Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Auto-inhibition of rat parallel fibre–Purkinje cell synapses by activity-dependent adenosine release

Auto-inhibition of rat parallel fibre–Purkinje cell synapses by activity-dependent adenosine release J Physiol 581.2 (2007) pp 553–565 553 Auto-inhibition of rat parallel fibre–Purkinje cell synapses by activity-dependent adenosine release Mark J. Wall and Nicholas Dale Neuroscience Group, Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL, UK Adenosine is an important signalling molecule involved in a large number of physiological functions. In the brain these processes are as diverse as sleep, memory, locomotion and neuro- protection during episodes of ischaemia and hypoxia. Although the actions of adenosine, through cell surface G-protein-coupled receptors, are well characterized, in many cases the sources of adenosine and mechanisms of release have not been defined. Here we demonstrate the activity-dependent release of adenosine in the cerebellum using a combination of electro- physiology and biosensors. Short trains of electrical stimuli delivered to the molecular layer in 2+ vitro, release adenosine via a process that is both TTX and Ca sensitive. As ATP release cannot be detected, adenosine must either be released directly or rapidly produced by highly localized and efficient extracellular ATP breakdown. Since adenosine release can be modulated by receptors that act on parallel fibre–Purkinje cell synapses, we suggest that the parallel fibres release adenosine. This activity-dependent adenosine release exerts feedback inhibition of parallel fibre–Purkinje cell transmission. Spike-mediated adenosine release from parallel fibres will thus powerfully regulate cerebellar circuit output. (Received 12 December 2006; accepted after revision 1 March 2007; first published online 8 March 2007) Corresponding author M. J. Wall: Neuroscience Group, Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL, UK. Email: mark.wall@warwick.ac.uk Adenosine is an important neuromodulator in the central The presence of adenosine, adenosine deaminase and nervous system, playing a role in a plethora of physio- A receptors in the cerebellar cortex (Braas et al. 1986; logical and pathophysiological processes. The action of Geiger & Nagy, 1986; Rivkees et al. 1995) strongly suggests adenosine on cell surface receptors is well defined with that adenosine plays an important role in cerebellar A ,A ,A and A receptors all cloned (Fredholm et al. function. The activation of A receptors inhibits synaptic 1 2a 2b 3 1 2000). Although extensively studied, the cellular source transmission between parallel fibres and Purkinje cells and mechanisms of adenosine release remain unclear (Kocsis et al. 1984). These receptors are tonically activated (for review see Latini & Pedata, 2001). Adenosine can by endogenous adenosine, since application of A receptor in principle gain access to the extracellular space by antagonists enhances synaptic transmission (Takahashi the breakdown of ATP, by translocation from cell cyto- et al. 1995; Dittman & Regehr, 1996). The source of this plasm via nucleoside transport proteins or possibly by the adenosine has not been determined but could arise from exocytosis of adenosine itself. There has been considerable the release of adenosine or the release of ATP and its investigation of adenosine release during pathological subsequent metabolism. A recent report has suggested episodes such as hypoxia, ischaemia and hypercapnia that ATP can be released from parallel fibres (Beierlein as adenosine is neuroprotective (Rudolphi et al. 1992; & Regehr, 2006). Fredholm, 1997; Dale et al. 2000; Dulla et al. 2005). To examine this issue we have used selective and sensitive 2+ Release under these conditions is often Ca independent, microelectrode biosensors (Llaudet et al. 2003) to measure relatively insensitive to TTX and is not mediated via the release of adenosine from cerebellar slices in real time. glutamate receptor activation. In contrast, little is known These biosensors are small enough (25–50 μm diameter) about the physiological release of adenosine with few to place either in or close to defined areas in cerebellar examples where a role and cellular source of adenosine slices. Here we report that adenosine can be released from have been identified (but see Dale, 1998). In many cases, the molecular layer by using a physiological stimulus, short adenosine release is evoked with stimuli such as high K , bursts (1–10 s) of focal electrical stimuli at the same voltage prolonged electrical stimulation and glutamate receptor used to elicit synaptic transmission. The adenosine release 2+ activation (Latini & Pedata, 2001). The physiological is both TTX and Ca sensitive and does not appear to arise relevance of these experiments is unclear. from the extracellular metabolism of ATP. Modulation C  C 2007 The Authors. Journal compilation 2007 The Physiological Society DOI: 10.1113/jphysiol.2006.126417 554 M. J. Wall and N. Dale J Physiol 581.2 of parallel fibre–Purkinje cell synaptic transmission can potentials in Purkinje cells and interneurones (Clark increase or decrease adenosine release, strongly suggesting & Barbour, 1997). Parallel fibre EPSP amplitude was that parallel fibres are involved in adenosine release. estimated from the CNQX/kynurenate-sensitive potential, which was measured by subtracting what remained in CNQX/kynurenate from control potentials. Confirmation Methods of PF EPSP identity was achieved by evoking pairs of EPSPs Slice preparation (interval 50 ms) and observing facilitation (20–30%) and by examining the pharmacological profile (inhibition by Transverse slices of cerebellum (400 μm) were prepared , GABA and mGlu4R receptor agonists). Sensor signals 1 B from male Wistar rats, at postnatal days 21–28 (P21–28), were acquired at 1 kHz with either a Digidata 1322A with modified methods based on Llinas & Sugimori (Axon) or a MiniDigi (Axon) using pCLAMP 9.2 (Axon) (1980). As previously described (Wall & Usowicz, 1997) or Axoscope 9.2 (Axon). Extracellular recordings were and in accordance with the UK Animals (Scientific made using an ISO-DAM extracellular amplifier (WPI, Procedures) Act 1986, male rats were killed by cervical Stevenage, UK), filtered at 1 kHz and digitized on line dislocation and decapitated. The cerebellum was rapidly (10 kHz) with a Digidata 1322A controlled by pCLAMP removed and transverse slices were cut on a Microm 9.2. HM 650V microslicer (Carl Zeiss, Welwyn Garden City, ◦ 2+ 2+ UK) in cold (2–4 C) high Mg ,low Ca aCSF, composed Biosensor characteristics of (mm): 127 NaCl, 1.9 KCl, 7 MgCl , 0.5 CaCl , 1.2 2 2 KH PO , 26 NaHCO ,10 d-glucose (pH 7.4 when bubbled 2 4 3 Biosensors were obtained from Sarissa Biomedical Ltd with 95% O and 5% CO ). Slices were stored in normal 2 2 (Coventry, UK). In brief the adenosine biosensor consisted aCSF (1.3 mm MgCl , 2.4 mm CaCl )atroomtemperature 2 2 of three entrapped enzymes (adenosine deaminase, for 1–6 h before recording. nucleoside phosphorylase and xanthine oxidase) within a matrix that was deposited around a Pt or Pt/Ir (90/10) Recording from slices wire etched to 25–50 μm (Llaudet et al. 2003). The biosensor had an exposed length of 500 μm that was An individual slice was transferred to a recording coated with enzymes and thus capable of detecting purines. −1 chamber, submerged in aCSF and perfused at 6 ml min Biosensors had an additional screening layer, which greatly (30–35 C). The slice was placed upon a suspended reduced the responses to non-specific electro active inter- grid to allow perfusion of the slice from above and ferents (such as 5-HT, dopamine, noradrenaline and below and thus reduce the likelihood of hypoxia. All ascorbate). Screened null sensors, possessing the matrix solutions were vigorously bubbled (95% O and 5% but no enzymes, were used to control for the release CO ) and all tubing had low gas permeability (Tygon; of any non-specific electro active interferents. The ATP Fisher Scientific, Loughborough, UK). For the stimulation biosensor consisted of the entrapped enzymes glycerol of purine release and parallel fibre–Purkinje cell (PF) kinase and glycerol-3-phosphate oxidase (Llaudet et al. EPSPs, square voltage pulses (2–8 V, 200 μs duration) 2005). Glycerol (2 mm) was included in solutions as were delivered by an isolated pulse stimulator (Model glycerol is a co-substrate required for ATP detection. 2100 AM systems; Olympic Peninsula, Washington, DC, Biosensors were calibrated with known concentrations USA) via a concentric bipolar metal stimulating electrode of adenosine and ATP (10 μm typically giving responses (FHC) placed on the surface of the molecular layer. of 2–3 nA for adenosine and 1.5 nA for ATP). Calibration Purine biosensors were either positioned just above the was performed before the slice was present in the perfusion surface of the slice (bent so their longitudinal surface was chamber and after the experiment (following slice parallel to the stimulated molecular layer) or carefully removal), this allowing quantification of any run-down in inserted (at an angle of ∼70 deg) into the stimulated sensitivity during the experiment. Previous reports have molecular layer. For the extracellular recording of PF detailed the properties of adenosine and ATP biosensors EPSPs, an electrode (aCSF-filled microelectrode) was (Llaudet et al. 2003, 2005): selectivity, linear responses to placed on the same track along which the parallel fibres increasing analyte concentration and rapid response. travel (for example see Yuan & Atchison, 1999). A typical extracellular field potential consisted of an initial Determination of ATP breakdown inhibition component which persisted in either 10 μm CNQX or 5mm kynurenate but was blocked by 1 μm TTX (parallel To determine which compounds were the most effective at fibre volley), followed by a component which could be blocking the extracellular metabolism of ATP, the break- blockedby1 μm TTX and greatly reduced by either 10 μm down of etheno-ATP (ε-ATP) was measured using HPLC. CNQX or 5 mm kynurenate. This component is probably Individual 400 μm cerebellar slices were incubated (on a produced by parallel fibre-mediated glutamatergic shaker, at room temperature) in 400 μl of aCSF containing excitatory synaptic currents and subsequent action either 5 or 50 μm ε-ATP (2 or 20 nmol of ε-ATP) with or C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 555 without inhibitors. At 0, 10, 25 and 45 min, 50 μl samples Fig. 1A). The current waveform was not determined by were taken and snap frozen with dry ice. sensor-response kinetics, as it was much slower than Samples were thawed out and diluted (1 in 10) in the sensor response to purine application (Llaudet et al. distilled water. HPLC analysis was performed with a Luna 2003). Since the adenosine biosensors can detect inosine C8 (2) reverse phase column and a Thermo Separation and hypoxanthine as well as adenosine, EHNA, a specific Products HPLC gradient pump (P2000) and fluorescence inhibitor of adenosine deaminase (the first enzyme in the detector (FL3000). The mobile phase consisted of 20 mm biosensor enzyme cascade), was used to determine the potassium phosphate pH 6.0 (solution A) and 75% purine detected (Agarwal et al. 1977; Safiulina et al. 2005). potassium phosphate pH 6.0 and 25% methanol (solution EHNA (20 μm) greatly reduced (85 ± 5%, n = 3) the B). A concave gradient was run going from 100% response of adenosine biosensors to exogenous adenosine solution A to 100% solution B in 10 min. ε-ATP typically (10 μm) but had no effect on the detection of inosine eluted around 3.5 min, ε-ADP around 4 min, ε-AMP (and by inference hypoxanthine, Fig. 1B). In five slices, 5 min and ε-adenosine around 9 min. The column was application of 20 μm EHNA greatly reduced the biosensor re-equilibrated with solution A for 10 min between runs. current following stimulation (mean inhibition 75 ± 3%, To quantify ε-ATP breakdown, the relative proportions of Fig. 1C ). Thus, most of the biosensor current results from the breakdown products was obtained from peak areas. adenosine detection. The remaining current is probably a combination of adenosine (as EHNA only partial blocks the adenosine sensor response) and inosine either directly Drugs released or as a consequence of adenosine breakdown (due All drugs were made up as 10–100 mm stock solutions, to incomplete inhibition of adenosine deaminase in the stored frozen and then thawed and diluted with aCSF on slice). the day of use. R(+)-baclofen hydrochloride, muscimol The amount of adenosine released was dependent on Evans Blue, α,β -methylene-ADP, adenosine, inosine, 6-[ both the frequency and duration of stimulation. With (4-nitrobenzyl)thiol]-9-β -d-ribofuranosylpurine (NBTI) 100 stimuli, no adenosine release could be detected 8-cyclopentyltheophylline (8CPT) and dipyridamole were purchased from Sigma. Erythro-9-(2-hydroxy- 3-nonyl) adenine (EHNA), 6-cyano-7-nitroquinoxaline- 2,3-dione (CNQX), ARL67165, l-AP4, d(–)-2- amino-5-phosphonopentanoic acid (AP5) were purchased from Tocris-Cookson. ATP was purchased from Roche and etheno-ATP (ε-ATP) was purchased as a 5mm solution from Invitrogen. Results Trains of electrical stimuli release adenosine in the molecular layer To investigate the release of purines in the cerebellum, trains (1–10 s, 2–20 Hz) of electrical stimuli were applied to areas of the cerebellum and purine release measured in real time using specific and selective microelectrode Figure 1. Electrical stimulation in the molecular layer releases biosensors (Llaudet et al. 2003). Stimuli applied to adenosine the molecular layer of transverse slices, consistently A, superimposed current traces from adenosine (Ado) and null produced a slow rising current on adenosine biosensors, biosensors following electrical stimulation of the molecular layer (5 V for 5 s at arrow). The lack of signal on the null sensor indicates that positioned on the surface of the molecular layer (Fig. 1A). the current on the adenosine biosensor is due to purine detection and Movement of the stimulator and sensor to the adjacent is not non-specific. B, application of 10 μM adenosine (Ado) and then granule cell layer caused a loss of current, suggesting the 10 μM inosine (Ino) produced current responses on an adenosine detected analyte was released in the molecular layer. The biosensor. Addition of 20 μM EHNA (to block adenosine deaminase) sensor-current (amplitude 20–200 pA) had a 10–90% rise almost abolished the response to adenosine but had little effect on the response to inosine. C, superimposed traces from an adenosine time of 14.4 ± 2.2 s and a 10–90% decay of 131.6 ± 14.2 s biosensor in control and in the presence of 20 μM EHNA (to block (n = 30). The current was due to purine detection, as no adenosine deaminase) following electrical stimulation of the molecular current was produced on a null sensor (identical to an layer (8 V for 8 s at arrow). EHNA reduced the current by ∼75% adenosine sensor but lacking the purine detection enzyme demonstrating that most of the signal is due to adenosine detection. All stimulations were at a frequency of 20 Hz. cascade) placed close to the adenosine biosensor (n = 8, C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 556 M. J. Wall and N. Dale J Physiol 581.2 at a frequency of 2 Hz but was detected at 5 Hz. release (n = 11, Fig. 3A) demonstrating that adenosine Maximal adenosine release occurred around 20 Hz with release is action potential dependent. In seven slices, little increase at higher frequencies (40–80 Hz, Fig. 2A). adenosine release was also abolished by substitution of 2+ 2+ Increasing the stimulus train length released more normal aCSF for Ca -free aCSF (Mg increased to 2+ adenosine (Fig. 2B). At 20 Hz, the shortest train to release 3.7 mm, Fig. 3B). To measure the Ca dependence of detectable adenosine was ∼1 s (20 stimuli) but in most slices 5–8 s (100–160 stimuli) were required. With an interval between trains of stimuli of 5 min, adenosine release was reliable and did not decrement. However, with shorter intervals the amount of adenosine released diminished. If it is assumed that all the sensor-current results from adenosine detection, then it is possible to calculate the adenosine concentration at the sensor. In 15 slices, an 8 s (20 Hz) stimulation produced concentrations ranging from 130 to 1650 nm (mean 492 ± 89 nm). This range of concentrations may stem from different distances between the adenosine release sites and the biosensor. Mechanisms of adenosine release 2+ We next examined the TTX-, Ca - and receptor- sensitivity of adenosine release. Application of the sodium channel blocker TTX (0.5–1 μm) abolished adenosine Figure 3. Mechanisms of adenosine release A, superimposed current traces from an adenosine biosensor placed on the surface of the molecular layer in control, in 10 μM CNQX + 50 μM AP5 and in 0.5 μM TTX. The adenosine release following electrical stimulation (5 V, 8 s at arrow) was abolished by TTX but was insensitive to the block of glutamate receptors. B, superimposed traces from an adenosine biosensor placed on the surface of the molecular layer in 2+ 2+ control, Ca -free aCSF (3.7 mM Mg ) and following reintroduction 2+ of Ca (wash). The adenosine release following electrical stimulation 2+ (7 V, 8 s at arrow) was abolished by removal of Ca . Inset, graph 2+ plotting normalized adenosine release against external Ca concentration. Adenosine release was normalized to what occurred at 2+ 2.4 mM Ca and summarizes data from 6 slices. C, superimposed traces from an adenosine biosensor placed on the surface of the Figure 2. Properties of adenosine release molecular layer in control and in the presence of the GABA receptor A, graph summarizing how adenosine release varies with stimulation agonist muscimol (30 μM). Adenosine was released by a 5 V, 7 s frequency (data from 4 slices). The number of stimuli was kept stimulus at the arrow. Inset, superimposed averages of EPSPs in constant (100) and the stimulation frequency was varied between 2 control and in 30 μM muscimol ( ). Muscimol reduced EPSP amplitude and 80 Hz. The data are normalized to the concentration of adenosine but had little effect on the volley. D, superimposed traces from an detected at a stimulus frequency of 20 Hz. B, superimposed traces adenosine biosensor placed on the surface of the molecular layer in from an adenosine biosensor following 20, 100 and 160 stimuli. The control and in the presence of 5 μM NBTI and 10 μM dipyridamole to stimulus frequency was kept constant at 20 Hz. The stimulation block equilibrative transport. Adenosine was released by a 5 V, 7 s strength was between 3 and 8 V (A) and 5 V (B). stimulus at the arrow. All stimuli were at a frequency of 20 Hz. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 557 2+ it seems very unlikely that adenosine is released as a adenosine release, extracellular Ca concentration was 2+ result of the downstream actions of another transmitter. changed between 0.5 and 5 mm with Mg kept at Furthermore it is unlikely to come from Purkinje cells or 1.3 mm. Adenosine release was examined in the presence 2+ interneurones as they will be shunted by GABA receptor of 1 μm 8CPT to block A receptors (see later). The Ca A activation. dependence was approximately linear (1.1 ± 0.1, n = 6, Fig. 3B inset) which is lower than that measured for Direct release of adenosine could occur via the classical neurotransmitters (Dodge & Rahamimoff, 1967; equilibrative nucleoside transporters ENT1 and ENT2 Mintz et al. 1995; Reid et al. 1998). translocating adenosine from the cytoplasm to the extracellular space. However, blockade of these molecules It is possible that the action potential-dependent 2+ 2+ by a combination of 5 μm NBTI and 10 μm dipyridamole opening of Ca channels and subsequent entry of Ca (for review see Noji et al. 2004) had no significant directly causes the release of adenosine. However, it is also (P = 0.71) effect on the amount of adenosine released possible that the stimulation indirectly evokes adenosine (n = 5, Fig. 3D, 41.6 ± 7.8 versus 38.3 ± 10.7 pA, 1 μm release through the downstream actions of another neuro- 8CPT was present to prevent A receptor activation). transmitter. We have taken two approaches to address this Adenosine was not released as a result of electroporation possibility. Firstly, we have tested the most likely neuro- and consequent release of cytoplasmic ATP (for example, transmitter candidates that could perform this role (mean see Hamann & Attwell, 1996) (with subsequent conversion current in control versus mean current with antagonist to adenosine by ecto-ATPases) since electroporation is present). Adenosine release was not sensitive to 10 μm insensitive to TTX and only occurred at much higher CNQX and 50 μm AP5 (AMPA and NMDA glutamate stimulus strengths (see later and Fig. 4C ). receptor antagonists, 53.5 ± 10 versus 51.8 ± 11 pA, n = 5, Fig. 3A), 100 μm CPCCOET or 300 μm AIDA (Group 1 mGluR antagonists, 44.2 ± 13 versus 52 ± 18.5 pA, n = 3), 10 μm PPADs (P2 receptor antagonist, 34.6 ± 5 versus ATP release cannot be detected 32.8 ± 4.3 pA, n = 3), 300 μml-NMMA (NO synthase An important potential source of adenosine is the inhibitor, n = 4, 25.3 ± 6.3 versus 21.6 ± 5 pA) and extracellular metabolism of ATP. To investigate whether prazosin (α adrenoceptor antagonist, 103 ± 13 versus adenosine is released in the form of ATP, a number of 99.6 ± 14 pA, n = 3). Furthermore application of 10 μm 5-HT or 10 μm noradrenaline did not induce adenosine approaches were taken. Firstly, ATP biosensors were used release (n = 3). Secondly, we have applied the GABA to test directly for ATP release (Llaudet et al. 2005). These receptor agonist muscimol (30 μm) and measured biosensors have a detection limit of ∼60 nm (producing adenosine release. The rationale for this approach is that ∼10 pA of current). In eight slices an ATP biosensor the activation of GABA receptors will hyperpolarize and was positioned close to the adenosine biosensor on the reduce the input resistance of neurones. The actions of surface of the molecular layer. Following stimulation, muscimol on glia are less certain, but Bergmann glia adenosine was detected but no signal was observed on express GABA receptors on their processes and therefore the ATP biosensor. In three further slices, after measuring may be inhibited (Riquelme et al. 2002). Thus, if adenosine adenosine release, the adenosine biosensor was carefully is released as a consequence of the downstream actions of removed and replaced with an ATP sensor (thus potential another transmitter, inhibition of the target cells should ATP release was measured in exactly the same place greatly reduce or abolish adenosine release. Furthermore, where adenosine was detected). Again there was no signal if GABA receptors are not present on parallel fibres following stimulation. It is possible that ATP is released and parallel fibres are the source of adenosine then deep within the slice and is then metabolized to adenosine muscimol should have no effect on adenosine release. before it reaches the sensor on the surface of the slice. The effectiveness of muscimol (30 μm) was confirmed by To investigate this, ATP biosensors were inserted into two observations: (1) the prolonged loss of spontaneous the molecular layer, in the region where adenosine was detected. Again no signal was observed on the ATP action potential firing in Purkinje cells following muscimol sensor following stimulation, although adenosine could application (n = 3) and (2) muscimol (30 μm) reduced still be detected on the slice surface (n = 8, Fig. 4A). To the amplitude of parallel fibre EPSPs by 56 ± 6.2% (n = 6) reduce the possible extracellular ATP breakdown we have with little effect upon the parallel fibre volley (8.1 ± 7.7% made use of three agents: Evans Blue and ARL67156 reduction, Fig. 3C inset, n = 6). The lack of effect of muscimol on the parallel fibre volley suggests few GABA (which inhibit ecto-ATPases, Crack et al. 1995; Bultmann receptors are expressed by parallel fibres and thus the et al. 1999) and α,β -methylene-ADP (which inhibits reduction in EPSP amplitude results from a reduction in ecto-5 -nucleotidase, for review see Zimmerman, 1996). In the resistance of Purkinje cells. Since muscimol (30 μm) the presence of Evans Blue (100 μm, n = 5) or ARL67156 had no significant effect on the magnitude of adenosine (100 μm, n = 3) no signal was detected on the inserted release (55.3 ± 8.5 versus 58.1 ± 8.1 pA, Fig. 3C , n = 5) ATP sensor and the adenosine signal was not significantly C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 558 M. J. Wall and N. Dale J Physiol 581.2 different from the control (P = 0.53, Fig. 4B). There was was converted to adenosine (n = 4). If adenosine were to also no significant difference between the amount of arise from ATP breakdown, this suggests that to give the adenosine detected in control, following application of observed signal of ∼0.5 μm adenosine, around 2.5–5 μm 100 μm α,β -methylene-ADP (47 ± 19 versus 54 ± 17 pA, ATP would have to be released. This quantity of ATP P = 0.06) and after wash (P = 0.22). would give an easily measurable signal on the ATP sensor. We have exploited the electroporation-induced release Therefore it seems unlikely that adenosine arises from the of ATP (Hamann & Attwell, 1996) to measure the kinetics extracellular breakdown of ATP. of ATP breakdown. An ATP biosensor was inserted into the molecular layer and an adenosine sensor was placed very Inhibitors slow ATP metabolism close to the ATP sensor on the surface of the molecular layer. TTX (0.5 μm) was present to prevent TTX-sensitive The inability to either directly detect ATP or reduce the adenosine release. Following 30–35 V stimulation, ATP adenosine signal could reflect an inability to block the was released and the resultant breakdown to adenosine was ecto-ATPases and 5 -nucleotidases which convert ATP measured by the adenosine biosensor (Fig. 4C ). Although to adenosine. Thus, we have assayed the effectiveness ATP breakdown was rapid, only 15 ± 5% of the ATP of Evans Blue, ARL67156 and α,β -methylene-ADP in preventing either ATP breakdown or conversion of ATP to adenosine using HPLC (Fig. 5). We added ε-ATP (5 and 50 μm) to cerebellar slices and observed the metabolism and resultant build up of ε-adenosine (Fig. 5A and D). Although the ecto-ATPase inhibitors ARL67156 and Evans Blue (100 μm) did not abolish ε-ATP breakdown they did slow its rate of metabolism (Fig. 5B and C ). At both 5 and 50 μm ε-ATP, Evans Blue was a more effective inhibitor than ARL67156. For example, after 25 min only 8.6% of 5 μm ε-ATP remained in control, compared with 39.3% in Evans Blue and 26.4% in ARL67156. Both inhibitors appeared more effective at lower substrate concentrations, suggesting a competitive mode of action. Other ecto-ATPase inhibitors, such as α,β -methylene-ADP, suramin and PPADs, were less effective than Evans Blue (not illustrated). Although the ecto-5 -nucelotidase inhibitor, α,β -methylene-ADP (100 μm), had little effect on ATP breakdown, it was the most effective in slowing adenosine formation (Fig. 5C and F ). α,β -Methylene-ADP (100 μm) Figure 4. ATP release is not detected appeared to slow conversion to adenosine by reducing the A, superimposed traces from an adenosine biosensor placed on the breakdown of ADP and reducing the conversion of AMP surface of the molecular layer and an ATP biosensor placed within the to adenosine. Again, inhibition of adenosine formation molecular layer. To maximize the likelihood of detecting ATP, adenosine was first detected (by an adenosine biosensor) then the ATP was more effective at lower substrate concentrations. biosensor was placed in the molecular layer where the adenosine was These enzyme inhibitors would be expected to slow detected. Following stimulation (arrow, 10 s) although adenosine was (but not block) the breakdown of ATP/formation of detected, no ATP release was measured. B, superimposed traces from adenosine, with their effectiveness dependent on substrate an adenosine biosensor placed on the surface of the molecular layer concentration. As they had no effect on the adenosine and an ATP biosensor placed within the molecular layer. Following the signal and did not reveal an ATP signal, we conclude that application of the ecto-ATPase inhibitor Evans Blue (100 μM) the adenosine current was not diminished and there was still no ATP ATP breakdown is most unlikely to contribute to the detection. The traces in B are from the same slice as in A. C, observed adenosine release. superimposed traces from an adenosine biosensor placed on the surface of the molecular layer and an ATP biosensor placed within the layer as in A and B. The biosensors were positioned at an approximately equal distance from the stimulating electrode. TTX Adenosine release can be decreased by inhibiting (0.5 μM) was present to block TTX-dependent adenosine release. parallel fibre transmission Stimulation (30 V, 10 s) caused cell damage and resulted in the electroporation of ATP, which produced a large current on the ATP Our data suggest that the most likely source of adenosine biosensor. A small proportion of the released ATP (∼20%) was broken is the parallel fibres (granule cell axons) which run down to adenosine and was detected by the adenosine biosensor. The along the molecular layer making glutamatergic synapses traces in C have been scaled by sensor calibration. All stimulations were at 20 Hz. onto Purkinje cells. These parallel fibre–Purkinje cell C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 559 synapses are inhibited by presynaptic GABA and mGlu4 release) then adenosine release should also be modulated receptors (Batchelor & Garthwaite, 1992; Neale et al. by these presynaptic receptors. Thus, adenosine release 2001). If adenosine is released from parallel fibres (or was measured before and after the addition of the parallel fibres release a transmitter required for adenosine GABA receptor agonist baclofen (10–50 μm). Baclofen Figure 5. Metabolism of ε-ATP by cerebellar slices at room temperature A, graph plotting percentage composition against time for the breakdown of 50 μM ε-ATP. At time zero there was ∼90% ATP () and 10% ADP (). Following incubation there was a fall in the proportion of ATP and a subsequent increase in AMP () and adenosine ( ) with little change in ADP. B, graph plotting percentage of 50 μM ε-ATP remaining against time in control and in the presence of inhibitors (Evans Blue and ARL67156). C, graph plotting percentage of adenosine against time in control and in the presence of inhibitors during metabolism of 50 μM ε-ATP. α,β -Methylene-ADP (100 μM) was the most effective in slowing the build-up of adenosine. D, graph plotting percentage composition against time for the breakdown of 5 μM ε-ATP. There was a more rapid breakdown of 5 μM ATP compared with 50 μM ε-ATP (compare with A). E, graph plotting percentage of ε-ATP (5 μM) remaining against time in control and in the presence of inhibitors. Note that both Evans Blue and ARL67156 (100 μM) were more effective at slowing the metabolism of 5 μM versus 50 μM ε-ATP (compare E and B). F, graph plotting percentage of adenosine against time in control and in the presence of inhibitors during metabolism of 5 μM ε-ATP. Again inhibitors are more effective at slowing build up of adenosine with lower concentrations of substrate. Graphs A, B and C summarize data from 4 experiments; graphs D, E and F are from 3 experiments. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 560 M. J. Wall and N. Dale J Physiol 581.2 reversibly reduced parallel fibre (PF) EPSP amplitude by 67.8 ± 4.9% (n = 6, Fig. 6A) and also significantly reduced adenosine release by 66.5 ± 8% (n = 5, Fig. 6B). Parallel fibres are also modulated by mGluR4 receptors: application of the mGluR4 receptor agonist l-AP4 (50 μm) caused a reversible decrease in PF EPSP amplitude (41.3 ± 8%, n = 4, Fig. 6C ). Application of l-AP4 ( 50 μm) caused a significant (P < 0.05) reversible decrease in the amplitude of adenosine signals (44.2%, 60 ± 17.7 versus 33.5 ± 10.6 pA, n = 4, Fig. 6D) demonstrating that adenosine release is also modulated by mGluR4 receptors. The matching of the pharmacological profile of parallel fibre–Purkinje cell synaptic transmission with that of adenosine release strongly suggests that activation of parallel fibres is essential for the activity-dependent adenosine release and this release can be modulated by endogenous transmitters. Endogenous adenosine release modulates information flow in the cerebellum Parallel fibre to Purkinje cell synaptic transmission is inhibited by presynaptic A receptors (Kocsis et al. 1984; Rivkees et al. 1995) which raises two possibilities: firstly adenosine release from parallel fibres should be modulated by endogenous adenosine and secondly, the released adenosine could inhibit parallel fibre–Purkinje cell synaptic transmission and inhibit adenosine release (negative feedback). To test the first possibility, the effects of an A receptor antagonist on adenosine release were investigated. Block of A receptors (8CPT, 1 μm) caused an increase in PF EPSP amplitude (45.8 ± 10%, n = 10, Fig. 7A) demonstrating continual adenosine-mediated inhibition and the presence of an extracellular adenosine tone (Takahashi et al. 1995; Dittman & Regehr, 1996). Application of 8CPT (1 μm) also caused a significant increase in the amplitude of adenosine signals (89 ± 15%, n = 8, Fig. 7B) demonstrating that adenosine release is also modulated by the level of extracellular adenosine. The much larger effect of 8CPT on adenosine release compared Figure 6. Adenosine release is modulated by parallel fibre with PF EPSPs may stem from the block of adenosine receptor agonists auto-inhibition (released adenosine binds to A receptors Parallel fibre (PF) EPSPs were evoked every 10 s by a stimulating and inhibits its own release). We investigated the feasibility electrode placed on the surface of the molecular layer and recorded with an extracellular electrode. A, superimposed averages of EPSPs in of this by examining modulation of parallel fibre trans- control, in 25 μM baclofen and following wash. Baclofen reversibly mitter release by released adenosine. If the concentration reduced EPSP amplitude by ∼60%. B, superimposed traces from an of adenosine released reaches a sufficient magnitude, it adenosine biosensor in control, baclofen (25 μM) and in wash. should inhibit glutamate release from parallel fibres (via Adenosine release was evoked by a 10 s stimulus at the arrow. the activation of A receptors). PF EPSPs were evoked Baclofen reversibly reduced adenosine release by ∼50%. C, superimposed averages of EPSPs in control, in 50 μML-AP4 and every 10 s (to measure baseline transmission) and then following wash. L-AP4 reversibly reduced EPSP amplitude by ∼40%. a 10 s train (20 Hz) of stimuli was delivered to release D, superimposed traces from an adenosine biosensor placed on the adenosine. Following the train of stimuli, the amplitude of surface of the molecular layer in control, 50 μML-AP4 and in wash. PF EPSPs was diminished but then slowly recovered back Following L-AP4 application adenosine release was decreased by to the baseline (mean recovery time 122 ± 6.5 s, n = 6) ∼40%. Adenosine release was evoked by a 5 s stimulus (20 Hz) at the arrow. with a time course very similar to the decay of adenosine C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 561 release (decay ∼130 s, Fig. 7C and D). Application of 1 μm physiological stimulus: short duration trains (1–10 s), 8 CPT, to block the A receptors, significantly increased localized to a small area of the slice and at the same minimal the speed of recovery following the train (recovery time strength that evokes transmitter release. reduced from 122 ± 6.5 to 44 ± 8.7 s, Fig. 7C and D). Thus, parallel fibre-dependent adenosine release is functionally What is the source of adenosine? important and acts to auto inhibit both the parallel fibre–Purkinje cell synapse and indeed adenosine release We have provided strong evidence that parallel fibre itself. activity is required for adenosine release. Firstly, the stimulating electrode and biosensors were arranged along a beam of parallel fibres; thus, parallel fibres are activated Discussion by the stimulus and the sensors are in the correct place to We have demonstrated that adenosine can be released measure what is released. Secondly, activation or inhibition from the molecular layer of cerebellar slices by electrical of the G-protein coupled receptors (GABA ,A and B 1 stimulation. This release of adenosine was via a process mGluR4) present on parallel fibre terminals modulates 2+ that is both TTX and Ca sensitive and as ATP release adenosine release by a similar magnitude as synaptic cannot be detected, adenosine is either released directly transmission. This proportionality in the reduction of or rapidly produced by efficient extracellular ATP break- EPSP amplitude and adenosine release suggests a common down. Since adenosine release was modulated by receptors mechanism for adenosine and glutamate release. However, that act on parallel fibre–Purkinje cell synapses, parallel there are apparent differences. Our results suggest that fibres are the most likely source of adenosine. No previous the dependence of adenosine release on extracellular 2+ study has measured adenosine release from cerebellar Ca is ∼1 whereas previous studies have shown that 2+ slices, although adenosine can be released from cultured the Ca dependence of glutamate release at parallel granule cells (Schousboe et al. 1989; Philibert & Dutton, fibre synapses is ∼3 (Mintz et al. 1995; Brown et al. 2+ 1989; Sweeney, 1996). Prolonged electrical stimulation 2004). This difference in Ca dependence may be more (1–5 min) can release adenosine in other brain regions apparent than real as different stimuli were used to elicit (the cortex, hippocampus and striatum, Pedata et al. 1990; glutamate and adenosine release. Single stimuli were used Lloyd et al. 1993). However, we have characterized the to evoke glutamate release (Mintz et al. 1995; Brown properties of adenosine release evoked by a plausibly et al. 2004) whereas in this study, trains of stimuli were Figure 7. The adenosine released by electrical stimulation modulates parallel fibre–Purkinje cell synaptic transmission Parallel fibre (PF) EPSPs were evoked every 10 s by a stimulating electrode placed on the surface of the molecular layer and recorded with an extracellular electrode. A, graph plotting PF EPSP amplitude against time. Application of 1 μM 8CPT, to block A receptors, increased EPSP amplitude by ∼30%. B, superimposed traces from an adenosine biosensor in control and in the presence of 1 μM 8CPT. Adenosine release was evoked by a 7 s stimulus at the arrow. Application of 8CPT increased adenosine release by 43%. C, graph plotting PF EPSP (evoked every 10 s) amplitude against time. At the asterisk a train of stimuli was delivered (10 s, 20 Hz) to cause adenosine release. Following the train there was a marked reduction in PF EPSP amplitude followed by a slow recovery back to control amplitude (PF EPSP amplitude during the train is not plotted). The time course of PF EPSP amplitude recovery was ∼150 s which is very similar to the time course of adenosine release. Application of the A receptor antagonist 8CPT increased PF EPSP amplitude by ∼25% and also markedly speeded recovery following a train of stimuli (∼50 s). Thus, the slow component of recovery results from the released adenosine activating A receptors. D, graph summarizing data from 6 recordings. Following blockade of A receptors there was significant speeding of EPSP recovery. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 562 M. J. Wall and N. Dale J Physiol 581.2 2+ sensors were pushed into the molecular layer (presumably required to release adenosine. The Ca dynamics during the sensor will be closer to the release sites). At the calyx a train of stimuli will be more complex than for a single 2+ of Held, trains of stimuli (10 Hz) release adenosine, which stimulus and may result in intracellular Ca accumulation can be detected indirectly through the activation of A thus reducing the apparent dependence on extracellular 1 2+ receptors and the consequent inhibition of transmitter Ca . Thirdly, it is physiologically plausible that parallel fibres maintain transmitter release at frequencies up to release. This inhibition only reaches significance after and above the stimulation frequency used in this study the first 20 stimuli in a train, suggesting that multiple (Kreitzer & Regehr, 2000). Thus, parallel fibre-dependent stimuli are required to release adenosine (Kimura et al. adenosine release could occur physiologically and act to 2003; Wong et al. 2006). In the hippocampus, 10 Hz but not 3 Hz stimulation released adenosine (measured by limit parallel fibre-dependent excitation of Purkinje cells. inhibition of EPSCs, Brager & Thompson, 2003). Secondly, Finally, moving the biosensor further away (along the same once released the time course of adenosine was very beam of parallel fibres) resulted in only a small reduction slow. The decay of adenosine takes around 130 s which in current amplitude with little change in rise time. is vastly longer than that for classical fast transmitters Other possible sources of adenosine include Purkinje like glutamate. This is similar to neuropeptides, which cells, interneurones and glia. Stimulation in the molecular have a long duration of action and act in a paracrine layer will activate all these cell types but the TTX sensitivity manner affecting many neurones. This may reflect the large of adenosine release makes Bergmann glia an unlikely number of parallel fibres that are stimulated resulting in source of adenosine as they do not fire action potentials the spill over of adenosine and diffusion over a large area. (Clark & Barbour, 1997). Purkinje cells are strongly immunoreactive for adenosine (Braas et al. 1986) and express GABA receptors in their dendrites (Lujan & Shigemoto, 2006) but there is little evidence that Purkinje Mechanism of adenosine release? cells express either A adenosine receptors or mGluR4 receptors. The restricted expression of mGluR4 receptors The inhibition of ENT1 and ENT2 had no effect on also means adenosine release from molecular layer inter- adenosine release and thus adenosine was not transported neurones is unlikely (Mateos et al. 1998). We can also across the cell membrane by these equilibrative trans- exclude glutamate released from parallel fibres activating porters. NBTI and dipyridamole do inhibit adenosine receptors on neurones and glia, since adenosine release was transport leading to greater synaptic inhibition in the not blocked by the glutamate receptor antagonists CNQX cerebellum (M. J. Wall, personal observation) and in and AP5. Furthermore Purkinje cells, interneurones and the hippocampus (Frenguelli et al. 2007). However, it (possibly) Bergmann glia will be shunted by application is possible that other transporters, which are insensitive of the GABA receptor agonist muscimol, which had no A to NBTI/dipyridamole, are involved in adenosine release. effect on adenosine release. Thus, the most likely source Adenosine release was blocked by TTX and by removal of 2+ of adenosine is the parallel fibres. However, on several Ca . The calcium dependence of adenosine release was occasions it has been possible to stimulate parallel fibres lower than that reported for classical neurotransmitters, and record EPSPs without observing release of detectable but was similar to that reported for the release of neuro- amounts of adenosine. Thus, perhaps only a subset of peptides (for example see Peng & Zucker, 1993). The parallel fibres release adenosine. adenosine release thus has the characteristics associated with conventional synaptic release. Nevertheless there are three alternate interpretations of our data. Firstly, an inter- mediate transmitter could in principle cause the down- Properties of adenosine release stream release of adenosine. This seems unlikely as we have Some of the characteristics of adenosine release appear eliminated the obvious candidates for such an intermediate different from parallel fibre–Purkinje cell synaptic trans- transmitter and have shown that inhibiting neurones and mission. Firstly, a single stimulus is sufficient to evoke glia (through GABA receptor activation) has little effect glutamate release from parallel fibres, yet a train of such on adenosine release. Secondly, action potential activity in stimuli is required to produce an adenosine signal on the parallel fibres could release potassium which depolarizes biosensor. This could simply reflect that enough adenosine other cells (including glia) resulting in adenosine release. has to be released to diffuse through the tissue to reach the This seems unlikely as activation of parallel fibre receptors sensor on the slice surface. Clearly the distance between (GABA ,A and mGlu4R) inhibits adenosine release but B 1 synaptic glutamate receptors and the release sites is much would not affect the amount of potassium released during smaller than that between the adenosine biosensor and action potentials. Furthermore any cells depolarized by the release sites. Thus, the requirement for trains of stimuli potassium efflux would probably be shunted by GABA could be an artefact of the recording system. However, receptor activation suggesting that this treatment should similar stimulation protocols were required when the diminish release if that were the case. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 563 Thirdly, adenosine could arise from the breakdown of greater inhibition of parallel fibre–Purkinje cell synapses. exocytotically released ATP. Beierlein & Regehr (2006) Although the increased synaptic inhibition will reduce 2+ have recently described parallel fibre-mediated Ca rises the amplitude of EPSPs, the effects on high-frequency in Bergmann glia that are blocked by PPADs (50 μm) synaptic transmission will probably be more important. and may thus depend on parallel fibre-mediated ATP During high-frequency parallel fibre activity the increased release. However, we have been unable to measure any ATP synaptic inhibition will reduce glutamate depletion and release (either with or without ecto-ATPase inhibition). thus maintain transmission to Purkinje cells (Kimura et al. Thus, the ATP released from parallel fibres, as reported 2003; Wong et al. 2006). Activation of A receptors by endo- genous adenosine at the calyx of Held causes a reduction by Beierlein & Regehr (2006), may be too small to detect by our methods. This contrasts with studies in the retina in the relative amplitude of EPSCs at the end of the train where glial cells release ATP which is rapidly converted compared with the EPSCs at the beginning. This is because by ectoenzymes into adenosine. In this case, the ATP can there is no ambient level of adenosine at the calyx of be detected using luciferin–luciferase chemiluminescence Held and adenosine is released and accumulates during and blockers of ATP breakdown reduce the production the stimulation, reaching a significant concentration late of adenosine (reviewed by Newman, 2004). In the in the train. In contrast, Billups et al. (2005) have shown hippocampus, it has been reported that exogenous ATP is that activation of presynaptic group III mGluRs has no converted to adenosine in less than a second (Dunwiddie net effect on the amplitude of EPSCs. However, there is an et al. 1997). More recent experiments demonstrated that increase in the size of the readily releasable vesicle pool only a small proportion (∼7%) of bath-applied ATP is which is balanced by a reduction in the probability of converted, albeit rapidly to adenosine by hippocampal release. Thus, the synaptic state of the synapse has changed slices (Frenguelli et al. 2007). Our experiments suggest that and the metabolic demand has been reduced. We have not in the cerebellum, only 10–20% of ATP is rapidly converted measured the actions of adenosine during a train of EPSPs to adenosine. Furthermore biosensor measurements of but have investigated recovery from depression following the breakdown of bath-applied ATP show that less than a train. We found that blocking A receptors greatly speeds 10% of the applied ATP is converted to adenosine (Wall the recovery of EPSP depression. Similar observations MJ & Dale N, personal observations). For the complete have been observed following trains of stimuli at the conversion of parallel fibre ATP to occur, high densities of calyx of Held, where blocking group III mGluRs increased ectonucleotidases would have to be localized at parallel the speed of recovery (Billups et al. 2005) while GABA fibre synapses around the site of release such that no receptor activation slowed recovery (Sakaba & Neher, ATP escaped the confines of the synapse. Although 2003). The action of GABA receptorsappearstobe an ecto-ATPase (CD39) is present on the soma and through retardation of synaptic vesicle recruitment during dendrites of Purkinje cells (Wang & Guidotti, 1998) sustained activity. and 5 -nucleotidase activity is present on Bergmann glia The dynamics of adenosine signalling are likely to be membranes (Schoen et al. 1987), the density of these complex as release of adenosine auto-inhibits its own enzymes is unclear and this explanation seems to us release. An intriguing question is whether adenosine unlikely. A direct mechanism of adenosine release is released from active parallel fibres could laterally inhibit supported by Wong et al. (2006) who demonstrated that neighbouring but inactive parallel fibres. Interestingly trains of stimuli release adenosine at the calyx of Held adenosine is also vaso-active and there is evidence that (detectedbyA receptor mediated-inhibition of EPSCs) periods of activity release adenosine resulting in the with no detectable release of ATP (measured by means dilatation of cerebellar blood vessels (Li & Iadecola, 1994). of the luciferin–luciferase method). In the hippocampus, the adenosine released by trains of stimuli (detected by References A receptor-mediated inhibition of EPSCs) could not be reduced by blocking 5 -ectonucleotidases (Mitchell et al. Agarwal RP, Spector T & Parks RE (1977). Tight-binding 1993; Brager & Thompson, 2003). Frenguelli et al. (2007) inhibitors-IV. Inhibition of adenosine deaminases by various have also shown that adenosine release during ischaemia inhibitors. Biochem Pharmac 26, 359–367. Batchelor AM & Garthwaite J (1992). GABAB receptors in the does not arise from the extracellular metabolism of ATP. parallel fibre pathway of rat cerebellum. EurJNeurosci 4, 1059–1064. The functional significance BeierleinM&RegehrWG (2006). Brief bursts of parallel fiber activity trigger calcium signals in Bergmann glia. JNeurosci Irrespective of the detailed mechanism of adenosine 26, 6958–6967. release, its dependency on parallel fibre activity is Billups B, Graham BP, Wong AYC & Forsythe ID (2005). potentially very significant. During periods of high parallel Unmasking group III metabotropic glutamate autoreceptor fibre activity adenosine will be released resulting in an function at excitatory synapses in the rat CNS. J Physiol 565, increased extracellular concentration of adenosine and 885–896. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 564 M. J. Wall and N. Dale J Physiol 581.2 Braas KM, Newby AC, Wilson VS & Snyder SH (1986). Kimura M, SaitohN&Takahashi T (2003). Adenosine A Adenosine-containing neurons in the brain localized by receptor-mediated presynaptic inhibition at the calyx of immunocytochemistry. JNeurosci 6, 1952–1961. Held of immature rats. J Physiol 553, 415–426. Brager DH & Thompson SM (2003). Activity-dependent Kocsis JD, Eng DL & Bhisitkul RB (1984). Adenosine selectively release of adenosine contributes to short-term depression at blocks parallel-fiber-mediated synaptic potentials in rat CA3-CA1 synapses in rat hippocampus. J Neurophysiol 89, cerebellar cortex. Proc Natl Acad SciUSA 81, 6531–6534. 22–26. Kreitzer AC & Regehr WG (2000). Modulation of transmission Brown SP, Safo PK & Regehr WG (2004). Endocannabinoids during trains at a cerebellar synapse. JNeurosci 20, inhibit transmission at granule cell to Purkinje cell synapses 1348–1357. by modulating three types of presynaptic calcium channels. LatiniS&Pedata F (2001). Adenosine in the central nervous JNeurosci 24, 5623–5631. system: release mechanisms and extracellular concentrations. Bultmann R, Trendelenburg M, Tuluc F, Wittenburg H & JNeurochem 79, 463–484. Starke K (1999). Concomitant blockade of P2X-receptors Li J & Iadecola C (1994). Nitric oxide and adenosine mediate and ecto-nucleotidases by P2-receptor antagonists: vasodilation during functional activation in cerebellar functional consequences in rat vas deferens. Naunyn cortex. Neuropharm 33, 1453–1461. Schmiedebers Arch Pharmacol 359, 339–344. Llaudet E, Botting NP, Crayston JA & Dale N (2003). A Clark BA & Barbour B (1997). Currents evoked in Bergmann three-enzyme microelectrode sensor for detecting purine glial cells by parallel fibre stimulation in rat cerebellar slices. release from central nervous system. Biosens Bioelectron 18, J Physiol 502, 335–350. 43–52. Crack BE, Pollard CE, Buekers MW, Roberts SM, Hunt SF, Llaudet E, Hatz S, Droniou M & Dale N (2005). Microelectrode Ingall AH, McKechnie KCW, Ijerzman AP & Leff P (1995). biosensor for real-time measurement of ATP in biological Pharmacological and biochemical analysis of FPL 67156, a tissue. Anal Chem 77, 3267–3273. novel, selective inhibitor of ecto-ATPase. Br J Pharm 114, Llinas RR & Sugimori M (1980). Electrophysiological 475–481. properties of in vitro Purkinje cell somata in mammalian Dale N (1998). Delayed production of adenosine underlies cerebellar slices. J Physiol 305, 171–195. temporal modulation of swimming in frog embryo. J Physiol Lloyd HGE, LindstomK&Fredholm BB (1993). Intracellular 511, 265–272. formation and release of adenosine from rat hippocampal Dale N, Pearson T & Frenguelli BG (2000). Direct measurement slices evoked by electrical stimulation or energy depletion. of adenosine release during hypoxia in the CA1 region of the Neurochem Int 23, 173–185. rat hippocampal slice. J Physiol 526, 143–155. Lujan R & Shigemoto R (2006). Localization of metabotropic Dittman JS & Regehr WG (1996). Contributions of GABA receptor subunits GABA and GABA relative to B1 B2 calcium-dependent and calcium-independent mechanisms synaptic sites in the rat developing cerebellum. EurJNeurosci to presynaptic inhibition at a cerebellar synapse. JNeurosci 23, 1479–1490. 16, 1623–1633. Mateos JM, Azkue J, Sarria R, Kuhn R, GrandesP&Knopfel T Dodge FA & Rahamimoff R (1967). Co-operative action a (1998). Localization of the mGlu4a metabotropic glutamate calcium ions in transmitter release at the neuromuscular receptor in rat cerebellar cortex. Histochem Cell Biol 109, junction. J Physiol 193, 419–432. 135–139. Dulla CG, Dobelis P, Pearson T, Frenguelli BG, Staley KJ & Mintz IM, Sabatini BL & Regehr WG (1995). Calcium control Masino SA (2005). Adenosine and ATP link P to cortical of transmitter release at a cerebellar synapse. Neuron 15, CO2 excitability via pH. Neuron 48, 1011–1023. 675–688. Dunwiddie TV, Diao L & Proctor WR (1997). Adenine Mitchell JB, Lupica CR & Dunwiddie TV (1993). Activity- nucleotides undergo rapid, quantitative conversion to dependent release of endogenous adenosine modulates adenosine in the extracellular space in rat hippocampus. synaptic responses in the rat hippocampus. JNeurosci 13, JNeurosci 17, 7673–7682. 3439–3447. Fredholm BB (1997). Adenosine and neuroprotection. Int Rev Neale SA, GarthwaiteJ&Batchelor AM (2001). Metabotropic Neurobiol 40, 259–280. glutamate receptor subtypes modulating neurotransmission Fredholm BB, Arslan G, Halldner L, Kull N, Shulte G & at parallel fibre-Purkinje cell synapses in rat cerebellum. Wasserman W (2000). Structure and function of adenosine Neuropharm 41, 42–49. receptors and their genes. Naunyn Schmeidebers Arch Newman EA (2004). Glial modulation of synaptic transmission Pharmacol 362, 364–374. in the retina. Glia 47, 268–274. Frenguelli BG, Wigmore G, Llaudet E & Dale N (2007). Noji T, Karaswa A & Kusaka H (2004). Adenosine uptake Temporal and mechanistic dissociation of ATP and inhibitors. Eur J Pharm 495, 1–16. adenosine release during ischemia in the mammalian Pedata F, Pazzagli M, TilliS&PepeuG (1990). Regional hippocampus. JNeurochem differences in the electrically stimulated release of DOI: 10.1111/j.1471-4159.2006.04425.x endogenous and radioactive adenosine and purine Geiger JD & Nagy JI (1986). Distribution of adenosine derivatives from rat brain slices. Naunyn Schmiedebers Arch deaminase activity in rat brain and spinal cord. JNeurosci 6, Pharmacol 342, 447–453. 2707–2714. Peng Y & Zucker RS (1993). Release of LHRH is linearly related 2+ Hamann M & Attwell D (1996). Non-synaptic release of ATP to the time integral of presynaptic Ca elevation above a by electrical stimulation in slices of rat hippocampus, threshold level in bullfrog sympathetic ganglia. Neuron 10, cerebellum and habenula. EurJNeurosci 8, 150–1515. 465–473. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 565 Philibert RA & Dutton GR (1989). Dihydropyridines modulate Takahashi M, Kovalchuk Y & Atwell D (1995). Pre- and K -evoked amino acid and adenosine release from cerebellar postsynaptic determinants of EPSC waveform at cerebellar neuronal cultures. Neurosci Lett 102, 97–102. climbing fiber and parallel fiber to Purkinje cell synapses. Reid CA, Bekkers JM & Clements JD (1998). N- and P/Q-type JNeurosci 15, 5693–5702. 2+ Ca channels mediate transmitter release with a similar Wall MJ & Usowicz MM (1997). Development of action cooperativity at rat hippocampal autapses. JNeurosci 18, potential-dependent and independent spontaneous GABA 2849–2855. receptor-mediated currents in granule cells of postnatal rat Riquelme R, Miralles CP & De Blas AL (2002). Bergmann glia cerebellum. Eur J Neurosci 9, 533–548. GABA receptors concentrate on the glial processes that Wang TF & Guidotti G (1998). Widespread expression of wrap inhibitory synapses. JNeurosci 22, 10720–10730. ecto-apyrase (CD39) in the central nervous system. Brain Res Rivkees SA, Price SL & Zhou FC (1995). Immunohistochemical 790, 318–321. detection of A1 adenosine receptors in rat brain with Wong AYC, Billups B, Johnston J, Evans RJ & Forsythe ID emphasis on localization in the hippocampal formation, (2006). Endogenous activation of adenosine A1 receptors, cerebral cortex, cerebellum, and basal ganglia. Brain Res 677, but not P2X receptors, during high-frequency synaptic 193–203. transmission at the calyx of Held. J Neurophysiol 95, Rudolphi KA, Schubert P, Parkinson FE & Fredholm BB 3336–3342. (1992). Neuroprotective role of adenosine in cerebral YuanY&Atchison WD (1999). Comparative effects of ischemia. Trends Pharm Sci 13, 439–445. methylmercury on parallel-fiber and climbing-fiber Safiulina VF, Kasyanov AM, GiniatullinR&Cherubini E responses of rat cerebellar slices. J Pharmacol Exp Ther 288, (2005). Adenosine down regulates giant depolarizing 1015–1025. potentials in the developing rat hippocampus by exerting a Zimmermann H (1996). Biochemistry, localization and negative control on glutamatergic inputs. J Neurophysiol 94, functional roles of ecto-nucleotidases in the nervous system. 2797–2804. Prog Neurobiol 49, 589–618. Sakaba T & Neher E (2003). Direct modulation of synaptic vesicle priming by GABA receptor activation at a glutamatergic synapse. Nature 424, 775–778. Schoen SW, Graeber MB, Reddington M & Kreutzberg GW (1987). Light and electron microscopical immunocytochemistry of 5 -nucleotidase in rat cerebellum. Histochem 87, 107–113. Schousboe A, Frandsen A & Drejer J (1989). Evidence for Acknowledgements evoked release of adenosine and glutamate from cultured We thank Dr Enrique Llaudet and Shakila Bibi of Sarissa cerebellar granule cells. Neurochem Res 14, 871–875. Biomedical Ltd for biosensor manufacture and Dr Robert Sweeney MI (1996). Adenosine release and uptake in cerebellar Eason and Matthew Simpson for assistance with the HPLC granule neurons both occur via an equilibrative nucleoside carrier that is modulated by G proteins. JNeurochem 67, experiments. This work was supported by the Wellcome Trust 81–88. (M.J.W. and N.D.). C  C 2007 The Authors. Journal compilation 2007 The Physiological Society http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journal of Physiology Pubmed Central

Auto-inhibition of rat parallel fibre–Purkinje cell synapses by activity-dependent adenosine release

The Journal of Physiology , Volume 581 (Pt 2) – Mar 8, 2007

Loading next page...
 
/lp/pubmed-central/auto-inhibition-of-rat-parallel-fibre-purkinje-cell-synapses-by-VaYVA3wUsN

References (60)

Publisher
Pubmed Central
ISSN
0022-3751
eISSN
1469-7793
DOI
10.1113/jphysiol.2006.126417
Publisher site
See Article on Publisher Site

Abstract

J Physiol 581.2 (2007) pp 553–565 553 Auto-inhibition of rat parallel fibre–Purkinje cell synapses by activity-dependent adenosine release Mark J. Wall and Nicholas Dale Neuroscience Group, Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL, UK Adenosine is an important signalling molecule involved in a large number of physiological functions. In the brain these processes are as diverse as sleep, memory, locomotion and neuro- protection during episodes of ischaemia and hypoxia. Although the actions of adenosine, through cell surface G-protein-coupled receptors, are well characterized, in many cases the sources of adenosine and mechanisms of release have not been defined. Here we demonstrate the activity-dependent release of adenosine in the cerebellum using a combination of electro- physiology and biosensors. Short trains of electrical stimuli delivered to the molecular layer in 2+ vitro, release adenosine via a process that is both TTX and Ca sensitive. As ATP release cannot be detected, adenosine must either be released directly or rapidly produced by highly localized and efficient extracellular ATP breakdown. Since adenosine release can be modulated by receptors that act on parallel fibre–Purkinje cell synapses, we suggest that the parallel fibres release adenosine. This activity-dependent adenosine release exerts feedback inhibition of parallel fibre–Purkinje cell transmission. Spike-mediated adenosine release from parallel fibres will thus powerfully regulate cerebellar circuit output. (Received 12 December 2006; accepted after revision 1 March 2007; first published online 8 March 2007) Corresponding author M. J. Wall: Neuroscience Group, Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL, UK. Email: mark.wall@warwick.ac.uk Adenosine is an important neuromodulator in the central The presence of adenosine, adenosine deaminase and nervous system, playing a role in a plethora of physio- A receptors in the cerebellar cortex (Braas et al. 1986; logical and pathophysiological processes. The action of Geiger & Nagy, 1986; Rivkees et al. 1995) strongly suggests adenosine on cell surface receptors is well defined with that adenosine plays an important role in cerebellar A ,A ,A and A receptors all cloned (Fredholm et al. function. The activation of A receptors inhibits synaptic 1 2a 2b 3 1 2000). Although extensively studied, the cellular source transmission between parallel fibres and Purkinje cells and mechanisms of adenosine release remain unclear (Kocsis et al. 1984). These receptors are tonically activated (for review see Latini & Pedata, 2001). Adenosine can by endogenous adenosine, since application of A receptor in principle gain access to the extracellular space by antagonists enhances synaptic transmission (Takahashi the breakdown of ATP, by translocation from cell cyto- et al. 1995; Dittman & Regehr, 1996). The source of this plasm via nucleoside transport proteins or possibly by the adenosine has not been determined but could arise from exocytosis of adenosine itself. There has been considerable the release of adenosine or the release of ATP and its investigation of adenosine release during pathological subsequent metabolism. A recent report has suggested episodes such as hypoxia, ischaemia and hypercapnia that ATP can be released from parallel fibres (Beierlein as adenosine is neuroprotective (Rudolphi et al. 1992; & Regehr, 2006). Fredholm, 1997; Dale et al. 2000; Dulla et al. 2005). To examine this issue we have used selective and sensitive 2+ Release under these conditions is often Ca independent, microelectrode biosensors (Llaudet et al. 2003) to measure relatively insensitive to TTX and is not mediated via the release of adenosine from cerebellar slices in real time. glutamate receptor activation. In contrast, little is known These biosensors are small enough (25–50 μm diameter) about the physiological release of adenosine with few to place either in or close to defined areas in cerebellar examples where a role and cellular source of adenosine slices. Here we report that adenosine can be released from have been identified (but see Dale, 1998). In many cases, the molecular layer by using a physiological stimulus, short adenosine release is evoked with stimuli such as high K , bursts (1–10 s) of focal electrical stimuli at the same voltage prolonged electrical stimulation and glutamate receptor used to elicit synaptic transmission. The adenosine release 2+ activation (Latini & Pedata, 2001). The physiological is both TTX and Ca sensitive and does not appear to arise relevance of these experiments is unclear. from the extracellular metabolism of ATP. Modulation C  C 2007 The Authors. Journal compilation 2007 The Physiological Society DOI: 10.1113/jphysiol.2006.126417 554 M. J. Wall and N. Dale J Physiol 581.2 of parallel fibre–Purkinje cell synaptic transmission can potentials in Purkinje cells and interneurones (Clark increase or decrease adenosine release, strongly suggesting & Barbour, 1997). Parallel fibre EPSP amplitude was that parallel fibres are involved in adenosine release. estimated from the CNQX/kynurenate-sensitive potential, which was measured by subtracting what remained in CNQX/kynurenate from control potentials. Confirmation Methods of PF EPSP identity was achieved by evoking pairs of EPSPs Slice preparation (interval 50 ms) and observing facilitation (20–30%) and by examining the pharmacological profile (inhibition by Transverse slices of cerebellum (400 μm) were prepared , GABA and mGlu4R receptor agonists). Sensor signals 1 B from male Wistar rats, at postnatal days 21–28 (P21–28), were acquired at 1 kHz with either a Digidata 1322A with modified methods based on Llinas & Sugimori (Axon) or a MiniDigi (Axon) using pCLAMP 9.2 (Axon) (1980). As previously described (Wall & Usowicz, 1997) or Axoscope 9.2 (Axon). Extracellular recordings were and in accordance with the UK Animals (Scientific made using an ISO-DAM extracellular amplifier (WPI, Procedures) Act 1986, male rats were killed by cervical Stevenage, UK), filtered at 1 kHz and digitized on line dislocation and decapitated. The cerebellum was rapidly (10 kHz) with a Digidata 1322A controlled by pCLAMP removed and transverse slices were cut on a Microm 9.2. HM 650V microslicer (Carl Zeiss, Welwyn Garden City, ◦ 2+ 2+ UK) in cold (2–4 C) high Mg ,low Ca aCSF, composed Biosensor characteristics of (mm): 127 NaCl, 1.9 KCl, 7 MgCl , 0.5 CaCl , 1.2 2 2 KH PO , 26 NaHCO ,10 d-glucose (pH 7.4 when bubbled 2 4 3 Biosensors were obtained from Sarissa Biomedical Ltd with 95% O and 5% CO ). Slices were stored in normal 2 2 (Coventry, UK). In brief the adenosine biosensor consisted aCSF (1.3 mm MgCl , 2.4 mm CaCl )atroomtemperature 2 2 of three entrapped enzymes (adenosine deaminase, for 1–6 h before recording. nucleoside phosphorylase and xanthine oxidase) within a matrix that was deposited around a Pt or Pt/Ir (90/10) Recording from slices wire etched to 25–50 μm (Llaudet et al. 2003). The biosensor had an exposed length of 500 μm that was An individual slice was transferred to a recording coated with enzymes and thus capable of detecting purines. −1 chamber, submerged in aCSF and perfused at 6 ml min Biosensors had an additional screening layer, which greatly (30–35 C). The slice was placed upon a suspended reduced the responses to non-specific electro active inter- grid to allow perfusion of the slice from above and ferents (such as 5-HT, dopamine, noradrenaline and below and thus reduce the likelihood of hypoxia. All ascorbate). Screened null sensors, possessing the matrix solutions were vigorously bubbled (95% O and 5% but no enzymes, were used to control for the release CO ) and all tubing had low gas permeability (Tygon; of any non-specific electro active interferents. The ATP Fisher Scientific, Loughborough, UK). For the stimulation biosensor consisted of the entrapped enzymes glycerol of purine release and parallel fibre–Purkinje cell (PF) kinase and glycerol-3-phosphate oxidase (Llaudet et al. EPSPs, square voltage pulses (2–8 V, 200 μs duration) 2005). Glycerol (2 mm) was included in solutions as were delivered by an isolated pulse stimulator (Model glycerol is a co-substrate required for ATP detection. 2100 AM systems; Olympic Peninsula, Washington, DC, Biosensors were calibrated with known concentrations USA) via a concentric bipolar metal stimulating electrode of adenosine and ATP (10 μm typically giving responses (FHC) placed on the surface of the molecular layer. of 2–3 nA for adenosine and 1.5 nA for ATP). Calibration Purine biosensors were either positioned just above the was performed before the slice was present in the perfusion surface of the slice (bent so their longitudinal surface was chamber and after the experiment (following slice parallel to the stimulated molecular layer) or carefully removal), this allowing quantification of any run-down in inserted (at an angle of ∼70 deg) into the stimulated sensitivity during the experiment. Previous reports have molecular layer. For the extracellular recording of PF detailed the properties of adenosine and ATP biosensors EPSPs, an electrode (aCSF-filled microelectrode) was (Llaudet et al. 2003, 2005): selectivity, linear responses to placed on the same track along which the parallel fibres increasing analyte concentration and rapid response. travel (for example see Yuan & Atchison, 1999). A typical extracellular field potential consisted of an initial Determination of ATP breakdown inhibition component which persisted in either 10 μm CNQX or 5mm kynurenate but was blocked by 1 μm TTX (parallel To determine which compounds were the most effective at fibre volley), followed by a component which could be blocking the extracellular metabolism of ATP, the break- blockedby1 μm TTX and greatly reduced by either 10 μm down of etheno-ATP (ε-ATP) was measured using HPLC. CNQX or 5 mm kynurenate. This component is probably Individual 400 μm cerebellar slices were incubated (on a produced by parallel fibre-mediated glutamatergic shaker, at room temperature) in 400 μl of aCSF containing excitatory synaptic currents and subsequent action either 5 or 50 μm ε-ATP (2 or 20 nmol of ε-ATP) with or C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 555 without inhibitors. At 0, 10, 25 and 45 min, 50 μl samples Fig. 1A). The current waveform was not determined by were taken and snap frozen with dry ice. sensor-response kinetics, as it was much slower than Samples were thawed out and diluted (1 in 10) in the sensor response to purine application (Llaudet et al. distilled water. HPLC analysis was performed with a Luna 2003). Since the adenosine biosensors can detect inosine C8 (2) reverse phase column and a Thermo Separation and hypoxanthine as well as adenosine, EHNA, a specific Products HPLC gradient pump (P2000) and fluorescence inhibitor of adenosine deaminase (the first enzyme in the detector (FL3000). The mobile phase consisted of 20 mm biosensor enzyme cascade), was used to determine the potassium phosphate pH 6.0 (solution A) and 75% purine detected (Agarwal et al. 1977; Safiulina et al. 2005). potassium phosphate pH 6.0 and 25% methanol (solution EHNA (20 μm) greatly reduced (85 ± 5%, n = 3) the B). A concave gradient was run going from 100% response of adenosine biosensors to exogenous adenosine solution A to 100% solution B in 10 min. ε-ATP typically (10 μm) but had no effect on the detection of inosine eluted around 3.5 min, ε-ADP around 4 min, ε-AMP (and by inference hypoxanthine, Fig. 1B). In five slices, 5 min and ε-adenosine around 9 min. The column was application of 20 μm EHNA greatly reduced the biosensor re-equilibrated with solution A for 10 min between runs. current following stimulation (mean inhibition 75 ± 3%, To quantify ε-ATP breakdown, the relative proportions of Fig. 1C ). Thus, most of the biosensor current results from the breakdown products was obtained from peak areas. adenosine detection. The remaining current is probably a combination of adenosine (as EHNA only partial blocks the adenosine sensor response) and inosine either directly Drugs released or as a consequence of adenosine breakdown (due All drugs were made up as 10–100 mm stock solutions, to incomplete inhibition of adenosine deaminase in the stored frozen and then thawed and diluted with aCSF on slice). the day of use. R(+)-baclofen hydrochloride, muscimol The amount of adenosine released was dependent on Evans Blue, α,β -methylene-ADP, adenosine, inosine, 6-[ both the frequency and duration of stimulation. With (4-nitrobenzyl)thiol]-9-β -d-ribofuranosylpurine (NBTI) 100 stimuli, no adenosine release could be detected 8-cyclopentyltheophylline (8CPT) and dipyridamole were purchased from Sigma. Erythro-9-(2-hydroxy- 3-nonyl) adenine (EHNA), 6-cyano-7-nitroquinoxaline- 2,3-dione (CNQX), ARL67165, l-AP4, d(–)-2- amino-5-phosphonopentanoic acid (AP5) were purchased from Tocris-Cookson. ATP was purchased from Roche and etheno-ATP (ε-ATP) was purchased as a 5mm solution from Invitrogen. Results Trains of electrical stimuli release adenosine in the molecular layer To investigate the release of purines in the cerebellum, trains (1–10 s, 2–20 Hz) of electrical stimuli were applied to areas of the cerebellum and purine release measured in real time using specific and selective microelectrode Figure 1. Electrical stimulation in the molecular layer releases biosensors (Llaudet et al. 2003). Stimuli applied to adenosine the molecular layer of transverse slices, consistently A, superimposed current traces from adenosine (Ado) and null produced a slow rising current on adenosine biosensors, biosensors following electrical stimulation of the molecular layer (5 V for 5 s at arrow). The lack of signal on the null sensor indicates that positioned on the surface of the molecular layer (Fig. 1A). the current on the adenosine biosensor is due to purine detection and Movement of the stimulator and sensor to the adjacent is not non-specific. B, application of 10 μM adenosine (Ado) and then granule cell layer caused a loss of current, suggesting the 10 μM inosine (Ino) produced current responses on an adenosine detected analyte was released in the molecular layer. The biosensor. Addition of 20 μM EHNA (to block adenosine deaminase) sensor-current (amplitude 20–200 pA) had a 10–90% rise almost abolished the response to adenosine but had little effect on the response to inosine. C, superimposed traces from an adenosine time of 14.4 ± 2.2 s and a 10–90% decay of 131.6 ± 14.2 s biosensor in control and in the presence of 20 μM EHNA (to block (n = 30). The current was due to purine detection, as no adenosine deaminase) following electrical stimulation of the molecular current was produced on a null sensor (identical to an layer (8 V for 8 s at arrow). EHNA reduced the current by ∼75% adenosine sensor but lacking the purine detection enzyme demonstrating that most of the signal is due to adenosine detection. All stimulations were at a frequency of 20 Hz. cascade) placed close to the adenosine biosensor (n = 8, C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 556 M. J. Wall and N. Dale J Physiol 581.2 at a frequency of 2 Hz but was detected at 5 Hz. release (n = 11, Fig. 3A) demonstrating that adenosine Maximal adenosine release occurred around 20 Hz with release is action potential dependent. In seven slices, little increase at higher frequencies (40–80 Hz, Fig. 2A). adenosine release was also abolished by substitution of 2+ 2+ Increasing the stimulus train length released more normal aCSF for Ca -free aCSF (Mg increased to 2+ adenosine (Fig. 2B). At 20 Hz, the shortest train to release 3.7 mm, Fig. 3B). To measure the Ca dependence of detectable adenosine was ∼1 s (20 stimuli) but in most slices 5–8 s (100–160 stimuli) were required. With an interval between trains of stimuli of 5 min, adenosine release was reliable and did not decrement. However, with shorter intervals the amount of adenosine released diminished. If it is assumed that all the sensor-current results from adenosine detection, then it is possible to calculate the adenosine concentration at the sensor. In 15 slices, an 8 s (20 Hz) stimulation produced concentrations ranging from 130 to 1650 nm (mean 492 ± 89 nm). This range of concentrations may stem from different distances between the adenosine release sites and the biosensor. Mechanisms of adenosine release 2+ We next examined the TTX-, Ca - and receptor- sensitivity of adenosine release. Application of the sodium channel blocker TTX (0.5–1 μm) abolished adenosine Figure 3. Mechanisms of adenosine release A, superimposed current traces from an adenosine biosensor placed on the surface of the molecular layer in control, in 10 μM CNQX + 50 μM AP5 and in 0.5 μM TTX. The adenosine release following electrical stimulation (5 V, 8 s at arrow) was abolished by TTX but was insensitive to the block of glutamate receptors. B, superimposed traces from an adenosine biosensor placed on the surface of the molecular layer in 2+ 2+ control, Ca -free aCSF (3.7 mM Mg ) and following reintroduction 2+ of Ca (wash). The adenosine release following electrical stimulation 2+ (7 V, 8 s at arrow) was abolished by removal of Ca . Inset, graph 2+ plotting normalized adenosine release against external Ca concentration. Adenosine release was normalized to what occurred at 2+ 2.4 mM Ca and summarizes data from 6 slices. C, superimposed traces from an adenosine biosensor placed on the surface of the Figure 2. Properties of adenosine release molecular layer in control and in the presence of the GABA receptor A, graph summarizing how adenosine release varies with stimulation agonist muscimol (30 μM). Adenosine was released by a 5 V, 7 s frequency (data from 4 slices). The number of stimuli was kept stimulus at the arrow. Inset, superimposed averages of EPSPs in constant (100) and the stimulation frequency was varied between 2 control and in 30 μM muscimol ( ). Muscimol reduced EPSP amplitude and 80 Hz. The data are normalized to the concentration of adenosine but had little effect on the volley. D, superimposed traces from an detected at a stimulus frequency of 20 Hz. B, superimposed traces adenosine biosensor placed on the surface of the molecular layer in from an adenosine biosensor following 20, 100 and 160 stimuli. The control and in the presence of 5 μM NBTI and 10 μM dipyridamole to stimulus frequency was kept constant at 20 Hz. The stimulation block equilibrative transport. Adenosine was released by a 5 V, 7 s strength was between 3 and 8 V (A) and 5 V (B). stimulus at the arrow. All stimuli were at a frequency of 20 Hz. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 557 2+ it seems very unlikely that adenosine is released as a adenosine release, extracellular Ca concentration was 2+ result of the downstream actions of another transmitter. changed between 0.5 and 5 mm with Mg kept at Furthermore it is unlikely to come from Purkinje cells or 1.3 mm. Adenosine release was examined in the presence 2+ interneurones as they will be shunted by GABA receptor of 1 μm 8CPT to block A receptors (see later). The Ca A activation. dependence was approximately linear (1.1 ± 0.1, n = 6, Fig. 3B inset) which is lower than that measured for Direct release of adenosine could occur via the classical neurotransmitters (Dodge & Rahamimoff, 1967; equilibrative nucleoside transporters ENT1 and ENT2 Mintz et al. 1995; Reid et al. 1998). translocating adenosine from the cytoplasm to the extracellular space. However, blockade of these molecules It is possible that the action potential-dependent 2+ 2+ by a combination of 5 μm NBTI and 10 μm dipyridamole opening of Ca channels and subsequent entry of Ca (for review see Noji et al. 2004) had no significant directly causes the release of adenosine. However, it is also (P = 0.71) effect on the amount of adenosine released possible that the stimulation indirectly evokes adenosine (n = 5, Fig. 3D, 41.6 ± 7.8 versus 38.3 ± 10.7 pA, 1 μm release through the downstream actions of another neuro- 8CPT was present to prevent A receptor activation). transmitter. We have taken two approaches to address this Adenosine was not released as a result of electroporation possibility. Firstly, we have tested the most likely neuro- and consequent release of cytoplasmic ATP (for example, transmitter candidates that could perform this role (mean see Hamann & Attwell, 1996) (with subsequent conversion current in control versus mean current with antagonist to adenosine by ecto-ATPases) since electroporation is present). Adenosine release was not sensitive to 10 μm insensitive to TTX and only occurred at much higher CNQX and 50 μm AP5 (AMPA and NMDA glutamate stimulus strengths (see later and Fig. 4C ). receptor antagonists, 53.5 ± 10 versus 51.8 ± 11 pA, n = 5, Fig. 3A), 100 μm CPCCOET or 300 μm AIDA (Group 1 mGluR antagonists, 44.2 ± 13 versus 52 ± 18.5 pA, n = 3), 10 μm PPADs (P2 receptor antagonist, 34.6 ± 5 versus ATP release cannot be detected 32.8 ± 4.3 pA, n = 3), 300 μml-NMMA (NO synthase An important potential source of adenosine is the inhibitor, n = 4, 25.3 ± 6.3 versus 21.6 ± 5 pA) and extracellular metabolism of ATP. To investigate whether prazosin (α adrenoceptor antagonist, 103 ± 13 versus adenosine is released in the form of ATP, a number of 99.6 ± 14 pA, n = 3). Furthermore application of 10 μm 5-HT or 10 μm noradrenaline did not induce adenosine approaches were taken. Firstly, ATP biosensors were used release (n = 3). Secondly, we have applied the GABA to test directly for ATP release (Llaudet et al. 2005). These receptor agonist muscimol (30 μm) and measured biosensors have a detection limit of ∼60 nm (producing adenosine release. The rationale for this approach is that ∼10 pA of current). In eight slices an ATP biosensor the activation of GABA receptors will hyperpolarize and was positioned close to the adenosine biosensor on the reduce the input resistance of neurones. The actions of surface of the molecular layer. Following stimulation, muscimol on glia are less certain, but Bergmann glia adenosine was detected but no signal was observed on express GABA receptors on their processes and therefore the ATP biosensor. In three further slices, after measuring may be inhibited (Riquelme et al. 2002). Thus, if adenosine adenosine release, the adenosine biosensor was carefully is released as a consequence of the downstream actions of removed and replaced with an ATP sensor (thus potential another transmitter, inhibition of the target cells should ATP release was measured in exactly the same place greatly reduce or abolish adenosine release. Furthermore, where adenosine was detected). Again there was no signal if GABA receptors are not present on parallel fibres following stimulation. It is possible that ATP is released and parallel fibres are the source of adenosine then deep within the slice and is then metabolized to adenosine muscimol should have no effect on adenosine release. before it reaches the sensor on the surface of the slice. The effectiveness of muscimol (30 μm) was confirmed by To investigate this, ATP biosensors were inserted into two observations: (1) the prolonged loss of spontaneous the molecular layer, in the region where adenosine was detected. Again no signal was observed on the ATP action potential firing in Purkinje cells following muscimol sensor following stimulation, although adenosine could application (n = 3) and (2) muscimol (30 μm) reduced still be detected on the slice surface (n = 8, Fig. 4A). To the amplitude of parallel fibre EPSPs by 56 ± 6.2% (n = 6) reduce the possible extracellular ATP breakdown we have with little effect upon the parallel fibre volley (8.1 ± 7.7% made use of three agents: Evans Blue and ARL67156 reduction, Fig. 3C inset, n = 6). The lack of effect of muscimol on the parallel fibre volley suggests few GABA (which inhibit ecto-ATPases, Crack et al. 1995; Bultmann receptors are expressed by parallel fibres and thus the et al. 1999) and α,β -methylene-ADP (which inhibits reduction in EPSP amplitude results from a reduction in ecto-5 -nucleotidase, for review see Zimmerman, 1996). In the resistance of Purkinje cells. Since muscimol (30 μm) the presence of Evans Blue (100 μm, n = 5) or ARL67156 had no significant effect on the magnitude of adenosine (100 μm, n = 3) no signal was detected on the inserted release (55.3 ± 8.5 versus 58.1 ± 8.1 pA, Fig. 3C , n = 5) ATP sensor and the adenosine signal was not significantly C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 558 M. J. Wall and N. Dale J Physiol 581.2 different from the control (P = 0.53, Fig. 4B). There was was converted to adenosine (n = 4). If adenosine were to also no significant difference between the amount of arise from ATP breakdown, this suggests that to give the adenosine detected in control, following application of observed signal of ∼0.5 μm adenosine, around 2.5–5 μm 100 μm α,β -methylene-ADP (47 ± 19 versus 54 ± 17 pA, ATP would have to be released. This quantity of ATP P = 0.06) and after wash (P = 0.22). would give an easily measurable signal on the ATP sensor. We have exploited the electroporation-induced release Therefore it seems unlikely that adenosine arises from the of ATP (Hamann & Attwell, 1996) to measure the kinetics extracellular breakdown of ATP. of ATP breakdown. An ATP biosensor was inserted into the molecular layer and an adenosine sensor was placed very Inhibitors slow ATP metabolism close to the ATP sensor on the surface of the molecular layer. TTX (0.5 μm) was present to prevent TTX-sensitive The inability to either directly detect ATP or reduce the adenosine release. Following 30–35 V stimulation, ATP adenosine signal could reflect an inability to block the was released and the resultant breakdown to adenosine was ecto-ATPases and 5 -nucleotidases which convert ATP measured by the adenosine biosensor (Fig. 4C ). Although to adenosine. Thus, we have assayed the effectiveness ATP breakdown was rapid, only 15 ± 5% of the ATP of Evans Blue, ARL67156 and α,β -methylene-ADP in preventing either ATP breakdown or conversion of ATP to adenosine using HPLC (Fig. 5). We added ε-ATP (5 and 50 μm) to cerebellar slices and observed the metabolism and resultant build up of ε-adenosine (Fig. 5A and D). Although the ecto-ATPase inhibitors ARL67156 and Evans Blue (100 μm) did not abolish ε-ATP breakdown they did slow its rate of metabolism (Fig. 5B and C ). At both 5 and 50 μm ε-ATP, Evans Blue was a more effective inhibitor than ARL67156. For example, after 25 min only 8.6% of 5 μm ε-ATP remained in control, compared with 39.3% in Evans Blue and 26.4% in ARL67156. Both inhibitors appeared more effective at lower substrate concentrations, suggesting a competitive mode of action. Other ecto-ATPase inhibitors, such as α,β -methylene-ADP, suramin and PPADs, were less effective than Evans Blue (not illustrated). Although the ecto-5 -nucelotidase inhibitor, α,β -methylene-ADP (100 μm), had little effect on ATP breakdown, it was the most effective in slowing adenosine formation (Fig. 5C and F ). α,β -Methylene-ADP (100 μm) Figure 4. ATP release is not detected appeared to slow conversion to adenosine by reducing the A, superimposed traces from an adenosine biosensor placed on the breakdown of ADP and reducing the conversion of AMP surface of the molecular layer and an ATP biosensor placed within the to adenosine. Again, inhibition of adenosine formation molecular layer. To maximize the likelihood of detecting ATP, adenosine was first detected (by an adenosine biosensor) then the ATP was more effective at lower substrate concentrations. biosensor was placed in the molecular layer where the adenosine was These enzyme inhibitors would be expected to slow detected. Following stimulation (arrow, 10 s) although adenosine was (but not block) the breakdown of ATP/formation of detected, no ATP release was measured. B, superimposed traces from adenosine, with their effectiveness dependent on substrate an adenosine biosensor placed on the surface of the molecular layer concentration. As they had no effect on the adenosine and an ATP biosensor placed within the molecular layer. Following the signal and did not reveal an ATP signal, we conclude that application of the ecto-ATPase inhibitor Evans Blue (100 μM) the adenosine current was not diminished and there was still no ATP ATP breakdown is most unlikely to contribute to the detection. The traces in B are from the same slice as in A. C, observed adenosine release. superimposed traces from an adenosine biosensor placed on the surface of the molecular layer and an ATP biosensor placed within the layer as in A and B. The biosensors were positioned at an approximately equal distance from the stimulating electrode. TTX Adenosine release can be decreased by inhibiting (0.5 μM) was present to block TTX-dependent adenosine release. parallel fibre transmission Stimulation (30 V, 10 s) caused cell damage and resulted in the electroporation of ATP, which produced a large current on the ATP Our data suggest that the most likely source of adenosine biosensor. A small proportion of the released ATP (∼20%) was broken is the parallel fibres (granule cell axons) which run down to adenosine and was detected by the adenosine biosensor. The along the molecular layer making glutamatergic synapses traces in C have been scaled by sensor calibration. All stimulations were at 20 Hz. onto Purkinje cells. These parallel fibre–Purkinje cell C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 559 synapses are inhibited by presynaptic GABA and mGlu4 release) then adenosine release should also be modulated receptors (Batchelor & Garthwaite, 1992; Neale et al. by these presynaptic receptors. Thus, adenosine release 2001). If adenosine is released from parallel fibres (or was measured before and after the addition of the parallel fibres release a transmitter required for adenosine GABA receptor agonist baclofen (10–50 μm). Baclofen Figure 5. Metabolism of ε-ATP by cerebellar slices at room temperature A, graph plotting percentage composition against time for the breakdown of 50 μM ε-ATP. At time zero there was ∼90% ATP () and 10% ADP (). Following incubation there was a fall in the proportion of ATP and a subsequent increase in AMP () and adenosine ( ) with little change in ADP. B, graph plotting percentage of 50 μM ε-ATP remaining against time in control and in the presence of inhibitors (Evans Blue and ARL67156). C, graph plotting percentage of adenosine against time in control and in the presence of inhibitors during metabolism of 50 μM ε-ATP. α,β -Methylene-ADP (100 μM) was the most effective in slowing the build-up of adenosine. D, graph plotting percentage composition against time for the breakdown of 5 μM ε-ATP. There was a more rapid breakdown of 5 μM ATP compared with 50 μM ε-ATP (compare with A). E, graph plotting percentage of ε-ATP (5 μM) remaining against time in control and in the presence of inhibitors. Note that both Evans Blue and ARL67156 (100 μM) were more effective at slowing the metabolism of 5 μM versus 50 μM ε-ATP (compare E and B). F, graph plotting percentage of adenosine against time in control and in the presence of inhibitors during metabolism of 5 μM ε-ATP. Again inhibitors are more effective at slowing build up of adenosine with lower concentrations of substrate. Graphs A, B and C summarize data from 4 experiments; graphs D, E and F are from 3 experiments. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 560 M. J. Wall and N. Dale J Physiol 581.2 reversibly reduced parallel fibre (PF) EPSP amplitude by 67.8 ± 4.9% (n = 6, Fig. 6A) and also significantly reduced adenosine release by 66.5 ± 8% (n = 5, Fig. 6B). Parallel fibres are also modulated by mGluR4 receptors: application of the mGluR4 receptor agonist l-AP4 (50 μm) caused a reversible decrease in PF EPSP amplitude (41.3 ± 8%, n = 4, Fig. 6C ). Application of l-AP4 ( 50 μm) caused a significant (P < 0.05) reversible decrease in the amplitude of adenosine signals (44.2%, 60 ± 17.7 versus 33.5 ± 10.6 pA, n = 4, Fig. 6D) demonstrating that adenosine release is also modulated by mGluR4 receptors. The matching of the pharmacological profile of parallel fibre–Purkinje cell synaptic transmission with that of adenosine release strongly suggests that activation of parallel fibres is essential for the activity-dependent adenosine release and this release can be modulated by endogenous transmitters. Endogenous adenosine release modulates information flow in the cerebellum Parallel fibre to Purkinje cell synaptic transmission is inhibited by presynaptic A receptors (Kocsis et al. 1984; Rivkees et al. 1995) which raises two possibilities: firstly adenosine release from parallel fibres should be modulated by endogenous adenosine and secondly, the released adenosine could inhibit parallel fibre–Purkinje cell synaptic transmission and inhibit adenosine release (negative feedback). To test the first possibility, the effects of an A receptor antagonist on adenosine release were investigated. Block of A receptors (8CPT, 1 μm) caused an increase in PF EPSP amplitude (45.8 ± 10%, n = 10, Fig. 7A) demonstrating continual adenosine-mediated inhibition and the presence of an extracellular adenosine tone (Takahashi et al. 1995; Dittman & Regehr, 1996). Application of 8CPT (1 μm) also caused a significant increase in the amplitude of adenosine signals (89 ± 15%, n = 8, Fig. 7B) demonstrating that adenosine release is also modulated by the level of extracellular adenosine. The much larger effect of 8CPT on adenosine release compared Figure 6. Adenosine release is modulated by parallel fibre with PF EPSPs may stem from the block of adenosine receptor agonists auto-inhibition (released adenosine binds to A receptors Parallel fibre (PF) EPSPs were evoked every 10 s by a stimulating and inhibits its own release). We investigated the feasibility electrode placed on the surface of the molecular layer and recorded with an extracellular electrode. A, superimposed averages of EPSPs in of this by examining modulation of parallel fibre trans- control, in 25 μM baclofen and following wash. Baclofen reversibly mitter release by released adenosine. If the concentration reduced EPSP amplitude by ∼60%. B, superimposed traces from an of adenosine released reaches a sufficient magnitude, it adenosine biosensor in control, baclofen (25 μM) and in wash. should inhibit glutamate release from parallel fibres (via Adenosine release was evoked by a 10 s stimulus at the arrow. the activation of A receptors). PF EPSPs were evoked Baclofen reversibly reduced adenosine release by ∼50%. C, superimposed averages of EPSPs in control, in 50 μML-AP4 and every 10 s (to measure baseline transmission) and then following wash. L-AP4 reversibly reduced EPSP amplitude by ∼40%. a 10 s train (20 Hz) of stimuli was delivered to release D, superimposed traces from an adenosine biosensor placed on the adenosine. Following the train of stimuli, the amplitude of surface of the molecular layer in control, 50 μML-AP4 and in wash. PF EPSPs was diminished but then slowly recovered back Following L-AP4 application adenosine release was decreased by to the baseline (mean recovery time 122 ± 6.5 s, n = 6) ∼40%. Adenosine release was evoked by a 5 s stimulus (20 Hz) at the arrow. with a time course very similar to the decay of adenosine C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 561 release (decay ∼130 s, Fig. 7C and D). Application of 1 μm physiological stimulus: short duration trains (1–10 s), 8 CPT, to block the A receptors, significantly increased localized to a small area of the slice and at the same minimal the speed of recovery following the train (recovery time strength that evokes transmitter release. reduced from 122 ± 6.5 to 44 ± 8.7 s, Fig. 7C and D). Thus, parallel fibre-dependent adenosine release is functionally What is the source of adenosine? important and acts to auto inhibit both the parallel fibre–Purkinje cell synapse and indeed adenosine release We have provided strong evidence that parallel fibre itself. activity is required for adenosine release. Firstly, the stimulating electrode and biosensors were arranged along a beam of parallel fibres; thus, parallel fibres are activated Discussion by the stimulus and the sensors are in the correct place to We have demonstrated that adenosine can be released measure what is released. Secondly, activation or inhibition from the molecular layer of cerebellar slices by electrical of the G-protein coupled receptors (GABA ,A and B 1 stimulation. This release of adenosine was via a process mGluR4) present on parallel fibre terminals modulates 2+ that is both TTX and Ca sensitive and as ATP release adenosine release by a similar magnitude as synaptic cannot be detected, adenosine is either released directly transmission. This proportionality in the reduction of or rapidly produced by efficient extracellular ATP break- EPSP amplitude and adenosine release suggests a common down. Since adenosine release was modulated by receptors mechanism for adenosine and glutamate release. However, that act on parallel fibre–Purkinje cell synapses, parallel there are apparent differences. Our results suggest that fibres are the most likely source of adenosine. No previous the dependence of adenosine release on extracellular 2+ study has measured adenosine release from cerebellar Ca is ∼1 whereas previous studies have shown that 2+ slices, although adenosine can be released from cultured the Ca dependence of glutamate release at parallel granule cells (Schousboe et al. 1989; Philibert & Dutton, fibre synapses is ∼3 (Mintz et al. 1995; Brown et al. 2+ 1989; Sweeney, 1996). Prolonged electrical stimulation 2004). This difference in Ca dependence may be more (1–5 min) can release adenosine in other brain regions apparent than real as different stimuli were used to elicit (the cortex, hippocampus and striatum, Pedata et al. 1990; glutamate and adenosine release. Single stimuli were used Lloyd et al. 1993). However, we have characterized the to evoke glutamate release (Mintz et al. 1995; Brown properties of adenosine release evoked by a plausibly et al. 2004) whereas in this study, trains of stimuli were Figure 7. The adenosine released by electrical stimulation modulates parallel fibre–Purkinje cell synaptic transmission Parallel fibre (PF) EPSPs were evoked every 10 s by a stimulating electrode placed on the surface of the molecular layer and recorded with an extracellular electrode. A, graph plotting PF EPSP amplitude against time. Application of 1 μM 8CPT, to block A receptors, increased EPSP amplitude by ∼30%. B, superimposed traces from an adenosine biosensor in control and in the presence of 1 μM 8CPT. Adenosine release was evoked by a 7 s stimulus at the arrow. Application of 8CPT increased adenosine release by 43%. C, graph plotting PF EPSP (evoked every 10 s) amplitude against time. At the asterisk a train of stimuli was delivered (10 s, 20 Hz) to cause adenosine release. Following the train there was a marked reduction in PF EPSP amplitude followed by a slow recovery back to control amplitude (PF EPSP amplitude during the train is not plotted). The time course of PF EPSP amplitude recovery was ∼150 s which is very similar to the time course of adenosine release. Application of the A receptor antagonist 8CPT increased PF EPSP amplitude by ∼25% and also markedly speeded recovery following a train of stimuli (∼50 s). Thus, the slow component of recovery results from the released adenosine activating A receptors. D, graph summarizing data from 6 recordings. Following blockade of A receptors there was significant speeding of EPSP recovery. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 562 M. J. Wall and N. Dale J Physiol 581.2 2+ sensors were pushed into the molecular layer (presumably required to release adenosine. The Ca dynamics during the sensor will be closer to the release sites). At the calyx a train of stimuli will be more complex than for a single 2+ of Held, trains of stimuli (10 Hz) release adenosine, which stimulus and may result in intracellular Ca accumulation can be detected indirectly through the activation of A thus reducing the apparent dependence on extracellular 1 2+ receptors and the consequent inhibition of transmitter Ca . Thirdly, it is physiologically plausible that parallel fibres maintain transmitter release at frequencies up to release. This inhibition only reaches significance after and above the stimulation frequency used in this study the first 20 stimuli in a train, suggesting that multiple (Kreitzer & Regehr, 2000). Thus, parallel fibre-dependent stimuli are required to release adenosine (Kimura et al. adenosine release could occur physiologically and act to 2003; Wong et al. 2006). In the hippocampus, 10 Hz but not 3 Hz stimulation released adenosine (measured by limit parallel fibre-dependent excitation of Purkinje cells. inhibition of EPSCs, Brager & Thompson, 2003). Secondly, Finally, moving the biosensor further away (along the same once released the time course of adenosine was very beam of parallel fibres) resulted in only a small reduction slow. The decay of adenosine takes around 130 s which in current amplitude with little change in rise time. is vastly longer than that for classical fast transmitters Other possible sources of adenosine include Purkinje like glutamate. This is similar to neuropeptides, which cells, interneurones and glia. Stimulation in the molecular have a long duration of action and act in a paracrine layer will activate all these cell types but the TTX sensitivity manner affecting many neurones. This may reflect the large of adenosine release makes Bergmann glia an unlikely number of parallel fibres that are stimulated resulting in source of adenosine as they do not fire action potentials the spill over of adenosine and diffusion over a large area. (Clark & Barbour, 1997). Purkinje cells are strongly immunoreactive for adenosine (Braas et al. 1986) and express GABA receptors in their dendrites (Lujan & Shigemoto, 2006) but there is little evidence that Purkinje Mechanism of adenosine release? cells express either A adenosine receptors or mGluR4 receptors. The restricted expression of mGluR4 receptors The inhibition of ENT1 and ENT2 had no effect on also means adenosine release from molecular layer inter- adenosine release and thus adenosine was not transported neurones is unlikely (Mateos et al. 1998). We can also across the cell membrane by these equilibrative trans- exclude glutamate released from parallel fibres activating porters. NBTI and dipyridamole do inhibit adenosine receptors on neurones and glia, since adenosine release was transport leading to greater synaptic inhibition in the not blocked by the glutamate receptor antagonists CNQX cerebellum (M. J. Wall, personal observation) and in and AP5. Furthermore Purkinje cells, interneurones and the hippocampus (Frenguelli et al. 2007). However, it (possibly) Bergmann glia will be shunted by application is possible that other transporters, which are insensitive of the GABA receptor agonist muscimol, which had no A to NBTI/dipyridamole, are involved in adenosine release. effect on adenosine release. Thus, the most likely source Adenosine release was blocked by TTX and by removal of 2+ of adenosine is the parallel fibres. However, on several Ca . The calcium dependence of adenosine release was occasions it has been possible to stimulate parallel fibres lower than that reported for classical neurotransmitters, and record EPSPs without observing release of detectable but was similar to that reported for the release of neuro- amounts of adenosine. Thus, perhaps only a subset of peptides (for example see Peng & Zucker, 1993). The parallel fibres release adenosine. adenosine release thus has the characteristics associated with conventional synaptic release. Nevertheless there are three alternate interpretations of our data. Firstly, an inter- mediate transmitter could in principle cause the down- Properties of adenosine release stream release of adenosine. This seems unlikely as we have Some of the characteristics of adenosine release appear eliminated the obvious candidates for such an intermediate different from parallel fibre–Purkinje cell synaptic trans- transmitter and have shown that inhibiting neurones and mission. Firstly, a single stimulus is sufficient to evoke glia (through GABA receptor activation) has little effect glutamate release from parallel fibres, yet a train of such on adenosine release. Secondly, action potential activity in stimuli is required to produce an adenosine signal on the parallel fibres could release potassium which depolarizes biosensor. This could simply reflect that enough adenosine other cells (including glia) resulting in adenosine release. has to be released to diffuse through the tissue to reach the This seems unlikely as activation of parallel fibre receptors sensor on the slice surface. Clearly the distance between (GABA ,A and mGlu4R) inhibits adenosine release but B 1 synaptic glutamate receptors and the release sites is much would not affect the amount of potassium released during smaller than that between the adenosine biosensor and action potentials. Furthermore any cells depolarized by the release sites. Thus, the requirement for trains of stimuli potassium efflux would probably be shunted by GABA could be an artefact of the recording system. However, receptor activation suggesting that this treatment should similar stimulation protocols were required when the diminish release if that were the case. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 563 Thirdly, adenosine could arise from the breakdown of greater inhibition of parallel fibre–Purkinje cell synapses. exocytotically released ATP. Beierlein & Regehr (2006) Although the increased synaptic inhibition will reduce 2+ have recently described parallel fibre-mediated Ca rises the amplitude of EPSPs, the effects on high-frequency in Bergmann glia that are blocked by PPADs (50 μm) synaptic transmission will probably be more important. and may thus depend on parallel fibre-mediated ATP During high-frequency parallel fibre activity the increased release. However, we have been unable to measure any ATP synaptic inhibition will reduce glutamate depletion and release (either with or without ecto-ATPase inhibition). thus maintain transmission to Purkinje cells (Kimura et al. Thus, the ATP released from parallel fibres, as reported 2003; Wong et al. 2006). Activation of A receptors by endo- genous adenosine at the calyx of Held causes a reduction by Beierlein & Regehr (2006), may be too small to detect by our methods. This contrasts with studies in the retina in the relative amplitude of EPSCs at the end of the train where glial cells release ATP which is rapidly converted compared with the EPSCs at the beginning. This is because by ectoenzymes into adenosine. In this case, the ATP can there is no ambient level of adenosine at the calyx of be detected using luciferin–luciferase chemiluminescence Held and adenosine is released and accumulates during and blockers of ATP breakdown reduce the production the stimulation, reaching a significant concentration late of adenosine (reviewed by Newman, 2004). In the in the train. In contrast, Billups et al. (2005) have shown hippocampus, it has been reported that exogenous ATP is that activation of presynaptic group III mGluRs has no converted to adenosine in less than a second (Dunwiddie net effect on the amplitude of EPSCs. However, there is an et al. 1997). More recent experiments demonstrated that increase in the size of the readily releasable vesicle pool only a small proportion (∼7%) of bath-applied ATP is which is balanced by a reduction in the probability of converted, albeit rapidly to adenosine by hippocampal release. Thus, the synaptic state of the synapse has changed slices (Frenguelli et al. 2007). Our experiments suggest that and the metabolic demand has been reduced. We have not in the cerebellum, only 10–20% of ATP is rapidly converted measured the actions of adenosine during a train of EPSPs to adenosine. Furthermore biosensor measurements of but have investigated recovery from depression following the breakdown of bath-applied ATP show that less than a train. We found that blocking A receptors greatly speeds 10% of the applied ATP is converted to adenosine (Wall the recovery of EPSP depression. Similar observations MJ & Dale N, personal observations). For the complete have been observed following trains of stimuli at the conversion of parallel fibre ATP to occur, high densities of calyx of Held, where blocking group III mGluRs increased ectonucleotidases would have to be localized at parallel the speed of recovery (Billups et al. 2005) while GABA fibre synapses around the site of release such that no receptor activation slowed recovery (Sakaba & Neher, ATP escaped the confines of the synapse. Although 2003). The action of GABA receptorsappearstobe an ecto-ATPase (CD39) is present on the soma and through retardation of synaptic vesicle recruitment during dendrites of Purkinje cells (Wang & Guidotti, 1998) sustained activity. and 5 -nucleotidase activity is present on Bergmann glia The dynamics of adenosine signalling are likely to be membranes (Schoen et al. 1987), the density of these complex as release of adenosine auto-inhibits its own enzymes is unclear and this explanation seems to us release. An intriguing question is whether adenosine unlikely. A direct mechanism of adenosine release is released from active parallel fibres could laterally inhibit supported by Wong et al. (2006) who demonstrated that neighbouring but inactive parallel fibres. Interestingly trains of stimuli release adenosine at the calyx of Held adenosine is also vaso-active and there is evidence that (detectedbyA receptor mediated-inhibition of EPSCs) periods of activity release adenosine resulting in the with no detectable release of ATP (measured by means dilatation of cerebellar blood vessels (Li & Iadecola, 1994). of the luciferin–luciferase method). In the hippocampus, the adenosine released by trains of stimuli (detected by References A receptor-mediated inhibition of EPSCs) could not be reduced by blocking 5 -ectonucleotidases (Mitchell et al. Agarwal RP, Spector T & Parks RE (1977). Tight-binding 1993; Brager & Thompson, 2003). Frenguelli et al. (2007) inhibitors-IV. Inhibition of adenosine deaminases by various have also shown that adenosine release during ischaemia inhibitors. Biochem Pharmac 26, 359–367. Batchelor AM & Garthwaite J (1992). GABAB receptors in the does not arise from the extracellular metabolism of ATP. parallel fibre pathway of rat cerebellum. EurJNeurosci 4, 1059–1064. The functional significance BeierleinM&RegehrWG (2006). Brief bursts of parallel fiber activity trigger calcium signals in Bergmann glia. JNeurosci Irrespective of the detailed mechanism of adenosine 26, 6958–6967. release, its dependency on parallel fibre activity is Billups B, Graham BP, Wong AYC & Forsythe ID (2005). potentially very significant. During periods of high parallel Unmasking group III metabotropic glutamate autoreceptor fibre activity adenosine will be released resulting in an function at excitatory synapses in the rat CNS. J Physiol 565, increased extracellular concentration of adenosine and 885–896. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society 564 M. J. Wall and N. Dale J Physiol 581.2 Braas KM, Newby AC, Wilson VS & Snyder SH (1986). Kimura M, SaitohN&Takahashi T (2003). Adenosine A Adenosine-containing neurons in the brain localized by receptor-mediated presynaptic inhibition at the calyx of immunocytochemistry. JNeurosci 6, 1952–1961. Held of immature rats. J Physiol 553, 415–426. Brager DH & Thompson SM (2003). Activity-dependent Kocsis JD, Eng DL & Bhisitkul RB (1984). Adenosine selectively release of adenosine contributes to short-term depression at blocks parallel-fiber-mediated synaptic potentials in rat CA3-CA1 synapses in rat hippocampus. J Neurophysiol 89, cerebellar cortex. Proc Natl Acad SciUSA 81, 6531–6534. 22–26. Kreitzer AC & Regehr WG (2000). Modulation of transmission Brown SP, Safo PK & Regehr WG (2004). Endocannabinoids during trains at a cerebellar synapse. JNeurosci 20, inhibit transmission at granule cell to Purkinje cell synapses 1348–1357. by modulating three types of presynaptic calcium channels. LatiniS&Pedata F (2001). Adenosine in the central nervous JNeurosci 24, 5623–5631. system: release mechanisms and extracellular concentrations. Bultmann R, Trendelenburg M, Tuluc F, Wittenburg H & JNeurochem 79, 463–484. Starke K (1999). Concomitant blockade of P2X-receptors Li J & Iadecola C (1994). Nitric oxide and adenosine mediate and ecto-nucleotidases by P2-receptor antagonists: vasodilation during functional activation in cerebellar functional consequences in rat vas deferens. Naunyn cortex. Neuropharm 33, 1453–1461. Schmiedebers Arch Pharmacol 359, 339–344. Llaudet E, Botting NP, Crayston JA & Dale N (2003). A Clark BA & Barbour B (1997). Currents evoked in Bergmann three-enzyme microelectrode sensor for detecting purine glial cells by parallel fibre stimulation in rat cerebellar slices. release from central nervous system. Biosens Bioelectron 18, J Physiol 502, 335–350. 43–52. Crack BE, Pollard CE, Buekers MW, Roberts SM, Hunt SF, Llaudet E, Hatz S, Droniou M & Dale N (2005). Microelectrode Ingall AH, McKechnie KCW, Ijerzman AP & Leff P (1995). biosensor for real-time measurement of ATP in biological Pharmacological and biochemical analysis of FPL 67156, a tissue. Anal Chem 77, 3267–3273. novel, selective inhibitor of ecto-ATPase. Br J Pharm 114, Llinas RR & Sugimori M (1980). Electrophysiological 475–481. properties of in vitro Purkinje cell somata in mammalian Dale N (1998). Delayed production of adenosine underlies cerebellar slices. J Physiol 305, 171–195. temporal modulation of swimming in frog embryo. J Physiol Lloyd HGE, LindstomK&Fredholm BB (1993). Intracellular 511, 265–272. formation and release of adenosine from rat hippocampal Dale N, Pearson T & Frenguelli BG (2000). Direct measurement slices evoked by electrical stimulation or energy depletion. of adenosine release during hypoxia in the CA1 region of the Neurochem Int 23, 173–185. rat hippocampal slice. J Physiol 526, 143–155. Lujan R & Shigemoto R (2006). Localization of metabotropic Dittman JS & Regehr WG (1996). Contributions of GABA receptor subunits GABA and GABA relative to B1 B2 calcium-dependent and calcium-independent mechanisms synaptic sites in the rat developing cerebellum. EurJNeurosci to presynaptic inhibition at a cerebellar synapse. JNeurosci 23, 1479–1490. 16, 1623–1633. Mateos JM, Azkue J, Sarria R, Kuhn R, GrandesP&Knopfel T Dodge FA & Rahamimoff R (1967). Co-operative action a (1998). Localization of the mGlu4a metabotropic glutamate calcium ions in transmitter release at the neuromuscular receptor in rat cerebellar cortex. Histochem Cell Biol 109, junction. J Physiol 193, 419–432. 135–139. Dulla CG, Dobelis P, Pearson T, Frenguelli BG, Staley KJ & Mintz IM, Sabatini BL & Regehr WG (1995). Calcium control Masino SA (2005). Adenosine and ATP link P to cortical of transmitter release at a cerebellar synapse. Neuron 15, CO2 excitability via pH. Neuron 48, 1011–1023. 675–688. Dunwiddie TV, Diao L & Proctor WR (1997). Adenine Mitchell JB, Lupica CR & Dunwiddie TV (1993). Activity- nucleotides undergo rapid, quantitative conversion to dependent release of endogenous adenosine modulates adenosine in the extracellular space in rat hippocampus. synaptic responses in the rat hippocampus. JNeurosci 13, JNeurosci 17, 7673–7682. 3439–3447. Fredholm BB (1997). Adenosine and neuroprotection. Int Rev Neale SA, GarthwaiteJ&Batchelor AM (2001). Metabotropic Neurobiol 40, 259–280. glutamate receptor subtypes modulating neurotransmission Fredholm BB, Arslan G, Halldner L, Kull N, Shulte G & at parallel fibre-Purkinje cell synapses in rat cerebellum. Wasserman W (2000). Structure and function of adenosine Neuropharm 41, 42–49. receptors and their genes. Naunyn Schmeidebers Arch Newman EA (2004). Glial modulation of synaptic transmission Pharmacol 362, 364–374. in the retina. Glia 47, 268–274. Frenguelli BG, Wigmore G, Llaudet E & Dale N (2007). Noji T, Karaswa A & Kusaka H (2004). Adenosine uptake Temporal and mechanistic dissociation of ATP and inhibitors. Eur J Pharm 495, 1–16. adenosine release during ischemia in the mammalian Pedata F, Pazzagli M, TilliS&PepeuG (1990). Regional hippocampus. JNeurochem differences in the electrically stimulated release of DOI: 10.1111/j.1471-4159.2006.04425.x endogenous and radioactive adenosine and purine Geiger JD & Nagy JI (1986). Distribution of adenosine derivatives from rat brain slices. Naunyn Schmiedebers Arch deaminase activity in rat brain and spinal cord. JNeurosci 6, Pharmacol 342, 447–453. 2707–2714. Peng Y & Zucker RS (1993). Release of LHRH is linearly related 2+ Hamann M & Attwell D (1996). Non-synaptic release of ATP to the time integral of presynaptic Ca elevation above a by electrical stimulation in slices of rat hippocampus, threshold level in bullfrog sympathetic ganglia. Neuron 10, cerebellum and habenula. EurJNeurosci 8, 150–1515. 465–473. C  C 2007 The Authors. Journal compilation 2007 The Physiological Society J Physiol 581.2 Adenosine release in cerebellum 565 Philibert RA & Dutton GR (1989). Dihydropyridines modulate Takahashi M, Kovalchuk Y & Atwell D (1995). Pre- and K -evoked amino acid and adenosine release from cerebellar postsynaptic determinants of EPSC waveform at cerebellar neuronal cultures. Neurosci Lett 102, 97–102. climbing fiber and parallel fiber to Purkinje cell synapses. Reid CA, Bekkers JM & Clements JD (1998). N- and P/Q-type JNeurosci 15, 5693–5702. 2+ Ca channels mediate transmitter release with a similar Wall MJ & Usowicz MM (1997). Development of action cooperativity at rat hippocampal autapses. JNeurosci 18, potential-dependent and independent spontaneous GABA 2849–2855. receptor-mediated currents in granule cells of postnatal rat Riquelme R, Miralles CP & De Blas AL (2002). Bergmann glia cerebellum. Eur J Neurosci 9, 533–548. GABA receptors concentrate on the glial processes that Wang TF & Guidotti G (1998). Widespread expression of wrap inhibitory synapses. JNeurosci 22, 10720–10730. ecto-apyrase (CD39) in the central nervous system. Brain Res Rivkees SA, Price SL & Zhou FC (1995). Immunohistochemical 790, 318–321. detection of A1 adenosine receptors in rat brain with Wong AYC, Billups B, Johnston J, Evans RJ & Forsythe ID emphasis on localization in the hippocampal formation, (2006). Endogenous activation of adenosine A1 receptors, cerebral cortex, cerebellum, and basal ganglia. Brain Res 677, but not P2X receptors, during high-frequency synaptic 193–203. transmission at the calyx of Held. J Neurophysiol 95, Rudolphi KA, Schubert P, Parkinson FE & Fredholm BB 3336–3342. (1992). Neuroprotective role of adenosine in cerebral YuanY&Atchison WD (1999). Comparative effects of ischemia. Trends Pharm Sci 13, 439–445. methylmercury on parallel-fiber and climbing-fiber Safiulina VF, Kasyanov AM, GiniatullinR&Cherubini E responses of rat cerebellar slices. J Pharmacol Exp Ther 288, (2005). Adenosine down regulates giant depolarizing 1015–1025. potentials in the developing rat hippocampus by exerting a Zimmermann H (1996). Biochemistry, localization and negative control on glutamatergic inputs. J Neurophysiol 94, functional roles of ecto-nucleotidases in the nervous system. 2797–2804. Prog Neurobiol 49, 589–618. Sakaba T & Neher E (2003). Direct modulation of synaptic vesicle priming by GABA receptor activation at a glutamatergic synapse. Nature 424, 775–778. Schoen SW, Graeber MB, Reddington M & Kreutzberg GW (1987). Light and electron microscopical immunocytochemistry of 5 -nucleotidase in rat cerebellum. Histochem 87, 107–113. Schousboe A, Frandsen A & Drejer J (1989). Evidence for Acknowledgements evoked release of adenosine and glutamate from cultured We thank Dr Enrique Llaudet and Shakila Bibi of Sarissa cerebellar granule cells. Neurochem Res 14, 871–875. Biomedical Ltd for biosensor manufacture and Dr Robert Sweeney MI (1996). Adenosine release and uptake in cerebellar Eason and Matthew Simpson for assistance with the HPLC granule neurons both occur via an equilibrative nucleoside carrier that is modulated by G proteins. JNeurochem 67, experiments. This work was supported by the Wellcome Trust 81–88. (M.J.W. and N.D.). C  C 2007 The Authors. Journal compilation 2007 The Physiological Society

Journal

The Journal of PhysiologyPubmed Central

Published: Mar 8, 2007

There are no references for this article.