP is not established and has been attributed to activation of Na+-Ca2+ exchange or a Ca2+-activated channel secondary to Ca2+ release from stores (Vranesic et al. 1991), although contrary evidence has also been reported (Hirono et al. 1998).
Recently we have developed a stable, fast, pharmacologically inert NI-caged glutamate, based on nitroindoline photochemistry (Papageorgiou et al. 1999). This permits the study of kinetics, mechanism and pharmacology of the postsynaptic events during the sEPSP independently of presynaptic processes. The experiments described here were made to determine the kinetics of postsynaptic events with AMPA receptors blocked and to investigate the ion conductance underlying the sEPSP. They provide evidence of a cation channel not directly linked to intracellular Ca2+. Preliminary accounts of some of this work have appeared in abstract form (Watkins & Ogden, 1999; Canepari et al. 2000).
Wistar rats, 19-22 or 12 days old, were killed by cervical dislocation, decapitated, and the cerebellum placed in ice-cold saline. Parasagittal slices, 200 μm thick, were cut in Hepes-buffered, 0.5 mm Ca2+ saline gassed with O2. External saline contained (mm): NaCl 135, KCl 4, MgSO4 or MgCl2 2, CaCl2 2, glucose 25, NaHCO3 2, Hepes-Na 10, pH 7.3, 305 mosmol kg−1. Experiments were carried out at 32 °C and a continuous stream of hydrated O2 was blown over the solution surface. NI-caged glutamate and antagonists were applied in 1 ml of solution (non-flowing) for 10 min prior to photolysis. Selective mGluR agonists were applied locally by pressure ejection from a patch pipette.
Slices were viewed with a Zeiss Axioskop 1FS, × 40 0.75w Achroplan objective and, to avoid photolysis, 500/40 nm bandpass illumination via a Reichert silica condenser 0.9 NA. A xenon arc flashlamp (Rapp OptoElektronik; Rapp & Güth, 1988) filtered with a UG11 (Schott, bandpass 290-370 nm) was focused into the slice from below, illuminating a spot of 200 μm diameter. The arc image was aligned and focused in the specimen plane with the condenser, optimised visually and by maximising the output of a photodiode. Photolysis calibration was from the fluorescence increase (470 nm excitation, > 530 nm emission) produced by photolysis of the 1-(2-nitrophenyl)ethyl ether of pyranine (NPE-HPTS; Jasuja et al. 1999) contained at 50 μm in 100 mm borate pH 9, in 10-20 μm diameter aqueous vesicles suspended in Sylgard. Conversion of NPE-HPTS is estimated in cuvette experiments as 0.7 times that of NI-caged glutamate. Transmission at 320 nm through 200 μm slices from 20-day-old rats was measured as 0.45 in the molecular layer, 0.4 in the granule cell layer. Flash lamp intensity was set to maximum, converting 7 % of NI-glutamate after correction for attenuation in the slice, and lower intensities were produced by neutral density filters in the condenser light path.
Whole cell patch clamp recordings were made with an Axoclamp-2A and 2.5 MΩ pipettes (Pyrex, 1.5 mm × 1.1 mm) were filled with internal solution (mm): potassium gluconate 110, Hepes 50, KCl 10, MgSO4 4, Na2ATP 4, creatine phosphate 10, GTP 0.05, pH 7.3 with KOH. The junction potential between this solution and external solution was measured as 12 mV, pipette negative.
Data were collected with Spike 2 software via a 1401+ interface (CED, Cambridge, UK; sampled at 10 kHz, lowpass filter 2 kHz, −3 dB). Data are given as means ±s.d. unless specified as s.e.m. Chemicals were Analar grade (BDH, Poole) and biochemicals and drugs obtained from Sigma (Poole), Tocris (Bristol) or RBI (Poole). Experiments with the Ca2+ channel blocker AGA4A (Peptide Institute, Osaka, Japan) were carried out in the presence of 0.1 mg ml−1 cytochrome c.
Photorelease of l-glutamate from NI-caged glutamate
The mGluR sEPSP was activated with rapid release of glutamate in the slice by photolysis of NI-caged glutamate, in the presence of TTx (0.5 or 1 μm), NBQX (2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulphonamide; 50 or 100 μm) to prevent AMPA receptor activation, and AP5 (dl-2-amino-5-phosphonopentanoic acid; 50 μm) and bicuculline (10 μm) to block NMDA and GABAA receptors, respectively. A membrane potential recording with 1 mm NI-caged glutamate and a high concentration of NBQX (100 μm) is shown in Fig. 1A. Photorelease of approximately 20 μm l-glutamate generated an immediate, transient 7 mV depolarisation, followed by a slower 23 mV depolarisation peaking in 650 ms and recovering in 6 s. A more intense flash, releasing approximately 70 μm l-glutamate, evoked a 12 mV transient depolarisation followed by a large slow depolarisation peaking at 28 mV in 600 ms and superimposed with delayed, TTx-insensitive regenerative spiking. With the soma voltage clamped at −65 mV (Fig. 1B) photorelease of 70 μm l-glutamate evoked an immediate inward current of −0.7 nA peaking in 5-10 ms, followed by partial recovery and a slowly rising inward current of approximately 1 nA amplitude, peaking in 1 s. This second component is associated with regenerative slow spiking blocked by Ca2+ channel blockers AGA4A or Cd2+ (data not shown). The mean values for the peak amplitude and peak times of the slow current at −65 mV in 19 Purkinje neurones in 20-day-old rats were −1.0 ± 0.5 nA and 0.7 ± 0.6 s, and the total duration was 7 ± 4 s.
Inhibitors of mGluR1 block the slow component
The slow current following l-glutamate release from NI-caged glutamate was fully and reversibly inhibited by the competitive mGluR antagonist MCPG (data not shown) and by the non-competitive mGluR1 inhibitor CPCCOEt (20 μm, Fig. 1B, middle trace, recovery lower trace; 7-(hydroxyimino)cyclopropan[b]chromen-1a-carboxylate ethyl ester; Litschig et al. 1999).
Photolysis controls and persistence of l-glutamate
NI-caged glutamate itself at 1 mm had no effect on the mGluR response. To test the effects of photolysis intermediates and by-products, control NI-caged 1 mm 5-dihydroxyphosphoryloxypentanoate (Papageorgiou et al. 1999) was photolysed under the same conditions and had no detectable action on PNs. The control compound has identical photolysis to the NI-caged glutamate except that it releases the inert dihydroxyphosphoryloxypentanoate instead of glutamate. Estimates of the time course of glutamate release from NI-glutamate by a 1 ms photolysis pulse gave t½ 0.7 ms for the rise of AMPA receptor current in cerebellar granule neurones (photochemical release rate > 106 s−1; P. Wan, J. Morrison, G. Papageorgiou & J. E. T. Corrie, unpublished data). l-Glutamate, released in the slice at a concentration of 70 μm in the 200 μm diameter release region, declined in concentration approximately exponentially with a t½ of 212 ms, indicating that glutamate released at a high concentration in the slice is removed rapidly.
Glutamate uptake inhibitors modify the fast component
The fast component was not blocked with 1 mm NBQX present and may result from immediate activation of electrogenic glutamate uptake. This was tested by applying threo-hydroxyaspartate (THA; 100-500 μm) which competes with l-glutamate for transport. In control the initial current rose to peak in 8.3 ± 3.7 ms (n= 8; Fig. 2A) with a mean amplitude of −488 ± 283 pA. In 300 μm THA, an initial fast rate of rise, similar to control but with an amplitude 5-40 % of the total, was followed by a slower rise to peak in 29 ± 16 ms, and a mean amplitude of 62 ± 22 % of control. The time course in the presence of MCPG, with and without THA, is shown on fast and slow time scales in Fig. 2A. The slow decline is similar in time course to the decline of l-glutamate concentration in the slice (see above); however, other processes cannot be excluded. Thus, THA modifies the time course and amplitude of the initial current, consistent with competition with photoreleased glutamate for binding to the electrogenic glutamate transporter. THA concentrations higher than 300-500 μm produced slow excitatory effects in PNs attributable to metabotropic receptor activation.
The transporter currents activated by photorelease of different glutamate concentrations were examined in the presence of a high concentration of NBQX and MCPG. Figure 2B shows the current evoked by pulses of 7-70 μm l-glutamate. Peak glutamate transporter current increased approximately linearly with increasing concentration in this range.
Permeability changes during the slow EPSP
The membrane permeability changes underlying the sEPSP were investigated by measuring the conductance and reversal potential of the sEPSC. The input resistance, measured at the end of 50 ms, 15 mV hyperpolarising potential steps in a train from −65 mV, decreased during the glutamate-evoked sEPSC from 50 to 30 MΩ (Fig. 3A). Input conductance in seven PNs increased by 32 ± 16 % showing that the net excitatory current is associated with a conductance increase, approximately doubling at the peak inward current after allowing for series resistance of 15 MΩ.
To investigate the ionic basis of membrane permeability changes, dendritic cable losses were minimised by making reversal potential measurements in 12-day-old rats (Llano et al. 1991), in which the dendritic tree has not extended to the external granule cell layer. A large, slow, mGluR-mediated current was seen in voltage clamp in response to photolytic l-glutamate release in 12 day as well as 20 day animals. The glutamate-evoked current was estimated at holding potentials between −70 and +50 mV after the resting currents had stabilised at each potential. Experiments were done with potassium gluconate internal solution (see Methods) with external NBQX (50 μm), TTx (1 μm) and CdCl2 (200 μm) to prevent regenerative dendritic potentials, and 5 mm BaCl2 to minimise outward holding current at positive potentials. The currents activated by release of 70 μm l-glutamate at different holding potentials in a PN are shown in Fig. 3B on two time scales. The initial current decreased in amplitude at potentials up to +50 mV and showed inward rectification and no reversal in this range, consistent with rapid activation of the electrogenic sodium-glutamate co-transporter. The subsequent slow current was reversed at +50 mV and reversal appeared close to +20 mV. The mean glutamate-evoked slow current for five PNs, normalised to that at −70 mV, is plotted against potential in Fig. 3C and shows reversal at a mean clamp potential of +20 mV, determined by regression. Making allowance for a pipette junction potential error of +12 mV corrects the reversal potential to +32 mV. For a mean net outward holding current of 0.7 nA at +20 mV, an error will result of > 10 mV (pipette positive) due to series resistance (≈15 MΩ) and cable attenuation, producing a further correction back towards more negative potentials. Thus, the true reversal potential for the mGluR conductance is close to 0 mV in these ionic conditions, suggesting a cation conductance with mainly Na+ and K+ ions carrying current.
Effect of membrane potential on mGluR1 activation
As seen in Fig. 3B, the time to peak activation of the mGluR-evoked conductance was reduced at positive potentials. In the experiments from 12 day animals the mean time to peak at −70 mV was 778 ± 190 ms (n= 4), and was shortened at more positive potentials by factors of 0.49 ± 0.27 at −30 mV, 0.48 ± 0.41 at +20 mV and 0.09 ± 0.06 at +50 mV.
Increase of current noise during mGluR1 responses
An increase of current noise above baseline was evident at the peak of the slow current evoked by photolytic release of l-glutamate or by selective mGluR agonists. This is illustrated in Fig. 4A and B, which shows traces recorded at DC to 1 kHz bandwidth (upper) and bandpass filtered at 1-200 Hz (-3 dB, lower) at high gain. Release of 70 μm l-glutamate, in the presence of NBQX, AP5, bicuculline, TTx and AGA4A toxin to block ionotropic receptors and regenerative potentials, showed an increase of current variance at the peak of inward current in a PN voltage clamped at −70 mV (Fig. 4A). The maximum variance recorded occurred at the peak current and declined as the mean current declined to baseline, indicating that the open probability was < 0.5 at the peak. The variance increase above baseline averaged 10.4 (pA)2 (s.e.m. ± 3.3; n= 7) and, assuming a low open probability, corresponds to a mean unitary current of 7 fA in the seven cells analysed. This is likely to underestimate single channel current because of cable losses in the dendritic tree and series resistance, and cannot be used to calculate the single channel conductance because poor voltage control in the dendrites will substantially reduce the driving potential. Local application of the selective mGluR agonists (R,S)-ACPD and (S)-DHPG (selective at group 1 mGluR; Schoepp et al. 1999) also produced an increase in current variance at the peak inward current (Fig. 4B), in the presence of TTx at −90 mV holding potential to minimise contamination with regenerative slow potentials.
Effect of channel blockers
Preliminary experiments to identify the ion channel showed that the mGluR-evoked conductance was not affected by divalent cations Mg2+ (5 mm), Ba2+ (5 mm), Co2+ (2 mm), Cd2+ (200 μm) nor Gd3+ (100 μm). The cyclic nucleotide channel antagonist ZD7288 at 20 μm was found to block IH in PNs after 5 min exposure but did not reduce the amplitude of the mGluR-evoked current after 20 min. The purinoceptor antagonist PPADS (pyridoxal phosphate-6-azophenyl-2′,4′-disulphonic acid) produced no block at 100 μm during 15-55 min exposure.
Effect of phosphoinositide pathway blockers
It has previously been shown that photorelease of InsP3 in PNs can evoke fast, large amplitude Ca2+ release from stores and this activates an inhibitory outward current due to Ca2+-activated K+ channels. Similarly, photorelease of Ca2+ from DM-nitrophen also activates an outward current (Khodakhah & Ogden, 1993, 1995; Watkins & Ogden, 1999; present study). The coupling between mGluR1 and the conductance increase was investigated pharmacologically. Purkinje neurones contain the PLC isoform β4 (Sugiyama et al. 1999) and this isoform is blocked by the inhibitor U-73122 (Cruzblanca et al. 1998; Haley et al. 2000). Neither U-73122 (2.5 or 5 μm, n= 6, 10-40 min exposure) nor the PKC inhibitors chelerythrine (20 μm, n= 4, 10-30 min exposure; Hansel & Linden, 2000) or bis-indolyl maleimide (0.5 μm, n= 2, 40 min exposure; Smith et al. 1998) applied in the pipette prevented the activation of the slow conductance, indicating that the PLC and PKC pathways are not involved.
The excitatory slow EPSP produced by brief tetanic stimulation of PFs was described by Batchelor & Garthwaite (1993) and shown to be mediated by mGluR1 receptors (Batchelor et al. 1994; Hirono et al. 1998). Here a response mediated by mGluR1 was elicited by photorelease of l-glutamate in the presence of NBQX. It was blocked by the antagonists MCPG and CPCCOEt. Furthermore, selective mGluR group 1 agonists also produced an excitation similar to the sEPSP. This excitation is shown to be due to the mGluR1-mediated opening of Na+- and K+-permeable cation channels through a pathway not requiring the activation of PLC. In view of the evidence of a role of mGluR1 in cerebellar function it is likely that the mGluR1-mediated conductance investigated here with photolytic l-glutamate release has a role in cerebellar motor coordination.
The rapid photolytic release of glutamate from NI-caged glutamate permits the kinetic distinction of fast and slowly developing postsynaptic currents. With AMPA receptors blocked two components were distinguished, a transient current peaking in 7-8 ms with pharmacological and electrophysiological properties characteristic of an electrogenic glutamate transporter, followed by a slow rising current, peaking in 0.7 s with the pharmacology of type 1 mGluR. Photolytically released glutamate persists in the slice with a half-time of 200 ms, implying that the mGluR1-activated current rises after the glutamate concentration has fallen substantially. This is consistent with the observation that the sEPSP rises with a delay (Batchelor & Garthwaite, 1993; present study, data not shown) at a time when the synaptic glutamate has declined following PF stimulation (Barbour et al. 1994). This delay indicates the involvement of intracellular messengers or slow, membrane-associated coupling steps.
The initial transient component was not affected by the AMPA receptor antagonist NBQX applied at concentrations up to 1 mm. The competitive glutamate transport inhibitor THA, at 100-300 μm, modified the time course as might be expected if the number of transporters initially available is reduced by bound THA and the glutamate persisting after photorelease competes with THA for transport. The inward rectification and failure to reverse the initial current at holding potentials as positive as 50 mV are consistent with concerted coupled transport of extracellular glutamate and Na+ into the cell immediately on glutamate release. A fast-rising, transient transporter current has been seen with climbing fibre stimulation in the presence of NBQX (Auger & Attwell, 2000) and with PF stimulation. Integrating the current in response to 70 μm l-glutamate with NBQX, bicuculline and MCPG present gave an estimate for the charge carried of 1.6 pC, corresponding to 16 × 10−18 mol univalent ions. If this is due to one cycle of transport (2 charges are translocated in each cycle; Takahashi et al. 1996; Auger & Attwell, 2000) it corresponds to 5 × 106 sites. This number is likely to be an underestimate of the number of sites because of the lower concentration of glutamate released here than synaptically (70 μmvs. 1 mm) and because of cable losses. However, the magnitude is not large when compared to the number of parallel fibre synapses, approximately 175 000, estimated anatomically to be present on a rat PN (Napper & Harvey, 1988).
The current underlying the sEPSP has previously been attributed to a Ca2+-activated non-selective cation conductance or to Na+-Ca2+ exchange, based on the effects of Na+ replacement with Li+ and block by high internal BAPTA concentrations (Vranesic et al. 1991). Recently mGluR activation has been shown to result in a dendritic increase of [Na+] (Knopfel et al. 2000) and in another study, inhibitors of Na+-Ca2+ exchange have been shown to be ineffective against the sEPSP (Hirono et al. 1998). Moreover, the effect of increasing intracellular Ca2+ photolytically, either with caged InsP3 or Ca2+-DM-nitrophen, is to activate an outward, inhibitory current (Khodakhah & Ogden, 1993, 1995; Watkins & Ogden, 1999) and not the excitatory slow current seen with mGluR receptor activation. This indicates that the sEPSP is not secondary to a rise of intracellular Ca2+ concentration, but is activated via another pathway. The results presented here show a conductance increase with reversal potential near 0 mV, indicating permeability to Na+, K+ and possibly Ca2+ but not Cl− (ECl=−85 mV). There was an increase of membrane current noise attributable to the stochastic opening of channels, with maximum variance at the peak of the mGluR1-activated inward current. Preliminary experiments to investigate the nature of the channels activated show no block by Cd2+, Co2+, Mg2+, Mn2+, Gd3+ or Ba2+, the inhibitor of the cyclic-nucleotide-gated conductance IH, ZD7288, or by the purinoceptor antagonist PPADS. The rate of activation of the conductance increased at positive potential, about 8-fold in 120 mV, indicating that membrane-delimited interactions are involved.
Thus, we have shown that metabotropic type 1 glutamate receptors slowly activate a cation conductance in PNs that underlies the slow EPSP at PF-PN synapses. The coupling between mGluR1 at parallel fibre synapses and the excitatory conductance underlying the sEPSP shows characteristics of indirect activation, rising after photolytic or synaptically generated glutamate concentrations have declined, and evidence presented here and in other studies (Hirono et al. 1998) indicates that the coupling is not via the phosphoinositide pathway. A non-selective cation conductance activated by mGluR1 independently of phosphoinositide and G-protein coupling has been demonstrated in hippocampal CA3 neurones (Heuss et al. 1999). In this case the GABAB receptors served as an internal control for G-protein inactivation, and activation of the conductance was via protein tyrosine kinases.
The most striking consequence of mGluR activation in these experiments was a large dendritic excitation, resulting in regenerative Ca2+ spikes within the dendritic tree. Further experiments need to be done to identify the ion channel, its potential role, directly or indirectly, in Ca2+ influx, and the coupling mechanism to mGluR1 receptors.
This work was supported by the MRC and an EU Marie Curie Fellowship (M.C.).