Abstract
Spreading depolarizations (SDs) of gray matter occur in the brain in different pathologic conditions, and cause varying degrees of tissue damage depending on the extent of metabolic burden on the tissue. As might be expected for such large depolarizations, neurons exhibit bursts of action potentials (APs) as the wave propagates. However, the specific role of APs in SD propagation is unclear. This is potentially consequential, since sodium channel modulation has not been considered as a therapeutic target for SD-associated disorders, because of ambiguous experimental evidence. Using whole-cell electrophysiology and single-photon imaging in acute cortical slices from male C57Bl6 mice, we tested the effects of AP blockade on SDs generated by two widely used induction paradigms. We found that AP blockade using tetrodotoxin (TTX) restricted propagation of focally induced SDs, and significantly reduced the amplitude of neuronal depolarization, as well as its Ca2+ load. TTX also abolished the suppression of spontaneous synaptic activity that is a hallmark of focally induced SD. In contrast, TTX did not affect the propagation of SD induced by global superfusion of high [K+]e containing artificial CSF (ACSF). Thus, we show that voltage-gated sodium channel (Nav)-mediated neuronal AP bursts are critical for the propagation and downstream effects of focally induced SD but are less important when the ionic balance of the extracellular space is already compromised. In doing so we corroborate the notion that two different SD induction paradigms, each relevant to different clinical situations, vary significantly in their characteristics and potentially their response to treatment.
SIGNIFICANCE STATEMENT Our findings suggest that voltage-gated sodium channel (Nav) channels have a critical role in the propagation and downstream neural effects of focally induced spreading depolarization (SD). As SDs are likely induced focally in many disease conditions, these studies support sodium channel modulation, a previously underappreciated therapeutic option in SD-associated disorders, as a viable approach.
- brain injury
- migraine
- neuronal action potentials
- neuronal excitability
- spreading depolarizations
- spreading depression
Introduction
Various pathologic conditions can lead to the eruption of slowly propagating waves of near complete depolarization of neurons and glia. Depending on the bioenergetic and metabolic state of tissue through which they propagate, spreading depolarizations (SDs) can exert different deleterious effects, or leave no detectable damage (Dreier, 2011; Pietrobon and Moskowitz, 2014; Hartings et al., 2017; Brennan and Pietrobon, 2018). SDs arise from a cascading chain of events leading to the near-complete breakdown of ion hemostasis and release of glutamate and K+ into the extracellular space (Somjen, 2001; Pietrobon and Moskowitz, 2014; Herreras and Makarova, 2020). Although not fully corroborated by experimental evidence, modeling studies have identified K+ and glutamate released during the passage of an SD wave as the main drivers of the depolarization of adjacent cells, which in turn release K+ and glutamate, leading to the wave's self-sustaining nature (Tuckwell, 1981; Wylie and Miura, 2006; Zandt et al., 2013; Wei et al., 2014; Newton et al., 2018).
In normally metabolizing tissue, SDs are typically induced using depolarizing stimuli that increase [K+]e above a critical threshold (Pietrobon and Moskowitz, 2014). The depolarization during SD propagation in normally metabolizing tissue is preceded by a burst of action potentials (APs), both in vivo and in vitro, regardless of the induction paradigm (Sugaya et al., 1975; Somjen, 2001; Hosseini-Zare et al., 2017; Sawant-Pokam et al., 2017). Such synchronous neuronal AP bursts during prolonged neuronal depolarizations can result in significant glutamate and K+ efflux in the extracellular space (Barreto and Cressman, 2011; Raimondo et al., 2015) and can aid further depolarization of adjacent neurons.
Although a potential role for neuronal AP bursts in SD ignition and propagation could be inferred based on the evidence discussed thus far, the experimental evidence has been ambiguous. While some studies show that voltage-gated sodium channel (Nav) blockers suppress APs and mitigate SDs (Sugaya et al., 1975; Aitken et al., 1991; Müller and Somjen, 2000; Akerman et al., 2008; Douglas et al., 2011), others report an insignificant effect of the sodium channel blockers on SD ignition/propagation (Tobiasz and Nicholson, 1982; Herreras and Somjen, 1993; Tozzi et al., 2012; Zhou et al., 2013). A potential explanation for the divergent data on Nav modulation in SD ignition/propagation the use of different methods to elicit SDs (Pietrobon and Moskowitz, 2014), with an apparent difference between focal and more diffuse induction stimuli, each known to generate a phenotypically distinct SD wave (Zhou et al., 2010). Both focal and diffuse origins of SD are relevant to the clinical condition (Dreier, 2011; Pietrobon and Moskowitz, 2014). In this study, we systematically examine the effect of Nav-mediated APs with two commonly used in vitro SD paradigms, focal high [K+] puff and diffuse high [K+] superfusion. Following a comparison of SDs generated by the two induction paradigms in identical experimental conditions, we find that the propagation of focally induced SD, but not SD induced by diffuse high [K+] superfusion, is dependent on Nav-mediated AP bursts.
Materials and Methods
Animal care and handling
All protocols were approved by the Institutional Animal Care and Use Committee at the University of Utah. Animals were housed in a temperature-controlled room on a 12/12 h light/dark cycle. Male C57Bl6 mice were used for every experiment, except for the in vitro epifluorescence calcium imaging experiments, which used transgenic mice on a C57Bl6 background, expressing GCamp3 under the Thy1 promoter (The Jackson Laboratory, strain: 017893).
In vitro brain slice preparation
Mice (two to three months old) were deeply anesthetized with 4% isoflurane, and the brain was removed for slice preparation. Coronal sections were cut in ice-cold dissection buffer (220 mm sucrose, 3 mm KCl, 10 mm MgSO4, 1.25 mm NaH2PO4, 25 mm NaHCO3, 25 mm D-glucose, and 0.2 mm CaCl2), Sections containing somatosensory cortex (Sawant-Pokam et al., 2017) were allowed to recover in a chamber containing normal artificial CSF (ACSF; 125 mm NaCl, 3 mm KCl, 1.3 mm MgSO4, 1.25 mm NaH2PO4, 25 mm NaHCO3, 25 mm D-glucose, 1.3 mm CaCl2, and saturated with 95% O2/5% CO2) at 35°C. For electrophysiology and imaging experiments, the sections were transferred to a submerged chamber (design based on Scientific Systems Design #MS-1; circular Plexiglas chamber with a glass coverslip base) constantly supplied with ACSF, saturated with 95% O2 and 5% CO2, also maintained at 35°C using a line heater (Scientific Systems Design #PTCO3). A slice anchor was placed on the slice (Warner Instruments: catalog #64-1420) to reduce tissue movement. ACSF inlet and the outlet were arranged approximately at a 90° angle to one another. ACSF inflow and outflow were maintained at the same rate (∼2.5 ml/min) throughout the entire experiment.
SD induction
A brief puff of high [K+] was used for the focal induction of SD (Tottene et al., 2009; Sawant-Pokam et al., 2017). A glass micropipette (0.5–1 MΩ resistance, tip size 10–20 μm) filled with 3 m KCl was placed on the superficial [layer (L)1–L2/3] cortical layers in acute slices. KCl was applied to the slices with a brief pulse (0.5 bar N2, 20 ms) using a pressure injector (MPPI-2, Applied Scientific Instruments).
For a global perturbation, acute brain slices were superfused with ACSF containing 40 mm K+ (88 mm NaCl, 40 mm KCl, 1.3 mm MgSO4, 1.25 mm NaH2PO4, 25 mm NaHCO3, 25 mm D-glucose, 1.3 mm CaCl2, and saturated with 95% O2/5% CO2), for 90 s, followed by regular ACSF wash (Zhou et al., 2010).
Pharmacology
Tetrodotoxin (TTX) powder (Abcam, catalog #ab120054) was stored at 4°C, in compliance with the Institutional Biosafety Committee approval; 2 mm stock in water was prepared and aliquoted. The stock was then diluted into ACSF for application in vitro. Control solutions contained identical vehicle concentrations to respective drug solutions (all ≤1%).
In vitro electrophysiology
All whole-cell patch-clamp recordings were obtained from regular spiking pyramidal neurons (Sawant-Pokam et al., 2017) in L2/3 somatosensory cortex. Neurons were visualized using differential interference contrast (DIC) microscopy and patched using glass micropipettes (4–6 MΩ resistance, tip size of 3–4 μm). Membrane voltages were recorded in current clamp (Iclamp) mode, using glass micropipettes filled with intracellular solution containing 130 mm K-gluconate, 5.5 mm EGTA, 10 mm HEPES, 2 mm NaCl, 2 mm KCl, 0.5 mm CaCl2, 0.5 mm GTP, 4 mm ATP, and 10 mm phosphocreatine (pH 7.2, osmolarity = 289–291 mOsm/l). Miniature EPSCs (mEPSCs) were recorded in voltage-clamp mode (Vclamp = –70 mV) in presence of 1 μm TTX. All whole-cell recordings were acquired at 20 kHz and filtered at 2 kHz (lowpass) using a Multiclamp 700B amplifier. Analog data were digitized using a Digidata 1330 digitizer and Clampex 9 software (Molecular Devices). Access resistance was monitored throughout recordings (5-mV pulses at 50 Hz). Recordings with access resistance higher than 25 MΩ or with >20% change in access resistance during baseline recordings were discarded from the analysis. Series resistance compensation (>70%) was applied to currents recorded in the voltage-clamp setting. Offline data processing was done with Clampfit10 (Molecular Devices).
In vitro intrinsic optical signal (IOS) imaging
Acute slices were trans-illuminated by a white-light source (Zeiss SNT tungsten, 12 V 100 W) and transmitted light (IOS) was collected on an upright microscope (Zeiss Axioskop 2) with a 4×/0.10 Achroplan objective lens (Zeiss) and focused on a high-sensitivity twelve-bit charge-coupled device camera (Mightex CCE-B013-U). Images were acquired at 2 Hz at a resolution of 1392 × 1040 pixels (2.9 × 2.17 mm) for the duration of the experiment. Offline image processing was performed with ImageJ (NIH). Grayscale values were obtained using circular regions of interest (ROIs). The first 20 frames of each image series were averaged and used as the baseline for subsequent baseline subtraction and normalization according to the following formula: ΔT = (T – T0)/T0, where T = transmittance at each frame and T0 = averaged baseline transmittance of image series. Circular ROIs were drawn at multiple distances from SD initiation to generate traces of the IOS signal. Linear ROIs were drawn from the SD initiation site up to the maximum recorded extent of SD propagation, unidirectionally across L2/3 (Fig. 1D). Kymographs showing the temporal progression of IOS change across time were generated using linear ROIs (Fig. 1E) and the velocity of SD propagation was quantified as slope of kymographs at fixed intervals (Fig. 1F).
In vitro epifluorescence imaging
For intracellular calcium imaging, acute slices were prepared from mice expressing GCamp3 under the Thy1 promoter (The Jackson Laboratory, strain: 017893). Microscope optics: Slices were epi-illuminated with blue light (excitation filter: 420–495 nm) and green fluorescence signal (emission filter: 520–570 nm) was focused on the camera through a filter cube set with a dichroic mirror. Images were acquired at 2 Hz at a resolution of 1392 × 1040 pixels for the duration of the experiment. Offline image processing was performed with ImageJ. Grayscale values were obtained using circular ROIs. The first 20 frames of each image series were averaged and used as the baseline for subsequent baseline subtraction and normalization according to the following formula: ΔF = (F – F0)/F0, where F = fluorescence at each frame and F0 = averaged baseline fluorescence of image series. Circular ROIs were drawn at multiple distances from SD initiation to determine propagation.
K+-sensitive microelectrodes
K+-sensitive bipolar microelectrodes were fabricated using double barrel electrodes (2BF100-75-10; Sutter Instruments) with tip sizes of ∼4 μm (∼2 μm/barrel). The glass micropipettes were rinsed with 1 m HCl and ethanol before silinization of with 5% hexamethyldisilazane (HMDS; Sigma-Aldrich) at 200°C overnight. The silinized K+-sensitive barrel was backfilled with 100 mm KCl (Haack et al., 2015) and the reference barrel (also used for DC measurements) was filled with regular ACSF. After filling, the tip of the K+-sensitive barrel was front-filled with ∼1 μl of liquid K+ ion exchanger (Potassium Ionophore I-cocktail B; Sigma-Aldrich). Before each experiment, the electrodes were calibrated with standard solutions of ACSF with known K+ concentration (3, 10, 30, and 100 mm) and equimolar NaCl substitution. Signal from the reference microelectrode was subtracted from that of the ion-sensitive microelectrode to obtain [K+]e-sensitive voltage signal, using a high impedance differential amplifier (HiZ-223, Warner Instruments). A log-linear fit from microelectrode calibration was used to calculate the actual [K+]e in each experiment. Signals were digitized by a Digidata 1330b acquisition board and further analyzed with pCLAMP 8.2 software (Molecular Devices, LLC).
Experimental design and statistical analysis
Data analysis was performed using GraphPad Prism 8 (GraphPad Software), MATLAB 7.8.0 (MathWorks), and Microsoft Excel (Microsoft Corp). Sample sizes were determined based on previous reports and pilot experiments, generally n = 6–10 for all in vitro experiments. Experimenters were not blinded to control/drug treatments for all physiology experiments because the effects of the drug treatment were often obvious. However, when blinding was not possible, slices receiving either control or drug treatments were randomized. Outliers were determined empirically based on data distribution (Grubbs' test) for exclusion from further statistical analysis. However, no datapoints were identified as statistical outliers in any of the datasets presented in this study. Most comparisons were made either across groups (grouped comparisons) or across conditions within the same slice (repeated measures).
The normality of distributions was determined using the D'Agostino–Pearson K2 test. Datapoints representing individual slices were compared across groups using a two-tailed, unpaired t test (parametric data) or Mann–Whitney U test (nonparametric data). Paired comparisons were performed using a two-tailed, paired t test. Normalized (%) distributions of individual events were compared across genotype using a two-sample Kolmogorov–Smirnov (KS) test. Data representing multiple time points within the same slice were compared across groups using two-way ANOVA (Friedman's test for nonparametric data) or within groups using one-way ANOVA (Kruskal–Wallis test for nonparametric data). Repeated measures were compared using Bonferroni's post hoc test (or Tukey's test, for equal sample sizes). For datasets with missing values, repeated measures ANOVA based on the mixed-effects model (rather than the canonical general linear model) was used.
Results
Distinct propagation kinetics of SD waves induced in vitro using different induction paradigms
To address the ambiguity surrounding the role of Nav and APs in SD propagation, we used two distinct, widely used chemical induction techniques in vitro (Fig. 1): focal high [K+] puff (20 ms, 0.5 bar N2; see to Materials and Methods; Tottene et al., 2009; Capuani et al., 2016; Sawant-Pokam et al., 2017) and diffuse high [K+] ACSF superfusion (Zhou et al., 2010, 2013; Tozzi et al., 2012; Andrew et al., 2017). SD is associated with transmembrane ion flows, transient depolarization of membrane potential (Vm), cell swelling and shrinkage of extracellular space (MacVicar and Hochman, 1991). These changes were recorded using whole-cell electrophysiology and IOS imaging in mouse brain slices containing somatosensory (barrel) cortex. Focal high [K+] puff consistently generated only a single SD wave (Movie 1), propagating away from the induction site. Increase in [K+]e measured at various distances from the induction site (from 300 to 800 µm), was coincident with the propagating SD measured optically. There was no change in [K+]e during the puff (Fig. 2D), confirming the relative “focality” of the induction stimulus. This was corroborated by imaging relatively limited spatial dispersion of a fluorescent dye following the puff (Fig. 1A). In contrast, the diffuse 40 mm [K+] superfusion paradigm resulted in either a single SD or multiple, concurrent SD waves (50% slices with multiple SDs; Fig. 1C; Movie 2), triggered 90–120 s from the beginning of superfusion.
SD induction using focal and global ignition stimuli. A, left, Example image showing the flow of ACSF and placement of puff (pressure ejection) pipette. Right, Representative images (grayscale) showing the acute spread of SR101 (emission: >605 nm) after a single puff (0.5 bar N2 pressure, pipette resistant 05–0.8MΩ, scale bar: 200 µm). B, Quantification of the area of SR101 signal, 5 min after a single puff. C, Example image sequence (ΔT) showing the progression of a single focally induced SD by 3 m K+ pressure ejection (top) as well as propagation of two SD waves induced by global 40 mm K+ perfusion (bottom). D, A schematic showing a circular ROI (ROI 1, 50 µm in diameter) as well as a linear ROI (ROI 2, 1337 µm in length) placed on cortical L2/3, to capture SD generated by focal high [K+] puff. E, top, Example trace (ΔT/T0) generated using a circular ROI, showing IOS change during SD at a particular location in cortical L/23. Bottom, A kymograph showing the temporal progression of SD-associated IOS change across time, generated using linear ROIs (distance vs time). The velocity of SD propagation was quantified as the slope of IOS change on the kymograph at fixed intervals, represented by three equidistant line segments (400 µm). F, Quantification of slopes ((y2 – y1)/(x2 – x1)) as propagation velocities (mm/min) for respective segments of distance. G, A schematic showing generation of linear ROI across L2/3 to capture SDs generated by high [K+] superfusion. H, Kymograph (distance vs time) derived from the ROI showing the temporal progression of two SD waves propagating from opposite directions.
Increase in [K+]e coincident with propagation of focally induced SD. A, Schematic showing focal SD induction using 3m [K+] pressure ejection and simultaneous acquisition of IOS (imaging) along with [K+]e and DC recording using a K+-sensitive bipolar microelectrode (see Materials and Methods). B, A calibration curve showing voltage (mV) change recorded with the K+-selective electrode in response to increasing concentrations of K+ (R2 = 0.9733, slope = 50.14, n = 3 electrodes). C, Peak [K+]e at the recording location during SD propagation is not correlated to the distance from the induction (p = 0.58, Pearson's correlation r = −0.23, n = 8). D, top, Image sequence showing placement of K+ induction pipette (indicated by *), K+-sensitive microelectrode (indicated by arrowhead) as well as ROI placement, along with real time propagation of SD (increase in ΔT) for 10 s following induction. The electrode and ROI placement was ∼474 μm away from the induction site. After induction, SD propagated in the direction of K+-sensitive electrode. Bottom, Representative trace showing voltage response to increase in extracellular K+ (K+-sensitive electrode) as well as DC signal (reference electrode). The [K+]e at this location did not rise at the time of SD induction (focal K+ puff) but increased coincident with the DC and ROI signal with the peaks of all three signals temporally aligned. E, top, Image sequence showing placement of K+ induction (puff) pipette, K+-sensitive microelectrode as well as ROI placement, 376 μm from induction site. Increase in ΔT following SD induction shows SD propagation away from the K+-sensitive electrode. Bottom, Representative trace showing no change in extracellular K+, DC, or IOS signal, either during or after SD induction, supporting our observations of [K+]e rise only during the SD propagation.
Propagation of an SD wave generated by focal high K+ puff in normal ACSF (control conditions). The movie shows propagating changes to the baseline-subtracted transmittance signal (ΔT) associated with the SD wave, following a pressure puff of 3M [K+]e through a glass pipette (left). The glass electrode for patch-clamp electrophysiology is to the right. The images were acquired using 4× objective, at a resolution of 1392 × 1040 pixels (2.9 × 2.17 mm). Real-time duration: 50 s (16.67×). Scale bar: 200 µm.
Propagation of an SD wave generated by diffuse high K+ superfusion in normal ACSF (control conditions). The movie shows propagating changes to the baseline-subtracted transmittance signal (ΔT) associated with the SD wave, following the superfusion of 40 mM [K+]e containing ACSF. The images were acquired using 4× objective, at a resolution of 1392 × 1040 pixels (2.9 × 2.17 mm). Real-time duration: 50 s (16.67×). Scale bar: 200 µm.
Although the induction stimulus during high [K+] superfusion was applied globally, SD always originated from one or more foci (Figs. 1C, 4A; Movie 2). This observation is consistent with previous reports (Anderson and Andrew, 2002; Zhou et al., 2010), suggesting that perhaps the high [K+] driven neuronal depolarization is not uniform across the tissue, and relatively smaller foci of significant metabolic burden ignite SD. A similar phenomenon has been observed in vivo; the mouse hemicortex, exposed to high [K+] superfusion, typically shows SD ignition from a single focus in the barrel cortex (Bogdanov et al., 2016).
Following a focal high [K+] puff, current-clamp recordings showed a sharp rise in membrane voltage (Vm) concurrent with an AP burst, followed by near-complete depolarization (Fig. 3A). Similar to prior work (Zhou et al., 2010), we found that the global high [K+] superfusion led to a slow ramped increase in membrane Vm (Fig. 3B), before the arrival of the wave. Consequently, globally induced SDs had a significantly higher time to peak from the beginning of the depolarization, compared with focal induction (p < 0.001, Mann–Whitney test, u = 0; Fig. 3G). SD induced by high [K+]e superfusion also revealed a significant increase in the duration of neuronal depolarization compared with focally induced SDs (p < 0.05, t test, t = 2.75; Fig. 3E).
Distinct rise kinetics and durations of neuronal depolarizations degenerated by the two different SD induction paradigms. A, Schematic showing induction of SD using 3 m [K+] focal pressure ejection (left), along with a representative trace of neuronal Vm during an SD wave (Iclamp) recorded >500 µm from the induction site. B, Schematic showing induction of SD using global 40 mm [K+] superfusion (left), along with a representative trace of neuronal Vm during SD. No difference in the peak amplitude (C, p = 0.14, Mann–Whitney test) as well as absolute peak Vm (D, p = 0.98, Mann–Whitney test) of SDs generated by the two induction paradigms. E, The total duration of SD depolarization was significantly higher with 40 mm K+ ACSF induction (p = 0.016, Mann–Whitney test), without a significant difference in the duration of AP bursts (F, p = 0.86, Mann–Whitney test) between the induction paradigms. G, Linear regression line fitted to the neuronal Vm traces showing distinct SD rise slopes for the two different induction paradigms. Time to peak was significantly reduced (inset, p = 0.0002, Mann–Whitney test) for SDs induced by focal 3 m K+ puff. (*P < 0.05, ***P < 0.001).
IOS imaging revealed that focal pressure ejection generated an SD wave at the K+ ejection site (Fig. 4A), which propagated (fully or partially) across the tissue (mean propagation: 2.29 mm, mean velocity: 3.42 mm/min; Fig. 4D) with a significant reduction in the propagation velocity by distance (p < 0.01, Kruskal–Wallis test, H = 13.04; Fig. 4E). Maximum SD propagation distance with diffuse [K+] superfusion was similar to the focal [K+] paradigm (mean distance 2.67 mm, mean velocity: 4.05 mm/min; Fig. 4D). However, the reduction in velocity was not significant with distance (p = 0.12, Kruskal–Wallis test, H = 5.69; Fig. 4E). Consistent with the neuronal Vm rise, the IOS signal also had a slow ramp following global high [K+] superfusion, whereas focal high [K+] puff led to a sharp rise in the IOS (Fig. 4B), confirming prior observations (Anderson and Andrew, 2002; Zhou et al., 2010). Therefore, SD waves generated by the two different induction paradigms were phenotypically distinct, with disparate kinetics of propagation across space, Vm depolarization, and corresponding IOS transients.
Distinct rise kinetics of IOS signal following SD induction by the two paradigms. A, IOS image sequence (ΔT) showing the progression of SD induced by 3 m K+ pressure ejection (top) as well as 40 mm K+ ACSF perfusion (bottom) across space and time (scale bar: 200 µm). B, Traces showing changes in transmittance (ΔT/T0, group means are bold) during the passage of an SD wave (ROI ∼1000 μm from the induction site). C, Linear regression line fitted to the mean ΔT/T0 traces showing distinct SD rise slopes for the two different induction paradigms. D, Maximum recorded propagation for SD waves induced by the two different induction paradigms is not different (unpaired t test, p = 0.53). E, Although the comparison of propagation velocities between the two induction paradigms, was not significant (p = 0.98, two-way ANOVA), the propagation velocity significantly reduced with distance for focally induced SD (p < 0.01, Kruskal–Wallis test).
TTX application selectively restricts propagation of focally induced SD waves
We then tested the effects of AP blockade on the propagation of SDs, using focal and global paradigms. Acute slices were treated with TTX (1 μm) for 5 min before SD induced by either focal or diffuse high [K+] application. The maximum propagation distance measured from SD induction site following a focal high [K+] puff was significantly reduced with TTX treatment compared with the control condition (p < 0.0001, t test, t = 5.59; Fig. 5D; Movie 3), though propagation velocity was unaffected (p = 0.63, two-way ANOVA, F = 0.029; Fig. 5E). In contrast, the application of TTX did not affect either the maximum propagation distance (p = 0.08, t test, t = 1.85; Fig. 6D; Movie 4) or the propagation velocity (p = 0.12, two-way ANOVA, F = 2.52; Fig. 6E) of SD waves generated by diffuse [K+] perfusion. Therefore, as global tissue conditions grow closer to those that occur around the SD threshold, as commonly observed with global stimuli, AP bursts likely become less necessary as a mechanism to sustain the propagation of SD waves.
TTX treatment restricts propagation of SD generated by focal high [K+] puff. A, Schematic showing induction of SD using focal 3 m [K+] pressure ejection (left) along with representative traces of transmittance signal (ΔT/T0) showing SD propagation in control and TTX condition (ROIs: 400, 800, and 1200 µm from the induction site). B, Example Image sequence (ΔT) showing the progression of a focally induced SD wave across space and time in control as well as TTX treatment conditions (scale bar: 200 µm). C, Traces showing changes in light transmittance (ΔT/T0) following focally induced SD, in control (n = 9, blue) and TTX (n = 11, gray) conditions (ROIs at ∼500 and 1000 µm, group means are bold). D, TTX treatment significantly reduced the maximum propagation of the SD wave generated by focal 3 m [K+] pressure ejection (unpaired t test, p < 0.001), without affecting SD velocity (E) between the two groups (two-way ANOVA, p = 0.07). (***P < 0.001).
Propagation of SD generated by global high [K+] is unaffected by TTX treatment. A, Schematic showing induction of SD using global 40 mm [K+] superfusion (left) along with representative traces of transmittance signal (ΔT/T0) showing SD propagation in control and TTX condition (ROIs: 400, 800, and 1200 µm from the induction site). B, Example image sequence (ΔT) showing the progression of a global 40 mm [K+] superfusion induced SD wave across space and time in control as well as TTX treatment conditions (scale bar: 200 µm). C, Traces showing changes in light transmittance (ΔT/T0) in response to SDs induced by 40 mm [K+] superfusion, in control (top, n = 10, red) as well as TTX (bottom, n = 11, gray) conditions (ROIs ∼1000 µm, group means are bold). In the case of SD generated by 40 mm K+ ACSF perfusion, TTX treatment did not affect either maximum propagation (D, unpaired t test, p = 0.08), or SD velocity (E, two-way ANOVA, p = 0.86).
Propagation of an SD wave generated by focal high K+ puff in 1 µM TTX. The movie shows propagating changes to the baseline-subtracted transmittance signal (ΔT) associated with the SD wave, following a pressure puff of 3M [K+]e through a glass pipette, in the presence of 1 µM TTX. The images were acquired using 4× objective, at a resolution of 1392 × 1040 pixels (2.9 × 2.17 mm). Real-time duration: 50 s (16.67×). Scale bar: 200 µm.
Propagation of an SD wave generated by diffuse high K+ superfusion in 1 µM TTX. The movie shows propagating changes to the baseline-subtracted transmittance signal (ΔT) associated with the SD wave, following the superfusion of 40 mM [K+]e containing ACSF, in the presence of 1 µM TTX. The images were acquired using 4× objective, at a resolution of 1392 × 1040 pixels (2.9 × 2.17 mm). Real-time duration: 50 s (16.67×). Scale bar: 200 µm.
TTX application reduces the amplitude and duration of neuronal depolarization in response to focally induced SD
We wanted to explicitly test whether the restricted propagation of focally induced SD after TTX treatment was because of incomplete or attenuated neuronal depolarization. During a propagating focally induced SD wave, we recorded neuronal voltage responses from cortical neurons at a distance of 200–1000 µm, with or without TTX (Table 1). As expected with TTX treatment, neurons failed to fire APs (p < 0.0001, Mann–Whitney test, U = 0; Fig. 7A,E). Importantly, TTX application significantly reduced the amplitude (p < 0.01, Mann–Whitney test, U = 7; Fig. 7C) and the absolute Vm (p < 0.01, Mann–Whitney test, U = 6; Fig. 7D) of peak depolarization as well as the total duration of the SD event (p < 0.05, Mann–Whitney test, U = 16; Fig. 7B). Depolarization amplitude in the presence of TTX was negatively correlated with the distance from the SD induction site (p < 0.05, Spearman's correlation, rs = −0.94; Fig. 7H), as the amplitude decreased in neuronal recordings obtained farther away from the induction site (Fig. 7F,G).
Whole-cell electrophysiological characteristics of SDs recorded from neurons under control condition and in the presence of 1 μm TTX
TTX treatment reduces the duration and amplitude of SD distant to the induction site. A, Schematic showing focal induction of SD using 3 m [K+] pressure ejection and data acquisition using whole-cell current clamp (left) along with typical current-clamp traces of SDs in control (blue) and TTX (gray) conditions (arrowheads indicate high [K+] puff). TTX treatment significantly reduced the SD duration (B, Mann–Whitney test, p = 0.043), maximum depolarization (C, Mann–Whitney test, p < 0.0038) and peak amplitude (D, Mann–Whitney test, p = 0.0024) as well as abolished AP bursts during SD (E, Mann–Whitney test, p < 0.0001, control N = 9, TTX N = 7). F, Schematic showing the current-clamp recordings of focally induced SD, obtained from proximal (∼200 µm) and distal (∼800 µm) neurons in acute slices. G, Representative traces of SD-associated Vm depolarization in presence of TTX in neurons proximal (top ∼200 µm) and distal (bottom ∼800 µm) to the induction site. H, Negative correlation between SD amplitude and distance from induction site (nonparametric Spearman's correlation coefficient rS = −0.9429, p = 0.017). I, Representative traces of Vm rise (filtered to eliminate APs) during SD showing multiphasic rise with two time constants under control condition and an exponential rise with one time constant tor TTX condition. J, Comparison of time constants (τ1) of single-phase exponential fits to the initial Vm rise (preceding AP bursts) between control and TTX conditions revealed no significant difference (p = 0.95, Mann–Whitney test, control n = 11, TTX n = 7). K, Comparison of time to peak between control and TTX groups showed no significant difference (p = 0.53, Mann–Whitney test, control n = 11, TTX n = 7). (*P < 0.05, **P < 0.01, ***P < 0.001).
Next, we wanted to determine whether TTX application affected the rise kinetics of neuronal depolarization. Neuronal depolarization during SD under control conditions was typically biphasic, with two different time constants. The two phases of neuronal Vm rise were typically separated by brief repolarization. The initial Vm rise typically culminated in AP bursts and could be fitted with a single exponential. The second phase typically followed the AP bursts and could also be fitted with a separate single exponential. In contrast, after TTX treatment, Vm depolarization only exhibited the initial rise, as the second phase of depolarization was abolished (Fig. 7I), suggesting that TTX treatment prevents neuronal depolarization beyond the initial rise. We used traces filtered to eliminate AP waveforms, to fit single exponentials to the initial Vm rise under both conditions. We found no significant difference between the time constants (tau; p = 0.94, Mann–Whitney test, U = 17; Fig. 7J) as well as time to peak (p = 0.53, Mann–Whitney test, U = 16; Fig. 7K) between control and TTX groups, suggesting TTX treatment does not significantly alter the initial (before AP bursts) Vm rise kinetics of SD-associated neural depolarization. However, TTX clearly affected the maximum Vm rise and the overall rise kinetics, by abolishing the second depolarization phase.
TTX reduces the propagation of calcium transients as well as intracellular calcium load elicited by focally induced SD
Our results thus far suggest that TTX-mediated Nav blockade, and consequent inhibition of neuronal AP bursts, limits the propagation of focally induced SD, likely by preventing near-complete depolarization in neurons away from the induction site. Neuro-glial depolarization during an SD wave evokes propagating intracellular calcium transients (Peters et al., 2003; Chuquet et al., 2007). Sustained neuronal depolarization along with action-potential bursts during SD contribute to neuronal Ca2+ transients under normal conditions (Jacobs and Meyer, 1997; Koester and Sakmann, 2000), and the onset of SD-associated neuronal calcium transients is coincident with glutamate and K+ efflux (Zhou et al., 2013; Enger et al., 2015). The SD-associated propagating Ca2+ waves can be effectively imaged using genetically encoded calcium indicators (GCaMP), both in vivo and in vitro (Enger et al., 2015; Khennouf et al., 2016). We wanted to test whether the suppression of AP bursts also reduced SD evoked [Ca2+]i transients in neurons. We prepared acute cortical slices from mice expressing GCaMP3 under a neuronal (Thy1) promoter. After the TTX application, the total propagation distance of calcium transients was significantly reduced (p < 0.01, Mann–Whitney test, U = 2; Fig. 8F; Movies 5, 6), consistent with insufficient neuronal depolarization away from the induction site, without a significant change in propagation velocity (p > 0.05, two-way ANOVA, F = 0.29; Fig. 8E). Moreover, the area under the curve (AUC) of the evoked calcium transient, representing tissue calcium load (Patel et al., 2015; Reinhart and Shuttleworth, 2018), was also significantly reduced after TTX application (500 µm: p < 0.05, 1000 µm: p < 0.001, t test, t = 2.42; Fig. 8G). As calcium load is often associated with the deleterious effects of SDs (Aiba and Shuttleworth, 2012; Reinhart and Shuttleworth, 2018), the blockade of AP bursts can potentially reduce SD severity.
SD-associated intracellular calcium load is significantly reduced by TTX treatment. A, Schematic showing focal induction of SD along with epifluorescence imaging in acute slices prepared from Thy1-GCamp3 animals. B, Representative traces of fluorescence signal (ΔF/F0) showing SD propagation in control and TTX treated slices (ROIs: 400, 800, and 1200 µm from the induction site). C, Example Image sequence (ΔF) showing the progression of a focally induced SD wave (3 m K+ pressure ejection) across space and time in control and TTX conditions (scale bar: 200 µm). D, Traces showing changes in fluorescence signal (ΔF/F0) in response to focally induced SD recorded from control (top, n = 8, blue) as well as TTX (bottom, n = 7, gray) slices (ROIs at ∼500 and 1000 µm from the induction site). E, TTX did not affect the propagation velocity of SD-associated Ca2+ signal (p = 0.65, two-way ANOVA). F, However, TTX treatment significantly reduced the maximum propagation distance of focally generated SD wave (p = 0.023, Mann–Whitney test). G, Similarly, the area under the curve (ΔF/F0*time) that corresponds to tissue Ca2+ load is also reduced after TTX treatment, at both 500 µm (p = 0.031, unpaired t test) as well as 1000 µm (p = 0.024, unpaired t test). (*P < 0.05, **P < 0.01).
Propagation of SD-associated Ca2+ transients generated by focal high K+ puff in normal ACSF (control conditions). The movie shows a propagating increase in the baseline-subtracted fluorescence signal (ΔF) in a Thy1-GCaMP3 slice, following a pressure puff of 3M [K+]e. The images were acquired using 4× objective, at a resolution of 1392 × 1040 pixels (2.9 × 2.17 mm). Real-time duration: 50 s (16.67×). Scale bar: 200 µm.
Propagation of SD-associated Ca2+ transients generated by focal high K+ puff in 1 µm TTX. The movie shows a propagating increase in the baseline-subtracted fluorescence signal (ΔF) in a Thy1-GCaMP3 slice, following a pressure puff of 3M [K+]e, in the presence of 1 µM TTX. The images were acquired using 4× objective, at a resolution of 1392 × 1040 pixels (2.9 × 2.17 mm). Real-time duration: 50 s (16.67×). Scale bar: 200 µm.
TTX treatment prevents SD-induced suppression of spontaneous EPSC frequency
In the normally metabolizing brain, the passage of an SD wave can lead to suppression of spontaneous and evoked activity (Somjen, 2001; Pietrobon and Moskowitz, 2014), with multiple in vitro studies reporting a reduction in the frequency of spontaneous postsynaptic currents (E/IPSCs) after focally induced SD (Aiba and Shuttleworth, 2012, 2014). Previously, under experimental conditions identical to that described in this study, we observed a short-term (5 min) and long-term (30 min) reduction in spontaneous EPSC frequency following focally induced SD in the normally metabolizing cortex, with a slight increase in the EPSC amplitude 30 min post-SD (Sawant-Pokam et al., 2017). The long-term changes in postsynaptic response, as well as sensory-evoked circuit responses (Theriot et al., 2012) following SD, resemble [Ca2+]i-dependent cortical plasticity (Buonomano and Merzenich, 1998; Feldman and Brecht, 2005). Given the effects of AP suppression on neuronal calcium, we suspected that the suppression of spontaneous neurotransmission mediated by SD might be altered as well. We used whole-cell voltage clamp (Vclamp = –70 mV) to record mEPSCs during TTX treatment (baseline) and continued recording up to 30 min following the focal induction of SD in the normally metabolizing cortical slices. TTX treatment abolished the short- as well as long-term reductions in mEPSC frequency (at 5 and 30 min, respectively) in cortical neurons (Fig. 9B,C) and had no effect on mEPSC amplitude (Fig. 9D,E; contrast with Sawant-Pokam et al., 2017). Therefore, neuronal AP bursts and subsequent suprathreshold depolarization are necessary for long lasting post-SD changes in synaptic transmission after focally induced SD.
TTX treatment prevents SD induced suppression of excitatory spontaneous synaptic transmission. A, Representative traces showing mEPSCs recorded before (baseline), 5 min after, and 30 min after focally induced SD, in presence of TTX. B, Comparison of mean mEPSC frequencies of individual neurons shows no change from baseline (p = 0.87, Wilcoxon matched pairs test). C, Cumulative frequency histogram of interevent intervals shows no difference in mEPSC frequency between baseline and 5 min (p = 0.061, two-sample KS test) and 30 min (p = 0.45, two-sample KS test) after SD (n = 4 neurons for amplitude and frequency). D, No change in mean mEPSC amplitudes at 5 and 30 min after SD induction (p = 0.38, Wilcoxon matched pairs test). E, Cumulative frequency histogram of amplitudes of individual mEPSC events revealed no difference at 5min (p = 0.076, two-sample KS test) and 30 min (p = 0.091, two-sample KS test) from baseline.
Discussion
We studied the role of Nav-mediated neuronal AP bursts in SD propagation, using two distinct in vitro induction paradigms (focal vs diffuse exposure to increased [K+]). We found that these two models showed very different consequences of AP blockade. Indeed, the two models generated phenotypically distinct SD waves with different propagation kinetics. Although this can be inferred from the examination of previous work, by putting prior literature side by side (Anderson and Andrew, 2002; Tottene et al., 2009; Zhou et al., 2010; Pietrobon and Moskowitz, 2014; Reinhart and Shuttleworth, 2018), an advantage of the current report lies in the systematic comparison of the two induction paradigms, with identical slice characteristics and experimental conditions, in one study.
Following focal induction, TTX-mediated Nav blockade restricted the propagation of SD, most likely by preventing suprathreshold neuronal depolarization in neurons distant to the induction site. Rise in [Ca2+]i is typically a concomitant of AP activity (Akerboom et al., 2012; Chen et al., 2012): when the AP burst of SD was suppressed we also observed a significant attenuation of intracellular calcium accumulation, which has implications for the long-term network effects of the event (Theriot et al., 2012; Sawant-Pokam et al., 2017). Indeed, we observed alterations in long-term network function, in that AP blockade prevented the changes in spontaneous activity mediated by SD that we have described in previous work (Sawant-Pokam et al., 2017).
In marked contrast to focally induced SD, TTX application did not affect SD induced by diffuse [K+] superfusion. These results help clarify the role of Nav-mediated AP bursts in SD propagation, and highlight physiological distinctions in SD waves generated by different induction paradigms.
Multiple lines of evidence suggest that SD propagation in normally metabolizing tissue is dependent on the activation of NMDA receptors and Cav channels. For example, propagation of SD induced by electrical or high [K+] application was completely blocked by NMDA receptor antagonists but not by other glutamate receptor antagonists (AMPA and kainate), both in vivo and in vitro (Lauritzen and Hansen, 1992; Krüger et al., 1999). Similarly, propagation of SD waves induced using similar focal stimuli, under normal metabolic conditions, was abolished after blocking Cav channels with Cd2+ or Ni2+, or in Ca2+ free medium, and with specific blockade of Cav2.1 channels (Footitt and Newberry, 1998; Tottene et al., 2002; Peters et al., 2003). Significant neuronal depolarization is necessary to activate a large population of Cav channels (Pietrobon, 2010; Hering et al., 2018), and recruit NMDA receptors by reversing the Mg2+ block of the receptor pore that is present at nondepolarized membrane potentials (Collingridge, 2003). AP bursts during SDs may be necessary to ensure sufficient depolarization of neurons distant to the induction site, to activate Cav and NMDA receptors, sustaining SD propagation. In support of this, we found that AP blockade by TTX significantly reduced the depolarization of distant neurons, restricting SD propagation.
Conclusions from studies investigating the role of Nav in SD ignition/propagation in normally metabolizing tissue have been ambiguous. SD propagation induced by focal stimuli such as pinprick (Akerman et al., 2008) or brief electrical stimulation (Sugaya et al., 1975) was blocked by TTX in vivo. However, SD induced by stimuli that cause an increase in baseline [K+]e, such as in vivo high K+ dialysis (Herreras and Somjen, 1993), topical application of KCl crystals (Sugaya et al., 1975) or in vitro high K+ superfusion of the whole-brain slice (Tozzi et al., 2012; Zhou et al., 2013) were insensitive to TTX. Similarly, a tissue-wide increase in baseline potassium renders SD increasingly insensitive to NMDA receptor antagonists (Petzold et al., 2005). High [K+] superfusion and subsequent neuronal depolarization maintains the entire tissue near SD-threshold conditions, for a physiologically significant time (Tang et al., 2014). Therefore, SD propagation may be AP-independent (or even NMDA receptor-independent) under these conditions, as the driving force necessary for near-complete neuronal depolarization is less. On the other hand, a focal high [K+] puff exposes only a limited area (Fig. 1B) to high [K+], leaving the tissue distant to the induction site with normal [K+]e until passage of the wave. Under these conditions, AP bursts and consequent efflux of [K+] as well as glutamate, appear to be necessary to drive SD propagation in tissue distant to the induction site. Consistent with this hypothesis, we found that TTX treatment reduced neuronal depolarization (Fig. 7) as well as consequent [Ca2+]i transients (Fig. 8) following SD induced by focal high [K+], away from the induction site.
Global superfusion of high [K+] can also impose a bioenergetic burden before SD induction, because of constant buffering of elevated [K+]e by Na+/K+ ATPases (Rose et al., 2009; Larsen et al., 2016). This contrasts with focal high [K+] puff, where the area of high [K+] exposure is limited. Indeed, it was interesting to note that the slow ramped increase in IOS and membrane Vm during high [K+] superfusion was phenotypically similar to that observed previously in SD induced using hypoxic conditions (Müller and Somjen, 1998; Anderson et al., 2005; Zhou et al., 2010; Andrew et al., 2017), suggesting that perhaps the global high [K+] superfusion recapitulates, to some extent, induction conditions similar to that seen in metabolically compromised tissue. Moreover, in mathematical modeling studies investigating the impact of metabolism on the neuronal transition from spiking to near-complete depolarizations during SDs (Wei et al., 2014; Ullah et al., 2015), the rates of transition from resting state to spiking to near-complete depolarization were found to be dependent on tissue metabolism. Neurons under hypoxic conditions bypass action-potential bursts and transition directly from the resting state to near-complete depolarization (Wei et al., 2014). These findings, along with the near SD-threshold neuronal depolarization, may explain the AP independent propagation of SD induced by high [K+].
SD underlying migraine aura presumably require largely intact energy metabolism before the onset (Dreier, 2011; Pietrobon and Moskowitz, 2014), or, put more conservatively, it is very unrealistic to assume the metabolic compromise that would be associated with global increases in [K+]e. Therefore, SD waves generated by focal and brief application of high [K+], in brain tissue with seemingly normal metabolism, may model migraine auras most accurately. Focality of the puff was confirmed by [K+]e measurements using K+-sensitive electrodes (Fig. 2) as well as imaging spatial dispersion of fluorescent dye following the puff (Fig. 1A). The true extent of spatial dispersion of [K+]e during induction puff and propagating SD could not be characterized, largely because of the limitations of ion-sensitive electrodes. With the advent of new fluorescent [K+]e indicators (Bischof et al., 2017; Shen et al., 2019), this should be addressable with optical measurements. In contrast, diffuse K+ superfusion models may be more appropriate for situations that prime the tissue before SD ignition, e.g., regions near contusions where cells have been lysed and high concentration intracellular K+ (∼145 mm, Tang et al., 2014) has been released into the extracellular space (Dreier, 2011; Pietrobon and Moskowitz, 2014; Hartings et al., 2017). As both induction paradigms represent plausible scenarios at different ends of a physiological continuum, our results may be relevant to different disease conditions that generate distinct SD.
SDs have been implicated in multiple neurologic disorders. In a metabolically normal brain, SD not only causes migraine auras (Lauritzen, 1994) but also sensitizes the trigeminovascular system (Bolay et al., 2002; Zhang et al., 2010) leading to craniofacial pain through the initiation of a proinflammatory cascade that activates nociceptors (Karatas et al., 2013; Brennan and Pietrobon, 2018). The magnitude of nociceptive network sensitization is higher in genetically susceptible animals (Zhang et al., 2011; Brennan et al., 2013) corroborating evidence in humans (Winawer et al., 2013). Beyond trigeminal activation, SD leads to dysfunctional plasticity of cortical sensory circuits (Faraguna et al., 2010; Sawant-Pokam et al., 2017), which may also underlie the chronification of migraine (Brennan and Pietrobon, 2018). Therefore, preventing SD ignition or limiting its propagation is likely important not only for alleviating migraine relevant pain but also its long-term circuit effect.
Similar considerations likely apply for SDs in brain injury situations (Müller and Somjen, 1998; Douglas et al., 2011), though they have not been fully explored experimentally. Our results strongly suggest that TTX application restricts SD propagation by preventing suprathreshold depolarization and the associated increase in [Ca2+]i in neurons away from the induction site. Neuronal depolarization, NMDA receptor activation, and the subsequent increase in [Ca2+]i are correlated to the severity of SD waves (Aiba and Shuttleworth, 2012; Reinhart and Shuttleworth, 2018) and may contribute to aberrant synaptic plasticity post-SD (Reid and Stewart, 1997; Fox et al., 2006; Theriot et al., 2012). Thus, for focally induced waves, preventing suprathreshold neuronal depolarization with Nav modulation can potentially reduce the deleterious effects of SD, in both the short and long term.
In contrast to propagation, we found that TTX application failed to prevent SD ignition to focal high [K+] puff. Our experiments were not designed to detect differences in initiation, e.g., we did not test whether TTX increased SD induction threshold. That said, SD thresholds are increased by perturbations that reduce neuronal [Ca2+]i downstream of Nav activation, such as NMDA receptor antagonists in rats (Marrannes et al., 1988) and a loss of function Cav2.1 mutation in mice (Ayata et al., 1999). Therefore, it is quite conceivable that the negative modulation of Nav will increase SD thresholds. Recent work in transgenic mice harboring a mutation in the SCN1A gene encoding α1 subunit of NaV1.1 (FHM3) highlights the role of Nav in SD ignition. These mice had had a reduced threshold to experimentally induced SDs, and awake FHM3 mice even exhibit apparently spontaneous SDs (Jansen et al., 2020). NaV1.1 loss of function preferentially reduced Na+ currents in inhibitory interneurons (Yu et al., 2006; Catterall, 2012), suggesting that aberrant interneuron activity may underlie the phenotype (Desroches et al., 2019). Interestingly, Nav channels contribute to SD induction under hypoxic conditions, as Nav blockers like TTX and dibucaine delay the onset of hypoxia-induced SDs (Aitken et al., 1991; Müller and Somjen, 1998; Jarvis et al., 2001; Douglas et al., 2011; Risher et al., 2011).
Although distinct SDs arise from disparate metabolic conditions, these events define a physiological continuum, with many common characteristics (Hartings et al., 2017). In addition to being an underappreciated element of SD propagation, Nav modulation may represent a potential strategy for SD suppression if our in vitro results are confirmed in more translational preclinical models.
Footnotes
This work was supported by National Institutes of Health Grants R01 NS102978 and NS104742 (to K.C.B.). We thank Patrick Parker and Punam Sawant-Pokam for providing technical direction.
The authors declare no competing financial interests.
- Correspondence should be addressed to K. C. Brennan at k.c.brennan{at}hsc.utah.edu
