A role for superoxide in protein kinase C activation and induction of long-term potentiation.

The induction of several forms of long-term potentiation (LTP) of synaptic transmission in the CA1 region of the mammalian hippocampus is dependent on N-methyl-D-aspartate receptor activation and the subsequent activation of protein kinase C (PKC), but the mechanisms that underlie the regulation of PKC in this context are largely unknown. It is known that reactive oxygen species, including superoxide, are produced by N-methyl-D-aspartate receptor activation in neurons, and recent studies have suggested that some reactive oxygen species can modulate PKC in vitro. Thus, we have investigated the role of superoxide in both the induction of LTP and the activation of PKC during LTP. We found that incubation of hippocampal slices with superoxide scavengers inhibited the induction of LTP. The effects of superoxide on LTP induction may involve PKC, as we observed that superoxide was required for appropriate modulation of PKC activation during the induction of LTP. In this respect, superoxide appears to work in conjunction with nitric oxide, which was required for a portion of the LTP-associated changes in PKC activity as well. Our observations indicate that superoxide and nitric oxide together regulate PKC in a physiologic context and that this type of regulation occurs during the induction of LTP in the hippocampus.

The induction of several forms of long-term potentiation (LTP) of synaptic transmission in the CA1 region of the mammalian hippocampus is dependent on N-methyl-D-aspartate receptor activation and the subsequent activation of protein kinase C (PKC), but the mechanisms that underlie the regulation of PKC in this context are largely unknown. It is known that reactive oxygen species, including superoxide, are produced by N-methyl-Daspartate receptor activation in neurons, and recent studies have suggested that some reactive oxygen species can modulate PKC in vitro. Thus, we have investigated the role of superoxide in both the induction of LTP and the activation of PKC during LTP. We found that incubation of hippocampal slices with superoxide scavengers inhibited the induction of LTP. The effects of superoxide on LTP induction may involve PKC, as we observed that superoxide was required for appropriate modulation of PKC activation during the induction of LTP. In this respect, superoxide appears to work in conjunction with nitric oxide, which was required for a portion of the LTP-associated changes in PKC activity as well. Our observations indicate that superoxide and nitric oxide together regulate PKC in a physiologic context and that this type of regulation occurs during the induction of LTP in the hippocampus.
Hippocampal long-term potentiation (LTP) 1 is a long-lasting synaptic enhancement that may be involved in certain types of mammalian learning and memory (1). It generally is agreed that in area CA1 of the hippocampus the induction of most forms of LTP induced by high frequency stimulation (HFS) is dependent on Ca 2ϩ entry into the postsynaptic neuron triggered by N-methyl-D-aspartate (NMDA) receptor activation (2)(3)(4). The exact biochemical sequence of events following Ca 2ϩ influx into the postsynaptic cell is unknown, but it is clear that Ca 2ϩ -dependent protein kinases are involved in the induction of LTP (5,6). The activation of protein kinase C (PKC) is one of the requisite biochemical steps necessary for the induction of LTP, as selective PKC inhibitors can block induction of LTP (5,7,8). Furthermore, both the second messenger-independent activity of PKC (autonomous activity; see Ref. 9) and total, cofactor-stimulated PKC activity (cofactor-dependent activity; see Refs. 9 and 10) are increased shortly after LTP-inducing HFS. Mechanisms that have been proposed for the activation of PKC immediately following the induction of LTP (15 s to 2 min after the final HFS) include translocation of PKC to the membrane (11,12), conformational changes in PKC that result in the unmasking of activator sites (10), and decreases in protein phosphatase activity that might result in increased phosphorylation and activity of PKC (10).
An additional mechanism that might be involved in the activation of PKC during the induction of LTP is an oxidative mechanism involving reactive oxygen species (ROS), particularly the superoxide anion (O 2 . ). A variety of evidence suggest this possibility. First, levels of superoxide may increase during the induction of LTP, as it has recently been demonstrated that NMDA receptor activation in area CA1 of hippocampal slices results in the production of superoxide (13). Second, if superoxide levels increase, PKC is a possible target, given that treatment of several different types of cells and of hippocampal homogenates with superoxide or other ROS has been shown to increase both autonomous and cofactor-dependent PKC activity (14 -16). Coupled with recent studies implicating superoxide dismutase (SOD), which catalyzes removal of superoxide, in neuronal function and the neuropathology of Alzheimer's disease (17) and amyotrophic lateral sclerosis (18), these findings prompted us to determine whether superoxide was involved in the induction of LTP, and if so, to determine whether superoxide was involved in the regulation of PKC activity during LTP.

EXPERIMENTAL PROCEDURES
Preparation of Hippocampal Slices and Induction of LTP-Hippocampi from male Sprague-Dawley rats (100 -150 g) were removed and 400-m slices were prepared with a McIlwain tissue chopper. The slices were perfused for 1 h with a standard saline solution (124 mM NaCl, 4.4 mM KCl, 26 mM NaHCO 3 , 10 mM D-glucose, 2 mM CaCl 2 , 2 mM MgCl 2 , gassed with 95% O 2 , 5% CO 2 , pH 7.4) in an interface tissue slice chamber at 32°C. After 1 h, baseline responses to Schaffer collateral stimulation in the CA1 region were monitored for at least 20 min before LTP-inducing HFS to assure a stable baseline. Test stimuli were given at a current that produced 50% of the maximum population EPSP (usually between 30 and 40 A). Responses were measured every 2.5 min as an average of four individual traces.
In our previous study (9), LTP-inducing HFS consisted of three sets of stimuli, each set was delivered 5 min apart. Each set included two 1-s trains of stimuli (100 Hz), given 20 s apart at the current needed to elicit a maximum population EPSP. Using this stimulus paradigm, cofactor-dependent PKC activity increased 2 min after the final HFS train. However, because this paradigm spanned a period of 10 min, the LTP-associated increase in cofactor-dependent PKC activity was actually measured 12 min after the first HFS train.
To determine changes in PKC activity that more accurately reflect changes during the induction of LTP, we modified our LTP-inducing HFS in this study to consist of three 1-s trains of stimuli (100 Hz), given 20 s apart at the current needed to elicit a maximal population EPSP (usually between 60 and 80 A). In experiments in which PKC activity was assayed, slices were frozen 2 min after the final HFS train, dissected, and homogenized. In experiments in which LTP was examined, responses were measured every 2.5 min as an average of four individual traces for 1 h after the final HFS. Post-HFS responses were generated with the same test stimuli used prior to HFS. LTP was defined as an increase of 20% or greater in the initial EPSP slope.
Application of ROS Scavengers, Spin Trap, or NOS Inhibitor to Hippocampal Slices-Slices were prepared as described above. After a 1-h incubation in standard saline solution, hippocampal slices were perfused with standard saline solution containing SOD (Calbiochem), SOD/catalase (Calbiochem), DMPO (OMRF Spin Trap Source), or NOArg (Sigma) for at least 30 min prior to LTP-inducing HFS. In more than 90% of the experiments, slices given HFS in the presence of SOD, SOD/catalase, DMPO, or NOArg were compared with slices from the same animal given HFS in the absence of the agents in an adjacent recording chamber (LTP controls).
Protein Kinase C Assays-Autonomous and cofactor-dependent PKC activity in control and potentiated tissue from hippocampal area CA1 was determined using the selective PKC substrate NG (28 -43) , as described previously (9). We defined autonomous PKC activity as activity assayed in the presence of 2 mM EGTA. Cofactor-dependent PKC activity was defined as activity assayed in the presence of 100 M CaCl 2 , 320 g/ml phosphatidylserine, and 30 g/ml dioctonoylglycerol with the autonomous PKC activity subtracted (9). Expressed in terms of absolute specific activities, autonomous NG (28 -43) phosphorylation was 0.65 Ϯ 0.10 pmol/min/g in control slices and 1.10 Ϯ 0.33 pmol/min/g in slices given LTP-inducing HFS; cofactor-dependent NG (28 -43) phosphorylation was 7.07 Ϯ 0.63 pmol/min/g protein in control slices and 11.27 Ϯ 1.30 pmol/min/g protein in slices given LTP-inducing HFS.

Superoxide Modulates the Induction of LTP-
The hypothesis that superoxide plays a role in the induction of LTP was investigated in area CA1 of rat hippocampal slices. LTP was induced with HFS consisting of three 1-s trains of 100-Hz stimulation delivered 20 s apart at a stimulus intensity that produced a maximum initial slope of the extracellular field excitatory postsynaptic potentials (EPSPs). This stimulus paradigm resulted in LTP in 96% (25 out of 26) of the control slices for the experiments shown in Fig. 1. The slope of the EPSPs was increased to 155.7 Ϯ 4.0% (n ϭ 26) of pre-HFS control levels (within-slice comparison) 60 min after the final HFS train (Fig.  1, A-C). To determine whether superoxide was involved in LTP induction, we gave slices HFS in the presence of the superoxide scavenger SOD. SOD is highly specific for superoxide and is not known to react enzymatically with other ROS (21). LTP was blocked in eight out of nine slices 60 min after the final HFS train (average slope of EPSPs ϭ 101.0 Ϯ 7.5% of control, n ϭ 9) (Fig. 1A). In contrast, slices from the same animal in an adjacent recording chamber given HFS without SOD exhibited LTP in all nine experiments (average slope of EPSPs ϭ 158.7 Ϯ 7.6% of control, n ϭ 9) (Fig. 1A). These data suggest that superoxide plays a role in the induction of LTP.
An alternative explanation of the results in Fig. 1A is that one of the products of the reaction between SOD and superoxide inhibits the induction of LTP. Specifically, SOD converts superoxide to H 2 O 2 and O 2 (22), and H 2 O 2 has been shown to inhibit LTP (23). To address this possibility, slices were given HFS in the presence of bath-applied SOD along with catalase, an enzyme which scavenges H 2 O 2 and catalyzes its conversion to O 2 and water (22). In these experiments, LTP was inhibited in 10 out of 15 slices (average slope of EPSPs ϭ 113.3 Ϯ 4.7% of control for all 15 slices and 102.8 Ϯ 2.4% of control for the 10 slices that did not meet our criteria for LTP) (Fig. 1B). In control experiments, LTP was produced in 12 out of 13 slices (average slope of EPSPs ϭ 150.4 Ϯ 4.7% of control, n ϭ 13) (Fig.  1B). Interestingly, the five slices incubated with SOD and catalase that did exhibit LTP after HFS were potentiated to a lesser extent when compared with LTP control slices (average slope of EPSPs ϭ 134.6 Ϯ 7.6% of control, n ϭ 5). Slices incubated with catalase alone exhibited LTP (average slope of EPSPs ϭ 168.5 Ϯ 12.5% of control, n ϭ 3) similar to that observed in LTP control slices (average slope of EPSPs ϭ 166.6 Ϯ 7.4% of control, n ϭ 3). Our interpretation of these data is that the effects of SOD on LTP induction are not solely a result of increased levels of H 2 O 2 . However, our findings do not exclude the possibility that H 2 O 2 not removed by catalase may contribute to the observed effects of SOD on LTP.
As mentioned, SOD is a specific enzyme not known to catalyze the removal of any ROS besides superoxide. However, SOD, like most other proteins, contains histidine, tryptophan, and methionine, each of which can react nonenzymatically with other ROS (21). To demonstrate that the effects of SOD are not due to these nonenzymatic reactions with other ROS, we incu- bated slices with boiled SOD and catalase and then examined their effect on the induction of LTP. In slices incubated with the heat-inactivated enzymes, LTP was indistinguishable (average slope of EPSPs ϭ 162.3 Ϯ 11.3% of control, n ϭ 4) (Fig. 1C) from LTP in control slices (average slope of EPSPs ϭ 154.4 Ϯ 9.4% of control, n ϭ 4) (Fig. 1C). Thus, the effects of SOD and catalase on LTP are not attributable to nonenzymatic removal of ROS other than superoxide.
Although the concentrations of SOD (121 units/ml) and catalase (260 units/ml) we used have been shown in previous studies to have no effect on baseline electrophysiological responses in area CA1 of guinea pig hippocampal slices (24), we performed a series of experiments to verify that the superoxide scavengers used in our studies did not have nonspecific effects on synaptic transmission. We observed no alteration of electrophysiological responses to test stimuli in area CA1 in rat hippocampal slices exposed to SOD and catalase ( Fig. 2A). Moreover, SOD and catalase had no effect on the NMDA component of the EPSP measured in the presence of 20 M 6-cyano-7nitroquinoxaline-2,3-dione (Fig. 2B). Finally, SOD and catalase had no effect on high-frequency synaptic transmission, as measured by either the total depolarization or the steady-state depolarization produced during HFS (Fig. 2C). These results suggest that inhibition of LTP induction by superoxide scavengers is not due to effects on baseline synaptic transmission or NMDA receptor function. Overall, these data suggest that superoxide has a role in modulating the induction of LTP in area CA1.
To demonstrate that a structurally and functionally distinct superoxide scavenger would also inhibit LTP induction, we incubated slices with 10 mM 5,5-dimethylpyrroline 1-oxide (DMPO), a superoxide and hydroxyl radical electron spin resonance (ESR) spin trap that previously has been shown to block NMDA-induced, superoxide-dependent neurotoxicity in cerebellar granule neurons (25). DMPO inhibited LTP in three out of five experiments and attenuated LTP in the other two experiments (average slope of EPSPs ϭ 120.6 Ϯ 4.8% of control, n ϭ 5) (Fig. 3). In contrast, all five control slices exhibited LTP  4). B, NMDA receptor-mediated EPSPs were isolated in 20 M 6-cyano-7-nitroquinoxaline-2,3-dione, 0 mM MgCl 2 , and 4 mM CaCl 2 . The EPSPs recorded in this manner were NMDA receptor-dependent, as they were blocked completely by 50 M APV (left). Representative traces taken during baseline recording (a) and after APV application (b) are shown at the top (calibration bars: 2 mV and 4 ms) and ensemble averages (n ϭ 5) at the bottom. Addition of SOD (121 units/ml) and catalase (260 units/ml) for 60 min had no effect on NMDA receptor-mediated EPSPs (right). Representative traces during baseline (a) and after SOD/catalase (b) are at the top and ensemble averages (n ϭ 7) at the bottom. C, after recording stable baseline responses for 10 min in standard saline solution, slices were stimulated with LTP-inducing HFS, as described under "Experimental Procedures." Left, the responses to the first HFS were analyzed by both integrating the entire HFS-response trace (Integral) and measuring the level of steady-state depolarization during HFS (SS Depol; averaged over the last 50 ms of the HFS). In addition, the integral measurement was made for the first and the second half of the HFS separately; no differences between control and SOD/catalase were observed in either measurement (not shown). Measurements in all slices were normalized to the amplitude of the EPSP produced by a single pulse at the stimulus intensity used for the HFS; data from SOD/catalase slices were then expressed as percent of control (n ϭ 3). Right, representative HFS response traces from control and SOD/catalase slices, and the two traces overlaid (calibration bar: 2 mV and 90 ms). (average slope of EPSPs ϭ 160.4 Ϯ 7.8% of control, n ϭ 5). In addition, DMPO had no effect on electrophysiological responses to test stimuli (data not shown). These results are consistent with results observed with SOD and catalase and provide further evidence of a role for superoxide in the induction of LTP. We conclude from this series of experiments that removing superoxide reduces the probability of LTP induction and the magnitude of LTP when potentiation occurs. Because neither SOD, catalase, nor DMPO can cross cell membranes, these data also suggest that superoxide may act as a transcellular messenger in LTP.
The results of the experiments in Figs. 1 and 3 indicate that superoxide modulates the induction of LTP, but they do not address the possibility of a role for superoxide in the maintenance of LTP. To test this possibility, we induced LTP and added SOD and catalase 20 min after the final LTP-inducing HFS. Pre-established LTP was unaffected by addition of SOD and catalase when monitored up to 1 h after the addition of the enzymes (Fig. 4). These results indicate that superoxide modulates the induction, but not the maintenance, of LTP.
Superoxide Contributes to PKC Activation during LTP-We next sought to begin to elucidate the biochemical targets of superoxide in LTP. We began by testing the hypothesis that superoxide regulates PKC during the induction of LTP. We previously have shown that the induction of LTP is associated with increases in both autonomous (9) and cofactor-dependent PKC activity (9). To investigate a role for superoxide in regulating PKC activity during the induction of LTP, we first measured PKC activity after LTP-inducing HFS (see "Experimental Procedures"). Slices were frozen 2 min after the final train of HFS, dissected, and homogenized. Homogenates from control slices and slices given HFS were assayed for autonomous and cofactor-dependent PKC activity using the selective PKC substrate peptide NG (28 -43) (9,26,27). In comparisons of slices given LTP-inducing HFS and control slices that received only test stimuli, we observed an increase in both autonomous and cofactor-dependent PKC activity 2 min after the HFS (183.4 Ϯ 18.6 and 144.8 Ϯ 11.6% of control, respectively, n ϭ 12, p Ͻ 0.05) (Figs. 5, A and B). These results agree with previous work demonstrating that HFS results in an increase in both autonomous (9) and cofactor-dependent PKC activity (9,10), and are consistent with physiological studies showing that blockade of PKC inhibits induction of LTP (5,7,8).
To test the hypothesis that superoxide mediates the increases in autonomous and cofactor-dependent PKC activity associated with the induction of LTP, we assayed PKC activity after incubating slices with SOD/catalase and delivering LTP-inducing HFS to area CA1. We observed an attenuated but significant increase in autonomous PKC activity in slices given HFS in the presence of SOD/catalase (146.8 Ϯ 20.1% of control, n ϭ 8, p Ͻ 0.05) (Fig. 5A). In contrast, the increase in cofactordependent PKC activity associated with the induction of LTP was blocked completely (91.4 Ϯ 9.6% of control, n ϭ 8) (Fig. 5B).
In control experiments, addition of SOD/catalase to both the slices after LTP-inducing HFS and to the homogenization buffer had no significant effect on the LTP-associated increases in either autonomous PKC activity (177.5 Ϯ 23.4% of control, n ϭ 3) or cofactor-dependent PKC activity (160.3 Ϯ 15.7% of control, n ϭ 3). These results suggest that the increase in cofactor-dependent PKC activity associated with the induction of LTP is mediated either directly or indirectly by superoxide. Furthermore, these data suggest that superoxide contributes to the generation of autonomously active PKC, consistent with the hypothesis that multiple mechanisms of PKC activation are associated with the induction of LTP (9,12,28).
Nitric Oxide Also Contributes to PKC Activation during LTP-The results described above indicate that superoxide can modulate induction of LTP, possibly through interactions with PKC. In many systems, superoxide exerts its effects in conjunction with nitric oxide (NO) (see Ref. 22). Furthermore, it is known that inhibitors of nitric oxide synthase (NOS) modulate

FIG. 5. Effect of SOD and catalase on LTP-associated increases
in PKC activity. LTP-inducing HFS was delivered to the CA1 region of hippocampal slices that were assayed for either autonomous (A) or cofactor-dependent (B) PKC activity using NG (28 -43) as an exogenous substrate. Experimental slices were compared with control slices from the same hippocampus. Autonomous and cofactor-dependent PKC activity are expressed as percent of control for control slices, slices given LTP-inducing HFS (n ϭ 12), and slices given HFS in the presence of bath applied SOD (121 units/ml) and catalase (260 units/ml) (n ϭ 8). Error bars are S.E. for the indicated number of determinations. Significance was determined with a paired Student's t test. * denotes p Ͻ 0.05. the induction of LTP induced by many stimulus paradigms (29 -34). In this study, slices given HFS in the presence of the NOS inhibitor N-nitro-arginine (NOArg, 100 M) exhibited LTP (average slope of EPSPs ϭ 135 Ϯ 5% of control, n ϭ 4). However, the slices were potentiated to a lesser extent (p Ͻ 0.05, paired Student's t test) when compared with LTP control slices from the same animal in an adjacent recording chamber (average slope of EPSPs ϭ 166 Ϯ 10% of control, n ϭ 4). Therefore, we next sought to determine whether NO might be involved in the regulation of PKC activity during the induction of LTP. To examine interactions between NO and PKC during the induction of LTP, hippocampal slices were incubated with NOArg (100 M), given LTP-inducing HFS, and assayed for changes in PKC activity. As with superoxide scavengers, in the presence of NOArg we observed an attenuated but significant increase in autonomous PKC activity (141.4 Ϯ 20.5% of control, n ϭ 10, p Ͻ 0.05) (Fig. 6A), and complete inhibition of the increase in cofactor-dependent PKC activity (98.7 Ϯ 14.6% of control, n ϭ 10) (Fig. 6B) associated with the induction of LTP. These results suggest that activation of NOS and production of NO play an important role in the activation of PKC during LTP induction.
There is an alternative interpretation of these results, as NOS is known to produce superoxide as a by-product (35). If NOS were in fact the source of superoxide production during LTP induction, the requirement for NOS activity could reflect the need for superoxide, rather than a direct role for NO in activating PKC. To distinguish between these two possibilities, we tested whether hemoglobin, which binds NO avidly (36), but does not block effects of superoxide alone on PKC activity 2 could block the changes in PKC activity associated with the induction of LTP. We delivered LTP-inducing HFS to hippocampal slices incubated with 20 M hemoglobin and then assayed PKC activity. The increase in autonomous PKC activity associated with the induction of LTP was attenuated (146.7 Ϯ 10.6% of control, n ϭ 8, p Ͻ 0.05), as was the increase in cofactor-dependent PKC activity (120.4 Ϯ 19.6% of control, n ϭ 8, p Ͼ 0.05) in slices incubated with hemoglobin. We interpret these results to suggest that NO, like superoxide, plays an important role in PKC activation during the induction of LTP. However, it should be noted that hemoglobin can also bind other ROS, including superoxide, albeit with a lower affinity than NO. Therefore, we cannot completely rule out the possibility that hemoglobin is blocking effects of superoxide or another ROS on PKC activity. It is interesting to note that hemoglobin, like SOD/catalase and DMPO, is confined to the extracellular space; the results therefore are consistent with the hypothesis that transcellular signaling plays a role in PKC activation by superoxide and NO during LTP. DISCUSSION The results in this report provide evidence of a role for superoxide in the induction of LTP. Our experiments with superoxide scavengers indicate that superoxide increases the probability of LTP induction and the magnitude of LTP when it occurs. Moreover, the demonstration of a superoxide-dependent change in PKC activity during the induction of LTP provides evidence that superoxide levels increase during the induction of LTP. Thus, we have shown that superoxide is likely to be produced during the induction of LTP, and that it has both biochemical and physiologic effects.
The idea of a role for superoxide in synaptic plasticity is consistent with a number of findings from previous studies. First, NMDA receptor activation, a key step in most forms of LTP, results in the production of superoxide in area CA1 of hippocampal slices (13) and in other types of neurons (25). Second, activation of other glutamate receptors can produce superoxide through a number of biochemical pathways (37). In addition, superoxide increases the release of glutamate in hippocampal slices (38), which suggests a means by which it might modulate synaptic efficacy. Finally, it is interesting to note that although the intracellular form of Cu/Zn-SOD, SOD1, is abundantly expressed in the dentate gyrus and area CA3 of the hippocampus, it is expressed to a much lesser degree in area CA1, the hippocampal subregion studied herein (39); thus, area CA1 is likely to have less capacity for removal of superoxide, consistent with a physiologic role for superoxide in this region.
There are a variety of possible sources for superoxide production during LTP induction. First, NOS, which is known to be activated during the induction of LTP (33), is a source of superoxide (35). The fact that NOS inhibitors and superoxide scavengers have similar effects is consistent with a role for NOS in producing superoxide, although it is difficult to dissect this role of NOS from its required role in producing NO. Another possible source of superoxide in LTP is arachidonic acid metabolism. Superoxide is produced by lipoxygenase, which initiates leukotriene synthesis from arachidonic acid, and prostaglandin endoperoxide synthase, which initiates prostaglandin synthesis from arachidonic acid (40). Interestingly, 2 L. T. Knapp and E. Klann, unpublished observations. FIG. 6. Effect of NOArg on LTP-associated increases in PKC activity. LTP-inducing HFS was delivered to the CA1 region of hippocampal slice that were assayed for either autonomous (A) or cofactordependent (B) PKC activity using NG (28 -43) as an exogenous substrate. Experimental slices were compared with control slices form the same hippocampus. Autonomous and cofactor-dependent PKC activity are expressed as a percent of control for control slices, slices given LTPinducing HFS (n ϭ 12), and slices given HFS in the presence of bathapplied 100 M NOArg (n ϭ 10). Error bars are S.E. for the indicated number of determinations. Significance was determined with a paired Student's t test. * denotes p Ͻ 0.05. arachidonic acid metabolism increases during LTP (41,42) and a lipoxygenase inhibitor, nordihydroguaiaretic acid, inhibits LTP induction (31,41,43), consistent with a possible role of arachidonic acid metabolism in superoxide production during the induction of LTP. Other possible sources of superoxide include the mitochondrial electron transport chain (13,44) and xanthine oxidase, which could be produced by protease activation during LTP induction (45).
In addition to demonstrating a role for superoxide in modulating LTP induction, we have shown that activation of PKC may be a biochemical effect of superoxide production during the induction of LTP. Superoxide is required for increases in cofactordependent PKC activity and contributes to increases in autonomous PKC activity during LTP induction. In addition, because superoxide appears to work in conjunction with NO to mediate PKC activity during the induction of LTP, our studies provide evidence that PKC could be a biochemical effector of NO in LTP induction.
Previous observations and our present studies suggest that superoxide and NO, and the subsequent increase in cofactordependent PKC activity (Figs. 5B and 6B), are likely to be modulatory rather than obligatory for the induction of LTP. First, although SOD/catalase blocked LTP in 10 out of 15 experiments, in five experiments an attenuated potentiation was observed (Fig. 1B). Second, although SOD/catalase, NO-Arg, and hemoglobin blocked the HFS-induced increase in cofactor-dependent PKC activity, none completely blocked the LTP-associated increase in autonomous PKC activity (Figs. 5A and 6A). This last observation would suggest that there are multiple mechanisms of PKC activation after LTP-inducing stimulation (9,12,28), 3 which include activation of autonomous PKC that is obligatory (5,7,8) and activation of cofactordependent PKC that is modulatory. Thus, it is tempting to speculate that the increase in cofactor-dependent PKC activity generated by superoxide and NO together, although transient, is involved in the generation of additional autonomously active PKC necessary for the maximum expression of LTP. For example, oxidation of PKC may modify the enzyme so that there is enhanced autophosphorylation, a mechanism that has been proposed to contribute to LTP (9).
How superoxide and NO might interact to regulate PKC activity is an interesting but unresolved question. Our data suggest that both superoxide and NO are necessary for the increase in PKC activity associated with the induction of LTP (Figs. 5 and 6). Superoxide and NO are known to react rapidly to form the highly reactive peroxynitrite intermediate, a potent oxidant that mediates oxidation of protein sulfhydryls (46). Furthermore, peroxynitrite decomposes to form hydroxyl radical (22), which is likely able to activate PKC (16). The extent to which peroxynitrite and hydroxyl radical are involved in PKC activation by superoxide and NO during LTP is an area of active research.
The cellular location of the superoxide-and NO-mediated changes in PKC activity is another interesting question. LTP is associated with enhanced phosphorylation of both pre-and postsynaptic PKC substrates (47)(48)(49)(50). In this regard, our results with SOD/catalase, DMPO, and hemoglobin are interesting, as all are confined to the extracellular space. The ability of these compounds to block the effects of superoxide and NO on PKC activity suggests that these species may serve as transcellular signals. It is well known that NO is a membranepermeable molecule. In contrast, it is not clear how the negatively charged superoxide molecule would be able to cross membranes, although it has been proposed that superoxide can traverse cell membranes through anion channels. Consistent with this notion, it has been reported that NMDA receptor activation in cerebellar granule neurons results in the generation of superoxide that can be trapped extracellularly (25). An alternative explanation for our results is that the scavenging compounds could be acting as extracellular sinks to pull ROS out of the postsynaptic neurons, preventing activation of PKC postsynaptically.
Although ROS traditionally have been thought of as toxic agents involved in cell injury, they recently have been speculated to act as cellular messengers. This is evidenced by reports showing that ROS increase: 1) the phosphorylation of the ribosomal S6 protein (51); 2) PKC activity in mouse epidermal JB6 cells (15); 3) tyrosine phosphorylation in human neutrophils (52); and 4) p21 ras GTPase activity in PC12 cells (53). In addition, it has been shown that generation of hydrogen peroxide is necessary for platelet-derived growth factor signal transduction (54) and may be involved in lysophosphatidic acid signal transduction (55). To our knowledge, the results in this report are the first to demonstrate ROS regulation of protein kinase activity in the brain and are consistent with the idea that ROS act as normal cellular messengers under certain physiological conditions.
Finally, our data are consistent with the hypothesis that ROS modulate the early expression of LTP via interactions with PKC. However, ROS also have been shown to inactivate protein phosphatases, including calcineurin (56). Calcineurin is a calcium/calmodulin-dependent protein phosphatase that has been shown to be necessary for NMDA receptor-dependent hippocampal long-term depression (57), a use-dependent decrease in synaptic strength. Thus, the production of ROS after induction of LTP, in addition to increasing PKC activity, might also serve to inactivate protein phosphatases such as calcineurin, thereby preventing long-term depression and favoring the expression of LTP.
In closing, the data in this report indicate that superoxide modulates the induction of LTP, and that superoxide, in conjunction with NO, mediates increases in PKC activity associated with the induction of LTP. These findings present an interesting physiologic and biochemical context for the effects of ROS in the nervous system.