Soomin C. Song, Joseph A. Beatty, Charles J. Wilson. J Neurophysiol. 116(4):1752-1764.

Striatal low-threshold spiking (LTS) interneurons spontaneously transition to a depolarized, oscillating state similar to that seen after sodium channels are blocked.  In the depolarized state, whether spontaneous or induced by sodium channel blockade, the neurons express a 3 – 7 Hz oscillation and membrane impedance resonance in the same frequency range.  The membrane potential oscillation and membrane resonance are expressed in the same voltage range (> -40 mV).  We identified and recorded from LTS interneurons in striatal slices from a mouse that expressed GFP under the control of the neuropeptide Y promoter.  The membrane potential oscillation depended on voltage-gated calcium channels.  Antagonism of CaV1 reduced the amplitude of the oscillation while blockade of CaV2.2 reduced the frequency.  Both calcium sources activate a calcium-activated chloride current (CaCC), the blockade of which abolished the oscillation.  Blocking any of these three channels abolished the membrane resonance.  Immunohistochemical staining indicated ANO2, and not ANO1, as the CaCC source.  Biophysical modeling showed that CaV1, CaV2.2 and ANO2 are sufficient to generate a membrane potential oscillation and membrane resonance, similar to that in LTS interneurons. LTS interneurons exhibit a membrane potential oscillation and membrane resonance that are both generated by CaV1 and CaV2.2 activating ANO2. They can spontaneously enter a state in which the membrane potential oscillation dominates the physiological properties of the neuron.  

Figure 1.  Striatal LTS interneurons spontaneously transition to a depolarized, oscillating state.  (A, left panel) An example of a striatal LTS interneuron as seen in a sagittal slice, identified by NPY-GFP fluorescence.  (A, right panel) Visualization of the same neuron filled with Alexa Fluor 594 after whole cell access.  Scale bar applies for both panels of Figure 1A.  (B, top trace) An example of a cell-attached recording of a silent LTS interneuron.  (B, bottom trace) Following perforated access to the neuron by gramicidin, a depolarized and oscillating membrane potential is revealed.  (C) LTS interneurons can spontaneously transition back and forth between silence and autonomous action potential firing in cell-attached recordings.  Note the lack of bursts when firing begins, indicating that the neurons are not being released from hyperpolarization.  (D, top trace and protocol) Neurons in the depolarized, oscillating state can be induced to fire by prolonged hyperpolarizations, followed by stepwise reduction of the hyperpolarizing current.  (D, bottom trace) LTS interneurons can maintain autonomous firing for prolonged periods after the transition from the depolarized, oscillating state.  This recording is 10 minutes after the above protocol was applied.  (E, top traces) Brief depolarization or hyperpolarization cannot induce a transition to sustained firing.  Note that the neuron can still fire action potentials during the LTS burst (black trace).  (E, bottom traces) After inducing a transition into sustained firing, LTS interneurons show the stereotypical electrophysiological properties of normal LTS interneurons.

Figure 2.  Membrane potential oscillations in striatal LTS interneurons are calcium dependent.  (A) An example of TTX application to an LTS interneuron in the depolarized, oscillating state.  (B) The average membrane potential was significantly hyperpolarized after TTX application.  (C) The frequency of the membrane potential oscillation was also significantly reduced by TTX.  (D) TTX had no effect on the amplitude of the membrane potential oscillation.  (E) Application of 1 μM TTX blocked autonomous action potentials and produced a depolarized, oscillating state.  (F, top trace) An increased magnification of the membrane potential oscillation shows the characteristic spikelet-like waveform of the oscillation and the depolarized membrane potential.  (F, bottom trace) Application of 400 µm cadmium blocked the membrane potential oscillations and caused a repolarization of the membrane potential.  (G) The average membrane potential was significantly hyperpolarized after the application of cadmium.  (H) Membrane variance (our measure of the amplitude of the oscillation) was significantly reduced in the presence of cadmium.  Error bars are SE. (* = p < 0.05, *** = p < 0.005)

Figure 3.  Membrane potential oscillations require L- and N-type calcium currents.  (A1) An example of the membrane potential oscillation unaffected by the application of 100 nM ω-agatoxin.  (A2, A3) The median oscillation frequency and membrane variance did not change after P/Q-type currents were blocked.  (B1) L-type calcium currents contribute to the membrane potential oscillations.  (B2) The median oscillation frequency was unchanged after 5 μM isradipine was applied, (B3) but the membrane variance was significantly reduced.  (C1) Blockade of N-type calcium currents with 1 μM ω-conotoxin GVIA disrupted the membrane potential oscillation frequency and shape.  The oscillation waveform changed from a series of spikelets to broad plateaus and valleys.  (C2) The median oscillation frequency was significantly reduced.  (C3) The membrane variance did not change after drug application.  Error bars are SE.  Scale bars apply to all traces.  (* = p < 0.05, *** = p < 0.005)

Figure 4.  Calcium-activated chloride currents, not potassium currents, are necessary for the membrane potential oscillation.  (A) Potassium currents were tested for participation in the membrane potential oscillations.  (B) Neither application of 50 nM iberiotoxin to block BK currents nor (C) 100 nM apamin to block SK currents had a significant effect on the membrane potential oscillation.  (D) KCNQ channels, which can be modulated by calcium, were blocked with 10 μM linopirdine with no effect.  (E, showing 3 mM TEA application) General potassium current contribution to the membrane oscillation was tested with a range of TEA concentrations.  The concentrations tested were 1, 3, 6, 8 and 10 mM.  TEA caused a significant depolarization of the average membrane potential and increased oscillation amplitude.  (F) Application of 100 μM niflumic acid, a CaCC blocker, abolished the membrane oscillation.  (G) The membrane variance was significantly reduced by niflumic acid.  Error bars are SE.  Scale bars apply to all traces.  (* = p < 0.05)

Figure 5.  Expression of ANO2 in LTS interneurons.  (A, arrow) A putative LTS interneuron labeled by NPY-GFP is shown to be co-labeled for ANO2 expression.  A large proportion of other cells were also positively labeled for ANO2 expression in the striatum.  (B) An example of a putative NGF cell (star) seen near an LTS interneuron (arrow), both co-labeled for ANO2 expression.  Scale bar applies for (A) and (B).  Images were taken at a single plane.

Figure 6.  Membrane resonance in LTS interneurons requires calcium currents.  (A) Comparison of the voltage sensitivity of the membrane potential oscillation to (B) the membrane resonance.  The impedance curves show relative impedance, normalized by the input resistance measured at each holding potential.  (C) An example of the membrane response to a chirp protocol shown before and after the application of 400 μM cadmium.  (D) The holding current to maintain the membrane potential at -30 mV was significantly increased after cadmium application.  (E) The sample data depicts the change in the impedance curves when voltage-gated calcium channels were blocked.  The zero frequency impedance indicates the input resistance.  (F) The phase of the membrane response lost the positive phase-shift after cadmium application.  Error bars are SE.  (*** = p < 0.005)

Figure 7.  L-type currents, not P/Q-type currents, are necessary for membrane resonance.  (A1) The sample impedance curves and the input resistance showed no effect of blocking P/Q-type currents.  The zero frequency impedance indicates the input resistance.  (A2) The phase of the membrane response was also unaffected by ω-agatoxin.  (A2, inset) The holding current was not changed by ω-agatoxin application.  (B1) In contrast, application of 5 μM isradipine significantly reduced the impedance at all frequencies but did not significantly change the input resistance.  (B2) The positive phase-shift in the membrane response was reduced after isradipine application.  (B2, inset) Isradipine significantly increased the required holding current.  Error bars are SE.  Scale bars apply to all traces.  (* = p < 0.05)

Figure 8.  N-type calcium currents and calcium-activated chloride channels are necessary for membrane resonance.  (A1) The sample impedance curve shows a blockade of membrane resonance after the application 1 μM ω-conotoxin GVIA.  The input resistance (zero frequency impedance) significantly increased.  (A2) The phase advance in the membrane response was prevented after application of ω-conotoxin GVIA.  (A2, inset) The holding current did not change after blocking N-type calcium currents. (B1) Blocking CaCCs with 100 μM niflumic acid prevented expression of membrane resonance, reducing the peak in the impedance profile.  The input resistance (zero frequency impedance) did not change after the application of niflumic acid.  (B1) Niflumic acid prevented the phase advance in the membrane response.  (B1, inset) The holding current was not significantly altered after niflumic acid application.  Error bars are SE.  Scale bars apply to all traces.

Figure 9.  A simplified model for oscillations and resonance in striatal LTS cells.  (A) Equivalent circuit diagram for the model.  (B) Oscillation of the model with all channels intact, using the parameters shown in Table 1.  The oscillation frequency is about 5.5 Hz.  (C) The oscillation of the model produced when the N-type calcium channel is removed, corresponding to treatment with ω-conotoxin GVIA.  The oscillation frequency slowed to about 2.5 Hz, and the duty cycle increased.  (D) The oscillation obtained with all calcium channels intact but the voltage-gated potassium channel reduced by 80% to represent the effect of TEA.  The frequency of the oscillation was almost unchanged, but the amplitude increased.  (E) Abolition of the oscillation by removal of the L-type channel (with the N-type channel intact), corresponding to isradipine treatment, and also by removal of the calcium-dependent chloride channel.  (F) Resonance of the model in the intact condition and after removal of L-type calcium channels (isradipine), N-type calcium channels (ω-conotoxin GVIA).