Long term depression in the CA1 field is associated with a transient decrease in pre- and postsynaptic PKC substrate phosphorylation.

Induction of homosynaptic long term depression (LTD) in the CA1 field of the hippocampus is thought to require activation of N-methyl-d-aspartate receptors, an elevation of postsynaptic Ca(2+) levels, and a subsequent increase in phosphatase activity. To investigate the spatial and temporal changes in protein phosphatase activity following LTD induction, we determined the in situ phosphorylation state of a pre- (GAP-43/B-50) and postsynaptic (RC3) protein kinase C substrate during N-methyl-d-aspartate receptor-dependent LTD in the CA1 field of rat hippocampal slices. We show that LTD is associated with a transient (<30 min) and D-AP5-sensitive reduction in GAP-43/B-50 and RC3 phosphorylation and that LTD is prevented by the phosphatase inhibitors okadaic acid and cyclosporin A. Our data provide strong evidence for a transient increase in pre- and postsynaptic phosphatase activity during LTD. Since the in situ phosphorylation of the calmodulin-binding proteins GAP-43/B-50 and RC3 changes during both LTD and long term potentiation, these proteins may form part of the link between the Ca(2+) signal and Ca(2+)/calmodulin-dependent processes implicated in long term potentiation and LTD.

Induction of homosynaptic long term depression (LTD) in the CA1 field of the hippocampus is thought to require activation of N-methyl-D-aspartate receptors, an elevation of postsynaptic Ca 2؉ levels, and a subsequent increase in phosphatase activity. To investigate the spatial and temporal changes in protein phosphatase activity following LTD induction, we determined the in situ phosphorylation state of a pre-(GAP-43/B-50) and postsynaptic (RC3) protein kinase C substrate during

N-methyl-D-aspartate receptor-dependent LTD in the CA1 field of rat hippocampal slices. We show that LTD is associated with a transient (<30 min) and D-AP5-sensitive reduction in GAP-43/B-50 and RC3 phosphorylation and that LTD is prevented by the phosphatase inhibitors okadaic acid and cyclosporin A. Our data provide strong evidence for a transient increase in pre-and postsynaptic phosphatase activity during LTD. Since the in situ phosphorylation of the calmodulin-binding proteins GAP-43/B-50 and RC3 changes during both LTD and long term potentiation, these proteins may form part of the link between the Ca 2؉ signal and Ca 2؉ /calmodulin-dependent processes implicated in long term potentiation and LTD.
Several forms of activity-dependent plasticity can be induced in the Schaffer collateral CA1 synapses of the hippocampus, including long term potentiation (LTP) 1 and long term depression (LTD) (1)(2)(3). In CA1, the major forms of LTP and LTD depend on activation of NMDA receptors and a subsequent elevation of the intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) (4). It has been suggested that specific properties of the Ca 2ϩ signal (magnitude and temporal structure) determine the direction of the synaptic change (5)(6)(7). During LTP, a large rise in [Ca 2ϩ ] i presumably activates protein kinases such as calcium/calmodulin-dependent protein kinase II and/or PKC, whereas during LTD a small rise in [Ca 2ϩ ] i favors an activation of protein phosphatases (8).
The involvement of kinases and, notably, PKC in LTP has been shown in several ways (3, 9 -11). Direct evidence for an involvement of phosphatases (PPases) in LTD is, however, less abundant. Pharmacological studies have shown that inhibitors of PPase 1, PPase 2A, and PPase 2B, applied intra-or extracellularly block the induction of LTD (12,13). In addition to this effect on LTD induction, there have been contradictory results on the effect of PPase inhibitors on basic synaptic transmission. Figurov et al. (14) report that calyculin A, in a concentration of 1 M, induced a potentiation of the field excitatory postsynaptic potentials (fEPSP) of 40%. In addition, it has been shown that calyculin A increased synaptic transmission in an activity-dependent manner (15). In contrast, no effect on synaptic transmission in the hippocampus and visual cortex have been reported by others (12,16). Because of the contradictory pharmacological data and the fact that it cannot be ruled out that the blocking effect of the PPase inhibitors on LTD induction was due to modulation of NMDA receptor function, we used another approach to investigate the role of PPases in LTD. We monitored substrate phosphorylation in hippocampal slices labeled with inorganic phosphate after the induction of LTD, a method we have exploited successfully in LTP (17,18).
The phosphoproteins of which the in situ phosphorylation state was monitored were GAP-43/B-50 (also known as F1, neuromodulin) and RC3 (also known as neurogranin, BICKS). GAP-43/B-50 is a nervous tissue-specific protein that is highly expressed in neurons during development and nerve regeneration and has been implicated in neurite outgrowth, long term potentiation, neuronal signal transduction, and neurotransmitter release. In mature, non-damaged nervous tissue GAP-43/B-50 is expressed in most neurons, but at lower levels; expression levels are highest in human associative brain areas and rat hippocampal and olfactory areas. In mature neurons, GAP-43/B-50 is predominantly found in presynaptic plasma membranes and not in dendrites. RC3 is a neuron-specific PKC substrate, mostly expressed in forebrain neurons during postnatal development and maturity and in hippocampus predominantly localized in dendritic spines (19 -23). Both GAP-43/ B-50 and RC3 are soluble in 2.5% perchloric acid and bind CaM at low Ca 2ϩ concentrations (19, 24 -27), which is prevented by PKC-mediated phosphorylation (24,25). Binding to CaM stabilizes the amphiphilic ␣-helix within RC3 and GAP-43/B-50 only when Ca 2ϩ is absent (28). Purified, PKC-phosphorylated RC3 and GAP-43/B-50 can be dephosphorylated by purified PPase 1, PPase 2A, and PPase 2B (29,30).
To test the involvement of pre-and postsynaptic phosphatases in LTD, we have examined changes in GAP-43/B-50 and RC3 phosphorylation at different times after the induction of LTD in the Schaffer collateral CA1 pathway. We show that both GAP-43/B-50 and RC3 phosphorylation are transiently decreased shortly after the induction of LTD, a reduction that is not observed when LTD induction is blocked by D-AP5. In parallel experiments we showed that inhibitors of PPases potently inhibit the induction of LTD. These data provide strong evidence for activation of both pre-and postsynaptic phosphatases immediately after the induction of LTD.
For electrophysiological recordings three slices were transferred to a recording chamber and superfused (1 ml/min) for at least 60 min with gassed phosphate-free ACSF at 30°C. fEPSPs were recorded in the radial layer of CA1 field in response to orthodromic stimulation as described before (18). In all three slices the maximal fEPSP was determined first. Only slices in which the maximal fEPSP was bigger than 1 mV were selected. Two slices were randomly chosen to be used for electrophysiological recordings. Stimulation intensity was adjusted to evoke half-maximal responses, and this intensity was kept constant throughout the experiment. Thirty min after placement of stimulation and recording electrodes, the superfusion medium was changed to ACSF containing 100 Ci/ml [ 32 P i ]orthophosphate, and after 90 min, base-line recordings of fEPSPs were started (0.05 Hz, 30 min). Subsequently one slice received a low frequency train of stimuli (1 Hz, 15 min) at test intensity, and recordings were continued for another 10, 30, 60, or 120 min. The other stimulated slice did not receive the 1-Hz stimulation, and the third slice served as an unstimulated control. In some experiments slices were treated with 50 M D-AP5 (Cambridge Research Biochemicals, Norwich, UK), 1 M okadaic acid (Anawa, Wangen, Switzerland), or 1 M cyclosporin A (gift from Dr. H. Boddeke, Sandoz AG, Basel, Switzerland) 10 min before the 1-Hz stimulation was applied or with 1 mM chelerythrine (Sigma) 45 min after the 1-Hz stimulation. At the end of the experiments, all three slices were removed from the recording chamber, washed, and homogenized (see below). Evoked fEPSPs were recorded using a conventional AC-coupled amplifier, and data were digitized using pCLAMP (Axon Instruments) software. Data were analyzed using the same software, and the initial slope of the fEPSPs was calculated. For graphical presentation, the data are expressed as percentages of the average of the base-line slope of the fEPSPs.
After the electrophysiological recordings had ended, the slices were rapidly washed twice in 2 ml of ice-old H 2 O containing 100 mM NaF, 1.2 mM NaH 2 PO 4 , 10 mM EDTA, 5 mM EGTA and homogenized in 100 l of H 2 O containing 100 mM NaF, 10 mM EDTA, and 5 mM EGTA. Of the total homogenate, 60 l was added to 40 l of stop-mix solution containing (final concentration) 62.5 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol (v/v), 5% ␤-mercaptoethanol (v/v), 0.001% bromphenol blue, and after boiling for 10 min, samples were stored at Ϫ20°C. The remainder of the homogenate was used to determine protein content (31) with BSA as a standard. The degree of GAP-43/B-50 and RC3 phosphorylation was determined using quantitative immunoprecipitation as described earlier (18,32). Briefly, for GAP-43/B-50 immunoprecipitation, 10 g of the protein homogenate was incubated overnight at 4°C with antibody 9527 (1:200 final dilution), and for RC3 immunoprecipitation, 0.33 g of protein homogenate was incubated overnight at 4°C with antibody 8420 (1:100 final dilution). Antigen complexes were precipitated with Pansorbin (Calbiochem) and solubilized, and immunoprecipitates were separated using 11% (GAP-43/B-50) or 15% (RC3) SDS-polyacrylamide gel electrophoresis. 32 P incorporation into proteins was detected using a Fuji BAS1000 imaging system (Raytest, Germany) and quantified using TINA analysis software. The total 32 P incorporation into proteins was determined by trichloroacetic acid precipitation as described previously, and 32 P incorporation into GAP-43/B-50 and RC3 was normalized accordingly (18). Statistical analysis were carried out using a Student's t test for paired or unpaired samples.

RESULTS
In the present experiments we monitored the phosphorylation state of GAP-43/B-50 and RC3 in 32 P-labeled hippocampal slices from young rats (less than 2 weeks old) at different times after the induction of LTD. Fig. 1 shows that prolonged (15 min) low frequency (1 Hz) stimulation of the Schaffer collaterals induced a robust and long-lasting decrease in the slope of the fEPSP, showing that LTD can be induced in 32 P-labeled slices. Ten minutes after the 1-Hz stimulation was stopped, the slope of the fEPSP was reduced to 57.6 Ϯ 4.8% (n ϭ 6), and this FIG. 1. LTD induction in the CA1 field of 32 P-labeled hippocampal slices. Three slices were placed in an electrophysiological recording chamber, and in two of these slices fEPSP were recorded in the dendritical layer of the CA1 in response to stimulation of the Schaffer collateral/commisural fibers. After the slices were labeled with 32 P, LTD was induced in one slice (filled circles) by repetitive low frequency stimulation (1 Hz, 15 min; indicated by the black bar) of the afferent fibers, whereas the other slice did not receive the conditioning stimuli (open circles). Experiments were stopped 10 (A), 30 (B), 60 (C), and 120 (D) min after LTD had been induced. At all time points, there was a significant reduction in the slope of the fEPSP in the slices receiving the conditioning stimulation, indicating that LTD had been induced in these slices. The slope of the fEPSP of the normal stimulated slices did not change. Typical traces of fEPSP in a slice in which LTD was induced are shown at the top corresponding to the time points indicated in D. Calibration, 2 mV, 5 ms. reduction in the slope of the fEPSP remained at 30 min (58.0 Ϯ 7.8%, n ϭ 6), 60 min (50.1 Ϯ 4.6%, n ϭ 6), and 120 min (38.1 Ϯ 2.3%). The normally stimulated slices, placed in the same recording chamber, did not show any run-down in synaptic responses (95.3 Ϯ 3.7%, 107.2 Ϯ 4.1%, 90.0 Ϯ 5.4%, 98.8 Ϯ 6.5% at 10, 30, 60 and 120 min after the LTD slice received the 1-Hz stimulation). The competitive NMDA receptor antagonist D-AP5 completely blocked the induction of LTD (data not shown). Therefore, we can conclude that the stimulation paradigm used in the present study induces a reproducible, robust, and NMDA receptor-dependent LTD in the CA1 field of 32 Plabeled hippocampal slices.
To examine the effect of LTD on the in situ GAP-43/B-50 and RC3 phosphorylation, GAP-43/B-50 and RC3 phosphorylation was determined 10 min after LTD induction. At this time point there is a decrease in the in situ phosphorylation of both GAP-43/B-50 and RC3 (Fig. 2). GAP-43/B-50 phosphorylation is reduced to 61.8 Ϯ 8.4% (Fig. 2A, n ϭ 6), and RC3 phosphorylation is reduced to 69.9 Ϯ 6.8% (Fig. 2B, n ϭ 6). When  Fig. 2, A and C. Ten minutes after LTD induction there is a significant correlation between the decrease in the in situ GAP-43/B-50 and RC3 phosphorylation and the decrease in the slope of the fEPSP (correlation coefficient ϭ 0.97, p Ͻ 0.001, n ϭ 6 for GAP-43/B-50 and correlation coefficient ϭ 0.89, p Ͻ 0.001, n ϭ 6 for RC3; Fig. 2, B and D). We also performed experiments on slices from older animals (4 to 6 weeks) in which the likelihood of LTD induction is reduced (33). Prolonged (15 min) low frequency (1 Hz) stimulation failed to induce significant LTD in these slices; the slope of the fEPSP returned to 93.8 Ϯ 7.4% (n ϭ 4) of the base-line value within 10 min after the end of the low frequency stimulation. In these slices we determined the phosphorylation of GAP-43/B-50 and RC3. There was no reduction of GAP-43/B-50 (101.2 Ϯ 4.3%, n ϭ 4) and RC3 (98.5 Ϯ 6.3%, n ϭ 4) phosphorylation in these slices. The decrease in GAP-43/B-50 and RC3 phosphorylation seems to be specific, since the in situ phosphorylation of another well defined PKC substrate, the myristoylated alanine-rich protein kinase C substrate, MARCKS (34), does not change after the induction to LTD (35).
To study the temporal relationship between the decrease in phosphorylation of pre-and postsynaptic PKC substrates, experiments were stopped at different times after low frequency stimulation (Fig. 3). The in situ phosphorylation state of postsynaptic protein RC3 is significantly reduced 10 min after LTD induction (see also Fig. 2) and is back to control levels at 30 min after LTD induction (98.3 Ϯ 10.0%) and remains unaltered at 60 and 120 min (100.2 Ϯ 11.6% and 92.4 Ϯ 8.3%, n ϭ 6 and 5, respectively). At the presynaptic site, GAP-43/B-50 phosphorylation, which is decreased at 10 min after LTD in- duction (see also Fig. 2), is still reduced at 30 min (74.0 Ϯ 4.9%, n ϭ 6). One hour after LTP induction GAP-43/B-50 phosphorylation is increased (132.7 Ϯ 8.8%, n ϭ 6) and returns to control levels 120 min after LTD induction (102.3 Ϯ 6.7%, n ϭ 5).
To investigate the involvement of phosphatases in the transient decrease in RC3 and GAP-43/B-50 phosphorylation, we determined the effect of two different phosphatase inhibitors known to increase basal RC3 and GAP-43/B-50 phosphorylation in hippocampal slices on LTD induction (Fig. 4). The addition of 1 M okadaic acid (Fig. 4A) or 1 M cyclosporin A (Fig. 4B) starting 10 min before and ending 5 min after the 15-min 1-Hz stimulation paradigm completely blocked the induction of LTD. The average slope of the fEPSP 60 min after the 1-Hz stimulation in the presence of okadaic acid or cyclosporin A was 100.5 Ϯ 5.4% and 96.0 Ϯ 4.8%, respectively (n ϭ 5), whereas the 1-Hz stimulation in phosphate-free ACSF resulted in a long-lasting decrease in the slope of the fEPSP (see Fig. 1).
The increase in GAP-43/B-50 1 h after LTD induction (see Fig. 3) could be caused by either decreased PPase activity or increased kinase activity. Since it is at present not possible to pharmacologically activate phosphatases, we decided to assess if increased PKC activity was causing the increase in GAP-43/ B-50 phosphorylation. We treated slices with chelerythrine, a specific PKC inhibitor known to block a phorbol ester-induced increase in GAP-43/B-50 phosphorylation (18). The addition of 1 M chelerythrine to the phosphate-free ACSF starting 30 min after LTD induction for 30 min did not affect the expression of LTD as compared with untreated control slices (Fig. 5). The slope of the fEPSP was still depressed to 43.7 Ϯ 8.1% (n ϭ 5) of the base line in chelerythrine-treated slices as compared with the depression to 38.1 Ϯ 2.3% in control slices (see Fig. 1). DISCUSSION In this study we show that LTD induction in the CA1 field of the hippocampus is accompanied by a transient decrease in GAP-43/B-50 and RC3 phosphorylation. This reduction in phosphorylation is firmly linked to LTD, because (i) it is NMDA receptor-dependent, (ii) it is significantly correlated with the degree of LTD, (iii) it is not elicited by test stimulation, (iv) it is not observed in slices from older animals in which low frequency stimulation fails to induce LTD, and (v) it is specific for certain PKC substrates since no effect was found on myristoylated alanine-rich protein kinase C substrate (MARCKS) phosphorylation (35). The transient reduction in the phosphorylation of GAP-43/B-50 and RC3 we find after LTD induction most likely reflects an increased phosphatase activity. Our pharmacological studies confirm the involvement of phosphatases (PPase 1, 2A, and 2B) in LTD, as already indicated by others, since okadaic acid and cyclosporin A block LTD. However, these pharmacological studies do not address the temporal and spatial aspects of PPase involvement in LTD. Our findings clearly show an NMDA receptor-dependent decrease in the in situ phosphorylation state of two identified pre-and postsynaptic phosphoproteins, providing independent evidence for an increased pre-and postsynaptic PPase activity following NMDA receptor activation. However, we cannot rule out the possibility that the decrease in GAP-43/B-50 and RC3 phosphorylation is caused by reduced kinase activity. Application of the kinase inhibitor H7 or the specific PKC inhibitor chelerythrine to naive slices mimics and occludes the maintenance of homosynaptic LTD in the CA1 field of the hippocampus. In addition, it was shown that LTD is accompanied with a reduction in the levels of protein kinase M (the constitutively active catalytic fragment of PKC) and the PKC isoforms PKC␥ and PKC⑀ in homogenates (mixed pre-and postsynaptic compartments) 30 min after LTD induction and a concomitant reduction in constitutive PKC activity (36). At present it is not known whether GAP-43/B-50 or RC3 are substrates for these forms of PKC.
Repetitive low frequency stimulation (15 min, 1 Hz) of the afferent fibers in the CA1 field of the hippocampus of young rats induced a robust and long-lasting decrease in synaptic transmission, which is comparable to that found by others (37). The stimulation and recording electrodes were separated as far as possible to guarantee the biggest possible area in which LTD was induced. The reduction in in situ GAP-43/B-50 and RC3 phosphorylation we observed probably is an underestimation of that in individual affected synapses, since our experimental approach averages phosphorylation in the whole slice. Although in the hippocampus GAP-43/B-50 and RC3 are found in high concentrations in the CA1 area, they are also present in other parts of the hippocampus (21,38,39). The use of subslices to increase the number of synapses expressing LTD was not successful because of the increase in signal to noise ratio (data not shown). Recently, a protocol to chemically induce LTD in the whole slice has been described (40 -42). This form of LTD is accompanied by a reduction in GluR1 phosphorylation (41).
Our data show that LTD in the CA1 field of the hippocampus is associated with pre-and postsynaptic reduction in the phosphorylation state of two identified PKC substrates. Since LTD induction is thought to be postsynaptic, the reduction in GAP-43/B-50 phosphorylation indicates an involvement of a retrograde messenger. The identity of the retrograde signal in LTD is still unknown (43)(44)(45)(46)(47).
The reduction in RC3 phosphorylation is transient and maximal 10 min after LTD induction. This indicates that there is a transient activation of protein phosphatase activity following the elevation in [Ca 2ϩ ] i due to NMDA receptor activation and presumably secondary release of Ca 2ϩ from intracellular Ca 2ϩ stores (48). Due to technical limitations, time points shorter than 10 min could not be measured; this is longer than the Ca 2ϩ signal (49). The mechanism of this transient PPase acti- vation and the nature of the PPase involved remain unknown. The calcium calmodulin-dependent phosphatase calcineurin has been implicated in the mechanisms underlying LTD (13,50) and is known to dephosphorylate RC3 and GAP-43/B-50 (29,30). The recent finding that mice lacking the A␣ isoform of calcineurin express normal LTD indicates that the A␤ isoform is required for LTD (51).
GAP-43/B-50 phosphorylation was increased 1 h after LTD induction. This finding can reflect either a decreased PPase activity or increased PKC activity. Because no potent and specific PPase activators are available at the moment, we tested the possibility of increased PKC activity using chelerythrine (52). Chelerythrine, applied 30 min after LTD induction, did not affect the maintenance of LTD, consistent with previous reports (36). Thus is seems unlikely that increased presynaptic PKC activity around 1 h after the induction of LTD is required for its maintenance. Therefore, we suggest that the increase in GAP-43/B-50 phosphorylation is due to an overshoot in inhibition of PPase activity after its activation following LTD induction.
GAP-43/B-50 and its PKC-mediated phosphorylation have been implicated in the regulation of neurite outgrowth and  neurotransmitter release (53,54), two processes that are important in activity-dependent forms of synaptic plasticity like LTP and LTD. It is tempting to speculate that reduced phosphorylation of GAP-43/B-50 would reduce synaptic efficacy during LTD by synaptic remodeling and/or reducing release. Indeed, some experimental evidence using quantal analysis indicates that LTD is accompanied by a decrease in neurotransmitter release (55). However, more recently no changes in the release probability were found (56). Interestingly, it has been shown that one of the phosphatases implicated in LTD, calcineurin, is very important in the process of neurotransmitter release (29,57). RC3 is thought be involved in regulating release of Ca 2ϩ from internal stores (58), a process important for LTP and LTD (48).
GAP-43/B-50 and RC3 contain an IQ domain, an amino acid sequence that is found in a growing subset of CaM-binding proteins and that contains a unique PKC phosphorylation site (59). Since both proteins bind CaM at resting [Ca 2ϩ ] i , they may function as a local CaM sink. Upon a rise in [Ca 2ϩ ] i GAP-43/ B-50 and RC3 will release CaM. The CaM binding capacity of both proteins is regulated by PKC phosphorylation; phosphorylation prohibits CaM binding. Therefore, increased phosphorylation of GAP-43/B-50 and RC3, as occurring during LTP, would result in higher free CaM levels, activating calcium/ calmodulin-dependent protein kinase II (phosphorylation of ␣-amino-3-hydroxy-5-methyl-4-isoxazole (AMPA)-type glutamate receptors (60), nitric-oxide synthase (47), and adenylate cyclase (higher cAMP levels and protein kinase A activation) (61,62). Reduced phosphorylation of GAP-43/B-50 and RC3, as observed during LTD, would result in lower free CaM levels, thereby favoring activation of the phosphatase calcineurin (dephosphorylation of e.g. AMPA-type glutamate receptors (41)) or activating phosphodiesterases (resulting in lower cAMP levels and reduced protein kinase A activity). Thus, the phosphorylation state of the two proteins could co-determine whether LTD-favorable CaM-dependent or LTP-favorable CaM-dependent processes would be activated.
In conclusion, we show for the first time changes in the in situ phosphorylation state of identified pre-and postsynaptic PKC substrates during LTD. This supports the idea that both pre-and postsynaptic changes occur during the expression of LTD. Our findings indicate that the decrease in GAP-43/B-50 as well as RC3 phosphorylation are part of the physiological changes occurring during LTD. The finding that two opposite forms of synaptic plasticity (LTP and LTD) are both accompanied by changes in the phosphorylation state of two neuronal phosphoproteins (GAP-43/B-50 and RC3) support the hypothesis that LTP and LTD might converge at the level of specific phosphoproteins (50).