Gating of CaMKII by cAMP-Regulated Protein Phosphatase Activity During LTP

See allHide authors and affiliations

Science  19 Jun 1998:
Vol. 280, Issue 5371, pp. 1940-1943
DOI: 10.1126/science.280.5371.1940


Long-term potentiation (LTP) at the Schaffer collateral–CA1 synapse involves interacting signaling components, including calcium (Ca2+)/calmodulin–dependent protein kinase II (CaMKII) and cyclic adenosine monophosphate (cAMP) pathways. Postsynaptic injection of thiophosphorylated inhibitor-1 protein, a specific inhibitor of protein phosphatase–1 (PP1), substituted for cAMP pathway activation in LTP. Stimulation that induced LTP triggered cAMP-dependent phosphorylation of endogenous inhibitor-1 and a decrease in PP1 activity. This stimulation also increased phosphorylation of CaMKII at Thr286 and Ca2+-independent CaMKII activity in a cAMP-dependent manner. The blockade of LTP by a CaMKII inhibitor was not overcome by thiophosphorylated inhibitor-1. Thus, the cAMP pathway uses PP1 to gate CaMKII signaling in LTP.

Multiple signaling pathways participate in LTP in the CA1 region of the hippocampus at both presynaptic and postsynaptic sites (1), with the CaMKII pathway playing a central role in transmitting the postsynaptic signals required for LTP (2, 3). In contrast, the role of other signaling pathways is not yet clear. The cAMP pathway is involved in LTP and memory in transgenic mice (4). In rats, the postsynaptic cAMP pathway is required for LTP induced by widely spaced trains of high-frequency synaptic stimulation (HFS), but activation of the pathway is not sufficient to induce LTP (5). Thus, the postsynaptic cAMP pathway does not transmit the signals for LTP but rather gates the transmittal pathway. It has been proposed (5, 6) that the cAMP-operated gate may use PP1.

The CA3-CA1 synapse of rat hippocampal slices was stimulated with widely spaced trains of HFS (7). The resulting LTP was blocked by inhibiting postsynaptic cAMP-dependent protein kinase [protein kinase A (PKA)] (Fig. 1A) (5). This requirement for PKA can be overcome by direct inhibition of postsynaptic phosphatases (5), suggesting that the cAMP pathway modulates LTP by blocking phosphatases. Protein phosphatase inhibitor-1 (I-1) is a candidate for mediating cAMP inhibition of phosphatase activity. I-1, upon phosphorylation by PKA at Thr35, is a specific blocker of PP1 (8). I-1 mRNA is expressed in CA1 neurons (9), and I-1 has already been implicated in plasticity at the CA3-CA1 synapse (10). We injected recombinant purified Thr35-thiophosphorylated I-1 (11) into the postsynaptic neuron and tested its ability to overcome the blockade of LTP by the specific PKA inhibitor Rp–cyclic adenosine monophosphorothioate (cAMPS). Thiophosphorylated I-1 completely reversed the effect of Rp-cAMPS, yielding LTP that was indistinguishable from that of the control (Fig. 1, B and C). In contrast, mutant [Thr35 → Ala (T35A)] nonphosphorylatable I-1 (11), when injected postsynaptically, did not reverse the effect of Rp-cAMPS. Thus, PP1 appears to be the postsynaptic phosphatase that negatively regulates LTP, and I-1 activation by PKA may facilitate LTP by inhibiting PP1.

Figure 1

Ability of Thr35-thiophosphorylated I-1 to substitute for cAMP pathway activation in LTP. (A) Intracellular summary data showing the blockade of LTP by postsynaptic Rp-cAMPS injection. Data were taken from the period of 20 to 30 min after the final train of HFS (consisting of measurements at three time points), when LTP had stabilized. Intracellular electrodes contained either no drug [control (Con), solid column, n = 4] or 10 mM Rp-cAMPS (hatched column, n = 7). The groups differed significantly (t test, P < 0.01), indicating that the postsynaptic cAMP pathway is required for LTP. (B) Time course of intracellular LTP, showing reversal of Rp-cAMPS blockade by thiophosphorylated I-1. After HFS, stable synaptic potentiation was obtained in cells recorded with control electrodes containing only KCl (triangles, n = 4). LTP was blocked when the electrode contained a combination of the PKA antagonist Rp-cAMPS (10 mM) and an inactive, nonphosphorylatable (T35A) form of I-1 (10 μM) (open circles, n = 5). However, when inactive I-1 was replaced by constitutively active Thr35-thiophosphorylated I-1 (10 μM), the blockade of LTP by Rp-cAMPS was overcome (filled circles, n = 8). The Rp-cAMPS + T35A I-1 group differed from the other two groups over the final three time points (Newman-Keuls test, P < 0.05). (C) Representative intracellular EPSPs (top row) and field EPSPs recorded simultaneously from the same slices (bottom row). Two superimposed traces are shown in each panel, one recorded during the baseline period and the other, indicated by the arrow, recorded 30 min after HFS. The intracellular recording electrode contained either KCl only (left traces), the combination of Rp-cAMPS and inactive T35A I-1 (middle traces), or Rp-cAMPS combined with Thr35-thiophosphorylated I-1 (right traces). LTP was observed in the field recording despite the blockade of intracellular LTP by the Rp-cAMPS + T35A mutant I-1. All slices used in this experiment showed normal field LTPs, with the EPSP slope measuring at least 153% of baseline at 30 min after HFS. There were no group differences in the field LTP. Calibrations, 5 mV intracellular, 250 μV extracellular, and 10 ms.

Next, we determined if LTP-inducing stimulation results in PKA phosphorylation of I-1. We examined the phosphorylation state of I-1 in the CA1 region after HFS (12). The same pattern of synaptic stimulation that induced cAMP-dependent LTP also raised the amount of phosphorylated I-1 in the CA1 region (Fig. 2A). The increase in phosphorylation of I-1 by HFS was dependent on PKA activity because it was blocked by the inclusion of Rp-cAMPS in the superfusate during stimulation (Fig. 2A). We then measured protein phosphatase activity in the CA1 region of stimulated and unstimulated slices (Fig. 2B). Under our assay conditions (13), greater than 90% of the phosphoprotein phosphatase activity in the CA1 homogenate was inhibited by thiophosphorylated I-1, defining it as PP1. HFS that activated I-1 also resulted in significant inhibition of PP1 activity. This effect was prevented by bath application of Rp-cAMPS during stimulation, indicating that PKA mediated the phosphatase inhibition.

Figure 2

cAMP-dependent phosphorylation of endogenous I-1 and inhibition of PP1 after HFS. (A) Tissue homogenates of area CA1 were probed with antibodies (29) recognizing either the phosphorylated form of I-1 selectively (top gel) or both phosphorylated and nonphosphorylated I-1 (bottom gel). HFS increased the amount of phosphorylated I-1, but this effect was prevented by inclusion of 100 μM Rp-cAMPS during stimulation. Total I-1 was not affected by either treatment. Similar results were obtained in two other experiments. (B) The superimposed columns on the left show that phosphatase activity measured in control CA1 tissue (open region) was predominantly due to PP1, because it was inhibited over 90% by 100 nM Thr35-thiophosphorylated inhibitor-1 (hatched column and dashed line). HFS significantly reduced phosphatase activity (center columns) (P < 0.05, Student's t test; indicated by an asterisk), an effect prevented by the inclusion of 100 μM Rp-cAMPS during stimulation (right columns). Results are representative of three independent assays.

Which PP1 targets are relevant for LTP? An important and well-established participant in LTP at the Schaffer collateral–CA1 synapse is CaMKII (14). Expression of constitutively active CaMKII increases synaptic efficiency and occludes LTP (2). Stimulation that induces LTP also increases Thr286phosphorylation and the consequent Ca2+-independent activity of CaMKII in area CA1 by 20 to 50% in a time-dependent fashion (15, 16). Autophosphorylation of CaMKII is reversed by PP1 in synaptic membranes (17), and the Thr286phosphorylation state of CaMKII plays an obligatory role in LTP and spatial memory (18). Thus, the cAMP pathway, through regulation of PP1, could modulate the phosphorylation state of Thr286 and thereby enhance Ca2+-independent CaMKII activity. If such a mechanism contributes to LTP, widely spaced HFS should increase phosphorylation of CaMKII as well as Ca2+-independent CaMKII activity in a cAMP-dependent manner. We determined the phosphorylation state of CaMKII by immunoblotting with an antibody that specifically recognizes Thr286-phosphorylated CaMKII (19). HFS that induced cAMP-dependent LTP also increased Thr286-phosphorylated CaMKII in the CA1 region (Fig. 3A). This effect of HFS was blocked by Rp-cAMPS, supporting the hypothesis that CaMKII phosphorylation at Thr286 is sustained in the presence of cAMP. Ca2+-independent CaMKII activity in the CA1 region was reliably increased by HFS (Fig. 3B) (20), an effect that was blocked by the inclusion of Rp-cAMPS during stimulation. Thus, the increase in autonomous CaMKII activity paralleled its autophosphorylation at Thr286and was mediated by a cAMP-dependent mechanism.

Figure 3

HFS-induced, cAMP-dependent increase in CaMKII phosphorylation and Ca2+-independent CaMKII activity. (A) CA1 homogenates were probed with an antibody specific for Thr286-phosphorylated CaMKII (top gel) or an antibody recognizing total CaMKII (bottom gel). HFS increased the concentration of Thr286-phosphorylated CaMKII by 38% as determined by densitometry, with little or no effect on total CaMKII (center lanes). In tissue stimulated in the presence of 100 μM Rp-cAMPS, the increase in Thr286-phosphorylated CaMKII was blocked (right lane). Results are representative of two independent assays. (B) HFS increased Ca2+-independent CaMKII activity. A 22% increase in CaMKII activity was observed, which was blocked in slices stimulated in the presence of 100 μM Rp-cAMPS. The asterisk indicates significant difference from other groups (Newman-Keuls test, P < 0.05). No significant group differences in total CaMKII activity were observed (control, 37.4 ± 11.3 pmol μg−1min−1; HFS, 43.7 ± 2.9; HFS + Rp-cAMPS, 55.5 ± 15.2; P > 0.05). Results are based on data from two independent experiments.

These results establish a locus of interaction between the cAMP and CaMKII pathways and indicate that the cAMP-operated gate in LTP, at least in part, acts at the level of CaMKII. In this model, if CaMKII activity were directly inhibited, then HFS would not induce LTP, regardless of phosphatase regulation. Thus, when LTP is blocked by a CaMKII inhibitor, the injection of activated I-1 should not restore LTP. Intracellular recordings were obtained with electrodes containing either a specific CaMKII inhibitor peptide (21) or a control peptide (Fig. 4). LTP was almost completely blocked by the inhibitor peptide, whereas the control peptide had no effect. The inclusion of thiophosphorylated I-1 in the recording electrode did not restore LTP to the control level, although a modest potentiation was obtained. Thus, the inhibition of PP1 has little effect on LTP in the absence of CaMKII activity, suggesting that CaMKII is a critical site for the cAMP-operated gate. The small recovery of LTP in the presence of thiophosphorylated I-1 may reflect incomplete inhibition of CaMKII by the peptide or an effect of PP1 on some other component of the LTP signal-transmitting pathway, such as the αamino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor (16). Although thiophosphorylated I-1 did not reverse the effect of the CaMKII inhibitor, it completely restored LTP in the presence of a PKA inhibitor (Fig. 1). These results are expected if PP1 is positioned between PKA and CaMKII.

Figure 4

Inability of thiophosphorylated I-1 to rescue LTP from CaMKII inhibition. (A) Time course graph of intracellular EPSP slope. The time of HFS is indicated by the gap in the abscissa. In control cells recorded with electrodes containing either KCl alone (filled triangles, n = 4) or 2.5 mM of inactive control peptide (open circles, n = 5), stable LTP was induced by HFS. However, when the electrode contained 2.5 mM of the CaMKII inhibitor autocamtide-3 (filled circles, n = 8), HFS did not induce LTP. The addition of Thr35-thiophosphorylated I-1 (10 μM) to electrodes containing autocamtide-3 (open triangles, n = 7) did not restore LTP to control levels, in contrast to its effectiveness in reversing the effect of a PKA inhibitor (see Fig. 1). Statistical analyses performed on the final three time points indicated that the modest reversal of the autocamtide-3 effect by thiophosphorylated I-1 was statistically reliable, whereas both autocamtide-3–treated groups differed significantly from the two control groups (Newman-Keuls tests, all P < 0.05). (B) Representative intracellular (top) and field (bottom) traces from slices in which the intracellular electrode contained, from left to right, KCl alone, inactive peptide, autocamtide-3, or autocamtide-3 combined with Thr35-thiophosphorylated I-1. Two superimposed traces are shown in each panel, one recorded during the baseline period and the other, indicated by the arrow, recorded 30 min after HFS. Intracellular and field traces were recorded simultaneously. Normal potentiation of the field EPSP was obtained in each case. For all slices used in this experiment, the field EPSP was greater than 145% of baseline at 30 min, and there were no group differences in the field LTP over the final three time points. Calibrations, 5 mV intracellular, 250 μV extracellular, and 10 ms.

Complex physiological functions are often regulated by cooperative interactions between multiple signaling pathways. We have proposed that the cAMP pathway uses protein phosphatases to gate signal flow through a transmittal pathway (5, 6). Here, we explicitly established that synaptic stimulation results in cAMP-dependent activation of I-1 and the concomitant inhibition of PP1, thus protecting the phosphorylation of CaMKII on Thr286 and maintaining increased CaMKII activity. Our data, in agreement with those of others (2, 3), indicate that CaMKII may be the transmittal pathway for LTP. Like other protein kinases in diverse physiological systems (22), CaMKII converts extracellular signals into physiological events upon sustained activation. Our data suggest that phosphatase regulation may be an important mechanism contributing to such persistent kinase activation.

Three phases of LTP have been proposed (23, 24). The coordinated actions of phosphatases, calcineurin and others, have been implicated in the transition between these phases (24,25). Of particular interest is calcineurin, which may negatively regulate the transition between early and late phases of LTP in a cAMP-dependent manner. Because calcineurin is a likely I-1 phosphatase in hippocampal neurons (26), cross-talk between calcium and cAMP at this point may be pivotal in synaptic plasticity. In any case, the cAMP pathway may bridge the early and late phases of LTP and provide an explanation for the involvement of cAMP signaling in memory in mice (4) and men (27).

  • * To whom correspondence should be addressed. E-mail: rb2{at}


View Abstract

Navigate This Article