Spinal cord injury significantly alters the properties of reticulospinal neurons: delayed repolarization mediated by potassium channels

Abstract

After rostral spinal cord injury (SCI) of lampreys, the descending axons of injured (axotomized) reticulospinal (RS) neurons regenerate and locomotor function gradually recovers. Our previous studies indicated that relative to uninjured lamprey RS neurons, injured RS neurons display several dramatic changes in their biophysical properties, called the “injury phenotype.” In the present study, at the onset of applied depolarizing current pulses for membrane potentials below as well as above threshold for action potentials (APs), injured RS neurons displayed a transient depolarization consisting of an initial depolarizing component followed by a delayed repolarizing component. In contrast, for uninjured neurons the transient depolarization was mostly only evident at suprathreshold voltages when APs were blocked. For injured RS neurons, the delayed repolarizing component resisted depolarization to threshold and made these neurons less excitable than uninjured RS neurons. After block of voltage-gated sodium and calcium channels for injured RS neurons, the transient depolarization was still present. After a further block of voltage-gated potassium channels, the delayed repolarizing component was abolished or significantly reduced, with little or no effect on the initial depolarizing component. Voltage-clamp experiments indicated that the delayed repolarizing component was due to a noninactivating outward-rectifying potassium channel whose conductance (g_K) was significantly larger for injured RS neurons compared to that for uninjured neurons. Thus, SCI results in an increase in g_K and other changes in the biophysical properties of injured lamprey RS neurons that lead to a reduction in excitability, which is proposed to create an intracellular environment that supports axonal regeneration.

NEW & NOTEWORTHY After spinal cord injury (SCI), lamprey reticulospinal (RS) neurons responded to subthreshold depolarizing current pulses with a transient depolarization, which included an initial depolarization that was due to passive channels followed by a delayed repolarization that was mediated by voltage-gated potassium channels. The conductance of these channels (g_K) was significantly increased for RS neurons after SCI and contributed to a reduction in excitability, which is expected to provide supportive conditions for subsequent axonal regeneration.

INTRODUCTION

For all vertebrates, reticulospinal (RS) neurons activate spinal neural circuits to initiate different types of motor functions, including locomotion (1). After a complete, rostral spinal cord injury (SCI), the descending axons of injured (axotomized) RS neurons are disrupted, resulting in a loss of locomotion as well as other motor functions (2). For “higher” vertebrates, including birds and mammals, axotomized RS neurons normally do not regenerate their axons, leading to permanent paralysis below the injury site (3, 4). In contrast, for “lower” vertebrates, including lampreys, fish, and certain amphibians, injured (axotomized) RS neurons regenerate their axons across a spinal injury site and reconnect with spinal neural circuits, resulting in virtually complete recovery of locomotion within a few weeks (reviewed in Refs. 5–7).

After rostral SCI, lampreys initially are paralyzed below the injury site but begin to display weak, incomplete locomotor-like movements in ∼2 wk, and by 8 wk locomotor function is nearly back to normal (8). With increasing recovery times, progressively greater numbers of RS neurons regenerate their axons across the injury site and for progressively greater distances below the injury site (9–11). However, even at relatively long recovery times (e.g., 32 wk), axonal regeneration of lamprey RS neurons is incomplete and does not restore the normal descending projection patterns of these neurons to the spinal cord (Refs. 8, 9; also see Refs, 12, 13). Several compensatory mechanisms, including descending propriospinal (PS) relay neurons, make up for incomplete axonal regeneration to mediate virtually full recovery (Refs. 14, 15; reviewed in Refs. 6, 7).

For spinal cord-injured lampreys, our studies have often focused on the biophysical properties of large, identified RS neurons, called Müller cells (Fig. 1A). Small, unidentified RS neurons, rather than Müller cells, are necessary and sufficient for the initiation of locomotor activity (16–19). However, in several respects, Müller cells can serve as models of lamprey RS neurons in general: 1) Large RS neurons are active during locomotor movements in vivo (20, 21) and during locomotor activity in vitro (22); 2) both large and small RS neurons receive direct or indirect inputs from higher locomotor areas in the brain (23, 24); 3) after SCI, these large lamprey RS neurons display changes in their biophysical properties similar to smaller, unidentified RS neurons (25); 4) both large identified and small unidentified RS neurons regenerate their axons after SCI (9–11).

Figure 1.

A, top: tracing of the brain (left) and rostral spinal cord (right) from a larval lamprey showing the four reticular nuclei: mesencephalic reticular nuclei (MRN) and anterior (ARRN), middle (MRRN), and posterior (PRRN) rhombencephalic reticular nuclei. The lamprey brain contains 28 large, identified reticulospinal (RS) neurons (14 left-right pairs) called Müller cells and Mauthner cells (dots). Bottom: enlargement of reticular nuclei showing Müller cells, which have ipsilateral descending axons: M cells (M1–M3) in MRN, I cells (I1–I4) in ARRN, and B cells (B1–B5) in MRRN. Mauthner (Mau) and auxiliary Mauthner (AM) cells in the MRRN have contralateral descending axons. All reticular nuclei, including the PRRN, contain small, unidentified RS neurons (omitted for simplicity). For experimental parts of the present study, “RS neurons” mainly refers to Müller cells. B: isolated brain-spinal cord preparation (not to scale) showing brain (left) and rostral spinal cord (right), intracellular micropipette (IC), right spinal cord hemitransection (HT) at 10% body length (BL, normalized distance from anterior tip of the head), and spinal cord suction electrode (SC). Evoked action potentials (APs) in uninjured (left) RS neurons (○) elicit orthodromic responses recorded from SC (suction electrode), whereas APs in injured (right) neurons (•) do not evoke responses below the spinal lesion site.

The descending axons of large identified lamprey RS neurons (Müller cells; Fig. 1A) project ipsilaterally in the spinal cord (19, 26). For some of our studies we have employed right rostral spinal cord hemitransections (HTs) so that most right large RS neurons were injured (axotomized) and their properties could be compared to those of uninjured (mostly left) large RS neurons (Fig. 1B). At 2- to 3-wk recovery times after right rostral spinal HTs, the changes in biophysical properties of injured RS neurons are maximal (25, 27). First, at these recovery times, uninjured (left) RS neurons fire a smooth, continuous train of action potentials (APs) in response to above-threshold current pulses (25, 28). Also, following the APs of uninjured neurons, there usually are three sequential afterpotential components: fast afterhyperpolarization potential (fAHP); afterdepolarization potential (ADP); and slow AHP (sAHP). In contrast, most injured (right) RS neurons respond to above-threshold current pulses by displaying distinct changes in their repetitive firing patterns (25, 27, 28): multiple relatively short bursts of APs; a single relatively short burst of APs; or a single AP. Also, following the APs of injured RS neurons, the fAHP is significantly larger and longer than for uninjured neurons, whereas the ADP and sAHP usually are virtually absent. After rostral SCI, injured RS neurons are largely disconnected from target-derived neurotrophic support from spinal synaptic targets, and this appears to be a stimulus for the changes in biophysical properties of these neurons and for regeneration of their axons (28).

Second, in addition to changes in repetitive firing patterns and afterpotentials, injured lamprey RS neurons also undergo a number of significant changes in their biophysical properties compared to those of uninjured neurons (27, 28): more hyperpolarized resting membrane potential (V_REST); larger membrane capacitance and time constant; larger and longer APs; higher threshold current (I_TH) and threshold voltage (V_TH) values for APs; decrease in average and peak spiking frequencies for a given suprathreshold depolarizing current pulse; and increase in the slope of the repolarizing (falling) phase of APs (dV_m/dt_fall, where V_m is membrane potential). Relative to uninjured RS neurons, the increase in I_TH and V_TH and the decrease in spiking frequencies for injured RS neurons indicate that SCI results in a decrease in excitability of these neurons following injury (27). Because high-voltage-activated (HVA) calcium channels are downregulated in injured lamprey RS neurons after SCI (25), a concurrent decrease in excitability would be expected to reduce and/or limit calcium influx in these neurons. Interestingly, experimentally elicited calcium influx in the growth cones or somata of lamprey RS neurons in culture inhibits neurite outgrowth or causes neurite retraction (29). Thus, SCI-induced changes in the biophysical properties of injured RS neurons result in intracellular conditions that very likely are supportive of axonal regeneration of these neurons (7).

For injured lamprey RS neurons after SCI, the increase in the amplitude of the fAHP (V_fAHP) and increase in the slope of the repolarizing phase of APs (dV_m/dt_fall) suggest that SCI may induce an increase in the conductance of voltage-gated potassium channels (g_K). This possibility was tested in the present study with current-clamp and voltage-clamp recordings from injured and uninjured RS neurons. For injured RS neurons, a possible increase in g_K following SCI could potentially contribute to the decrease in excitability mentioned above. After SCI, the changes in the properties of lamprey RS neurons might provide clues for enhancing axonal regeneration in higher vertebrates, including perhaps humans.

METHODS

Animal Care

Larval sea lampreys (Petromyzon marinus, N = 106 animals, length = 90–140 mm) were collected from streams and rivers in Michigan and Massachusetts. Animals initially were maintained at ∼4°C and subsequently were acclimated to 23°C for several days to several weeks before experimentation. Before all surgical procedures, animals were anesthetized in ∼200 mg/L tricaine methanesulfonate (MS-222; Crescent Research Chemicals; Phoenix, AZ). The Animal Care and Use Committee (ACUC) at the University of Missouri has approved the procedures in this study.

Spinal Cord Lesions

To compare the electrical properties of injured and uninjured large identified lamprey RS neurons (Müller cells), three animal groups were used: 1) normal animals (N = 49 animals) with no spinal lesions; 2) experimental animals (N = 45) with right spinal cord hemitransections (HTs) at 10% body length (BL, normalized distance from the anterior tip of the head); and 3) experimental animals (N = 12) with left-right staggered spinal HTs at 10% BL that injured (axotomized) all RS neurons. For experimental animals, a ∼5-mm dorsal incision was made at 10% BL to expose the spinal cord. Fine forceps and iridectomy scissors were used to lesion the cord, and subsequently the incision was manually closed. After surgery, experimental animals were placed in their home aquaria to recover at room temperature (∼23°C) for 2–3 wk before experiments were conducted, at which time the changes in electrical properties of injured RS neurons are maximal (25, 27).

Recordings were made from Müller cells (usually M2, M3, I1, B1, B3, B4; see Fig. 1A), which are left-right pairs of large, identified RS neurons with ipsilateral descending axons (19, 26). For experimental parts of the present study, “RS neurons” mainly refers to Müller cells. Thus, for experimental animals with right spinal cord HTs, most right RS neurons were injured (axotomized), leaving most left RS neurons uninjured (Fig. 1B). However, for animals with right spinal cord HTs, the injury status of RS neurons was always determined (see below) so that uninjured and injured neurons could be compared accurately. Animals with double HTs were used to ensure that all RS neurons were in fact injured when the injury status could not be determined [e.g., tetrodotoxin (TTX) in the bath; see below).

Isolated Brain-Spinal Cord Preparation

Dissection.

Normal animals, or experimental animals that had recovered for 2–3 wk, were anesthetized, and the brains and rostral spinal cords were exposed in cold, oxygenated lamprey Ringer solution (6–8°C) (mM): 10 HEPES, 130 NaCl, 2.1 KCI, 2.6 CaCl₂, 1.8 MgCl₂, and 4.0 dextrose (pH = 7.4). The brain and rostral spinal cord were then removed and pinned dorsal side up on a small strip of Sylgard (Corning Co; Midland, MI). Subsequently, the choroid plexus was removed, the cerebellar commissure was cut, and the obex was extended caudally, as previously described (9, 25, 27). Finally, the brain was pinned flat, and the Sylgard strip was transferred to a recording chamber containing cold, oxygenated lamprey Ringer solution (6–8°C) (Fig. 1B).

Recording setup.

Sharp micropipettes were pulled with a horizontal puller (model P-80; Sutter Instruments, Novato, CA) from thin-walled glass (World Precision Instruments, Inc. Sarasota, FL). Micropipettes were filled with 5 M K-acetate and then inserted into an electrode holder containing 3 M KCl. The electrode holder was plugged into either an X0.1L or an X1.0L head stage, which was connected to an intracellular amplifier (Axoclamp 2A; Axon Instruments, Foster City, CA). For recordings made in discontinuous current-clamp mode [DCC; sampling frequency (f_s) = 2–6 kHz]) or continuous current-clamp (i.e., bridge) mode, micropipettes had resistances of ∼50–70 MΩ, whereas for recordings made in discontinuous single-electrode voltage clamp (dSEVC; f_s = 2 kHz), micropipette resistances were ∼10–15 MΩ. The tip of a micropipette was manually positioned near the surface of an RS neuron, and a dc motor mounted on a manipulator was then used to rapidly advance the electrode in ∼1-µm increments until the cell was impaled. For neurons to be included in the analysis, V_REST had to be more negative than −65 mV, and the peak of action potentials (APs) had to be more positive than +20 mV. However, most of the neurons had AP amplitudes > 100 mV.

Individual action potentials (APs) were elicited in large, identified RS neurons (Müller cells) by applying single current pulses (+10 nA, 0–10 ms) using continuous current clamp, as previously described (25, 27, 28). A suction electrode was placed around the caudal end of the spinal cord (SC) to record possible orthodromic action potentials elicited by RS neurons (Fig. 1B). Thus, for preparations from animals with right spinal cord HTs, the absence (presence) of orthodromic APs caudal to the spinal lesion site confirmed that a given RS neuron (Müller cell) was injured (uninjured) (see Fig. 1B). Intracellular recordings were stored on VHS tape (NeuroData DR890; Cygnus Technologies, Delaware Water Gap, PA; 11 kHz sampling frequency per channel) and also were acquired with a custom data acquisition/analysis system (DT-3016 data acquisition board; Data Translations, Marlboro, MA).

Pharmacology

For some manipulations, the following blockers were added to the bath: 1) 0.5 mM kynurenic acid (KYN; Sigma Chemical Co., St. Louis, MO): blocked ionotropic excitatory amino acid receptors and reduced spontaneous electrical activity (30) when APs were not pharmacologically blocked; 2) 3 µM tetrodotoxin (TTX; Sigma): blocked voltage-gated sodium channels and APs; 3) 200 µM CdCl₂ and 400 µM NiCl₂: blocked high-voltage-activated (HVA) and low-voltage-activated (LVA) calcium channels; and 4) 10 mM tetraethylammonium chloride (TEA; Sigma) and 5 mM 4-aminopyridine (4-AP; Sigma): partially or completely blocked certain voltage-gated potassium channels.

Intracellular Recording and Analysis

In the present study, for large, identified injured lamprey RS neurons after SCI, at the onset of applied depolarizing current pulses that were just below as well as above threshold for APs, a transient depolarization was elicited consisting of an initial depolarizing component followed by a delayed repolarizing component (see Fig. 2A). In contrast, for most uninjured RS neurons, a substantial transient depolarization was mostly evident at membrane potentials above threshold when APs were pharmacologically blocked (see Fig. 3). For uninjured and injured RS neurons, current-clamp and voltage-clamp experiments were performed to characterize the transient depolarization and its underlying currents (31). Because uninjured RS neurons in normal animals and uninjured (mostly left) neurons in animals with right spinal cord HTs at 10% BL (see Fig. 1B) had very similar properties, their datasets were combined for statistical analysis.

Figure 2.

Gradation of the transient depolarization, which was mainly displayed by injured reticulospinal (RS) neurons (see text and Fig. 3, A1 and B1). The transient depolarization was elicited at the onset of applied subthreshold depolarizing current pulses and consisted of an initial depolarizing component (a) followed by a delayed repolarizing component (b) (see traces 2 and 3 in A). A: membrane potential (V_m) responses to three different 0.2-s depolarizing current pulses (I_m) for an injured (right) I1 RS neuron [resting membrane potential (V_REST) = −80 mV; see Fig. 1]. When the transient depolarization was relatively small and had a monophasic decay (trace 2), the amplitude was measured from the peak to the steady-state voltage at the end of the applied pulse. When the transient depolarization was relatively large and had a multiphasic decay (trace 3), the amplitude was measured from the peak to the first trough of the waveform (two horizontal dashed lines). The depolarizing voltage (ΔV_m) was the difference between V_REST and the steady-state voltage if there was no transient depolarization (trace 1) or between V_REST and the peak of the transient depolarization (traces 2 and 3). B and C: curves plotting the amplitude of the transient depolarization vs. the amplitude of the depolarizing current pulse (I_m; B) or depolarizing voltage (ΔV_m; C) for 8 different injured RS neurons that had relatively large transient depolarizations. For C, the rightmost point for each curve represents the amplitude of the maximal transient depolarization for each neuron that was elicited at just below threshold voltage (V_TH), where ΔV_m ≈ ΔV_TH = V_TH − V_REST (see methods).

Figure 3.

A1, B1, and C1: membrane potential (V_m) of uninjured and injured reticulospinal (RS) neurons in response to symmetrical 0.2-s hyperpolarizing and depolarizing current pulses (I_m). A2, B2, and C2: corresponding V-I plots of steady-state V_m (measured at the end of the applied pulse) vs. applied I_m. A1: response of an uninjured (left) B1 neuron (see Fig. 1). A2: the corresponding V-I plot (solid line and dots) displaying relatively linear depolarization up to the threshold voltage (V_TH; horizontal dashed line) for action potentials (APs). B1: response of an injured (right) B1 neuron (same brain as in A) showing transient depolarization (initial depolarizing component followed by delayed repolarizing component) elicited at the beginning of applied subthreshold depolarizing current pulses. B2: the corresponding V-I plot (solid line and dots) displaying nonlinear depolarization up to V_TH (horizontal dashed line). C1: response of an uninjured (left) B1 neuron (different neuron than in A) in the presence of 3 µM TTX, 400 µM NiCl₂, and 200 µM CdCl₂ showing transient depolarization (initial depolarization followed by delayed repolarization). C2: the corresponding V-I plot (solid line and dots) displaying nonlinear depolarization above V_TH [horizontal dashed line; estimated in TTX from the average ΔV_TH = V_TH – resting membrane potential (V_REST) for uninjured RS neurons; see text]. A1 and B1: dashed horizontal lines = clipped APs. A2, B2, and C2: dashed sloped lines = linear extrapolation of the slope of V_m polarization at V_REST. A1, B1, and C1: vertical/horizontal scale bars, 20 mV, 5 nA/40 ms.

Various parameters of APs and repetitive firing patterns were determined according to previously described procedures and definitions (see Refs. 25, 27, 28). In particular, for most injured RS neurons, and some uninjured neurons, that displayed a transient depolarization at the beginning of subthreshold depolarizing current pulses, voltage threshold (V_TH) was obscured when APs were present in response to suprathreshold currents (see Fig. 3B1). Therefore, V_TH was measured at the peak of the transient depolarization for depolarizing current pulses that were just subthreshold, and ΔV_TH = V_TH – V_REST (see Ref. 27).

First, with the DCC mode (f_s = 2 kHz), depolarizing current pulses (+0.01–10 nA, 0.2–0.6 s) were applied to uninjured RS neurons (n = 122 neurons) and injured neurons (n = 94). At the beginning of depolarizing current pulses that were just subthreshold, almost all injured RS neurons, and some uninjured neurons, produced a maximal transient depolarization (e.g., see V_m trace 3 in Fig. 2A). The percentages of uninjured and injured neurons that displayed a maximal transient depolarization for current pulses just subthreshold were determined.

Second, for the maximal transient depolarization elicited at just-subthreshold voltages, several parameters were determined (see Fig. 5 and Fig. 6): 1) amplitude: difference in voltage from the peak of the transient depolarization to the immediately following voltage trough or to the steady-state membrane potential at the end of the applied current pulse (see Fig. 2A); 2) delay-to-peak: interval from the onset of the current pulse to the peak of the transient depolarization; and 3) half-amplitude duration: duration of the transient depolarization at half-amplitude. For uninjured and injured neurons, the maximal amplitudes of the transient depolarization at just-subthreshold voltages were compared (Kruskal–Wallis test with Dunn’s multiple comparisons post-test). In addition, with the DCC mode (f_s = 6 kHz), suprathreshold depolarizing current pulses (+0.01–10 nA, 2.0 s) were applied to uninjured and injured neurons to characterize their repetitive firing patterns (see Fig. 5, A2, B2, C2, and D2). Virtually all uninjured neurons displayed smooth repetitive firing patterns. However, for each of the different types of altered firing patterns displayed by injured RS neurons (n = 91 neurons), the corresponding properties of the maximal transient depolarization were compared by one-way ANOVA with Bonferroni multiple comparisons post-test or Kruskal–Wallis with Dunn’s multiple comparison post-test (InStat; GraphPad Software, Inc., La Jolla, CA) (see Fig. 6).

Figure 5.

A1, B1, C1, and D1: membrane potential (V_m) of injured reticulospinal (RS) neurons in response to 2.0-s depolarizing current pulses (I_m) that were just subthreshold (insets are ×2 enlargements). A2, B2, C2, and D2: firing patterns of same injured neurons as in A1, B1, C1, and D1 in response to suprathreshold depolarizing current pulses [horizontal dashed lines = clipped action potentials (APs)]. A: injured I3 neuron (see Fig. 1A) that displayed no transient depolarization (A1; see inset) and smooth repetitive firing of APs (A2). B: injured B1 neuron that displayed a relatively small transient depolarization (B1; see inset) and fired multiple short bursts of APs (B2). Note V_m resonance between bursts (arrowhead). C: injured B4 neuron that displayed moderate-sized transient depolarization (C1; see inset; initial depolarizing component followed by delayed repolarizing component) and fired a single, short burst of APs (C2). Note short resonance-like response following the burst of APs (arrowhead). D: injured B3 neuron that displayed a relatively large transient depolarization (D1; see inset) and fired only a single AP (D2). See Fig. 6 for relationships between the properties of the transient depolarization and the different repetitive firing patterns. Vertical/horizontal scale bars, 25 mV, 10 nA/0.5 s.

Figure 6.

Histograms (bars = means; error bars = SDs) showing the relationships between the amplitude (A), delay-to-peak (B), and half-amplitude duration (C) of the maximal transient depolarization for injured reticulospinal (RS) neurons elicited by depolarizing current pulses at just subthreshold and the different repetitive firing patterns (x-axes of graphs; numbers in parentheses indicate n = numbers of neurons) (see Fig. 5). See text for definitions of properties of the transient depolarization. Note that for four of five neurons that displayed smooth repetitive firing (leftmost bars), the transient depolarization was absent or too small for the delay-to-peak or half-amplitude duration to be measured. *P < 0.05, **P < 0.01, ***P < 0.001; one-way ANOVA with Bonferroni multiple comparisons post-test or Kruskal–Wallis test with Dunn’s multiple comparisons post-test. AP, action potential.

Third, with the DCC mode (f_s = 2 kHz), symmetrical hyperpolarizing current pulses and subthreshold depolarizing current pulses (±0.01–30 nA, 0.2–0.6 s) were applied to uninjured RS neurons (n = 94 neurons) and injured neurons (n = 87). The steady-state membrane potentials (V_m) and applied currents (I_m) were used to construct V-I plots (see Fig. 3 and Fig 7). The membrane input resistance (R_in = ΔV_m/ΔI_m) between each consecutive pair of points on the V-I plots was calculated. To determine the linearity of the V-I plots for each neuron, the average R_in values for membrane hyperpolarization were compared to those for membrane depolarization (unpaired t test, with Welch correction when necessary, or Mann–Whitney test; InStat). The sign test (one-tail P value; https://www.graphpad.com/quickcalcs/binomial1.cfm) was applied to the fraction of uninjured or injured RS neurons for which R_in was significantly smaller for depolarized regions of the V-I plots compared to hyperpolarized regions. If the sign test indicated no significant difference, the membrane was classified as “linearly polarizing,” whereas if R_in for depolarizing potentials was significantly less than that for hyperpolarizing potentials, the membrane was classified as “nonlinearly polarizing” (see Fig. 3, A and B).

Figure 7.

Transient depolarization (initial depolarizing component followed by delayed repolarizing component) elicited at the beginning of a subthreshold depolarizing current pulse appears to depend mainly on activation of voltage-gated K⁺ channels and not appreciably on Na⁺ and/or Ca²⁺ voltage-gated channels. A1, B1, and C1: membrane potential (V_m) of an injured (right) I1 neuron (see Fig. 1) in response to symmetrical hyperpolarizing and depolarizing 0.2-s current pulses (I_m). A2, B2, and C2: corresponding V-I plots. Solid lines and dots = steady-state V_m at the end of the applied pulses; dashed lines = V_m values at the peak of the initial depolarization, before the delayed repolarization. A: V_m and I_m recordings with no blockers of voltage-gated channels in the bath (A1) and V-I plot (A2) [threshold voltage (V_TH) was −52 mV]. B: after addition of 3 µM TTX, 400 µM NiCl₂, and 200 µM CdCl₂, V_m and I_m traces (B1) and V-I plot (B2) were not appreciably altered from those in A. C: subsequently, after addition of 10 mM tetraethylammonium chloride (TEA) and 5 mM 4-aminopyridine (4-AP), the initial depolarization was still present, but the delayed repolarization was mostly blocked (C1) and the V-I plot (C2) was largely linear below V_TH, typical for passive channels. A1, B1, and C1: vertical/horizontal scale bars, 15 mV, 7.5 nA/50 ms.

In addition, with the DCC mode (f_s = 6 kHz), triangular current waveforms (±5–10 nA; 0.1 Hz, 0.5 Hz, 1.0 Hz) were applied to uninjured RS neurons (n = 80 neurons) and injured neurons (n = 57). Hyperpolarizing parts of the voltage waveforms and depolarizing parts of the waveforms, up to threshold, were used to further examine the linearity of polarization of the membrane potential (see Fig. 4).

Figure 4.

Membrane potential (V_m) of uninjured and injured reticulospinal (RS) neurons in response to 0.5-Hz triangular current waveforms (I_m). Freq = instantaneous firing frequency of action potentials. A and B: membrane polarization for an uninjured (left) B1 neuron (A) and an injured (right) B1 neuron (B) recorded from the same brain. C: membrane polarization for an injured (right) I1 neuron recorded from a different brain. Sloped dashed lines = linear extrapolation of V_m polarization for hyperpolarized potentials. Note relatively linear V_m vs. I_m for A and nonlinear V_m vs. I_m for B and C (arrowheads). It should be noted that for a given amplitude of injected suprathreshold depolarizing current, uninjured neurons had significantly higher average and peak spiking frequencies than injured neurons (27). Vertical/horizontal scale bars, 20 mV, 5 nA, 12.5 Hz/0.5 s.

Fourth, with the DCC mode (f_s = 2 kHz), the pharmacology of the transient depolarization for uninjured (n = 10 neurons) and injured (n = 13) RS neurons was determined for three different bath conditions: 1) without voltage-gated channel blockers present: recordings from uninjured and injured RS neurons for depolarizing current pulses that were just subthreshold for APs; 2) TTX, NiCl₂, and CdCl₂ in the bath: recordings to characterize parts of the V-I plot above threshold and determine whether components of the transient depolarization were due to sodium and/or calcium voltage-gated channels; and 3) TTX, NiCl₂, and CdCl₂ before and after application of TEA and 4-AP to the bath: recordings to test whether voltage-gated potassium channels contributed to components of the transient depolarization.

Fifth, with the dSEVC mode (f_s = 2 kHz), RS neurons were held at V_REST, and voltage steps (±1–30 mV, 0.2–0.6 s) were applied to uninjured and injured RS neurons. Current recordings were made under three different bath conditions: 1) absence of voltage-gated channel blockers in the bath (uninjured RS neurons: n = 8; injured RS neurons: n = 11); 2) TTX, NiCl₂, and CdCl₂ (uninjured: n = 23; injured: n = 17); and 3) TTX, NiCl₂, and CdCl₂ before and after addition of TEA and 4-AP (injured: n = 5). The passive leak currents were subtracted from the total measured currents. For depolarizing voltage steps, delayed outward currents (I_K) were recorded and measured for injured and uninjured RS neurons (see Figs. 8, 9, and 11).

Figure 8.

A and B: voltage-clamp recordings [discontinuous single-electrode voltage clamp (dSEVC)] of membrane currents (I_m) in response to 0.6-s hyperpolarizing and subthreshold depolarizing voltage pulses (V_m) for an uninjured (□) and injured (•) pair of reticulospinal (RS) neurons in the same brain in the absence of blockers for voltage-gated ion channels. Leakage currents were subtracted from total currents. A: uninjured (left) I1 neuron [held at resting membrane potential (V_REST) = −71.8 mV] displayed virtually no activation of outward currents at subthreshold depolarizing voltages. B: injured (right) I1 neuron (held at V_REST = −69.6 mV) exhibited activation of delayed outward (positive) currents at subthreshold depolarizing voltages. C: I-V plots for the neurons in A and B illustrating activation of outward current at subthreshold voltages for injured I1 neuron (•) but not for uninjured I1 neuron (□). Note that threshold voltage (V_TH; vertical dashed lines) was more depolarized for the injured RS neuron (•) than for the uninjured neuron (□), as previously demonstrated (27, 28). A and B: vertical/horizontal scale bars, 6.67 nA, 20 mV/0.2 s.

Figure 9.

A and B: voltage-clamp recordings of membrane currents (I_m) for an uninjured (□) and an injured (•) reticulospinal (RS) neuron in response to 0.4-s voltage pulses (V_m) in the presence of 3 µM TTX, 400 µM NiCl₂, and 200 µM CdCl₂ (leakage currents subtracted from total currents). Uninjured I1 neuron [held at resting membrane potential (V_REST) = −69.3 mV] (A) and injured B3 neuron (held at V_REST = −69.4 mV) (B) exhibited activation of delayed outward currents for depolarizing voltage pulses. These two neurons are illustrated because they had similar V_REST values. C: I-V plots indicate that the injured neuron (•) had an apparent activation voltage for the delayed outward current of approximately −65 mV (V_K; left arrow), whereas the uninjured neuron (□) had an apparent activation voltage of approximately −55 mV (V_K; right arrow) (see Fig. 10 and text). Thin sloped dashed lines indicate linear regression through last three points for each I-V plot to estimate V_K for the delayed outward currents (see methods). Thick curved dashed line (arrowhead) indicates that scaling the currents of the I-V plot for the uninjured RS neuron (□) by a factor of 10 resulted in a relatively good match to the I-V plot for the injured neuron (•) (the average scaling factor was 4.0; n = 23 uninjured neurons, n = 17 injured neurons; see text). A and B: vertical/horizontal scale bars, 10 nA, 20 mV/0.2 s.

Figure 11.

Current-voltage (I-V) plot from a voltage-clamp experiment for an injured B1 neuron in the presence of 3 µM TTX, 400 µM NiCl₂, and 200 µM CdCl₂ before (•) and after (○) the addition of 10 mM tetraethylammonium chloride (TEA) and 5 mM 4-aminopyridine (4-AP) to the bath. Scaling the current of the I-V plot before application of TEA and 4-AP (•) by a factor of 0.06 (6%) resulted in a relatively good match (curve denoted by △ symbols) to the I-V plot after application of K⁺ channel blockers (○) (the average scaling factor was 14.9% ± 9.0%; n = 5 neurons; see text). Note that before partially blocking potassium channels (•), the large amplitude of the outward current and the limitations of discontinuous single-electrode voltage clamp (dSEVC) prevented membrane potential (V_m) from being depolarized to the same levels as after partially blocking these channels (○).

Sixth, in the presence of blockers for sodium and calcium voltage-gated channels, dSEVC recordings were made (see Fig. 9, A and B) and I-V plots were constructed to estimate the apparent activation voltage (V_K) at which the delayed outward current (I_K) began to appear (uninjured: n = 23; injured: n = 17) (see Fig. 9C). The slopes (ΔI_m/ΔV_m) between sequential points on the I-V plots were calculated and normalized to the maximum slope, which occurred between the two rightmost (i.e., most depolarized) points of the plots. The pairs of points that had a normalized slope of ≥90% of the maximum slope were included in a linear regression analysis, and the x-axis intercept of the linear regression was used to estimate V_K (see Fig. 9C and Fig. 10).

Figure 10.

Box-and-whisker graphs comparing the properties derived from voltage-clamp experiments (see Fig. 9) for uninjured (n = 23 neurons) and injured (n = 17 neurons) reticulospinal (RS) neurons recorded in the presence of 3 µM TTX, 400 µM NiCl₂, and 200 µM CdCl₂. A: the apparent activation voltage (V_K) for the delayed outward current (I_K; see Fig. 9C) was at a significantly more hyperpolarized membrane potential for injured RS neurons (mean ± SD = −62.1 ± 2.3 mV) compared to that for uninjured neurons (−56.4 ± 5.8 mV; **P < 0.01; Kruskal–Wallis with Dunn’s multiple comparisons post-test). B: the amplitude of the depolarization for activating the delayed outward current (ΔV_K = V_K − V_REST) was significantly smaller for injured RS neurons (mean ± SD = 5.8 ± 1.6 mV) compared to that for uninjured RS neurons (12.6 ± 4.3 mV; ***P < 0.001; unpaired t test with Welch correction). Note that for injured RS neurons threshold voltage (V_TH) was significantly more depolarized and ΔV_TH (= V_TH − V_REST) was significantly larger compared to those for uninjured neurons (see text), as previously described (27, 28).

Statistics

All data are presented as mean ± standard deviation (SD), except for the box-and-whisker plots in Fig. 10. The properties of uninjured and injured RS neurons were compared by either an unpaired t test, with Welch correction when necessary, or a Mann–Whitney test (InStat). Multiple datasets usually were compared by either one-way ANOVA with Bonferroni multiple comparisons post-test or a Kruskal–Wallis test with Dunn’s multiple comparisons post-test (InStat). Significance was assumed for P ≤ 0.05.

RESULTS

Characteristics of the Transient Depolarization

Gradation of the transient depolarization.

For almost all injured RS neurons, and some uninjured neurons, at the beginning of subthreshold depolarizing current pulses (0.01–10 nA, 0.2–0.6 s or 2.0 s; DCC mode), a transient depolarization was elicited consisting of an initial depolarizing component followed by a delayed repolarizing component. In most cases in the present study, measurements were made of the maximal amplitude of the transient depolarization, which was elicited at just below threshold voltage (V_TH). However, the amplitude of the transient depolarization was graded with subthreshold depolarizing currents and voltages (Fig. 2A). This graded effect was most clearly shown for injured RS neurons with the largest maximal transient depolarizations (n = 8 neurons) (Fig. 2, B and C). For example, increases in the amplitude of the subthreshold depolarizing current pulse (I_m) resulted in an exponential-like graded increase in the amplitude of the transient depolarization (Fig. 2B). Similar results were shown for plots of the amplitude of the transient depolarization versus membrane depolarization (ΔV_m = V_m − V_REST) (Fig. 2C). For each of the curves, which represent different neurons (Fig. 2, B and C), the rightmost point represents the maximal amplitude of the transient depolarization elicited by current pulses that were just subthreshold, at which ΔV_m ≈ ΔV_TH = V_TH − V_REST (see methods for how V_TH was measured).

Effects on membrane potential.

The transient depolarization was investigated with the DCC mode by applying depolarizing current pulses (0.01–10 nA, 2.0 s) to both uninjured and injured RS neurons. For all uninjured RS neurons that were depolarized by current pulses just subthreshold, 66.4% (n = 81/122 neurons) did not display a transient depolarization (Fig. 3A1), and the average maximal value of this potential for all uninjured neurons was 0.40 ± 1.05 mV (n = 122). However, 33.6% (n = 41/122) of uninjured RS neurons did display a relatively small transient depolarization whose maximal amplitude for those particular uninjured neurons was 1.19 ± 1.05 mV (n = 41). In contrast, 97.9% (n = 92/94 neurons) of injured RS neurons depolarized by current pulses that were just subthreshold displayed a maximal transient depolarization. The average amplitude of this potential for all injured neurons was 4.87 ± 2.14 mV (Fig. 3B1), which was significantly larger than the two amplitudes mentioned above for uninjured neurons (P < 0.001, Kruskal–Wallis test with Dunn’s multiple comparisons post-test).

For uninjured and injured RS neurons, V-I plots were constructed from application of symmetrical hyperpolarizing and subthreshold depolarizing current pulses (±0.01–30 nA, 0.2–0.6 s). For most uninjured RS neurons, the membrane potential depolarized approximately linearly up to V_TH (Fig. 3A2). In contrast, the transient depolarization observed for most injured neurons, and some uninjured neurons, resisted or counteracted depolarization, causing the membrane potential to depolarize nonlinearly as it approached V_TH (Fig. 3B2). To quantify the linearity of the membrane potential polarization, for each of the V-I plots average R_in values (= ΔV_m/ΔI_m) for the hyperpolarized region of the plots were statistically compared to those for the depolarized regions of the plots, from V_REST to V_TH (unpaired t tests; see methods). A significant percentage of all uninjured RS neurons (n = 65/94 neurons = 69%) had linear V-I plots (P = 0.0001, one-tail sign test; see methods) (Fig. 3, A1 and A2) for which R_in was not significantly different for hyperpolarized and depolarized regions of the plots (see methods). In contrast, a significant percentage of injured RS neurons (n = 61/87 neurons = 70%) had nonlinear V-I plots (P = 0.0001; sign test) (Fig. 3, B1 and B2) for which R_in was significantly less for depolarized regions of the plots compared to the hyperpolarized regions. In addition, compared to most uninjured RS neurons, for injured neurons the subthreshold, nonlinear regions of the V-I plots (Fig. 3B2) extended the threshold current (I_TH) to higher values, and this factor plus the higher average V_TH values for injured neurons (27, 28) likely contributed to a decrease in excitability (see discussion).

For uninjured RS neurons, the possibility of a substantial transient depolarization for voltages more depolarized than V_TH (i.e., above V_TH) was explored in the presence of blockers for voltage-gated sodium channels (3 µM TTX) and voltage-gated calcium channels (400 µM NiCl₂ and 200 µM CdCl₂). Under these conditions, all uninjured RS neurons tested (n = 10 neurons) now displayed a transient depolarization, largely at membrane potentials more depolarized than V_TH (Fig. 3C1). Thus, the transient depolarization probably was not due appreciably to voltage-gated sodium and/or calcium channels but was more likely due to an initial depolarizing component that was countered by a delayed repolarizing component, possibly mediated by potassium channels (see below). Furthermore, after addition of the above drugs, a significant fraction of uninjured RS neurons tested (n = 10/10 neurons = 100%) displayed nonlinear regions in their V-I plots (P = 0.001, one-tail sign test), similar to the V-I plots for injured neurons, except mostly for membrane potentials more depolarized than V_TH (Fig. 3C2).

To further characterize the effects of the transient depolarization on membrane potential and the linearity of V-I plots, triangular current waveforms (0.1, 0.5, 1.0 Hz; ±5–10 nA) were applied to injured and uninjured RS neurons (no voltage-gated channel blockers in the bath) to roughly mimic the alternating excitatory and inhibitory synaptic inputs these neurons receive during the generation of spinal locomotor activity (22). For uninjured neurons (n = 80 neurons), injection of triangular current waveforms resulted in relatively linear membrane hyperpolarization and subthreshold depolarization (dashed line, Fig. 4A) (27) and elicited a smooth train of APs when the membrane potential was depolarized above V_TH. For injured RS neurons (n = 57), injection of triangular current waveforms resulted in relatively linear membrane hyperpolarization (dashed lines, Fig. 4, B and C), but in the depolarizing direction, between V_REST and V_TH, the membrane potential (V_m) clearly deviated from linearity and polarized nonlinearly (arrowheads, Fig. 4, B and C). This nonlinearity in the depolarizing direction presumably resulted because the delayed repolarizing component resisted or counteracted further depolarization. It should be noted that for a given amplitude of injected suprathreshold depolarizing current, uninjured neurons had significantly higher average and peak spiking frequencies than injured neurons (27).

For 37% (n = 21/57) of injured RS neurons, the delayed repolarizing component was sufficient to prevent the membrane potential from reaching threshold in response to ±10-nA peak triangular current waveforms (Fig. 4B). For 63% (n = 36/57) of injured neurons, the membrane potential could be driven above V_TH to elicit APs for at least some frequencies of the applied triangular current waveforms, but the membrane potential still polarized nonlinearly between V_REST and V_TH (Fig. 4C).

Interestingly, for injection of ±10-nA triangular current waveforms, 11% (n = 4/36) of injured RS neurons did not fire APs in response to applied relatively low-frequency triangular current waveforms (0.1, 0.5 Hz) but did fire at higher waveform frequencies (1.0 Hz). A similar phenomenon was observed in a previous study (27). Apparently, for injection of lower-frequency triangular waveform currents there was sufficient time for the delayed repolarizing component to more fully develop during depolarization to counteract further depolarization to V_TH.

Correlation of the Transient Depolarization with Altered Repetitive Firing Patterns

At 2–3 wk recovery times following right spinal cord HTs at 10% BL, 97% (n = 114/118 neurons) of uninjured (mostly left) RS neurons (see Fig. 1B) fired a smooth train of APs in response to 2.0-s suprathreshold depolarizing current pulses, similar to that previously reported (25, 27, 28). In contrast, almost all injured (mostly right) RS neurons adopted the injury phenotype, which included several types of repetitive firing patterns in response to 2.0-s suprathreshold depolarizing current pulses (+0.01–8.0 nA) (Fig. 5): 1) ∼5.5% (n = 5/91 neurons) of injured neurons still fired a smooth, continuous train of APs (Fig. 5A2); 2) 27.5% (n = 25/91) of neurons fired multiple short bursts (Fig. 5B2); 3) ∼52.7% (n = 48/91) of neurons fired a single relatively short burst (Fig. 5C2); and 4) ∼14.3% (n = 13/91) of neurons fired a single AP at all current levels above threshold, up to +10 nA (Fig. 5D2).

Interestingly, for injured RS neurons (n = 91), the four types of repetitive firing patterns described above were correlated with the amplitude, delay-to-peak, and half-amplitude duration of the maximal transient depolarization elicited for depolarizing current pulses that were just subthreshold. First, there was a progressive increase in the average amplitude of the maximal transient depolarization (∼0.6–5.7 mV) for injured RS neurons that fired a smooth train of APs, multiple short bursts, a single short burst, and a single AP (Fig. 5, A1, B1, C1, and D1). Specifically, the average amplitude of this potential was significantly larger for neurons that fired multiple bursts compared to those that fired smoothly (P < 0.001) and for neurons that fired a single burst compared to those that fired multiple bursts (P < 0.01; one-way ANOVA with Bonferroni multiple comparisons post-test, InStat; Fig. 6A). Second, there was a progressive decrease in the delay-to-peak for the transient depolarization such that this parameter was significantly shorter for injured RS neurons that fired a single burst compared to those that fired multiple bursts (P < 0.01) and for neurons that fired a single AP compared to those that fired a single burst (P < 0.01; Fig. 6B). Third, there was a progressive decrease in the half-amplitude duration of the transient depolarization such that this parameter was significantly shorter for injured RS neurons that fired a single burst compared to those that fired multiple bursts (P < 0.05) and for neurons that fired a single AP compared to those that fired a single burst (P < 0.001; Kruskal–Wallis test with Dunn’s multiple comparison post-test, InStat; Fig. 6C). Furthermore, there were significant negative correlations for the maximal amplitude of the transient depolarization (n = 90 neurons) versus the delay-to-peak (P = 0.0143, linear regression; InStat) and versus the half-amplitude duration (P = 0.0007), although the r² values were relatively small (0.066 and 0.126, respectively).

For injured RS neurons, the amplitudes of the maximal transient depolarization (n = 94 neurons) elicited by depolarizing current pulses that were just subthreshold had significant positive correlations versus the absolute threshold voltage (V_TH, P < 0.0001, linear regression, InStat), versus the relative threshold voltage (ΔV_TH = V_TH − V_REST, P = 0.0002), and versus the threshold current (I_TH, P < 0.0001), although the r² values were relatively small (0.174, 0.140, and 0.215, respectively). In general, a more depolarized threshold voltage might be expected to produce a larger transient depolarization by allowing for a larger initial depolarizing component and greater activation of the delayed repolarizing component (see Fig. 3B2), which appears to be mediated by voltage-gated potassium channels (see Ion Channels Mediating the Transient Depolarization). For example, there was a significant positive correlation for the amplitude of the maximal transient depolarization versus the absolute value of the fast AHP (|V_fAHP|; P = 0.031, r² = 0.051, linear regression, n = 92 neurons), which is mediated largely by fast voltage-gated potassium channels (25).

In our previous study, we determined that different Müller and Mauthner cells (see Fig. 1A) have different axonal regenerative capacities (i.e., percentages) for regenerating from 10% BL (SCI) to 20% BL at 32 wk after injury (10): I3 (94%); I4 (89%); M1 and AM (∼72%); I2, B1, B2, and B4 (∼60%); M2, B3, and B5 (∼30%); I1 and Mau (∼20%); and M3 (11%). [Note that these percentages generally are higher than those in a later study from a different laboratory (32), which used recovery times of ∼9–14 wk.] In the present study, we were interested to determine whether different injured Müller cells had different firing patterns or different amplitudes for the transient depolarization. First, in the present study, for each of the injured Müller cell types that were the main focus of the present study (M2, M3, I1, B1, B3, B4), the different firing patterns were numerically scored: smooth firing (= 1), multiple bursts (= 2), single burst (= 3), and single AP (= 4). The average numerical scores (2.4–2.9) for the different Müller cell types (n = 5–28 neurons per cell type) were not significantly different (P > 0.05, Kruskal–Wallis with Dunn’s multiple comparisons post-test), although the M3 (worst regenerator) and I1 neurons had the lowest scores (i.e., least change from normal, smooth firing). These data suggest that the different cell types did not have preferentially different firing patterns in response to SCI, similar to previous results (27). Second, the average amplitudes of the maximal transient depolarization for the different Müller cell types were not significantly different from each other (P > 0.05, one-way ANOVA with Tukey–Kramer multiple comparisons post-test), although Müller cell M3 had the lowest mean amplitude. These results suggest that injury-type biophysical properties and the appearance of the transient depolarization for injured lamprey RS neurons may be important conditions for axonal regeneration but are not predictors of successful regeneration by themselves (see discussion).

Ion Channels Mediating the Transient Depolarization

To determine the contribution of various voltage-gated ion channels to the transient depolarization, continuous DCC recordings were made first without and then with blockers for different voltage-gated ion channels in the bath. Because the amplitude of the transient depolarization was voltage dependent (Fig. 2C), for all bath conditions for a given neuron the maximal transient depolarization was measured for the same ΔV_m ≈ ΔV_TH (= V_TH − V_REST) (see methods for definition of V_TH). First, for one set of injured RS neurons (n = 6 neurons), the amplitude of the maximal transient depolarization under control conditions was 5.34 ± 2.31 mV. After application of sodium and calcium channel blockers, the transient depolarization was not significantly decreased (P > 0.05, Kruskal–Wallis with Dunn’s multiple comparisons post-test; Fig. 7, A and B). In contrast, subsequent application of potassium channel blockers (Fig. 7C) mainly diminished or abolished the delayed repolarizing component of the transient depolarization but appeared to have little or no effect on the initial depolarizing component. As a result of nearly abolishing the delayed repolarizing component, the amplitude of the transient depolarization (0.11 ± 0.22 mV) was significantly reduced compared to its original amplitude (P < 0.01). Second, for a different set of injured neurons (n = 7), only blocking sodium channels (but not calcium channels) did not significantly diminish the amplitude of the transient depolarization (P > 0.05), whereas subsequently blocking potassium channels significantly reduced the delayed repolarizing component and transient depolarization to ∼3.7% of their original amplitudes (P < 0.01). Third, for uninjured neurons that displayed a clear transient depolarization (n = 4), blocking voltage-gated sodium, calcium, and potassium channels significantly reduced the transient depolarization (P = 0.03, Mann–Whitney test). Thus, the initial depolarizing component of the transient depolarization likely is due to applied depolarizing current passing through passive channels. However, the delayed repolarizing component very likely is due to voltage-gated potassium channels and not appreciably to delayed inactivation of voltage-gated sodium and/or calcium channels. The incomplete block of the delayed repolarizing component by TEA and 4-AP in the present study might be due to a partial block of potassium channels (33) or residual potassium channels that are not blocked by these agents (34).

Current(s) Mediating the Delayed Repolarizing Component

Voltage-clamp recordings (dSEVC; see methods) were performed for uninjured and injured RS neurons to characterize the current(s) responsible for the delayed repolarizing component. First, with no blockers for voltage-gated channels in the bath, at membrane potentials just below V_TH (see vertical dashed lines in Fig. 8C), uninjured RS neurons (n = 8 neurons) displayed an average current of −0.24 ± 0.25 nA (Fig. 8A), which was significantly less than the delayed outward current (I_K) of +2.74 ± 2.18 nA (Fig. 8B) displayed by injured neurons (n = 11) (P = 0.001, unpaired t test with Welch correction) (27, 28). For injured RS neurons, there was virtually no inward current, and the delayed outward current did not display inactivation over ∼600 ms for membrane potentials approximately −70 to −50 mV (Fig. 8B). Also, for injured neurons (n = 11), at membrane potentials just below threshold there was a significant positive correlation between the amplitude of the delayed outward current (I_K measured with dSEVC) and the maximal transient depolarization (measured with DCC) (P = 0.023, r² = 0.46; linear regression). Because of the relatively low signal-to-noise ratio of the current recordings, it was difficult to precisely quantify the time constant (τ) for the delayed outward current (I_K), but fitting each of the current traces (n = 9 neurons) with a single-exponential curve yielded an average τ of ∼40 ms (e.g., see Fig. 8B).

Second, in the presence of blockers for voltage-gated sodium and calcium channels, delayed outward currents (I_K) were measured for uninjured (Fig. 9A) and injured (Fig. 9B) RS neurons with voltage clamp (see methods). For both injured and uninjured RS neurons, there was little or no inward current, and the delayed outward currents did not appear to inactivate (Fig. 9, A and B). Uninjured RS neurons displayed delayed outward currents (Fig. 9A) with an apparent activation voltage (V_K; see methods) of −56.4 ± 5.8 mV [n = 23 neurons; Fig. 10B (27, 28); see right arrow in Fig. 9C]. For injured neurons, the delayed outward currents (Fig. 9B) had an apparent activation voltage of −62.1 ± 2.3 mV (n = 17; Fig. 10A; see left arrow in Fig. 9C), which was significantly more hyperpolarized than that for uninjured RS neurons (P < 0.01, Kruskal–Wallis with Dunn’s multiple comparisons post-test; Fig. 10A). Additionally, the amount of membrane depolarization (ΔV_K = V_K − V_REST) required to activate the delayed outward current was significantly greater for uninjured RS neurons compared to that for injured neurons (P < 0.001, unpaired t test with Welch correction; Fig. 10B).

For the present dataset, V_TH for injured RS neurons (−57.2 ± 5.0 mV, n = 94) was significantly more depolarized than that for uninjured neurons (−61.8 ± 4.8 mV, n = 122) (P < 0.001; Kruskal–Wallis with Dunn’s multiple comparisons post-test), verifying previous results (27, 28). Consequently, for uninjured RS neurons V_TH was significantly more hyperpolarized than V_K (P < 0.001), whereas for injured neurons V_TH was significantly more depolarized than V_K (P < 0.001, Kruskal–Wallis with Dunn’s multiple comparisons post-test). Thus, for uninjured RS neurons I_K was activated mostly above threshold voltages, whereas for injured neurons I_K was activated below as well as above threshold voltages (Fig. 8 and Fig. 9), similar to the above-described features of the transient depolarization (Fig. 3, A1, B1, and C1).

Compared to uninjured RS neurons, for injured RS neurons 1) the actual activation voltage (V_K) for I_K might have shifted to more hyperpolarized potentials after injury or 2) the conductance of the delayed outward-rectifying channel might have increased after injury. Under the conditions in which sodium and calcium voltage-gated channels were blocked, separate scatterplots were constructed with I-V data points from a group of uninjured RS neurons (n = 11 neurons) and from a group of injured RS neurons (n = 6) that had V_REST values within ∼4 mV of each other (not shown). The data points for each group were fitted with a composite curve, and scaling the currents of the composite I-V curve for uninjured RS neurons by a factor of 4.0 resulted in a good fit to the composite I-V curve for injured neurons (for the specific pair of currents in Fig. 9C this scaling factor was 10; see dashed line and arrowhead). This analysis, in addition to other evidence such as a significant increase in the amplitude of the fAHP and slope of the falling phase of the AP (dV_m/dt_fall) for injured RS neurons compared to uninjured neurons (25, 27, 28), suggests that the conductance of an outward-rectifying potassium channel (g_K) increased for injured RS neurons after SCI (see discussion). In contrast, shifting the composite I-V curve for uninjured RS neurons to the left (i.e., more hyperpolarized; see Fig. 9C) did not result in a good fit to the composite I-V curve for injured neurons. Thus, the actual activation voltages for the delayed outward current probably were similar for uninjured and injured RS neurons, but for injured neurons less depolarization was needed to result in substantial activation of this current. For example, with the equation g_K = I_K/(V_m – E_K), with E_K = −90 mV, for Fig. 9C g_K was equal to ∼0.18 µS at V_m = −59.5 mV for the injured RS neuron, but the uninjured neuron required depolarization to −45.8 mV to reach this same conductance.

Third, the pharmacology of the delayed outward current was further examined with voltage-clamp recordings from injured RS neurons (n = 5 neurons) in the presence of 3 µM TTX, 400 µM NiCl₂, and 200 µM CdCl₂, before and after the addition of 10 mM TEA and 5 mM 4-AP (see methods). Under these conditions, injured RS neurons displayed a noninactivating delayed outward current before and after partial block of voltage-gated potassium channels (Fig. 11). Because TEA and 4-AP may not have completely blocked voltage-gated potassium channels in the present study (35–37), the delayed outward current was substantially reduced but not completely abolished in the presence of these blockers. For each neuron, the outward currents that were generated before application of potassium channel blockers were scaled (see methods), so the resultant curve was a good fit to the corresponding outward currents after addition of these blockers. On average, these potassium channel blockers reduced the amplitude of the delayed outward current to 14.9 ± 9.0% of its value before these blockers were applied (P < 0.0001, one­sample t test compared to 100%) (for the specific neuron in Fig. 11 this scaling factor was 6%; see curve marked with △ symbols). Conversely, multiplying the delayed outward currents after blockade of potassium channels by an average factor of ∼8.8 resulted in a good match to the currents before the block.

DISCUSSION

Injury Phenotype for Lamprey RS Neurons after SCI

Our previous studies demonstrated that after SCI for the lamprey injured RS neurons displayed several dramatic changes in their properties, described as the injury phenotype, compared to uninjured neurons (25, 28). These changes include altered firing patterns; changes in afterpotential components, including a significant increase in the amplitude of the fAHP (V_fAHP); and downregulation of calcium-activated potassium (SK) channels and HVA calcium channels. In our subsequent study, neurophysiological experiments indicated that relative to uninjured lamprey RS neurons, injured neurons displayed several significant changes in their biophysical properties (27): 1) increased membrane capacitance (C_in) and time constant (τ_in) but no significant change in membrane resistance (R_in); 2) increased threshold voltage (V_TH and ΔV_TH) and threshold current (I_TH); 3) larger amplitude (V_AP) and longer duration (D_AP) for action potentials; 4) higher slope for the repolarizing (falling) phase of action potentials (dV_m/dt_fall); and 5) lower average and peak spiking frequencies for a given amplitude of a suprathreshold depolarizing current pulse.

The data from the present study extended the above findings regarding the underlying neuronal mechanisms for the injury phenotype following SCI. In particular, the results indicate that at the beginning of depolarizing current pulses that were just below as well as above threshold (V_TH), injured (axotomized) lamprey RS neurons displayed a transient depolarization (i.e., initial depolarizing component followed by a delayed repolarizing component) that was mostly only evident to a substantial degree at above-threshold voltages for uninjured neurons when APs were blocked. The initial depolarizing component could elicit APs, and the delayed repolarizing component, whose current activated with a time constant (τ) of ∼40 ms, could terminate firing. Interestingly, the amplitude, delay-to-peak, and half-amplitude duration of the maximal transient depolarization elicited by depolarizing currents that were just subthreshold were correlated with the different repetitive firing patterns displayed by injured RS neurons (Fig. 5 and Fig. 6). The initial depolarizing component of the transient depolarization likely was due to applied depolarizing current passing through passive channels. The delayed repolarizing component was not appreciably dependent on a delayed inactivation of voltage-gated sodium and/or calcium channels. Rather, this component was mediated by a noninactivating (or very slowly inactivating), outward-rectifying potassium channel whose conductance (g_K) for injured RS neurons was significantly larger than that for uninjured neurons. At present, it is not known whether the increase in g_K following SCI is due to an increase in the number of K⁺ channels and/or an increase in the unit conductance of individual channels.

After SCI, injured RS neurons display an increase in V_TH and ΔV_TH (see above), presumably due to changes in voltage-gated sodium channels, and an increase in g_K. Together, these changes allow voltage-gated potassium channels to be effectively activated at subthreshold membrane potentials, at which voltage-gated sodium channels are not yet activated. Relative to uninjured neurons, for injured RS neurons the increase in g_K was shown to counteract depolarization and probably contributed to the increase in current thresholds (I_TH) and decrease in spiking frequencies described above (27). The changes in properties of axotomized RS neurons following SCI described in our present and previous studies very likely will reduce excitability, which is expected to provide a supportive intracellular environment for axonal regeneration (see Implications for Axonal Regeneration following SCI in the Lamprey).

After SCI, an increase in g_K for injured lamprey RS neurons is also indicated by a significant increase in the fAHP (25, 28) and dV_m/dt_fall for APs (27). Thus, the increase in delayed outward K⁺ current for RS neurons following SCI probably is associated with one or more of the noninactivating, or very slowly inactivating, delayed outward rectifier Kv channel subtypes that is responsible, in part, for repolarization of the action potential in lamprey neurons (Ref. 38; see Refs. 33, 39). However, one of the difficulties in determining the specific channel(s) underlying the delayed outward K⁺ current based on kinetics is that expression in oocytes of some of the K⁺ channel subtypes from different animals can result in somewhat different kinetics (40). In addition, the delayed outward K⁺ current described in the present study might be due to a combination of several potassium channel subtypes. A recent study used RNA sequencing (RNA-Seq) to describe changes in expression of a large number of ion channels in the lamprey brain (41), but at present it is not known which of these changes in expression might occur in RS neurons or other brain cells (neuronal and/or nonneuronal). Also, possible posttranslational modifications of ion channel activity (e.g., phosphorylation) would not be directly detectable with this technique.

At 12–16 wk, the majority of injured RS neurons display smooth, regular repetitive firing, similar to that for uninjured neurons (25, 27). In addition, at these recovery times the fAHP for injured RS neurons is not significantly different than that for uninjured neurons (25, 27), suggesting that g_K is no longer elevated. Finally, for injured RS neurons that have recovered for 12–16 wk and display smooth repetitive firing, we have not observed the transient depolarization, which contributes to the reduction in excitability that is most evident at 2–3 wk after SCI.

Implications for Axonal Regeneration following SCI in the Lamprey

First, after SCI injured lamprey RS neurons display significant changes in several properties, relative to those of uninjured neurons, that are expected to reduce excitability (see above in discussion and Ref. 27). Results from the present study indicate that an increase in the conductance of voltage-gated potassium channels (g_K) contributes to this reduction in excitability. Second, injured RS neurons display a significant decrease in expression of HVA calcium channels (25). In addition, preliminary results suggest that calcium currents are reduced in injured RS neurons compared to those for uninjured neurons (42). Third, for lamprey RS neurons in culture, experimental stimulation of calcium influx in growth cones or somata via voltage-gated and/or chemically gated ion channels inhibits neurite outgrowth or causes neurite retraction (29). Thus, elevation of intracellular calcium appears to be detrimental for axonal outgrowth of lamprey RS neurons (reviewed in Ref. 7). Taken together, these results suggest that after SCI the increase in g_K, decrease in excitability, and decrease in expression of calcium channels for injured lamprey RS neurons reduce and/or limit calcium influx and result in intracellular conditions that contribute to the ability of these neurons to regenerate their axons (25, 27). Notably, for several types of embryonic chick neurons, the normal expression of voltage-gated K⁺ channels limits excitability and restricts calcium influx, thereby maintaining intracellular calcium in an optimal range for axonal extension (43), whereas suppression of these K⁺ channels has an inhibitory effect on axonal extension.

For injured lamprey RS neurons, a decrease in g_K, reduction in excitability, and resultant reduction and/or limitation of calcium influx likely are not the only factors contributing to successful axonal regeneration. For example, all injured Müller cells display a decrease in excitability (Refs. 25, 27, 28; see results), but some of these RS neurons (e.g., M3; see Fig. 1A) exhibit markedly poorer axonal regenerative capacity compared to other large RS neurons (Ref. 10; also see Ref. 32 and results). Perhaps the changes in excitability and calcium influx contribute to a switch of injured RS neurons into a regenerative mode whereby the neurons attempt to regenerate their axons but together with other factors determine whether regeneration is successful. For example, the largest Müller cells, which tend to be the poorest regenerators because, in part, of delayed apoptosis (Refs. 10, 32; also see results), require an extended period of time for their transected axons to seal. Application of polyethylene glycol (PEG) to the spinal cord injury site to enhance axon sealing reduces RS neuron apoptosis (44). In addition, several factors or signaling pathways are preferentially activated in the poorest Müller cell regenerators (45, 46). Also, certain manipulations can enhance axonal regeneration of lamprey RS neurons (Ref. 47; also see Ref. 48), whereas others inhibit regeneration (49).

Finally, after SCI in the lamprey axonal regeneration of RS neurons is incomplete (9, 10), and several mechanisms appear to compensate for reduced supraspinal descending projections. For example, RS neurons and descending propriospinal (PS) relay neurons act in parallel to activate spinal locomotor networks (15). After rostral SCI, uninjured descending PS neurons below the injury site sprout and extend their axons (14), and this likely will partially compensate for incomplete axonal regeneration of RS neurons (Refs. 9, 14, 15, 50; reviewed in Refs. 6, 7). Also, there appears to be an increase in excitability for presumably uninjured spinal interneurons and motoneurons (MNs) below the injury site as well as potentiation of proprioceptive inputs (51). These changes might be part of a compensatory response of the spinal motor networks allowing them to be more responsive in the face of incomplete axonal regeneration of RS neuron projections (see Fig. 5 in Ref. 6). But, because RS and PS neurons can both activate spinal locomotor networks (9, 14, 15, 50), after rostral SCI, descending PS neuron plasticity, and behavioral recovery, the degree to which the net descending drive to these networks is reduced remains unclear. Finally, for spinal cord-injured lampreys, regeneration of descending axons across a rostral injury site is essential for behavioral recovery because physically blocking regeneration prevents recovery (unpublished observations from Ref. 29).

Relationship of Intracellular Calcium and Axonal Regeneration

For many neurons, there is a range of intracellular calcium levels ([Ca²⁺]_i) that supports axonal outgrowth, referred to as the “calcium set-point hypothesis” (Ref. 52; reviewed in Refs. 53–55). Thus, increases (decreases) in internal calcium levels above (below) this optimal range are detrimental to axonal growth. For example, for molluscan neurons in culture, calcium influx mediated by application of neurotransmitters, electrical stimulation, or application of calcium ionophores inhibits neurite outgrowth (reviewed in Refs. 53–57). For mammalian central nervous system (CNS) neurons in culture, contact of growth cones with mature myelin increases intracellular calcium and causes growth cone collapse (58), which can be negated by blockers for voltage-gated calcium channels (59). Finally, in rats after spinal nerve injury, peripherally axotomized dorsal root ganglion (DRG) neurons display a reduction in electrical activity as well as a decrease in L-type voltage-gated calcium channels (60). Both of these changes in the properties of peripherally axotomized DRG neurons contribute to the ability of the central branches of their axons to regenerate in the CNS after SCI. Interestingly, for many neurons, there is a reduction in calcium channel currents, conductances, and/or expression levels after axotomy (see references in Ref. 25).

For a few types of neurons, increases in calcium influx enhance axonal outgrowth and blocking calcium influx inhibits axonal elongation (61, 62). For these neurons, the baseline [Ca²⁺]_i might be at the lower end of the optimal range for axonal outgrowth, such that increasing Ca²⁺ influx augments outgrowth whereas blocking calcium influx lowers the [Ca²⁺]_i below the optimal range.

Biophysical Properties after SCI for Other Types of Lamprey Neurons

A few weeks after SCI for the lamprey, injured spinal dorsal cells (DCs), which are centrally located sensory neurons (63), and injured giant interneurons (GIs), which are spinal interneurons (63), display a significant reduction in V_REST (i.e., less negative) (64, 65). In addition, injured DCs also display a significant increase in R_in, ΔV_TH, I_TH, and D_AP (64). Thus, DCs appear to undergo a decrease in excitability after injury (64), despite the decrease in V_REST.

Axotomy-Induced Changes in the Properties of Neurons in Other Vertebrates

Unlike injured lamprey RS neurons, for many vertebrate neurons axotomy results in a reduction in threshold (i.e., I_TH, V_TH, and/or ΔV_TH = V_TH − V_REST), which typically increases excitability. For example, peripherally axotomized DRG neurons display a decrease in threshold and increase in excitability in rat (66–69), mouse (70), and hamster (71) (see Ref. 72 for similar results for axotomized mouse trigeminal sensory neurons). Similarly, there is a decrease in thresholds and an increase in excitability after axotomy of spinal motoneurons (MNs) in cat (Refs. 73–76; see Ref. 77 for similar results for axotomized cat facial MNs). Axotomy of sympathetic B cells in bullfrog results in an increase in I_TH but apparently little change in excitability (78), possibly because of compensatory changes in other properties. After axotomy of rat rubrospinal neurons, there appears to be an initial decrease followed by a later increase in excitability (79).

After injury, there often is a decrease in conductance, currents, and/or expression of delayed-rectifier, voltage-gated potassium channels of axotomized neurons, including DRG neurons (80–86) and sympathetic B cells (87). In addition, axotomized vagal and facial MNs display a decrease in A current (88, 89). For injured DRG neurons and MNs, a decrease in conductance/current/expression of potassium channels, as well as a concurrent decrease in threshold, would be expected to increase excitability.

The reasons for the difference in responses to axotomy of lamprey RS neurons and the neurons in other vertebrates mentioned above are unclear. Perhaps these other vertebrate neurons, many of which are capable of axonal regeneration after axotomy, require an increase in excitability to elevate intracellular calcium ([Ca²⁺]_i) into the range that supports axonal growth.

Summary

After SCI, at the beginning of applied depolarizing current pulses just below as well as above threshold, injured lamprey RS neurons exhibited a transient depolarization, consisting of an initial depolarizing component followed by a delayed repolarizing component. In contrast, for uninjured neurons a substantial transient depolarization was evident mostly at membrane potentials above threshold when APs were blocked. Pharmacological experiments suggested that the initial depolarizing component was due to applied depolarizing current passing through passive channels. The delayed repolarizing component was not appreciably dependent on voltage-gated sodium and/or calcium channels. Rather, the delayed repolarizing component was mediated by a noninactivating, outward-rectifying potassium channel whose conductance (g_K) was significantly larger for injured RS neurons compared to that for uninjured neurons. After SCI, an increase in g_K, as well as other changes in the properties of injured lamprey RS neurons, contributed to a decrease in excitability, which is expected to reduce and/or limit calcium influx and provide a cellular environment that is supportive for axonal regeneration. The present study identified altered properties of injured RS neurons after SCI in the lamprey that are presumed to support regeneration of their axons, and these results may provide insights into improving regeneration and behavioral recovery in spinal cord-injured higher vertebrates, including perhaps humans.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

This work was supported by NIH Grant NS 29043 and American Paralysis Association Grant MB1-9108 awarded to A.D.M.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

R.A.H. and A.D.M. conceived and designed research; R.A.H. performed experiments; R.A.H. and A.D.M. analyzed data; R.A.H. and A.D.M. interpreted results of experiments; R.A.H. and A.D.M. prepared figures; R.A.H. and A.D.M. drafted manuscript; R.A.H. and A.D.M. edited and revised manuscript; R.A.H. and A.D.M. approved final version of manuscript.