INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Long-term potentiation (LTP) is the best candidate for the cellular mechanism of memory storage (Bliss and Collingridge 1993; Martin et al. 2000). LTP processes can be divided into at least two steps: the induction that triggers strengthening of synaptic transmission and the maintenance that causes the strengthening to be persistent. It is the maintenance processes that are most relevant to the molecular basis of memory storage, but not much is known about them.¹ A critical test of whether a process is involved in LTP maintenance is to alter this process after LTP has been induced. If the mechanism responsible for maintaining the strength of the synapse is impaired, the potentiated response should return toward the prepotentiated level. A question of what happens to the basal transmission depends on still unresolved issues regarding the mechanisms of potentiation. If LTP is due to a purely quantitative change, then both potentiated and nonpotentiated transmission would be similarly affected. If, however, LTP is due to a purely qualitative change, then only potentiated responses would be altered. A combination of both mechanisms would result in alteration of both basal and potentiated transmission, with greater effect on potentiated transmission.

Using these kinds of tests, several agents have been found to disrupt the maintenance of potentiated transmission. However, the data are sparse and, in some cases, contradictory. There have been reports that potentiation can be reversed specifically (without affecting the basal transmission) by inhibitors of heme oxygenase (Stevens and Wang 1993) and brain-derived neurotrophic factor (BDNF) (Kang et al. 1997). In later tests, however, the same or similar inhibitors showed no effect on LTP maintenance (Chen et al. 1999; Meffert et al. 1994). In another study, an integrin inhibitor specifically depressed potentiated transmission (Staubli et al. 1998) but was effective only if applied immediately after LTP induction but not 25 min later. Recently actin polymerization inhibitors have been shown to depress potentiated transmission with no or smaller effect on the non-potentiated transmission (Kim and Lisman 1999; Krucker et al. 2000). The time dependence of these inhibitors on LTP maintenance is yet to be studied. Considerable efforts have been made to investigate the role of protein kinases in the LTP maintenance. Several protein kinases have been shown to be persistently activated after LTP induction, and it has been postulated that one or more of these kinases may be responsible for the maintenance of LTP (Lisman 1994; Roberson and Sweatt 1996; Soderling and Derkach 2000). Initial studies used bath applications of inhibitors that could affect both pre- and postsynaptic sites. Although some of these studies showed that LTP maintenance could be specifically reversed (Hrabetova and Sacktor 1996; Huang et al. 1992; Huber et al. 1995; Malinow et al. 1988; Matthies et al. 1991), others reported no effect (Denny et al. 1990; Lopez-Molina et al. 1993; Matthies and Reymann 1993; Muller et al. 1992) or nonspecific effects on both potentiated and nonpotentiated synaptic transmission (Kim 1999; Leahy and Vallano 1991; Muller et al. 1990; Perkel and Nicoll 1991). In different studies, kinase inhibitors were applied intracellularly to the postsynaptic cell after LTP induction, but the results were also contradictory: some investigators (Feng 1995; Wang and Feng 1992; Wang and Kelly 1996) observed specific reversal of LTP, but others (Chen et al. 2001; Huang et al. 1992; Malgaroli et al. 1992; Malinow et al. 1989; Otmakhov et al. 1997) did not see any effect on LTP maintenance even though the same inhibitors blocked LTP induction.

We previously suggested (Chen et al. 2001; Otmakhov et al. 1997) that our difficulty in reversing LTP maintenance by CaMKII inhibitors might be due to low phosphatase activity after LTP induction. The main phosphatase dephosphorylating postsynaptic density (PSD)-associated CaMKII is protein phosphatase 1, PP1 (Strack et al. 1997). Because this phosphatase can be inhibited after LTP induction due to PKA-dependent phosphorylation of inhibitor 1 (Blitzer et al. 1998), we wanted to boost the phosphatase activity by inhibiting PKA. We report here that postsynaptic application of PKA inhibitor, Rp-cAMPS, indeed partially reverses LTP. Our results, however, indicate that this effect does not appear to be primarily due to PKA inhibition but must involve other cyclic nucleotide-regulated processes.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Slice preparation

The methods of slice preparation, electrical stimulation, recording, and intracellular perfusion were similar to ones used previously (Otmakhov et al. 1997) and were conducted in accordance to the Guiding Principles for Research Involving Animals and Human Beings. Male Long-Evans rats (14-23 days old) were anesthetized by isoflurane and decapitated as approved by the Institutional Animal Care and Use committee. The brain was quickly removed and submerged in ice-cold artificial cerebrospinal fluid (ACSF) containing 2 mM Ca²⁺ (see following text for other constituents). Transverse hippocampal slices (350 µm thick) were prepared at 06°C using a vibratome (Vibratome 1000). The CA3 region was surgically removed from each slice immediately after preparation. Slices were kept for 2 h on cell culture inserts (Falcon, 8-µm pore diameter) covered by a thin layer of ACSF and surrounded by a humidified 95% O₂-5% CO₂ atmosphere at room temperature (20°C). For recording, slices were transferred to a submerged-type recording chamber with continuous flow (1.5 ml/min) of ACSF. The ACSF contained (in mM) 124 NaCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 2.5 KCl, 4 CaCl₂, 4 MgSO₄, 20 D-glucose, and 0.1 picrotoxin. ASCF was saturated with 95% O₂-5% CO₂ (pH 7.4). Experiments were carried out at 22-25°C.

Electrical recording and stimulation

Whole cell recording was performed using an Axopatch-1D amplifier (Axon Instruments, Foster City, CA) with low-pass filter set at 1 kHz. The patch pipettes had a resistance of 2.5-3.5 M when filled with pipette solution. The pipette solution contained (in mM) 120 Cs-methanesulfonate, 20 CsCl, 10 HEPES, 4 MgATP, 0.3 Na₃GTP, 0.2 EGTA, and 10 phosphocreatine, pH, 7.3; osmolarity, 300 mosM. Patching was performed under visual control using infrared oblique illumination and a CCD video camera. Recordings were made from cell bodies in the CA1 pyramidal layer, 20-80 µm beneath the slice surface. Whole cell currents were measured in voltage-clamp mode at a holding potential of 65 mV. Series and input resistances were monitored every 6 s by measuring the peak and steady-state currents in response to 2- to 4-mV, 38-ms hyperpolarizing steps. The accuracy of these measurements was confirmed using a model cell. For monitoring the stability of the slice responsiveness, field potentials were simultaneously recorded using a glass pipette filled with ACSF (300 k resistance). The pipette was positioned in stratum radiatum near the dendritic region of the recorded cell at ~100 µm from the cell body. An Axoclamp2A (Axon Instruments) and a custom (×1,000) amplifier (0.5 Hz to 1 kHz) were used for the amplification of field potentials. Both, intracellular and extracellular recorded signals were digitized at 5-10 kHz and then stored and analyzed using custom software written in Axobasic (Axon Instruments). Two stimulating electrodes (glass pipettes filled with ACSF, 300 k) were placed in st. radiatum to activate two independent pathways of Schaffer collaterals. The electrodes were positioned in the dendritic region ~70 and 150 µm from the cell body and ~50 µm lateral from the dendritic tree of the recorded cell. Each pathway was stimulated every 6 or 12 s with 2- to 50-µA, 150-µs square pulses delivered through current isolation units (Isolator-11, Axon Instruments). The two pathways were stimulated alternatively. LTP was induced using a pairing procedure in which a cell was depolarized to 0 mV, and 200 stimuli were delivered to one of the pathways at 1.4 Hz. Responses in this pathway (the "LTP pathway") showed large and reliable potentiation. The second pathway, which was not stimulated during the depolarization (the "non-LTP pathway"), normally showed no potentiation.

Intracellular perfusion

The intracellular drug application was performed using a custom-made perfusion system (Otmakhov et al. 1997).

Drug preparation

Several phosphodiesterase-resistant cAMP phosphorothioate stereoisomers were used: Rp-adenosine 3', 5',-cyclic monophosphorothioate sodium salt (Rp-cAMPS); Sp-8-hydroxyadenosine 3', 5',-cyclic monophosphorothioate sodium salt (Sp-8OH-cAMPS) and Rp-8-hydroxyadenosine 3', 5',-cyclic monophosphorothioate sodium salt (Rp-8OH-cAMPS; all from BioLog, La Jolla, CA). Stock solutions (50 mM) of all cAMP analogues were prepared in deionized water and stored in aliquots at 70°C. On the day of an experiment, a fresh aliquot was diluted in standard internal solution concentrated by 10%, and the osmolarity was adjusted to 300 mosM. The addition of the hydroxy group in Rp-8-OH-cAMPS and Sp-8-OH-cAMPS makes these isomers less membrane permeable than Rp-cAMPS. The effects of these compounds on PKA activity were tested in a biochemical assay by Joel C. Selcher in the laboratory of Dr. David Sweatt. The assay was performed using brain homogenate in the presence of 1 µM of cAMP (Roberson and Sweatt 1996). Rp-cAMPS (100 µM) produced 60% and Rp-8-OH-cAMPS (100 µM) produced 10% of the level of PKA inhibition produced by a potent and specific PKA peptide inhibitor, IP20 (100 µM). At this concentration of IP20, PKA activity was completely blocked. Sp-8-OH-cAMPS (10 µM) and cAMP (10 µM) activated PKA similarly (105 and 100%, respectively). The absence of inhibitory activity for Rp-8-OH-cAMPS was unexpected. Stock solutions of SQ 22536 (Biomol) and regulatory subunit type II of PKA (Promega) were prepared in distilled water at concentrations 125 mM and 250 U/µl, respectively, and stored in aliquots at 70°C until the day of an experiment. 8(p-sulfophenyl)theophylline (8p-SPT; RBI, Natick, MA) was dissolved in ACSF on the day of an experiment.

Statistical analysis

The peak amplitude of a synaptic response was calculated by subtracting the average value of data points in a 15-ms window before the stimulus from the average value of data points in a 3-ms window at the peak of the synaptic response. In figures describing individual experiments, each symbol represents the amplitude of a single response. In figures that summarize several experiments, the amplitude measurements from individual experiments were first averaged during 2-min periods, then normalized to the baseline period before pairing, and finally averaged across experiments. The data were presented as means ± SE. For calculations of the statistical significance of differences, averages for longer periods (from 4 to 20 min, as indicated in the text and figures) were used. The Student t-test for two independent populations was applied in all cases. The duration of excitatory postsynaptic current (EPSC) was measured at the half-maximum of the EPSC peak amplitude. All calculations and plots were performed using the spread-sheet program "Origin 5.0" (MicroCal Software)

Quantification of the effects of drugs on LTP maintenance

METHOD OF CALCULATION. When we calculated effects of drug on LTP and non-LTP responses we made an adjustment for the "natural" drift (usually decay) of LTP responses and non-LTP responses that was normally observed in control experiments. Test experiments (in which a drug was applied) and control experiments were normally alternated. The level of drug-induced depression of synaptic responses in both pathways was expressed as a percentage of the responses in each pathway just before the drug application but after LTP induction in the LTP pathway. The drug effect was considered as a specific for LTP if the fractional depression of LTP-responses was larger than the fractional depression of non-LTP responses.

View larger version (31K): [in this window] [in a new window]   Fig. 1. Calculation of drug effects on long-term potentiation (LTP) maintenance. Simulated experiments, in which control internal solution () or drug containing solution () were perfused after induction of LTP. , the period of the perfusion. A: data for the pathway in which LTP was induced (LTP responses). B: data for the nonpotentiated pathway (non-LTP responses). Rbd and Rad, responses before and after drug perfusion; Rbc and Rac, responses before and after control solution perfusion; Radn and Racn, normalized responses after drug or control solution perfusion; DepLTP and Depnon-LTP, the adjusted drug-induced depressions of potentiated (A) and nonpotentiated (B) synaptic responses. C: ratios of LTP to non-LTP excitatory postsynaptic current (EPSC) calculated from A and B for control () and drug () experiments. The baseline values were subtracted from these ratios, and the ratios were then normalized to the values after LTP induction. Cr and Dr, the degrees of LTP/non-LTP ratios for control and drug experiments; respectively; DepLTPsp, the adjusted drug-induced "specific" depression of LTP. DepLTP, Depnon-LTP, and DepLTPsp are the values used in RESULTS.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The pairing procedure for LTP induction (see METHODS) produced reliable and large LTP (>300%). For the purpose of standardization, only experiments with the amplitude of potentiated responses >250% but <600% of the baseline amplitude were used for the analysis. Typically, LTP was fully selective for the pathway synaptically stimulated during the pairing (the LTP pathway). Occasionally, very weak (10-15%) potentiation was also observed in nonpaired pathway (non-LTP pathway).

Rp-cAMPS specifically depresses potentiated synaptic transmission

To test the effect of Rp-cAMPS on LTP maintenance, the drug was applied intracellularly after LTP induction. Figure 2A shows a representative experiment. LTP was induced in one synaptic pathway, and the second pathway was left unpotentiated as a control. Ten minutes after induction, Rp-cAMPS (4 mM) was perfused into the patch pipette. This produced a dramatic decrease in the LTP-pathway responses ("LTP responses") over the next 20 min. Two additional effects were also observed: a smaller decrease in the magnitude of non-LTP pathway responses ("non-LTP responses") and an increase in holding current (Fig. 2). Perfusion of standard internal solution did not affect these parameters (Fig. 2B). Summary data (n = 7) are shown in Fig. 3. Perfusion of Rp-cAMPS resulted in a significant decrease in LTP responses within 20 min of the drug application. Because a small decay of LTP and non-LTP responses were observed in control experiments, we made an adjustments for these drifts during the calculation of the actual effect of drug (see METHODS ). At 40 min after the start of the drug application, the LTP level was depressed by 45% from the LTP level measured before the drug application [Dep_LTP = 45%, (P < 0.01) see METHODS]. Non-LTP responses were also depressed by the drug, but the effect was weaker [Dep_non-LTP = 30%, P < 0.05; Fig. 3B].

View larger version (32K): [in this window] [in a new window]   Fig. 2. Postsynaptic application of Rp-cAMPS reverses maintenance of LTP. Representative experiments are shown, in which Rp-cAMPS (4 mM; A) or control internal solution (B) were perfused into recorded cell 10 min after LTP was induced in 1 pathway (), while the 2nd pathway was left unpotentiated ( and , respectively, in A1 and B1). LTP was induced by a pairing procedure 12 min after the beginning of recording (). Potentiated synaptic responses were dramatically depressed by the drug (A1), while LTP remained stable in the control experiment (B1). , the perfusion period. Holding current was slightly increased by the drug (A4) but not by the control perfusion (B4). Series (Rs) and input (Ri) resistances were stable in both test and control experiments (A, 2 and 3, and B, 2 and 3). A5 and B5: representative synaptic responses (averages of 10) taken at the times indicated in A1 and B1. A6 and B6: the same traces but normalized to the peak. Trace 2 is for the potentiated EPSC before perfusion. Trace 3 is for the potentiated EPSC ~40 min after perfusion.

View larger version (25K): [in this window] [in a new window]   Fig. 3. Postsynaptic application of Rp-cAMPS (4 mM) depresses potentiated synaptic responses more dramatically than nonpotentiated responses. A-F: summary of the experiments shown in Fig. 2. A: superimposition of the graphs, showing the time course of synaptic responses in the pathways, in which LTP was induced (LTP-path). Postsynaptic perfusion of Rp-cAMPS () resulted in significantly larger decay of LTP than the "natural" decay observed in control experiments (). B: time course of nonpotentiated synaptic responses (non-LTP path) in the experiments shown in A. Series (C) and input resistances (D) were not significantly affected by the perfusion, but holding current (E) increased significantly during perfusion of Rp-cAMPS. F: LTP/non-LTP EPSC ratio (calculated from A and B) demonstrates that the drug () depresses LTP significantly stronger than the "natural" decay of LTP in the control experiments (). , the time where comparisons of the control and the test data were performed.

Because the drug depressed the non-LTP responses, this nonspecific effect must contribute to the depression of the LTP responses. We address the issue of specificity by calculating the proportional effects of the drug on LTP and non-LTP responses. Two methods have been used. One method is based on the calculation of the differences in drug-induced depression of the LTP and non-LTP responses (Dep_LTP  Dep_non-LTP, see METHODS). The second method is based on the calculation of the LTP/non-LTP EPSC ratio. If the percentages of the depression of LTP and non-LTP responses are different, then the LTP/non-LTP EPSC ratio should change after drug application. Therefore according to this method a drug has a specific component of the effect if the difference between the LTP/non-LTP EPSC ratios calculated from the control and test experiments is statistically significant. The degree of this difference is a measure of the specific component of the effect (Dep_LTPsp, see METHODS). The advantage of the second method is that it visualizes the dynamic change in the specific component of the drug effect. The LTP/non-LTP EPSC ratio, normalized as described in METHODS, is shown in Fig. 3F. The graph clearly demonstrates that the level of LTP was specifically depressed by the drug (Dep_LTPsp = 39%).

Rp-cAMPS application also significantly increased inward holding current (from 50 ± 2 to 73 ± 3 pA, P < 0.001, Fig. 3E), while series and input resistances were not significantly affected (P > 0.05 for both; Fig. 3, C and D). It should be noted that the duration of EPSCs slightly but significantly increased after induction of LTP in potentiated (18 ± 3%, P < 0.05) but not in control pathway (3 ± 1.2%, P > 0.05). There was also a small additional increase in the EPSC duration 40 min after the drug application. This increase (15 ± 5.4%), however, was not significantly different from the increase observed in the control experiments (7 ± 1.8%, P > 0.05)

Time dependence after LTP induction

Several lines of evidence suggest that LTP has a vulnerable period of ~15 min after induction when it can be more easily reversed by several manipulations (Arai et al. 1990; Staubli and Chun 1996; see Huang and Hsu 2001 for review). It has also been demonstrated that a tetanus-producing LTP results in a transient (for 2-10 min) increase in the production of cAMP and PKA activation (Blitzer et al. 1995; Roberson and Sweatt 1996). We, therefore tested if the inhibition induced by Rp-cAMPS depends on the time of the drug application after LTP induction. Figure 4 shows a summary of experiments in which Rp-cAMPS (or control solution) was intracellularly applied 2, 10, and 25 min after LTP induction. The levels of LTP response depression (Dep_LTP) were 53, 49, and 52%, respectively (Fig. 4A). They were all measured 35 min after drug application. Non-LTP responses were also significantly depressed by drug applications (Dep_non-LTP = 33, 34, and 45%, respectively). The plots of LTP/non-LTP EPSC ratios (Fig. 4C) calculated from the experiments shown in Fig. 4A indicate that the inhibitory effect of Rp-cAMPS has a specific component and is independent of the time of drug application after LTP induction; the decrease of the LTP/Control EPSC ratio (Dep_LTPsp) was 39, 41, and 38%, respectively for 2-, 10-, and 25-min delays in the drug application. These depressive effects of the drug on LTP maintenance were statistically significant for all delay periods (P < 0.05, for each time delay). We conclude that postsynaptically applied Rp-cAMPS specifically reverses previously induced LTP and that this reversal is not dependent on the time (from 2 to 25 min) of drug application after LTP induction.

View larger version (32K): [in this window] [in a new window]   Fig. 4. The inhibitory affect of Rp-cAMPS on LTP maintenance does not depend on time of drug application after LTP induction. A: summary of the experiments in which control solution or Rp-cAMPS (4 mM) were perfused into cells 2, 10, or 25 min after pairing. Open symbols, the LTP pathways; closed symbols, the non-LTP pathways. B: time course of series resistance during experiments shown in A. C: superimposition of the LTP/non-LTP ratios calculated from A. Open and gray symbols, experiments in which Rp-cAMPS was applied inside of cells. Closed symbols, control experiments. Thin lines indicate the periods used for calculation of the LTP levels. Gray bars show the time of the drug or control solution intracellular perfusion. Arrows, the times when pairing were applied.

Concentration dependence of the effects of Rp-cAMPS

Rp-cAMPS is a membrane permeable compound and due to its leakage out of the recorded neuron only a fraction of its initial concentration is likely to reach the dendritic region. We therefore started with a high concentration of Rp-cAMPS (4 mM) but also tested lower concentrations (1.0 and 0.4 mM). These also induced a significant depression of LTP responses (Dep_LTP = 45 and 33%, respectively, P < 0.01, Fig. 5A). The non-LTP responses were also significantly depressed (Dep_non-LTP = 48 and 37%, respectively, P < 0.01). The depression developed more slowly and with a longer delay than that induced by 4 mM of the drug. This depression, however, did not have a specific component for LTP. LTP/non-LTP EPSP ratios show that either concentration (1.0 or 0.4 mM) of Rp-cAMPS induced a similar depression of LTP (Dep_LTPsp = 20% for both concentrations) that was not significantly different from the natural decay of potentiated responses observed in control experiments (13%, P > 0.05; Fig. 5C). The change in I_h in these experiments was not significantly different from control experiments in which control internal solution was perfused (45% ± 4 in control and 47 ± 5 in 0.4 mM of Rp-cAMPS experiments, P > 0.5)

View larger version (33K): [in this window] [in a new window]   Fig. 5. Concentration dependence of the Rp-cAMPS effects. A: summary of the experiments in which control solution or 4, 1, or 0.4 mM of Rp-cAMPS were intracellularly perfused after LTP induction. Open symbols, LTP pathways; closed symbols, non-LTP pathways. B: time course of series resistance during experiments shown in A. C: superimposition of the LTP/non-LTP ratios calculated from A. Open and gray symbols, experiments in which Rp-cAMPS was applied; closed symbols, control experiments. Thin lines indicate the periods used for calculation of LTP levels. Gray bars show the time of the drug or control solution intracellular perfusion. Arrows, the times when pairing were applied.

Activity dependence of the effects of Rp-cAMPS

We next tested the possibility that the inhibitory effect of Rp-cAMPS is dependent on the activity of synaptic activation. That might happen if processes triggered by synaptic activation itself (for example metabotropic glutamate receptor activation, Ca²⁺ entry or others) are required for the weakening of the synaptic transmission produced by Rp-cAMPS. If this is the case, the synaptic response could decay faster with more frequent synaptic stimulation for the same period of time. Figure 6 shows a summary of six experiments in which one synaptic pathway was stimulated 10 times more frequently than the other (1 in 6 s vs. 1 in 60 s). Perfusion of Rp-cAMPS started 6 min after simultaneous induction of LTP in both pathways and resulted in a similar depression of LTP responses in both pathways. Twenty minutes after the drug application the remaining levels of LTP responses were 48 ± 17 and 46 ± 14%, respectively, for stimulus frequency 1 in 6 and 1 in 60 s (P > 0.05). Therefore the inhibitory affect of Rp-cAMPS on synaptic transmission does not show activity dependence in the frequency range tested.

View larger version (23K): [in this window] [in a new window]   Fig. 6. The inhibitory effects of Rp-cAMPS on LTP maintenance are not activity dependent. Summary of experiments in which 2 synaptic inputs were stimulated with different frequencies (1/6 and 1/60 s). Pairing was applied at time 0 () simultaneously for both inputs. Intracellular perfusion of Rp-cAMPS (4 mM) induced a similar depression of LTP in both inputs. , the time of the drug perfusion.

Inhibitory effect of Rp-cAMPS on LTP is not amplitude dependent

Another form of activity dependence might arise from the difference in amplitude of LTP and non-LTP responses. The greater amplitude of LTP responses might produce some additional dendritic depolarization (due to the inadequate voltage-clamp) that might trigger a process leading to the depression of synaptic responses in the presence of the drug. In this case, more pronounced inhibition of potentiated responses by Rp-cAMPS might simply be due to the greater amplitude of the potentiated response in comparison to the control responses. To test this hypothesis, we lowered the stimulation strength in one pathway so that it produced a response much smaller than in the other pathway. After induction of LTP in this weakly stimulated pathway, the amplitude of LTP responses became comparable with the amplitude in non-LTP pathway responses. Even in this case, intracellular perfusion of Rp-cAMPS (4 mM) induced a faster and greater depression of LTP responses (Dep_LTP = 53%) than of non-LTP responses (Dep_non-LTP = 40%, P < 0.05; Fig. 7). The LTP/non-LTP EPSP ratio in these experiments decreased 50% in comparison to the more stable level (only 19% decay, P < 0.001) in the control experiments in which standard internal solution was perfused (Fig. 7C). Therefore the specific component of the drug-induced depression of LTP (Dep_LTPsp) was 38%, which was very similar to the LTP depression in experiments with large-amplitude responses (Dep_LTPsp = 39%, P > 0.05, Fig. 3F). We conclude that the specific effect of Rp-cAMPS on LTP responses is not amplitude dependent.

View larger version (43K): [in this window] [in a new window]   Fig. 7. The inhibitory effect of Rp-cAMPS on LTP does not depend on the amplitude of response. The stimulus strength applied to LTP path was initially adjusted to induce ~3-fold less synaptic response than the one in non-LTP path. A and B, top: (A1 and B1) representative experiments and summary of several similar experiments (A2 and B2), in which Rp-cAMPS (4 mM) or standard internal solution (control) were internally applied after induction of LTP in weakly stimulated synaptic input.  and , LTP pathways;  and , non-LTP pathways. A and B, bottom: the time course of series resistance change during the above experiments. A3 and B3: representative traces (averages of 10) taken at the times indicated in A1 and B1 and normalized to the peak. Insets: non-normalized the same traces. C: LTP/non-LTP EPSC ratios calculated from A2 and B2 show significantly larger depression of LTP induced by Rp-cAMPS () in comparison to experiments in which standard internal solution was perfused. , the time of the internal perfusion. , pairing procedure.

When smaller initial synaptic responses were used, the shape of synaptic responses did not change after induction of LTP (EPSC duration increased only 4%, P > 0.05, n = 15, Fig. 7, A3 and B3) as was observed with large potentiated responses (see Fig. 1). This suggests that a large synaptic current results in some error in the voltage-clamp that prolongs the apparent response. Perfusion of Rp-cAMPS produced a prolongation in the duration of synaptic responses. This increase in duration was statistically significant; in drug experiments EPSC duration increased 35 ± 3%, whereas in control experiments, EPSC duration increased only 13 ± 4% (P < 0.01; Fig. 7, A3 and B3).

Adenosine receptors are not involved in the effect of Rp-cAMPS

It has been demonstrated that overproduction of cAMP (Gereau and Conn 1994) or infusion of a large amount of adenosine into a postsynaptic cell (Brundege and Dunwiddie 1996) results in the appearance of adenosine (or its homologues) outside the cell; this can then inhibit transmitter release by acting on presynaptic adenosine receptors. To check if Rp-cAMPS or products of its degradation affect adenosine receptors, we tested the effect of intracellularly applied Rp-cAMPS on synaptic transmission during bath application of the adenosine receptor antagonist, 8p-SPT (100 µM). First, in separate experiments, we confirmed that this concentration of 8p-SPT completely abolished the strong (95%) inhibitory effect of adenosine (50 µM) on field excitatory postsynaptic potentials (EPSPs, n = 2, data not shown) (see also Brundege and Dunwiddie 1996). Another confirmation of the effectiveness of 8p-SPT in our preparation is that this inhibitor also abolished the dramatic (100%) depression of field EPSPs usually observed after pipette solution is pushed into the slice before establishing the seal (data not shown). This inhibition only occurs if the patch pipette solution contains ATP, which is known to activate adenosine receptors involved in presynaptic inhibition.

In this condition of strong inhibition of the adenosine receptors, intracellular application of Rp-cAMPS still induced a large decrease of both LTP responses and non-LTP responses (Fig. 8, A and B). In cells perfused with Rp-cAMPS, both LTP responses and non-LTP responses decreased by 44% (Dep_LTP) and 36% (Dep_non-LTP), respectively (P < 0.01 for both), and the adjusted LTP/non-LTP EPSC ratio (Dep_LTPsp) decreased by 24% (P < 0.001) in the presence of 8p-SPT. These experiments demonstrate that neither the specific nor the nonspecific components of the depressive effect of Rp-cAMPS is due to the activation of adenosine receptors outside the recorded cell.

View larger version (28K): [in this window] [in a new window]   Fig. 8. Bath application of adenosine receptor inhibitor 8(p-sulfophenyl)theophylline (8p-SPT, 100 µM) does not prevent the inhibitory effect of Rp-cAMPS on LTP maintenance. A: superimposition of the graphs, showing time course of synaptic responses in the pathways, in which LTP was induced (LTP path). Postsynaptic perfusion of Rp-cAMPS () resulted in significantly larger depression of LTP responses than decay of LTP in experiments in which standard internal solution was perfused (). B: time course of nonpotentiated synaptic responses (non-LTP path) in the same experiments. Series resistance (D) was not significantly affected by the perfusion, but the holding current (E) increased significantly during perfusion of Rp-cAMPS. C: superimposition of the graphs of LTP/non-LTP EPSC ratios (calculated from A and B) demonstrates that the drug () depresses LTP significantly more strongly than the "natural" decay of LTP in the control experiments (). , the time where comparisons of the control and the test data were performed.

Is the effect of Rp-cAMPS PKA dependent?

One action of Rp-cAMPS is to compete with cAMP for the binding to PKA resulting in the inhibition of the kinase. To determine whether the effects of Rp-cAMPS that we observe are due to the action on PKA, we tested two newly developed cAMP analogues with different functional effects on PKA: Sp-8OH-cAMPS and Rp-8OH-cAMPS. Both compounds have additional OH groups, which should make them less membrane permeable. The first is a derivative of Sp-cAMPS, which is an activator of PKA. The second is a derivative of Rp-cAMPS, a PKA inhibitor. Biochemical assays confirmed that Sp-8OH-cAMPS is indeed a strong PKA activator. Surprisingly, Rp-8OH-cAMPS proved to have very little inhibitory effect toward PKA (see METHODS). It was even more surprising that both compounds produced similar effects on cell physiology and that these effects were similar to the effects of Rp-cAMPS. Figure 9 shows representative experiments in which Rp-8OH-cAMPS, Sp-8OH-cAMPS (both 4 mM), or control internal solution were perfused into postsynaptic cells 10 min after the induction of LTP in one pathway. Both Rp-8OH-cAMPS and Sp-8OH-cAMPS induced a large decrease in synaptic responses. Similar to Rp-cAMPS, both compounds also induced an increase in holding currents. The effect on the duration of EPSC was seen in some cells (Fig. 9A5) but on average was not statistically significant (P > 0.05 for both drugs). Figures 10 and 11 show summary data for the experiments shown in Fig. 9. After the perfusion of Rp-8OH-cAMPS, the responses decreased similarly in the LTP (Dep_LTP = 38%) and non-LTP (Dep_non-LTP = 35%) pathways. This depression was statistically significant for both pathways (P < 0.001 in both cases). The change in the LTP/non-LTP EPSC ratio after the drug application (Dep_LTPsp = 4%) was not significant (P > 0.05; Fig. 9F). When Sp-8OH-cAMPS was perfused, LTP responses and non-LTP responses were depressed by 40% (Dep_LTP) and 28% (Dep_non-LTP), respectively. These effects were statistically significant for both LTP and non-LTP pathways (P < 0.001 for both cases). LTP/non-LTP ratios demonstrated a weak (Dep_LTPsp = 16%) but statistically significant (P < 0.01) specific effect of Sp-8OH cAMPS on LTP (Fig. 10F). Therefore these experiments demonstrate that a PKA inhibitor (Rp-cAMPS), PKA activator (Sp-8OH-cAMPS), and weak PKA inhibitor (Rp-8OH-cAMPS) produce similar effects. This suggests the depressive effects of the cyclic nucleotide analogues on synaptic responses and LTP may not solely depend on their effect on PKA.

View larger version (39K): [in this window] [in a new window]   Fig. 9. Postsynaptic application of either a weak protein kinase A (PKA) inhibitor (Rp-8OH-cAMPS) or a PKA activator (Sp-8OH-cAMPS) depresses both potentiated and basal synaptic responses. Representative experiments are shown, in which Rp-8OH-cAMPS (4 mM; A), Sp-8OH-cAMPS (4 mM; B), or control internal solution (C) were perfused into the recorded cell 10 min after LTP was induced in 1 pathway ( and ), while the 2nd pathway was left unpotentiated ( and  in A1-C1). Synaptic responses were dramatically depressed by the drugs (A1 and B1), while they remained stable in the control experiment (C1). , time of LTP induction. , perfusion periods. Holding current (Ih) was increased by the drugs (A4 and B4) but not by the control perfusion (C4). Series (Rs) and input (Ri) resistances were stable in both test and control experiments (A, 2 and 3, B, 2 and 3, and C, 2 and 3). A5-C5: representative synaptic responses (averages of 10, normalized to the peak) taken at the times indicated in A1-C1.  the potentiated EPSC before perfusion. Insets: the same traces before the normalization

View larger version (24K): [in this window] [in a new window]   Fig. 10. Postsynaptic application of Rp-8OH-cAMPS (4 mM) depresses potentiated and nonpotentiated synaptic responses equally. A-F: summary of experiments like that shown in Fig. 9, A and C. A: superimposition of the graphs, showing the time course of synaptic responses in the pathways in which LTP was induced (LTP path). Postsynaptic perfusion of Rp-8OH-cAMPS () resulted in significantly larger depression of LTP than natural decline of LTP in the control experiments, in which standard internal solution was perfused (). B: time course of nonpotentiated synaptic responses (non-LTP path) in the experiments shown in 9, A and C. Series (C) and input resistance (D) was not significantly affected by the perfusion, but holding current (E) increased significantly during perfusion of Rp-8OH-cAMPS. F: LTP/non-LTP EPSC ratio (calculated from A and B) for both drug () and control experiments () demonstrates that the drug depresses LTP and non-LTP responses equally. -, time where comparisons of the control and the test data were performed.

View larger version (24K): [in this window] [in a new window]   Fig. 11. Postsynaptic application of Sp-8OH-cAMPS (4 mM) depresses potentiated synaptic responses more strongly than nonpotentiated. A-F: summary of experiments like that shown in Fig. 9, A and C. A: superimposition of the graphs showing the time course of synaptic responses in the pathways in which LTP was induced (LTP path). Postsynaptic perfusion of Sp-8OH-cAMPS () resulted in significantly larger depression of LTP than natural decay of LTP in experiments, in which standard internal solution was perfused (). B: time course of nonpotentiated synaptic responses (non-LTP path) in the experiments shown in Fig. 9, B and C. Series (C) and input resistance (D) were not significantly affected by the perfusion, but the holding current (E) increased significantly during perfusion of Sp-8OH-cAMPS. F: LTP/non-LTP EPSC ratio (calculated from A and B) for both drug () and control experiments (). The drug depresses potentiated responses only slightly more than nonpotentiated. , time where comparisons of the control and the test data were performed.

To further test this conclusion, we used two more compounds involved in regulation of PKA: the regulatory subunit type II of PKA, which is a potent PKA inhibitor (Francis and Corbin 1994), and SQ 22536, which is a specific inhibitor of adenylyl cyclase (the enzyme that produces cAMP and thereby activates PKA). As a control for the active regulatory subunit, we used the subunit after it was inactivated by boiling. Perfusion of the active PKA regulatory subunit (25 U/µl) resulted in a very slow developing and weak but statistically significant depression of both LTP and non-LTP responses (Dep_LTP = 11%, Dep_non-LTP = 20%; P < 0.05 for both; Fig. 12). LTP/non-LTP control ratios calculated for test and control experiments showed no significant difference (Dep_LTPsp = 1%, P > 0.05), indicating no specific effect toward LTP pathway. Holding currents were not affected by this PKA inhibitor (P > 0.05; Fig. 12D).

View larger version (27K): [in this window] [in a new window]   Fig. 12. Postsynaptic application of a non-competitive PKA inhibitor (regulatory type II subunit of PKA, PKARS) depresses potentiated and nonpotentiated synaptic responses weakly and equally. A-F: average results from experiments in which active () and inactive () PKARS of PKA were intracellularly perfused 2 min after LTP induction. A: superimposition of the graphs, showing the time course of synaptic responses in the pathways, in which LTP was induced (LTP path). Postsynaptic perfusion of active PKARS (25 U/µl) resulted in significantly larger depression of LTP than in experiments in which inactive PKARS (25 U/µl) was perfused. For clarity, the data were normalized to the initial level of LTP. B, time course of nonpotentiated synaptic responses (non-LTP path) in the experiments shown in A. Series resistance (C) and holding current (D) were not significantly affected by the perfusion. F: LTP/non-LTP EPSC ratios (calculated from A and B) for both drug () and control experiments () demonstrates that the drug depresses LTP and non-LTP responses equally. , time where comparisons of the control and the test data were performed.

Contrary to the regulatory PKA subunit, perfusion of SQ 22536 (1.25-5 mM, Fig. 13) induced no significant effects on neither LTP nor non-LTP synaptic responses (Dep_LTP = 6%, Dep_non-LTP = 3%; P > 0.05 in both cases). Conceivably, the LTP/non-LTP ratio method also showed no significant specific effect of the drug (Dep_LTPsp = 0.5%, P > 0.05). The holding current, however, was significantly decreased in these experiments (from 77 ± 36 pA in the control to 60 ± 4 pA in the test experiments, P < 0.02); this was opposite to the increase of the current observed after perfusion of cAMP analogues (see preceding text). The change in the holding current indicates the effectiveness of the inhibitor inside the cell.

View larger version (30K): [in this window] [in a new window]   Fig. 13. Postsynaptic application of an inhibitor of adenylyl cyclase (SQ 22536, 1.25-5 mM) depresses potentiated and nonpotentiated synaptic responses equally. A-F: summary of the experiments in which SQ 22536 () or control internal solution () were intracellularly perfused 10 min after induction of LTP. A: superimposition of the graphs, showing time course of synaptic responses in the pathways, in which LTP was induced (LTP path). Postsynaptic perfusion of both SQ 22536 or control internal solution resulted in no effect on LTP maintenance. B: time course of nonpotentiated synaptic responses (non-LTP path) in the same experiments shown in A. Series (C) and input (D) resistances were not significantly affected by the perfusion, but holding current (E) was significantly decreased. F: LTP/non-LTP EPSC ratios (calculated from A and B) for both drug () and control () experiments demonstrate that the drug depresses LTP and non-LTP responses equally. , time where comparisons of the control and the test data were performed.

The weak and nonspecific effect of the regulatory PKA subunit in comparison to strong and specific effects of cyclic nucleotide analogues, which are both an activator and an inhibitor of PKA, suggests that the effects of Rp-cAMPS might be not primarily due to inhibition of PKA.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Rp-cAMPS reverses maintenance of LTP

Our results demonstrate that processes that maintain LTP can be impaired by postsynaptic drug application, supporting the notion that postsynaptic mechanisms are involved in the maintenance of LTP. Specifically, we demonstrate that Rp-cAMPS is able to almost completely reverse responses in previously potentiated pathway to the basal level. The non-LTP responses are also affected but to a lesser extent. Control experiments indicate that this specificity toward potentiated responses cannot be attributed simply to their larger amplitude. This ability to produce a specific effect on LTP places Rp-cAMPS on a short list of substances that have this ability (see INTRODUCTION). The fact that the LTP pathway is affected more strongly than the non-LTP pathway implies that synapses that have been potentiated must differ from unpotentiated synapses in a qualitative way and argues against models of LTP in which only quantitative changes occur. However, because Rp-cAMPS does affect non-LTP pathway, our results suggest that both qualitative and quantitative changes occur and are responsible for maintaining of LTP. We find that only the highest concentration of Rp-cAMPS used (4 mM) produced specific effect on LTP pathway. Lower concentrations (1 and 0.4 mM) depressed both LTP and non-LTP pathways equally. This different sensitivity of the basal and potentiated transmission to the drug also suggests that the mechanisms underlying them are different.

It has been suggested that there might be a narrow time window shortly after LTP induction during which potentiation can be more easily reversed. For example, some low-frequency stimulation protocols can reverse LTP only if applied within a time window of ~1-15 min after the induction (Staubli and Chun 1996). Similarly, some other factors that reverse LTP can only do so when applied within this time window (Arai et al. 1990; Staubli et al. 1998; reviewed in Huang and Hsu 2001). In contrast, we find that Rp-cAMPS is equally able to reverse LTP whether it is applied 2 or 25 min after induction, suggesting that the LTP maintenance processes affected by this compound are not time sensitive during this period. We further find that the effect of Rp-cAMPS is not dependent on the frequency of synaptic activity. This indicates that Rp-cAMPS directly affects the processes that maintain the strength of synapses rather than activity-dependent processes that weaken potentiated synapses.

Is activation of presynaptic adenosine receptors involved?

The effect of Rp-cAMPS on basal or potentiated synaptic transmission has not been previously reported (Blitzer et al. 1995; Bolshakov et al. 1997; Frey et al. 1993; but see Lessmann and Heumann 1997). Presumably, we see the effect because of the more efficient method of drug delivery through the patch pipette. It has been reported, however, that bath application of Sp-cAMPS (Frey et al. 1993) or cAMP itself (Gereau and Conn 1994) depresses synaptic transmission during the application, but the effect was attributed to the activation of extracellular adenosine receptors (Gereau and Conn 1994). There have also been reports that an excessive production of cAMP by a cell or intracellular infusion of high concentration of exogenous adenosine can result in a "leak" of these compounds (or products of their degradation) through the membrane and produce decrease in neurotransmitter release due to the activation of presynaptic adenosine receptors (Brundege and Dunwiddie 1996; Gereau and Conn 1994). Our results, however, show that the effect of Rp-cAMPS does not require functional adenosine receptors.

Involvement of PKA

Several experiments indicate that PKA is involved in the maintenance of the basal AMPA-mediated synaptic transmission. Ser-845 site on the GluR1 subunit of the AMPA channel is phosphorylated under basal conditions (Kameyama et al. 1998; Lee et al. 1998) and the phosphorylation of this site directly upregulates channel function (Banke et al. 2000; Roche et al. 1996). Induction of long-term depression (LTD) results in the dephosphorylation of Ser-845, but the following dedepression of synaptic transmission rephosphoryles the site (Kameyama et al. 1998; Lee et al. 1998; see also Ehlers 2000). Basal phosphorylation of Ser-845 site by PKA may explain why postsynaptic application of the peptide inhibitors of PKA depresses AMPA-mediated transmission and occludes LTD induction (Kameyama et al. 1998; Rosenmund et al. 1994). In agreement with these data, we found that regulatory PKA subunit II, which is a strong PKA inhibitor, depresses basal synaptic transmission.

Very little is known about the involvement of PKA in the maintenance of LTP. An early study showed no effect of PKA inhibitor, KT5720, on previously induced LTP even though induction of late phase of LTP was blocked (Huang and Kandel 1994). Effect on the basal transmission was not tested in this study. Our experiments demonstrate that PKA inhibitor (PKA regulatory subunits II) does slightly depress previously induced LTP, but the effect was not specific toward potentiated transmission because basal transmission was also depressed. It has also been reported that that PKA phosphorylation site of GluR1 (Ser-845) is not further phosphorylated after induction of LTP (Lee et al. 2000). All this suggests that PKA activity might be mostly involved in the processes maintaining basal but not potentiated synaptic transmission.

In contrast to the absence of the specific effect of inhibitors of catalytic PKA activity on the maintenance of LTP (Huang and Kandel 1994; our data with regulatory subunit II of PKA), we found that the cAMP analogue, Rp-cAMPS, depressed potentiated transmission significantly stronger than basal transmission. That indicates that Rp-cAMPS is likely to affect some other target in addition to PKA. Our experiments with other cAMP analogues support this suggestion. We found that both a PKA activator (Sp-8OH-cAMPS) and a weak PKA inhibitor (Rp-8OH-cAMPS) also dramatically decreased potentiated and basal synaptic transmission. This suggests that Rp-cAMPS may produces depression of both potentiated and nonpotentiated synaptic transmission through a mechanisms independent of PKA.

Although we found an effect of PKA inhibitors on synaptic responses, an inhibitor of cAMP production (SQ 22536) did not affect synaptic responses, although it did decrease holding current indicating its effectiveness inside the cell. One possibility is that it is constitutive PKA activity independent of cAMP levels that is responsible for the maintaining a component of basal transmission. Such activity might be produced due to proteolysis of the PKA regulatory subunit (Chain et al. 1995).

Involvement of PKA-independent, cyclic nucleotide-dependent processes

Although the Rp- and Sp-cAMPS analogues are thought of as agents that affect PKA, they can also affect other targets. These analogues are relatively minor chemical modifications of cAMP (Botelho et al. 1988) and are thus likely to interact with any other cAMP binding proteins. Increasing evidence indicate that there are many cyclic nucleotide-dependent processes that are not dependent on PKA. Hippocampal neurons contain at least two types of cyclic nucleotide-gated channels (CNGC) directly controlled by cAMP: the olfactory type of CNGC and the hyperpolarization-activated channel or I_h (Bradley et al. 1997; Pape 1996; Zufall et al. 1997). It has been shown that both Rp- and Sp-AMP analogues can bind and activate both of these channels (Ingram and Williams 1996; Kramer and Tibbs 1996). All CNGC are nonselective cation channels and their activation produces inward current. Therefore it is tempting to suggest that inward current produced during the application of both Rp- and Sp- cAMPS analogues might be due to the activation of such channels. Further evidence consistent with this idea is that the adenylyl cyclase inhibitor, SQ 22536, which should decrease cAMP concentration, produces an outward current (Figs. 3E, 10E, 11E, and 13E). However, the change in inward current in our experiments was not accompanied by a change in input resistance that usually occurs when channels are activated. In addition, although we always observed strong inward current when specific LTP depression was induced, similar magnitude current sometimes occurred when only nonspecific synaptic depression occurred. This suggests that inward current activated by cAMP analogues might be involved in, but is not sufficient for, the production of the specific component of LTP depression. The mechanism of this possible involvement also is not clear. One possibility is that activation of CNGC might result in an increase in the postsynaptic Ca²⁺ level (Bradley et al. 1997) that activates depotentiation processes similar to that produced by low-frequency stimulation and is thought to involve phosphatase activation (Huang and Hsu 2001; Lisman 1994; Luscher et al. 2000). Recently, however, several new actions of cyclic nucleotides have been reported, suggesting the existence of wide variety of cyclic nucleotide-dependent processes. For example, a study on the neuromuscular junction demonstrated that Rp-AMPS could modulate presynaptic function through a PKA-independent activation of I_h channels by some unknown mechanism, possibly by activation of actin-dependent transport (Beaumont and Zucker 2000). Another example of a cyclic nucleotide-dependent, kinase-independent processes has been recently found in hippocampal neurons. Here, cGMP analogues induced strong depression of AMPA currents also through unknown mechanisms (Lei et al. 2000). Another example of PKA-independent affects of cAMP analogues is the regulation of MAP kinase activity through the cAMP binding proteins that affect Rab activation (Kawasaki et al. 1998; Pham et al. 2000). An additional unexplained effect of cyclic nucleotides is LTD produced by transient bath application of Sp-cAMPS in the presence of an inhibitor of GABA-A mediated inhibition (Yu et al. 2001). This is opposite to the PKA-dependent potentiation produced by the drug when GABA-A inhibition is intact (Frey et al. 1993; Yu et al. 2001). This new effect is in line with our finding that postsynaptically applied Sp-cAMPS (also in the presence of GABA-A inhibitors) depresses basal synaptic transmission. Interestingly, involvement of protein phosphatases through unknown mechanism has been implicated in this effect of Sp-cAMPS.

In summary, we found that a postsynaptic application of a cyclic nucleotide analogue, Rp-cAMPS, can dramatically reverse preestablished LTP. This reversal is not activity-dependent and is not dependent on the time (25 min) of the drug application after LTP induction. The effect seems to involve a new mechanism and is not primarily dependent on activation of PKA. Understanding this mechanism would appear to be a promising line of research for unraveling the still mysterious processes which maintain LTP.