Differential regulation of synaptic transmission by pre- and postsynaptic SK channels in the spinal locomotor network

Abstract

The generation of activity in the central nervous system requires precise tuning of cellular properties and synaptic transmission. Neural networks in the spinal cord produce coordinated locomotor movements. Synapses in these networks need to be equipped with multiple mechanisms that regulate their operation over varying regimes to produce locomotor activity at different frequencies. Using the in vitro lamprey spinal cord, we explored whether Ca²⁺ influx via different routes in postsynaptic soma and dendrites and in presynaptic terminals can activate apamin-sensitive Ca²⁺-activated K⁺ (SK) channels and thereby shape synaptic transmission. We show that postsynaptic SK channels are tightly coupled to Ca²⁺ influx via NMDA receptors. Activation of these channels by synaptically induced NMDA-dependent Ca²⁺ transients restrains the time course of the synaptic current and the amplitude of the synaptic potential. In addition, presynaptic SK channels are activated by Ca²⁺ influx via voltage-gated channels and control the waveform of the action potential and the resulting Ca²⁺ dynamics in the axon terminals. The coupling of SK channels to different Ca²⁺ sources, pre- and postsynaptically, acts as a negative feedback mechanism to shape synaptic transmission. Thus SK channels can play a pivotal role in setting the dynamic range of synapses and enabling short-term plasticity in the spinal locomotor network.

computation in the central nervous system (CNS) is the consequence of precise modulation of synaptic transmission and neuronal activity. The conductance of specific ion channels is regulated to produce complex cognitive and motor functions. In the spinal locomotor networks, apamin-sensitive Ca²⁺-activated K⁺ (SK) channels control the firing properties of the constituent neurons and regulate the frequency of the locomotor rhythm (Diaz-Rios et al. 2007; El Manira et al. 1994; Hill et al. 1992; Hounsgaard et al. 1988; Meer and Buchanan 1992; Miles et al. 2005; Wall and Dale 1995). In these networks, voltage-activated Ca²⁺ channels have been considered the main source for activation of the postsynaptic SK channels (El Manira et al. 1994, 2010; Harris-Warrick 2010; Hill et al. 1992; Wall and Dale 1995). Specifically, in lamprey and Xenopus tadpoles, SK channels are activated by Ca²⁺ influx via N- and P/Q-type channels (El Manira et al. 1994; Hill et al. 1992; Wall and Dale 1995; Wikström and El Manira 1998). Blockade of SK channels with apamin affects locomotor activity by increasing the locomotor burst duration and decreasing the burst frequency predominantly at low frequencies (El Manira et al. 1994; Grillner 2006; Hill et al. 1992; Wallén 1989). Both Ca²⁺ and, indirectly, SK channels are the target of modulatory systems that impact locomotor output (El Manira et al. 2010; Harris-Warrick 2010; Lebeau et al. 2005; Miles and Sillar 2011; Toledo-Rodriguez et al. 2005).

Voltage-activated Ca²⁺ channels are not the only source of Ca²⁺ influx into spinal neurons, but it is still unclear if SK channels are also activated by Ca²⁺ arising from other sources such as ionotropic receptors. The generation of locomotion is dependent on activation of NMDA receptors, which represent another major source of Ca²⁺ influx (Bacskai et al. 1995; Cazalets et al. 1992; Grillner 2006; Issberner and Sillar 2007; Kiehn 2006; Kyriakatos et al. 2011; Li et al. 2010; Roberts et al. 2010). However, it is unclear if NMDA receptors and SK channels are colocalized in spinal cord neurons and how such an interaction would affect synaptic transmission within the locomotor network (Cangiano et al. 2002). Such a colocalization and the resultant coupling between synaptically driven NMDA receptor-mediated Ca²⁺ influx will have profoundly different computational implications for intrinsic pattern generation in the neuron. In addition, it is not known if SK channels are also located presynaptically on axon terminals and how they might affect synaptic transmission in the spinal cord. At the frog neuromuscular junction, presynaptic charybdotoxin-sensitive Ca²⁺-activated K⁺ channels have been shown to be activated by Ca²⁺ entering the terminal and to regulate transmitter release (Robitaille and Charlton 1992; Robitaille et al. 1993).

In this study we used a combination of physiological and imaging approaches to determine if Ca²⁺ through NMDA receptors activates postsynaptic SK channels and whether presynaptic SK channels regulate synaptic transmission. Our results show that postsynaptic SK channel activation is coupled to Ca²⁺ influx via NMDA receptors and shapes synaptic responses. In addition, we show that presynaptic SK channels affect Ca²⁺dynamics in presynaptic terminals to modulate synaptic transmission. Thus SK channels act both pre- and postsynaptically to shape synaptic transmission and subsequent final locomotor output.

MATERIALS AND METHODS

Experiments were performed on 80 sea lampreys (Petromyzon marinus). All protocols were approved by the animal research ethical committee in Stockholm. Lampreys were anesthetized with MS 222 (100 mg/l; Sigma-Aldrich, St Louis, MO) and eviscerated, and the lateral muscle walls were removed. The brain stem and spinal cord were dissected and pinned down in a cooled (8–10°C) Sylgard-lined experimental chamber continuously perfused with physiological solution.

Cell dissociation and electrophysiology.

Spinal motoneurons and interneurons were cultured by dissociation of the spinal cord of late larval stages or transformer lampreys. The spinal cords were treated first by collagenase (1 mg/ml, 30 min; Sigma-Aldrich) and then by protease (2 mg/ml, 45 min; Sigma-Aldrich) diluted in Leibovitz's L-15 culture medium (Sigma-Aldrich) with penicillin-gentamicin (2 μl/ml; Sigma-Aldrich) added. The osmolarity of the culture medium was adjusted to 270 mosM, matching the extracellular recording solution. After enzyme treatment, the tissue was subsequently washed with the culture medium and triturated through a sterilized glass pipette. The dissociated neurons were then distributed in 15 petri dishes and incubated at 10–12°C for 1–2 days.

Whole cell voltage-clamp recordings were made in a Mg²⁺-free extracellular solution while the membrane potential of neurons was held at −60 mV. The extracellular solution was applied through a gravity-driven microperfusion system and contained (in mM) 124 NaCl, 2 KCl, 1.2 MgCl₂, 2 CaCl₂, 10 glucose, and 10 HEPES (pH 7.60). The intracellular solution contained (in mM) 110 KCl, 1.2 MgCl₂, 10 glucose, and 10 HEPES (pH 7.59). NMDA (Sigma-Aldrich) was applied for 4 s at a concentration of 200 μM together with 1 μM glycine (Sigma-Aldrich). The solutions were applied through a gravity-driven microperfusion system with the tip placed close to the recorded cell. The recording pipettes had resistances in the range of 6–8 MΩ and series resistance (R_s) was monitored during the whole experiment with values around 10 MΩ, which was compensated by 65–80% using a lag value of 10 μs in the R_s compensation circuit of the Axopatch 200 B amplifier.

Spinal cord in vitro.

For whole mount intracellular recordings, the spinal cord of transformer lampreys was dissected free of the notochord and mounted in a chamber, whereas for patch-clamp recordings, an additional thin slice from the ventral horn was removed with the use of a Vibratome to provide access to the gray matter. The spinal cord was then mounted ventral side up in a cooled (8–10°C) Sylgard-lined chamber. A glass suction electrode was placed on the ventral surface of the spinal cord to stimulate descending reticulospinal axons and elicit excitatory postsynaptic potentials (EPSPs) and currents (EPSCs) in the recorded neurons. The preparation was continuously perfused with a Mg²⁺-free extracellular solution containing (in mM) 110 NaCl, 2 KCl, 2 CaCl₂, 10 glucose, and 10 HEPES (pH 7.6). Whole cell patch-clamp recordings were made from somata of spinal neurons. The intracellular solution contained (in mM) 102 KCH₃SO₃, 1.2 MgCl₂, 10 glucose, and 10 HEPES (pH 7.4). ATP (3 mM; Sigma-Aldrich), GTP (1 mM; Sigma-Aldrich), phosphocreatine (5 mM; Sigma-Aldrich), QX-314 (5 mM; Tocris), BAPTA (10 mM; Invitrogen), and EGTA (10 mM; Sigma-Aldrich) were added to the patch intracellular solution. GYKI-52466 (30 μM; Tocris), strychnine (5 μM; Sigma-Aldrich), 2-amino-5-phosphonopentanoate (AP-5; 50 μM; Tocris), and apamin (1 μM; Tocris) were added to the perfusing solution. The glass capillaries had resistances in the range of 4–6 MΩ, and R_s was monitored during the whole experiment with values <15 MΩ, which was compensated by 50–65% using a bandwidth of 15 kHz in the R_s compensation circuit of the Multiclamp amplifier. Reticulospinal axons were recorded using sharp electrodes filled with 3 M KCl intracellular solution. Action potentials were elicited in the recorded reticulospinal axons by using an extracellular glass suction electrode placed on the ventromedial part of the spinal cord. These axons were identified by their medial position and fast conduction velocity.

Fictive locomotion in vitro.

Fictive locomotion was induced by chemical activation of the mesencephalic locomotor region (MLR) in brain stem-spinal cord preparations from transformer animals (Gariepy et al. 2012; Sirota et al. 2000; Smetana et al. 2010). The chemical stimulation consists of the excitatory amino acid l-glutamate (1 mM; Sigma-Aldrich) diluted in Ringer solution at pH 7.6, pressure-ejected through a glass micropipette using a Picospritzer (General Valve). Fast Green (Sigma-Aldrich) was also included in the ejection solution to monitor the location and diffusion through the tissue. The resulting activity was recorded from ventral roots on both sides of the spinal cord. A brief (1- to 3-s puff) activation of MLR induced long-lasting locomotor activity. In these experiments a double-pool recording chamber was used to allow separate perfusion of the brain stem and spinal cord. Apamin (diluted in extracellular solution as described above with the addition of 0.1% bovine serum albumin; Sigma-Aldrich) was applied only to the spinal cord to determine the effect of blockade of SK channels on the locomotor activity. Fast green (Sigma-Aldrich) was included with apamin in the perfusion solution to monitor any seepage into the brain compartment.

Ca²⁺ imaging.

Presynaptic Ca²⁺ transients were recorded using line-scan confocal imaging with a modified Bio-Rad MRC-600 confocal microscope. The photomultiplier outputs were amplified with low-noise current amplifiers (Stanford Instruments) and digitized to 12 bits (National Instruments; custom software written under Matlab, The MathWorks). The scan head mirrors were driven though the MR-600 scan head amplifiers with the same custom software. This software is available online (http://alford.bios.uic.edu/Research/software.html). Reticulospinal axons were recorded in whole mount spinal cord as described above using sharp electrodes filled with KCl (1.5 M) and fluo 5F (2.5 mM). The dye affinity to Ca²⁺ (K_d = 3.1 μM) was determined at 10°C using Ca²⁺ standards (Invitrogen) in the same optical path as the experiments were performed. This relatively low-affinity dye was chosen to avoid saturation during repetitive stimulation. Dye was applied to the axons by pressure pulses through the recording microelectrode, after which the axon was left unstimulated for 10 min to allow dye diffusion. Action potentials were evoked in reticulospinal axons by brief depolarizing pulses (2 ms) applied through the microelectrode. Imaging was performed by line scanning (500 Hz) over presynaptic terminals identified as discrete points of Ca²⁺ entry (Takahashi et al. 2001).

Data analysis.

Data were recorded using pClamp9 software (Molecular Devices). The analysis was performed using Clampfit (Molecular Devices) and Origin (Microcal Software). A single standard exponential by Clampfit was used to fit decay time constants. Data are means ± SE, and n indicates the number of neurons tested. All the experiments were analyzed by paired t-test. In brain stem evoked fictive locomotion experiments, statistical significance is indicated both for activity evoked repetitively in each animal analyzed (paired t-test) and also for the means obtained across groups (control and apamin) (Student's t-test).

RESULTS

Blockade of SK channels decreases the frequency of MLR-induced locomotor rhythm.

A brain stem-spinal cord preparation was used to test the effect of blockade of SK channels on fictive locomotion induced by MLR stimulation (Fig. 1A). A brief pressure ejection of glutamate into the MLR induced a long episode of locomotor activity (Fig. 1B). Application of apamin (5 μM) selectively to the spinal cord produced a significant decrease in the frequency of the locomotor rhythm that then became irregular (Fig. 1C) and recovered after washout (Fig. 1D). The frequency of the first 5 bursts of the locomotor episode, which were the most reliable to evoke, was 1.89 ± 0.35 Hz in control, was significantly (P < 0.05) decreased to 0.86 ± 0.18 in apamin, and recovered to 1.61 ± 0.27 Hz after 30 min of wash (n = 5). Similarly, the frequency of the activity throughout the episodes was significantly reduced in apamin in each of the five animals tested (P < 0.01), as well as the mean frequency of activity measured across all five animals (apamin reduced mean frequency to 35 ± 10% of control, P < 0.01). In addition, the variance of the burst frequency weighted for the change in mean burst frequency (measured as μ²/σ²) was increased (P < 0.05) for four to six sequential episodes obtained in control and then in apamin in three of five animals. In the remaining two animals the change in μ²/σ² was not significant. This effect was also significant when the means of these values obtained from all five animals investigated were compared (P < 0.05). This irregularity of left/right coordination in apamin (Fig. 1C) reflects prolonged excitation of both excitatory and inhibitory interneurons, motoneurons, and descending reticulospinal neurons (see Fig. 6). These results show that SK channels contribute to setting the excitability level of the spinal networks and determine the baseline locomotor frequency. Furthermore, these results demonstrate that SK channels play an important role in coordinating excitation and inhibition to produce regular output of the spinal network.

Fig. 1.

Blockade of apamin-sensitive Ca²⁺-activated K⁺ (SK) channels decreases the burst frequency. A: experimental setup with a double-pool chamber. MLR, mesencephalic locomotor region; vr, ventral root. B: pressure injection of l-glutamate (Glu; 1 mM) to the MLR induces a long episode of locomotor activity. C: application of apamin (5 μM) decreases the locomotor burst frequency. D: burst frequency recovers after washout of apamin.

Fig. 6.

Presynaptic SK channels control the action potential waveform in axon terminals. A: apamin (1 μM) significantly increases the width of a single action potential in presynaptic axon terminals. B: graph showing the increase in the action potential width induced by apamin. C: superimposed 1st and 5th action potentials recorded in a presynaptic axon terminal in control and in apamin. D: graph showing the lack of activity-dependent change in the width of presynaptic action potentials in the presence of apamin. Traces in A and C are averages of 30 sweeps.

Activation of SK channels by NMDA in dissociated spinal neurons.

It had been previously thought that SK channels are activated only by Ca²⁺ influx via voltage-activated Ca²⁺ channels and exert their effect in the spinal locomotor networks by regulating the afterhyperpolarization and pacemaker oscillations (Cangiano et al. 2002; El Manira et al. 1994; Hill et al. 1992; Meer and Buchanan 1992; Wallén et al. 1989). NMDA receptors represent another major source of Ca²⁺ that can activate SK channels. To examine if postsynaptic SK channels in spinal neurons can be activated by Ca²⁺ influx via NMDA receptors, we tested the effect of apamin on NMDA-induced current in isolated spinal cord neurons, which include both motoneurons and interneurons (El Manira and Bussières 1997; Nanou and El Manira 2007). Neurons were held at a membrane potential of −60 mV and perfused with Mg²⁺-free extracellular solution. Application of NMDA (200 μM; 4 s) induced an inward current with average peak amplitude of 1.3 ± 0.2 nA that decayed to 0.7 ± 0.1 nA at the end of application (n = 23; Fig. 2A). Blockade of SK channels with apamin (1 μM) slowed the decay of the NMDA receptor current without affecting its peak amplitude (1.2 ± 0.1 nA). The amplitude of the current at the end of the NMDA application was 0.9 ± 0.1 nA in apamin compared with 0.7 ± 0.1 nA in control (Fig. 2A). The calculated ratio of the NMDA-induced current amplitude (offset/onset) was 0.5 ± 0.03 in control and increased to 0.7 ± 0.03 by apamin (P < 0.001; Fig. 2B). Similar experiments were performed with longer application of NMDA (30 s), which resulted in an inward current followed by an outward current on termination of the NMDA application (Fig. 2C). The average density of the NMDA-induced outward current was 58.1 ± 8.4 pA/pF in control and was reduced to 6.8 ± 2.4 pA/pF by apamin (P < 0.001; n = 5; Fig. 2, C and D). Similarly, the decay of the NMDA current was significantly slower in the presence of apamin (P < 0.05; Fig. 2C). The calculated ratio of the NMDA-induced current amplitude (offset/onset) was 0.2 ± 0.02 in control and increased to 0.4 ± 0.08 in apamin (P < 0.05). These results suggest that Ca²⁺ influx through NMDA receptors can activate SK channels, resulting in a faster decay of the NMDA receptor current.

Fig. 2.

Ca²⁺ influx via NMDA receptors activates SK current in dissociated spinal neurons. A: application of apamin (1 μM) slows down the decay of the inward current induced by NMDA (200 μM; 4 s; horizontal bar). B: graph showing the ratio of the current amplitude at the offset and onset of NMDA application. ***P < 0.001, significant difference. C: longer application of NMDA (200 μM; 30 s; horizontal bar) induces an inward current followed by an outward current. Application of apamin (1 μM) blocks the outward current and slows down the decay of the NMDA-induced inward current. D: average density of the NMDA-induced outward current in control and in apamin. ***P < 0.001, significant difference.

Synaptic activation of postsynaptic SK channels.

To determine if the coupling between Ca²⁺ influx via NMDA receptors and SK channels seen in dissociated neurons can be recruited synaptically, we tested the effect of apamin on NMDA receptor-mediated EPSCs and EPSPs. In these experiments, NMDA receptor-mediated EPSCs/EPSPs were induced in spinal neurons by extracellular stimulation of multiple reticulospinal axons. The preparations were perfused with a Mg²⁺-free extracellular solution containing inhibitors for glycine (strychnine, 5 μM) and AMPA (GYKI-52466, 30 μM) receptors (Fig. 3, A and B). Stimulation of reticulospinal axons induced mainly a chemical EPSC, although some neurons also had an electrical component (Fig. 3A). The chemical EPSCs were blocked by the NMDA receptor antagonist (AP-5, 50 μM) leaving intact the electrical component (n = 6; Fig. 3, A–C; the electrical component persists in cadmium, see Cochilla and Alford 1997; Krieger et al. 1999).

Fig. 3.

Synaptic activation of NMDA receptors in the spinal cord. A: stimulation of reticulospinal axons by a glass electrode placed on the ventral surface of the spinal cord induced an excitatory postsynaptic current (EPSC) with a chemical and an electrical component in the presence of glycine (strychnine, 5 μM) and AMPA (GYKI-52466, 30 μM) receptor antagonists. Application of 2-amino-5-phosphonopentanoate (AP-5; 50 μM) decreases the amplitude of the NMDA-induced EPSC, leaving only the electrical component. Traces are averages of 30 sweeps. V_h, holding potential. B: subtraction of the current recorded in the presence of AP-5 from that in control reveals the NMDA-induced EPSC. C: graph showing the amplitude of EPSCs in control and in AP-5 recorded in 6 different preparations.

The fact that NMDA receptor-mediated EPSCs/EPSPs can be isolated pharmacologically allows testing the possibility of coupling between Ca²⁺ influx via NMDA receptors and activation of SK channels and its impact on synaptic response. Indeed, blockade of SK channels with apamin (1 μM) significantly increased the amplitude of NMDA receptor-induced EPSPs from 3.7 ± 0.3 to 4.8 ± 0.4 mV (P < 0.001; n = 12; Fig. 4, A and B). This enhancement was caused solely by a change of the decay time constant of the synaptically induced NMDA receptor current that increased from 112.2 ± 6.3 ms in control to 161.5 ± 9.4 ms in apamin (P < 0.001) without affecting the peak EPSC amplitude (P > 0.05, n = 16; Fig. 4, C and D). Subtraction of the EPSC in apamin from that in control showed that the increase in the decay time constant is due to a block of an outward current (Fig. 4C). Apamin did not alter the resting membrane potential or conductance in the recorded neurons. These results suggest that Ca²⁺ influx via synaptically activated NMDA receptors can activate postsynaptic SK channels that limit the amplitude of the synaptic response by controlling the decay time constant of the NMDA receptor current.

Fig. 4.

Activation of SK channels by Ca²⁺ influx via synaptically activated NMDA receptors. A: application of apamin (1 μM) enhances the amplitude of the NMDA receptor-induced excitatory postsynaptic potentials (EPSPs) by stimulation of reticulospinal axons in neurons in the spinal cord. Traces are averages of 30 sweeps. B: graph showing the amplitude of the evoked EPSPs in control and in apamin from 12 different neurons. C: application of apamin slows down the decay time constant (tau) of NMDA receptor-induced EPSCs. Bottom trace shows the current blocked by apamin. Traces are averages of 30 sweeps. D: graph showing the decay time constant of the NMDA receptor-mediated EPSC in control and in apamin from 16 different neurons.

SK channels are located within close proximity to NMDA receptors.

To determine if the apamin-induced change in the NMDA-mediated synaptic current, described above, is indeed the result of activation of postsynaptic SK channels and to assess the proximity of these channels to the Ca²⁺ source, the fast (BAPTA) and slow (EGTA) Ca²⁺ chelators were used. BAPTA is considerably more effective in preventing diffusion of free Ca²⁺ away from its site of entry at the plasma membrane. BAPTA interferes with Ca²⁺-dependent processes located within ∼20–50 nm from the Ca²⁺ source (Ca²⁺ nanodomains), whereas those affected by EGTA are located at distances beyond 50 nm (Augustine et al. 2003; Fakler and Adelman 2008; Neher 1998). In neurons dialyzed with BAPTA, apamin (1 μM) had no effect on the decay time constant of the NMDA receptor-induced EPSC evoked by stimulation of descending reticulospinal axons (time constant was 117.9 ± 4.9 ms compared with 115.8 ± 4.9 ms in control; n = 8; P > 0.05; Fig. 5, A and B). In neurons dialyzed with EGTA, apamin was still able to increase the decay time constant of the EPSCs to 158.4 ± 13.3 ms from 119.6 ± 8.8 ms in control (n = 5; P < 0.05; Fig. 5, C and D). Hence, chelating Ca²⁺ with BAPTA, but not with EGTA, abolished the effect of apamin on the decay time constant of NMDA receptor-induced EPSCs. These data further support that SK channels are activated by Ca²⁺ influx via NMDA receptors and show that these two proteins are located in close proximity postsynaptically.

Fig. 5.

SK channels are located in close proximity with NMDA receptors. A: chelating intracellular Ca²⁺ with BAPTA (10 mM) prevents the activation of SK channels and blocks the effect of apamin on the decay time constant of the NMDA receptor-mediated EPSC. Traces are averages of 30 sweeps. B: graph showing the lack of an effect of apamin on the decay time constant of NMDA receptor-induced EPSCs in neurons dialyzed with BAPTA. C: the slow Ca²⁺ chelator EGTA (10 mM) does not prevent the apamin-induced increase of the decay time constant of NMDA-mediated EPSC. Bottom trace shows the current blocked by apamin in a neuron dialyzed with EGTA. Traces are averages of 30 sweeps. D: graph showing the apamin-induced increase in the decay time constant of NMDA receptor-mediated EPSCs in the presence of EGTA.

Presynaptic SK channels shape the action potential waveform.

If SK channels are also located on presynaptic axons, their activation by Ca²⁺ influx via voltage-activated Ca²⁺ channels may restrict the duration of action potentials in axon terminals. To test this possibility, intracellular recordings were made from presynaptic reticulospinal axons and the effect of apamin on the action potential waveform was examined. Blockade of SK channels by apamin increased the duration of a single action potential from 2.6 ± 0.1 to 2.8 ± 0.1 ms (P < 0.05; n = 6; Fig. 6, A and B). The change in the action potential duration induced by apamin did not display any activity dependence during stimulation trains. The increase in duration induced by apamin was similar between the first and the fifth action potentials of a train of 5 stimuli at 20 Hz (n = 6; Fig. 6, C and D). These results show that SK channels are present presynaptically and act to limit the duration of action potentials in axon terminals.

Presynaptic SK channels limit synaptic transmission and Ca²⁺ dynamics.

To determine if presynaptic SK channels can affect synaptic transmission, spinal neurons were dialyzed with BAPTA to prevent postsynaptic synaptic plasticity. In these experiments, NMDA receptors were blocked with AP-5 (50 μM) to allow the isolation of AMPA receptor-mediated EPSCs elicited by stimulation of multiple reticulospinal axons with an extracellular glass electrode. In 11 of 13 neurons, blockade of SK channels with apamin (1 μM) significantly increased the amplitude of single AMPA receptor-mediated EPSCs from 59.2 ± 7.9 to 72.5 ± 8.2 pA (P < 0.01; Fig. 7, A and B) at −60 mV without affecting its decay time constant (7.9 ± 0.3 ms in control; 7.9 ± 0.3 ms in apamin; P > 0.05; n = 13).

Fig. 7.

Presynaptic SK channels regulate synaptic transmission. A: recording of AMPA-mediated EPSC induced by stimulation of reticulospinal axons in a spinal neuron dialyzed with BAPTA (10 mM) recorded in the presence of the NMDA (AP-5, 50 μM) and glycine (strychnine, 5 μM) receptor blockers. Application of apamin (1 μM) significantly increases the amplitude of AMPA receptor-mediated EPSC without affecting its decay time constant. B: graph showing the significant increase in the amplitude of synaptically induced AMPA receptor EPSCs between control and apamin. **P < 0.01, significant difference. C: apamin increases the amplitude of AMPA receptor-mediated EPSCs induced by a train of 5 stimuli at 20 Hz in a neuron dialyzed with BAPTA and with NMDA and glycine receptors blocked with AP-5 and strychnine, respectively. D: graph showing activity-dependent increase in the amplitude of AMPA-EPSC induced by a train of 5 stimuli at 20 Hz. Traces in A and C are averages of 30 sweeps.

We then tested if Ca²⁺ accumulation in presynaptic terminals during repetitive firing of action potentials leads to increased activation of SK channels and restricts activity-dependent changes. Multiple EPSCs were elicited by repetitive stimulation of presynaptic reticulospinal axons (5 stimuli at 20 Hz). Apamin increased the amplitude of the AMPA receptor-induced EPSCs in an activity-dependent manner with the maximum increase reached at the third EPSC (Fig. 7, C and D). These results show that presynaptic SK channels limit the strength and alter the dynamics of synaptic transmission.

Although the action potential waveform did not display any activity-dependent change in apamin (see Fig. 6, C and D), the activity-dependent increase in the EPSC amplitude could still be the result of presynaptic Ca²⁺ dynamics due to residual Ca²⁺ remaining in the terminal (Katz and Miledi 1968). To test this possibility, we performed Ca²⁺ imaging experiments to monitor the change in Ca²⁺ in presynaptic terminals using a low-affinity dye (fluo 5F). Repetitive stimulation of single presynaptic axons (5 action potentials, 20 Hz) induced Ca²⁺ transients (Fig. 8A) in which single action potential-dependent Ca²⁺ transients are resolved and are known to correlate with presynaptic active zones and transmitter release (Photowala et al. 2005; Takahashi et al. 2001). Blockade of SK channels with apamin increased the amplitude of the Ca²⁺ transient (Fig. 8, A and B). The increase in the Ca²⁺ transient was activity dependent, with the first response showing no significant increase (6 ± 6%; n = 4), whereas the fifth response significantly increased by 38 ± 15% (P < 0.05; n = 4). The change in the Ca²⁺ transient induced by apamin mirrors the activity-dependent increase in the EPSC amplitude. These results thus show that presynaptic SK channels can play an important role in controlling the dynamics of presynaptic Ca²⁺ and synaptic transmission.

Fig. 8.

Apamin enhances the amplitude of presynaptic Ca²⁺ transients evoked by repetitive stimulation of reticulospinal axons. A: 5 action potentials (20 Hz) evoked a series of 5 summating Ca²⁺ transients recorded using confocal line scanning at the plasma membrane of a reticulospinal axon. Top image was recorded in control and bottom image after addition of apamin to the superfusate. Note that Ca²⁺ enters the axons at discrete regions associated with active zones. B: data from the regions between the white lines in A were integrated and expressed as ΔF/(F + 1), where F is prestimulus fluorescence. After addition of apamin, the Ca²⁺ signal shows an enhanced summation during the stimulus train.

DISCUSSION

A major finding of this study is that pre- and postsynaptic SK channels are activated by different Ca²⁺ sources to endow the synapse with a negative feedback mechanism that limits synaptic strength. Presynaptic SK channels are coupled to Ca²⁺ influx via voltage-activated Ca²⁺ channels. Their activation limits the action potential duration and Ca²⁺ accumulation in the presynaptic terminal. Postsynaptic SK channels interact with NMDA receptors, potentially located within nanodomains. Their activation determines the time course of the synaptic NMDA receptor current and thus sets the amplitude of the resulting membrane potential depolarization. Previous studies reported conflicting results of apamin on locomotor activity induced pharmacologically with glutamate receptor agonists (El Manira et al. 1994; Hill et al. 1992; Meer and Buchanan 1992). Our present results show that SK channels play a role in setting the baseline frequency of the locomotor activity induced by stimulation of the MLR in lamprey. Blockade of these channels decreases the burst frequency and increases the irregularity of the rhythm. The change in the locomotor frequency can be explained by an increase in NMDA-induced synaptic transmission that leads to long-lasting depolarization in spinal neurons (see El Manira et al. 1994). In addition to mediating synaptic transmission, activation of NMDA receptors also induces intrinsic pacemaker oscillations in spinal neurons. The termination of the depolarization phase of these oscillations is dependent on SK channels, and a blockade of these channels increases the duration of the up state of the oscillation. Prolonged plateau periods in excitatory and inhibitory interneurons in addition to motor neurons in apamin may nonlinearly disrupt the balance of excitation and inhibition driving the locomotor rhythm. Additionally, enhancement of reticulospinal excitatory drive in apamin will also contribute to the imbalance of excitation and inhibition in spinal interneurons and motor neurons controlling spinal output. The net effect may underlie the change in the locomotor frequency and irregularity induced by apamin. Thus it seems that the coupling of SK channels with NMDA channels postsynaptically and with voltage-activated Ca²⁺ channels presynaptically restrains synaptic strength within an appropriate range to set the baseline burst frequency. Blockade of SK channels results in a deviation from this set point, leading to profound changes in the excitability of the locomotor network.

SK channels have been shown to mediate the afterhyperpolarization following the action potential and control spike frequency adaptation in neurons in various regions of the CNS (Bond et al. 1999). In the spinal cord they also have been shown to set the duration of pacemaker oscillations induced by NMDA (Bond et al. 1999; El Manira et al. 1994). However, their role in controlling synaptic transmission has been elusive. Although the possible interaction between NMDA and SK channels was previously examined, no direct evidence supporting such an interaction was found because it was reported that apamin does not affect the amplitude of the NMDA-mediated EPSP (Cangiano et al. 2002). Our present results show that such an interaction exists in the spinal cord. This apparent discrepancy could be due to difference in the experimental setup. In the earlier study, NMDA receptor-mediated EPSPs were induced in motoneurons by stimulation of single presynaptic reticulospinal axons using paired recordings (Cangiano et al. 2002). It is possible that the Ca²⁺ increase induced by activation of NMDA receptors at a single synapse does not reach sufficient levels to activate postsynaptic SK channels. In our present study, stimulation of multiple presynaptic reticulospinal axons was used to reveal the coupling between Ca²⁺ influx via NMDA receptors and SK channels and its effect on the decay time constant of EPSCs and the amplitude of EPSPs. A postsynaptic effect of SK channels on synaptic transmission has been shown in hippocampus and amygdala (Bloodgood and Sabatini 2007; Faber et al. 2005; Fakler and Adelman 2008; Lujan et al. 2009; Ngo-Anh et al. 2005).

The close interaction between synaptically induced Ca²⁺ influx through NMDA receptors and SK channels is supported by the apamin-induced outward current suppression. The SK current impacts both the time course of the synaptic current and the amplitude of the resulting depolarization. Dialysis with BAPTA, which interferes with Ca²⁺-dependent processes located within ∼20–50 nm from the Ca²⁺ source (Augustine et al. 2003; Fakler and Adelman 2008; Neher 1998), also demonstrates the site of Ca²⁺ entry via NMDA receptors is closely coupled to SK channels. Our results from isolated neurons in culture also corroborate such an interaction by showing that activation of NMDA receptors not only produces the characteristic inward current but also induces an apamin-sensitive outward current. It should be noted that postsynaptic SK channels can also be activated by Ca²⁺ influx via voltage-activated Ca²⁺ channels, allowing them to act as a processing interface of both electrical activity and synaptic inputs.

Another important finding of this study is the parallel presynaptic negative feedback loop involving activation of SK channels by Ca²⁺ influx during action potentials. Activation of these presynaptic SK channels determines the duration of presynaptic spikes and the associated Ca²⁺ dynamics. The potentiation of synaptic transmission was more pronounced when monitored on isolated AMPA than on NMDA synaptic responses. This is likely to be due the difference in affinity of these receptors to glutamate. The high affinity of NMDA receptors could preclude any additional increase of glutamate release, which would be more apparent on the low-affinity AMPA receptors. Thus SK channels on both sides of the synapse are coupled to different Ca²⁺ sources and act in a complementary manner to limit the short-term plasticity and activity-dependent changes of synaptic transmission. The existence of a close interaction between Ca²⁺-activated K⁺ channels and Ca²⁺ influx in presynaptic terminals has been shown at the frog neuromuscular junction. At this synapse, Ca²⁺-activated K⁺ channels with large and intermediate conductance that are sensitive to charybdotoxin, but not SK channels, regulate transmitter release by limiting the duration of the action potential (Robitaille and Charlton 1992; Robitaille et al. 1993).

In a previous study, we showed that AMPA receptors are tightly coupled with Na⁺-activated K⁺ (K_Na) channels (Nanou and El Manira 2007, 2010). The Na⁺ influx through AMPA receptors activates Slack channels and changes the decay time constant of synaptic current and the amplitude of the synaptic potential (Nanou et al. 2008). These data together with our present findings show that ion influxes through ionotropic glutamate receptors do not merely serve as a charge carrier, but they also mediate activation of specific K⁺ channels that underlie a negative feedback loop to prevent the synapse from operating at full strength, thereby increasing its dynamic range. Both SK and K_Na channels possess distinct regulatory sites that may be targeted by different modulatory systems, thus providing the synapse with a powerful mechanism for tuning and adapting its strength for an optimal output. The physiological outcome of a direct coupling between Ca²⁺ entry through NMDA receptors and subsequent SK activation is substantially different from the coupling between voltage-operated Ca²⁺ channels and SK. The latter is essentially constitutive in nature, coupling a depolarization to subsequent afterhyperpolarization. In contrast, coupling the NMDA receptor to SK requires the synaptic activation of the NMDA receptor, linking late outward currents to only synaptic locations activated during the behavior in question. In this case we describe a mechanism involved in locomotion, but the result has implications also, for example, to local control of membrane potential (Hosy et al. 2011).

The generation of the locomotor rhythm is the result of both synaptic interactions between spinal network neurons and their intrinsic properties. SK channels have been mostly linked with changing the firing properties in many locomotor networks; however, their role in controlling synaptic transmission has remained undetermined. In the present study we demonstrate a prominent impact of SK channels postsynaptically via their interaction with NMDA receptors, and presynaptically we define additional sites of action of these channels. The multitude of effects of SK channels and their coupling to various sources of Ca²⁺ influx into neurons endow them an important function in shaping the operation of the locomotor network.

GRANTS

This work was supported by the Swedish Research Council and Karolinska Institutet (to A. El Manira) and National Institute of Neurological Disorders and Stroke Grant R01 NS 052699 (to S. Alford).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.N. and A.E.M. conception and design of research; E.N., M.H.A., and S.A. performed experiments; E.N., M.H.A., S.A., and A.E.M. analyzed data; E.N., M.H.A., S.A., and A.E.M. interpreted results of experiments; E.N., M.H.A., S.A., and A.E.M. prepared figures; E.N. and A.E.M. drafted manuscript; E.N., M.H.A., S.A., and A.E.M. edited and revised manuscript; E.N., M.H.A., S.A., and A.E.M. approved final version of manuscript.