Transient and sustained types of long-term potentiation in the CA1 area of the rat hippocampus

doi:10.1113/jphysiol.2003.044214

Transient and sustained types of long-term potentiation in the CA1 area of the rat hippocampus

Arturas Volianskis and Morten S. Jensen

Department of Neurobiology, Institute of Anatomy, University of Aarhus, Wilhelm Meyers Allé, Building 233/234 DK-8000 Aarhus C, Denmark

ABSTRACT

Top
Abstract
Introduction
Methods
Results
Discussion
References

Synaptic potentiation induced by high frequency stimulation was investigated by recording field excitatory postsynaptic potentials (f-EPSPs) in rat hippocampal slices. Potentiation consisted of a transient period of decaying f-EPSPs (short-term potentiation, STP) that led to a plateau of continuously potentiated f-EPSPs (long-term potentiation, LTP). Here we show that a previously unknown type of transient, use-dependent, long-lasting potentiation (t-LTP) can account for STP. t-LTP could be stored for more than 6 h and its decay was caused by synaptic activation. Both the expression and the decay of t-LTP were input specific. t-LTP was induced differently from conventional LTP in that the amplitude of t-LTP was dependent upon the stimulation frequency, whereas the magnitude of LTP was dependent on the number of stimuli in the induction train. The decay of t-LTP could not be prevented by the blockage of glutamate receptors, but was prevented by the blockage of stimulus-evoked neurotransmitter release, suggesting that t-LTP is expressed presynaptically. Paired-pulse stimulation experiments showed that the decay of t-LTP was mediated by a decrease in the probability of neurotransmitter release. The decline of t-LTP could be prolonged by the activation of NMDA receptors. Hence, both single and paired-pulse stimuli prolonged the decline of the t-LTP. This decline could be prevented by high frequency burst stimulation (200 Hz). We conclude that t-LTP allows dynamic modulation of synaptic transmission by providing not only spatial association but also temporal convergence between synaptic inputs. Therefore, t-LTP might be a substrate for the encoding of synaptic memory.

(Resubmitted 1 April 2003; accepted after revision 24 April 2003; first published online 6 June 2003)
Corresponding author M. S. Jensen: Department of Neurobiology, Institute of Anatomy, University of Aarhus, Wilhelm Meyers Allé, Building 233/234 DK-8000, Aarhus C, Denmark. Email: msj{at}neuro.au.dk

INTRODUCTION

Top
Abstract
Introduction
Methods
Results
Discussion
References

Long-term potentiation (LTP), which was first described in 1973 (Bliss & Lomo, 1973), is currently the main model of long-lasting synaptic memory in the mammalian brain. Potentiation, defined as an increase in synaptic efficacy, is readily induced by high frequency stimulation (HFS) of the synapses between the Schaffer collaterals and the pyramidal cells in the hippocampus CA1 area (Collingridge & Bliss, 1995; Malenka & Nicoll, 1999). This increase can be evaluated in terms of changes in the rate of rise of extracellulary recorded field excitatory postsynaptic potentials (f-EPSPs, Andersen et al. 1980). Potentiation, induced by HFS, has three phases: (1) a second-long phase of decaying f-EPSPs termed post-tetanic potentiation (PTP) that is followed by (2) a longer lasting (30-60 min) period of a slow decrease in synaptic efficacy. The latter phase leads to (3) a plateau of potentiated field potentials (LTP). The slow decay of potentiation is usually termed short-term potentiation (STP) or early LTP (e-LTP) and has been considered to be an epiphenomenon of LTP induction.

Post-tetanic potentiation is caused by a presynaptic accumulation of Ca²⁺ during tetanisation and the rapid decay of PTP reflects the clearance of the Ca²⁺ from the presynaptic terminals (Stevens et al. 1994; Zucker, 1999). Differently from PTP, the induction of both STP and LTP can be blocked by N-methyl-D-aspartate (NMDA) receptor antagonists (Collingridge et al. 1983; Anwyl et al. 1989; Stevens et al. 1994; Tsien et al. 1996). Activation of NMDA receptors leads to a subsequent influx and rise in the postsynaptic levels of Ca²⁺, which is obligatory for the induction of both STP and LTP (Malenka et al. 1988; Anwyl et al. 1988; Malenka et al. 1992).

The nature of STP and its relationship to LTP remains under discussion. Two main hypotheses explaining STP have been put forward. According to the first view, the transient phase is an unstable phase of LTP that leads to a consolidation of potentiation (Gustafsson & Wigstrom, 1990; Hanse & Gustafsson, 1994). In contrast, it has also been suggested that the transient phase might be different from the stable phase of LTP (Kauer et al. 1988; Schulz & Fitzgibbons, 1997). Studies, which compare the two phenomena, imply that STP and LTP have a different threshold for their induction in that different numbers of high frequency stimuli are needed to induce STP and LTP (e.g. Schulz & Fitzgibbons, 1997). Nevertheless, the mechanistic nature of STP and its role in synaptic encoding are unknown. Notably, a phenomenon that resembles STP (and not LTP) has been observed after exploratory learning in rats (Moser et al. 1993, 1994) but no further attempts have been made to explain how STP can be involved in the formation of memories.

We have investigated potentiation induced by HFS in the CA1 area of rat hippocampal slices and demonstrate here that STP, the transient phase of synaptic potentiation, is a previously unknown type of long-lasting potentiation. This transient LTP (t-LTP) can be stored for many hours in an input-specific manner and its disappearance is caused by synaptic activation. t-LTP is induced differently from the sustained type of LTP (s-LTP) in that the amplitude of t-LTP is dependent on the frequency of tetanisation whereas the amplitude of s-LTP is dependent on the number of stimuli that are used during a tetanus. We further examined the site for synaptic expression of t-LTP and found that its induction depends on NMDA receptor activation during tetanisation. We also found that the induction of t-LTP decreases the ability of synapses to express paired-pulse facilitation and that the decline of t-LTP is accompanied by concomitant increases in the amount of paired-pulse facilitation in synaptic populations. The decay of t-LTP, furthermore, requires presynaptic influx of calcium, and blockage of glutamate receptors during test stimulation cannot prevent this decay. When taken together these results show that t-LTP, in a population of synapses, is a consequence of an increase in the probability of neurotransmitter release and that the decay of t-LTP is explained by a decrease in the probability of neurotransmitter release. We further show that the decay of t-LTP is dynamically regulated by NMDA- receptor activation during test stimulation. Therefore, the amount of t-LTP that is present at any given time after its induction depends on a balance between a use-dependent reduction in the probability of neurotransmitter release and a concurrent re-induction of potentiation via the activation of NMDA receptors.

METHODS

Top
Abstract
Introduction
Methods
Results
Discussion
References

Subjects

One hundred and eighty-eight adult (73 ± 0.9 days old, 316 ± 3 g) male Wistar rats were used in the study and all experiments were conducted according to national and EU guidelines for animal care.

Slice preparation

The rats were deeply anaesthetised with halothane and killed by decapitation. The hippocampi were dissected from the brain and cooled down in Hepes-buffered saline solution (0-4 °C) consisting of (mM): 5 Hepes, 124 NaCl, 3.5 KCl, 1.25 NaH₂PO₄, 2 CaCl₂, 2 MgSO₄ and 26 NaHCO₃, saturated with 95 % O₂-5 % CO₂. Slices of the septal end of the hippocampus (400 µm thick) were cut on a McIllwain tissue chopper and incubated at room temperature in the Hepes solution for at least 2 h before they were transferred to the recording chamber. During the experiments the slices were submerged in a medium that was heated to 33 °C. The medium consisted of (mM): 124 NaCl, 3.5 KCl, 1.25 NaH₂PO₄, 2 CaCl₂, 2 MgSO₄, 26 NaHCO₃, equilibrated with 95 % O₂-5 % CO₂ at 37 °C. In the experiments in which a calcium-free medium (Ca²⁺-free) was used calcium was replaced by equimolar concentrations of MgSO₄.

Chemicals and their application

The chemicals used were obtained from Sigma, with the exception of D(-)-2-amino-5-phosphonopentanoic acid (D-AP5), DL-2-amino-5-phosphonovaleric acid (DL-AP5), 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulphonamide disodium (NBQX), ( $alpha$ S)- $alpha$ -amino- $alpha$ [(1S,2S)-2-carboxy -cyclopropyl-9H-xanthine-9-propanoic acid (LY341495), 4-hydroxyquinoline-2-carboxylic acid (kynurenate) and picrotoxin, which were obtained from Tocris. Kynurenate and picrotoxin were added directly to the perfusion medium before use. Other chemicals were prepared as stock solutions, stored frozen and added to the perfusion medium as indicated in Results.

Experimental procedures

The Schaffer collaterals were stimulated with a bipolar tungsten electrode placed on the border between the areas CA1 and CA2 using a stimulus isolation unit (A385, WPI) and a pulse generator (Master 8, A.M.P.I). The stimulus duration was always 100 µs and the stimulation current (150-450 µA) was fixed to three times threshold for evoking f-EPSPs in the CA1-B area of the stratum radiatum, if not stated otherwise. The f-EPSPs were recorded with glass electrodes that were filled with saline solution (3-5 M $Omega$ ). The signals were amplified using an Axoclamp 2B amplifier (Axon Instruments, Inc.) and filtered at 10 kHz (Cyber Amp 380, Axon Instruments, Inc.). Stimulus-evoked f-EPSPs were digitised (Digidata 1200b, Axon Instruments, Inc.) and recorded on a PC using pCLAMP software (Axon Instruments, Inc.).

Test stimulation during the control periods of the experiments was given once every 15 s. Tetanisation was given as theta-burst stimulation (four stimuli at 100 Hz, repeated 10 times, with a frequency of 5 Hz) that was routinely used to induce potentiation. Other induction protocols were used also and are described in the appropriate Results sections. Averages of four f-EPSPs were acquired at 2 min post-tetanus in order to estimate the maximal amounts of NMDA receptor-dependent potentiation (P_max) in slices. The subsequent frequencies and patterns of the test stimulation were dependent on the experimental hypothesis. For convenience, low frequency test stimulation is given as a fraction of times per number of seconds (e.g. 1/15 s = 0.67 Hz). Field responses were recorded as averages of four responses to stimulation (four pairs when paired-pulse stimulation was applied).

'Two pathway' experiments were performed by placing an additional stimulation electrode in the CA1-C area of the Schaffer collaterals (on the border of the subiculum). The two stimulation electrodes were activated alternately once every 15 s with stimuli separated by 7.5 s in order to stimulate two separate synaptic populations. Crosstalk between synaptic populations was minimised by assuring that there was no paired-pulse facilitation when stimuli were separated by 80 ms.

Data analysis and plotting

In our analysis of the experiments the rates of rise of the f-EPSPs were measured as the deflections of the slopes of the f-EPSPs at the points after the termination of the afferent fibre volleys (AVs) divided by the time interval from the peaks of the AVs to the measured points on the f-EPSPs. Data, obtained in the individual experiments, were normalised to their respective control periods that were set to 100 %. The actual gain in synaptic transmission (potentiation) was expressed as the amplitude of potentiation by subtracting the control level (100 %) from the normalised rising rates of the f-EPSPs. This permitted direct comparison and statistical evaluation of the initial, the transient and the sustained amplitudes of potentiation.

Experimental results are visualised by charts of the amplitudes of potentiation, which are plotted over time. The data are presented in terms of experimental group averages. Single data points show the mean values of the groups and the error bars show the standard errors of the mean values (S.E.M.). The data are plotted with a time interval of 2 min after the induction of potentiation. Note that the average potentiation in the control periods is zero (no gain in synaptic transmission).

The amplitudes of the transient and of the sustained phases of potentiation, and the decay time constants of the transient phases, were determined by applying a mono-exponential fitting routine (Levenberg-Marquardt method) to data in the individual experiments. The mean estimated amplitudes of the different phases of potentiation and the mean decay time constants of the transient phases, which were obtained in different experimental groups, are summarised in histograms (±S.E.M.) throughout the Results. Dashed bars in the histograms indicate that an experimental group has been shown in one of the preceding figures and is now reproduced in the particular histogram for the purposes of comparison.

Analysis, statistical evaluation and graphical representation of the data were performed with Platin (a custom-built software package), Excel (Microsoft) and SPSS (SPSS Inc.). Student's (two-tailed, either paired or unpaired) t test, ANOVA and the Kolmogorov-Smirnov test were used for statistical comparisons. If not otherwise stated, unpaired t tests were used for the statistics. Significant differences are indicated in the histograms by asterisks (*P < 0.05; **P < 0.01; ***P < 0.001).

RESULTS

Top
Abstract
Introduction
Methods
Results
Discussion
References

NMDA receptor-dependent potentiation in the CA1 area of the hippocampus

The field response in the stratum radiatum of the CA1 area of the hippocampus is a compound waveform. It consists of the sum of an afferent fibre volley (AV) and a f-EPSP. It also reflects somatic spiking in a form of an inverted population spike (PS). Therefore, the basic concern of any extracellular study of potentiation of synaptic transmission is the extent to which the rising phase of a f-EPSP can be isolated from both the AV and the population spike in order to quantify changes in synaptic transmission in a given experimental situation.

In order to investigate the above, field responses were evoked by stimulating the Schaffer collaterals and recorded in the stratum radiatum (st.r.) and the stratum pyramidale (st.p.) of the CA1 area (Fig. 1A). After a recording of a stable period of the control responses (Fig. 1Aa) synaptic transmission was potentiated by theta-burst stimulation (tetanisation, see inset, Fig. 1B). Potentiation did not affect the initial slopes of the AVs and their peaks did not change (Fig. 1Ab-d). However, potentiation increased the amplitudes and reduced the latencies of the f-EPSPs when compared with those of the control. Similarly, the amplitudes of the population spikes were enhanced and their latencies were reduced.

View larger version
[in this window]
[in a new window]

Figure 1. Potentiation in the CA1 area of the hippocampus
A, schematic illustration of the experimental situation that was used to study potentiation of synaptic transmission in the CA1 area of hippocampal slices. The stimulation electrode was placed in the Schaffer collaterals on the border between the areas CA1 and CA2. Recording electrodes were placed in the stratum pyramidale (St.p.) and the stratum radiatum (St.r.) of the CA1 area in order to record the population spikes and the field excitatory postsynaptic potentials, respectively. The waveforms, recorded before and after tetanisation of the Schaffer collaterals, are shown on the right of the corresponding recording electrodes. Their labelling and colours correspond to different time points before and after the tetanisation: 'a' before tetanisation (black), 'b' 2 min after the tetanisation (red), 'c' 15 min post-tetanus (green), 'd' 1 h post-tetanus (blue). Three components of the waveforms are indicated by the arrows: (1) the afferent fibre volley (AV), (2) the field excitatory postsynaptic potential (f-EPSP) and (3) the population spike (PS). The points of the initial rising phase of the f-EPSPs (indicated by the dashed line through the stratum radiatum responses) that were neither affected by the afferent volley nor by the population spikes were used to calculate changes in synaptic transmission. The calibration bar is shown in the inset. B, a group of experiments (n = 16, filled circle mean estimates of potentiation, the error bars indicate S.E.M.) depicting the control level of synaptic transmission (Ctrl), the initial amounts of potentiation at 2 min post-tetanus (P_max, 128 ± 8.8 %), the decay of STP (81 ± 6.7 %) and the LTP (47 ± 3.7 %). Potentiation of synaptic transmission was calculated by subtracting the 100 % level from the relative change in synaptic transmission. The amplitudes of STP were estimated by subtracting the amplitudes of the LTP from those of the P_max in the individual experiments. Labelling and colours correspond to the sample waveforms in A. Filled arrow indicates the time of tetanisation and the tetanisation protocol, consisting of a number of theta-bursts, is depicted in the inset. Ca, in this group of experiments (n = 16) 50 µM D-AP5 (indicated by the continuous line) was applied in order to isolate the PTP. The amplitude of PTP (open arrowhead) was estimated at 2 s after the first tetanisation (left filled arrow) and amounted to 80 ± 4.9 % (PTP₁). Twenty stimuli were then delivered at a frequency of 0.5 Hz. The stimulation was then resumed at 2 min post-tetanus (8 ± 3.9 %) and the decline in potentiation was followed for 30 min (6 ± 3.7 %). The slices were tetanised again (right filled arrow) and the amplitude of PTP₂ was estimated after 2 s (80 ± 11.0 %). The stimulation was resumed at 2 min post-tetanus (9 ± 6.9 %) and the potentiation was followed for the next 30 min (3 ± 6.9 %). The dashed line indicates the control level. Cb, the decay of the PTP₁ (from Ca) is shown on another time-scale ( circle ). PTP₁ declined in a mono-exponential manner (continuous line) with a time constant ( $tau$ ) of 6.2 ± 0.5 s.

On the basis of the above we concluded that the post-tetanic population spikes might affect the peaks of the f-EPSPs rather than their initial phases (Fig. 1A, illustrated in Fig. 11 too). We knew also that AVs are nearly symmetrical around their peaks (see sample traces in Fig. 9 and Fig. 10). Hence, the efficacy of synaptic transmission can be quantified by measuring the deflection of the f-EPSP at the point immediately after the termination of the AV, which is not affected by the population spikes (indicated by the dashed line). To do that, we estimated the rate of rise of the control f-EPSP in terms of the deflection of the slope of the f-EPSP (at the point just after the termination of the AV). This amplitude was then divided by the time interval from the peak of the AV to the point that was measured on the f-EPSP. Relative changes in synaptic transmission were evaluated on a 100 % scale in this and all the subsequent experiments. This was done by normalising the initial rising rates of the post-tetanic f-EPSPs (Fig. 1A), which were calculated at the same latency as the control rising rates, to their respective pre-tetanic controls. The actual gain in synaptic transmission (potentiation) was calculated by subtracting the 100 % level (Fig. 1B).

Potentiation, measured 2 min after its induction by theta-burst stimulation in standard medium, was 128 ± 8.8 % (P_max, n = 16, Fig. 1B). The potentiation declined hereafter and exposed the STP (81 ± 6.7 %). The decay of STP lasted for about 1 h before a sustained phase of potentiation was reached (LTP, 47 ± 3.7 %).

The main objective of this investigation was to describe STP and LTP, the induction of which is dependent on NMDA receptor activation during tetanisation. Post-tetanic potentiation (PTP) is another type of synaptic potentiation that can be induced in the CA1 area of the hippocampus by tetanisation. PTP is different from both STP and LTP in that its induction is independent of the NMDA receptor activation. It is observed immediately after tetanisation, declines over a relatively short period of time (Stevens et al. 1994) and reflects the clearance of Ca²⁺ that accumulates in the presynaptic terminals during the tetanisation (Zucker, 1999). Although it is believed that PTP declines rapidly we were interested in knowing the exact extent to which PTP might affect the measurements of STP and LTP in our experimental conditions.

The NMDA receptor antagonist D-AP5 (50 µM, Fig. 1Ca, n = 16) was applied in a group of experiments in order to investigate PTP in isolation from STP and LTP. After recording the control period, the slices were tetanised and 2 s after the tetanisation 20 stimuli were delivered at a frequency of 0.5 Hz. Then, the stimulation was stopped for 80 s and was resumed again at 2 min post-tetanus at the control frequency of 1/15 s. After a 30 min recording the slices were tetanised again (tetanus 2, Fig. 1Ca) and the amplitude of potentiation was estimated at 2 s after the tetanisation. The stimulation was then discontinued and resumed again at 2 min post-tetanus.

The results showed that 2 s after the first tetanisation, the amplitude of the PTP was 80 ± 4.9 % (PTP₁, Fig. 1Ca). PTP declined in a mono-exponential fashion with a time constant of 6.2 ± 0.5 s (Fig. 1Cb) and 2 min after the tetanisation the amplitude of the potentiation was 8 ± 3.9 %. This amplitude was greater than the baseline (P = 0.05, paired t test) and, over the next 30 min, declined to 6 ± 3.7 %, which was not different from the baseline (P > 0.1). The second tetanisation induced PTP₂ (80 ± 11.0 %), which was of a similar magnitude when compared with that of PTP₁ (P > 0.9, paired t test, Fig. 1Ca). Two minutes after the second tetanisation, the amplitude of potentiation was 9 ± 6.9 % and similar to that induced by the first tetanisation (P = 1, paired t test). Potentiation declined to 3 ± 6.6 % after 30 min.

In summary, the data show that STP and LTP can be isolated from the PTP by delaying post-tetanic test stimulation for a few minutes and that little, if any, STP and LTP can be induced by tetanisation after blockage of NMDA receptors by AP5. The results also show that PTP declines rapidly and in a manner that is independent of the application of the test stimulation during the post-tetanic period.

Long-term storage of the transient phase of potentiation

We wanted to determine whether or not STP, like PTP, decayed over time. To do this, we estimated the amplitude of the initial potentiation (isolated from PTP) from an average of four f-EPSPs evoked 2 min after tetanisation (P_max, Fig. 2A). We then introduced an additional delay in the post-tetanic test stimulation. Resuming the test stimulation after 1 h revealed a phase of potentiation that declined to a stable level of LTP. Both the transient phase of potentiation and the LTP were similar to those of the control (Fig. 2B). This indicates that, in contrast to PTP (Fig. 1Ca), both STP and LTP can be stored. It further suggests that the decay of STP is triggered by the application of the test stimulation. Because of this, terms such as short-term potentiation or early-LTP seem inadequate to describe the persistent nature of the transient phase of LTP. We have therefore decided to refer to this phase of LTP as 'transient long-term potentiation' (t-LTP) throughout the rest of this article and refer to the stable phase of LTP as 'sustained long-term potentiation' (s-LTP).

View larger version
[in this window]
[in a new window]

Figure 2. Transient and sustained phases of potentiation
A, a single experiment ( circle ) in which, after recording the control period (Ctrl), the slice was tetanised (theta-burst stimulation, filled arrow) and the mean amplitude of potentiation was recorded at 2 min post-tetanus (P_max). Test stimulation (1/15 s) was then discontinued for a period of 1 h. After resuming the test stimulation, transient (t-LTP) and sustained (s-LTP) phases of synaptic potentiation were followed for a period of 1.5 h. Potentiation was fitted to a mono-exponential function (see text), which is shown superimposed on the data (thick line). Constants (P_t₀, t-LTP, s-LTP and $tau$ ) that were estimated by the fitting are visualised in the histograms. B, an experiment similar to that in A except that the test stimulation was not delayed after the P_max amplitude was recorded.

In a conventional analysis of LTP experiments, the amplitude of the sustained synaptic potentiation (s-LTP) is estimated from an average of its value at the end of a recording session (as in Fig. 1B). An estimate of this type is not suitable to describe the transient phase of long-term potentiation (t-LTP). From the individual experiments it appears that the transient phase of potentiation can be described by a mono-exponential function. The exponential decay function (P = s-LTP + t-LTP e^-t/) could represent the individual data sets (Fig. 2A and B, thick line) given that t₀ (time zero) was defined as the time when test stimulation was initiated (excluding the P_max). A fitting routine was used to estimate the amplitude of the transient (t-LTP) and sustained (s-LTP) synaptic potentiation. Furthermore, it was also used to resolve the decay time constant of the transient phase ( $tau$ ). The total amplitude of potentiation at time zero was P_t₀ = t-LTP + s-LTP. Constants, which were derived by fitting the potentiation to the above function, are depicted in the histograms in Fig. 2A and B.

Groups of experiments, consisting of controls (i.e. 2 min delay before the test stimulation, Fig. 3A) and experiments with a delay of test stimulation after recording of P_max (Fig. 3B and C, filled circles), were used to confirm the storage of the potentiation. Mean estimates of the constants obtained from fitting the mono-exponential function to the individual experiments (as in Fig. 2A and B) are summarised in Fig. 3D and E. These data and the mean estimates of P_max were used to make statistical evaluations of different experimental situations.

View larger version
[in this window]
[in a new window]

Figure 3. Long-term storage of the transient phase of potentiation
A, a group of 13 experiments in which potentiation of synaptic transmission was tested after a 2 min delay in test stimulation after tetanisation (as in Fig. 2B, ±S.E.M.). Mean amplitudes of potentiation that were estimated in the individual experiments were: P_max = 118 ± 11.5 %, P_t₀ = 116 ± 10.7 %, t-LTP = 71 ± 10.6 %, s-LTP = 44 ± 6.3 %. The $tau$ of t-LTP was 20 ± 1.2 min. Averaged f-EPSPs, from the experiment in Fig. 2B, for the time periods as indicated by the letters, are displayed in the inset together with the calibration bar. B, in a group of 12 experiments the testing of potentiation was delayed for a period of 1 h (as in Fig. 2A) after recording the P_max ( filled circle ). P_max = 125 ± 17.1 %, P_t₀ = 110 ± 11.7 %, t-LTP = 69 ± 9.0 %, s-LTP = 41 ± 7.3 %. The $tau$ of t-LTP was 20 ± 3.2 min. Averaged f-EPSPs, from the experiment in Fig. 2A, are displayed in the inset. circle , control experiments without tetanisation (n = 11). In these, resuming stimulation after 1 h revealed a small facilitation of the f-EPSPs (18 ± 4.7 %) that decayed ( $tau$ = 4.4 ± 0.9 min) to -6 ± 5.4 % of potentiation. C, experiments in which stimulation was discontinued for a period of 6 h (n = 6). In order to ensure a correct estimation of potentiation, single experiments in this group were adjusted for changes in the afferent fibre volley off-line. P_max = 148 ± 21.3 %, P_t₀ = 100 ± 19.9 %, t-LTP = 59 ± 9.7 %, s-LTP = 42 ± 22.4 % and $tau$ of t-LTP = 19 ± 4.0 min. D, mean amplitudes of potentiation that were reached under the experimental conditions in A, B and C (±S.E.M.). Entire bars represent P_t₀, filled bars s-LTP and open bars t-LTP (according to the histograms in Fig. 2). The amplitudes of potentiation in the experiments with either 1 h (B) or 6 h (C) delay in test stimulation after tetanisation, were not different from those in the control group (A, P 0.5, all cases). These amplitudes were different from those associated with facilitation of the f-EPSPs in the non-tetanised control (B-Ctrl, P < 0.0001, all cases). Group sizes are indicated in parentheses. E, decay time constants of t-LTP in B (B) or C (C) were not different from those in A (A, P 0.8, both cases) but were different from the $tau$ values in experiments without tetanisation (B-Ctrl, P < 0.01, both cases). Note that only 6 of 11 non-tetanised experiments could be fitted by the mono-exponential function.

Initial potentiations (P_max), recorded 2-4 min after tetanisation in the experimental groups (Fig. 3B and C), were similar to those of the control (Fig. 3A, P 0.2), indicating that the initial potentiations were uniformly expressed in the three experimental situations. In spite of the additional delay in the test stimulation in the experimental groups, the amplitudes of P_t₀, t-LTP and s-LTP were not significantly different from those of the control (Fig. 3D). In addition, in all the groups t-LTP had a similar decay time constant of about 20 min (Fig. 3E). Varying the delay of test stimulation between 0.5 and 6 h did not change the expression of t-LTP.

It has been reported that a period without stimulation in magnesium-free medium can induce a transient form of synaptic potentiation (Niu et al. 1999). We therefore determined whether or not a silent period, under the experimental conditions used in our experiments, could induce potentiation. (Fig. 3B, open circles). In these experiments a small facilitation of the f-EPSPs was detected after a 1 h period without stimulation (18 ± 4.7 %, n = 11). In five out of the 11 experiments the fitting routine did not resolve a decay of potentiation. In the remaining six experiments the responses stabilised after 4.4 ± 0.9 min. Both the facilitation of the f-EPSPs and the decay time constants, which were estimated by the fitting routine, were different from the respective estimates of the tetanus-induced potentiation (P < 0.01, Fig. 3D and E).

We also determined whether or not the apparent decrease in the total amplitude of potentiation (P_t₀, Fig. 3C) represented a true reduction of potentiation over time or, alternatively, could be explained as a run down. The reduction of potentiation was evaluated in terms of the difference between P_max and the total potentiation obtained from our fitting routines (P_max - P_t₀) in the individual experiments. A slight decrease in P_t₀ was observed in the standard experiments (2.9 ± 3.7 %) in which potentiation was tested with a 2 min delay in test stimulation. P_t₀ was further reduced when potentiation was tested after a 1 h delay (14.7 ± 13.5 %, P = 0.4) and a significant decrease was seen in experiments in which testing was delayed for 6 h (47.9 ± 14.0 %, P < 0.001). In the experiments in which there was no tetanisation, the rate of rise of f-EPSPs was reduced by 6.0 ± 5 % to a potentiation of -6 % (Fig. 3B, open circles) during an experimental period of 2 h. The run down of approximately 3 % per hour, together with an increased variance in the long-lasting experiments, may account for some of the reduction of the total amplitude of potentiation.

In summary, our data show that the transient phase of LTP can be preserved over a relatively long period of time (> 6 h). This long-term storage of synaptic potentiation outlasts the time course of protein synthesis-dependent consolidation of the sustained phase of LTP (Frey et al. 1988; Frey & Morris, 1998).

Decay of transient long-term potentiation is dependent on post-tetanic control stimulation

The results of the above experiments demonstrate that t-LTP can be stored for long periods of time and imply that its decline is caused by post-tetanic test stimulation. In order to test the latter, the decay of t-LTP was investigated by applying test stimulation at frequencies of 1/7.5 s (Fig. 4A) and 1/30 s (Fig. 4B). The data were compared with those from the standard experiments in which the post-tetanic stimulation was given at 1/15 s (Fig. 3A).

View larger version
[in this window]
[in a new window]

Figure 4. Transient LTP decays in an activity dependent manner
A, potentiation, induced by theta-burst stimulation (filled arrow), was investigated with test stimuli delivered once every 7.5 s (n = 14) and resulted in P_t₀ = 104 ± 10.2 %, t-LTP = 58 ± 7.7 %, s-LTP = 46 ± 7.4 %. The $tau$ of t-LTP was 10 ± 1.6 min. B, potentiation tested with test stimulation delivered once every 30 s (n = 14). P_t₀ = 112 ± 12.0 %, t-LTP = 71 ± 8.6 %, s-LTP = 41 ± 6.9 %. The $tau$ of t-LTP was 39 ± 3.1 min. C, no differences were found between the amplitudes of potentiation (P_t₀, t-LTP and s-LTP) in A (A) and B (B) and those in standard experiments tested at 1/15 s (Fig. 3A, P 0.3 in all cases). Data, imported from the preceding figures, are presented as dashed bars in this and all the following histograms. D, the decay time of t-LTP, tested without a delay in stimulation, was inversely related to the frequency of test stimulation. The $tau$ of t-LTP tested at 1/7.5 s (A), and the $tau$ of t-LTP tested at 1/30 s (B), were different from the $tau$ of t-LTP tested at 1/15 s (Fig. 3A, P < 0.0001, both cases). E, the mean numbers of stimuli that were needed to reduce t-LTP to 37 % of its amplitude were constant: A, 79.7 ± 12.6 stimuli at 1/7.5 s; Fig. 3A, 79.0 ± 4.7 stimuli at 1/15 s; B, 74.9 ± 6.6 stimuli at 1/30 s; t-LTP tested at 1/15 s, with a delay in test stimulation for either 1 h (Fig. 3B) or 6 h (Fig. 3C), was reduced to 37 % of its amplitude in response to 78.6 ± 12.9 stimuli and 76.4 ± 16.0 stimuli, respectively. F, the numbers of stimuli needed to decrease t-LTP to 37 % of its amplitude were not correlated (P = 0.9, ANOVA, R = 0.01) to the amount of initial potentiation (P_max).

Analysis of the group with the faster test stimulation (1/7.5 s) showed that the decay time of t-LTP decreased twofold in comparison with that of the control group (P < 0.0001, Fig. 4D). We found no differences between the amplitudes of P_t₀ or between the amplitudes of t-LTP and s-LTP when compared between the two groups (P > 0.3, all cases, Fig. 4C). The results from the experiments with slow test stimulation (1/30 s), in contrast to the preceding results, showed a twofold increase in the decay time of t-LTP, when compared with that of the control (P < 0.0001, Fig. 4D). P_t₀, and the amplitudes of t-LTP and s-LTP were again not different from those of the control (P > 0.7, Fig. 4C). In conclusion, the results showed that the decay time of t-LTP was inversely related to the frequency of stimulation during the post-tetanic test period (Fig. 4D).

We estimated the number of stimuli that were needed to decay t-LTP to 37 % of its maximal amplitude (Fig. 4E). Irrespective of either the stimulation frequency (Fig. 3A, Fig. 4A and B) or the length of the time periods during which there was no stimulation (Fig. 3B and C), similar numbers of stimuli were needed to decrease t-LTP, when compared with those that were used in the control group, in which potentiation was tested at 1/15 s (Fig. 4E, P > 0.6 in all cases). Therefore, the decay of t-LTP was directly related to the number of the post-tetanic stimuli and, on average, 78 ± 4.5 stimuli were needed to decrease t-LTP to 37 % of its maximal amplitude.

It has been suggested that the decay time of STP depends upon the size of the initial potentiation (P_max, Malenka, 1991). We tested this in our data set by examining the relationship between P_max and the number of stimuli that were needed to reduce t-LTP to 37 % of its maximal amplitude. Linear regression analysis showed that P_max and the decay of t-LTP were independent variables (Fig. 4F).

In spite of a change in either the frequency of the post-tetanic test stimulation or in the duration of the time periods during which there was no stimulation, the mean amplitudes of P_max, t-LTP and s-LTP did not change. We were interested, therefore, in whether or not the amplitudes of P_max, t-LTP and s-LTP were dependent on each other. Data from the single experiments (Fig. 3 and Fig. 4) were tested by a linear regression analysis (n = 59). The amplitudes of both t-LTP and s-LTP were linearly dependent on the initial amounts of potentiation in slice preparations (Fig. 5A and B, P < 0.001 for both cases). Notably, t-LTP and s-LTP were expressed independently of each other in the single experiments (P > 0.6, R = 0.06, Fig. 5C).

View larger version
[in this window]
[in a new window]

Figure 5. Dependence of transient and sustained LTP on the ability of slices to express potentiation
A, the amplitude of t-LTP in single experiments (Figs 3 and 4, n = 59) was positively correlated with the initial amount of potentiation (P_max) induced in slice preparations (P < 0.0001, ANOVA, R = 0.63). B, the amplitude of s-LTP was dependent on the amplitude of P_max (P < 0.001, ANOVA, R = 0.45). C, the amplitudes of t-LTP and s-LTP were expressed independently of each other (P > 0.6, ANOVA, R = 0.06).

Transient long-term potentiation is synapse specific

It is currently believed that changes of specific synaptic inputs underlie the encoding of memories in the brain (Martin et al. 2000). Both STP and LTP can be induced in axonal pathways that converge independently on the same neurons (Andersen et al. 1980). If t-LTP is involved in memory processes, this type of specificity might not only include its induction but also its storage and its decay. We tested this, in 'two pathway' experiments (see Methods) by delaying test stimulation in the tetanised pathway for a period of 1 h (Fig. 6A). Our results showed that t-LTP could be induced and stored in an input-specific manner, despite continuous activation of synapses in the non-tetanised pathway. The amplitudes of P_t₀, t-LTP and s-LTP were similar to those of the standard experiments, in which testing of potentiation was discontinued for a period of 1 h (Fig. 3B, P 0.2 for all cases). In contrast, t-LTP decreased faster when compared with the decay time constant of the control (Fig. 6D).

View larger version
[in this window]
[in a new window]

Figure 6. Transient LTP can be stored in synaptic sub-populations
A, two-pathway experiments (n = 11) in which potentiation was induced in one of the pathways ( filled circle ) with theta-burst stimulation (filled arrow). Potentiation was tested 1 h after tetanisation at a frequency of 1/15 s. In the tetanised pathway: P_t₀ = 106 ± 15.0 %, t-LTP = 57 ± 6.4 %, s-LTP = 50 ± 11.2 % and the $tau$ of t-LTP = 10.9 ± 2.2 min. The non-tetanised control pathway ( circle ) showed little change. Average f-EPSPs from one of the experiments are shown in the inset. Ba, single-pathway experiments (n = 16) tested with two different stimulation intensities (Low = 2 times threshold and High = 4 times threshold for evoking f-EPSPs). The open arrow indicates intensity change. Tetanisation (filled arrow) was given at high intensity. Potentiation was tested at low intensity for 1 h followed by a 1 h test at high intensity. Data were normalised to their respective controls and presented as potentiation. Potentiation at low intensity ( filled circle ) was: P_t₀ = 107 ± 5.5 %, t-LTP = 62 ± 6.5 % and s-LTP = 45 ± 6.6 %. The $tau$ of t-LTP was 14.8 ± 1.7 min. Potentiation at high intensity ( circle was: P_t₀ = 78 ± 9.2 %, t-LTP = 43 ± 9.9 % and s-LTP = 35 ± 4.5 %. The $tau$ of t-LTP was 12.4 ± 2.0 min. Bb, subtraction (see text) was used to estimate t-LTP and s-LTP in synapses that were additionally recruited (Calc, diamond ) by the high intensity test stimulation. P_t₀ = 99 ± 12.5 %, t-LTP = 64 ± 10.6 %, s-LTP = 36 ± 7.8 % and the $tau$ of t-LTP = 12.5 ± 2.0 min. Potentiation that was tested with low intensity stimulation is reproduced from Ba for comparison ( filled circle ). C, the amplitudes of P_t₀, t-LTP and s-LTP from the tetanised pathway in A (A), from the low stimulation intensity (Ba-L) and the high stimulation intensity (Ba-H) pathways in Ba, and from the calculated pathway (Bb-C) in Bb. The changes in synaptic transmission that were caused by an increase in the stimulation intensity during the control periods (Bb-Ctrl) in Bb were different from the potentiations (Ba-L, Ba-H and Bb-C) obtained after the tetanisation (P < 0.001, all cases, paired t test). D, the decay time constants of t-LTP that were found in the experiments in A (A) were different from those of t-LTP that were tested 1 h after tetanisation in the control experiments (Fig. 3B, dashed bar, P < 0.05). The decay time constants of t-LTP from the experiments in B (Ba-L, Ba-H and Bb-C) were different from those that were caused by an increase in the stimulus intensity during the control periods (Bb-Ctrl, P < 0.01, in all cases, paired t test).

In another set of experiments, we attempted to determine whether or not t-LTP, once induced and stored, could be used in an input-specific manner. In order to achieve this, two synaptic populations were recruited with a single stimulation electrode. A first synaptic population was assessed with stimuli at low intensity (Low, Fig. 6Ba). Increasing the stimulus intensity recruited the first population and an additional synaptic population (High, Fig. 6Ba). Both synaptic populations were then tetanised, using the high intensity stimulation, and the P_max was recorded. The potentiation was then tested using the low intensity stimulation. After a period of 1 h, the high intensity stimulation was applied in order to evaluate potentiation in synapses that remained untested during the period of low intensity stimulation.

The f-EPSPs that were evoked by the high intensity stimulation consisted of the summed responses from two synaptic populations. In order to separate the respective contributions of the two synaptic populations the f-EPSPs that were evoked by the low intensity stimulation were subtracted from the f-EPSPs that were evoked by the high intensity stimulation. This was done in the individual experiments. In the control period, the calculated synaptic response was estimated by subtracting the average of the low intensity responses from the individual responses obtained at high intensity. After the low intensity post-tetanic stimulation, t-LTP has almost completely decayed in the tested synapses. The potentiation in the untested synapses (Calc), therefore, could be calculated as the response at high intensity minus the average s-LTP at low intensity. The calculated data were normalised to the control periods set to 100 % and the amplitudes of the potentiation were calculated by subtracting the control level (100 %) from the normalised rising rate of f-EPSPs (Calc, Fig. 6Bb).

Statistical evaluation confirmed that t-LTP and s-LTP were present to the same extent in both the low and calculated synaptic populations (Fig. 6C, P 0.4, paired t test) and the decay time constants of t-LTP were not different when compared between the groups (Fig. 6D, P 0.3, paired t test). Changing the stimulation intensity during the control periods resulted in a brief facilitation of the f-EPSPs. This facilitation was in all respects different from the t-LTP that was induced by tetanisation (Fig. 6C and D). This indicates that t-LTP, after its induction, can be temporally separated and used in a synapse-specific manner.

Induction of transient and sustained long-term potentiation

Earlier studies have suggested that different numbers of high frequency stimuli are needed to induce STP and LTP (Schulz & Fitzgibbons, 1997). The effects of high frequency stimulation are dependent on various experimental conditions (Hanse & Gustafsson, 1992). Out of those, three basic factors have to be considered: (1) the intensity of stimulation, (2) the frequency of stimulation, and (3) the number of stimuli given in a train.

We investigated whether or not t-LTP and s-LTP could be separated in relation to their induction by using two different frequencies (30 or 200 Hz) and two different numbers of stimuli (8 or 200 stimuli) during tetanisation. Each experimental group comprised eight experiments and tetanisation was given as a single train of continuous stimuli (Fig. 7). The amplitude of s-LTP increased when the number of stimuli that were used for induction was increased from 8 (Fig. 7A and C) to 200 (Fig. 7B and D). In contrast, larger amplitudes of t-LTP were seen after stimulation at 200 Hz when compared with those after 30 Hz stimulation. Despite the difference in the expression of t-LTP and s-LTP in the four groups, the decline in t-LTP always followed a mono-exponential time course.

View larger version
[in this window]
[in a new window]

Figure 7. Induction of transient and sustained LTP
Experiments (n = 8, circle ) showing t-LTP and s-LTP expressed after different patterns of induction as depicted in the insets. Tetanisation (filled arrow) was: A, 8 stimuli at 30 Hz; B, 200 stimuli at 30 Hz; C, 8 stimuli at 200 Hz and D, 200 stimuli at 200 Hz. Mono-exponential decaying functions were fitted to the mean potentiations in the experimental groups and are shown superimposed on the data (black lines). The calculated constants, given as best fit ± estimated error of constant were: A, t-LTP = 37.0 ± 9.3 %, s-LTP = 20.3 ± 1.7 % and $tau$ = 8.2 ± 2.8 min. The correlation coefficient R was 0.93. B, t-LTP = 37.7 ± 5.12 %, s-LTP = 57.3 ± 3.1 %, $tau$ = 32.9 ± 11.1 min and R = 0.99. C, t-LTP = 52.7 ± 7.2 %, s-LTP = 20.5 ± 1.2 %, $tau$ = 8.32 ± 1.5 min and R = 0.98. D, t-LTP = 70.3 ± 3.7 %, s-LTP = 56.7 ± 2.1 %, $tau$ = 29.3 ± 3.6 min and R = 0.99.

In order to quantify the effects of the induction parameters on the expression of t-LTP and s-LTP, we extended the data set by adding additional groups of experiments and depicted the results in 3D diagrams (Fig. 8). The relationships between the frequencies of stimulation and P_max, t-LTP, s-LTP and $tau$ were statistically evaluated using pooled data from the 30 Hz (n = 36) and the 200 Hz (n = 36) groups. The pooled data from groups with 8 (n = 24) and 200 stimuli (n = 24) were used to evaluate the effects of the number of stimuli per train on the expressions of potentiation. The data were tested for differences using a two-sample Kolmogorov-Smirnov test.

View larger version
[in this window]
[in a new window]

Figure 8. Relationships between tetanisation and potentiation
A, relationship between P_max, frequency of tetanisation and the number of induction stimuli. filled circle , mean estimates from single experiments in the separate groups (error bars = S.E.M.). The grid shows the mean functional relationship between expression of potentiation and the induction parameters. The numbers of experiments per group were: 8 stimuli at 30-200 Hz (n = 8), 16 stimuli at 30 and 200 Hz (n = 6), 40 stimuli at 30-200 Hz (n = 8), 100 stimuli at 30-200 Hz (n = 6) and 200 stimuli at 30-200 Hz (n = 8). B, similar to A, showing that the amplitude of t-LTP is dependent on the frequency of tetanisation and independent of the number of stimuli given. C, the amplitude of s-LTP is dependent on the number of induction stimuli and independent of the frequency of tetanisation. D, decay time constants of t-LTP ( $tau$ ) are related to the number of induction stimuli and independent of the frequency of tetanisation.

Analysis showed that P_max increased with both the number of stimuli per train (P < 0.05) and the frequency of tetanisation (P = 0.01, Fig. 8A). The expression of t-LTP was largely dependent on the frequency of the applied stimuli (P < 0.01, Fig. 8B) and appeared somewhat reduced after long trains of low frequency tetanisation. However, t-LTP was independent of the number of the induction stimuli (P = 0.9). The amplitude of s-LTP was, in contrast to that of t-LTP, strictly related to the number of stimuli in a train (P < 0.001, Fig. 8C). In the tested range of frequencies, this relationship was linear and independent of the frequency of tetanisation (P = 0.98). The decay of t-LTP ( $tau$ ) was related to the number of stimuli that were applied in a train (P < 0.001, Fig. 8D), but was not related to the frequency of tetanisation (P = 0.9). A correlation analysis confirmed that the amplitudes of t-LTP and s-LTP were unrelated (R = 0.17). Expressions of t-LTP and s-LTP could therefore be separated in relation to their induction parameters.

Effects of glutamate receptor antagonists on the decline of transient long-term potentiation

Changes in the phosphorylation of $alpha$ -amino-3-hydroxy-5-methyl-4-isoxolepropionic acid (AMPA) receptors (Stanton, 1996; Soderling & Derkach, 2000) and/or the insertion and recycling of the postsynaptic AMPA receptors (Nishimune et al. 1998; Noel et al. 1999; Hayashi et al. 2000) may be involved in the expression and maintenance of LTP (Mainen et al. 1998; Luscher et al. 1999, 2000; Carroll et al. 1999). Internalisation of AMPA receptors, mediated by glutamate binding, is also an attractive mechanistic explanation for the stimulation-dependent decay of t-LTP. This hypothesis was tested by investigating whether or not t-LTP would be preserved if test stimuli were applied in the presence of blockers of glutamate receptor-mediated synaptic transmission.

We first tested whether or not the blockage of ionotropic glutamate receptors, during maintenance of t-LTP, would affect its expression (Fig. 9A). After tetanisation, the initial potentiation (P_max) was recorded and stimulation was stopped. The ionotropic glutamate receptor antagonist kynurenate (5 mM) was then applied for a period of 25 min followed by a 35 min washout of the drug. One hour after tetanisation, test stimulation was resumed and the expressions of t-LTP and s-LTP were recorded. Potentiations obtained in this group of experiments were compared with those observed in the experiments without kynurenate perfusion (Fig. 3B). There were no differences between the two groups (Fig. 9D and E). The speed of clearance of kynurenate from the perfusion chamber was tested in a separate set of control experiments (n = 2, not shown). In these experiments test f-EPSPs were blocked by application of 5 mM kynurenate for a period of 25 min. The f-EPSPs returned to their control values with a time constant of 11.6 ± 3.6 min during the washout of the antagonist.

View larger version
[in this window]
[in a new window]

Figure 9. Glutamate receptor antagonists do not prevent decay of transient LTP
A, experiments (n = 12) in which t-LTP and s-LTP were induced by theta-burst stimulation (filled arrow). After the induction of potentiation P_max was recorded (132 ± 14.6 %) and kynurenate (5 mM, continuous line) was applied for 25 min and then washed out (35 min). After the wash-out of kynurenate potentiations were: P_t₀ = 121 ± 13.3 %, t-LTP = 67 ± 14.1 % and s-LTP = 55 ± 6.9 %. The $tau$ of t-LTP was 18 ± 4.0 min. B, similar to A except that a series of 80 test stimuli (1/7.5 s, dashed line) were delivered in the presence of kynurenate. 5 mM kynurenate blocked f-EPSPs 15 min after its application (inset trace b). Note that the amplitude of t-LTP was reduced compared with that in A, when tested after the wash-out of kynurenate. P_max = 145 ± 11.4 %, P_t₀ = 76 ± 9.5 %, t-LTP = 26 ± 5.3 %, s-LTP = 51 ± 5.7 % and $tau$ of t-LTP = 11.5 ± 2.7 min. C, similar to B except that the metabotropic glutamate receptor antagonist LY341495 (100 µM) was applied together with kynurenate. P_max = 135 ± 13.8 %, P_t₀ = 64 ± 8.5 %, t-LTP = 25 ± 8.2 %, s-LTP = 40 ± 7.5 % and $tau$ of t-LTP = 6.0 ± 2.0 min. D, P_t₀, t-LTP and s-LTP in the control experiments (Fig. 3B) were compared with the respective amplitudes of potentiation in A, B and C. Application of kynurenate in A (A) did not affect the amplitudes of potentiation (P > 0.2, all cases). The experimental groups B and C showed lower P_t₀ and t-LTP (P < 0.05, all cases), whereas the amplitudes of s-LTP were similar to those of the control (P > 0.2, in both cases). E, decay time constants of t-LTP for the experimental groups in D. Decay time constants in A (A) and B (B) were similar to those of the control (P > 0.07). The $tau$ in C (C) was smaller than that of the control (P < 0.01).

In another group of experiments, we tested whether or not a series of 80 test stimuli (1/7.5 s), given after the blockage of the evoked f-EPSPs by kynurenate, would affect the amplitude of t-LTP (Fig. 9B). We chose to use 80 stimuli, because in the standard control experiments (Fig. 3B) this number of stimuli reduced the amplitude of t-LTP to 37 % of its initial amplitude (Fig. 4E). The amplitude of t-LTP, after washout of kynurenate, was 26 ± 5.3 % or only 38 % of the amplitude of t-LTP in the control experiments (69 ± 9.0 %, P < 0.001, Fig. 9D). The amplitude of t-LTP that was seen after 80 post-tetanic stimuli in the control experiments was, on average, 25 ± 5.2 % (Fig. 3B) and this amplitude of t-LTP was not significantly different from the reduced amplitude of t-LTP that was observed after stimulation in the presence of kynurenate (P = 0.9). The amplitude of s-LTP (Fig. 9B) was not affected by the application of the stimulation in the presence of kynurenate and similar to that of the control (Fig. 9D). When compared with that of the control the decay time constant of t-LTP was slightly, although not significantly, reduced after the stimulation in kynurenate (Fig. 9E). In conclusion, the reduction of t-LTP that was caused by presynaptic activation could not be prevented by blockage of the ionotropic glutamate receptors.

We tested whether or not the decay of t-LTP depended on an activation of the metabotropic glutamate receptors. To do this, we conducted a series of experiments in which 80 post-tetanic test stimuli were given in the presence of 5 mM kynurenate and 100 µM of the metabotropic glutamate receptor antagonist LY341495 (Fig. 9C). In these experiments, the amplitudes of t-LTP were reduced to an extent that was similar to that in Fig. 9B whereas the amplitudes of s-LTP were unaffected (Fig. 9D). The $tau$ of t-LTP was reduced when compared with that of the controls (Fig. 9E). Since a combination of kynurenate (5 mM) and LY341495 (100 µM) blocks all identified glutamate receptors (Fitzjohn et al. 1998; Bortolotto et al. 1999), our data suggest that the dynamic decay of t-LTP occurs independently of glutamate receptor activation and may be mediated by a presynaptic mechanism.

Modulation of transient and sustained long-term potentiation by calcium-free medium

The above experiments demonstrated that testing, during blockage of glutamate receptors, could reduce t-LTP and suggested that t-LTP was expressed presynaptically. Therefore, the decline of t-LTP might be dependent on the vesicular release of glutamate. If this were true, then t-LTP should not be reduced by testing during a presynaptic blockage of synaptic transmission. Otherwise, t-LTP might be dependent on some other changes in the presynaptic terminals that might affect the probabilities for neurotransmitter release without being directly related to the release of a neurotransmitter per se. Stimulus-dependent neurotransmitter release can be blocked reversibly by perfusion of Ca²⁺-free medium. This procedure, however, has been shown to disrupt LTP when applied shortly after tetanisation (Katsuki et al. 1997). The degree to which t-LTP is affected by Ca²⁺-free medium is currently not known.

In the first group of experiments we determined whether or not the perfusion of Ca²⁺-free medium, during the period when t-LTP is maintained, would affect its expression (Fig. 10A). After tetanisation, the P_max was recorded and the stimulation was stopped. Ca²⁺-free medium was then applied for a period of 25 min. Potentiation was tested 35 min after returning to the standard medium and the data were compared with those from the control experiments, which are presented in Fig. 3B. Analysis of the results, showed that the amplitudes of P_t₀, t-LTP and s-LTP were reduced when compared with those of the control experiments (P < 0.05, all cases, Fig. 10D). The amplitude of t-LTP (44 ± 6.0 %), however, decreased to a lesser extent than the amplitude of s-LTP (5 ± 5.6 %) when compared with those of the control (69 ± 9.0 % and 41 ± 7.3 %, respectively). Hence, 64 % of t-LTP was still present after the treatment with Ca²⁺-free medium, whereas only 12 % of s-LTP remained preserved. The decay time constants of t-LTP (10.8 ± 2.6 min) were also affected by treatment with Ca²⁺-free medium and were significantly smaller than those of the control (20 ± 1.2 min, P < 0.05, Fig. 10E). These results confirm that an early application of Ca²⁺-free medium can reduce s-LTP (Katsuki et al. 1997) and demonstrate that a substantial amount of t-LTP is preserved in spite of the treatment with Ca²⁺-free medium.

F10 View larger version
[in this window]
[in a new window]

Figure 10. Decrease of transient and sustained LTP in calcium-free medium
A, in a group of experiments (n = 12), potentiation was induced by theta-burst stimulation, P_max (147 ± 14.6 %) was recorded and Ca²⁺-free medium was applied for 25 min (0 mM Ca²⁺, 4 mM Mg²⁺, filled bar). Potentiations, 35 min after returning to normal medium, were: P_t₀ = 49 ± 7.2 %, t-LTP = 44 ± 6.0 %, s-LTP = 5 ± 5.6 % and $tau$ of t-LTP = 10.7 ± 2.6 min. B, similar to A ( filled circle , n = 12) except that 80 stimuli (1/7.5 s) were delivered during the last 10 min of the exposure to Ca²⁺-free medium (dashed line, inset trace b). P_max = 157 ± 15.3 %, P_t₀ = 48 ± 6.3 %, t-LTP = 43 ± 4.2 %, s-LTP = 5 ± 4.6 % and $tau$ of t-LTP = 11.2 ± 1.5 min. circle , experiments in which slices were not tetanised (n = 13). Testing, after returning to normal medium, revealed a small facilitation of f-EPSPs (2 ± 3.5 %) that decayed to a potentiation of -11 ± 5.1 % with a $tau$ of 4.1 ± 1.2 min. C, a group of experiments (n = 14) in which, 15 min after tetanisation, Ca²⁺-free medium was applied for 25 min. Testing, 20 min after returning to normal medium, showed both t-LTP and s-LTP: P_max = 133 ± 5.0 %, P_t₀ = 67 ± 8.7 %, t-LTP = 46 ± 6.0 % and s-LTP = 20 ± 4.8 %. $tau$ of t-LTP = 19.6 ± 2.5 min. D, the amplitudes of P_t₀, t-LTP and s-LTP in A (A) were smaller than those in the standard experiments (Fig. 3B, P < 0.05, all cases). The amplitudes of potentiation after the application of 80 stimuli in B (B) were not different from those in A (A) (P 0.9, all cases). Estimates of the amount of potentiation in the experiments in which slices were not tetanised (B-Ctrl) were different from those recorded in B (B) (P < 0.0001, all cases). A later application of Ca²⁺-free medium in C (C) resulted in a larger s-LTP (P < 0.05) when compared with s-LTP in B (B), whereas P_t₀ and t-LTP were not affected (P > 0.1, both cases). E, the $tau$ of t-LTP in A (A) was smaller than that of the control (P < 0.05, Fig. 3B). Decay time constants in B (B) were similar to those in A (A) (P = 0.9) and larger than the decay time constants observed in experiments without tetanisation (P < 0.001, B-Ctrl). Decay time constants of t-LTP in C (C) were larger than those in A (A) and B (B) (P < 0.05, both cases).

In the second group of experiments we tested whether or not the amplitude of t-LTP could be reduced by stimulation during the blockage of stimulus-evoked neurotransmitter release (Fig. 10B, filled circles). In contrast to the control experiments, which are shown in Fig. 10A, 80 stimuli (1/7.5 s) were applied during the last 10 min of the exposure to the Ca²⁺-free medium. Analysis of the results showed that the stimulation did not reduce the amplitude of t-LTP (43 ± 4.2 %) relative to that observed in experiments that did not include the application of the stimuli (P = 0.9, Fig. 10D). The amplitude of s-LTP (5 ± 4.6 %) and the decay time constants of t-LTP (11.2 ± 1.5 min) were also not affected by the application of the stimuli (P 0.9, Fig. 10D and E).

In a group of control experiments, we investigated the extent to which perfusion of Ca²⁺-free medium affected the baseline responses (Fig. 10B, open circles). In those experiments, the slices were not tetanised and Ca²⁺-free medium was applied as in the two previous groups. During blockage of synaptic transmission 80 stimuli were delivered. Analysis of the results showed a small enhancement of the recorded f-EPSPs (2 ± 3.5 %) that declined rapidly (4.1 ± 1.2 min) to a potentiation of about -11 % (Fig. 10D and E). All those estimates were smaller when compared with the respective estimates that were found in experiments in which slices were tetanised and treated by stimulation in Ca²⁺-free medium (P < 0.001, all cases, Fig. 10D and E).

It has been reported that a late application of Ca²⁺-free medium reduces the amplitude of LTP to a lesser extent when compared with the reduction of LTP that is observed after an early application of Ca²⁺-free medium (Katsuki et al. 1997). We investigated how t-LTP was affected by this experimental situation. The slices were tetanised, P_max was recorded and, 15 min later, Ca²⁺-free medium was applied for 25 min (Fig. 10C). After blocking of synaptic transmission, stimulation was applied in an attempt to reduce t-LTP. The late application of the Ca²⁺-free medium resulted in a greater amount of s-LTP (20 ± 4.8 %) when compared with that in the experiments in which Ca²⁺-free medium was applied early and stimulation was given during the blockage of neurotransmission (P < 0.05, Fig. 10D). However, the amplitude of s-LTP was still smaller than that which was found in experiments that did not involve the application of Ca²⁺-free medium (P < 0.05). The amplitude of t-LTP (46 ± 6.0 %) was not different (P > 0.6), whereas the $tau$ of t-LTP (20 ± 2.5 min) was larger (P < 0.01) when compared with those in experiments in which Ca²⁺-free medium was applied earlier (Fig. 10D and E). The decay time constants of t-LTP, in this group of experiments, were not different from those found in experiments that did not involve the application of Ca²⁺-free medium (P = 1).

In summary, our results confirm that the application of Ca²⁺-free medium can disrupt LTP and that treatment with Ca²⁺-free medium is most effective when applied shortly after tetanisation (Katsuki et al. 1997). In contrast, a substantial amplitude of t-LTP is preserved in spite of the treatment with Ca²⁺-free medium. This amplitude of t-LTP is resistant to stimulation during blockage of stimulus-evoked neurotransmitter release. t-LTP declines with a small decay time constant when Ca²⁺-free medium is applied shortly after tetanisation. However, a later application of Ca²⁺-free medium does not affect the decay kinetics of t-LTP. The findings from the experiments in which Ca²⁺-free medium was used support the hypothesis that both the expression and the decline of t-LTP involve changes in the intracellular Ca²⁺-mediated dynamics of the transmitter release, rather than some other types of change in the presynaptic terminals that do not involve the release of the neurotransmitter. The data furthermore suggest that t-LTP and s-LTP are different in that the two phenomena can be modulated independently of each other when Ca²⁺-free medium is applied.

Paired-pulse facilitation in the CA1 area of the hippocampus

The studies described above suggest that t-LTP is expressed presynaptically in that the decay of t-LTP can be evoked by stimulation, in spite of a full blockage of glutamate receptors (Fig. 9). In support of this, the stimulus-induced decay of t-LTP is prevented in the presence of Ca²⁺-free medium, which blocks stimulus-evoked neurotransmitter release, suggesting that intracellular Ca²⁺-mediated dynamics of neurotransmitter release might be involved in the decay of t-LTP (Fig. 10). The decay of t-LTP is dependent upon the number of stimuli given during the post-tetanic test period (Fig. 4 and Fig. 9). We therefore hypothesised that facilitating synaptic release, by means of increasing the presynaptic levels of [Ca²⁺]_i during the post-tetanic test period, would accelerate the decay of t-LTP.

When two pulses are given with a short inter-pulse interval (IPI), the synaptic release of transmitter, in a population of synapses, can be modulated in a predictable fashion (Stevens & Wang, 1995; Hjelmstad et al. 1997). The gain in synaptic transmission, from the first to the second pulse, can be evaluated as a paired-pulse facilitation (PPF). PPF, expressed as a percentage, is equal to (f-EPSP₂/f-EPSP₁ - 1) times 100 %. PPF is dependent on the IPI used during PP stimulation and is determined by: (1) the average release probability in a synaptic population, (2) the refractoriness of the synapses that have released a neurotransmitter and (3) the stimulation-dependent accumulation of Ca²⁺ in the presynaptic terminals. At very short IPIs (< 5 ms), PPF is negative because f-EPSP₂ is reduced due to a large number of refractory synapses. Synapses become capable of release after about 30 ms (Stevens & Wang, 1995; Hjelmstad et al. 1997). Maximal PPF is usually recorded at longer IPIs (40-100 ms) and is primarily dependent on the speed with which Ca²⁺ is cleared from the presynaptic terminals (Zucker, 1999; Emptage et al. 2001). These features of PPF provide a probe for investigating the relationship between the decay of t-LTP and the intracellular Ca²⁺-mediated dynamics of neurotransmitter release.

In order to evaluate the dependency of t-LTP on paired pulse (PP) stimulation we wanted to ensure that the rising phases of f-EPSPs, recorded in the stratum radiatum of the CA1 area, could be used reliably to estimate the amounts of PPF. More specifically, we were concerned about the ability of postsynaptic spiking to affect the rising rate of the second f-EPSP (f-EPSP₂) in a non-linear fashion when compared with that of the first f-EPSP (f-EPSP₁).

The above was tested in experiments in which PP stimulation was delivered with an IPI of 80 ms (Fig. 11A). After recording the control f-EPSPs (black waveforms 'a' and 'c', Fig. 11A) a perfusion medium that contained the GABA_A receptor antagonist picrotoxin (100 µM) was applied to the slices. Blockage of the GABA_A receptors increased the decay times of f-EPSP₁ and f-EPSP₂ and led to an enhancement of the population spikes, which were seen reflected in both potentials (red waveforms 'b' and 'd', Fig. 11A). The amplitudes of the f-EPSPs appeared to be reduced by the population spikes when compared with those before the application of the picrotoxin (as indicted by the open arrowheads in Fig. 11A). However, the initial parts of the rising phases of the f-EPSPs were not affected by the application of the antagonist indicating that the rising rates of the two potentials can be reliably quantified (as described in Fig. 1).

F11 View larger version
[in this window]
[in a new window]

Figure 11. Measurement of paired-pulse facilitation
A, superimposed f-EPSPs (indicated by the two arrows) from an experiment in which paired-pulse (PP) stimulation was given with an inter-pulse interval (IPI) of 80 ms. Black waveforms show control f-EPSP₁ (a) and f-EPSP₂ (c) in standard medium. Red waveforms (b and d) show the respective potentials after the application of the GABA_A receptor antagonist picrotoxin (100 µM). The dashed line indicates the point on the initial rising phase of the f-EPSPs that was used to analyse this and other experiments (as described in Fig. 1). The open arrowheads point to the decreases in amplitudes of the f-EPSPs that were caused by the application of picrotoxin. Calibration bar is shown in the inset. B, a group of experiments in which the effects of the picrotoxin on PPF were established (n = 16, IPI = 80 ms). Picrotoxin (100 µM) was applied at the time indicated by the red continuous line. The letter labels correspond to the sample waveforms presented in A. PPF, shown by the black continuous line (±S.E.M.), was 104 ± 3.6 % in control conditions and 103 ± 3.7 % after the application of picrotoxin (P = 0.8, paired t test). The filled circle show potentiation of f-EPSP₁ (9 ± 5.4 %) whereas the green circles show the potentiation of f-EPSP₂ (11 ± 4.7 %). The two potentiations were similar (P = 0.6, paired t test).

In a group of experiments, f-EPSP₁ was potentiated to a mean value of 9 ± 5.4 % when measured 20-30 min after the application of picrotoxin (black circles, n = 16, Fig. 11B). This enhancement of f-EPSP₁ was not reliably observed in all the experiments (e.g. Fig. 11A) and was not significantly different from the baseline (P > 0.1, paired t test). Potentiation of f-EPSP₂ was evaluated as that of f-EPSP₁, by normalising the rising rates of the f-EPSPs after the application of picrotoxin to their control values (green circles, Fig. 11B). f-EPSP₂ showed a similar amount of increase, when compared with that of f-EPSP₁ (11 ± 4.7 %, P = 0.6, paired t test). In support of this, the PPF (continuous line in Fig. 11B) before the application of picrotoxin was 104 ± 3.6 %, and this did not differ significantly from PPF after the application of GABA_A receptor antagonist (102 ± 3.7 %, P = 0.8, paired t test).

In conclusion, the initial rising phases of f-EPSP₁ and f-EPSP₂ and the corresponding measurement of PPF were not affected by the increase in postsynaptic spiking caused by picrotoxin.

Expression of transient and sustained long-term potentiation in paired-pulse facilitation experiments

In order to investigate the effects of PPF on the decline of t-LTP, PP stimulation was given once every 15 s with an IPI of either 20 ms (Fig. 12A) or 80 ms (Fig. 12B) during the control and during the post-tetanic periods of the experiments. Responses that were evoked by the first pulse were measured in terms of the rising rate of f-EPSP₁ and were presented as mean amplitudes of potentiation, in the same way as in the previous experiments (filled circles, Fig. 12A and B). The responses to the second pulse before tetanisation were amplified by PPF. However, after tetanisation they were affected both by expression of potentiation and by PPF. In order to present the dynamic changes in the responses to the second pulse, the rising rate of f-EPSP₂ was normalised to the control rising rate of f-EPSP₁ and the degree of the facilitated potentiation was determined by subtracting the control level of f-EPSP₁ (100 %). Note that in the control periods the facilitated potentiation of f-EPSP₂ is equal to the actual gain in synaptic transmission, which was caused by PPF (open triangles, Fig. 12A and B).

F12 View larger version
[in this window]
[in a new window]

Figure 12. Paired-pulse facilitation in relation to the expression of transient and sustained LTP
A, experiments in which slices were tested with paired-pulse (PP) stimulation (n = 16) delivered once every 15 s with an inter-pulse interval (IPI) of 20 ms. The figure shows the mean potentiation of f-EPSP₁ ( filled circle ) and the mean facilitated potentiation of f-EPSP₂ ( up triangle ). The filled arrow indicates the time of tetanisation. The amplitudes of potentiation of f-EPSP₁ were: P_t₀ = 102 ± 9.9 %, t-LTP = 62 ± 5.8 % and s-LTP = 39 ± 6.2 %. The $tau$ of t-LTP was 18.7 ± 2.2 min. Averaged f-EPSPs from a single experiment are shown in the inset. B, a group of experiments in which t-LTP and s-LTP were tested with an IPI of 80 ms (n = 16). P_t₀ = 124 ± 11.5 %, t-LTP = 76 ± 11.2 %, s-LTP = 48 ± 6.3 % and $tau$ of t-LTP = 20.0 ± 2.5 min. C, the $tau$ of t-LTP ( filled circle ), in the IPI range from 5 to 500 ms, was not dependent on the amount of PPF in the control periods of the experiments ( up triangle ) and was greater when compared with the $tau$ of t-LTP at 1/7.5 s (from Fig. 4A, indicated by the dashed line, P < 0.01, all cases). The $tau$ of t-LTP in experiments in which PP stimulation was given with an IPI of 1000 ms, was similar to that of t-LTP at 1/7.5 s (P > 0.5). The numbers of experiments are indicated in parentheses. D, PPF (±S.E.M.) before and after tetanisation for the experimental groups in A (20 ms IPI, thick line) and B (80 ms IPI, thin line). When PP stimulation was given with an IPI of 20 ms the PPF was reduced from a pre-tetanic control value of 103 ± 9.4 % to 26 ± 4.9 % after the tetanisation (P < 0.0001, paired t test). It increased progressively to 80 ± 6.2 % over a period of 2 h and remained different from the control PPF (P < 0.01, paired t test). In PP experiments, in which IPI was 80 ms, the PPF decreased from a baseline value of 111 ± 7.0 % to 61 ± 4.8 % after tetanisation (P < 0.0001, paired t test). The PPF increased to 102 ± 8.1 % over a period of 2 h, and was not different from the control PPF (P = 0.4, paired t test).

The amount of PPF (~100 %) recorded in the control periods of the experiments, in which PP stimulation was given with an IPI of either 20 or 80 ms, was similar (Fig. 12A and B, n = 16 in both groups). In both situations PP stimulation resulted in a release that was about three times larger than that which was seen with a single stimulus. PP stimulation, after the tetanisation, revealed that t-LTP, which was measured at pulse one, declined with a $tau$ value of about 20 min in both of the groups. The results were compared with those of the control group, in which the same number of stimuli as used in the PP stimulation conditions were applied as single stimuli with a frequency of 1/7.5 s (Fig. 4A). The estimates of $tau$ in the PP experiments were about two times larger than the decay time constants of the control group (10 ± 1.6 min, P < 0.01 for both cases). Similarly, the decay of t-LTP was longer than expected in the experiments in which the post-tetanic PP stimulation was delayed for a period of 30 min. In these experiments, the estimates of $tau$ were 18 ± 3.0 min and 16 ± 2.0 min when t-LTP was tested with an IPI of either 20 ms (n = 13) or 80 ms (n = 13), respectively. The amplitudes of t-LTP and s-LTP and the estimates of $tau$ in the experiments in which there was a 30 min delay in test stimulation, were similar to the respective estimates in the experiments without a delay in the test stimulation (P 0.3, all cases).

The relationship between the decay kinetics of t-LTP and the amount of pre-tetanic PPF was investigated in experimental groups in which PP stimulation was given with IPIs that ranged from 5 to 1000 ms (Fig. 12C). In the range of IPIs from 5 to 500 ms the decay time constants of t-LTP were all similar, that is about two times larger than the decay time constants that were observed in the controls (indicated by the dashed line in Fig. 12C). Furthermore, the $tau$ of t-LTP did not correlate with the amounts of PPF that were expressed in these experiments (Fig. 12C). In contrast, PP stimulation that was given with an IPI of 1000 ms showed little PPF (8.2 ± 1.4 %). In this case the $tau$ of t-LTP was 8.3 ± 2.1 min and was not different from that of the control experiments (P > 0.5). This indicates that the decline of t-LTP is not a simple function of the number of stimuli given, but can be modulated by both the frequency and the pattern of the test stimulation during the post-tetanic period.

In spite of similar amounts of PPF (~100 %, P = 0.5), the amplitude of the facilitated potentiation after tetanisation was smaller in the group in which PP stimulation was given with an IPI of 20 ms (Fig. 12A) than it was in the experiments in which PP stimulation was given with an IPI of 80 ms (Fig. 12B). This difference was reflected by changes in PPF after the induction of potentiation (Fig. 12D). The amount of PPF 2 min after the tetanisation was smaller in the PP experiments tested with an IPI of 20 ms when compared with that in the experiments in which potentiation was tested with an IPI of 80 ms (26 ± 4.9 % and 61 ± 4.8 %, respectively, P < 0.0001). In both groups the decay of t-LTP was accompanied by a progressive increase in the PPF towards its control values. However, 2 h after the tetanisation, the amount of PPF at 20 ms IPI remained smaller when compared with the amount of PPF at 80 ms IPI (P < 0.05, Fig. 12D). At this time after the tetanisation, the amount of PPF at 80 ms IPI was slightly, although not significantly, reduced when compared with the respective amount of PPF before the tetanus (Fig. 12D). In contrast, the amount of PPF at 20 ms IPI was smaller than the baseline amount of PPF (P < 0.01, paired t test). Persistent differences in the amounts of PPF (Fig. 12D) and of facilitated potentiation (Fig. 12A and B), between the groups that were tested with an IPI of 20 and 80 ms, can be explained by a post-tetanic increase in the proportion of refractory synapses at pulse two in the experimental group that was tested with an IPI of 20 ms. These observations indicate that there is an enhanced probability of neurotransmitter release after tetanisation (Stevens & Wang, 1995; Hjelmstad et al. 1997). They support the finding that t-LTP is expressed presynaptically (Fig. 9 and Fig. 10) and suggest that an enhanced probability of neurotransmitter release is also involved in the expression of s-LTP.

It has been shown previously that the expression of LTP depends on the amount of PPF before tetanisation, suggesting that LTP is dependent on the initial probability of neurotransmitter release in slice preparations (Kleschevnikov et al. 1997). It is not known if t-LTP shows a similar dependency, however. In order to investigate the dependencies of P_max, t-LTP, s-LTP and the $tau$ of t-LTP on the amount of pre-tetanic PPF, we analysed the data from experimental groups in which potentiation was tested with PP stimulation that showed a substantial amount of PPF (IPI from 20 to 500 ms, Fig. 12C). In each group, the PPF during the control period of the individual experiments was normalised to the mean PPF observed in that group. All single experiments were binned with respect to the normalised PPF (Fig. 13). Statistical analysis of this data set confirmed that the amplitudes of P_max and s-LTP were linearly dependent on the amount PPF that was seen before tetanisation (P < 0.02, Fig. 13A and C). The amplitude of t-LTP was largest in the group that showed the highest PPF (Fig. 13B). However, the amplitude of t-LTP was not linearly related to the amount of PPF (P > 0.05) and this relationship was more complex than that of s-LTP (Fig. 13C). In the group with the highest PPF the amplitude of s-LTP was about five times larger than that in the group with the lowest PPF. In contrast, in the group with the highest PPF, t-LTP increased by only about 1.5 times when compared with that expressed in the group with the lowest PPF. This indicates that the expressions of s-LTP and t-LTP are differentially dependent on the initial probability of neurotransmitter release in slice preparations. The $tau$ of t-LTP was not dependent on the amount of initial PPF (Fig. 13D).

F13 View larger version
[in this window]
[in a new window]

Figure 13. Initial probability of neurotransmitter release and expression of transient and sustained LTP
A, P_max was positively correlated (R = 0.980) to the amount of PPF in the control periods of the experiments. It was also linearly dependent on the amount of PPF (P < 0.02, ANOVA). B, t-LTP was positively correlated to the amount of PPF (R = 0.914), but not linearly dependent on the amount of PPF (P > 0.05, ANOVA). C, s-LTP was both positively correlated (R = 0.995) to the amount of PPF and linearly dependent on the amount of PPF (P < 0.01, ANOVA). D, the $tau$ of t-LTP was neither correlated to PPF (R = 0.095) nor dependent on its amount (P > 0.9, ANOVA).

Decay of transient long-term potentiation depends on activation of NMDA receptors

The unexpected slow decay of t-LTP that was observed in experiments involving PP stimulation suggested that the decay kinetics of t-LTP might be modulated by an unknown mechanism (Fig. 12C). NMDA receptors are activated during low frequency test stimulation (Andreasen et al. 1989) and provide calcium influx and release of calcium from internal stores during single synaptic events (Emptage et al. 1999). On the basis of this, we hypothesised that PP stimulation, which enhances synaptic release of neurotransmitter (Clark et al. 1994), might prolong the decay of t-LTP through a frequency-dependent calcium signal that is mediated by NMDA receptor activation.

Two groups of experiments, in which the test stimulation consisted of PP stimuli with an IPI of 80 ms, were used to investigate whether or not the activation of NMDA receptors prolonged the decay kinetics of t-LTP. The post-tetanic test stimulation