Enhanced high-frequency membrane potential fluctuations control spike output in striatal fast-spiking interneurones in vivo

Abstract

Fast-spiking interneurones (FSIs) constitute a prominent part of the inhibitory microcircuitry of the striatum; however, little is known about their recruitment by synaptic inputs in vivo. Here, we report that, in contrast to cholinergic interneurones (CINs), FSIs (n = 9) recorded in urethane-anaesthetized rats exhibit Down-to-Up state transitions very similar to spiny projection neurones (SPNs). Compared to SPNs, the FSI Up state membrane potential was noisier and power spectra exhibited significantly larger power at frequencies in the gamma range (55-95 Hz). The membrane potential exhibited short and steep trajectories preceding spontaneous spike discharge, suggesting that fast input components controlled spike output in FSIs. Spontaneous spike data contained a high proportion (43.6 ± 32.8%) of small inter-spike intervals (ISIs) of <30 ms, setting FSIs clearly apart from SPNs and CINs. Cortical-evoked inputs had slower dynamics in SPNs than FSIs, and repetitive stimulation entrained SPN spike output only if the stimulation was delivered at an intermediate frequency (20 Hz), but not at a high frequency (100 Hz). Pharmacological induction of an activated ECoG state, known to promote rapid FSI spiking, mildly increased the power (by 43 ± 55%, n = 13) at gamma frequencies in the membrane potential of SPNs, but resulted in few small ISIs (<30 ms; 4.3 ± 6.4%, n = 8). The gamma frequency content did not change in CINs (n = 8). These results indicate that FSIs are uniquely responsive to high-frequency input sequences. By controlling the spike output of SPNs, FSIs could serve gating of top-down signals and long-range synchronisation of gamma-oscillations during behaviour.

Figure 1. Electrophysiological and morphological characteristics of three classes of striatal neurones in vivo

A, membrane potential response of a FSI to linear current steps. Note the immediate and subsequently stuttering action potential discharge as well as the subthreshold membrane potential fluctuations in between. B, merged Z-projections from confocal image stacks of the same biocytin-filled neurone. The dendrites are smooth. C, membrane potential response of a SPN to linear current steps. Note the delayed action potential discharge. Inset shows the current–voltage relation; the input resistance was derived from the slope of the regression line. D, merged Z-projections from confocal image stacks of the same biocytin-filled neurone. The dendrites are densely studded with spines. E, electrophysiological characteristics of a CIN. Note the I_H-dependent sag response to hyperpolarising current injections. F, grandmeans of the spike waveform (top) and the first derivative (bottom) are shown for SPNs (blue; n = 74), FSIs (red; n = 9) and CINs (green; n = 12). Lighter colours indicate the SEM in all panels. G, plot of wavelength of the mean AP derivative versus ratio between positive and negative peak showing that the FSI spike waveforms were clearly different from other neurons.

Figure 3. A comparison of slow and fast membrane potential fluctuations in SPNs and FSIs recorded during ECoG slow-wave activity

A, measurements of parameters of Up state transitions on the smoothed membrane potential recording (3rd order Savitzky–Golay, 30 ms window). For definition of parameters see Methods. B and C, three Up states aligned to the transition threshold in the same FSI (B) and SPN (C) as in Fig. 2. Action potentials have been removed from traces for the calculation of the membrane potential power spectra. D, mean log-scale power spectra of the membrane potential over the whole recording segment for FSIs (red) and SPNs (blue; both: n = 9). Note the non-linear decay in FSIs in this chart and the cross-over of the spectra from FSIs and SPNs at ∼30 Hz (dashed line). E and F, the mean normalized time-resolved power spectrum relative to the Up state transition are shown for the same neurons as in B and C. The grey coding is according to a linear scale (right), where 1 represents the average power for the respective frequency band. G, mean Up state transition aligned to the transition threshold in FSIs (red) and SPNs (blue; both: n = 9). Lighter shades indicate SEM. Upper panel, FSIs tended to larger spike rates during the Up state, while membrane potential trajectories during the transition were very similar between FSIs and SPNs (central panel). Lower panel, the mean time-resolved power spectrum from FSIs normalized relative to SPNs. The colour coding according to log-scale is shown on the right. FSIs exhibited larger fast membrane potential oscillations than SPNs specifically during the Up state (indicated by the white stars, P < 0.01). H, spike-triggered average of the membrane potential (top), the slope (middle), and the standard deviation of the slope (bottom) show that fast membrane potential fluctuations precede spike discharges in FSIs.

Figure 2. Spontaneous activity patterns of three classes of striatal neurones during urethane-induced slow-wave ECoG activity in vivo

A–C, segments of intracellular and simultaneous ECoG recordings (left), membrane potential distribution over the whole 90 s recording (top right), and cross-correlograms between intracellular and extracellular ECoG activity (bottom right) for a representative FSI (A), SPN (B) and CIN (C). Cardiovascular artefacts are evident in the ECoG traces. D, mean distribution of interspike intervals (ISI) for all three neurone classes. Note the bimodal distribution in SPNs (blue; n = 74) and FSIs (red; n = 9) in contrast to the unimodal distribution at intermediate ISIs in CINs (green; n = 12). E, mean distribution of CV2 values. Note the high proportion of small CV2 values in CINs and the high proportion of high CV2 values in SPNs. F, dependence of spike rate on the membrane potential depolarisation during the Up state in SPNs and FSIs.

Figure 4. Entrainment of striatal neurones by repetitive synaptic stimulation

A, example traces of responses to stimulation of the contralateral motor cortex at 100 Hz or 20 Hz in a FSI and two SPNs. Left inset shows the mean PSP evoked by a single stimulation during the Down state. Central trace shows a response to repetitive stimulation and concomitant intracellular current injection. Right inset shows the spike response in six repetitions. The scale is the same for all three examples. For clarity, stimulus artefacts have been truncated. B, grand mean of the evoked PSP in four FSIs and SPNs recorded in the same or a subsequent experiment. Lighter traces indicate the SEM. C, histograms of ISIs evoked by repetitive synaptic stimulation. Note the absence of a peak at 10 ms, when SPNs were stimulated at 100 Hz, and the presence of a peak at 50 ms, when stimulated at 20 Hz.

Figure 5. Effects of pharmacologically induced ECoG activation on fast membrane potential fluctuations in three classes of striatal neurones

A, immediate effects of the BIC ejection onto the membrane potential and spiking of a FSI. The time of the ejection is indicated. Note the dramatically increased spike frequency as soon as the ECoG activity loses the dominant slow wave oscillations. For clarity, cardiovascular artefacts have been removed from the ECoG trace by low-pass filtering (cutoff at 45 Hz). B, 1.5 s long membrane potential traces before and after BIC for the same neuron. Right inset shows the effect on the mean power spectrum of the membrane potential. Note the dramatically increased noise in the membrane potential post BIC (black). C, membrane potential traces before and after BIC for a SPN. Note the mild increase in high frequencies (>50 Hz) after the BIC ejection for the group of SPNs (n = 13). D, no change could be detected in the mean power spectra of CINs (n = 8). The scale is the same for all traces and neurons. The membrane potential offset is different for traces post BIC. Action potentials have been truncated.