(Received for publication, March 21, 1997, and in revised form, May 5, 1997)
From the Division of Neuroscience, Baylor College of Medicine, Houston, Texas 77030
The mitogen-activated protein kinase (MAPK) cascade has been intensely studied as a primary biochemical pathway through which a variety of extracellular stimuli initiate and regulate processes of cellular transformation. That MAPKs are abundantly expressed in postmitotic neurons, however, suggests different yet currently unknown functions for this cascade in the mature nervous system. Here we report that the MAPK cascade is required for hippocampal long term potentiation (LTP), a robust and widely studied form of synaptic plasticity. We observed that PD 098059, a selective inhibitor of the MAPK cascade, blocked MAPK activation in response to direct stimulation of the NMDA receptor as well as to LTP-inducing stimuli. Furthermore, inhibition of the MAPK cascade markedly attenuated the induction of LTP. PD 098059, however, had no effect on the expression of established LTP, and the MAPK cascade was not persistently activated during LTP expression. Our observations provide the first demonstration of a role for the MAPK cascade in the activity-dependent modification of synaptic connections between neurons in the adult mammalian nervous system.
The mitogen-activated protein kinase (MAPK)1 cascade has been classically studied as a critical biochemical pathway involved in cellular transformation events such as cell proliferation and determination. Such work has delineated a pathway by which growth factor receptor activation initiates a complex cascade leading to the activation of Ras, Raf, and MEK, a dual-specific kinase that activates MAPKs via phosphorylations on both threonine and tyrosine residues (reviewed in Refs. 1 and 2).
Although this cascade is typically studied in the context of mitotic cell regulation, its components are actually most abundantly expressed in postmitotic neurons of the developed nervous system (3, 4). At present, however, little is known about the physiologic roles of this cascade in mature neurons. We have begun to investigate the possible involvement of the MAPK cascade in the activity-dependent modulation of synaptic connections between neurons, a putative mechanism for the neuronal basis of learning and memory. In particular, we have examined the role of the MAPK cascade in hippocampal long term potentiation (LTP), a widely studied form of synaptic plasticity (reviewed in Refs. 5 and 6).
Recently, we reported that p42 MAPK (extracellular signal-regulated kinase 2) is activated during the induction of LTP in area CA1 of the hippocampus (7). Though this observation identifies the MAPK cascade as a potential component of the LTP induction cascades in area CA1, the physiologic necessity of MAPK activation during LTP induction remains to be established. To address this question, we have utilized the compound PD 098059 (8, 9), a recently described inhibitor of MEK, to block activation of the MAPK cascade during the delivery of LTP-inducing stimuli. Here we report that inhibition of the MAPK cascade greatly attenuates the induction but not expression of LTP in area CA1. Our observations provide an initial insight into a physiological role for the MAPK cascade in postmitotic neurons in the adult mammalian nervous system: the activity-dependent regulation of synaptic strength.
For hippocampal slice preparation, pharmacology, and electrophysiology, transverse hippocampal slices (400 µm) from 4-8 week old male Sprague-Dawley rats were prepared and maintained as described (7). Drug application, area CA1 subregion microdissection, tissue sonication, fractionation of soluble extracts, Western blotting, and densitometric analysis of phosphotyrosine immunoreactivity were conducted as described previously (7). All data are expressed as means ± S.E. In the present studies, we also utilized an antibody that selectively recognizes tyrosine phosphorylated extracellular signal-regulated kinases (New England Biolabs).
PD 098059 was dissolved in Me2SO and diluted into
artificial cerebrospinal fluid (ACSF; in mM 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 10 D-glucose, 2 CaCl2, 2 MgCl2,
saturated with 95% O2/5% CO2) to give the
desired final concentration (with a final Me2SO
concentration of 0.33%). For drug application (either PD 098059 or
0.33% Me2SO alone), ACSF solutions were maintained in a
32 °C water bath (to assure the complete solubility of PD 098059). For biochemical analysis of the effect of PD 098059 upon p42 MAPK activation, slices were preincubated in 50 µM PD 098059 or 0.33% Me2SO for 45 min to 1 h prior to NMDA
application (100 µM, 4) or HFS (strong induction
paradigm, see below). 0.33% Me2SO alone had no effect on
NMDA- or HFS-mediated p42 MAPK activation (not shown).
For extracellular field recordings of the Schaffer collateral synapses
in area CA1, stimulus intensity was adjusted to elicit a population
excitatory postsynaptic potential (pEPSP) that was approximately 25%
of the maximum response (initial slope typically about 1 mV/ms), and
responses (obtained at 0.05 Hz) were monitored for at least 15 min to
ensure a stable base line. Two LTP induction paradigms were employed:
1) modest: two trains of 1-s, 100-Hz stimulation at the test stimulus
intensity, with an intertrain interval of 20 s; and 2) strong:
three sets of tetani, each set spaced 10 min apart and consisting of
two trains of 1-s, 100-Hz stimulation at a stimulus intensity that
generated 75% of the maximal pEPSP, with an intertrain interval of
20 s. PD 098059 or 0.33% Me2SO was applied for 1 h prior to tetanization and was maintained throughout the recording
period. In initial experiments testing the effect of PD 098095 on LTP
elicited with the strong induction protocol, 30-ml solutions of normal
ACSF, 0.33% Me2SO, or 100 µM PD 098059 were
prepared and recycled through the recording chambers. Similar results
were also obtained using 50 µM PD 098059 (which, as with
all other experiments, was not recycled), and these data were pooled as
noted. For control experiments illustrated in Fig. 3B, an
NMDA receptor-mediated pEPSP was isolated by perfusing the slices with
a modified ACSF (as described, with the following exceptions: 4 mM CaCl2, 0 mM MgCl2,
and 20 µM 6-cyano-7-nitroquinoxaline-2,3-dione).
To test the
physiologic significance of MAPK activation during LTP induction, we
employed a recently described inhibitor of MEK, the dual-specific
protein kinase responsible for phosphorylating and activating MAPKs. We
first determined whether this MEK inhibitor, PD 098059, blocks MAPK
activation in area CA1 of the hippocampus. PD 098059 completely
abolished basal p42 MAPK phosphotyrosine immunoreactivity as well as
the NMDA- and HFS-mediated increase in p42 MAPK tyrosine
phosphorylation (Fig. 1A,
APT Western). Similar results were obtained using an
antibody that selectively recognizes phosphorylated, activated MAPKs
(Fig. 1A, -PMAPK Western).
Quantitative analysis of p42 MAPK APT immunoreactivity is shown in Fig. 1B. Application of either NMDA or HFS led to a significant increase in p42 MAPK phosphotyrosine immunoreactivity (NMDA application: Me2SO alone, 100 ± 13; n = 8; Me2SO/NMDA, 290 ± 38; n = 10 (p < 0.01, Student's t test); HFS application: Me2SO alone, 100 ± 13; n = 6; Me2SO/HFS, 159 ± 20; n = 6; (p < 0.05, Student's t test)). In the presence of the MEK inhibitor PD 098059, however, essentially all phosphotyrosine immunoreactivity (basal as well as the NMDA- and HFS-mediated increases) was eliminated (NMDA application: PD 098059 alone, 6 ± 12; n = 7; PD 098059/NMDA, 16 ± 6; n = 9; HFS application: PD 098059 alone, 9 ± 3; n = 3; PD 098059/HFS, 9 ± 3; n = 3). Together, these data demonstrate that NMDA receptor stimulation or LTP-inducing HFS elicits p42 MAPK activation in area CA1 and that PD 098059 blocks this activation.
PD 098059 Markedly Attenuates the Induction of LTP in Area CA1Having established that PD 098059 blocks MAPK activation in
area CA1, we next determined the effect of this inhibitor on LTP induction. LTP observed in control slices was not significantly different at any time point from LTP observed in slices treated with
the drug vehicle, 0.33% Me2SO (pEPSP slope at
t = 90 min, control: 173 ± 9%, n = 6; 0.33% Me2SO: 174 ± 11%, n = 4). Overall, stable LTP was elicited in nine of ten control slices,
with the initial slope of the pEPSP potentiated to 173 ± 7% of
pre-tetanus levels, 90 min following the tetanus (n = 10; Fig. 2A, closed circles). In slices treated with 50 µM PD 098059, however, a decaying potentiation was observed in six of six slices; at
t = 90 min, the pEPSP slope was 116 ± 7% of
base-line responses (Fig. 2A, open circles).
Responses in PD 098059-treated slices were significantly different from
control responses for each time point from t = 20 to 90 min (time 0-18 min, p > 0.05; 20-46 min,
p < 0.05; 48-90 min, p < 0.01.)
These data suggest that the MAPK cascade is critical for the induction
of stable LTP in hippocampal area CA1.
The Effect of PD 098059 upon LTP Induction Is Not Overcome with a Stronger Induction Protocol
In our initial studies, we employed a modest LTP-inducing paradigm consisting of a single set of tetani. The inhibition of certain enzymes (e.g. nitric-oxide synthase (10, 11), fyn kinase knockout mice (12)) blocks LTP induced with similar modest paradigms but not LTP induced with stronger paradigms (e.g. multiple sets of tetani), suggesting a modulatory rather than an obligatory function in LTP induction. To begin to determine whether MAPK activation plays an obligatory or modulatory role in LTP induction, we examined the effect of PD 098059 on LTP elicited with a strong induction protocol: multiple sets of tetani. PD 098059 also markedly attenuated the magnitude of LTP induced with this stronger induction protocol. In control slices, stable LTP was elicited in nine of nine slices, with the initial slope of the pEPSP potentiated to 226 ± 12% of pre-tetanus levels, 60 min following the final tetanus (n = 9; Fig. 2B, closed circles). These data are pooled averages of six control slices and three drug vehicle-treated slices (pEPSP at t = 60 min, control: 228 ± 7%, n = 6; 0.33% Me2SO: 224 ± 38%, n = 3). In slices treated with PD 098059, a decaying potentiation was observed in nine of ten slices; at t = 60 min, the pEPSP slope was 132 ± 13% of base-line responses (n = 10). These data are pooled averages of seven slices treated with 100 µM PD 098059 and three slices treated with 50 µM PD 098059 (100 µM: blocked in six of seven slices; pEPSP slope at t = 60 min, 132 ± 19%; 50 µM: blocked in three of three slices; pEPSP slope at t = 60 min, 132 ± 7%; these data were not significantly different and were thus combined). Responses in PD 098059-treated slices were significantly different from control slices for all time points from 2 to 60 min (p < 0.01 for each). That the effect of PD 098059 is similar on LTP elicited with both modest and strong stimuli (i.e. normal initial responses followed by a slowly decaying potentiation) suggests that the MAPK cascade is an obligatory component of the biochemical induction cascades that subserve this form of synaptic potentiation.
PD 098059 Does Not Affect Basal Synaptic Transmission, NMDA Receptor-mediated Transmission, or HFS-mediated DepolarizationIn control experiments, we found that prolonged application of PD 098059 had no effect upon basal synaptic transmission (Fig. 3A). Following 2.5 h of PD 098059 application (50 µM), average initial slope of the pEPSP were 87 ± 10% of base-line responses (n = 3, not significant). Similar results were also obtained with 100 µM PD 098059 applied for 1 h (not shown).
Similarly, PD 098059 had no effect upon NMDA receptor-mediated pEPSPs elicited in the presence of 6-cyano-7-nitroquinoxaline-2,3-dione, an antagonist of non-NMDA receptor subtypes of glutamate receptors (Fig. 3B). NMDA receptor-mediated responses following 1 h of 100 µM PD 098059 application were not different from base-line levels (amplitude: 99 ± 2% of base line; initial slope: 94 ± 4% of base line; n = 3; p > 0.05 for each). Application of a competitive antagonist of the NMDA receptor, D,L-2-amino-5-phosphonovalerate (D,L-APV, 50 µM), completely and reversibly blocked the pEPSP.
Finally, PD 098059 did not alter the depolarization responses to high frequency stimulation (Fig. 3C). No differences were observed for either the total depolarization (integral over entire HFS response) or the steady-state depolarization (average over last 50 ms) observed during the initial tetanus for control, Me2SO- and PD 098059-treated slices (Integral: (control: 100 ± 6.5%, n = 6; Me2SO: 88 ± 12%, n = 4; PD 098059: 100.5 ± 13.5%, n = 6). Steady-state depolarization: (control: 100 ± 7.8%, n = 6; Me2SO: 88 ± 14%, n = 4; PD 098059: 97.4 ± 11.7%, n = 6)). Overall, these results demonstrate that PD 098059 does not affect basal synaptic transmission, NMDA receptor-mediated responses, or HFS-mediated depolarization.
PD 098059 Does Not Affect Protein Kinase C, Protein Kinase A, or Calcium/Calmodulin-dependent Protein Kinase IIPrevious reports have demonstrated that PD 098059 shows a high degree of selectivity for MEK, because 50 µM PD 098059 did not alter the activity of over 20 protein kinases, including serine/threonine kinases such as PKC, PKA, c-Raf, and c-Jun N-terminal kinase and tyrosine kinases such as the epidermal growth factor receptor (8, 9). We replicated and extended these observations by testing the effect of PD 098059 on the activity in vitro of three kinases required for LTP induction: PKA, PKC, and CaMKII. PD 098059 (used at concentrations ranging from 1 to 300 µM) did not affect the activity of these protein kinases (data not shown), suggesting that PD 098059 does not attenuate LTP induction by inhibiting any of these three protein kinases previously implicated in LTP induction.
The MAPK Cascade Is Not Persistently Activated during LTP ExpressionOngoing kinase activity is thought to underlie LTP
expression (13, 14), and both PKC and CaMKII are persistently activated during LTP expression (15-17). To determine whether ongoing MEK activation is required for LTP expression, we examined the effect of PD
098059 on established LTP in area CA1. In these experiments, stable LTP
was elicited in six of six slices using the modest induction protocol,
and PD 098059 was applied for 1 h beginning 30 min after
tetanization. PD 098059 had no effect on LTP expression (Fig.
4A). Consistent with these
observations, PD 098059 also had no effect upon the expression of LTP
elicited with the stronger induction paradigm (not shown). Note that a
1-h application of PD 098059 is sufficient for complete reversal of
basal p42 MAPK tyrosine phosphorylation (Fig. 1). These results suggest
that ongoing activation of the MAPK cascade is not necessary for
LTP expression.
One important caveat to this conclusion, however, is that MAPKs, like PKC and CaMKII, can be persistently activated via phosphorylations (18). Because MEK inhibition would have no effect on persistently activated MAPKs, it remains a possibility that persistently activated MAPKs contribute to LTP expression. We tested this hypothesis by examining p42 MAPK activation levels during LTP expression. No difference in either phosphotyrosine or phospho-MAPK immunoreactivity for p42 MAPK was observed between control and LTP samples taken 45 min following delivery of the strong HFS paradigm, indicating that p42 MAPK is not persistently activated during LTP expression (Fig. 4, B and C; normalized p42 MAPK phosphotyrosine immunoreactivity: 114 ± 20% of control; n = 6; not significant).
The abundant expression of MAPKs in mature neurons of the developed nervous system suggests a specialized function for the MAPK cascade in these postmitotic cells. At present, however, little is known regarding either the regulatory mechanisms or physiologic roles of MAPKs in neurons. Using both electrophysiology and biochemistry, we have found that the MAPK cascade is involved in the activity-dependent modulation of synaptic strength, the leading candidate for the neuronal basis of information storage. In particular, we have identified p42 MAPK as a critical component of the biochemical cascades that underlie the induction of hippocampal long term potentiation.
Finally, it is well established that important early events in LTP induction include NMDA receptor stimulation, subsequent postsynaptic calcium influx, and the initiation of several protein kinase cascades. MAPKs appear to be localized in both presynaptic and postsynaptic structures (4, 19) and are thus well positioned to participate in these early events. Interestingly, consideration of several MAPK substrates suggest possible functions of MAPKs in these induction cascades: generation of putative retrograde messengers (e.g. cytosolic phospholipase A2 and the formation of arachadonic acid/platelet-activating factor) (20-22), synaptic vesicle regulation (e.g. synapsin) (19, 23), cytoskeletal modulation (e.g. MAP2 protein) (24), and regulation of gene expression (e.g. transcription factors such as Elk-1) (reviewed in Ref. 25), and CREB (26). In short, the necessity of MAPK activation for LTP induction and the known localization of MAPKs in postmitotic neurons strongly suggest that MAPKs play a critical role in LTP induction cascades. An important step in understanding the function of MAPKs in the strengthening of synaptic connections will be the determination of which MAPK substrates are utilized during LTP.
We thank Dr. Alan Saltiel for generously providing PD 098059 for these studies and Erik Roberson, Coleen Atkins, and Anne Anderson for critical readings of the manuscript. We also thank Richard Dickason, David Leopold, William J. Martin, and Mark Jacobs for useful discussions.