Adrenergic Gate Release for Spike Timing-dependent Synaptic Potentiation

Abstract

Spike timing-dependent synaptic plasticity (STDP) serves as a key cellular correlate of associative learning, which is facilitated by elevated attentional and emotional states involving activation of adrenergic signaling. At cellular levels, adrenergic signaling increases dendrite excitability, but the underlying mechanisms remain elusive. Here we show that activation of β2-adrenoceptors promoted STD long-term synaptic potentiation at mouse hippocampal excitatory synapses by inactivating dendritic Kv1.1-containing potassium channels, which increased dendrite excitability and facilitated dendritic propagation of postsynaptic depolarization, potentially improving co-incidental activation of pre- and postsynaptic terminals. We further demonstrate that adrenergic modulation of Kv1.1 was mediated by the signaling scaffold SAP97, which, through direct protein-protein interactions, escorts β2-signaling to remove Kv1.1 from the dendrite surface. These results reveal a mechanism through which the postsynaptic signaling scaffolds bridge the aroused brain state to promote induction of synaptic plasticity and potentially to enhance spike timing and memory encoding.

eToc Blurb

Liu et al. report a molecular mechanism through which adrenergic signaling increases dendritic excitability to facilitate LTP induction, a finding explaining how attentional or emotional arousal facilitates learning.

Introduction

Spike timing-dependent synaptic plasticity (STDP) is triggered by the temporally ordered activation of pre- and postsynaptic neurons, which renders it a likely cellular mechanism for associative learning (Bi and Poo, 1998; Debanne et al., 1996; Markram et al., 1997). Associative learning is facilitated by aroused attentional or emotional states, involving activation of adrenergic signaling (O’dell et al., 2015). While the expression of STD long-term potentiation (LTP) at excitatory synapses is facilitated by β-adrenergic signaling-mediated priming of synaptic insertion of AMPA-type glutamate receptors (AMPARs; Hu et al., 2007; Makino et al., 2011; Seol et al., 2007), the induction of LTP can also be facilitated by this signaling. Specifically, the β-adrenergic signaling inhibits dendritic potassium conductances and thus facilitates the generation and propagation of dendritic spikes (Hoffman and Johnston, 1999; Watanabe et al., 2002), helping relieve the magnesium blockade of NMDA-type glutamate receptors (NMDARs) for the induction of NMDAR-dependent LTP. However, it remains unclear which potassium channels mediate the adrenergic effects on dendrite excitability, and how these potassium channels are mobilized by adrenergic signaling to potentially release the gates for STDP.

STD-LTP at hippocampal glutamatergic synapses is induced upon presynaptic activation, contingent with short-delayed postsynaptic backpropagating action potential (AP) spiking (Bi and Poo, 1998). Dendritic AP propagation is facilitated by inhibition of potassium conductances and may promote the induction of STD-LTP (Chen et al., 2006; Hoffman et al., 1997; Watanabe et al., 2002). The Kv1 family of potassium channels are richly expressed in somato-dendritic compartments (Guan et al., 2006). They are locally synthesized within the dendrites of hippocampal pyramidal neurons (Raab-Graham et al., 2006), and their expression is dynamically regulated during associative learning (Kourrich et al., 2005). The C-terminus of Kv1 interacts with the PDZ-domain of SAP97, a scaffold protein thought to regulate membrane expression and stabilization of PDZ-ligand-containing receptors and ion channels (Kim et al., 1995; Tiffany et al., 2000).

Our present study demonstrates that, linked by SAP97, activation of β2-adrenergic signaling removed cell surface Kv1.1 to improve dendritic AP propagation and thus facilitated the induction of STDP. These results narrow down Kv1.1 as the neuronal substrate mediating adrenergic gate release of STDP, reveal an unexpected role of SAP97 in linking adrenergic signaling to dendrite excitability, and thus may provide a mechanistic understanding how attentional, emotional and motivational states regulate associative learning.

Results

β2-adrenoceptor activation reduces the latency of action potential firing

The D-current (I_D) is activated at voltages near the AP threshold (Locke and Nerbonne, 1997; Wu and Barish, 1992) and mediated by Kv1 channels (Cudmore et al., 2010; Golding et al., 1999; Storm, 1988). We tested whether Kv1.1 mediates I_D to regulate the latency of the 1^st AP in CA1 pyramidal neurons. Using patch clamp electrophysiology, we recorded in current clamp mode and adjusted the injected current that triggered a single AP (rheobase). The onset of single APs of multiple trials (Δt=15s) was delayed with a mean onset time of 580 ± 81ms (n=7). Additionally, the latency varied with a jitter, calculated as the mean standard deviation of AP latency of 160 ± 16ms (Fig. 1A, H). Such a jittered delay confers a low spike-timing precision. Dendrotoxin-K (DTX-K) is a relatively selective blocker of Kv1.1-containing potassium channels (Wang et al., 1999). Perfusion of DTX-K decreased the latency of the 1^st AP as well as the jitter (p<0.001; Fig. 1A, G, H). These results indicate that Kv1.1 is critical for I_D in CA1 pyramidal neurons, activation of which dampens the spike-timing precision.

Fig. 1.

Norepinephrine modulation of Kv1.1-mediated latency of APs. A–C, E, F. Repeated rheobase current injections elicited single APs in CA1 pyramidal neurons (second AP marked with “#”). Sample recordings before (grey, left traces) and after different treatments (color, right traces). 100nM dendrotoxin-K (+ DTX-K) (A) to block Kv1.1 or 20μM NE (+ NE) (B, C) to activate adrenoceptors was applied for 10 min as treatment. E, F. 473nm laser light pulses were applied onto the stratum radiatum to trigger endogenous NE release in DBH-Cre x iChR2 mice (E) or no release in iChR2 (F) control mice. 1μM ICI118551 blocked NE modulation (C). D. Coronal slices of the brain stem (brain scheme with position of slices, upper panel) were cut from DBH-Cre x iChR2 (left) and iChR2 (right) mice and ChR2Y expression visualized under fluorescent light. G. Summary graph shows normalized AP latency modulation after different triggers as in panels A–C and E, F normalized to latency in control. H. AP latency jitter of control is illustrated in grey and after different triggers in color as in panel A–C and E, F. Scale bar 20mV and 250ms.

Since norepinephrine (NE) can increase CA1 pyramidal neuron excitability (Hoffman and Johnston, 1999), we tested whether this effect is at least partially mediated via Kv1.1 inhibition. Similar to DTX-K, perfusion of NE reduced the latency and jitter of the 1^st AP (p<0.001; Fig. 1B, G, H). Perfusion of a β2-selective adrenoceptor antagonist (ICI118551) blocked the effect of NE (p=0.32; Fig. 1C, G, H), indicating that β2-adrenoceptors are responsible for I_D inhibition. Unexpectedly, NE also decreased the input resistance to 89.2 ± 1.8% (p<0.001), so that the rheobase current was slightly increased in some recordings to trigger single APs. However, the effect on the input resistance was unrelated to I_D for two reasons. First, DTX-K perfusion and thus Kv1.1 inhibition, did not alter the input resistance (p=0.47). Second, ICI118551 did not block the NE-mediated decrease in input resistance (p<0.001), while it blocked NE-mediated I_D inhibition. Thus, via β2-adrenoceptors, NE increases CA1 pyramidal neuron excitability, likely by reducing Kv1.1 activity.

Using organotypic rat slice cultures, a previous study shows that inhibiting I_D in CA3 pyramidal neurons increases the slope of the membrane potential before AP onset (Cudmore et al., 2010). However, when examined in CA1 pyramidal neurons from acute mouse slices, NE did not alter the slope significantly (n=7; p=0.56; Fig. S1B, C), indicating that the expression of Kv1 subunits or adrenergic regulation is different in these two types of neurons. Despite this difference, both results indicate that I_D delays the AP onset and maintains a membrane potential plateau below the AP threshold. Furthermore, the AP amplitude was not altered after perfusion of NE (ctr. 113 ± 1.8mV vs. NE: 110 ± 1.9mV; n=28; p=0.52; Fig. S1F, H), indicating that I_D is already inactivated when the AP is triggered.

To test whether endogenous NE modulates the AP latency, we crossed an inducible (i) Venus-tagged channelrhodopsin 2 H134R (ChR2Y) reporter mouse line with a noradrenergic neuron-selective Cre driver mouse line (DBH-Cre) (Madisen et al., 2012; Parlato et al., 2007). ChR2Y expression, detected by yellow fluorescence, was selectively induced in neurons of the locus coeruleus in double-transgenic mice, but not in the iChR2Y mono-transgenic mice (controls; Fig. 1D). To evoke endogenous NE release, we applied a train of light pulses through the microscope objective (60x, NA0.9), which illuminated a small spot of the hippocampal slice (Fig. S1A). Light-triggered endogenous NE release reduced the latency of the 1^st AP (p<0.001), but did not reduce the jitter significantly (p=0.15; Fig. 1E, G, H). Notably, similar to NE-perfusion, endogenous NE release decreased the input resistance (p<0.05), indicating that input resistance modulation by NE is physiological, but mediated by other channels than Kv1.1. In contrast, in mono-transgenic control mice without induction of ChR2Y expression, the light trains did not alter the AP onset latency (p=0.39) and the AP onset jitter was increased (p<0.05; Fig. 1F, G, H). These results reveal that similar to bath application, endogenously released NE reduced I_D in CA1 pyramidal neurons.

β receptor-induced inhibition of I_D is mediated by the C-terminal interaction of Kv1.1 with SAP97

SAP97 regulates Kv1.5 surface expression in myocytes (Abi-Char et al., 2008), and SAP97 is expressed in the somato-dendritic compartment of hippocampal neurons (Waites et al., 2009). In light of these results, we tested whether SAP97 interacts with Kv1.1 to regulate its surface expression in CA1 pyramidal neurons. In an immunoprecipitation assay of mouse hippocampal protein extracts, SAP97 was co-immunoprecipitated with Kv1.1, indicating an interaction either directly or in a complex with other proteins (Fig. 2A). As a specificity control, we used extracts from conditional SAP97 KO mice. We crossed a floxed SAP97 mouse line with a NEX-Cre driver line, which deletes SAP97 expression from principal glutamatergic neurons (Goebbels et al., 2006; Zhou et al., 2008). The protein levels of SAP97 in the hippocampus were reduced in the SAP97-NEX mice (97KO^Nex: 40 ± 5.8% of 97^flx; n=3; p<0.01; one sample T-test). This result is in agreement with selective deletion in principal neurons but not in other neuronal types and non-neuronal cells, given that SAP97 expression is ubiquitous (Fig. 2B). Similarly, only a fraction of SAP97 was co-immunoprecipitated with Kv1.1 from SAP97-NEX hippocampal extracts (97KO^Nex: 38 ± 7.5% of 97^flx; n=3; p<0.01; one sample T-test; Fig. 2A). These results confirm that the Kv1.1-immunoprecipitated band contained SAP97, and that the interaction of SAP97 and Kv1.1 is primarily in principal neurons of the hippocampus.

Fig. 2.

The SAP97-Kv1.1 interaction is required for norepinephrine modulation of action potential latency. A. Kv1.1 was immunoprecipitated from mouse hippocampal extracts from floxed SAP97 (97^flx) mice and SAP97-Nex (97KO^Nex) knockout mice, which lacks SAP97 in forebrain glutamatergic neurons. The western blot was decorated with a Kv1.1 antibody or SAP97 antibody to test the complex formation with SAP97. Lane “w/o ab” presents immunoprecipitation without primary antibody as control and “input” presents the hippocampal extract used for immunoprecipitation. B. Analysis of SAP97 knock-out in SAP97-Nex hippocampal P2 fractions. C. Scheme of Kv1.1 and SAP97 interaction for full-length protein (left) and Kv1.1Δ4 (right). D. Kv1.1 was immunoprecipitated from HEK293 cell extracts, transfected with SAP97 and Kv1.1 or Kv1.1Δ4. Samples with cell extract (input) and without primary antibody (w/o ab) were used as controls for protein expression and non-specific bead interactions. E, F, I, J. Repeated rheobase current injections elicited single APs in CA1 pyramidal neurons. Sample recordings before trigger stimulus are presented in black (E, F) or green (I, J) (left traces) and after different stimuli in orange (E) or red (F, I, J) (right traces). 100nM DTX-K (+ DTX-K) to block Kv1.1 (E) or 20μM NE (+NE) to activate adrenoceptors (F, I, J) was applied for 10min as a trigger. E, F. CA1 pyramidal neurons from SAP97-Nex knockout mice were analyzed. G. Scheme of AAV molecular replacement vector to express Kv1.1 or Kv1.1Δ4 in CA1 pyramidal neurons. shRNA targeting endogenous Kv1.1 is driven by a human U6 promoter and recombinant Kv1.1 with silent mutations (*****) in the shRNA target sequence from the CAG promoter. H. Efficiency of the Kv1.1 shRNA was tested in dissociated hippocampal cultures, which were transduced at DIV5 and harvested at DIV14. I, J. Modulation of AP latency was analyzed in transduced CA1 pyramidal neurons, expressing GFP-tagged Kv1.1 (I) or GFP-tagged Kv1.1Δ4 (J). K. Summary of normalized AP latency modulation after different triggers as in panels E, F, I and J normalized to latency in control. L. AP latency jitter of control in black and green and after different triggers in orange and red as in panel E, F, I and J. Scale bar 20mV and 250ms.

The C-terminus of Kv1 interacts with the PDZ domains of the DLG-MAGUKs (Kim et al., 1995). To test whether the SAP97-Kv1.1 interaction is also mediated through the C-terminus of Kv1.1, we constructed KvΔ4, in which the last four amino acids of Kv1.1 were deleted (Fig. 2C). We expressed a GFP-tagged full-length Kv1.1, or GFP-tagged KvΔ4, together with recombinant SAP97 in HEK293 cells. While the immunoprecipitation of Kv1.1 pulled down SAP97, the co-immunoprecipitation of SAP97 was reduced in KvΔ4-expressing cells (KvΔ4: 17.9 ± 0.3% of Kv1.1 wt; n=3; p<0.01; one sample T-test; Fig. 2D), indicating that the interaction was primarily, albeit not exclusively, mediated by the C-terminal PDZ-ligand of Kv1.1.

We then tested the requirement of SAP97 for NE-induced inhibition of I_D. Rheobase current injections induced a delayed AP (564 ± 77.6ms) with an AP onset jitter of 111 ± 15.6ms in CA1 pyramidal neurons of SAP97-NEX mice (Fig. 2E). DTX-K reduced the AP onset latency (p<0.001) and the AP onset jitter (p<0.05; Fig. 2E, K, L). These results indicate that in SAP97-deficient neurons, Kv1.1 surface expression that mediated I_D was unaltered. In contrast, the NE-induced inhibition of I_D was impaired in SAP97-NEX mice. The AP onset latency (p=0.47) and AP onset jitter (p=0.088) were not modulated by NE perfusion (Fig. 2F, K, L). The NE-mediated decrease of the input resistance was preserved in SAP97-deficient neurons (p<0.01), indicating that SAP97 is not involved in other NE-mediated conductance changes in CA1 pyramidal neurons, but rather is specifically required for NE-mediated Kv1.1 inhibition.

To test whether the interaction between SAP97 and Kv1.1 is required for the NE-mediated modulation of I_D, we developed a viral vector-based molecular replacement construct to replace endogenous Kv1.1 with mutant constructs (Fig. 2G; Schlüter et al., 2006). We first designed and validated an shRNA sequence targeting Kv1.1 (shKv1.1), which, expressed from a recombinant adeno-associated viral vector (AAV), knocked down endogenous Kv1.1 from hippocampus primary cultures almost completely, while viral vector expression itself, tested with a GFP-expressing AAV, had no effect on Kv1.1 expression (AAV-shKv1.1: 2.8 ± 1.0% of AAV-GFP; n=4; p<0.01; one sample T-test; Fig. 2H). To prevent silencing of the recombinant Kv1.1 in the replacement vector, we introduced silent mutations in the targeted coding region (Fig. 2G). AAVs expressing either the wild-type replacement with a GFP-tagged full-length Kv1.1 or the GFP-tagged KvΔ4 were injected stereotactically into the dorsal hippocampus of wild-type mice (scheme for KvΔ4 in Fig. 2C). Eight days later, CA1 pyramidal neurons expressing AAVs were identified by their green fluorescence and whole cell recordings were performed from hippocampal slices. In neurons with Kv1.1 wild type-to-wild type replacement as well as in uninfected neurons, perfusion of NE reduced the AP onset latency (p<0.001) and AP onset jitter (p<0.001; Fig. 2I, K, L). In contrast, the KvΔ4 replacement compromised NE-induced inhibition of I_D. Importantly, the latency of the 1^st AP was preserved in neurons with KvΔ4 replacement (570 ± 96.1ms), a result consistent with normal surface expression of KvΔ4. But, NE perfusion was without significant effect on the 1^st AP latency (p=0.084) and AP onset jitter (p=0.073; Fig. 2J–L). These results indicate that the C-terminal PDZ ligand was not required for surface expression of Kv1.1, nor was the interaction with SAP97. Rather, the NE-induced effects on I_D, which were mediated by Kv1.1, required the SAP97-Kv1.1interaction.

Norepinephrine facilitates STD-LTP

Activation of β-adrenoceptors increases dendrite excitability (Hoffman and Johnston, 1999), which facilitates STDP and extends the time window for the association of pre- and postsynaptic stimulation (Lin et al., 2003; Seol et al., 2007). To test the role of NE and Kv1.1 in STDP, we adapted an STDP protocol, which is sensitive to dendritic spike propagation (Carlisle et al., 2008). After a 5–10-min baseline recording of excitatory postsynaptic potentials (EPSP), we applied an induction train at 5Hz, each stimulation containing a presynaptic pulse to trigger EPSPs, and short postsynaptic current to trigger a single backpropagating AP (Fig. 3A). This protocol did not trigger STD-LTP in CA1 pyramidal neurons, measured by the EPSP amplitude relative to the baseline (p=0.25; Fig. 3B, F; Tab. S1). However, a longer postsynaptic stimulation, which triggered a burst of up to 4 backpropagating APs (Fig. 3C), induced LTP (p<0.05; Fig. 3D, F; Tab. S1). While a single backpropagating AP, which likely insufficiently depolarized the synaptic membrane for NMDAR activation, failed to induce STDP (Carlisle et al., 2008), the same protocol induced STD-LTP during perfusion of NE (p<0.001; referred to as NE-LTP hereafter; Fig. 3E–F; Tab. S1).

Fig. 3.

Norepinephrine gates spike timing-dependent long-term synaptic potentiation. A, C. Schematic presentation of EPSP followed by single back propagating AP (A) or AP burst (C) conditioning stimulus with time course of events (left) and example trace (right). Scale bar 10ms (left) and 20mV and 50ms (right). B, D, E, G, I. EPSP amplitude versus time of sample recordings with 100 conditioning stimuli at 5Hz with 1 AP (B, E, G, I) or 1 burst (D), applied at vertical line. Averaged (3 min) EPSP before conditioning at position 1 and 25 min after conditioning at position 2 presented in inset. Scale bar 5mV and 50ms. F, H, J. Summary graphs for different conditions with 1 AP pair conditioning (B, F), 1 AP pair conditioning with 20μM NE application (including interleaved controls) during baseline recording (horizontal line) (E – J), 1 burst pair conditioning (D, F), NMDA receptor blockade with 100 μM APV (G, H) and β2-adrenoceptor blockade with 1μM ICI118551 (I, J). NE-LTP control summary graph from panel F is illustrated in copy as a red line (mean) with grey box (SEM) in panels H and J.

This NE-LTP shares some core features with associative plasticity. First, neither pre- nor postsynaptic stimulation alone was sufficient to induce LTP in the presence of NE, indicating a requirement of coincident pre- and postsynaptic signaling (Fig. S2). Second, NE-LTP was prevented by the NMDAR antagonist APV and ICI118551 (Fig. 3G–J; Tab. S1). Thus, NE-facilitated STD-LTP likely relies on activation of β2-adrenoceptor-induced inhibition of I_D, which can increase dendrite excitability and improve AP backpropagation.

SAP97 is required for norepinephrine-facilitated STD-LTP

To test whether NE-LTP requires SAP97, we used the NE-LTP conditioning protocol in SAP97-deficient CA1 pyramidal neurons of SAP97-NEX mice. While floxed SAP97 littermate controls expressed NE-LTP (n=5; p<0.05), NE-LTP was absent in SAP97-NEX neurons (n=5; p=0.39; Fig. 4A–C). Thus, SAP97 is essential for NE-facilitated STD-LTP.

Fig. 4.

Norepinephrine-facilitated STD-LTP requires SAP97 and Kv1.1 C-terminal SAP97 interaction motif. A, B, E. EPSP amplitude versus time of sample recordings with 20μM NE application during baseline (horizontal line) and 100 conditioning stimuli with 1 AP at 5Hz, applied at vertical line. Average (3min) EPSP before conditioning at position 1 and 25min after conditioning at position 2 presented in inset. Scale bar 5mV and 50ms. C, F. Summary graphs for different conditions with floxed SAP97 (97^flx) mice (B, C), SAP97-Nex knock-out mice (A, C), floxed SAP97 mice (F) and single cell SAP97 knock out (97KO^AAV) with control from Fig. 1F (E, F). D. Schematic presentation of recording configuration and sample fluorescent image of AAV-Cre transduced CA1 pyramidal neurons to generate single cell SAP97 knockouts. G, H. EPSP amplitude versus time of sample recordings with 20μM NE application during baseline (horizontal line) and 100 conditioning stimuli with 1 AP at 5Hz, applied at vertical line. Average (3min) EPSP before conditioning at position 1 and 25min after conditioning at position 2 presented in inset. Scale bar 5mV and 50ms. I. Summary graphs for different conditions with AAV-mediated Kv1.1 wild-type replacement (Kv1.1) (G, I) and Kv1.1Δ4 replacement (H, I).

SAP97 dynamically regulates postsynaptic AMPARs and NMDARs, which directly or indirectly influence excitatory synaptic strength (Nakagawa et al., 2004; Schlüter et al., 2006). To test whether there are direct effects of SAP97 on AMPARs and NMDARs, we used organotypic slice cultures from floxed SAP97 mice, in which we knocked out SAP97 in CA1 pyramidal neurons with a Cre-expressing lentivirus. Using a dual recording configuration, we measured the AMPAR- and NMDAR-mediated excitatory postsynaptic currents (EPSCs) from a transduced neuron lacking SAP97 and a neighboring uninfected control neuron (Schlüter et al., 2006). The average amplitudes of the AMPAR (p=0.18) and NMDAR (P=0.51) EPSCs in SAP97-lacking neurons were not different from that of the control neurons (Fig. S3), indicating that SAP97 does not regulate basal synaptic transmission, a result echoing previous findings (Howard et al., 2010).

We then examined the so-called pairing LTP in SAP97-deficient neurons, in which we used an induction protocol comprising a presynaptic stimulation, paired with a postsynaptic depolarization to −10mV. This pairing LTP is different from STD-LTP because the postsynaptic neuron was clamped at −10mV during the induction such that the activation of NMDARs was independent of dendrite excitability. We stereotactically injected an AAV expressing Cre into the hippocampus of floxed SAP97 mice to knock out SAP97 in individual pyramidal neurons in vivo. The pairing protocol induced LTP in non-transduced control neurons (p<0.05; Fig. S4A, C), which was prevented by APV (p=0.16; Fig. S4B, C). Transduced neurons, which lacked SAP97 and were identified in the collected hippocampal slices by co-expressed GFP (Fig. 4D), exhibited an LTP (p<0.001) with a similar magnitude as in control neurons (Fig. S4D, E). Finally, we tested whether SAP97 was essential in a NE-independent form of STD-LTP, which is induced by multiple backpropagating APs paired with a single presynaptic stimulation (Carlisle et al., 2008). Similar to wild type mice, which were recorded as interleaved controls (Fig. S5A, C), a spike burst of up to 4 backpropagating APs triggered LTP in CA1 pyramidal neurons of SAP97-NEX mice (p<0.01; Fig. S5B, C).

The above results indicate that while SAP97 is not essential for the constitutive machineries underlying the induction and expression of pairing LTP and spike burst STD-LTP, it is essential for NE-LTP.

In SAP97-NEX mice, SAP97 was absent in all principal neurons including presynaptic neurons. To test whether SAP97 cell-autonomously regulates NE-mediated facilitation of STD-LTP and whether SAP97 in the postsynaptic compartment is responsible, we analyzed CA1 pyramidal neurons with single cell deletion of SAP97. We stereotactically injected the AAV-Cre into the hippocampus of floxed SAP97 mice (Fig. 4D). SAP97-lacking neurons did not exhibit NE-facilitated STD-LTP (p=0.14; Fig. 4E, F; Tab. S1), whereas cre-expressing neurons from wild type mice did (p<0.05; Fig. S6; Tab. S1). These results narrowed down an essential function of SAP97 in postsynaptic neurons.

β2-adrenoceptor-dependent LTP of the dentate gyrus requires SAP97

Tetanic stimulation in the perforant path triggers NE release (Bronzino et al., 2001). The noradrenergic innervation in the hippocampus is the densest in the medial molecular layer of the dentate gyrus, the target area of medial perforant path (MPP) synapses (Haring and Davis, 1985). LTP of the medial perforant path requires β-adrenoceptor activation (Bramham et al., 1997; Walling et al., 2004), potentially due to β-adrenoceptor-mediated increase in dentate granule cell excitability (Lacaille and Harley, 1985). To test whether other forms of NE-dependent LTP also require SAP97, we analyzed MPP LTP. Using field potential recordings, we identified MPP synapses based on their signature paired-pulse ratio properties (Colino and Malenka, 1993). After a stable baseline recording of field EPSPs, a high frequency induction stimulation induced LTP (p<0.01; Fig. 5A, D). The MPP LTP required β2-adrenoceptor activation, evidenced as that ICI118551 prevented MPP LTP (p=0.062; Fig. 5B, D). In conditional SAP97 KO mice, created by crossing the floxed SAP97 mice with the CaMKIIα Cre driver mice (Minichiello et al., 1999), SAP97 was deleted in principal neurons including the dentate gyrus granule cells, and MPP LTP was abolished. The post-conditioning responses expressed a small but significant depression (p=0.014; Fig. 5C–D). These results indicate that SAP97 is required for β2-adrenoceptor-associated LTP, regardless of the brain region or the source of NE.

Fig. 5.

β2-adrenoceptor-dependent LTP of dentate gyrus requires SAP97. A, B, C. Field EPSP slope versus time of sample recordings from medial perforant path of dentate gyrus with 3 times 100Hz for 1 sec conditioning stimuli 5 min apart, were recorded under different conditions: floxed SAP97; control slices (A), β2-adrenoceptor blockade with 1μM ICI 118551 (B) and in conditional SAP97 knock-out mice with CaMKIIα-Cre driver (97KO^CaMKII, C). Average (3 min) field EPSP before conditioning at position 1 and 60min after conditioning at position 2 presented in inset. Scale bar 0.5mV and 10ms. D. Summary graphs for different conditions as in panels A–C.

Kv1.1 C-terminal PDZ ligand is required for norepinephrine-facilitated LTP

To determine a potential causal link of NE-LTP to SAP97-mediated Kv1.1 inhibition, we tested the C-terminal interaction of Kv1.1 with SAP97 using a replacement strategy similar to the NE-induced inhibition of I_D (Fig. 2). CA1 pyramidal neurons expressing the Kv1.1 wild type replacement exhibited normal NE-LTP (p<0.001; Fig. 4G, I). In contrast, in KvΔ4 expressing neurons, in which the Kv1.1-SAP97 interaction was impaired (Fig. 2), NE-LTP was abolished (p=0.29; Fig. 4H–I). Thus, the C-terminal interaction of Kv1.1 with SAP97 was required for NE-LTP. Since the expression of KvΔ4 does not alter I_D (Fig. 2) and thus the surface expression of this potassium channel, we conclude that NE-induced inhibition of Kv1.1 is necessary for both reduction of I_D and induction of NE-LTP.

SAP97 is required for the norepinephrine-induced surface removal of Kv1.1

To explore the mechanisms underlying NE-mediated inhibition of Kv1.1, we measured the surface expression of Kv1.1. Previous results regarding the subcellular localization of Kv1.1 are inconsistent. Some suggest an exclusive localization of the Kv1 protein family and Kv1.1 particularly in the axon initial segment (Kirizs et al., 2014; Kole et al., 2007; Shu et al., 2007), while others show Kv1.1 expression in CA1 pyramidal neuron dendrites (Guan et al., 2006; Raab-Graham et al., 2006). To determine the Kv1.1 subcellular localization in our case, we co-labeled low-density hippocampal neuron cultures with marker antibodies for axonal neuro-filaments (SMI-312 or Ankyrin-G) and dendritic microtubules (MAP-2) with an antibody against Kv1.1. The fluorescence signals were analyzed with a confocal microscope. The axonal neuro-filament marker decorated the initial segment of one to two neurites, while the majority of neurites and soma were only weakly stained (Fig. 6A, S7A). The Kv1.1 antibody decorated all neurites and somas of the hippocampal neurons. In contrast, the dendritic microtubule marker decorated all neurites except for one or two thin ones originating from the soma and labeled with the Kv1.1 antibody (Fig. 6B). Kv1.1 was co-localized both with MAP-2 and the axonal neuro-filament. These results indicate that Kv1.1 is localized in axons, soma, and dendrites of hippocampus neurons.

Fig. 6.

Norepinephrine triggered reduction in surface Kv1.1 requires SAP97. A, B. Dissociated hippocampal cultures were fixed, permeabilized and decorated with antibodies directed against SMI-312 and Kv1.1 (A) or MAP-2 and Kv1.1 (B). Fluorescence images were acquired with a confocal microscope. Arrows mark dendrite, while arrowheads mark axons. C. Neurons were fixed and decorated with a Kv1.1 antibody recognizing an extracellular epitope in non-permeabilizing conditions. Three primary dendrites were analyzed as the exemplified one with the boxed dendrite, magnified at bottom. Reduction of Kv1.1 surface expression was induced with 20μM NE and it was blocked with 1μM ICI118551. D. Summary graph of normalized surface fluorescence relative to control for different conditions as in panel C. E. Analysis of AAV-Cre transduced cultures of floxed SAP97 knockout mice. Single cell knockout cell (sKO) were identified by their green fluorescence. Surface expression of Kv1.1 was analyzed in untreated control cultures and NE treated sister cover slips. F. Summary graph of surface fluorescence for the conditions in panel E, normalized to control condition.

To measure the surface expression of Kv1.1, we decorated Kv1.1 in non-permeabilized neurons from floxed SAP97 mice with an antibody against an extracellular epitope (Fig. 6C). In primary dendrites, a 10-min incubation of NE decreased the fluorescence intensity compared to that of control neurons (p<0.001; Fig. 6C, D). Furthermore, the NE-triggered Kv1.1 surface reduction was prevented in the presence of ICI118551 (p=0.08; Fig. 6C, D).

We tested the specificity of the Kv1.1 surface labeling in two experiments. First, in primary neuron cultures, knocking down Kv1.1 with shKv1.1 prevented the antibody decoration (Fig. S7B). Second, a Kv1.1 antibody against an intracellular Kv1.1 epitope did not decorate non-permeabilized neurons, while after membrane permeabilization and antibody access to the intracellular epitopes, neurons were decorated (Fig. S7B). These results verify that our experimental conditions allowed a specific detection of Kv1.1 and surface Kv1.1. Notably, some axons adjacent to the dendrites were also labeled with Ankyrin-B immunostaining, but with much lower intensity compared to dendritic labeling (Fig. S7C). Thus, some axonal Kv1.1 may have contributed to a small portion of surface dendritic Kv1.1 (Fig. 6C). Despite this, SAP97 is primarily localized in the somatodendritic compartment with little amounts in axons (Waites et al., 2009). As such, SAP97-dependent Kv1.1 modulation is likely restricted to the somatodendritic compartment (see Fig. 7 below).

Fig. 7.

β-adrenoceptor-facilitated increase of dendritic excitability requires SAP97. A. AAV-ChR2Y was stereotactically injected into the CA1 area of SAP97-Nex (97KO^Nex) knockout mice or littermate floxed SAP97 (97^flx) controls. ChR2Y expression was visualized in hippocampal slices with fluorescent illumination. B. Schematic representation of recording configuration with somatic patch clamp recording from CA1 pyramidal neurons and 473nm laser light stimulation in a 40-μm diameter spot in stratum radiatum (blue). C. Sample traces of triggered APs in 97^flx mice with 2 ms current injection (1nA; left) or 1–2 ms 473nm laser light stimulation (blue mark) on soma (middle) or dendrite (right). Scale bar: 20mV and 50ms. D. Summary graph of AP threshold of somatic current injections (ΔI_soma) and somatic (hν_soma) or dendritic (hν_dendrite) ChR2Y activation in 97^flx control or 97KO^Nex mice. Statistical analysis was performed with two-factor ANOVA and Sidak post-hoc analysis. E, F. Modulation of AP threshold and latency with 10min β-adrenoceptor activation in 97^flx (E) and 97KO^Nex (F) mice. Sample traces with AP before 10μM isoproterenol (ISO) perfusion (control) and after 5min ISO perfusion. Scale up of time axes right. Scale bar: 20mV, 5ms. G, H. Summary graph of AP latency (G), calculated relative to control condition and change in AP threshold (H) after ISO perfusion. I. Schematic recording configuration of dendritic spikes with dendritic patch clamp electrophysiology of apical CA1 pyramidal neuron dendrites and electrical stimulation in the alveus to trigger back-propagating APs. J–L. Dendritic spikes before and after ISO (2μM, J, K) or DTX-K (L) perfusion of 97^flx (J, L) and 97KO^Nex (K) mice. Scale bar 10mV and 5ms. M–N. Summary graph of time course of change (M) and relative change after 10min (N) in dendritic spike amplitude after drug perfusion. O–R: 0.5 ms light (blue bar) pulses triggered somatic subthreshold depolarization in 97^flx neurons (M) or 97KO^Nex neurons (N). Scale bar: 10mV, 25ms. After 10min stimulation of β-adrenoceptors, depolarizations with same light pulse intensities became superthreshold in 97^flx (M) but not in 97KO^Nex (N). O. Summary graph of average subthreshold depolarizations before ISO application. P. Summary graph of fraction of APs triggered before ISO (−ISO) and after ISO (+ISO) perfusion.

To knock out SAP97, we transduced the neuronal cultures with a lentivirus, expressing a bicistronic expression cassette with GFP and Cre. The fluorescence was normalized to the signal from control neurons without Cre-expressing viral vectors. Incubation with NE did not change the fluorescence intensity of surface Kv1.1 in SAP97-deficient neurons (p=0.19; Fig. 6E, F). Thus, NE removed Kv1.1 from the dendrite surface and this removal was abolished in SAP97-deficient neurons.

Norepinephrine induces a SAP97-mediated increase in dendrite excitability

β-adrenoceptor activation increases dendrite excitability (Hoffman and Johnston, 1999; Yuan et al., 2002). To test whether the NE-induced alteration in dendrite excitability relies on SAP97, we used two approaches: Optogenetics and dendritic patch clamp recordings. The optogenetic approach allows a subcellular-specific analysis of the excitability by restricting the illumination spot (Petreanu et al., 2009). We selectively analyzed dendritic or somatic depolarization by focusing a collimated laser beam (~40μm; Fig. S1A) onto the targeted subcellular compartment. To do this, we stereotactically injected an AAV1/2 with a ChR2Y expression cassette into the hippocampus of SAP97-NEX and floxed SAP97 control mice. Nine days later, we patched on ChR2Y-expressing CA1 pyramidal neurons (Fig. 7A). Using current-clamp mode, we adjusted the injection current to hold the neurons at −70mV, and synaptic transmission was blocked pharmacologically. We then depolarized the neurons by either applying a laser beam on the apical dendrites at a distance of ~200μm from the stratum pyramidale or onto the soma (Fig. 7B). 1–2ms light pulses triggered APs (Fig. 7C), with several electrophysiological properties, such as the depolarization/repolarization dynamics and the AP threshold similar to somatic current-evoked APs (F_2,62=25.4 for treatment, p<0.0001; I_inj 97^flx vs. hν_soma 97^flx p=0.99; Fig. 7C, D). In SAP97-Nex mice, the AP thresholds for somatic current injection-evoked and somatic ChR2-evoked APs were similar to that in floxed control mice (F_1,62=0.216 for genotype, p=0.64; I_inj 97KO^Nex vs. I_inj 97^flx p=0.99; hν_soma 97KO^Nex vs. hν_soma 97^flx p=0.98; Fig. 7D). These results indicate that the somatic ChR2-evoked APs were indistinguishable from APs generated by somatic current injections, thus validating the use of this optogenetic approach.

We then placed the laser spot on the apical dendrites of CA1 pyramidal neurons of floxed SAP97 control mice. The AP initiation threshold of the dendritic, ChR2-evoked APs was lower compared to the one evoked by somatic current injection (F_2,62=25.4 for treatment, p<0.0001; hν_dendrite 97^flx vs. I_inj 97^flx p<0.001; Fig. 7C, D). The narrow shape of the APs triggered by dendritic ChR2 activation was similar to that of the somatic ChR2-evoked APs, which was likely triggered from the axon hillock (Kole and Stuart, 2008). It is important to note that the reduction in the AP threshold measured from the soma does not indicate an actual change of the AP threshold at the axon hillock, as demonstrated in a previous study, measuring AP generation from different distances to the axon hillock (Kole and Stuart, 2008). Rather, this reduction is likely consequence of dendritic conductances that prevented further depolarization. This apparent reduction in the AP threshold allowed us to measure dendritic depolarization remotely from the soma. In SAP97-Nex mice, the threshold of the dendritic ChR2-evoked AP was similar to that of control floxed SAP97 mice (F_1,62=0.216 for genotype, p=0.64; hν_dendrite 97KO^Nex vs. hν_dendrite 97^flx p=0.99; Fig. 7D).

We tested whether the propagation of ChR2-induced dendrite depolarization is modulated by β-adrenoceptor activation. We recorded and compared dendritic ChR2-evoked APs before and after perfusion with the β-adrenoceptor agonist isoproterenol (ISO). ISO reduced the AP threshold (p<0.05; Fig. 7E, H) and the latency of APs (p<0.001; Fig. 7E, G), whereas these effects of ISO were absent in SAP97-Nex mice (threshold: p=0.73; latency: p=0.11; Fig. 7F–H). Thus, SAP97 is essential for β-adrenoceptor-mediated modulation of APs induced by dendrite depolarization.

In the above experiments, while the depolarization was triggered in dendrites, the AP was recorded from the soma. To directly measure dendrite excitability, we performed dendritic patch-clamp recordings to examine the apical CA1 pyramidal neuron dendrites in response to back-propagating APs, stimulated in the alveus with an extracellular electrode (Yuan et al., 2002; Fig. 7I). We established dendrite recordings under visual guidance at a distance ~200μm from the pyramidal cell layer. Compared to somatic APs (see below), the amplitude of the dendritic spikes was much lower but with comparable size in floxed SAP97 control (27.9 ± 6.5mV, n=6) and SAP97-Nex mice (27.9 ± 4.9 mV, n=7; 97^flx vs. 97KO^Nex, p=0.997; Fig. 7J, K). However, perfusion with ISO increased the amplitude of the backpropagated dendritic spikes in floxed SAP97 control mice, but not in SAP97-Nex mice (97^flx vs. 97KO^Nex p<0.001; Fig. 7M–N). To verify the role of Kv1.1 in dendrite excitability, we perfused the slices from floxed SAP97 control mice with DTX-K. Pharmacological inhibition of Kv1.1 increased the amplitude of the dendritic spike (AP amplitude; before, 33.2 ± 3.0mV vs. during, 44.0 ± 2.5mV; p<0.01; Fig. 7L–N), indicating that controlling the activity of Kv1.1 sensitively regulates dendrite excitability. Importantly, the increase in the dendritic spike was independent of somatic APs; the amplitude of somatic APs triggered by alveus stimulation was 83 ± 5.2mV and did not increase after ISO perfusion (p=0.42). These results support the notion that β-adrenergic signaling directly modulates dendritic spikes through SAP97-mediated inactivation of Kv1.1.

In the final experiments, we tested whether the β-adrenoceptor-induced increase in the dendrite excitability is critical for promoting AP generation. We adjusted the laser intensity to trigger a modest depolarization of ~20mV (Fig. 7M, N). This modest depolarization only triggered APs occasionally (Fig. 7O, P). Perfusion of ISO increased the ChR2-induced depolarization, resulting in additional depolarization and the summed depolarization triggered reliable AP firing in control mice (Fig. 7M, P). However, in SAP97-Nex mice, perfusion of ISO did not affect the dendritic ChR2-induced depolarization and did not affect the AP firing (Fig. 7N, P). These results, taken together with above results, showing dendritic localization of Kv1.1 and its β2-adrenoceptor-dependent modulation (Fig. 6), indicate that β-adrenoceptor-mediated regulation of dendrite excitability is mediated by the SAP97-Kv1.1 interaction.

Discussion

The back-propagating AP is a critical second factor for associative synaptic plasticity to sufficiently depolarize the synaptic membrane for NMDAR unblocking (Bi and Poo, 1998; Magee and Johnston, 1997; Markram et al., 1997). Postsynaptic depolarization is paired with synaptic glutamate release, which is the first factor of the contingency. However, dendritic potassium currents, including Kv4.2 as shown previously (Hoffman et al., 1997), and Kv1.1 as shown here (Fig. 7), damp dendritic excitation and thus prevent the invasion of individual APs into the dendritic trees of CA1 pyramidal neurons. Neuromodulatory systems, including β-adrenergic signaling, increase dendrite excitability and thus promote the spread of depolarization (Dunwiddie et al., 1992; Hoffman and Johnston, 1999). However, the identity of the involved potassium channels remained unknown. Our present results show that Kv1.1-containing potassium channels gated STD-LTP and that β2-adrenoceptor-mediated inhibition of these potassium channels released the gate. Furthermore, SAP97 is the likely common signaling scaffold that regulates the gating mechanism of different forms of β2-adrenoceptor-dependent synaptic plasticity. This SAP97-dependent adrenergic modulation integrates the internal state of the brain as a third factor into synaptic plasticity. Since learning is facilitated during attentional and emotional arousal, when noradrenaline is released, synaptic plasticity with all these three factors might better represent behaviorally relevant mechanisms (O’dell et al., 2015; Pawlak et al., 2010; Sara and Bouret, 2012).

β2-adrenoceptor signaling modulates dendrite excitability through Kv1.1

Prior studies characterize that axon-located Kv1-containing potassium channels modulates AP generation (Bekkers and Delaney, 2001; Gu et al., 2003; Smart et al., 1998). Recent studies using electrophysiology, immunocytochemistry and local translation analysis reveal that Kv1 subunits, including Kv1.1, are also expressed in the somato-dendritic compartment (Chen and Johnston, 2004; Guan et al., 2006; Raab-Graham et al., 2006). Since dendrite excitability has long been thought to be dominantly regulated by the A-type potassium channels containing Kv4.2 (Hoffman et al., 1997), the function of dendritic Kv1 remains underexplored. Our present results demonstrate a key role of Kv1.1 in regulating dendrite excitability. Kv1.1 was prominently expressed in the dendrites of hippocampal neurons (Fig. 6), and interacted with a preferentially dendrite-expressed signaling scaffold, SAP97 (Fig. 2; Waites et al., 2009). Furthermore, β2-adrenoceptor activation reduced the dendritic surface expression of Kv1.1 (Fig. 6). It has long been known that β-adrenoceptor activation increases dendrite excitability by potassium channel inhibition (Hoffman and Johnston, 1999; Yuan et al., 2002), but it remained unclear which type of potassium channels are the key. Using a molecular replacement approach to substitute endogenous Kv1.1 with KvΔ4, our results reveal the requirement of Kv1.1, and furthermore, the critical role of its C-terminus. After KvΔ4 replacement, the somatic membrane excitability, as measured by the AP latency (Fig. 2), was not altered, but the β2-adrenoceptor-mediated modulation was abolished. This modulation primarily occurred on dendrites, demonstrated as β2-adrenoceptor-induced Kv1.1 surface removal on dendrites (Fig. 6) and as changes in the membrane excitability confined in the dendrite but not soma (Fig. 7). Furthermore, in SAP97 KO mice, β-adrenoceptor-induced modulation of I_D, the dendritic AP amplitude, and ChR2-triggered dendrite depolarization were all abolished (Fig. 2, 7). This complete abolishment indicates an essential role of the Kv1.1-SAP97 interaction in these adrenergic modulations. While it remains to be tested whether other potassium channels, such as delayed rectifier, D-, M- and A-type that are expressed in CA1 pyramidal cell apical dendrites (Chen and Johnston, 2004), are also under β-adrenoceptor-mediated modulation, our results identify Kv1.1 as the key potassium channel type that mediates I_D in CA1 pyramidal neurons (Fig. 1, 2). The contribution of dendritic Kv1.1 to I_D is in agreement with previous studies, showing an attribution of other paralogs of the Kv1 family, such as dendritic Kv1.2, to I_D in other cortical neurons (Bekkers and Delaney, 2001; Hyun et al., 2013).

SAP97 links β2-adrenoceptor signaling to Kv1.1

The Kv1 family of potassium channels binds through their C-terminal PDZ-ligand to the PDZ domains of the DLG-MAGUKs SAP97 and PSD-95, to form multimeric protein clusters (Kim et al., 1995; Tiffany et al., 2000). Previous studies in heterologous systems report that SAP97 and PSD-95 differentially regulate surface expression of Kv1 (Tiffany et al., 2000). While SAP97 prevents surface expression of Kv1 subunits, PSD-95 clusters them on the cell surface. Our present results in hippocampal neurons reveal a different function of the PDZ interaction. The 4 amino acids in the C-terminus of Kv1.1 were not required for surface expression, as the KvΔ4 replacement did not alter the I_D-mediated AP latency (Fig. 2). Additionally, the surface expression was not altered in SAP97-lacking CA1 pyramidal neurons (Fig. 2, 6). In contrast, surface removal of Kv1.1 requires its C-terminus to bind to SAP97, evidenced as that the β2-adrenoceptor-mediated surface removal of Kv1.1 was abolished by the KvΔ4 replacement or SAP97 knock out (Fig. 2, 6). These results indicate that the Kv1.1-DLG-MAGUK interaction mediates the surface removal rather than surface expression of Kv1.1. The most parsimonious interpretation is that the Kv1.1 C-terminus recruits SAP97 rather than SAP97 mobilizing Kv1.1. At the dendritic membrane, SAP97 then may act as a signaling complex to coordinate the signaling machinery for β2-adrenoceptor-mediated Kv1.1 surface removal. Consistent with this notion, Kv1.1 surface expression is reduced after β-adrenoceptor activation (Fig. 6). This interpretation does not exclude the possibility that SAP97 might additionally tether multiple Kv1.1 on the surface into clusters, which might otherwise be dispersed.

Norepinephrine-induced Kv1.1 inhibition facilitates STD-LTP

Consistent with previous observations (Carlisle et al., 2008), our results confirm that repetitive presynaptic activation followed by a single postsynaptic backpropagating AP, the canonical induction procedure of STDP at glutamatergic synapses, was insufficient to trigger LTP in the mature hippocampus (Fig. 3). Thus, this canonical form of STDP appears to be restricted to the developing neural circuits (Bi and Poo, 1998; Debanne et al., 1996), whereas in mature neural circuits, STD-LTP typically requires presynaptic activation followed by a burst of backpropagating APs (Buchanan and Mellor, 2007; Carlisle et al., 2008; Pike et al., 1999; Watanabe et al., 2002; Wittenberg and Wang, 2006; Fig. 3) or pre- and postsynaptic stimulations at high frequencies (Bauer et al., 2001; Chen et al., 2006; Markram et al., 1997). Our present results reveal that this constrain is released by β2-adrenoceptor activation (Fig. 3), which through the signaling scaffold SAP97 results in dendritic Kv1.1 inhibition (Fig. 6) and an increase of dendrite excitability (Fig. 7). Similarly, MPP LTP in the dentate gyrus requires both β2-adrenoceptor activation and the presence of SAP97 (Fig. 5), and thus is likely gated through similar mechanisms.

While our results reveal a molecular mechanism underlying β-adrenoceptor-mediated facilitation of STD-LTP induction, β-adrenoceptors also facilitate the expression phase of LTP (Hu et al., 2007; Makino et al., 2011; Seol et al., 2007; Thomas et al., 1996). Activation of β-adrenoceptors triggers the phosphorylation of the AMPAR subunit GluA1 at the S845 site, which facilitates synaptic incorporation of AMPARs during LTP expression (Hu et al., 2007). To facilitate LTP expression, the β-adrenoceptor activation can happen even after STD-LTP induction in the visual cortex (He et al., 2015). Thus, β-adrenoceptors facilitate LTP at multiple steps, the induction by Kv1.1 inhibition and the expression by AMPAR priming. At least in hippocampal CA1 pyramidal cells, both mechanisms seem to be required, as disrupting either mechanism prevents STD-LTP or field potential LTP.

Norepinephrine-induced Kv1.1 inhibition increases spike precision

Additional to its role in mnemonic processes, locus coeruleus neurons are activated upon salient stimuli, inducing a noradrenergic sharpening of sensory perceptions (Sara and Bouret, 2012). Priming stimulations of the locus coeruleus reduce the spontaneous activity of neurons both in the somatosensory and piriform cortices to regulate evoked sensory responses (Lecas, 2004; Sara and Bouret, 2012). The spike latency is shorter and the latency jitter is reduced, increasing the synchronous precision of consecutive spikes. Precise and reliable spiking is critical for stimulus encoding (Tiesinga et al., 2008). We show that Kv1.1 activity causes a high jitter of somatically triggered APs, potentially by preventing the depolarization of the membrane to reach the threshold for AP initiation (Fig. 1). In support of this notion, rheobase current injections resulted in a prolonged plateau of the membrane potential, while NE-induced Kv1.1 inhibition decreased the AP onset jitter (Fig. 1). Additionally, NE-induced Kv1.1 inhibition increases dendrite excitability, decreasing the spike latency and facilitating AP initiation, both consequences contributing to the increased precision and reliability of sensory responses for perceptual acuity.

In conclusion, we report a novel role of SAP97 in coupling β2-adrenergic signaling with Kv1.1, with a cellular consequence that may increase spike precision and gate release STD-LTP. These mechanisms may provide cellular interpretation of how neuromodulators regulate synaptic plasticity and how the brain state gates memory encoding.

Experimental Procedures

Mice

Mouse lines of conditional SAP97 KO mice (Zhou et al., 2008), Nex-Cre (Goebbels et al., 2006), CaMKIIα-Cre (Minichiello et al., 1999) and DBH-Cre (Parlato et al., 2007) driver lines and ChR2Y reporter mice (Madisen et al., 2012) were bred on a mixed 129SV/C57Bl6 background. Experimental animals with littermate controls were generated from heterozygous breeding pairs, and crossed between the lines. All animal procedures were performed by following the procedures approved by the animal care and use committees of the University Medical School Göttingen, the Lower Saxony State Office for Consumer Protection and Food Safety, or the University of Pittsburgh.

Immunocytochemistry

Hippocampal cultures were prepared from P0–1 mice as described (Krüger et al., 2013). For immune staining, neurons were fixed with 4% paraformaldehyde. Primary antibodies were incubated with non-permeabilized neurons or after Triton X-100 permeabilization and followed by incubations of fluorescent secondary antibodies before mounting.

Image analysis and quantification

Images of immune-labeled coverslips were acquired using a confocal microscope at 4096×4096 resolution and analyzed using NIH ImageJ software with normalization to the average surface fluorescence of untreated control cells.

Biochemistry

Immunoprecipitation was performed as described previously (Zhang et al., 2008). Immunoprecipitates were separated by Bis-Tris gels (Bonnet et al., 2013), transferred onto nitrocellulose and analyzed using standard procedures (Krüger et al., 2013).

Virus injection

P0 or P22–27 mice were anesthetized by hypothermia or avertin, respectively. AAV was injected relative to the lambda suture (P0) or Bregma (P22–27). After recovery, mice were transferred to home cage.

Slice preparation

Hippocampal transverse slices were prepared from P28–40 for whole-cell recordings, and P60–90 for field potential recordings in ice-cold sucrose cutting buffer (in mM: NaCl 25, sucrose 168, KCl 1.9, MgSO₄ 10, NaHCO₃ 26, NaH₂PO₄ 1.2, D-glucose 25, equilibrated with 95% O₂, 5% CO₂) or in a previously described choline chloride based cutting solution for dendritic patch clamp recordings (Makino et al., 2011). After slicing, slices were recovered in recovery solution (in mM: NaCl 119, KCl 1.9, CaCl₂ 1, MgSO₄ 10, NaHCO₃ 26, NaH₂PO₄ 1.2, D-glucose 20, equilibrated with 95% O₂, 5% CO₂). Slices were kept in a holding chamber, containing recording ACSF (in mM: NaCl 119, KCl 2.5, CaCl₂ 2.5, MgSO₄ 1.3, NaHCO₃ 26, NaH₂PO₄ 1, D-glucose 10, equilibrated with 95% O₂, 5% CO₂).

Whole-cell patch-clamp recording

Slices were maintained in a submerged-slice recording chamber with warm ACSF (30±0.5°C), containing picrotoxin (50 μM). Patch electrodes (4–6 MΩ or 12–15 MΩ for dendritic patch clamp recordings), filled with a solution containing current-clamp internal solution (in mM: potassium gluconate 130, KCl 10, EGTA 0.2, Mg-ATP 4, Na-GTP 0.5, HEPES 10 pH 7.2–7.4). EPSPs were elicited once every 20 sec using a bipolar stimulating electrode placed in the stratum radiatum. Constant injections of hyperpolarizing current were used to maintain membrane potentials between −70 and −80mV. STDP was induced as described previously (Carlisle et al., 2008). NE (20μM) was applied for 10 min after formation of the whole-cell recording until the start of the STD-LTP induction protocol.

To study membrane excitability AMPAR/NMDAR transmission was blocked. APs were elicited by current injections, which size was adjusted to trigger ≤ 2 APs. 20–30 APs were elicited every 15sec before bath application of NE or optogenetic stimulation. To trigger antidromic APs for dendritic patch clamp recordings, tungsten electrodes were placed in the alveus. 0.1msec constant current pulses were applied every 15 sec.

Compartment-specific optogenetic recording

We stereotactically injected AAV-ChR2Y into the hippocampal CA1 area of the mice at P22–27. After 9–18d hippocampal slices were collected. We recorded CA1 pyramidal neurons in the current-clamp mode. ChR2Y was activated with light pulses, applied to the stratum radiatum 150–230μm distal from the cell soma.

Data acquisition and statistics

Electrophysiological data was on-line filtered at 4kHz, collected with custom routines in Igor (Wavemetrics) and digitized at 10kHz with an ITC-18 (HEKA). Data are presented as mean±SEM. For statistical analysis, the intra group before/after comparisons were performed with paired t-test with unequal variance. The differences of normalized data was analyzed by single wave t-test. For absolute values normal distribution was assumed, and statistical analysis was performed with two-tailed t-test for two experimental groups or two-factor ANOVA and Sidak’s post-hoc multiple comparisons test for multiple groups. Relative values were analyzed with two-tailed Mann-Whitney test for two experimental groups and Kruskal-Wallis test with Dunn’s post-hoc multiple comparisons test. The levels of significance were set as *: p<0.05; **: p<0.01.

Highlights.

• Kv1.1 mediates dendritic potassium current in hippocampal neurons.

• Kv1.1 gates spike timing-dependent LTP at hippocampal synapses.

• SAP97 enables β2-adrenoceptor-induced inhibition of dendritic Kv1.1.

• SAP97/β2/Kv1.1-mediated regulation of dendrite excitability gates LTP.