|Home | About | Journals | Submit | Contact Us | Français|
The transient, A-type K+ current (IA) controls the excitability of CA1 pyramidal neuron dendrites by regulating the back-propagation of action potentials and by shaping synaptic input. Dendritic A-type K+ channels are targeted for modulation during long-term potentiation (LTP) and we have recently shown that activity-dependent internalization of the A-type channel subunit Kv4.2 enhances synaptic currents (Kim et al., 2007). However, the effect of changes in IA on the ability to induce subsequent synaptic plasticity (meta-plasticity) has not been investigated. Here we show that altering functional Kv4.2 expression level leads to a rapid, bidirectional remodeling of CA1 synapses. Neurons exhibiting enhanced IA showed a decrease in relative synaptic NR2B/NR2A subunit composition and did not exhibit LTP. Conversely, reducing IA by expression of a Kv4.2 dominant negative or through genomic knockout of Kv4.2 led to an increased fraction of synaptic NR2B/NR2A and enhanced LTP. Bidirectional synaptic remodeling was mimicked in experiments manipulating intracellular Ca2+ and dependent on spontaneous activation of NMDA receptors and CaMKII activity. Our data suggest that A-type K+ channels are an integral part of a synaptic complex that regulates Ca2+ signaling through spontaneous NMDAR activation to control synaptic NMDAR expression and plasticity.
In hippocampal CA1 pyramidal neurons, the total outward K+ current consists of a transient or rapidly inactivating (A-type) current, which is enhanced in dendrites, and a sustained or slow/non-inactivating current expressed at a constant somatodendritic density (Hoffman et al., 1997). Previously we have shown that proteins encoded from the shal family of K+ channels (most likely Kv4.2) underlie the A-current (IA) in CA1 hippocampal pyramidal neurons (Kim et al., 2005) and dendritic A-type currents were abolished in Kv4.2 knockout mice (Chen et al., 2006). A-type currents in neurons expressing Kv4.2 have important roles in dendritic signal processing, including the regulation of AP propagation, synaptic integration, filtering of fast synaptic potentials (Cash and Yuste, 1998; Goldberg et al., 2003; Hoffman et al., 1997; Kim et al., 2005; Ramakers and Storm, 2002; Schoppa and Westbrook, 1999), and in LTP (reviewed by Kim and Hoffman, 2008).
Recent studies have shown that genetic down- or up-regulation of Kv4.2 alters dendritic Ca2+ influx during back-propagation (Chen et al., 2006; Kim et al., 2005). As intracellular Ca2+ elevation is necessary for LTP, these results indicate that regulation of back-propagation by Kv4.2 may not only be important for the modulation of dendritic depolarization but also for subsequent cellular cascades to influence the induction and/or the maintenance of synaptic plasticity (Bliss and Collingridge, 1993; Malenka and Nicoll, 1999). We have recently shown that Kv4.2 is expressed in spines of adult CA1 hippocampal pyramidal neurons and that altering the functional level of Kv4.2 influences miniature excitatory postsynaptic currents (mEPSCs) in cultured hippocampal neurons (Kim et al., 2007). Consequently, subthreshold activity of Kv4.2 may also regulate synaptic excitability and Ca2+ influx. Although Chen and colleagues (Chen et al., 2006) show adult Kv4.2 knockout mice have a lowered threshold for LTP induction, whether and how acute changes in Kv4.2 activity affect LTP induction are still unknown.
The importance of NMDA-type glutamate receptors (NMDARs) in synaptic plasticity and memory are well described (reviewed by Malenka and Bear, 2004; Malenka and Nicoll, 1999). At excitatory synapses, Ca2+ influx occurs through both NMDARs and voltage-dependent Ca2+ channels (Frick et al., 2004; Lisman and Spruston, 2005; Magee and Johnston, 1997). This Ca2+ elevation is detected by Ca2+/Calmodulin-dependent kinase II (CaMKII), which is activated by Ca2+/Calmodulin binding and initiates the biochemical cascade of synaptic potentiation (Lisman et al., 2002). CaMKII activated by intracellular Ca2+ elevation is then rapidly translocated to active synaptic sites and binds to NMDARs (Barria and Malinow, 2005; Bayer et al., 2001; Mayadevi et al., 2002; Otmakhov et al., 2004; Shen and Meyer, 1999; Strack et al., 2000). As NR2B-containing NMDAR currents have slower kinetics, allowing for greater temporal summation and Ca2+ influx (Lisman et al., 2002; Quinlan et al., 2004), Ca2+ dependent synaptic potentiation potentially relies on the interaction between NR2B and CaMKII at postsynaptic domains. In fact, ifenprodil, a use-dependent NR2B-selective blocker (Williams, 1993) prevents LTP induction in young CA1 neurons of organotypic slice cultures (Barria and Malinow, 2005).
Existing data suggests a general neuronal mechanism whereby synaptic NR2A-containing receptors become more prevalent with development, learning or experience (Monyer et al., 1994; Quinlan et al., 1999) and recently it has been shown that, in young rats, LTP induces an immediate switch in the composition of synaptic NR2 subunits in favor of NR2A over NR2B (Bellone and Nicoll, 2007). However, synaptic NR2B-containing receptors are still present in adult hippocampal synapses (Al-Hallaq et al., 2007; Kohr et al., 2003; Thomas et al., 2006), indicating a mechanism exists for NR2B subunits to traffic into synapses.
Here, we show that IA is an important regulator of synaptic signaling. Within one day of altering IA activity, CA1 neurons in organotypic slice cultures responded with changes in synaptic NMDAR-mediated current levels. Increasing IA decreased the NR2B contribution to the total synaptic NMDAR current, while IA knockdown increased the relative NR2B fraction. This remodeling was specific to synaptic NMDARs and was blocked by APV but not TTX pretreatment. Thus synaptic NR2 subunit composition appears to be regulated by spontaneous synaptic transmission acting on NMDARs. These results were supported by data collected from Kv4.2 knockout mice, which exhibit enhanced NR1 and NR2B levels in synaptic membranes, compared to wild type mice. IA-dependent synaptic receptor remodeling was accompanied by changes in cellular levels of active CaMKII and in the ability to induce LTP. These findings are consistent with the proposal that learning induced changes in synaptic NMDAR composition limits subsequent synaptic plasticity to maintain memories encoded by experience (Quinlan et al., 2004). Together with our previous results showing the activity-dependent trafficking of Kv4.2 out of spines, these results suggest that neurons may traffic A-type K+ channels to regulate spine-compartmentalized Ca2+ signaling through NMDARs during the learning process.
Previously we reported expression of the A-type voltage-gated K+ channel subunit Kv4.2 in spines of hippocampal neurons (Kim et al., 2007). Kv4.2 localization in spines suggests a role for A-type currents beyond their established role in regulating action potential back-propagation (Chen et al., 2006; Hoffman et al., 1997; Kim et al., 2005). To determine if IA influences synaptic glutamate receptor function, we recorded EPSCs from pairs of neighboring uninfected and Kv4.2-targeted neurons in hippocampal organotypic slice cultures (Hayashi et al., 2000). AMPAR and NMDAR EPSCs were compared in three experimental groups: 1) uninfected or eGFP-infected CA1 pyramidal neurons, 2) those expressing eGFP-labeled Kv4.2 (Kv4.2g) and 3) neurons expressing a dominant negative pore mutant of Kv4.2 (Kv4.2gW362F). In whole-cell voltage-clamp recordings, no change in average resting potential upon break-in was observed between the three groups (Kv4.2g, -60.7 ± 1.4 mV; Kv4.2gW362F, -61.4 ± 1.4 mV; uninfected, -62.9 ± 1.3 mV; p > 0.05). Figure 1 shows that functional IA expression level negatively influences synaptic NMDAR-mediated current amplitude without affecting AMPAR currents. In whole-cell recordings from uninfected neurons, stimulation intensity was adjusted to induce ~ -100 pA EPSCs at -60 mV. We then measured either AMPAR or NMDAR EPSCs in the uninfected neuron before patching the neighboring infected neuron and recording EPSCs with the same stimulation electrode placement and intensity (Figure 1A-C). NMDAR EPSCs in these experiments were recorded at a +40 mV holding potential after washing-in DNQX (10 μM). Comparison of AMPAR EPSC amplitudes between uninfected and infected neurons did not show any differences among the three experimental groups (Figure 1B, uninfected: infected, in pA; uninfected (n = 2) or eGFP (n = 3), 94.73 ± 2.47 : 100.4 ± 4.86; Kv4.2gW362F (n = 8), 99.18 ± 2.66 : 99.06 ± 7.32; Kv4.2g (n = 10), 95.76 ± 3.46 : 95.15 ± 3.83). However, NMDAR EPSCs were significantly increased in Kv4.2gW362F neurons (57.48 ± 2.09: 86.29 ± 3.96, n = 8, p < 0.01 compared with uninfected or eGFP), while Kv4.2g neurons showed decreased amplitudes compared to the control group (Figure 1C; 51.45 ± 2.83 : 29.46 ± 2.88, n = 7, p < 0.01). For further investigation, we recorded isolated AMPAR and NMDAR EPSCs and prepared synaptic intensity-response curves in Kv4.2-targeted neurons (Supplemental Figure 1). These experiments again established specific changes to NMDAR EPSCs, showing enhanced NMDAR currents in cells expressing the Kv4.2 dominant negative but reduced NMDAR currents in Kv4.2g expressing neurons.
To further characterize the observed changes in NMDARs, in another set of experiments we measured the decay rate and pharmacological profile of isolated NMDAR currents (Supplemental Figure 2). In these experiments, stimulation intensity for evoking synaptic currents was adjusted to half-minimum intensity necessary to initiate an AP at -65 mV (100 – 600 μA). After recording EPSCs at -65 mV, DNQX (10 μM) was bath applied to block AMPARs, and NMDAR EPSCs were recorded at the same intensity of stimulation. NMDAR EPSC amplitudes (measured after application of 10 μM DNQX) were again enhanced in Kv4.2gW362F expressing neurons (76.2 ± 6.8 pA, p < 0.05, n = 7) and reduced in Kv4.2g expressing neurons (32.0 ± 3.9 pA, p < 0.05, n = 8) compared to uninfected neurons (48.4 ± 3.7 pA, n = 6). However, we found no significant difference in the average amplitude of AMPAR-mediated currents between groups (Kv4.2g, 114.0 ± 10.8 pA; Kv4.2gW362F, 124.2 ± 8.9 pA; uninfected, 118.8 ± 15.5 pA; p > 0.05). We note that these experiments may underestimate the Kv4.2-mediated changes in NMDAR currents given Kv4.2's role in regulating neuronal excitability (Kim et al., 2005).
In these experiments, the decay rate of isolated NMDAR-mediated EPSCs in Kv4.2gW362F recordings was noticeably slower than that measured in uninfected neurons, while faster kinetics were found for Kv4.2g expressing neurons (“DNQX” in Figure 1D and Table 1). Because the NR2 subunits determine NMDAR decay kinetics (Kohr, 2006), we pharmacologically isolated NR2A and NR2B subunit components from total NMDAR-mediated EPSCs by bath applying the NR2B antagonist ifenprodil (3 μM), subsequent to DNQX, to gauge the relative contribution of the two subunits to the total synaptic NMDAR current (“DNQX+ifenprodil” in Figure 1D and Table 1). This concentration of ifenprodil shows a selective block on NR1/NR2B receptors upon heterologous expression (Williams, 1993). Here, we found application of ifenprodil completely abolished the difference in decay rate among the three groups. To confirm that the residual currents after DNQX were mediated by NMDARs, we bath applied APV (50 μM) after DNQX in some experiments (Supplemental Figure 3). Synaptic, ifenprodil-sensitive NMDAR currents are thus negatively affected by functional Kv4.2 expression level, without appreciable change in synaptic ifenprodil-insensitive currents.
To estimate the relative amplitude of each NR2 subunit, total current charge (integral summation of an EPSC, pA * ms) was calculated for each current and normalized to total AMPAR-mediated current charge (“control” – “DNQX”) recorded at +40 mV (Figure 1E). The ifenprodil sensitive current was obtained by subtracting the “DNQX + ifenprodil” trace from “DNQX” trace. The validity of this analysis relies on our finding that evoked AMPAR-current amplitudes and decay kinetics were unaffected by functional Kv4.2 expression level (Figure 1B and Supplemental Figures 1 and 2, and Table 1). The results presented in Figure 1E suggest that group differences in NMDAR/AMPAR current ratio originate from altered synaptic NR2B expression (“ifen sensitive” ratio, uninfected, 0.57 ± 0.05, n = 7; Kv4.2gW362F, 0.93 ± 0.06, n = 7, p < 0.05; Kv4.2g, 0.25 ± 0.05, n = 9, p < 0.01), with no change in NR2A/AMPAR current ratio found between three groups (“ifen insensitive” ratio, uninfected, 0.79 ± 0.08; Kv4.2gW362F, 0.77 ± 0.08; Kv4.2g, 0.76 ± 0.12). Acute block of predominately A-type K+ currents with intracellular 4-aminopyridine (4-AP) application also led to synaptic remodeling of evoked NMDAR EPSCs (Supplemental Figure 4).
These results suggest that IA does not cause a “switch” or exchange between NR2A- and NR2B-containing NMDARs but, rather, to simply impact synaptic NR1/NR2B functional expression. These striking, bidirectional changes in NR2 subunit composition could be through altered gene expression and/or lateral or intracellular trafficking of NR2 subunits. However, that a single point mutation (Kv4.2gW362F) enhanced synaptic NR2B fraction while Kv4.2g expression has the opposite effect, suggests that Kv4.2 regulation of NR2 subunits is activity-dependent rather than through direct protein-protein interactions with Kv4.2 or intermediates.
Previously we have shown that functional Kv4.2 expression level affects a number of measures of neuronal excitability including AP threshold, onset, half-width and firing rate. The pronounced effect of functional Kv4.2 expression level on AP repolarization and back-propagation lead to clear differences in dendritic Ca2+ influx during single and short trains of APs (Kim et al., 2005). These results, together with those presented in Figure 1, suggest the possibility that altering the amplitude of IA induces an activity- and/or Ca2+-dependent response to remodel synaptic NMDARs in CA1 neurons. Block of AP firing might then mimic Kv4.2 upregulation, reducing excitability and associated Ca2+ influx, leading to a reduced synaptic NR2B subunit fraction. To test for this, organotypic slice cultures were incubated for 20-24 hrs in TTX (0.5 μM) immediately after viral infection with either Kv4.2g or Kv4.2gW362F. Again, for electrophysiological recordings, we adjusted the stimulus intensity to establish a baseline evoked EPSC amplitude of ~ -100 pA at -60 mV before washing in DNQX to isolate NMDAR-mediated synaptic currents. Unexpectedly, we found that TTX pretreatment did not block either the kinetic or NMDAR/AMPAR current ratio changes induced by either Kv4.2-targeted group compared to uninfected neurons (Figure 2A and D, and Table 1). This finding argues against a role for AP firing (evoking Ca2+ influx through voltage-gated Ca2+ channels) in our experiments showing IA regulation of synaptic NR2 subunit composition.
Another source of Ca2+ influx in CA1 neurons is through synaptic activation of NMDARs. A recent report shows NMDAR signaling during spontaneous synaptic transmission stabilizes synaptic function (Sutton et al., 2006). IA control of spine depolarization during spontaneous activity could then determine NMDAR activation and any subsequent signaling. To test for this we blocked spontaneous NMDAR activation overnight with APV (100 μM, 18-24 hrs) or high MgCl2 (additional 4 mM, 18-24 hrs). After washing out APV or high Mg2+ for 15-20 min in the recording chamber before recording, EPSCs were again recorded at a holding potential of +40 mV. Both APV and high Mg2+ pretreatment had no effect on synaptic NMDAR-mediated EPSC decay or NMDAR/AMPAR current ratio in control recordings from uninfected neurons (Figure 2B-D and Table 1). Remarkably, however, both the enhanced rate of NMDAR-mediated EPSC decay in Kv4.2g expressing neurons and the delayed decay in Kv4.2gW362F expressing neurons, along with IA-mediated changes in NMDAR/AMPAR current ratios, were completely blocked by overnight APV or high Mg2+ pretreatment (Figure 2B-D and Table 1). Spontaneous activation of NMDARs is therefore required for IA-mediated rapid changes in synaptic NR2 subunit composition.
To identify Ca2+ as the intracellular secondary messenger involved in NMDAR signaling during spontaneous synaptic transmission we looked at the effect of altering intracellular Ca2+ levels on NMDAR decay rates in uninfected, control neurons (Figure 3A and C and Table 1). Pretreatment with the fast membrane-permeable Ca2+ chelator BAPTA-AM (100 μM) to the culture media (18-24 hrs) mimicked the Kv4.2g-mediated effects on synaptic NR2 subunit composition, enhancing NMDAR decay rates (Figure 3A and Table 1) and reducing ifenprodil-sensitive currents (Figure 3A and C, “ifen sensitive” ratio, non-treated, 0.53 ± 0.09; BAPTA, 0.24 ± 0.05, p < 0.05 compared with non-treated). Conversely, doubling the external [Ca2+] to 3.6 mM overnight had the opposite effect, mimicking our results with Kv4.2gW362F on decay rates (Figure 3A and Table 1) and enhancing ifenprodil-sensitive currents (Figure 3A and C, “ifen sensitive” ratio, 0.85 ± 0.06, p < 0.05 compared with non-treated). Unlike the results of our experiments manipulating Kv4.2 functional expression level, the experiments designed to alter synaptic Ca2+ influx also resulted in small changes in the measured ifenprodil-insensitive currents (Figure 3A and C, p < 0.05). This result suggests the involvement of another Ca2+ dependent mechanism in the control of synaptic NR2A function (such as NR2A-specific current potentiation by tyrosine kinases (Kohr and Seeburg, 1996; Zheng et al., 1998)).
For further investigation, the effects of APV (100 μM) and TTX (0.5 μM) were tested on the high Ca2+-mediated NR2 subunit modification (Figure 3A and C and Table 1). The delayed decay and enhanced ifenprodil-sensitive current by high Ca2+ pretreatment were completely blocked by APV pretreatment (high Ca2+ + APV, “ifen sensitive” ratio, 0.46 ± 0.07, p < 0.01 compared with “high Ca2+”) but not by TTX (high Ca2+ + TTX, 0.95 ± 0.07, p > 0.05). APV also abolished Ca2+-mediated NR2 subunit modification in the presence of TTX (n = 5, NMDA/AMPA ratio, total, 1.03 +/- 0.12; ifen sensitive, 0.45 +/- 0.07; ifen insensitive, 0.57 +/- 0.06; p < 0.05 compared with “high Ca2+ + TTX”, data not shown). Together, these data demonstrate that synaptic NR2 subunit remodeling depends specifically on spontaneous activation and Ca2+ entry through NMDARs.
The Ca2+ and calmodulin protein kinase II (CaMKII) is an initial target of NMDAR-mediated Ca2+-entry, perhaps triggering synaptic plasticity through its interaction with NR2B subunits (Barria and Malinow, 2005; Lisman et al., 2002). To further characterize the mechanism of spontaneous activity-dependent synaptic NR2 subunit remodeling, we considered the possibility that CaMKII is the Ca2+ sensor given its persistent activation by spontaneous activity following action potential block by TTX (Murphy et al., 1994). First, we pretreated slices with the CaMKII inhibitor, KN93 (20 μM, 18-24 hrs; Figure 3B,C and Table 1). The decay rates of NMDAR EPSCs were significantly accelerated in both uninfected and Kv4.2gW362F neurons by KN93, through a reduction of synaptic NR2B fraction (Figure 3C, “ifen sensitive” ratio, uninfected, 0.21 ± 0.03, p < 0.05; Kv4.2gW362F, 0.20 ± 0.0 3, p < 0.05; compared to the non-treated, uninfected group in Figure 1E), mimicking the effect of Kv4.2 overexpression. Next, KN93 was co-applied overnight with high Ca2+. In these experiments (Figure 3A and C, Table 1), KN93 completely prevented high Ca2+-induced increases in decay rates and ifenprodil-sensitive currents (high Ca2+ + KN93, “ifen sensitive” ratio, 0.30 ± 0.04, p < 0.01 compared with Ca2+-pretreated group).
To see if CaMKII activity is sufficient to drive synaptic NR2 subunit remodeling, we expressed constitutively active CaMKII (tCaMKII) in hippocampal organotypic slice cultures (Figure 3D,E). After infecting CA1 pyramidal neurons with EGFP-tagged tCaMKII (36-48 hrs), total and pharmacologically isolated NMDAR EPSCs were recorded at a +40 mV holding potential. tCaMKII-infected neurons showed an enhanced ifenprodil-sensitive fraction, mimicking Kv4.2gW362F expression (uninfected, 39.79 ± 0.05; Kv4.2gW362F, 53.06 ± 3.87, p < 0.05; tCaMKII, 59.23 ± 2.49 %, p < 0.05). Taken together, these results, along with those presented in Figure 2 demonstrate that IA-mediated remodeling of NMDAR subunit composition is dependent on spontaneous NMDAR activation and CaMKII signaling. The NMDAR-dependence and TTX-independence of these changes distinguish our findings from other commonly observed activity-dependent synaptic scaling mechanisms of homeostatic synaptic plasticity (Turrigiano and Nelson, 2004).
The significant reduction of ifenprodil sensitive NMDA current by KN93 shown in Figure 3, along with our IA results shown in Figures 1 and and2,2, predict that IA would control active CaMKII levels in CA1 neurons. To test this prediction, we investigated resting active CaMKII levels in primary hippocampal neurons (DIV 14) using immunolabeling (phospho-T286 antibody recognizing the active form of CaMKII) (Figure 4). Representative linear plot analysis and a comparison of anti-phosphorylated (T286) CaMKII labeled fluorescent intensity showed that active CaMKII levels in Kv4.2g expressing neurons were less than that in uninfected neighbor neurons (0.71 ± 0.03 of uninfected, n = 30, p < 0.05 in soma; 0.74 ± 0.03, n=33 in dendrites, p < 0.05), with no change in total CaMKII level (1.00 ± 0.03 of uninfected, n = 44, p > 0.05, Supplemental Figure 5). In contrast to Kv4.2g, active CaMKII levels in Kv4.2gW362F expressing neurons were increased compared to that in uninfected neighbor neurons (1.36 ± 0.12 of uninfected, n = 26, p < 0.05 in soma; 1.38 ± 0.07, n=21, p < 0.05 in dendrites). Again these effects occurred without a change in total CaMKII level (1.09 ± 0.04 of uninfected, n = 37, p > 0.05, Supplemental Figure 5). Acute (4 h) application of 4-AP (5 mM) also resulted in an increase of active CaMKII staining in dissociated hippocampal neurons (Supplemental Figure 4G).
Consistent with our results showing spontaneous activation of NMDARs is required for synaptic NMDAR subunit remodeling, the differences in active CaMKII between Kv4.2g- or Kv4.2gW362F-infected and uninfected neurons was abolished with APV but not TTX pretreatment (APV-pretreated Kv4.2g: 1.03 ± 0.03 of uninfected, n = 38, p > 0.05 in soma, 1.04 ± 0.06, n=27, p > 0.05 in dendrites; TTX-pretreated Kv4.2g: 0.78 ± 0.05 of uninfected, n = 37, p < 0.05 in soma, 0.74 ± 0.04, n=23, p < 0.05 in dendrites; APV-pretreated Kv4.2gW362F: 1.13 ± 0.08 of uninfected, n = 35, p > 0.05 in soma, 1.00 ± 0.04, n=33, p > 0.05 in dendrites; TTX-pretreated Kv4.2gW362F: 1.23 ± 0.11 of uninfected, n = 28, p < 0.05 in soma, 1.18 ± 0.05, n=22, p < 0.05 in dendrites; Figure 4). For total CaMKII levels, APV pretreatment resulted in no significant change in Kv4.2g- or Kv4.2gW362F-infected neurons (APV-pretreated Kv4.2g: 1.01 ± 0.02 of uninfected, n = 32; APV-pretreated Kv4.2gW362F: 1.04 ± 0.03 of uninfected, n = 26; Supplemental Figure 5).
Since spines act as chemical compartments (Segal, 1995), the dependence of synaptic NR2 subunit remodeling on NMDAR-mediated Ca2+ influx could indicate that extrasynaptic NMDAR subunit composition is unchanged upon altering IA. We determined this by isolating extrasynaptic NMDARs in control and Kv4.2 mutant-expressing primary hippocampal neurons (Figure 5). In these experiments, NMDA currents were evoked by picospritzer application of NMDA (100 μM, 2s duration, 20s interval, 10-15 PSI). Glycine (100 μM), a co-agonist of NMDARs, was added to the external recording solution. Extrasynaptic NMDAR currents were isolated from total currents by applying the rapid, irreversible open NMDAR channel blocker, MK801 (60 μM) and high KCl (90 mM) to block active synaptic NMDARs (Thomas et al., 2006) (Figure 5).
Total (synaptic + extrasynaptic) NMDA-evoked current was significantly reduced in Kv4.2g expressing neurons (n = 25) compared to uninfected neurons (n = 20), while a slight increase of amplitude was observed in Kv4.2gW362F (n = 14) (Figure 5B,C; “control”, “before MK801”: uninfected, 0.92 ± 0.06; Kv4.2gW362F, 1.00 ± 0.06; Kv4.2g, 0.51 ± 0.03 nA, p < 0.05 compared with uninfected). As was found for synaptic stimulation (Figure 1 and Supplemental Figures 1 and 2), these differences in NMDA-evoked total currents were eliminated after ifenprodil (3 μM) wash-in (Figure 5B-D; “ifenprodil”, “before MK801”: uninfected, 0.38 ± 0.03; Kv4.2gW362F, 0.34 ± 0.03; Kv4.2g, 0.29 ± 0.02 nA; p > 0.05). However, subsequent to selective MK801 block of active synaptic NMDAR-mediated currents, no difference in both total currents and the ifenprodil-sensitive fraction was found among three groups (Figure 5B-D; “control”, “after MK801”: uninfected, 0.42 ± 0.09; Kv4.2gW362F, 0.42 ± 0.04; Kv4.2g, 0.38 ± 0.03 nA, p > 0.05; “ifenprodil”, “after MK801”: uninfected, 0.18 ± 0.04; Kv4.2gW362F, 0.22 ± 0.03; Kv4.2g, 0.19 ± 0.02 nA; p > 0.05). In some cells we applied APV (50 μM) instead of MK801 to determine if high KCl stimulation itself affects NMDA-evoked currents. Total NMDA-evoked current (1.15 ± 0.09 nA, n = 6) was not changed after high KCl stimulation (1.12 ± 0.08 nA, p > 0.05, not shown), indicating that KCl application itself did not affect the NMDA-induced current (Thomas et al., 2006). These results show that IA-induced remodeling of NR2 subunit composition is specific to synaptic NMDARs.
If the level of IA alters NMDAR synaptic subunit composition, it may also affect the ability to induce long-lasting, NMDAR-dependent changes in synaptic strength (Malenka and Bear, 2004). Notably, previous results from organotypic slice cultures have implicated NR2B and CaMKII as required for LTP induction in young CA1 neurons in organotypic slice cultures (Barria and Malinow, 2005). We therefore measured LTP in CA1 neurons of organotypic slices expressing Kv4.2g or it's dominant negative (Kv4.2gW362F). In control, uninfected neurons (n = 9), a robust potentiation is observed with a depolarization-pairing protocol (Figure 6A-C). Importantly, in this method of LTP induction the neuron is voltage-clamped to 0 mV during pairing and so A-type K+ channels will be inactivated and LTP induction should not be directly affected by Kv4.2 expression level. However, distinct patterns of potentiation were observed in Kv4.2g (n = 11) and Kv4.2gW362F (n = 10) expressing neurons after delivering the pairing protocol in organotypic hippocampal slices. The initial increase of EPSC amplitude in Kv4.2g expressing neurons diminished to baseline amplitude by ~20 min post-pairing (Figure 6A-C). Unlike control neurons where peak potentiation occurs within 5 min of pairing, EPSC amplitude continued to rise for the entire recording after pairing in Kv4.2gW362F expressing neurons (Figure 6A-C, uninfected, 66.26 ± 14.48; Kv4.2gW362F, 112.36 ± 18.78, p < 0.05; Kv4.2g, -3.03 ± 12.93%, p < 0.01 compared with uninfected, potentiation at 40-50 min post-induction).
Our finding that LTP is compromised in NR2B-deficient (Kv4.2g expressing) cells is consistent with previous findings in CA1 neurons of acute hippocampal slices from young animals (Ito et al., 1996; Kohr et al., 2003) and in hippocampal organotypic slice cultures (Barria and Malinow, 2005). However, the dependence on NR2B for LTP decreases with age as NR2B is developmentally replaced with NR2A (Ito et al., 1996; Liu et al., 2004). We wondered then, if IA influences synaptic NMDA signaling in adult animals. First, we checked for Kv4.2-dependent synaptic remodeling in immunoblot analyses of subcellular synaptic fractions from wild type and Kv4.2 knockout (Kv4.2-/-) mice (Figure 7). In Kv4.2-/- mice, total and synaptic NR2B levels were increased (1.23 ± 0.07 of wild type in synaptic membranes, 1.22 ± 0.08 of wild type in total, n=6), along with NR1 (1.32 ± 0.06 of wild type in synaptic membranes, 1.58 ± 0.13 of wild type in total, n=6). No significant change in NR2A was observed in the Kv4.2-/- mice (0.95 ± 0.07 of wild type in synaptic membranes, 1.00 ± 0.03 of wild type in total, n=6). In addition, an increase in PSD-95 (an NMDAR anchoring protein) was detected in the Kv4.2-/- mice (1.25 ± 0.05 of wild type in synaptic membranes, 1.19 ± 0.05 of wild type in total, n=6), while no change was found for total CaMKII level (1.08 ± 0.06 of wild type in synaptic membranes, 1.10 ± 0.05 of wild type in total, n=6). It is also important to note synaptic expression Kv4.2 in wild type mice (Figure 7), consistent with our previous imaging data (Kim et al. 2007).
Synaptic NR2B subunit enrichment in Kv4.2-/- mice was confirmed in electrophysiological recordings. In acute hippocampal slice recordings from WT and Kv4.2-/- mice (P20-23), AMPAR and NMDAR EPSCs were recorded in CA1 neurons at -60 mV and +40 mV (with the addition of DNQX), respectively (Figure 8 and Supplemental Figure 6). Although we did not observe any difference in the NMDA/AMPA current amplitude ratio between WT (Supplemental Figure 6, 0.92 ± 0.08, n = 6) and Kv4.2-/- (0.89 ± 0.07, n = 7) recordings, the decay rate of NMDAR EPSCs was slower in Kv4.2-/- neurons (“DNQX” WT, 47.53 ± 2.0, n = 6; Kv4.2-/-, 62.00 ± 2.59 ms, n = 8, p < 0.01) (Figure 8A,B). Bath application of ifenprodil indicated that synaptic NMDARs of Kv4.2-/- neurons are predominately composed with NR2B subunits, while NR2A is the dominant subunit in WT neurons at this age (Figure 8C, “ifen sensitive” ratio to AMPA; WT, 0.27 ± 0.05; Kv4.2-/-, 1.21 ± 0.18, p < 0.01), consistent with ifenprodil sensitivity of NMDAR EPSCs evoked by 0.1 Hz stimulation in Mg2+-free external solution containing DNQX at -60 mV (Figure 8D, 10 min after ifenprodil application, WT, 21.07 ± 5.52, n = 5; Kv4.2-/-, 60.57 ± 3.12 %, n = 7, p < 0.01).
Despite these changes in synaptic NR2 subunits, no difference in LTP magnitude between WT (n = 6) and Kv4.2-/- (n = 7) neurons was observed when induced using a paring protocol similar to that used in organotypic slice cultures (4 Hz stimulation paired with 0 mV holding potential for 1 min) (Figure 8E). That LTP induction and expression is normal in Kv4.2-/- mice is consistent with a previous paper demonstrating that strong induction protocols overcome of the LTP induction deficient found in between WT and Kv4.2-/- (Chen et al., 2006). Next, we tested the effect of blocking NR2B subunits on LTP induction (Figure 8F). The maximal effect of ifenprodil occurred after ~ 10 min wash-in (Figure 8D). Therefore, we applied ifenprodil to the recording chamber at least 20 min before attempting to induce LTP. Application of ifenprodil completely blocked the induction of synaptic LTP in Kv4.2-/- neurons, while only slightly reducing potentiation of EPSCs in wild type neurons (40 min, wild type, 65.46 ± 10.49%, n = 5; Kv4.2-/-, -1.56 ± 7.51%, n = 8, p < 0.01).
Our results showing that induction and expression of LTP was normal in control (no drug) conditions in Kv4.2-/- mice but completely blocked by ifenprodil, indicate that while a minimal level of synaptic NMDAR current and associated Ca2+ influx is necessary for LTP induction, the relative composition of NR2 subunits is less important in acute slice recordings. These results support the conclusion from a previous study that hippocampal LTP induction can be generated by either NMDAR subtype (Berberich et al., 2005).
Here we show that the activity level of subthreshold, A-type voltage-gated K+ channels can bidirectionally reorganize synaptic NMDAR subunit composition: expression of a dominant negative mutant of the A-type K+ channel subunit Kv4.2 resulted in greater synaptic NR2B/NR2A fraction while increased Kv4.2 expression lead to a smaller NR2B/NR2A fraction in CA1 pyramidal neurons from hippocampal organotypic slice cultures. These changes are specific to synaptic NR2 subunits and require spontaneous NMDAR and CaMKII activation. They are mimicked in experiments designed to alter postsynaptic intracellular Ca2+ concentration but are not blocked by overnight TTX, indicating that the Ca2+-dependent changes do not require neuronal firing. These effects appear to be strictly due to an IA influence on synaptic NR2B-containing NMDAR expression rather than promotion of NR2 subunit switching. Consistent with these findings, we found a distinct increase in NR1 and NR2B (but not NR2A) proteins along with the NMDAR scaffolding protein PSD-95 in immunoblots of hippocampal synaptic fractions from Kv4.2 knockout mice.
In addition, we show that augmenting IA reduces cellular active CaMKII levels and blocks LTP whereas knocking down IA enhances active CaMKII and results in greater potentiation. These effects of IA activity too were overcome by blocking spontaneous NMDAR signaling. Together, our results show that A-type K+ channels are an integral part of a synaptic complex regulating Ca2+ signaling through spontaneous NMDAR activation to control synaptic NMDAR expression and plasticity.
Our primary finding is a rapid change in synaptic NR2 subunit composition in hippocampal CA1 pyramidal neurons of organotypic slice cultures upon altering the strength of A-type K+ currents that depends on spontaneous activation of NMDARs and CaMKII activation. These striking changes are activity-dependent rather than through direct protein-protein interactions with Kv4.2 or intermediates since a single amino acid pore mutation (Kv4.2gW362F) enhanced synaptic NR2B fraction while Kv4.2g expression has the opposite effect.
A full understanding of how IA is involved in synaptic signaling will require further study but we propose that subthreshold A-type K+ channels in spines act to gate NMDAR activation by opposing Mg2+-unblock. Recent evidence suggests that subthreshold synaptic potentials can flux Ca2+ through NMDARs at resting potential in the hippocampus (Kovalchuk et al., 2000; Losonczy and Magee, 2006). By countering depolarization and providing repolarization, Kv4.2 channels can shape Ca2+ influx in spines and therefore their trafficking or modulation may be a normal target for the regulation of synaptic Ca2+-dependent signaling (Cai et al., 2004; Hoffman and Johnston, 1998, 1999; Kim et al., 2007; Losonczy et al., 2008; Merrill et al., 2005).
Our results show that spontaneous NMDAR activation but not AP firing is necessary to induce synaptic remodeling after altering Kv4.2 activity, suggesting the requirement for spontaneous NMDAR activation. This finding follows other recent studies showing that signaling through NMDARs during spontaneous synaptic transmission: 1) stabilizes synaptic function by tonically suppressing dendritic protein synthesis and subsequent synaptic AMPAR trafficking (Sutton et al., 2006; Sutton et al., 2004) and 2) prevents the synaptic incorporation of recombinant NR2A subunits (Barria and Malinow, 2002). As with our results, AP firing was not required for the repression of dendritic protein synthesis or synaptic recombinant NR2A incorporation.
It is known that NR2A- and NR2B-containing receptors have different internalization motifs with NR2B subunits showing higher turnover rates (Lavezzari et al., 2004; Perez-Otano and Ehlers, 2005). Our data suggest that spontaneous activation of NMDARs and CaMKII is important for the trafficking and thus ratio of synaptic NR2 subunits. Furthermore, Figure 3 shows that raising intracellular Ca2+ enhances synaptic NR2B/NR2A fraction while Ca2+ chelation has the opposite effect. Therefore, less excitable spines, expressing enhanced IA, would be expected to have diminished NMDAR activation and associated Ca2+ influx leading to decreased synaptic NR2B/NR2A fractions. IA knockdown, increasing spine excitability, leads to enhanced NMDAR-dependent Ca2+ signaling and an increased synaptic NR2B/NR2A ratio (Figure 2).
In this scenario, NMDAR block, abolishing Ca2+ signaling, should result in reduced NR2B/NR2A levels. However, no change in synaptic NR2 subunit ratios is found in control, uninfected neurons after overnight APV. To account for this finding, we suggest that ligand binding of NMDARs by APV halts or dramatically slows NMDAR trafficking. Support for this suggestion comes from Kv4.2-targeted experiments where APV applied right after viral infection of our organotypic slices prevented any IA-dependent changes in NR2 subunit composition (NR2 levels are nearly identical to control, Table 1). However, if APV is applied the next day, four hrs before recording, no affect of APV is found (NMDA decay in DNQX at +40 mV: Kv4.2g, 37.73 ± 1.76, n = 7; control 44.31 ± 1.89 ms, n = 6; p < 0.05, data not shown).
Existing data suggests a general neuronal mechanism whereby synaptic NR2A-containing receptors become more prevalent with development, learning or experience (Monyer et al., 1994; Quinlan et al., 1999), although synaptic NR2B-containing receptors are still present in the adult hippocampus (Thomas et al., 2006). Whether and how our results, showing a rapid change in NR2 subunit composition is related to these developmental/activity-related changes will require a separate investigation. However, electron microscopy data shows no change in the number of synaptic NMDARs with development, suggesting that the NR2 subunit differences are enabled through other mechanisms (Petralia et al., 1999). Our finding that NR2 levels are readily, rapidly and bidirectionally modifiable suggests that synapses over the long term must have the ability to maintain a target ratio for synaptic NR2A-NR2B subunits during development and in the face of different activity levels.
Dendritic IA activity, by regulating neuronal excitability and associated Ca2+ influx, has the potential to impact LTP on many levels (Chen et al., 2006; Kim et al., 2007; Kim et al., 2005; Watanabe et al., 2002). Previous work from the Johnston lab has shown Kv4.2 to be important for the induction of LTP induced by theta-burst pairing by regulating action potential backpropagation (Chen et al., 2006). The protocol used here to induce LTP, however, is unlikely to be impacted by Kv4.2 expression as it does not rely on action potential back-propagation and Kv4.2 channels would be expected to be inactivated during the pairing period (with the cell held at 0 mV). Our finding that Kv4.2 expression level affects the degree of LTP induced thus likely depends on secondary consequences of Kv4.2 expression rather than activity of Kv4.2-channels during induction. In addition to altering synaptic NR2 subunits and active CaMKII levels, Kv4.2 may play a direct role in LTP expression, perhaps through activity-dependent internalization of Kv4.2, affecting dendritic integration (Kim et al., 2007).
It is well established that NMDAR activation is required for the induction of most forms of synaptic plasticity (Malenka and Bear, 2004) although there is conflicting evidence for the role of specific NR2 subunits (Berberich et al., 2005; Liu et al., 2004; Massey et al., 2004; Morishita et al., 2007). In the hippocampus, overexpression of NR2B in transgenic mice leads to enhanced LTP and learning in behavioral tasks (Tang et al., 1999) and it is more difficult to induce plasticity at synapses with a lower synaptic NR2B/NR2A fraction (Barria and Malinow, 2002), perhaps due to decreased Ca2+ flux through receptors expressing the NR2A subunit (Kohr, 2006; Sobczyk et al., 2005). Our LTP results, showing increased potentiation in NR2B enriched synapses (Kv4.2gW362F cells) but a complete loss of LTP after ~ 20 min in relatively NR2B-deprived synapses (Kv4.2g cells), support this tendency (Figure 6).
It has recently been shown that, in developing hippocampal neurons, NR2A subunits are swapped-in to replace synaptic NR2B subunits after LTP induction (Bellone and Nicoll, 2007). Together with our findings, these results suggest a model whereby basal synaptic NR2 subunit composition is determined by spontaneous NMDAR activity with Ca2+ influx enhancing synaptic NR2B-containing NMDAR incorporation. However, with experience or over the course of development, LTP “activates” synapses with NR2A replacing NR2B. Subsequently, LTP becomes harder to induce. Our findings also suggest that synaptic NMDAR configuration (and thus the ability to induce plasticity) is readily modifiable with changes in IA. Known hippocampal functions (memory formation and spatial navigation) necessitate such adaptability and so it would be expected that a mechanism exists for the rapid bidirectional remodeling of synaptic NMDARs. Our results would further indicate an important role for spontaneous NMDAR and CaMKII activation in these types of synaptic plasticity.
Organotypic hippocampal slices (400 μm thick) were prepared from postnatal day 7-8 Sprague-Dawley rats. Primary hippocampal pyramidal neurons for NMDA-evoked current recordings and immunostaining were prepared from embryonic day 18 as in (Kim et al., 2007). An attenuated Sindbis virus system was employed to express Kv4.2g and Kv4.2gW362F. All electrophysiological measurements from organotypic slices were made 1-2 days after viral infection of CA1 pyramidal neurons (4-6 DIV). We note here that the Mg2+ level in the culture media (0.67 mM) that our slices were maintained in between infection and recording is less than in our standard recording solution (1mM). More detailed protocols for tissue preparation and viral infections are available in our previous paper (Kim et al., 2005). The constitutively active CaMKII (tCaMKII) was a kind gift from Dr. Esteban (U. Michigan). Other constructs used in present experiments were described previously (Kim et al., 2007; Kim et al., 2005).
Acute hippocampal slices (350 μm thick) were made from P20-23 129/SvEv wild type and Kv4.2-/- mice (Guo et al., 2005) in a cutting solution containing (in mM): 125 NaCl, 2.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 25 Glucose, 0.5 CaCl2, 5 MgCl2 (pH, 7.2). After 30 min incubation in normal ACSF (bubbled with 95 % O2, 5% CO2) at 37 °C, slices were transferred to room temperature ACSF solutions. The National Institute of Child Health and Human Development's Animal Care and Use Committee approved all animal protocols.
For patch-clamp recordings from organotypic or acute hippocampal slices, slices were transferred to a submerged recording chamber with continuous flow of ACSF containing (in mM): 125 NaCl, 2.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 25 Glucose, 2 CaCl2, 1 MgCl2. 5 μM 2-chloroadenosine and 5 μM bicuculine (10 μM for acute slice) were added in some recordings. To record NMDA currents in primary cultured neurons the external solution contained (in mM): 145 NaCl, 5 KCl, 10 Glucose, 2 CaCl2, 10 HEPES, 0.0005 TTX, 0.002 strychnine, 0.01 bicuculine, 0.1 glycine. Patch electrodes (4-6 MΩ) were filled with (in mM): organotypic slices and acute slices; 20 KCl, 125 Kglu, 10 HEPES, 4 NaCl, 0.5 EGTA, 4 ATP, 0.3 TrisGTP and 10 Phosphocreatin; for NMDA currents in primary cultured neurons; 100 Cs-gluconate, 5 MgCl2, 0.6 EGTA, 8 NaCl, 40 HEPES, 2 NaATP, 0.3 TrisGTP. pH and osmolarity were adjusted to 7.2-7.3 and 280-300 mOsm, respectively, in all experiments.
In organotypic slices and acute slices, EPSCs were induced by stimulation on Schaffer-collateral pathway via bipolar electrodes located in 100-200 μm from the recorded cell soma. The test stimulation in all EPSC experiments was set at 0.1 Hz and 0.2 ms duration and its intensity (100-900 μA) was adjusted to induce about 100 pA AMPAR EPSCs at -60 mV holding potentials (-65 mV in some cases). AMPAR EPSC amplitude was measured from an averaged trace of at least 10 sweeps. Amplitude, current charge (pA * ms) and decay of total and NMDA EPSCs recorded at +40 mV were measured in traces averaged from 30 sweeps. After recording total EPSCs at a +40 mV holding potential, DNQX (10 μM) was added to external recording solution. NMDA EPSCs were recorded after confirming the effect of DNQX by observing the complete blockade of AMPA EPSCs recorded at -60 mV (~5 min after application). Putative NR2A-mediated EPSCs were then acquired by adding ifenprodil (3 μM) to external solution including DNQX. NR2A-EPSCs were recorded for 15 min after washing in DNQX + ifenprodil and the last 30 sweeps averaged, as ifenprodil exhibits use-dependent block (Williams, 1993). For recording cell pairs shown in Figure 1, two cell bodies within ~20 μm were selected (one uninfected, one infected). After adjusting stimulation intensity in uninfected neurons as described above, EPSCs were sequentially recorded from uninfected and infected neurons with same stimulation intensity and location (Hayashi et al., 2000). Whole-cell recording parameters were monitored throughout each experiment and recordings where series resistance varied by more then 10% were rejected.
In this study the pairing protocol for LTP induction consisted of low frequency stimulation (2 Hz for organotypic preparation and 4 Hz for acute slices, 0.2 ms duration) paired with depolarization to 0 mV for 1 min. Potentiation of EPSCs was measured at a -65 or -60 mV holding potential, using the same 0.1 Hz testing stimulation protocol for EPSCs mentioned above. All EPSC recordings and LTP experiments in slices were performed at 31-32 °C and NMDA-evoked current in dissociated neurons at 23-25 °C.
All recordings were low-pass filtered at 5 kHz and digitized at 10 kHz by an Instrutech ITC-18 A/D board controlled by software written for Igor Pro (WaveMetrics). An Axopatch-200B amplifier was employed for whole-cell patch recordings in this study. Command pulse generation, data acquisition and analysis were performed using IGOR Pro (Wavemetrics). SPSS (SPSS Inc.) and Excel (Microsoft) software were used for further data and statistical analysis. Statistical tests performed were unpaired t-tests or ANOVAs followed by LCD post-hoc analysis. Significance was set to p < 0.05.
Synaptic fractionation of hippocampi from 27 adult mice each of wild type and Kv4.2-/- was performed as described previously (Blackstone et al., 1992; Carlin et al., 1980). Total hippocampal membranes were obtained from 4~5 animals by homogenizing in ice-cold Buffer A (0.32 M sucrose, 20 mM HEPES (pH 7.4), 5 mM EDTA, and protease inhibitor cocktail tablet (Roche)). The homogenate was cleared by centrifugation at 1,000 g for 10 min at 4 °C. The second centrifugation (10,000 g) was carried out for 10 min at 4 °C. The resulting pellet, crude synaptosomes, was resuspended and sequentially fractionated in sucrose gradients (0.8 M, 1.0 M, and 1.2 M sucrose) by centrifugation at 65,000 g for 2 hr. The postsynaptic membranes were layered between 1.0 M and 1.2 M sucrose. Proteins were separated in Novex Tris-Acetate 3-8 % gradient gels (Invitrogen). Immunoblotting was performed as described (Kim et al., 2005).
Antibodies: anti-NR1 (PharMingen, 1:1,000), anti-NR2A (Upstate, 1:1,000), anti-NR2B (Novus, 1:1,000), anti-Kv4.2 (K57/1, NeuroMab, 1:1,000), anti-PSD 95 (ABR, 1:1,000), anti-α-CaMKII (abcam, 1:300). Quantification was performed using ImageJ v1.40. The values of total protein were normalized to that of total β–actin, as a homogenate control on the same blots, and presented as ratios between wild type and Kv4.2-/-. For synaptic fractions, the ratio of wild type and Kv4.2-/- were directly compared.
Neurons were immediately fixed with 4% paraformaldehyde and 0.1% glutaraldehyde in PBS containing 0.12 M sucrose for 8 min on ice and permeabilized with 0.5% Triton X-100 in PBS for 5 min. After preblocking with PBS containing 5% NGS, 0.05% Triton X-100, and 450 mM NaCl for 1 h at 4°C, neurons were incubated with anti-CaMKII (abcam, 1:300), anti-activeCaMKII (phospho T286, abcam, 1:100) and neuron-specific enolase (NSE; abcam 1:100) in blocking solution overnight at 4 °C and followed by incubation with Alexa 555-conjugated anti-rabbit IgG (Molecular Probes Inc.) for 2 h at RT. Immunolabeled cell images were acquired with Leica TCS RS and Zeiss LSM 510-Meta confocal microscopes. The instrument parameter settings to acquire fluorescent signals were optimized in non-treated cells and further adjusted in the brightest experimental sample in order to avoid image saturation. The same parameter settings were used for infected- or treated-neurons and kept for each experiment. Every experiment was repeated a minimum of five times. Linear Profile Plot and ROI manager of ImageJ v1.40 were used for analysis of total or active CaMKII level, as described previously (Kim et al., 2005). Paired students t-tests were performed to determine significance, set at p < 0.05.
We thank Drs. Georg Koehr, Chris McBain and members of John Lisman's lab for their critical review of earlier versions of this manuscript. We thank Drs. Diana Medrano-Velasquez, Young Ho Suh and Jeff Magee for technical support, Dr. Thomas L. Schwarz for providing Kv4.2-/- mice and Dr. José Esteban for providing the tCaMKII construct. This work was supported by the National Institute of Child Health and Human Development Intramural Research Program.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.