RAPID TEMPERATURE CHANGES INDUCEADENOSINE-MEDIATED DEPRESSION OF SYNAPTIC

TRANSMISSION IN HIPPOCAMPAL SLICES FROM RATS(NON-HIBERNATORS) BUT NOT IN SLICES FROM

GOLDEN HAMSTERS (HIBERNATORS)

A. GABRIEL, F. W. KLUßMANN and P. IGELMUND*Institut fur Neurophysiologie der Universitat zu Koln, Robert-Koch-Str. 39, D-50931 Koln, Germany

Abstract––Disturbances in neuronal communication induced by rapid temperature changes are a risk inthe context of accidental hypothermia and would be fatal for hibernators during arousal from hibernation.Therefore, we investigated the effects of rapid temperature changes on synaptically induced CA1population spikes in hippocampal slices from golden hamsters (hibernators) and rats (non-hibernators).Temperature was changed ramp-like by 0.3)C/min, which corresponds to the rise of body temperature ingolden hamsters during arousal from hibernation. During cooling from 35 to 10–15)C, the populationspike amplitude increased, reached maximal values at 25–30)C and 20–25)C in hamster and rat slices,respectively, and then decreased with further cooling. During rewarming, hamster slices displayed thesame temperature dependence as during cooling. In contrast, in rat slices dynamic effects of thetemperature change occurred. These were most obvious in a strong depression of the spike amplitudeduring rewarming as compared to cooling. Above 26–29)C, the depression was superimposed by anexcitatory effect. The depression was largely attenuated by theophylline (100–200 µM) and thus seems tobe based on an increase of the concentration of endogenous adenosine, which in turn may result from animbalance in energy metabolism during warming.

The lack of warming-related depression in hamster slices can be explained by a lower sensitivity foradenosine as compared to rat slices. In addition, a better resistance of metabolic balance against rapidtemperature changes may prevent large elevations of endogenous adenosine in the hamster hippocampus.For hibernators, the avoidance of temperature change-induced disturbances of neuronal communicationmay be a prerequisite for safe arousal from hibernation. ? 1998 IBRO. Published by Elsevier Science Ltd.

Key words: hippocampus, synaptic transmission, temperature change, adenosine, rat, hamster.

Synaptic transmission depends on a multitude ofprocesses with a variety of different temperaturecharacteristics. Thus, temperature changes may easilycause imbalances in neuronal function and therebydisturb neuronal communication. In humans, suchdisturbances are of importance in the context ofhyperthermia, as well as clinical and accidental hypo-thermia. In animal experiments with artificial generalhypothermia, transient hyperresponsiveness, some-times leading to epileptic seizures, has been observedduring cooling as well as during rewarming.26 Inaddition, transient effects of local temperaturechanges on synaptic transmission have beenobserved in spinal motoneurons in vivo,21 as well asin hippocampal neurons in vitro.35

In contrast to other mammals, hibernatorsperiodically reduce their body temperature to valuesslightly above 0)C for several days or weeks andspontaneously rewarm to the euthermic state. In

golden hamsters, the heating velocity during arousalreaches up to 0.5)C/min. Thus, these animals physio-logically undergo large and rapid changes of theirbody temperature. In contrast to rewarming fromhypothermia, epileptiform activity does not seem tooccur during arousal from hibernation.26 To surviveentrance into and arousal from hibernation, thenervous system of hibernators also seems to beable to provide ordered cell function during rapidtemperature changes. The higher susceptibility ofnon-hibernators, as compared to hibernators, totemperature change-induced neuronal misfunctionsmay, on the one hand, be an indirect consequenceof disorders in other organs concerning, for example,oxygen and energy supply. On the other hand, itmay result directly from differences in neuronalfunction.

To address this topic, we investigated the effects ofrapid temperature changes in isolated brain tissue.Stimulus-induced synaptic transmission was investi-gated as a function of ramp-like temperature changes(0.3)C/min) in area CA1 of hippocampal slicesprepared from golden hamsters and rats.

*To whom correspondence should be addressed.Abbreviations: ACSF, artificial cerebrospinal fluid; FPSP,

field postsynaptic potential; PDE, phosphodiesterase.

Pergamon

Neuroscience Vol. 86, No. 1, pp. 67–77, 1998Copyright ? 1998 IBRO. Published by Elsevier Science Ltd

Printed in Great Britain. All rights reserved0306–4522/98 $19.00+0.00PII: S0306-4522(98)00011-6

67

Preliminary results of this study have beenpresented previously.9,10,18

EXPERIMENTAL PROCEDURES

The experiments were performed on hippocampal slicesprepared from Wistar rats (n=43 slices) and goldenhamsters (Mesocricetus auratus, n=21 slices). Hamsterswere obtained from the Zentrale Versuchstieranstalt(Hannover, Germany). They were housed in individualcages in a long-day photoperiod (17 h light, 7 h dark) at22&2)C (warm-acclimated hamsters). Six hamsters weretransposed to a short photoperiod (7 h light, 17 h dark)and cold (7&2)C) to induce hibernation, as describedpreviously.17 Rats were taken from our institute’s breed.They were kept in cages with up to six animals at 22&2)Cwith a 12-h light, 12-h dark photoperiod. All animals wereprovided with food and water ad libitum.

Rats and warm-acclimated hamsters were decapitatedunder ether anaesthesia, hibernating hamsters in deephibernation. The brain was quickly removed and chilled inice-cold artificial cerebrospinal fluid (ACSF). The hippo-campus was isolated and transverse slices (400 µm) wereprepared with a McIlwain tissue chopper and immediatelytransferred into a recording chamber, where they werecontinuously superfused (1.6 ml/min) with an aerated (95%O2/5% CO2) ACSF containing (in mM): NaCl 124,NaH2PO4 1.25, NaHCO3 26, KCl 3, CaCl2 1.6, MgSO4 1.8,glucose 10. To allow recovery from the preparation, sliceswere incubated at 32)C for at least 1 h prior to a recordingsession. The temperature of the chamber was varied by acomputer-controlled heating device (developed in our insti-tute on the basis of wiring diagrams obtained from HansBraun, Marburg, Germany). The slice temperature wasmeasured with thermocouples (diameter 200 µm) in thesuperfusing ACSF next to the slices. Temperature waschanged ramp-like with a slope of 0.3)C/min. In some of theexperiments, the temperature was continuously changedfrom 35)C to the temperature below which transmission wasblocked and vice versa. In others, the ramp was paused for15 min at various temperatures, or the direction of tempera-ture change was repetitively inverted after short periods ofcooling or warming by 5 or 10)C (a zig-zag ramp). In thepharmacological experiments, two temperature cycles, acontrol and a test cycle, were successively conducted oneach slice. To prevent problems due to the long duration ofthe experiments, the ramps were limited in these exper-iments to the range of 17–30)C. After the control cycle, thesubstance under investigation was continuously appliedwith the ACSF and equilibration was allowed at constanttemperature for at least 30 min before starting the test cycle.

For the pharmacological investigation, the following sub-stances were used: adenosine (Serva), ù-agatoxin IVA(Pfizer), bicuculline methiodide (RBI), CPG55845A (CibaGeigy), norepinephrine bitartrate (RBI) and theophylline(Sigma).

Extracellular field potentials were recorded with micro-electrodes positioned in the stratum pyramidale of areaCA1. Pyramidal cells were synaptically activated by stimu-lation of the Schaffer collateral/commissural fibre bundle(pulse duration 100 µs, interval 30 s) via bipolar glass-insulated platinum wire electrodes inserted into the stratumradiatum. In some experiments, a second stimulus electrodewas positioned in the alveus for antidromic stimulation ofpyramidal cells, and synaptic and antidromic stimulationwere applied alternately. The stimulus intensity was set toevoke a population spike of approximately half-maximalamplitude at the beginning of an experiment. In pharmaco-logical experiments, the stimulus amplitude was readjustedto half-maximal before starting the test cycle. Data wereconventionally amplified, digitized with a CED 1401 inter-face [Cambridge Electronic Design (CED), Cambridge,

U.K.] and recorded on a personal computer with theprogram SIGAVG (CED). For analysis, they were trans-ferred to the program SPIKE2 (CED), and the populationspike amplitude was calculated semi-automatically with aself-written macro script.

Calculation of averaged curves (temperature vs popula-tion spike amplitude; Figs 1B, D, 5A, B) was performed intwo steps. First, the individual curves were smoothed byaveraging the amplitudes of six to seven population spikeselicited in the range of &0.5)C around each full )C. Then,these values were averaged across the different experiments.Averaged data are indicated as mean&S.E.M.

RESULTS

In hippocampal slices, electrical stimulation of theSchaffer collaterals/commissural fibres synapticallyinduces action potentials in the CA1 pyramidal cells,which can be recorded extracellularly as populationaction potentials (population spikes; see Fig. 2). Ithas been described previously that, with cooling,starting at 37)C, the population spikes induced withconstant stimulus intensity first increase, reachmaximal values in the temperature range of 25–30)C,then decrease with further cooling, and finallyvanish.14,16,17,43 The durations of the afferent volley,of the field postsynaptic potential (FPSP) and of thepopulation spike increase continuously with cooling,whereas the slope of the FPSP decreases con-tinuously. The effects of cooling are reversible. Thetemperature below which synaptic transmission isblocked (the ‘‘threshold temperature’’) stronglydepends on stimulus intensity. In previous studiesfrom our laboratory, population spikes induced withhigh stimulus intensities could be recorded down totemperatures of 14)C and 16–18)C in rat and hamsterslices, respectively, while with low intensities, thespikes could already vanish at temperatures above20)C.15–17

In the present study, the temperature was changedramp-like (0.3)C/min) and the stimulus intensitywas generally adjusted to evoke approximately half-maximal population spikes at the starting tempera-ture (30–35)C). In this condition, population spikescould be evoked down to 12–15)C and 14–19)C in ratand hamster slices, respectively. These values arelower than those measured previously.17 The differ-ence is, at least in major part, based on the differencesin the ionic composition of the ACSF used in thesestudies.16

The temperature dependence of the populationspikes in hamster slices measured in this study isshown in Fig. 1A and B. It agrees well with theprevious reports. With cooling, the spike amplitudeincreased, reached maximal values between 25 and30)C, and decreased to zero with further cooling.During rewarming, similar amplitude values wereobtained at the corresponding temperatures, result-ing in a mirror image of the time-course as shown inFig. 1A and an overlap of the cooling and rewarmingcurves as shown in Fig. 1B. Such behaviour wasobtained in all of 21 hamster slices, with no difference

68 A. Gabriel et al.

in this respect between slices from warm-acclimated hamsters (n=12) and slices from hiber-nating hamsters (n=9). In different experiments, the

threshold temperature varied between 14 and 19)Cdue to interindividual variations and due to slightdifferences in the stimulus intensity. As a conse-quence, averaging all experiments would lead totremendous error bars in the lower temperaturerange, which would mask the information obtainedin the individual experiments. Therefore, Fig. 1Bcontains only experiments in which transmission wasblocked at the same temperature, namely 15)C (n=4).

In rat slices, in contrast, the dependence of thepopulation spike amplitude on temperature wasmarkedly different during cooling and rewarming(Fig. 1C, D). As in hamster slices, continuous coolinginduced first an increase and then a decrease of thespike amplitude. Maximal amplitudes were, however,obtained at lower temperatures (19–22)C) and thespikes vanished at lower temperatures (12–15)C) ascompared to hamster slices. During rewarming, thepopulation spikes reappeared, but their amplitudewas drastically reduced as compared to the valuesduring cooling (Fig. 1C, D). The spikes increasedup to a temperature of 19–22)C, then consistentlydecreased until 26–29)C and finally increasedagain. Average data showing the population spike

Fig. 1. Synaptic transmission during ramp-like temperature change. Population spike (PS) amplitude inhippocampal slices from warm-acclimated hamsters (A, B) and rats (C, D) as a function of time (A, C) andas a function of temperature (B, D). In the experiments shown in A and C, temperature was lowered witha slope of 0.3)C/min until the spikes disappeared, then raised with the same velocity, and finally lowereda second time. The temperature ramp is indicated and the rewarming phase is highlighted by the dashedarea. In B and D, the cooling (thin line) and rewarming (thick line) phases are indicated by correspondingarrows. (A, C) Single experiments. (B, D) Mean values&S.E.M. of the relative population spike amplitude

(percentage of the maximal amplitude), n=4 and n=6 slices, respectively.

Fig. 2. Field potential recordings from a rat hippocampalslice at 25)C during cooling (thin line) and rewarming (thickline), respectively. The stimulus artefact (SA) is followed bythe afferent volley (AV; representing the population actionpotential of the stimulated Schaffer collaterals/commissuralfibres) and the positive-going field postsynaptic potential(FPSP) superimposed by the negative-going populationspike (PS; representing the population action potential of

the pyramidal cells).

Effects of temperature on synaptic transmission 69

amplitude as a function of temperature are shown inFig. 1D. For the same reasons as in Fig. 1B, Fig. 1Dcontains only experiments in which transmission wasblocked at a certain temperature, 13)C (n=6). Asshown in Fig. 1C, a second cooling phase followingthe rewarming phase abolished the depression andevoked the same response as the first cooling phase.

At 25)C, the spike amplitude in rat slices duringrewarming measured 42&3% of the amplitude dur-ing cooling (n=35). The depression of the spikeamplitude during rewarming as compared to coolingwas associated with a depression of the slope of theFPSP and no change in the afferent volley (Fig. 2).This finding was confirmed by simultanous record-ings with a second electrode positioned in the stratumradiatum (n=8; not shown).

In four rat slices and one hamster slice, anadditional stimulus electrode was placed in the alveusfor antidromic stimulation of the pyramidal cells, andorthodromic and antidromic stimuli were deliveredalternately at 30-s intervals. In contrast to the ortho-dromically induced responses in rat slices, the anti-dromically induced population spikes were notdepressed during rewarming and the spike amplitudewas maximal at 27–34)C (not shown).

Duration of the temperature change-induced effects

To investigate whether the warming-relateddepression in rat slices is a direction-specific short-term effect or a long-lasting hysteresis effect,we applied intermittent temperature ramps with15-min halt phases. In hamster slices (n=2), theramp-induced modulation of the population spikeamplitude stopped shortly after halt of the tempera-ture ramp or continued slowly in the previous direc-tion (Fig. 3A, B). In rat slices (n=8), in contrast,great changes of the spike amplitude occurred duringthe halt periods in the warming phase (Fig. 3C, D). Indetail, as shown in Fig. 3C and D, halting the rampat 27)C during cooling had virtually no effect,whereas halting the ramp at the same temperatureduring rewarming was accompanied by a strongincrease of the spike amplitude near to the valuesobtained during cooling. At 22)C, halt during coolinginduced a decrease and halt during rewarminginduced an increase of the spike amplitude.Generally, the changes were in the direction of asteady-state which resembled the average tempera-ture dependence measured for hamster slices(Fig. 1B), with maximal spike amplitudes at 25–30)C.

Fig. 3. Synaptic transmission during intermittent ramp-like temperature change. Population spike (PS)amplitude in hippocampal slices from a warm-acclimated hamster (A, B) and a rat (C, D) as a function oftime (A, C) and as a function of temperature (B, D). In A and C, the time-course of the temperature isindicated and the halt phases are highlighted by the dashed areas. In B and D, showing the same data asin A and C, the cooling (thin line) and rewarming (thick line) phases are indicated by corresponding

arrows.

70 A. Gabriel et al.

The changes started immediately with halt of thetemperature ramp and came near to, but did notreach, an equilibrium within the 15-min pause. Withresumption of the ramp-like temperature change, thedynamic effects on the spike amplitude reappearedimmediately and developed within 10–15 min (seeFig. 3C and D, increase and depression duringcooling and rewarming, respectively, after halt at22)C). Note that, during warming from 27 to 35)Cafter the pause at 27)C, the population spikeamplitude was larger at corresponding temperaturesthan during the cooling phase (see Discussion).

In the experiment shown in Fig. 3C and D, thepyramidal cells were activated alternately by ortho-dromic (synaptic) and antidromic (direct) stimula-tion. In contrast to the synaptically evokedpopulation spikes (Fig. 3C, D), the antidromicallyinduced population spikes (not shown) were notdepressed during rewarming and did not changeduring the halt phases. As seen in experiments withsimultaneous recording from the stratum radiatum(n=2), the changes of the population spike amplitude

during the halt phases were associated with changesof the slope of the FPSP in the same sense. Theafferent volley was not modified during the haltphases.

Temperature range of the temperature change-inducedeffects

In a third experimental series, we wanted toelucidate if the depression during rewarming isassociated with a certain temperature range or with acertain duration of rewarming. To address thisquestion, we applied zig-zag ramps with repetitivecooling and warming phases of approximately 10 and5)C, respectively, and vice versa in two rat slices.The results were similar in both experiments. One ofthem is shown in Fig. 4A, where the time-course ofthe spike amplitude is depicted together with thetemperature. In Fig. 4B and C, the spike amplitudeis plotted as a function of temperature. To allowthe discrimination of individual phases, the individ-ual cooling and warming phases are numbered

Fig. 4. Synaptic transmission in a rat hippocampal slice during zig-zag temperature changes. Populationspike (PS) amplitude as a function of time (A) and as a function of temperature (B, C). The phases ofcooling (thin lines) and warming (thick lines) are numbered consecutively (c, cooling; w, warming). Forclarity, the data of the first half (net cooling phase) and the second half of the experiment (net rewarming

phase) are drawn separately in B and C, respectively.

Effects of temperature on synaptic transmission 71

consecutively according to the numbering in Fig. 4A,and the data of the first and second halves of theexperiment have been split into two diagrams,Fig. 4B and C, respectively. Figure 4B and C clearlyshows that the depression as compared to the valuesduring cooling is fixed to a certain temperature range,independent of the preceding events. In this exper-iment, the depression similarly started at 20)C andwas maximal at 25)C in three different warmingsituations: (i) after cooling from 30 to 20)C (4c-5w inFig. 4B); (ii) after cooling from 20 to 15)C (8c-9w inFig. 4C); and (iii) after cooling from 26 to 20)C(10c-11w in Fig. 4C). In contrast, there was nodepression during warming from 15 to 20)C (6c-7w inFig. 4B, 8c-9w in Fig. 4C; in 9w, the depression isonly present above 20)C) and from 25 to 30)C (2c-3win Fig. 4B, 12c-13w in Fig. 4C). Note that, in thetemperature range of 30–35)C, the values duringcooling are lower than during the preceding warmingphase (1w-2c in Fig. 4B, 13w-14c in Fig. 4C; seeDiscussion).

Pharmacology of the temperature change-inducedeffects

We have tried to block the warming-induceddepression in rat slices with several substances whichinfluence neurotransmitter/neuromodulator receptorsor ion channels. In these experiments, the slices werefirst cooled and rewarmed under control conditions(control cycle). Then, the respective substance wasbath applied at 30)C and, after equilibration atconstant temperature, the stimulus was readjustedto half-maximal if necessary. Finally, the slice wassimilarly cooled and rewarmed during substanceapplication (test cycle). The substances includedthe neuromodulator norepinephrine (20 µM), theGABAA antagonist bicuculline (5 µM), the GABAB

antagonist CPG55845A (10 µM), the P-type Ca2+

channel antagonist ù-agatoxin IVA (0.4–1 µM) andthe adenosine antagonist theophylline (100–200 µM).All of these substances, except CPG55845A, affectedthe population spike amplitude. The warming-induced depression was, however, only influenced bytheophylline.

In all of nine rat slices, application of the adeno-sine antagonist theophylline (100–200 µM) inducedan increase of the population spike amplitude at 30)Cby 110&37% and, after readjustment of the stimulusintensity, a strong attenuation of the warming-relateddepression in the test cycle. The average effect oftheophylline in four of these slices is shown in Fig. 5(in the other experiments, rundown problems due toa defective isolation unit accumulator prevented athorough quantitative analysis). In these exper-iments, the spike amplitude at 25)C during rewarm-ing was depressed, as compared to cooling, by54&13% in the control cycle (Fig. 5A). In the testcycle, the depression was only 26&11% (Fig. 5B).Thus, theophylline attenuated the warming-relateddepression at 25)C by 57&10%, which is statisticallysignificant at the level of P<0.001 (paired t-test).

Effects of adenosine at constant temperature

To reveal differences between rat and hamsterslices concerning the adenosine sensitivity, adenosinewas bath applied at a constant temperature of 22)Cto slices from rats, as well as from warm-acclimatedand hibernating hamsters. In all animal groups,adenosine reversibly depressed the population spikeand the slope of the FPSP. At a concentration of10 µM, the depression of the population spikeamplitude amounted to 40% in rat slices, whereashamster slices were unaffected (Table 1). Also, atconcentrations of 50 and 100 µM, hamster sliceswere significantly less sensitive to adenosine than ratslices. In addition, application of 100 µM adenosine

Fig. 5. Effect of theophylline on synaptic transmission in rat hippocampal slices during ramp-liketemperature changes. Population spike (PS) amplitude as a function of temperature during control (A)and during bath application of 200 µM theophylline (B). Mean values&S.E.M. of the relative populationspike amplitude (percentage of the maximal amplitude in the respective temperature cycle), n=4 slices. Ineach experiment, a control cycle was first performed. Then, theophylline was applied at constanttemperature, inducing an increase of the population spike amplitude. After stabilization of the effect, thestimulus amplitude was readjusted to half-maximal and a second temperature cycle was performed.

Temperature ramps were limited to the range of 17–30)C.

72 A. Gabriel et al.

revealed a lower sensitivity in slices from hiber-nating hamsters as compared to slices fromwarm-acclimated hamsters (Table 1).

DISCUSSION

To investigate disturbances in neuronal com-munication induced by rapid temperature changes,we studied the effects of rapid ramp-like temperaturechanges (0.3)C/min) on synaptic transmission inhippocampal slices prepared from golden hamstersand rats. While in hamster slices the population spikeamplitude showed a temperature dependence accord-ing to previously published results,14,16,17,43 we founda prominent dynamic influence of the rapid tempera-ture change in rat slices. As compared to the steady-state temperature dependence, the spikes wereslightly augmented at low temperatures during cool-ing and strongly depressed during the whole rewarm-ing phase, with maximal depression at temperaturesbelow 30)C. The depression of the population spikeamplitude during rewarming as compared to coolingwas associated with a decrease of the slope of theFPSP, indicating that the depression is mediated at apresynaptic site. This is supported by the finding thatpopulation spikes directly evoked by antidromicstimulation of CA1 pyramidal cells did not showthe dynamic effects of temperature changes. Theattenuation by theophylline indicates an involvementof endogenous adenosine in the warming-relateddepression. Application of exogenous adenosine atconstant temperature revealed a lower adenosinesensitivity in hamster slices as compared to rat slices.Thus, the lack of dynamic effects of temperature inhamster slices seems to be due to the lower sensitivityto adenosine in this species.

Effects of temperature changes in hippocampal slices

The first reports on temperature effects on synaptictransmission in the hippocampus in vitro appeared adecade ago. In these and later works, it has beenfound in various animal species (rat, hamster,chipmunk, guinea-pig, rabbit) that the amplitude ofsynaptically induced population spikes in hippo-

campal slices was larger at temperatures near 30)C ascompared to higher temperatures.3,14,16,17,35,39,42,43

In the few studies which also included lower tempera-tures, the population spike amplitude decreasedagain with further lowering of tempera-ture.14,16,17,39,43 In contrast to the biphasic behaviourof the population spike amplitude, the slope and theamplitude of the FPSP were positively correlatedwith temperature over the whole range of investi-gated temperatures.16,17,35 In intracellular studies,the input resistance and the amplitude of actionpotentials were found to be larger at temperaturesnear 30)C than at higher temperatures.39,45 Theresting potential was not significantly differentbetween 27 and 37)C in one study,45 whereas inanother study39 a number of cells depolarized withcooling from 35 to 20–25)C and hyperpolarized withwarming.

These results are in line with previous studiesreporting enhanced synaptic transmission duringmoderate hypothermia in other systems, like guinea-pig olfactory cortex in vitro8 and cat spinal cordin vivo,21 and a transient stage of hyperresponsivenessoccurring between the temperatures of 25 and 35)Cat practically all levels of the nervous system (forreviews see Refs 22 and 26).

A transient effect of rapid temperature change inrat hippocampal slices has been reported by Schiffand Somjen.35 During warming from 29 to 33 and37)C, population spikes were transiently increasedbefore they attained a smaller steady-state amplitudeafter 10–15 min. This transient effect was observedduring rapid warming (21)C/min, as inferred fromFig. 4 in Ref. 35), but not during slow warming, andthus appeared to be related to the rate of temperatureincrease. At first sight, our results seem to be incontrast to the enhancement of synaptic transmissionin that study. However, this apparent contradiction isresolved by considering the different tempera-ture ranges of the effects in both studies and bythoroughly regarding two minor points in the resultsfrom our experiments with intermittent ramps andzig-zag ramps (Figs 3, 4). The study of Schiff andSomjen35 was restricted to temperatures above 29)C.At this temperature, there is a marked turning point

Table 1. Change of population spike amplitude (%) induced by adenosine in hippocampal slicesfrom warm-acclimated golden hamsters, hibernating golden hamsters and rats

Adenosineconcentration Change of population spike amplitude (%)

(µM) Rat WGH HGH10 "40&14 (6)* "2&2 (5)† 0& 0 (5)50 "63&10 (9)* "16&7 (5)† "10& 8 (6)100 "87& 4 (12)** "57&8 (14)** "21&10 (10)

Mean&S.E.M. Number of slices is given in parentheses. Statistical significance (Student’st-test) between adjacent values: †P>0.05, *P<0.02, **P<0.01. Parts of the data with 50 and100 µM adenosine have been published in Spangenberger et al.40 and Igelmund et al.18 HGH,hibernating golden hamsters; WGH, warm-acclimated golden hamsters.

Effects of temperature on synaptic transmission 73

in our curves showing the population spike amplitudeduring rewarming as a function of temperature(Figs 1D, 5A). Thus, at higher temperatures,the warming-related depression seems to be eitherreleased or superimposed by a counteractingexcitatory mechanism. The second of these two pos-sibilities is supported by the fact that in five of eightintermittent-ramp experiments, after a 15-min pauseat 27)C, further rewarming resulted in an augmenta-tion rather than a depression of the spikes as com-pared to the cooling phase (Fig. 3D). Similarly, atransient excitatory effect is suggested by the obser-vation that, in the zig-zag ramp experiments, coolingfrom 35 to 30)C yielded smaller spikes than theimmediately preceding short warming periods, whichstarted at 30 and 25)C, respectively (phases 1w-2c inFig. 4B and 13w-14c in Fig. 4C). Together, theseresults indicate an excitatory effect of rapid warmingin the temperature range above 26–29)C, but not atlower temperatures, which is fully in line with andcomplementary to the results of Schiff and Somjen.35

To our knowledge, there are no reports aboutdynamic effects of temperature on synaptic transmis-sion in mammalian central neurons at lower tempera-tures. Accordingly, the depression during warming ascompared to cooling, which was very obvious atlower temperatures in our experiments, does notseem to be mentioned in the literature. There is,however, a study in which comparable effects mayhave occurred. Buldakova et al.3 warmed rat hippo-campal slices from 25 to 32)C with 1)C steps and0.7 min intervals between steps. They measured adecrease of the population spike amplitude with eachstep. Subsequent recooling reversed the depression.While population spike amplitudes returned progres-sively to their initial amplitudes in slices from aged(24–28 months old) rats, a long (>30 min) and con-siderable increase of the population spike amplituderelative to the control at 25)C was observed in slicesfrom young (one to two months old) rats. The largerspikes during cooling as compared to warming are inline with our results. The duration of the effects,however, is much longer than in our study. Thisdifference may result from the different velocities oftemperature change. In unpublished experimentsfrom our laboratory, stepwise warming from 25 to30)C with a slope of 27)C/min produced a depres-sion of synaptic transmission lasting for >30 min(Gabriel A. and Igelmund P., unpublished obser-vations). On the other hand, the difference may bebased on the different age of the animals. Buldakovaet al.3 used young rats with an age of one to twomonths, whereas our animals were two to fourmonths old.

Pharmacology

We have shown that application of 100–200 µMof the adenosine antagonist theophylline largelydiminishes the depression occurring during rewarm-

ing in rat slices. This suggests that the depression ismediated by endogenous adenosine. In this concen-tration, theophylline only partially blocks the effectsof adenosine (20–100 µM) applied exogenously atconstant temperature in rat as well as in hamsterslices (Gabriel A., Zhao Y. Q. and Igelmund P.,unpublished observations). Thus, an incompleteblock does not indicate the involvement of othermechanisms. At higher concentrations, theophyllinecan also directly inhibit phosphodiesterases (PDEs).However, although we have not excluded this effect,a direct action on PDEs seems unlikely in our exper-iments, since theophylline is a relatively weak PDEinhibitor; the potency of theophylline in inhibitingbrain PDEs is quite different from the potency inblocking adenosine receptors.4,27 The opinion thatthe effect of theophylline was restricted to the blockof adenosine receptors is further supported byanalogous results: hypoglycaemia-induced failure oftransmission, which is also attenuated by theo-phylline, is similarly attenuated by the selective A1

antagonist 8-(p-sulphophenyl)theophylline.48 Thisantagonist penetrates cells to a smaller extent andtherefore has less inhibitory effect on PDEs thantheophylline.48

Like other xanthines, theophylline can, at higherconcentrations (and, with a lower Mg2+ level in theACSF, also at lower concentrations), inducespontaneous epileptiform activity in hippocampalslices.27 In our experiments, no epileptiform activitywas observed in response to theophylline. Sinceepileptiform activity in hippocampal slices stronglydepends on temperature,23 its absence in our exper-iments may be due to the low temperatures used.

At the Schaffer collateral/pyramidal cell synapse,adenosine has presynaptic as well as postsynapticeffects, which seem to be mediated primarily by A1

receptors.6,30,37,44 The main presynaptic effect is adepression of Ca2+ currents and thereby a reductionof transmitter release.33,36,47 Postsynaptically, adeno-sine increases a K+ conductance, leading to hyper-polarization of the pyramidal cells.1,11,44 For syn-aptically induced field potentials of CA1 pyramidalcells, the effect of adenosine is almost completelybased on the inhibition of transmitter release38

mediated by block of presynaptic Ca2+ channels.47

It has repeatedly been shown that synaptic trans-mission in acute5,12,13 as well as cultured hippo-campal slices29 underlies a tonic depression mediatedby endogenous adenosine. The release of adenosine isremarkably related to the metabolic state, indicatingthat adenosine is a link between metabolic andneurophysiologic activities. Pathological conditionslike hypoxia, hypoglycaemia and seizure activity,which increase metabolic demand relative to metabo-lite availability, increase the adenosine concentrationin the extracellular fluid of nervous tissue.12,34 Thephysiological adenosine level in the rat brain, asmeasured in the striatum, has been estimated to be40–460 nM.2

74 A. Gabriel et al.

Hypoxia-induced7 as well as hypoglycaemia-induced48 block of neuronal activity in hippocampalslices can be strongly delayed and reduced withtheophylline, as well as with specific A1 antagonists,indicating that this kind of depression is mediated byendogenous adenosine. Similarly, action potential-dependent spontaneous glutamatergic excitatorypostsynaptic currents appear in CA1 pyramidal cellsduring hypoxia in the presence of A1 antagonists.20

The mechanism involved seems to be the presynapticaction of adenosine, since block by pertussis toxin ofpostsynaptic adenosine action had no effect onhypoxia/hypoglycaemia-induced hyperpolarizationof pyramidal cells.41 Our results suggest that thewarming-related depression of the population spike issimilarly based on presynaptic mechanisms. This isindicated by (i) the association of the depression witha decrease of the slope of the FPSP and (ii) thelack of dynamic effects on antidromically evokedpopulation spikes.

Possible coupling between temperature changes andadenosine release

The action of temperature on synaptic transmis-sion is based on a bouquet of different mechanismswith different temperature characteristics. Thesemechanisms include ion channel permeabilities forNa+, K+ and Ca2+, as well as the electrogenicNa+/K+ pump and other transporters, in thedendrites, somata and axons of excitatory as well asinhibitory cells. All these mechanisms influence theduration and amplitude of pre- and postsynapticalpotentials, and thereby transmitter release and excit-ability.32,35,39,45 As a result, a non-linear temperaturedependence of synaptic transmission is not surpris-ing. However, the temperature dependence of allthese mechanisms provides no explanation for thedynamic effects of temperature changes.

Schiff and Somjen35 proposed that the differentsurface-to-volume ratios of the presynaptic terminalsand pyramidal cells may create the transient increaseof excitatory transmission during rapid warming attemperatures above 29)C. If small-volume pre-synaptic terminals hyperpolarize faster than largepyramidal cell somata, they might release more trans-mitter before the pyramidal cells hyperpolarize andbecome less sensitive. Similarly, a different reactiontime due to different compartment properties orreaction properties is a possible explanation for theadenosine-dependent dynamic effects of warming andcooling on synaptic transmission at lower tempera-tures. If energy-consuming mechanisms are influ-enced faster by temperature changes than energy-producing mechanisms, then warming will lead toenhanced energy demand as compared to energyavailability, whereas during cooling the opposite willoccur. As a consequence, release of endogenousadenosine will be enhanced during warming and

attenuated during cooling, leading to depression andenhancement, respectively, of synaptic transmission.

In addition, a possible increase of glutamaterelease could contribute to the rise of the endogenousadenosine level during warming. Glutamate, actingon N-methyl--aspartate receptors, has been shownto elevate the extracellular adenosine concentrationin hippocampal slices, presumably by increasing theadenosine release from interneurons and possiblypyramidal neurons.24

The warming-related depression found in thepresent study was maximal at temperatures around27)C. The attenuation of the depression at highertemperatures may be due to various mechanisms:(i) a decrease of adenosine release; (ii) an increase ofadenosine elimination; (iii) a decrease of sensitivity toadenosine; and (iv) to excitatory effects mediated bydifferent mechanisms. Measuring the effects of 50 µMadenosine on submaximal population spikes inrat hippocampal slices at constant temperatures,Spangenberger et al.40 found a mean depression of69&8% and 39&12% at 22 and 37)C, respectively.The difference between the mean effects at the twotemperatures was not significant. However, compar-ing the effects for each slice tested at both tempera-tures, evaluation with the paired t-test reveals asignificance of P<0.02 for the difference betweenthe effects at 22 and 37)C (Spangenberger H. andIgelmund P., unpublished observations). This sug-gests that the attenuation of the warming-relateddepression at temperatures above 27)C is, at least inpart, due to a lower sensitivity to adenosine at highertemperatures.

Differences between hamsters and rats

A striking result of our investigation is theobservation that the warming-related, adenosine-dependent depression was present in rat slices but notin hamster slices. Exogenously applied adenosine at aconcentration of 10 µM at 22)C depressed populationspikes in rat slices by 40%, which is in the range of thewarming-related depression at this temperature (seeFigs 1D and 5A). At this concentration, adenosinehad no effect on hamster slices. Thus, the lack ofwarming-related depression in hamster slices may becompletely explained by the difference in sensitivity.In addition to the lower adenosine sensitivity, hiber-nators like hamsters might have an energy metab-olism which does not easily get out of balance duringrapid temperature changes and thus does not causeenhanced release of adenosine during warming.

Striking differences between hamsters and rats,as well as between other hibernators and non-hibernators, are also obvious in the responses ofintact animals to induced hypothermia, resulting indifferent metabolic reactions and a conspicuouslylonger survival time of hibernators as comparedto non-hibernators at body temperatures of 10–19)C.19,28,46 In the context of the present paper, it is

Effects of temperature on synaptic transmission 75

highly interesting that, in the study of Massopustet al.,25 cats and guinea-pigs (non-hibernators) werenot able to rewarm from hypothermia below 24)C,while prairie dogs and ground squirrels (hibernators)rewarmed from temperatures below 16)C withoutresuscitation. Similarly, European hamsters arereported to be able to rewarm from induced hypo-thermia down to 7–10)C.31 Concerning the differentsurvival times, Popovic28 points out the possibility‘‘that it is not an organ which fails first but that thereis failure in coordination and integration, whichbecomes critical at the moment of biological death’’.Apparently, the non-hibernator brain is not able toprovide an orderly control of homeostatic functionsat low temperatures and/or with temperaturechanges. It is plausible to assume that adenosine-mediated depression of synaptic transmission,induced by a disequilibrium of energy metabolism,

could be a major source of disturbances of neuronalcommunication in non-hibernators in and afterinduced or accidental hypothermia. For hibernators,the reduction of sensitivity to adenosine or othermechanisms which avoid temperature change-induced disturbances of neuronal communica-tion may be a prerequisite for safe arousal fromhibernation.

Acknowledgements—We thank Holger Spangenberger andYongqiang Zhao for carrying out some experiments withadenosine application at constant temperature, and ManuelMetzner for participation in some ramp experiments.Thanks are also due to Erika Walde for excellent technicalassistance, Uwe Heinemann and Holger Spangenberger forvaluable discussions, and Michaela Bohm-Pinger for read-ing the manuscript. ù-Agatoxin IVA and CPG55845Awere gifts of Pfizer (Groton, CT) and Ciba Geigy (Basel,Switzerland), respectively. This research was supported byDFG grants Ig10/1-2 and Ig10/1-3.

REFERENCES

1. Ameri A. and Jurna I. (1991) Adenosine A1 and non-A1 receptors: intracellular analysis of the actions of adenosineagonists and antagonists in rat hippocampal neurones. Brain Res. 546, 69–78.

2. Ballarin M., Fredholm B. B., Ambrosio S. and Mahy N. (1991) Extracellular levels of adenosine and its metabolitesin the striatum of awake rats: inhibition of uptake and metabolism. Acta physiol. scand. 142, 97–103.

3. Buldakova S., Dutova E., Ivlev S. and Weiss M. (1995) Temperature change-induced potentiation: a comparativestudy of facilitatory mechanisms in aged and young rat hippocampal slices. Neuroscience 68, 395–397.

4. Choi O. H., Shamim M. T., Padgett W. L. and Daly J. W. (1988) Caffeine and theophylline analogues: correlation ofbehavioural effects with activity as adenosine receptor antagonists and as phosphodiesterase inhibitors. Life Sci. 43,387–398.

5. Dunwiddie T. V. (1980) Endogenously released adenosine regulates excitability in the in vitro hippocampus. Epilepsia21, 541–548.

6. Dunwiddie T. V. and Haas H. L. (1985) Adenosine increases synaptic facilitation in the in vitro rat hippocampus:evidence for a presynaptic site of action. J. Physiol. 369, 365–377.

7. Fowler J. C. (1989) Adenosine antagonists delay hypoxia-induced depression of neuronal activity in hippocampalbrain slice. Brain Res. 490, 378–384.

8. Fujii T. and Yoshizaki K. (1976) Temperature influence on the development of electrical activities in mammalian brainslice during incubation. Jap. J. Physiol. 26, 355–365.

9. Gabriel A., Lutke K. and Igelmund P. (1993) Dynamic effects of temperature in hippocampal slices of rats and goldenhamsters. In Genes, Brain and Behavior. Proceedings of the 21st Gottingen Neurobiology Conference (eds Elsner N. andHeisenberg M.), p. 510. Georg Thieme, Stuttgart.

10. Gabriel A., Metzner M. and Igelmund P. (1995) Dynamic effects of temperature changes on synaptic transmission inrat hippocampal slices are partially based on endogenous adenosine. In Gottingen Neurobiology Report 1995.Proceedings of the 23rd Gottingen Neurobiology Conference (eds Elsner N. and Menzel R.), p. 658. Georg Thieme,Stuttgart.

11. Gerber U., Greene R. W., Haas H. L. and Stevens D. R. (1989) Characterization of inhibition mediated by adenosinein the hippocampus of the rat in vitro. J. Physiol. 417, 567–578.

12. Greene R. W. and Haas H. L. (1991) The electrophysiology of adenosine in the mammalian central nervous system.Prog. Neurobiol. 36, 329–341.

13. Greene R. W., Haas H. L. and Hermann A. (1985) Effects of caffeine on hippocampal pyramidal cells in vitro. Br. J.Pharmac. 85, 163–169.

14. Hooper D. C., Martin S. M. and Horowitz J. M. (1985) Temperature effects on evoked potentials of hippocampalslices from euthermic chipmunks, hamsters and rats. J. therm. Biol. 10, 35–40.

15. Igelmund P. (1995) Modulation of synaptic transmission at low temperatures by hibernation-related changes in ionicmicroenvironment in hippocampal slices of golden hamsters. Cryobiology 32, 334–343.

16. Igelmund P. and Heinemann U. (1995) Synaptic transmission and paired-pulse behaviour of CA1 pyramidal cells inhippocampal slices from a hibernator at low temperature: importance of ionic environment. Brain Res. 689, 9–20.

17. Igelmund P., Heinemann U. and Klußmann F. W. (1993) Hibernation-related modification of activity-dependentproperties of synaptic transmission in hamster hippocampus. Neurosci. Res. Commun. 13, 167–173.

18. Igelmund P., Spangenberger H., Nikmanesh F. Gh., Gabriel A., Lutke K., Zhao Y. Q., Bohm-Pinger M. M.,Heinemann U., Hescheler J. and Klußmann F. W. (1996) Hibernation and hippocampal synaptic transmission. InAdaptation to the Cold: Tenth International Hibernation Symposium (eds Geiser F., Hulbert A. J. and Nicol S. C.),pp. 159–166. University of New England Press, Armidale.

19. Jourdan M. L., McAllister D. J. and Wang L. C. H. (1989) Physiological changes in cations and water balance duringstable long term hypothermia in rats. In Thermoregulation: Research and Clinical Applications (eds Lomax P. andSchonbaum E.), pp. 157–161. Karger, Basel.

20. Katchman A. N. and Hershkowitz N. (1996) Adenosine A1 antagonism increases specific synaptic forms of glutamaterelease during anoxia, revealing a unique source of excitation. Hippocampus 6, 213–224.

76 A. Gabriel et al.

21. Klee M. R., Pierau F. K. and Faber D. S. (1974) Temperature effects on resting potential and spike parameters of catmotoneurons. Expl Brain Res. 19, 478–492.

22. Koizumi K., Brooks C. M. and Ushiyama J. (1959) Hypothermia and reaction patterns of the nervous system. Ann.N. Y. Acad. Sci. 80, 449–456.

23. Leschinger A., Stabel J., Igelmund P. and Heinemann U. (1993) Pharmacological and electrographic properties ofepileptiform activity induced by elevated K+ and lowered Ca2+ and Mg2+ concentration in rat hippocampal slices.Expl Brain Res. 96, 230–240.

24. Manzoni O. J., Manabe T. and Nicoll R. A. (1994) Release of adenosine by activation of NMDA receptors in thehippocampus. Science 265, 2098–2101.

25. Massopust L. C., Wolin L. R. and Meder J. (1965) Spontaneous electrical activity of the brain in hibernators andnon-hibernators during hypothermia. J. exp. Neurol. 12, 25–32.

26. Mihailovic L. T. (1972) Cortical and subcortical electrical activity in hibernation and hypothermia. In Hibernation–Hypothermia. Perspectives and Challenges (eds South F. E., Hannon J. P., Willis J. R., Pengelley E. T. and AlpertN. R.), pp. 487–534. Elsevier, Amsterdam.

27. Moraidis I. and Bingmann D. (1994) Epileptogenic actions of xanthines in relation to their affinities for adenosine A1

receptors in CA3 neurons of hippocampal slices (guinea pig). Brain Res. 640, 140–145.28. Popovic V. (1960) Physiological characteristics of rats and ground squirrels during prolonged lethargic hypothermia.

Am. J. Physiol. 199, 467–471.29. Prestwich S. A., Forda S. R. and Dolphin A. C. (1987) Adenosine antagonists increase spontaneous and evoked

transmitter release from neuronal cells in culture. Brain Res. 405, 130–139.30. Proctor W. R. and Dunwiddie T. V. (1987) Pre- and postsynaptic actions of adenosine in the in vitro rat hippocampus.

Brain Res. 426, 187–190.31. Raths P. (1957) Die bioelektrische Hirntatigkeit des Hamsters im Verlaufe des Erwachens aus Winterschlaf und

Kaltenarkose. Z. Biol. 110, 62–80.32. Reid K. H., Edmonds H. L. J., Schurr A., Tseng M. T. and West C. A. (1988) Pitfalls in the use of brain slices. Prog.

Neurobiol. 31, 1–18.33. Ribeiro J. A. and Sebastiao A. M. (1986) Adenosine receptors and calcium: basis for proposing a third (A3) adenosine

receptor. Prog. Neurobiol. 26, 179–209.34. Rudolphi K. A., Schubert P., Parkinson F. E. and Fredholm B. B. (1992) Adenosine and brain ischemia. Cerebrovasc.

Brain Metab. Rev. 4, 346–369.35. Schiff S. J. and Somjen G. G. (1985) The effects of temperature on synaptic transmission in hippocampal tissue slices.

Brain Res. 345, 279–284.36. Scholz K. P. and Miller R. J. (1991) Analysis of adenosine actions on Ca2+ currents and synaptic transmission in

cultured rat hippocampal pyramidal neurones. J. Physiol. 435, 373–393.37. Schubert P., Heinemann U. and Kolb R. (1986) Differential effect of adenosine on pre- and postsynaptic calcium

fluxes. Brain Res. 376, 382–386.38. Schubert P. and Mitzdorf U. (1979) Analysis and quantitative evaluation of the depressive effect of adenosine on

evoked potentials in hippocampal slices. Brain Res. 172, 186–190.39. Shen K. F. and Schwartzkroin P. A. (1988) Effects of temperature alterations on population and cellular activities in

hippocampal slices from mature and immature rabbit. Brain Res. 475, 305–316.40. Spangenberger H., Nikmanesh F. Gh. and Igelmund P. (1995) Effects of adenosine on synaptic transmission in

hippocampal slices from hibernating and warm-acclimated Turkish hamsters and rats. Neurosci. Lett. 185, 217–219.41. Spuler A. and Grafe P. (1989) Adenosine, ‘‘pertussis-sensitive’’ G-proteins, and K+ conductance in central mammalian

neurons under energy deprivation. Neurosci. Lett. 98, 280–284.42. Tanimoto M. and Okada Y. (1987) The protective effect of hypothermia on hippocampal slices from guinea pig during

deprivation of oxygen and glucose. Brain Res. 417, 239–246.43. Thomas M. P., Martin S. M. and Horowitz J. M. (1986) Temperature effects on evoked potentials of hippocampal

slices from noncold-acclimated, cold-acclimated and hibernating hamsters. J. therm. Biol. 11, 213–218.44. Thompson S. M., Haas H. L. and Gahwiler B. H. (1992) Comparison of the actions of adenosine at pre- and

postsynaptic receptors in the rat hippocampus in vitro. J. Physiol. 451, 347–363.45. Thompson S. M., Masukawa L. M. and Prince D. A. (1985) Temperature dependence of intrinsic membrane

properties and synaptic potentials in hippocampal CA1 neurons in vitro. J. Neurosci. 5, 817–824.46. Uchida K., Shibuya I. and Doi K. (1987) Difference in the mode of acute cold-induced hypothermia between rat and

hamster. Jap. J. Physiol. 37, 207–222.47. Wu L. G. and Saggau P. (1994) Adenosine inhibits evoked synaptic transmission primarily by reducing presynaptic

calcium influx in area CA1 of hippocampus. Neuron 12, 1139–1148.48. Zhu P. J. and Krnjevic K. (1993) Adenosine release is a major cause of failure of synaptic transmission during

hypoglycaemia in rat hippocampal slices. Neurosci. Lett. 155, 128–131.

(Accepted 5 January 1998)

Effects of temperature on synaptic transmission 77