Department of Preclinical and Clinical Pharmacology, University of Florence, Viale Pieraccini 6, 50139 Florence, Italy. E-mail: email@example.com
The application of an ischaemic insult in hippocampal slices results in the depression of synaptic transmission, mainly attributed to the activation of A1 adenosine receptors by adenosine released in the extracellular space.
To estimate the concentration of endogenous adenosine acting at the receptor level during an ischaemic episode, we recorded field e.p.s.ps (fe.p.s.ps) from hippocampal slices, and evaluated the ability of the selective A1 receptor antagonist, 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), to reverse the fe.p.s.p. depression induced by in vitro ischaemia. A relationship between the IC50 of an antagonist and the endogenous concentration of a neurotransmitter has been used for pharmacological analysis.
The complete and reversible depression of fe.p.s.p. in the CA1 region induced by 5 min ischaemia was decreased in the presence of DPCPX (50–500 nM). 8-Phenyltheophylline (10 μM) abolished the depression of fe.p.s.ps during the ischaemic period, while a small (peak effect 12±4%) decrease in fe.p.s.ps was observed during the initial phase of reperfusion.
In the time-interval of maximal depression of fe.p.s.ps., IC50 and adenosine concentration changed as function of time with a good degree of correlation. The maximal value of adenosine concentration was 30 μM.
Our data provide an estimation of the adenosine concentration reached at the receptor level during an ischaemic episode, with a higher time discrimination (15 s) than that achieved with any biochemical approach. This estimation may be useful in order to establish appropriate concentrations of purinergic compounds to be tested for their pharmacological effects during an ischaemic episode.
In the Central Nervous System (CNS) adenosine is an important neuromodulator which exerts an inhibitory tonus on synaptic transmission, principally mediated by an inhibition of neurotransmitter release and by a reduction of postsynaptic excitability (Corradetti et al., 1984a; Dunwiddie, 1985; Proctor & Dunwiddie, 1987). This well established action of adenosine on neurotransmission is classically associated with the activation of the first (A1) of four adenosine receptor types which have been cloned and identified in the CNS: A1, A2A, A2B, A3 (Fredholm et al., 1994). In the brain region such as the hippocampus, under physiological conditions, adenosine A1 receptors are tonically exposed to extracellular endogenous adenosine, as demonstrated by excitatory effects of adenosine receptor antagonists on electrophysiological responsiveness (Corradetti et al., 1984b; Dunwiddie & Hoffer, 1980; Schubert, 1988; Wu & Saggau, 1994). On the other hand, an excitatory action of A2A adenosine receptors on synaptic responses has been demonstrated in the hippocampus by use of the selective agonist CGS 21680 (Cunha et al., 1994; Sebastião & Ribeiro, 1992), while A3 adenosine receptors may be responsible for a desensitization of A1 receptor-mediated responses (Dunwiddie et al., 1997). No information has been provided until now on the role of A2B adenosine receptors on synaptic transmission.
Brain extracellular concentration of adenosine can be increased by several stimuli, including electrical stimulation, hypoxia and ischaemia (Pedata et al., 1991; 1993; Phillis et al., 1987). It appears that in the hippocampus the application of ischaemic or hypoxic insults is associated with depressed synaptic transmission, mainly evoked by the activation of A1 adenosine receptors by endogenous adenosine released in the extracellular space. Non-selective antagonists of adenosine receptors, like 8-phenyltheophilline (8-PT), have been found to delay or to prevent the hypoxia- or ischaemia-induced depression of synaptic transmission in the hippocampus (Fowler, 1989; 1990; Pedata et al., 1993), and the selective A1 receptor antagonist, DPCPX, has been found to prevent the inhibition of synaptic responses evoked by hypoxia (Canhao et al., 1994; Lucchi et al., 1996). While an important neuroprotective role of A1 adenosine receptor activation against ischaemic damage has been clearly demonstrated (Domenici et al., 1996; Heron et al., 1994), very few reports describe the role of A2 and A3 adenosine receptor during cerebral ischaemia. Since A1, A2 and A3 adenosine receptors may be activated by different ranges of concentrations of endogenous adenosine (Fredholm et al., 1994), knowledge of adenosine concentrations reached in the extracellular space during an ischaemic event, is critical to evaluate the involvement of different adenosine receptors.
In a recent work, Barlow (1995) explicitly set out the relationship between the IC50 of an antagonist and the endogenous concentration of a neurotransmitter and suggested that by using an antagonist of known KB it is possible to estimate the concentration of an agonist of known KA and [A50] acting at receptor level for producing the recorded pharmacological effect. By using this relationship, in the present study we estimated the concentrations of adenosine acting at the receptor level at various times during an ischaemia-like episode in vitro. To monitor the effect of endogenously released adenosine during ischaemia we recorded fe.p.s.ps and we evaluated the ability of the selective A1 receptor antagonist, DPCPX, to reverse fe.p.s.p. depression induced by an ischaemic episode on hippocampal slices.
Preparation of hippocampal slices
Experiments were carried out on rat hippocampal slices, prepared as previously described (Corradetti et al., 1983). Charles River male Wistar rats, 150–200 g body weight, were killed by decapitation and their hippocampi rapidly removed and placed on ice-cold oxygenated (95% O2-5% CO2) artificial cerebral spinal fluid (aCSF) of the following composition (mM): NaCl 124, KCl 3.33, KH2PO4 1.25, MgSO4 2, CaCl2 2, NaHCO3 25 and D-glucose 10. Slices (400 μm thick) were cut by a McIlwain tissue chopper and kept in oxygenated aCSF for at least 1 h at room temperature. A single slice was then placed on a nylon mesh, completely submerged in a small chamber (0.5 ml) and superfused through a peristaltic pump with oxygenated aCSF (30–32°C) at a constant flow rate of 2 ml min−1. The treated solutions reached the preparation in 90 s and this delay was taken into account in our calculations. In vitro ischaemia-like conditions were applied by superfusing the slice for 5 min with aCSF without glucose and gassed with nitrogen (95% N2-5% CO2) (Pedata et al., 1993). At the end of the ischaemic period, the slice was again superfused with normal oxygenated aCSF. Each slice was exposed to two periods of ischaemia-like conditions with a time interval of 50 min.
Test pulses (80 μs, 0.06 Hz) were delivered through a bipolar nichrome electrode positioned in the stratum radiatum. Evoked extracellular potentials were recorded with glass microelectrodes (2–10 MΩ) filled with 3 M NaCl, placed in the CA1 region of the stratum radiatum. Responses were amplified (Neurolog NL 104, Digitimer Ltd), digitized (sample rate, 33.33 kHz), and stored on floppy disks for later analysis using pCLAMP 6 software facilities (Axon Instruments Inc.). Stimulus-response curves were obtained by gradual increases in stimulus strength. The test stimulus pulse was then adjusted to produce a field excitatory post synaptic potential (fe.p.s.p.) whose slope was 40–50% of the maximum and was kept constant throughout the experiment. The amplitude of fe.p.s.p. was routinely measured and expressed as the percentage of the average amplitude of the potentials measured during 10 min that preceded exposure of the hippocampal slices to ischaemic conditions. In some experiments both the amplitude and the initial slope of fe.p.s.p. were quantified, but since no appreciable differences were observed in the effect of drugs and of in vitro ischaemia, usually only the measure of the amplitude was expressed in figures.
The first step of our study was to estimate the concentration of endogenous adenosine in our preparations using an approach similar to that described by Dunwiddie & Diao (1994). This allowed us to compare our conditions with those of previously published studies, and permitted us to implement our calculations with the pharmacological parameters obtained in these works. In parallel we tested whether DPCPX was able to antagonize a high concentration of adenosine (20 μM) under conditions of substantial block of adenosine uptake and deamination. This permitted the choice of antagonist concentrations to be used to block the effects of adenosine released by the ischaemic episode. Finally, to estimate the concentrations of endogenous adenosine, data were analysed as suggested by Barlow (1995).
Collection of data and pharmacological analysis
To generate data for DPCPX concentration-response curves, the amplitude of fe.p.s.ps evoked by test stimuli was measured while slices were superfused with increasing concentrations of the antagonist using a cumulative protocol (unless otherwise stated). The per cent changes in amplitude of recorded potential were fitted to a hyperbolic function (equation 1):
where E is the per cent change in fe.p.s.p. amplitude produced by the antagonist at the concentration [B], Emax is the maximum change in response, [B50] is the concentration of the antagonist producing a half-maximum effect and n is the slope index. Non-linear regression fitting was carried out with Prism 2.0 (GraphPad) software facilities.
The maximum response achievable in the preparation was used to calculate the maximum fractional increase in the response. To estimate the concentration of endogenous adenosine acting on A1 receptors under control conditions, this value was introduced in equation 2 (Dunwiddie & Diao, 1994):
where [Aend] is the concentration of endogenous adenosine, FI is the fractional increase in the response produced by an antagonist in the presence of endogenous adenosine (e.g.: if DPCPX increases the control response by 16% FI=0.16), and H is the Hill slope of the concentration-response curve of the agonist. In our calculations we introduced a Hill slope of 1.52 as obtained in the work by Dunwiddie & Diao (1994).
The estimation of adenosine concentration during the ischaemic episode was based on the relationship among the degree of agonist stimulation, the concentration of an antagonist producing 50% inhibition of agonist stimulation, and the dose ratio as defined by the Gaddum-Schild equation. These relationships have been explicitly set out by Barlow (1995). In brief, the relationship between response, E, and the concentration of the agonist [A] is described by equation 3:
where Emax is the maximum response, [A50] is the agonist concentration producing a half-maximum response, and n is the slope index (Hill coefficient). In the presence of an antagonist producing a dose ratio of r, the response is described by equation 4:
which, resolved for the percentage of reduction (Z) defined by [(E−E′)/E]*100, can be reduced to equation 5:
If the antagonist is competitive, the Gaddum-Schild equation defines the dose ratio as (equation 6):
and allows the relation between percentage of inhibition, Z, and the antagonist concentration expressed as a fraction of the dissociation constant (KB), that is [B]/KB to be expressed by using equation 7:
Therefore when KB of the antagonist and [A50] are known, the equation can be resolved and the concentration of the agonist [A] calculated.
The assumptions and limitations in application of this equation are fully developed in Barlow (1995). In the context of our approach the use of equation 7 requires that the value of [B]/KB is >1 which justifies the use of the logistic expression. This can be assumed when the IC50 is >>KB. In our experiments the IC50 of the competitive antagonist DPCPX can be calculated and the assumption verified.
We calculated the IC50 of DPCPX using the logistic equation 1 adapted for inhibition curves.
Adenosine, 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), 8-phenyltheophylline (8-PT), dipyridamole (DPY), erythro-9-(2-hydroxy-3-nonyl)adenine hydrochloride (EHNA) and S-(4-nitrobenzyl)-6-thioinosine (NBTI) were purchased from Research Biochemicals International (Natick, MA, U.S.A.). Adenosine deaminase (ADA; 220 U mg−1) was from Boehringer Mannheim (Monza, Italy). DPCPX and NBTI were dissolved in DMSO and stock solutions were made to obtain concentrations of DMSO of 0.05 and 0.01% in aCSF, respectively. Experiments carried out in parallel for an unrelated project showed that this concentration of DMSO did not affect the depression induced by the second ischaemic episode. Dipyridamole was dissolved in ethanol and stock solution were made to obtain a final concentration of ethanol of 0.05% in aCSF. Stock solutions of 8-PT (10 mM) were made up in alkalyne ethanol.
All numerical data are expressed as the mean±s.e.mean. Statistical significance was determined by paired Student's t-test or by analysis of variance followed by the Fisher post hoc test.
The major task of the present work was to produce experimental data whose analysis could allow estimation of endogenous adenosine concentrations acting at the receptor level during an ischaemia-like episode in vitro. However, since we needed to introduce into our calculations some pharmacological parameters already available in the literature, we considered it necessary to evaluate whether our experimental conditions were similar enough to include previously published values in our analysis. In addition, we investigated whether the adenosine A1 receptor antagonist was able to substantially antagonize high concentrations of adenosine within a range of concentrations which insured maintenance of A1 receptor selectivity.
Effects of DPCPX on fe.p.s.ps in normoxic conditions
In a first series of experiments we tested the effects of DPCPX (10–500 nM) on the fe.p.s.p. evoked in normoxic conditions. DPCPX was allowed to equilibrate for at least 15 min, and a steady-state effect was considered to be reached when eight consecutive fe.p.s.ps (2 min) showed a reproducible, constant amplitude. As shown in Figure 1a and b, in one typical of five experiments, superfusion of slices with the antagonist elicited a concentration-dependent increase in fe.p.s.p. amplitude.
DPCPX also antagonized the effects of 20 μM adenosine applied in the presence of DPY (5 μM), NBTI (100 nM), and EHNA (5 μM), which substantially blocked the uptake and deamination of adenosine (Dunwiddie & Diao 1994). Figure 1c shows the time course of the changes in fe.p.s.p. amplitude in one typical experiment. The application of uptake and deaminase inhibitors significantly decreased fe.p.s.p. amplitude (−47.7±9.2% vs control, P<0.04, Student t-test, n=4), and subsequent application of adenosine 20 μM produced a further, almost complete, inhibition of fe.p.s.ps (−77.2±3.1% vs uptake and deaminase inhibitors effect, P<0.04; −90±1.9% vs control P<0.005, Student t-test, n=4). In the absence of uptake and deaminase inhibitors, 20 μM adenosine induced a 55.1±6.5% depression of fe.p.s.p. amplitude (data not shown, P<0.006, Student t-test, n=3). Superfusion of DPCPX (5–500 nM), still in the presence of 20 μM adenosine, elicited a concentration-dependent reappearance of fe.p.s.ps, which in the presence of 500 nM were increased over control (pre-treatment) values by 15.5±0.1% (Figure 1d; P<0.05; Student t-test n=4). Therefore, we showed that DPCPX was able to antagonize the effect of a high concentration of adenosine.
Effects of in vitro ischaemia on fe.p.s.ps
The application of ischaemia-like conditions for 5 min on hippocampal slices resulted in a progressive depression of synaptic potentials recorded in the CA1 region, with a characteristic time-course. As illustrated in Figure 2b, the depression of synaptic potentials started about 45 s after the beginning of the ischaemic period and reached complete inhibition of fe.p.s.ps after about 3 min. Immediately after starting reperfusion with the oxygenated aCSF solution, there was a transient reappearance of synaptic potentials, followed by complete inhibition of fe.p.s.p. amplitude for about 3 min. After this period, fe.p.s.p. progressively reappeared and reached a complete recovery of amplitude within 10 min from the beginning of reperfusion. The application of a second period of 5 min in vitro ischaemia, 50 min after the end of the first period, resulted in a similar depression of fe.p.s.p. amplitude, followed by a faster recovery of fe.p.s.ps in comparison with the first period, i.e. after 7–8 min of reperfusion. This difference was consistently reproducible and, though small, was statistically significant from the fifth min of reperfusion with normoxic aCSF (P<0.05, Student t-test) until complete recovery.
In each experiment a first ischaemic insult was applied to evaluate the responsiveness of slice preparation to in vitro ischaemia, and all the pharmacological evaluations were carried out on the second ischaemic episode.
Effects of DPCPX on fe.p.s.p. depression elicited by an ischaemic episode
The application of the selective A1 receptor antagonist DPCPX (50–500 nM) resulted in a concentration-dependent increase in fe.p.s.p. amplitude and slope, and in a significant reduction of the depression of fe.p.s.ps produced by ischaemia. Figure 3 summarizes the effects of 100 nM DPCPX on baseline fe.p.s.p. after recovery from the first ischaemic period and on the depression induced by the second ischaemic treatment. As shown in Figure 3a, 5 min of ischaemia completely and reversibly (see Figure 3b) depressed the fe.p.s.p. Subsequent application of DPCPX increased the field synaptic response and partially prevented the effects of ischaemia. The time-course of fe.p.s.p. modifications produced by ischaemia-like treatment in the absence and presence of 100 nM DPCPX is depicted in Figure 3b. The concentration-dependent effects of DPCPX on ischaemia-induced depression of fe.p.s.ps is shown at an expanded time scale in Figure 4a. Statistical significance of the antagonism of the ischaemia-induced depression produced by DPCPX at various times of the ischaemic treatment is shown in Figure 4b. DPCPX antagonized in a concentration-dependent manner the ischaemia-induced depression of fe.p.s.p. In the interval between the second min of ischaemia and the fourth min of recovery the changes were statistically significant for all concentrations tested.
Evaluation of the contribution of adenosine receptors to the fe.p.s.p. depression induced by in vitro ischaemia
A separate set of experiments was addressed to assess if, in our experimental conditions, the depression of excitatory neurotransmission observed during the ischaemic episode was due to the action of adenosine. Due to the poor solubility of DPCPX, the concentration could not be raised above 500 nM. We therefore tested whether the presence of adenosine deaminase (ADA) during the second ischaemic episode, by decreasing the concentration of adenosine at the receptor level, allowed for a full block of A1 receptors by 500 nM DPCPX. As shown in Figure 5a, the superfusion of ADA (4 U ml−1+DPCPX 500 nM) elicited a reproducible increase in the amplitude of the fe.p.s.ps (15±4%; n=7). When the second ischaemic episode was induced in the presence of DPCPX+ADA, the depression was completely antagonized for the first 3–4 min of ischaemia (Figure 5a). In the following 2–3 min (i.e. starting from the fifth min of ischaemia through the third min of reperfusion), the fe.p.s.ps were contaminated by dendritic and/or somatic spikes. This hampered precise measuring of fe.p.s.ps, but confirmed that the depressant effects of ischaemia on excitatory neurotransmission are mediated by adenosine. To further confirm this conclusion, we tested the effects of 8-phenyltheophylline (8-PT, 10 μM; n=5), a non-selective, potent adenosine antagonist on the fe.p.s.p. depression induced by ischaemia. As illustrated by Figure 5b, 8-PT increased the amplitude of the fe.p.s.ps (22±3%) and delayed any decrease in fe.p.s.ps for at least 4 min. In three out of five preparations, a decrease was found only after the fifth min of ischaemia (i.e. during the first 2 min of reperfusion: Figure 5b and c). The peak of the depression was observed within the first min of reperfusion (12±4%, n=5), and the values of residual depression observed in the interval between the fourth min of ischaemia and the second min of reperfusion were used for correcting our calculations of adenosine concentration (see below).
Pharmacological analysis and estimation of adenosine concentrations in normoxic conditions
To estimate the concentration of adenosine under control normoxic conditions we calculated the maximum fractional increase in fe.p.s.p. amplitude which was estimated by the effects of the A1 receptor antagonist DPCPX. As shown in Figure 1d the changes in fe.p.s.p. amplitude observed in the presence of DPCPX in normoxic aCSF were fitted using the logistic equation (equation 1) and led to estimation of a maximum fractional increase of about 25%. By applying equation 2, the calculated concentration of endogenous adenosine in control conditions was 240 nM; 95% C.L.: 225–255 nM (n=5). This value was in agreement with those (120–200 nM) found by Dunwiddie & Diao (1994) and that calculated from the experiments devoted to studying the effects of DPCPX during ischaemia (180 nM; 95% C.L.: 175–185 nM; see e.g. Figure 3b).
Pharmacological analysis and estimation of adenosine concentrations in ischaemic conditions
The depression of fe.p.s.ps which developed during in vitro ischaemia and reperfusion is temporally coincident with the increase in adenosine flowing out of the cells (Latini et al., 1998a). During the first minutes of ischaemia the changes in fe.p.s.p. amplitude produced by the ischaemic episode, and their modification by DPCPX, were too small to allow calculations of adenosine extracellular concentrations at any experimental times sampled by fe.p.s.p. recording. Therefore we chose the time interval between the last minutes of ischaemia and the beginning of reperfusion (from 3.25 min of ischaemia to 2.50 min of reperfusion), when maximal depression of fe.p.s.p. is reached and when the effects of DPCPX on the ischaemia-induced depression of fe.p.s.ps were measurable and permitted accurate calculations of adenosine concentration.
Since it is known that the IC50 of an antagonist depends on the concentration of the agonist, it was conceivable that the IC50 of DPCPX was changing as a function of time, according to the concentration of adenosine present at each time. We therefore tested whether our data could generate estimations of IC50 which were sensitive to the presumed changes in adenosine concentration which developed over time.
The depression of fe.p.s.ps (expressed as per cent of baseline values) in the absence and presence of DPCPX was fitted to a logistic equation and led to an IC50 estimation for DPCPX (e.g. Figure 6a) which changed over time with a good degree of correlation (Figure 7). The IC50 of DPCPX varied according to the initial degree of synaptic depression and the values were greater at times when the fe.p.s.ps were abolished in control ischaemia. For clarity of illustration only the curves obtained every 30 s in the 1–4 min interval of reperfusion are shown in Figure 6a. The slopes of the curves are constrained to the average value (0.68±0.05, n=7) obtained from fitting the same data with a variable slope. The curves obtained in the time interval 1.5–5 min of ischaemia are almost symmetrical and gave similar IC50 values (range: 53–353 nM; see also Figure 7). These results indicated that our protocol and measurements were sensitive enough to accurately follow the changes in the IC50 of DPCPX. In addition, at any time considered the IC50 of DPCPX was much higher than its KB (0.45–3.6 nM, see below). This allowed the application of the equation explicitly set out by Barlow (1995).
We therefore fitted, by using equation 7, the values of fe.p.s.p. depression induced by ischaemia in the control and in the presence of DPCPX. This equation permits estimation of the concentration of an agonist (whose A50 is known) from the IC50 of an antagonist of known KB. For our calculations we introduced the following parameters: A50 of adenosine 610 nM, KB of DPCPX 1.5 nM (−8.82). These values were taken from Dunwiddie & Diao (1994) and from Zocchi et al. (1996), respectively (see also Methods). Figure 6b illustrates examples of the curves obtained with this procedure for the same times shown in Figure 6a. The adenosine concentrations estimated by these calculations in the interval between the fifth min of ischaemia (peak effect) and the third min of recovery ranged between 30 and 7.5 μM and changed as a function of time (Figure 7). Similar results were obtained by introducing into the equations the values of KB for DPCPX measured in functional experiments in hippocampal slices (3.3 nM: −8.48 from Alzheimer et al., 1991; 0.45 nM: −9.35 from Sebastião et al., 1990). Using these alternative parameters, the ranges of adenosine concentrations in the time interval shown in Figure 7 were 20–5 and 45–10 μM respectively.
A residual depression of the fe.p.s.ps was still seen in the presence of 8-PT and although the concentration (10 μM) of the antagonist did not insure full occupancy of the A1 receptors, the participation of other neurotransmitters or mechanisms in the transient depression observed could not be discarded (see Discussion). We corrected the calculations by subtracting the residual depression observed in 8-PT (during the interval between the fourth min of ischaemia and the second min of reperfusion) from the values implemented in Barlow's equation. In the interval considered, the resulting concentrations of adenosine were accordingly lower than those in Figure 7 and peaked at 30 s reperfusion with 23 μM.
Our results provide the first estimation of the concentration of adenosine acting at the receptor level during an ischaemia-like episode in the CNS.
We monitored the effects of an ischaemia-like episode on the amplitude of extracellularly recorded fe.p.s.ps, which mainly reflect the currents generated by the inflow of cations into CA1 pyramidal cell dendrites due to activation of glutamate receptors (Hubbard et al., 1969). The reduction of the fe.p.s.p. amplitude can be produced through several mechanisms whose principles are: (i) competitive or non competitive antagonism of glutamate receptors; (ii) depolarization of the postsynaptic cell elicited by excitatory receptor stimulation or by potassium ion accumulation in the extracellular space and subsequent spreading depression; (iii) decrease in glutamate release by actions at presynaptic terminals. Among these possibilities the most likely mechanism of fe.p.s.p. reduction during ischaemia is a decrease in glutamate release caused by activation of presynaptic A1 receptors (Corradetti et al., 1984a). In addition, it has been shown that the postsynaptic action of adenosine does not affect the size of miniature postsynaptic excitatory currents recorded from CA1 pyramidal cells (Prince & Stevens, 1992; Scholz & Miller, 1992), which suggests that the main effect of A1 receptor stimulation on fe.p.s.p. is a decrease in glutamate release.
Other mechanisms not involving adenosine receptors might participate in the decrease in fe.p.s.ps in our experiments. For instance, ATP-dependent potassium channels are known to open during a hypoxic and/or hypoglycaemic episode (Ben-Ari, 1989; Mourre et al., 1989; Tromba et al., 1992). However it is generally accepted that synaptic depression during ischaemia is due to A1 adenosine receptor stimulation (Fowler, 1990), and since the presence of the adenosine receptor antagonist 8-phenyltheophylline (8-PT) fully prevents the disappearance of fe.p.s.p. during an ischaemic episode (Pedata et al., 1993), it is conceivable that additional non adenosine-mediated mechanisms represent a negligible source of error in our conditions.
This assumption was confirmed by our experiments performed in the presence of DPCPX+ADA or 8-PT alone where the effect of the ischaemia was blocked for as long as 5 min (i.e. the whole duration of the ischaemic episode). However, at the end of the 5 min ischaemic period and during the first 2–3 min of reperfusion, a depression was still observed in the presence of any of the treatments. In addition, the treatment with DPCPX+ADA produced, mainly during reperfusion, a contamination of the fe.p.s.p. by dendritic and/or somatic spikes. It is known that the ischaemic insult increases spontaneous release of glutamate and that A1 receptor stimulation can greatly limit the outflow of glutamate (Heron et al., 1994). On the other hand, once the inhibitory effect of adenosine is completely blocked, glutamate can freely flow out of cells and, by stimulating ionotropic and metabotropic receptors, depolarize pyramidal neurones and increase their excitability. This produces a decrease in the amplitude of the fe.p.s.ps and the appearance of multiple action potentials. Furthermore, ADA might enhance neurone excitability through unspecific effects, as suggested by the modest, if any, hyperexcitation observed during ischaemia when the adenosine receptors are blocked by 8-PT. On the basis of this interpretation, we considered the experiments with 8-PT more reliable for correcting our calculations for the participation of unidentified mechanism(s) of fe.p.s.p. depression during reperfusion. However, an alternative explanation should be taken into account based on the fact that 8-PT is a competitive antagonist. From our calculations, the concentrations of adenosine at the end of 5 min ischaemia approached 20 μM and, considering the affinity of 8-PT for A1 receptors (Ki=86 nM) relative to that of adenosine (EC50=610 nM) the fraction of A1 receptors still occupied by the agonist (calculated with Gaddum's equation; see Kenakin (1993)) is not negligible (about 20%). Under our conditions of competitive antagonism, it is likely that the residual depression (10–15%) observed during reperfusion was still due to A1 receptor stimulation.
Since A2A adenosine antagonists, (e.g. ZM 241385), do not modify the depression of fe.p.s.p. induced by the ischaemic episode (Latini et al., 1998b; Latini et al., manuscript in preparation) and since 8-PT does not antagonize A3 receptors (Fredholm et al., 1994), A2A and A3 receptors do not appear directly involved in the depression of fe.p.s.p. induced by ischaemia. The role of A2B receptors, which are uniformly distributed in the CNS (Dixon et al., 1996), remains to be elucidated.
In our experiments, we found that the recovery of the fe.p.s.ps after the second ischaemic episode was slightly faster than after the first episode. This suggests a small but statistically significant ‘preconditioning effect’ by the first ischaemic episode, possibly due to desensitization of the A1 receptors. It has been shown that prolonged (>20 min) stimulation of A3 receptors by mM concentrations of adenosine blocks A1 receptor-mediated responses in the CA1 region of the hippocampus (Dunwiddie et al., 1997). Such a mechanism is unlikely to occur in our conditions because the first ischaemic episode lasted only 5 min and it is difficult to envisage that adenosine reached mM concentrations. The stimulation of A2A adenosine receptors also might desensitize A1 receptors in the CA1 (Cunha et al., 1994; O'Kane & Stone, 1998). However, O'Kane & Stone (1998) showed that stimulation of A2A receptors does not modify A1 receptor-mediated inhibitory action on fe.p.s.ps, and convincingly demonstrated that the possible interaction between A2A and A1 receptors is selectively postsynaptic in the CA1 region. We favour an alternative interpretation based on measurements of endogenous adenosine flowing out of hippocampal slices during two ischaemic episodes in similar (though not identical) experimental conditions. We found a 36±8% (P<0.05; n=7) reduction in the total amount of adenosine released over basal levels during the second ischaemic insult when compared to the first. Consistent with the present findings, the time-course of the outflow of adenosine after the second ischaemic insult was faster than after the first (unpublished observations). These results suggest a reduction in adenosine availability after the first ischaemic episode.
Our approach is based on the assumption that to decrease fe.p.s.ps, adenosine acted on a single class of receptors with identical properties, an assumption which has been supported by previous work (Alzheimer et al., 1991; Dunwiddie et al., 1986; Dunwiddie & Diao, 1994; Dunwiddie & Fredholm, 1989; Scholz & Miller, 1992; Wu & Saggau, 1994). For our calculations we introduced several parameters which were obtained from previous work (Alzheimer et al., 1991; Dunwiddie & Diao, 1994) conducted in hippocampal slices by using methods similar to those applied in the present research. Our estimation of the concentrations of adenosine is based on the analysis proposed by Barlow (1995); (see Methods). On the basis of this analysis the calculated concentrations of adenosine represent the degree of receptor stimulation. From our data, the most conservative estimation of adenosine concentrations at the peak, after the ischaemic episode, at the receptor level is between 20 and 30 μM, depending on the parameters used for estimation and/or on the correction of raw data for the residual depressant effects observed in the presence of 8-PT. The rise in adenosine concentration was time-locked to the ischaemic episode, reaching its maximum within 45 s after reintroduction of oxygenated control aCSF and rapidly declining thereafter. These results confirm the observation by Latini et al. (1998a) that an increase in adenosine efflux from hippocampal slices closely parallels in time the depression of fe.p.s.ps produced by an ischaemic episode similar to that used in this study.
Under normoxic conditions the adenosine extracellular concentration estimated in the brain by microdialysis 24 h after fibre implantation (40–120 nM, values corrected for recovery of the fibre) (Ballarin et al., 1991; Pazzagli et al., 1993; 1994; 1995) or by the cortical cup technique (30–50 nM) (Phillis, 1989) is in about the same concentration range as that calculated in the present work (180–240 nM) and by Dunwiddie & Diao (1994) (120–200 nM) who used the same preparation. This supports the interpretation that in normoxic conditions the released adenosine equilibrates in the extracellular fluid and therefore the degree of tonic stimulation at its receptors can be inferred from the measure of adenosine concentration in the CSF using the microdialysis technique.
After the in vitro ischaemic episode, adenosine increases about 100–150 fold, reaching an estimated value of 30 μM. This increase is much higher than that calculated in the fluid collected from hippocampal slices (4–6 fold) (Latini et al., 1998a; Pedata et al., 1993) or from the hippocampus in vivo (19–23 fold) (Andinè et al., 1990; Dux et al., 1990) after inducing ischaemia. However, it should be noted that the time discrimination of adenosine concentrations (15 s) allowed by the present approach is greater than that achieved with any biochemical approach, where the sampling periods are more likely to represent the equilibrium reached in the CSF after partial reuptake and degradation of released adenosine. Taking into consideration these observations, it is likely that the adenosine concentration of 30 μM estimated in our study represents an adenosine concentration acting on receptors before diffusion and equilibration in the CSF. Therefore the degrees of receptor stimulation is greater than that extrapolated by the microdialysis adenosine measurement after ischaemia. It deserves mention that adenosine concentrations as high as 24–40 μM have been measured by microdialysis after induction of global ischaemia in the rat and gerbil (Dux et al., 1990; Hagberg et al., 1987). However these values were reached on the first day after microdialysis tube implantation when the estimated basal adenosine extracellular concentration is 20 times higher than those found in experiments conducted 24 h after surgery (Pazzagli et al., 1993). The effect of ischaemia immediately after tube implantation of the probe may be therefore compounded by that of trauma due to the microdialysis probe insertion.
In agreement with the view that the presence of adenosine on receptors, close to release sites, is functionally important, Brundege & Dunwiddie (1996) have demonstrated, by applying adenosine by the recording pipette, that the degree of synaptic depression in the CA1 hippocampus is related to the adenosine concentration in the electrode. Their data strongly suggest that the principal source of adenosine, acting at the presynaptic level to decrease the excitatory input on the recorded CA1 pyramidal neurone, is the outflow of adenosine from the cell from which the postsynaptic current was recorded. This implies that during the ischaemic episode the outflow of adenosine from CA1 pyramidal neurones is likely to greatly affect excitatory neurotransmission.
In the CNS adenosine may act through four receptor subtypes to which the endogenous ligand shows varying affinity: a higher affinity for A1 and A2A (1–30 nM) than for A2B and A3 (>1 μM) receptors (Fredholm et al., 1994). Under normoxic conditions the adenosine present at the receptor level is therefore likely to preferentially activate A1 and A2A receptors. On the other hand, if we assume that our estimation of adenosine concentration reached during an ischaemia-like episode reflects the order of magnitude of adenosine present at the receptor level during an in vivo ischaemic episode, it is conceivable that under these conditions all adenosine receptors subtypes are activated during in vivo ischaemia. Stimulation of at least A1, A2A and A3 adenosine receptors may modify the outcome of an ischaemic episode. It appears that during in vivo ischaemia A1 receptor stimulation protects cerebral tissue from damage (Rudolphi et al., 1992) whereas the role of A2A (O'Regan et al., 1992; Phillis, 1989) and/or A3 adenosine receptors remains to be clarified (Von Lubitz et al., 1994).
In conclusion our data provide a ‘dynamic’ estimation of the concentration of adenosine reached at the receptor level during a transient ischaemic episode. This information may be useful in designing appropriate concentrations of purinergic compounds to be tested for their pharmacological effects during in vitro and/or in vivo ischaemic episodes.
This investigation was supported by grants from the University of Florence and MURST (60%).