Concussive Head Trauma Deranges Axon Initial Segment Function in Axotomized and Intact Layer 5 Pyramidal Neurons

Experimental animals

Two mouse lines were used to generate the data of the present study. The first line expresses Thy1-driven yellow fluorescent protein (YFP) within a subset of layer 5 pyramidal neurons ²⁸ and was acquired from Jackson Labs (B6Cg-TgN[Thy1-YFPH] 2Jrs, stock number 003782; Bar Harbor, ME). Ear punches were inspected by fluorescence microscopy to confirm the presence of the YFP transgene.

To generate the second line, these same YFP-H mice were crossed with cyclophilin-D (CypD) knockout mice, which have been shown to be more resistant to the assembly and opening of the mitochondrial permeability transition pore and the consequent derangement of intracellular calcium dynamics and mitochondrial swelling. ^29,30 The gene encoding CypD is ppif, and thus these knockout mice are also known as ppif^-/-. Here the YFP-H strain is referred to as wildtype (WT), and the CypD knockout mice crossed with the YFP-H are referred to as CypDKO.

For this study, male mice (P56-P90) were used. The mice were group housed in 12h/12h non-reversed light cycle conditions on corncob bedding with access to food and water ad libitum. All animal procedures were approved by the Institutional Animal Care and Use Committee of Virginia Commonwealth University.

Central fluid percussion injury (cFPI)

Mild cFPI was induced as described previously. ^{31 –34} Mice were anesthetized with 4% isoflurane in 100% O₂, with anesthesia maintained with 2% isoflurane during surgery. The body temperature was thermostatically controlled via a heating pad (Harvard Apparatus, Holliston, MA) and maintained at 37°C. A pulse oximetry sensor (STARR Life Sciences, Oakmont, PA) was used to monitor pulse rate, respiratory rate, and blood oxygenation intraoperatively.

A 3.0 μm circular craniectomy was made along the sagittal suture midway between bregma and lambda. Care was taken to avoid any insult to the underlying dura. The cFPI injury at this location consistently produces diffuse axonal injury throughout the primary somatosensory cortex. A sterile Leur-Loc syringe hub cut from a 20-gauge needle was affixed to the craniectomy site using cyanoacrylate and dental cement then filled with sterile saline. This procedure required 45–75 min. The animal was then removed from anesthesia and monitored in a warmed cage until fully ambulatory (60–80 min of recovery). Injury or sham procedure was applied thereafter.

For injury induction, each animal was reanesthetized with 4% isoflurane in 100% O₂ for 4 min four, and the hub was filled with sterile saline and attached to a fluid percussion apparatus (Custom Design and Fabrication; Virginia Commonwealth University; Richmond, VA). A mild severity injury (1.7 ± 0.06 atmospheres) was induced by a brief fluid pressure pulse on the intact dura. The peak pressure was measured by the transducer (Tektronix 5111) and displayed on an oscilloscope.

After injury, the animals were monitored visually for loss of spontaneous respiration or the emergence of clonus. Transient derangement of respiratory rhythm that resolved into eupnea was routinely observed. The duration of transient unconsciousness was measured by the time interval between the injury and the return of the spontaneous righting reflex. While sham righting times occur within approximately 1 min, within this cFPI injury paradigm, injury is considered effectively mild when righting occurs within 8 min, a standard that held true for all injured mice included here.

After recovery of the righting reflex, animals were placed in a warmed holding cage and monitored during recovery (typically ∼60 min) before being returned to the vivarium. For sham injury, all the above steps were followed except for the release of the pendulum to induce the injury.

Acute slice preparation and patch-clamp recording

Our previous work has shown that the effect of this mild CFP injury progresses with intact neurons at one day manifesting physiological properties similar to axotomized neurons while intact neurons at two days manifest properties similar to controls. ^32,33 Thus we continued the investigation at these two survival times.

One or two days after injury or sham, mice were deeply anesthetized with isoflurane and perfused through the aorta with semi-frozen, sucrose substituted artificial cerebrospinal fluid (aCSF) previously saturated by a mixture of 95% O₂/5% CO₂ (carbogen) containing (in mM): 2.5 KCl, 10 MgSO₄, 0.5 CaCl₂, 1.25 NaH₂PO₄, 234 sucrose, 11 glucose, and 26 NaHCO₃. Subsequently the mice were decapitated, and the brain was rapidly extracted and chilled in this same aCSF.

Scrupulous care was undertaken to generate anatomically coronal slices in which vibratome-induced axotomy was minimized. To this end, the brain was carefully docked within an acrylic mouse brain slicer matrix (Zivic Instruments). 300-μm thick coronal slices were cut with a Leica VT 1200 vibratome (Leica Microsystems, Wetzlar, Germany). These slices were incubated for 30 min at 34°C in aCSF containing (in mM): 126 NaCl, 3 KCl, 2 MgCl₂, 2 CaCl₂, 1.25 NaH₂PO₄, 10 glucose, and 26 NaHCO₃, bubbled continuously with carbogen.

The slices remained at room temperature thereafter until transferred to the recording chamber, which was maintained at 30 ± 0.5^o C. The slices were secured by a harp, and a 63X water-immersion objective (Zeiss) was used to visually identify YFP⁺ layer 5 pyramidal neurons of primary somatosensory cortex ventral to the injury site. Neurons possessing an axon that could be traced along its trajectory into the subcortical white matter were considered intact, while those terminating in an axonal bleb were considered axotomized. Care was taken to select only neurons axotomized deep to the surface of the slice to exclude any neurons axotomized by the vibratome during slicing. Among YFP+ layer 5 pyramidal neurons in the YFP-H condition, differentiating between these two morphologies is readily feasible within the living brain slice. ³³

Within the recording chamber, slices were perfused with aCSF bubbled continuously with carbogen. Patch electrodes (final resistance 2–4 MΩ) were pulled from borosilicate glass (World Precision Instruments, Sarasota, FL). The intracellular solution contained (in mM): 130 K-gluconate, 10 Hepes, 11 EGTA, 2.0 MgCl2, 2.0 CaCl2, 4 Na-ATP, 0.2 Na-GTP, 0.4% biocytin (Sigma, cat# B4261). Electrode capacitance was electronically compensated. Recordings were made via a MultiClamp 700B (Molecular Devices, Sunnyvale, CA) and digitized via a Digidata 1440A and pClamp software (Molecular Devices).

During current clamp recordings, the membrane potential was maintained at -60 mV with continuous depolarizing or hyperpolarizing current as needed. Access resistance was continuously monitored. If the access resistance increased by 15% at any time between break in and the culmination of data collection, an attempt was made to clear the pipette tip via gentle positive pressure. If this attempt was not successful, the recording was terminated.

Most neurons had a resting membrane potential (RMP) between -60 mV and -70 mV (Table 1), except for a few severely axotomized neurons that rested between -60 mV and -55 mV. The axotomized neurons for both one- and two-day survival times were significantly more depolarized than those of the two-day sham condition (Table 1). On the measure of input resistance, there was no significant difference in subject groups or experimental conditions (WT or CypDKO) and no significant interaction between these two (Table 1).

To probe for the existence of trauma-induced perturbation of AIS function, we recorded from ex-vivo brain slices prepared from mice subjected to mild cFPI one or two days before experimentation. In the resulting slices, we used a 63X water immersion objective to follow the trajectory of the axons of YFP+, layer 5 pyramidal neurons to entrance into the subcortical white matter (intact neurons), or to the axonal bleb/swelling (axotomized neurons); 33 neurons thus visualized were targeted for whole cell patch clamp recording. From these neurons, high-resolution APs were generated by a brief depolarizing current injection at a level that typically evoked 1–2 APs, which were recorded unfiltered and digitized at 200 kHz (high resolution APs for 2^o analysis). This procedure was repeated 5–9 times at 0.2 Hz.

In addition, the membrane voltage response to a series of hyperpolarizing and depolarizing pulses of 400 msec duration to identify the AP firing pattern and other intrinsic cellular properties was recorded. These recordings were digitized at 20 kHz.

We note that all YFP+, layer 5 pyramidal neurons tested possessed some degree of the hyperpolarization-activated cyclic nucleotide-gated current (Ih), which manifests as a depolarizing sag in response to hyperpolarizing current injection. The amplitude of this sag was added to the amplitude of the depolarizing afterpotential in the manner described by Gee and colleagues. ⁶⁴ It was determined that all neurons recorded in this study could be classified as Type A (Supplementary Fig. 1). In previous studies neuronal firing pattern has also been classified as regular spiking (RS), regular spiking with an initial doublet (RSd), or intrinsically bursting (IB). ^31,33 In the current cohort we observed all three of these firing types in each subject group, with no significant differences in the proportions (Supplementary Fig. 2).

OPEN IN VIEWER

FIG. 1.

The biphasic components of the action potential (AP) upstroke, visualized with a 2^o analysis are caused by sequential depolarization of the axon initial segment (AIS) and the soma. Results shown here are for application of tetrodotoxin (TTX) to the AIS. (A) Experimental schematic, including image taken during recording. The TTX is puffed onto the AIS (position 1) or the soma (position 2, see Fig. 2). (B) Maximum projection of the confocal image for same cell shown in A. Arrows indicate axon. (C) 2^o of the AP from the cell shown in A and B, during sequential application of TTX. Green dashed lines show linear slope from 99% to 10% of Cp2 peak. Symbols in first panel show the associated peaks for D. (D-F) Measurements made on the same cell as A–C. The pre-TTX and wash measurements are each an average of 5, while individual measurements of TTX application are shown. Black markers are plotted on the left axis and brown on the right axis. Shown in F is the summed difference points from the real data to a line drawn from the 99% of the second peak to 10% of the peak. Note the dashed line for the left axis that indicates a threshold level of 2.25 dividing 2^o with double peaks versus a shoulder+peak or single peak. (G) Decrement of Cp1 with TTX application was typical of control cells (n = 3 neurons from two slices from one mouse), with some decrement of the Cp2 peak but significantly more for the Cp1 peak. A two-way repeated measures ANOVA showed a significant effect of TTX timing and of the interaction of TTX location (AIS vs. soma) with the TTX timing. (H–K) A similar effect was found in two intact cells one day after injury that had double peaks pre-TTX. Here all five cells (from four slices from two mice) are grouped together. In H the peaks are normalized to the pre-TTX level. This shows the relatively selective effect on the Cp1 peak of the TTX applied to the AIS. Here also a repeated measures two-way ANOVA showed a significant effect of TTX timing and of the interaction between TTX location (with results in Fig. 2 considered) and TTX timing. (I) The ratio of Cp1/Cp2 peaks also shows a significant reduction, with repeated measures one-way ANOVA. (J) The time of the Cp2 peak was not significantly different after TTX. (K) The LS_diff measure falls below the 2.25 threshold, indicating a single peak, at the late TTX time point, and this measure is significantly different with TTX application (repeated measures ANOVA).

OPEN IN VIEWER

FIG. 2.

Effect of TTX applied to the soma of a layer 5 pyramidal neuron. (A) 2^o of the action potential (AP) recorded before, during and after tetrodotoxin (TTX) application to the soma. (B) Same waveforms as shown in A after normalization to the Cp1 peak, showing the relative effect on Cp2. Data in C–E are from this cell, with averages of five evoked APs measured for pre-TTX and Wash points. (C) Left axis shows percent of the pre-TTX peak for Cp1 (black) and Cp2 (gray). Right axis shows Cp2 peak (brown). (D) Because the TTX applied to the soma has a greater effect on Cp2 than Cp1, the Cp1/Cp2 ratio increases with each TTX application (left axis, black). Time between Cp1 and Cp2 peaks also increases (right axis, brown). (E) The LS_diff (black, left axis) increases, while the minimum value between peaks relative to Cp1 peak (brown, right axis) decreases with TTX application to the soma. (F) The effect on the Cp1/Cp2 ratio showed a consistent increase (t test, one tailed, p < 0.05). (G) The minimum value between peaks relative to Cp1 peak (t test, p < 0.05). For F and G, n = 3 neurons from three slices from two mice.

To correlate the biphasic components of the AP upstroke to the regional depolarization of the AIS and the soma, a patch pipette containing tetrodotoxin (TTX, 2 μM) was positioned approximately 30 μm distal to the soma (to dissociate the AIS component) or over the soma (to dissociate the somatic component) of visualized YFP⁺ layer 5 pyramidal neurons within the WT sham condition. Subsequently, TTX was applied through the patch pipette via a Picospritzer (Parker Hannifin, Mayfield Heights, Ohio) at 30 psi for a duration of 8 msec and at a rate of 0.1 Hz.

Immediately after each puff of TTX, an AP was evoked via square wave depolarizing current injection and subjected to 2^o analysis. The slice was positioned within the chamber with the apical dendrites projecting toward the aCSF inflow and the axon projecting toward the outflow to minimize TTX spillage onto the soma during TTX application to the AIS.

Immunohistochemistry, confocal and super-resolution microscopy

After each recording, neurons were depolarized slowly to – 40 mV, and the patch pipette was gradually withdrawn over the course of approximately 20 sec to reseal the neuronal membrane. Slices were fixed in 4% paraformaldehyde for 70 min, then washed six times in 0.1M phosphate-buffered saline (PBS) and stored in 0.1M PBS at 4°C until processing. To reconstruct the morphology of recorded neurons, slices were washed three times for 5 min with 0.1M PBS, then reacted overnight at 4°C with a solution of 0.1M PBS containing Texas Red-conjugated streptavidin (1:500, ThermoFisher, cat# S32356) in the presence of 0.5% triton-X.

To visualize the structure of the AIS, slices were then treated with heat-based epitope retrieval. Slices were immersed within a solution of 10 mM sodium citrate dihydrate (294.10 g/mol; pH = 8.0) and heated on a block to 80°C for 10 min. The slices were allowed to cool for 10 min, then blocked in a solution of 10% normal horse serum and 0.5% triton-X within 0.1M PBS at room temperature for 1h. Slices were then washed three times with 0.1M PBS for 10 minutes and incubated in a solution containing anti-ankyrin-G primary antibody (1:400; mouse IgG2a, clone N106/36, NeuroMab, cat# 73-146) in 0.1M PBS with 0.5% triton-X and 1% NHS for 24h at 4°C.

The slices then were washed three times with 0.1M PBS and reacted with a solution containing the secondary antibody (Alexa Fluor 405; 1:500, Life Technologies, cat# A31553) in 0.1M PBS with 0.5% triton-X and 1% NHS for 2h at room temperature. Subsequently, slices were washed three times for 10 min in 0.1M PBS and mounted on glass slides with Vectashield non-hardening antifade mounting medium, and the slides sealed with clear nail polish.

Slices thus prepared were imaged at 63X via a Zeiss 880 laser scanning confocal microscope. Serial scanning of each of the three channels (Texas Red, YFP, and Alexa Fluor 405) was employed, and care was taken to image the entire axon of all successfully recovered neurons.

For measurement of the distance to axotomy, the distance from the emergence of the axon to the site of the first retraction bulb was measured using the freehand line tool on the biocytin or YFP channel within ImageJ. For axotomized neurons of club-shaped morphology and lacking classical retraction bulbs, the distance from the emergence of the axon to the end of the axon was measured. For measurement of the AIS regional axonal width, the width of the axon of neurons axotomized more than 30μm from the soma was measured at 30μm using the straight-line tool within ImageJ. For neurons axotomized within 30μm of the soma, the width was measured at its widest point along the axon excluding the retraction bulb itself.

For experiments examining the AIS-regional distribution and superstructure of ankyrin-G and Na_V1.6, one day after injury mice were perfused through the aorta with 0.1M PBS until exsanguination followed by 8 min of 4% PFA. Brains were post-fixed in 4% PFA for 150 min, washed six times in 0.1M PBS, and sectioned at 50 μm via a Leica VT1000S vibrating microtome. Slices of the primary somatosensory cortex ventral to the site of injury induction were washed three times in 0.1M PBS and immersed in 10 mM sodium citrate dihydrate (294.10 g/mol; pH = 8.0) heated on a block to 80°C for 12 min.

Slices were allowed to cool for 10 min, washed twice, blocked for 1h at room temperature, then reacted with solutions containing 0.5% triton-X, 1% NHS and primary antibodies to ankyrin-G (1:250; mouse IgG2a, clone N106/36, NeuroMab, cat# 73-146) and Na_V1.6 (1:250; rabbit polyclonal, Alomone, cat#ASC-009) for at least 18h. Subsequently, slices were washed three times and reacted with appropriate secondary antibodies (1:500) at room temperature for 2h. After a final wash, slices were mounted on glass slides via Vectashield non-hardening anti-fade mounting medium and sealed with clear nail polish.

Slices thus prepared were imaged via sequential scanning at Nyquist criteria by a Zeiss 880 laser scanning confocal microscope, with the pinhole of the channel corresponding to Na_V1.6 set to one Airy unit. Subsequently, cells of interest were imaged via the Zeiss Elyra 7 lattice structured illumination microscope to achieve superior resolution of the AIS-regional membrane distribution of these protein species. Microscopy was performed at the VCU Department of Anatomy and Neurobiology microscopy facility, supported, in part, with funding from the NIH-NCI Cancer Support Center grant P30 CA016059. For pixel intensity analysis, using the freehand tool within Fiji, lines were drawn along the AIS within tiffs corresponding to the channel of interest and quantified via the “Plot Profile” function.

Data analysis

Data analysis was performed with Clampfit software (Molecular Devices, Sunnyvale, CA) and custom programs in Visual Basic for Applications, within Microsoft Excel (Microsoft, Redmond, WA). For the APs generated for 2^o analyses, in all cases only the first AP in the sweep was analyzed. The digitized AP was filtered with a 7-point boxcar filter repeated three times before calculation of the first derivative and the 2^o. The AP threshold was measured as the membrane potential at 10 mV/msec. For each cell typically 5 APs were analyzed (3-5APs), with measurements made of each individual AP and then those measurements averaged to create the record per cell.

Unless otherwise specified, significance was tested using one or two-way analysis of variance (ANOVA), with Bonferroni post-hoc test (SPSS, IBM). Results are reported as mean ± standard error of the mean (SEM).

Isolation of the AIS and somatic contributions to the AP upstroke

The 2^o analysis of the AP upstroke has been demonstrated to contain two components (Cp1, Cp2), which have been shown to represent the sequential depolarization of the AIS and the soma, respectively, in CA3 pyramidal neurons ²⁶ and cerebellar Purkinje cells. ²⁷ To extend this analysis to layer 5 neocortical pyramidal neurons and to confirm that our methods allowed the differentiation of AP generation at the AIS and the soma, APs were recorded before, during, and after the focal application of TTX (2 μM) to either the axon at the expected location of the AIS (30 μm distal to soma) or to the soma (Fig. 1A,B).

In sham controls, the 2^o of the AP always contained two components with clearly separate peaks (Fig. 1C). Application of TTX to the AIS reduced the peak amplitude of Cp1 to a greater degree than that of Cp2 (Fig. 1C–E). Continued application of TTX produced APs with a gradually diminishing Cp1, which ultimately was abolished outright, giving rise to a monophasic waveform. Within seconds of cessation of TTX application, 2^o analysis again demonstrated two peaks. This process of eliminating the first peak then allowing its recovery could be repeated within the same neuron.

Before elimination of Cp1, TTX application also decreased the time between component peaks and depolarized the AP threshold potential (Fig. 1E,F). Just before elimination of Cp1, the separation of the peaks became less apparent, such that Cp1 and Cp2 began to meld into one another, giving rise to a shoulder and a peak (Fig. 1C, 20 sec). To quantify this transition from double peaks to a single peak, we used a new measure in which the 2^o was first normalized to the second (or only) peak. A line was then drawn from 99% to 30% of the second (or only) peak (green lines in Fig. 1C). The difference between the actual data and this theoretical line was then calculated for each point and summed for all points creating the summed difference from linear slope (LS_diff).

In cases of two clear peaks separated by a negative intervening slope, the summed difference from this linear slope was always greater than 2.25. The LS_diff dropped below this threshold in cases of a shoulder+peak and in cases of a single peak. These results were consistent in all neurons tested, whether from sham controls or the IN1D group (Fig. 1G–K).

Movement of the TTX puffer to the location of the soma reduced the amplitude of Cp2, but typically did not eliminate it (Fig. 2). Because the puffer electrode tip size was small relative to the soma, the entire soma could not be covered by TTX application; thus, we did not expect to abolish Cp2 outright. While targeting the soma, some spillage of TTX onto the region of the AIS did occur, as the aCSF flowed toward the AIS. Consequently, Cp1 was also affected; however, the effect was greatest on Cp2 (Fig. 2C, D). This result can be appreciated on normalization of the 2^o to the Cp1 peak (Fig. 2B).

The TTX application to the soma gave rise to the converse effects compared with TTX application to the AIS on measures of the Cp1/Cp2 ratio, time between peaks and LS_diff. Further, there evolved a deepening of the valley between the Cp1 and Cp2 peaks, consistent with an increase in the spatial separation between each zone of activation. To quantify this phenomenon, we measured the minimum value between peaks after normalization to the Cp1 peak (Fig. 2E, right axis, brown squares, 2G). The reduction of this minimum was typical for all neurons tested with TTX application to the soma.

From these results, we considered the 2^o Cp1 to represent depolarization of the AIS and Cp2 to represent depolarization of the soma. Thus, hereafter Cp1 will be referred to as the AIS-regional peak acceleration of the membrane potential and Cp2 will be referred to as the soma-regional peak acceleration of the membrane potential.

Our remaining results describe the effect of mTBI on the function of the AIS and AP generation. Sham1D and 2D refer to the survival times of one or two days after the sham injury. Experimental groups AX1D and IN1D were taken from the same animals, surviving to one day after injury, but with the endogenous YFP allowing the state of the axon to be identified before recording as axotomized (AX, having a disconnected axon and swelling) or intact (IN) down to the subcortical white matter with no swelling. The same pre-recording identification was performed for AX2D and IN2D but on survival day 2 after the injury.

AIS peak is reduced or lost after mTBI in both intact and axotomized neurons

Injection of square wave depolarizing current gave rise to a qualitatively similar AP in sham controls and in axotomized and intact neurons one and two days after TBI (Fig. 3). Calculation of the first derivative of the AP revealed a change in slope of the AP rise to peak for sham neurons. This change in slope was present in some TBI neurons but not in others (Fig. 3B). This result was more evident when plotted against the membrane potential and when compared to the 2^o (Fig. 3C, D). When the 2^o was double-peaked, a notch could typically be seen in the plot of the first derivative versus membrane potential, indicating a distinct threshold potential for AIS versus soma activation zones.

OPEN IN VIEWER

FIG. 3.

Examples of recordings from each subject group. Waveforms labeled with # indicate that they are shown at 2X vertical scale. (A) Action potential (AP) produced by brief depolarizing pulse. (B) First derivative of the AP shown in A. (C) First derivative as in B but plotted against membrane potential. (D) 2^o of AP shown in A. All shams had double-peaked 2^o, some neurons from the mild traumatic brain injury (mTBI) AX1 and 2D groups had single-peaked 2^o, while others had a shoulder and then a peak, and some were double-peaked. The mTBI IN1 and 2D groups had some cells with a shoulder and then a peak and some clear double-peaked 2^o.

Because concussive TBI-induced axotomy manifests at the AIS within the majority of axotomized neurons, ⁷ we expected that the AIS function would be perturbed in axotomized neurons, especially those axotomized at the region of the AIS. We discovered a subset of the AX1D and AX2D neurons in which there was only a single peak, and another group in which the first peak was not clearly separated from the second peak, creating a shoulder melding into a peak (see examples in Figs. 3 and 6).

A new measure (LS_diff) distinguishes the cells with an absent AIS 2^o peak

Because of the variability in the shapes of the 2^o for TBI neurons, we sought a quantitative means to distinguish single- versus double-peaked 2^o waveforms. This led to the development of the LS_diff measure (see description in text above for Fig. 1). The threshold line of 2.25 separating single-peaked 2^o below that level and double-peaked 2^o above it was determined in our TTX experiments that eliminated the AIS peak (see Fig. 1).

This measure is shown for experimental groups in Fig. 4, where this quantitative measure agreed with qualitative classification: All single-peaked neurons shown as triangles fell below the 2.25 threshold while all double-peaked neurons shown as squares fell above this threshold. For the neurons that had a 2^o with a shoulder+peak (Fig 4, circles), many had LS_diff values that fell below the threshold, similar to the single-peaked 2^o, while others were just above this threshold.

OPEN IN VIEWER

FIG. 4.

Measurement that differentiates between single and double-peaked 2^o. The sum of the difference of actual values from the linear slope calculated from 99% of the second or only peak to 30% of that peak (LS_diff). Individual neurons in solid symbols and mean ± standard error of the mean in open symbols. All double-peaked 2^o fall above the 2.25 line (purple dashed line), and all single-peaked 2^o fall below this line. Shoulder+peak neurons tend to straddle the line. Subject numbers are reported as #neurons(#slices)[#mice]: Sham1D 23(16)[6]; Sham2D 17(13)[4]; TBI_AX1D 23(20)[8], and within this subject group 10 neurons had a double-peaked 2^o, eight had a shoulder+peak, and five were single-peaked; TBI_AX2D 15(13)[3], and within this subject group, 10 neurons had double-peaked 2^o, four had a shoulder+peak, and one was single-peaked; TBI_IN1D 23(21)[11], and within this subject group 15 neurons had a double-peaked 2^o, and eight had a shoulder+peak; TBI_IN2D 14(11)[4], and within this subject group 12 neurons had a double-peaked 2^o and two had a shoulder+peak.

A two-way ANOVA found no significant difference for LS_diff between subject groups, but p < 0.001 for the number of peaks, where shoulder+peak was considered 1.5 peaks. There was also no significant interaction between subject group and peak number. A Bonferroni post-hoc showed that LS_diff for double-peaked 2^o was significantly greater than that for both the single-peaked and the shoulder+peak .

Timing of single-peaked 2^o corresponds to the somatic peak

We next sought to determine whether the single-peaked 2^o corresponded to the AIS or somatic peak or something in between, by assessing the time from AP threshold to the time of the 2^o peak(s) (Fig 5). For the sham neurons, the AIS 2^o peak always occurred less than 0.144 msec (dashed line in Fig. 5) after AP threshold. This was also true for all but one of the TBI neurons with double-peaked-2^o (see filled diamonds in Fig. 5A).

OPEN IN VIEWER

FIG. 5.

Most axon initial segment (AIS) or first peaks of the 2^o occur less than 0.144 msec after threshold (dashed line), while most somatic or 2nd 2^o peaks occur later than this. The single-peaked 2^o occur as late as or later than the soma time to peak. Time to peak for individual cells shown in (A) and means ± standard error of the mean shown in (B). Color indicates subject group, as labeled. The AIS peaks are shown as diamonds, filled for double-peaked and open for shoulder+peak 2^o. The somatic peak is shown as filled squares for double-peaked and open circles for shoulder+peak 2^o. Single-peaked 2^o are filled triangles. Subject numbers are reported as #neurons(#slices)[#mice]: Sham1D 23(16)[6]; Sham2D 17(13)[4]; TBI_AX1D 23(20)[8], and within this subject group 10 neurons had a double-peaked 2^o, eight had a shoulder+peak, and five were single-peaked; TBI_AX2D 15(13)[3], and within this subject group, 10 neurons had double-peaked 2^o, four had a shoulder+peak, and one was single-peaked; TBI_IN1D 23(21)[11], and within this subject group 15 neurons had a double-peaked 2^o, and eight had a shoulder+peak; TBI_IN2D 14(11)[4], and within this subject group 12 neurons had a double-peaked 2^o and two had a shoulder+peak.

For shams, the somatic peak always occurred after this time of 0.144 msec. The same was true for the somatic peak for all but one of the TBI neurons with double-peaked-2^o (Fig. 5A, filled squares). Thus, there was a clear time separation between AIS and somatic peaks. For the shoulder+peak neurons, the timings were similar to the double-peaked 2^o neurons; that is, the shoulder (open diamonds in Fig. 5A) occurred less than 0.144 msec from AP threshold in all but one neuron. The peak for the shoulder+ peak neurons occurred in a time consistent with the somatic peak (open circles in Fig. 5A) for most neurons.

Within the IN1D group, however, several occurred slightly before this time (see open light blue circles in Fig. 5A). Importantly, the timing of single-peaked 2^o (filled triangles in Fig. 5) was most consistent with the timing of the somatic (second) peak; in fact, on average, the timing of the single peak times was even later than that of the second peak for double-peaked 2^o.

To examine this phenomenon further, the time of the second or only peak was statistically compared across neurons within the AX1D group. A one-way ANOVA showed a significant effect of peak number on relative time to peak, and the Bonferroni post-hoc showed that the time of the peak for single-peaked neurons was significantly greater than that of the second component of the double-peaked and the shoulder+peak neurons, suggesting that in the absence of an AIS-generated AP, more time is required to discharge the soma.

The slope between the 2^o peaks distinguishes the shoulder+peak neurons

We next sought a more quantitative means of identifying the shoulder + peak 2^o waveforms (see examples in Fig. 6A). Qualitatively, it appeared that the slope between the peaks was positive for the neurons manifesting a shoulder and might specifically distinguish them. Thus, we calculated the slope of the 2^o from the AIS peak (or shoulder) to the midpoint between the peaks (Fig. 6B).

OPEN IN VIEWER

FIG. 6.

Some 2^o waveforms manifest a shoulder before a peak. (A) Examples of 2^os with a shoulder (arrowhead) before the peak. (B) Slope of a linear line from the first peak to the midpoint between peaks is an effective measure to identify these shoulder neurons. Colors indicate the subject group. The 2^o with double peaks are shown as squares, while those with a shoulder+peak are shown as circles. All of the cells with a shoulder had positive slopes, while all of the non-shoulder cells had negative slopes. Filled circles are from individual neurons, while empty circles shown mean + standard error of the mean. Note this measure cannot be made on cells with a single peak. Subject numbers are reported as #neurons(#slices)[#mice]: Sham1D 23(16)[6]; Sham2D 17(13)[4]; TBI_AX1D 23(20)[8], and within this subject group 10 neurons had a double-peaked 2^o, eight had a shoulder+peak, and five were single-peaked; TBI_AX2D 15(13)[3], and within this subject group, 10 neurons had double-peaked 2^o, four had a shoulder+peak, and one was single-peaked; TBI_IN1D 23(21)[11], and within this subject group 15 neurons had a double-peaked 2^o, and eight had a shoulder+peak; TBI_IN2D 14(11)[4], and within this subject group 12 neurons had a double-peaked 2^o and two had a shoulder+peak.

For all sham and most IN2D neurons, this slope was negative, allowing the peaks to be unambiguously distinguished. This measure could only be made on neurons whose 2^o manifested either double peaks or a shoulder and a peak; consequently, a subset of neurons from the AX1D and AX2D groups with a single peak did not qualify. Eight of the 23 neurons within the AX1D group had positive slopes (Fig 6) and the shoulder appearance. We note that a majority of neurons in the AX1D group had either a single peak or a shoulder+peak (Fig. 7A). The numbers were smaller for the other TBI groups, but all had some “shoulder” neurons, indicating consistently altered AIS function for all TBI groups.

OPEN IN VIEWER

FIG. 7.

(A) Proportion of neurons having double, shoulder+peak, or single peak(s) in the 2^o of the action potential (AP). Subject numbers are reported as #neurons(#slices)[#mice]: Sham1D 23(16)[6]; Sham2D 17(13)[4]; TBI_AX1D 23(20)[8]; TBI_AX2D 15(13)[3]; TBI_IN1D 23(21)[11]; TBI_IN2D 14(11)[4]. (B) The 2^os for all APs from a cell were normalized to the 2nd (or only) peak, aligned at the time of that peak and then averaged. The average was then taken for each subject group. The vertical lines indicate the standard error of the mean of this mean. Subject groups are: Sham1D (a), Sham2D (b), AX1D (c), AX2D (d), IN1D (e), IN2D (f). Z-tests showed that the TBI_AX1D and TBI_IN1D had significantly fewer double-peaked 2^o neurons compared with Sham1D, and that the TBI_AX2D had significantly fewer double-peaked 2^o neurons than Sham2D.

The degree of these changes can be seen from the waveforms averaged across all neurons within a subject group (Fig. 7B). Here the waveforms were normalized to the second (or only) peak and aligned to the time of this peak and then averaged across all cells within the group. For groups containing many shoulder neurons (AX1D, IN1D), this produced a relatively flat region of the curve where the expected AIS peak would occur (Fig. 7Bc, e).

By the second day after injury, the axotomized group showed a smaller subset with either a single peak or shoulder+peak (Fig. 7A), resulting in greater variability in the normalized AIS peak amplitude for the averaged waveform of this group (Fig. 7Bd, see error bars). For groups AX1D, AX2D, and IN1D, significantly fewer neurons had double peaks with a negative intervening slope, compared with the corresponding shams (z-tests, p < 0.05).

Decreased amplitude of 2^o peaks suggests reduced sodium current density after injury

The analyses up to this point allowed classification of the neurons and proper assignment of the single-peaked 2^o neurons (somatic peak) for quantitative measurement of the peak voltage acceleration of the 2^o. To examine the overall effects of concussive brain trauma on the relative voltage acceleration during the AP upstroke at the AIS and the soma, the peak amplitudes of the 2^o for AIS and somatic locations were examined. The 2^o peak amplitude at the AIS was significantly lower than sham for both AX1D and IN1D groups (Fig. 8A, ANOVA, Bonferroni post-hoc, p < 0.05).

OPEN IN VIEWER

FIG. 8.

The 2^o peak values suggest that action potential (AP) generation at both the axon initial segment (AIS) and the soma are altered for some experimental groups. (A) AIS 2^o peak amplitude. * = analysis of variance (ANOVA), with Bonferroni post-hoc, p < 0.05. (B) Soma 2^o peak amplitude. *indicates that these groups are significantly different from all three other groups (ANOVA, Bonferroni post-hoc, p < 0.05). (C) Action potential threshold was more depolarized in the AX groups at one and two days and in the IN1D group compared with the Shams and IN2D group. *indicates that these groups are significantly different from all three other groups (ANOVA, Bonferroni post-hoc, p < 0.05). (D) The sum of AIS and soma peaks shown per neuron. Neurons with double-peaked 2^o shown as squares, neurons with a 2^o containing a shoulder and peak are shown as circles, and those with a single peak are shown as triangles. Mean ± standard error of the mean shown in open symbols with error bars. Note that in the TBI_AX1D group, the single-peaked 2^o were similar in amplitude to the sum of both peaks for the double-peaked 2^o. For all data shown in this figure, the subject numbers are reported as #neurons(#slices)[#mice]: Sham1D 23(16)[6]; Sham2D 17(13)[4]; TBI_AX1D 23(20)[8]; TBI_AX2D 15(13)[3]; TBI_IN1D 23(21)[11]; TBI_IN2D 14(11)[4]. Because five neurons in the TBI_AX1D group and one neuron in the TBI_AX2D group had no AIS peak, those neurons could not be measured in part A. Thus, for A, the subject numbers for those two groups were: TBI_AX1D 18(15)[7]; and TBI_AX2D 14(12)[3].

While we previously saw a decrease in the IN2D group compared with sham, ²⁴ this result was not replicated in this larger cohort. The somatic 2^o peak amplitude was also significantly lower than sham for both AX1D and IN1D groups, as well as for the AX2D group (Fig. 8B, ANOVA, Bonferroni post-hoc, p < 0.05). This difference returned to sham levels for the IN2D group. The three TBI groups showing reduced soma-regional AP acceleration also had a significantly more depolarized AP threshold that returned to near-sham levels for the IN2D group (Fig. 8C, ANOVA, Bonferroni post-hoc, p < 0.05).

When the sum of the AIS- and soma-regional 2^o peak amplitudes was calculated, presumably reflecting the total sodium current density during the AP upstroke, the sums for the AX1D, AX2D, and IN1D groups were significantly less than sham and IN2D groups (Fig. 8D, ANOVA, Bonferroni post-hoc, p < 0.05). We also observed, however, that the amplitude of single-peaked 2^o was similar to that of the sum of the two peaks of double-peaked 2^o within the AX1D group (Fig. 9A).

OPEN IN VIEWER

FIG. 9.

Effect of 2^o peak number and shape for neurons from the AX1D group. In all cases a one-way analysis of variance with Bonferroni post-hoc was performed, * = p < 0.05, with subject numbers reported as #neurons(#slices)[#mice], 10(9)[6] double-peaked, 8(7)[5] shoulder + peak, and 5(5)[3] single-peaked neurons. (A) The 2^o amplitude of the somatic (or only) peak is shown in gray, and the combined peak amplitude of the axon initial segment (AIS) and somatic (or only) peak(s) is shown in white. The total 2^o peak amplitude is not different for single-peaked neurons compared with that of double-peaked neurons or compared with those with a shoulder+peak. The somatic peak alone is significantly higher for the single-peaked neurons compared with the other groups. This result suggests that the AIS generation site may have moved toward the somatic generation site such that the two are no longer differentiated, or that abolition of AIS function gives rise to a homeostatic upregulation of somatic sodium channels. (B) The threshold for AP generation was significantly different for the single-peaked neurons compared with the groups. (C,D) There were no significant differences in resting membrane potential (RMP) or input resistance among these groups, suggesting that single-peaked neurons are healthy by conventional criteria. (E) The frequency of APs generated during a 400 msec depolarizing pulse at 200 pA was significantly less in the single-peaked compared with double-peaked neurons. (F) Rheobase, or the lowest depolarizing current to generate an AP was not significantly different for these groups. (G) The slope of the plot of AP frequency versus injected current level was significantly lower for single-peaked neurons compared with double-peaked neurons. (H) The time to first AP at rheobase was not significantly different across the different peak shapes, but there was a trend toward an increase for the shoulder+peak group compared with two peaks.

This result suggested that the AIS and somatic generation zones may approach each other more closely. If so, it follows that the peak amplitude of the single-peaked 2^o of the AX1D group should be larger than that of the somatic peak of the double-peaked and shoulder+peak 2^o in the AX1D group. We find this result to be the case (Fig. 9A, ANOVA, Bonferroni post-hoc, p < 0.05).

The impairment of AIS-regional AP generation in the AX1D, AX2D, and IN1D groups did not arise as a function of unhealthy neurons as judged by conventional criteria. The input resistance in these groups was similar to that of the sham and IN2D groups (Table 1), although the resting membrane potential (RMP) was slightly more depolarized in these groups (Table 1). In addition, within the AX1D subject group, the single-peaked or shoulder+peak neurons were not significantly different from the double-peaked AX1D neurons on RMP and input resistance (Fig. 9C, D), suggesting the differences in the shape of the 2^o cannot be explained by differences in global metrics of neuronal health.

Abolition of the AIS peak results in reduced overall neuronal excitability

We next sought to correlate the consequences of trauma-induced AIS-regional functional perturbation to the associated AP discharge patterns of affected neurons. We observed differences in the AP discharge patterns of single-peaked and shoulder+peak neurons compared with double-peaked neurons within the AX1D subject group. In particular, the AP threshold for single-peaked AX1D neurons was significantly depolarized compared with either shoulder+peak or double-peaked AX1D neurons (Fig. 9B), yet the rheobase was not significantly different between these three AX1D groups (Fig. 9F).

The single-peaked AX1D neurons discharged APs at a significantly lower frequency in response to 200 pA depolarizing current injection compared with double-peaked AX1D neurons (Fig. 9E, ANOVA, Bonferroni post-hoc, p < 0.05). In addition, the slope of the plot of AP frequency versus current injection was significantly lower in single-peaked compared with the double-peaked neurons, with a similar trend in a lower value for shoulder+peak compared with double-peaked (Fig. 9G, ANOVA, Bonferroni post-hoc, p = 0.054). These results suggest that the single-peaked neurons are less excitable.

The time to the first AP at rheobase was not significantly different between these three peak shapes for AX1D neurons, but there was a trend toward a longer time for the shoulder+peak group compared with double-peaked neurons (Fig. 9H, ANOVA, Bonferroni post-hoc, p = 0.072).

Because the pattern of action potentials varies for different firing types of RS, RSd, and IB and that can affect some of these intrinsic/cellular measures, the proportion of different 2^o peak numbers for each firing type was determined. Double-peaked and shoulder+peak 2^o forms were found for all three firing types of cells (RS, RSd and IB, Supplementary Fig. S3). Single-peaked 2^o were found only in RS and RSd types. We also sought to determine whether the significant effects shown in Figure 9 were attributable to firing type instead of 2^o peak number.

To examine this, two-way ANOVAs for firing type and peak number were performed for the AX1D subject group. Note other subject groups did not have a full complement of all peak numbers. The results of these twp-way ANOVAs are shown in Supplementary Table S1, where there were no significant effects of neuronal firing type nor any significant interactions between neuronal firing type and 2^o peak number. For this reason, we feel confident that the effects shown in Figure 9 are associated with 2^o peak number.

Abolition of the AIS 2^o peak is associated with morphological change of the axon

In examining the filled neurons that had shown a single-peaked 2^o, we noted a specific morphological aberration in the form of an AIS-regional stump. All of the neurons manifesting a single peak were characterized by a wider, distended AIS and diminished staining intensity of ankyrin-G compared with neighboring neurons (Fig. 10, C1,C2, also see 10G; Fig. 11 D,E).

OPEN IN VIEWER

FIG. 10.

Axon initial segment (AIS) morphology and associated action potential (AP) upstroke kinetics for sham and axotomized mild traumatic brain injury (mTBI) neurons. Examples of one neuron recorded within the sham condition (1) three neurons axotomized near the soma (2, 3, 5), and one far from the soma (4). (A) Targeted neurons as visualized by endogenous YFP expression within the living brain slice just before recording. The inset in yellow shows the 2^o of the AP recorded from the same cell. (B) Maximum intensity projections of the entire image stacks of the same neurons in A after recovery of biocytin (red) and immunohistochemical staining of ankyrin-G (violet). Endogenous YFP fluorescence is green. (C–E) Morphological details of the AIS of neurons 1–5 imaged at 63X. Arrows within D indicate the AIS of recorded neurons as visualized by immunohistochemical labeling of ankyrin-G, while E depicts the merged image of ankyrin-G and endogenous YFP. For neurons 2 and 3, single optical slices are displayed. For neurons 1, 4, and 5, maximum intensity projections of subsets of the complete image stack exhibiting the entire AIS in optimal clarity are displayed. Note that neurons 2 and 3 manifest a diminished intensity of staining for ankyrin-G and a wider, distended AIS. These morphological properties correlate with an absence of discernable AIS-regional voltage acceleration during the AP upstroke. Note also the filipodia- or spine-like sprouts emanating from the AIS of neurons 2 and 3 (arrowheads in C). Neuron shown in 2 here is the same as that from which recordings are shown in row 3 of Fig. 3, while neuron 4 here is the same as row 6 of Fig. 3, and neuron 5 here is the same as row 4 in Fig. 3. Scalebar in 1B = 20 μm, for all images in column B. Scalebar in 1E = 10 μm for all images in columns C-E. (F–H) Subject numbers reported as #neurons(#slices)[#mice]: measurements made on 15(13)[4] all within the AX1D subject group so that survival time and condition would not contaminate the comparison between 2^o peak types. Both F and H show that distance from soma to axotomy does not predict the absence or presence of the AIS peak. G suggests that the increased width of the axon at the AIS does predict loss of AIS peak (1-peaked 2^o).

OPEN IN VIEWER

FIG. 11.

Traumatically axotomized layer 5 pyramidal neurons exhibiting axon initial segment (AIS)-regional distention stained less intensely for ankyrin-G and NaV1.6 compared with intact counterparts within the vicinity. (A) Traumatically axotomized neuron (neuron x) in the vicinity of an intact counterpart (neuron y). Note the persistence of the disconnected distal portion of the axon from neuron x (yellow arrowhead). White arrows point to axon of neuron x, while orange arrows point to axon of neuron y. Neuron x exhibits diminished staining intensity for ankyrin-G (B) and NaV1.6 (C) compared with neuron y. Scale bar in C for A–C is 20 μm. (D) Elyra-7-lattice structured illumination microscopy reveals fine-scale structure of the organization of ankyrin-G (cyan) and NaV1.6 (magenta) along the AIS of neuron x (D1-3) and neuron y (D4-6). Note the fragmentation of ankyrin-G (yellow arrowheads in 1) and smaller, dimmer conglomerations of NaV1.6 channels (yellow arrowheads in 2) along the AIS of neuron x. D7-9 depict Elyra 7 super-resolution micrographs of a neuron from the sham condition. Scale bar in Elyra-7 images is 2 μm. (E) Quantification of pixel intensity along the AIS region versus distance from the soma for neuron x. (F) Quantification of pixel intensity along the AIS region versus length from the soma for neuron y. Cyan corresponds to ankyrin-G, and magenta corresponds to NaV1.6. (G–I) The same quantification shown in E and F was performed on a population of axotomized neurons exhibiting a club-shaped, distended AIS (G), axotomized neurons without AIS-regional distention (H), and for intact neurons (I). Darker lines are the mean with lighter vertical lines of the same color showing the standard error of the mean. (J) Ankyrin-G data shown in H and I is graphed in a single plot so cell types can be directly compared. Subject numbers reported as (#neurons[#mice]). Club-shaped axotomized neurons in peach (n = 6[3)), classical axotomized neurons in bright green (n = 4[1]), and intact neurons in gray (n = 5[2]). (K) Colors have same definition as J, but for NaV1.6 staining. Subject numbers #neurons[#mice]: club-shaped 3[2], classical 4[1], and intact neurons 2[2].

To directly compare the potential effects of 2^o peak number, we quantified measurements of distance from soma to axotomy and width of the AIS for 15 neurons within the AX1D group (Fig. 10F–H). We hypothesized that axotomy within the AIS might account for the single-peaked 2^o, while more distant axotomy might allow for a double-peaked 2^o within the AX1D subject group. Indeed, all of the single-peaked AX1D neurons were axotomized in the vicinity of the soma (Fig. 10F). Some of the double-peaked AX1D neurons, however, were axotomized within the AIS and 40 μm or less from the soma (Fig. 10F). This result thus suggests that intracellular signaling cascades rather than mechanical disruption per se affect that ability of the AIS to generate an AP after trauma.

The apparent width of the AIS was measured 30 μm from the soma for neurons axotomized beyond 30 μm, while for those axotomized within 30 μm of the soma, the apparent width was measured at its widest point along the axon excluding the retraction bulb (Fig. 10G). The single-peaked 2^o neurons had a significantly wider AIS compared with the double-peaked and shoulder+peak neurons (ANOVA, Bonferroni post-hoc, p < 0.05). The apparent width of the AIS was not significantly correlated with the distance from soma to axotomy (Fig. 10H).

Morphologically altered axons are associated with reduced ankyrin-G and Nav1.6 staining

Within the morphological subclass corresponding to neurons manifesting a single peak, we sought to correlate the diminished staining intensity of ankyrin-G to the AIS-regional structure of Na_V1.6 channels. To this end, we imaged ankyrin-G and Na_V1.6 simultaneously within animals processed for standard immunohistochemistry one day after injury. We found that both protein species stain less intensely at the AIS in the morphological subclass corresponding to single-peaked neurons compared with intact neurons within the vicinity (Fig. 11).

We next employed lattice structured illumination microscopy to resolve the AIS-regional fine-scale organization of these protein species beyond the diffraction barrier of conventional confocal microscopy. This technique allowed the discernment of apparent fragmentation of the architecture of ankyrin-G within an injured neuron of the morphological subclass corresponding to neurons manifesting a single peak (Fig. 11D1). Simultaneously, the puncta representing conglomerations of Na_V1.6 channels were decreased in apparent area and intensity (Fig. 11D2). These findings were consistent across neurons manifesting the distended club-shaped axon indicative of those with one 2^o peak (Fig. 11G). Both axotomized without the club shape and intact neurons showed more intense staining for these markers and a greater length of staining (Fig. 11 G-K).

These experiments constitute evidence that neurons of this morphological subclass represent neurons that have sustained an especially profound form of axonal injury.

In the same population of neurons shown in Fig. 11, we analyzed the distance from the soma to the start of ankyrin-G and Nav1.6 staining and the distance to the termination of staining (Supplementary Fig. 4). While there was no significance difference between club-shaped, classical axotomized and intact groups for the distance from soma to ankyrin-G start, there was a significance difference from the soma to the ankyrin-G termination of staining for the club-shaped compared with both the classical axotomized and the intact groups (ANOVA, Bonferroni post-hoc, p < 0.05). There was no significant difference in either distance for the Nav1.6 staining (Supplementary Fig. 4).

Loss or reduction of the AIS 2^o peak after mTBI is not observed in CypDKO mice

We previously found that aberrations of intrinsic cellular and membrane properties were prevented or ameliorated when experiments were conducted on brains extracted from CypDKO mice subjected to the equivalent concussive TBI. ³¹ Here, we examined whether CypDKO might mitigate trauma-induced effects on AIS- and soma-regional voltage acceleration. Within CypDKO mice, we found that even when axotomy occurred just proximal to the soma (not shown), the AP 2^o still manifested double peaks, or in a few cases, a shoulder+peak (Fig. 12). Single-peaked 2^o were not observed in any neurons from CypDKO mice. In fact, the overall shape of the 2^o was fairly similar in axotomized and intact groups compared with the shams (Fig. 12A, also compared with WT shown in Fig. 7).

OPEN IN VIEWER

FIG. 12.

Measures in CypDKO tissue. (A) Effects of mild traumatic brain injury (mTBI) on the morphology of 2^o waveforms in CypDKO tissue. The 2^os for all action potentials (APs) from a cell were normalized to the second (or only) peak, aligned at the soma peak of the 2^o, and then averaged. For each subject group, the average across cells was then taken, with the vertical lines indicating the standard error of the mean (SEM). Subject numbers for this figure, reported as #neuron(#slices)[#mice]: (a) CypDKO_Sham1D 7(6)[2]; (b) CypDKO_Sham2D 8(7)[2]; (c) CypDKO_AX1D 11(9)[3]; (d) CypDKO_AX2D 6(5)[2]; (e) CypDKO_IN1D 13(8)[3]; (f), CypDKO_IN2D 7(7)[2]. (B) For neurons recorded in slices from CypDKO animals, proportion of neurons having double, shoulder+peak, or single peak(s) in the 2^o of the AP, numbers of neurons as indicated in A. (C) Slope from the axon initial segment (AIS) peak to the midpoint between AIS and soma peaks. Means ±SEM shown in open symbols with error bars. All but four of the CypDKO neurons' 2^o had negative slopes. (Dtwo Summed difference from linear slope for CypDKO groups, mean ± SEM shown in open symbols with error bars. Here only two neurons' 2^o had a summed difference from linear slope below the threshold of 2.25.

The measure allowing identification of the shoulder+peak 2^o neurons (slope from AIS peak to midpoint between the peaks) reveals only four neurons with positive slopes measured from the AIS peak to the midpoint between the AIS and somatic peaks (Fig. 12C). Among those four shoulder+peak CypDKO neurons, only two fell just below the 2.25 threshold on the LS_diff measure (Fig. 12D). Thus, CypDKO was associated with a striking amelioration of trauma-induced perturbation of AP upstroke kinetics (compare Figs: 7B,A and 12A,B; 4 and 12D; 6 and 12C).

Neuronal excitability and the distance between AIS and somatic activation zones is altered after TBI in CypDKO compared with WT

To compare further between WT and CypDKO on a series of measures (Fig. 13), two-way ANOVAs with Bonferroni post-hoc tests were performed, with the main effects listed in Table 2. For the number of cells in each group, see Table 2. For the post-hoc interactions between subject group and the experimental conditions of WT or CypDKO, significance is shown in Fig. 13, with * denoting p < 0.05.

OPEN IN VIEWER

FIG. 13.

Comparison between WT and CypDKO experimental conditions for all subject groups. WT is in lighter gray and CypDKO is in darker gray. Subject groups are marked at the bottom of the columns. For each measure, a two-way analysis of variance was performed. See Table 2 for subject numbers and p values for subject group, experimental condition, and interactions. When a significant interaction between subject group and experimental condition was present, a Bonferroni post-hoc analysis was performed, with significant effects indicated here by an *.

For the AIS peak amplitude there was a significant effect of subject group, with AX1D and IN1D being significantly lower than Sham1D. The AIS peak for AX2D as well as AX1D and IN1D was significantly lower than Sham2D. For the IN2D group, the AIS peak was significantly different from AX1D and IN1D.

The interaction between subject group and experimental condition was also significant, with the post-hoc test showing the following: The reduced 2^o for the AIS peak was recovered in the CypDKO mice only for the IN1D group. Surprisingly, while the AIS peak was increased in CypDKO compared with WT for the IN1D neurons, it was significantly decreased in CypDKO compared with WT for the IN2D neurons. It appears that restricting the assembly of the mitochondrial pore transition complex within CypDKO mice may be associated with non-linear effects on neuronal excitability. ³¹

For the AP acceleration at the soma (Fig. 13B), there was a significant effect of subject group, experimental condition, and the interaction between these two. The IN1D group had a significantly smaller soma-regional peak 2^o compared with the Sham1D and Sham2D groups. The IN2D group was significantly different from AX1D, AX2D, and IN1D groups. For the interaction between subject group and experimental condition, surprisingly there was an effect on the Sham2D neurons, with a reduction in the somatic 2^o peak in the CypDKO condition compared with WT. There was also a reduction in this peak under CypDKO condition for the IN2D group.

For the AP threshold (Fig. 13C), there was a significant effect of subject group, no effect of experimental condition, but a significant interaction of subject group and experimental condition. The AX2D group had a significantly more depolarized AP threshold compared with the Sham2D group. For the interaction, the CypDKO condition ameliorated the more depolarized AP threshold for both AX1D and IN1D groups but exacerbated it in the AX2D group and produced a more depolarized AP threshold in the Sham2D group. The overall result of the CypDKO condition was a very similar AP threshold level between subject groups.

In previous examinations of the 2^o of the AP upstroke, authors have characterized the shoulder+peak 2^o waveform as possessing a range of bumpiness, depending on whether the slope of the 2^o passes through zero. ³⁵ The phenomenon of less bumpiness was attributed to a closer relative location of the AIS and somatic AP initiation zones. ³⁵ Here we made this measurement explicitly for the 2^o of each AP, then calculated the percent of APs per cell that were bumpy, which we defined as any 2^o waveform in which the intervening slope between AIS and somatic peaks passed through zero. This measure was then averaged among neurons of each subject groups (Fig. 13D).

For some groups, every single 2^o slope passed through zero, thus putting the average at 100% without error (WT: Sham1D, Sham2D, and IN2D; CypDKO: Sham2D, AX2D, and IN1D). The three experimental groups of the WT condition most affected by TBI (AX1D, AX2D, and IN1D) all contained some non-bumpy 2^o neurons. The two-way ANOVA showed no significant effect of subject group or of experimental condition, but did show an interaction between these two, with a significant increase in bumpiness for both AX1D and IN1D groups. Thus, within these groups the CypDKO condition gave rise to enhanced bumpiness, presumably because of a greater spatial separation between AIS and soma.

While the bumpiness quotient can only be measured in double-peaked 2^o, the LS_diff measure can determine how close the 2^o is to a single peak for all 2^o. This metric provided a more inclusive measurement of bumpiness (Fig. 13E). We found a threshold level at 2.25, where lower values were typical for single-peaked 2^o, and higher values were typical for double-peaked 2^o (Figs. 1F, 4). A one-way ANOVA with Bonferroni post-hoc for the WT condition showed a significantly lower LS_diff for the AX1D group compared with the Sham1D. This result was ameliorated in the CypDKO condition.

The two-way ANOVA considering both experimental conditions showed no significant effect of subject group but a significant effect of experimental condition, with an increase in LS_diff under the condition of CypDKO. There was a trend toward an interaction between subject group and experimental condition.

The examination of the CypDKO 2^o suggested that under this condition, there were more double-peaked 2^o and that the minimum between the peaks was lower than for the WT condition. We evaluated this by measuring the duration of time between the two peaks (only for double-peaked or shoulder+peak 2^o) during which the value was less than that of the AIS peak (Fig. 13F).

Here, there was a significant effect of experimental condition. Relative to the WT condition, CypDKO produced 2^o with greater durations of time during which the 2^o was less than that of the AIS peak. This result suggests that the current producing the AP at the AIS decays to a greater degree before the generation of the current producing the AP at the soma. Thus, this result supports the bumpiness quotient result in suggesting that the AIS and soma AP initiation zones are further apart in the CypDKO condition compared with WT.