K+ channels at the axon initial segment dampen near-threshold excitability of neocortical fast-spiking GABAergic interneurons

Abstract

Fast-spiking cells (FS cells) are a prominent subtype of neocortical GABAergic interneurons with important functional roles. Multiple FS cell properties are coordinated for rapid response. Here, we describe an FS cell feature that serves to gate the powerful inhibition produced by FS cell activity. We show that FS cells in layer 2/3 barrel cortex possess a dampening mechanism mediated by Kv1.1-containing potassium channels localized to the axon initial segment. These channels powerfully regulate action potential threshold and allow FS cells to respond preferentially to large inputs that are fast enough to "outrun" Kv1 activation. In addition, Kv1.1 channel blockade converts the delay-type discharge pattern of FS cells to one of continuous fast spiking without influencing the high-frequency firing that defines FS cells. Thus, Kv1 channels provide a key counterbalance to the established rapid-response characteristics of FS cells, regulating excitability through a unique combination of electrophysiological properties and discrete subcellular localization.

Figure 1. Salient Electrophysiological Properties of FS Cells in Barrel Cortex

(A) Discharge pattern of an FS cell in layer 2/3 barrel cortex. Shown are voltage responses to 600 ms current injections of, from bottom to top, −100, 100, 345 (I_TH; blue), 360, 410, and 440 pA. Gray bars highlight delay to first spike. Inset in the topmost trace shows the first two spikes on an expanded timescale (scale bar, 10 ms). (B) Firing pattern of an adjacent PC. Shown are responses to current injections of −40, 50 (I_TH), 60, 110, and 210 pA. Scale bars as in (A). (C) f–I curves for a subset of representative FS cells (blue) and PCs (red), with the cells in (A) and (B) indicated by dots (red, FS cell; blue, PC). (D) Summary data illustrating initial firing frequency (f_i; n = 12) and the slope of the f–I relation (in Hz/nA; n = 16) for FS cells and PCs. Error bars here and in subsequent figures indicate SD; *p < 0.01 level; **p < 0.05. (E) Plot of instantaneous firing frequency (IFF; quantified as the inverse of the ISI) (same FS cell as in [A]), calculated for each AP in a train elicited in response to current injections of 440, 460, 480, and 500 pA, illustrating spike-frequency acceleration. (F) Summary data comparing spike-frequency adaptation for FS cells (n = 31) and PCs (n = 4).

Figure 2. Delay-Type Firing Is Associated with a Slow Ramp Depolarization and a Positive Shift in AP Voltage Threshold

(A) Response of an FS cell in layer 2/3 barrel cortex (different from cell in Figure 1) to current injections of, from bottom to top, 680, 700 (I_TH), 720, 760, and 800 pA. The red dot in the threshold trace indicates the first spike at I_TH, shown in red on an expanded timescale; the blue dot in the uppermost trace indicates a spike time-matched to the red dot in the threshold trace and is shown in blue on an expanded timescale in the inset. (B) Shown are the indicated single APs from (A) at increased magnification. (C) Phase plane (V_m versus dV/dt) for the first AP generated during delayed firing at I_TH (red; same spike in [A] and [B]) and all spikes generated during cFS discharge (blue; all spikes from the uppermost trace in [A]). (D) Summary data (n = 12) illustrating the difference between AP threshold calculated for the first spike generated during delayed firing at I_TH versus the first spike generated in response to larger current injections that elicited cFS discharge. Error bars indicate SD. (E) Plot of delay to first spike versus current injection for the cell shown in (A). (F) Plot of AP voltage threshold versus current injection (same cell). Solid red dot, I_TH; open red dot, dFS; open blue dot, onset spike with pause in firing; solid blue dot, cFS. (G) Detail (5-fold increase in vertical scale) of the subthreshold voltage to the 680 pA current injection shown in Figure 2A (bottom-most trace), illustrating the amplified response to a near- but just subthreshold current injection, with a single-exponential fit to the slow depolarization (red; τ = 118 ms). The amplitude of the slow depolarization (indicated by the arrow) is 4.9 mV.

Figure 3. DTx Blocks Delayed Firing by FS Cells

(A) voltage response of an FS cell in layer 2/3 mouse barrel cortex to current injections of, from bottom to top, −200, 200, 260, 280 (I_TH), 300, and 400 pA. (B) Same cell as in (A), after bath application of 50 nM DTx-I. Current injections are, from bottom to top, −100, 100, 170, 175 (new I_TH), 200, and 300 pA. Note decrease in I_TH and elimination of delay, but preservation of high-frequency discharge at suprathreshold current injection. (C) Summary data (n = 7). DTx-I produces no change in R_m or f_i but decreased I_TH and shifted the voltage threshold for AP generation to hyperpolarized potentials. Error bars indicate SD. (D) Plot of IFF for each AP in an FS cell spike train in response to current injections of various amplitudes (green, 400 pA; red, 460 pA; blue, 500 pA). Note the spike-frequency acceleration early in the train, followed by spike-frequency adaptation later in the train. (E) As in (D), after application of 100 nM DTx-I. Note elimination of spike-frequency acceleration.

Figure 4. Kv1.1 Protein Is Enriched at the FS Cell AIS

(A) Confocal image showing immunohistochemical staining for ankyrin-G (AnkG; blue, left) and Kv1.1 (red, center) in barrel cortex of a transgenic mouse expressing EGFP (green, left) in FS cells. Shown is a maximum intensity projection of seven images at a z separation of 1.03 µm. AnkG appeared to label initial segments of all neurons in neocortex, including those of FS cells. Kv1.1 staining of cortical neuropil was widespread. Scale bar, 100 µm. (B) Higher-magnification image (maximum projection of 23 images; z separation, 0.48 µm) of the region indicated in (A), showing two GFP-positive cells from layer 2/3; Kv1.1 staining is enriched at the AIS and colocalizes with AnkG (arrowheads). (Insets) Single optical sections through the cell bodies of indicated Kv1.1-immunoreactive cells. Scale bar, 20 µm (10 µm in the inset). (C) Typical image (projection of 25 images; z separation, 0.43 µm) illustrating enrichment of Kv1.1 at the AIS of a GFP-positive FS cell. Scale bar, 10 µm.

Figure 5. Kv1.1 Protein Is Present at the AIS but Absent from Somatic Membrane of FS Cells in Layer 2/3 Barrel Cortex

(A) Composite electron micrographs of a Kv1.1-immunopositive nonpyramidal cell found in ultrathin sections spaced throughout the depth of the cell. An ~20 µm extent of the AIS is outlined and pseudocolored in four successive sections. Rectangles represent regions magnified in panels (B)–(E). Scale bar, 5 µm. (B) DAB staining product accumulated in discrete clusters on the cytosolic side of the nuclear membrane (long arrows) and on the endoplasmic reticulum (short arrows). Scale bar, 1 µm for panels (B) and (C). (C) Cytoplasmic Kv1.1 immunoreactivity was also present around mitochondria (m). In the cortical neuropil, labeled terminals (t) that may be presynaptic (*) to the indicated cell soma were occasionally observed; however, the cytosolic side of the perikaryal membrane (indicated with red arrows from outside of cell) was clearly devoid of Kv1.1 labeling. Note the abundance of mitochondria, a hallmark of PV-positive interneurons. (D) Kv1.1 was also present in the cytoplasm at the base of the axon hillock within the first 10 µm of the AIS (short arrow) and on microtubule bundles (long arrows), yet avoided axolemma (red arrowheads). Scale bar, 1 µm for panels (D) and (E). (E) Dense Kv1.1 immunoreactivity (white arrows) was found on the membrane of the AIS (red arrowheads) beyond a distance of ~10 µm from the soma, lining the undercoating of the axolemma.

Figure 6. In a Two-Compartment FS Cell Model, Kv1 Conductance Effectively and Specifically Dampens Near-Threshold Excitability If Present at the Axon

(A) FS cell model in the absence of Kv1. Current injections were delivered in steps of 10 pA; shown are voltage responses to current injections of, from bottom to top, −80, 80 (10 pA below threshold), 90 (current threshold for AP generation), 110 (20 pA above threshold), and 180 pA (two times threshold). (B) Addition of a Kv1 current to the axonal compartment (see Experimental Procedures) produces an increase in current threshold for AP generation and delay-type firing over a broad range of near-threshold current injections, without altering high-frequency discharge. Shown are voltage responses to current injections of −80, 80, 130 (10 pA below threshold), 140 (threshold), 160 (20 pA above), and 280 pA, (two times threshold). Delayed firing is high-lighted with gray bars, as in Figure 1A. (C) Kv1 in the soma (but not in the axon) cannot reproduce robust delay-type firing, but instead produces an onset spike at I_TH. Shown are voltage responses to current injections of −80, 80, 120, 130 (threshold), 150, and 260 pA. Note that in (B) and (C), the first spike following a pause in firing is smaller than the second (and subsequent) spikes. This feature is also observed experimentally, but to a lesser extent (average 2.1 mV difference; range 0.5–4.4 mV; n = 10) and is due to accumulation of Na⁺ channel inactivation during the slow ramp depolarization. The discrepancy in spike heights could be essentially normalized in the model by increasing the proportion of persistent Na⁺ current from 0.8% to 3%.

Figure 7. Kv1 Current Functions to Suppress the Response of FS Cells to Slowly Rising Inputs

(A) Current ramps (500 pA peak amplitude in this example; variable rise time) elicited spiking in FS cells if sufficiently rapid (slope ≥ 25.0 nA/s). However, spiking in response to slower ramps (from <25.0 to >14.3 nA/s) was suppressed. Sufficiently slow inputs (14.3 nA/s or less) produced long-latency spikes preceded by a delay. (B) The window of spike suppression was abolished by 50 nM DTx-K. (C) PC firing increased progressively with current ramps of increasing duration (but identical amplitude).

Figure 8. Kv1 Current Functions to Dynamically Regulate the Near-Threshold Excitability of FS Cells in Response to Time-Varying Inputs

(A) Current sinusoids (4 Hz; shifted via DC offset to produce a positive deviation only) illustrate near-threshold suppression of excitability early in stimulus trains. At I_TH (peak current = 315 pA), response to the first cycles in a train were suppressed, with successive cycles achieving progressively greater depolarization (see also the boxed region in [B]). (B) Same cell, after 50 nM DTx-K. Note that the first cycle in the series at I_TH (285 pA) now elicits a spike. The boxed region illustrates suppression of responsiveness to early cycle presentations under control conditions (blue) that is eliminated by DTx (red). (C) Behavior of a PC to the same simulation paradigm (see text). (D) Plot of the phase (θ) of the peak of each AP elicited in response to trains of sinusoids of increasing amplitude. Note phase advancement within a train for all current amplitudes in the dFS cell under control conditions (D₁), a phenomenon that is eliminated by DTx (D₂) and not present in PCs (D₃).