Spreading depolarization triggered by elevated potassium is weak or absent in the rodent lower brain

Abstract

We examined in live coronal slices from rat and mouse which brain regions generate potassium-triggered spreading depolarization (SD_Kt). This technique simulates cortical spreading depression, which underlies migraine aura in the intact brain. An SD_Kt episode was evoked by increasing bath [K⁺]_o and recorded as a propagating front of elevated light transmittance representing transient neuronal swelling in gray matter of neocortex, hippocampus, striatum, and thalamus. In contrast, SD_Kt was not imaged in hypothalamic nuclei or brainstem with exception of those nuclei near the dorsal brainstem surface. In rat slices, single neurons were whole-cell current clamped during SD_Kt. “Higher” neurons depolarized to near zero millivolts indicating SD_Kt generation. In contrast, seven types of neurons in hypothalamus and brainstem only slowly depolarized without generating SD_Kt, supporting our imaging findings. Therefore, SD_Kt is not a default of CNS neurons but rather displays a region-specific susceptibility, similar to anoxic depolarization, which we have proposed is correlated with a region’s vulnerability to traumatic brain injury. In the higher brain, SD_Kt may be a vestigial spreading depolarization that originally evolved to shut down and vasoconstrict gray matter regions more exposed to impact and contusion.

Introduction

Spreading depolarization (SD) is a sudden and profound loss of membrane potential by neurons and neighboring astrocytes that migrates at 1.7–9.2 mm/min across gray matter.^1,2 It is important clinically because SD can arise repeatedly following focal stroke, subarachnoid hemorrhage, or traumatic brain injury (TBI) to expand the original focal damage.^3,4 In live brain slices under simulated ischemia (oxygen/glucose deprivation, OGD), a propagating wave of anoxic depolarization (AD) arises, which leaves neurons damaged in its wake.^5–7 Under less metabolic stress, higher gray matter supports cortical spreading depolarization (CSD), which can be experimentally evoked in vivo and in slices by elevating [K⁺]_o. This “potassium-triggered spreading depolarization” (SD_Kt) is the common technique to evoke CSD-like events.^8–10 CSD is responsible for the marching sensory deficit that often dominates migraine aura although CSD itself does not induce neural injury. AD and SD_Kt represent extremes along a continuum of SD^6,11 whereby ultimate neuronal injury is dependent upon the metabolic state of the gray matter through which the SD event propagates.^11–14 Several studies indicate that more rostral brain regions are more susceptible to ischemia^15,16 and generate stronger AD.^17–20 In this article, we examine whether this is also true regarding SD_Kt and if there are regional differences in the ability to inhibit SD_Kt in the presence of the glutamate receptor antagonist kynurenate. Live brain slices allow study of the intrinsic propensity of neurons to generate AD because potential variability in blood flow among brain regions is not a factor. SD_Kt can be induced repeatedly by brief bath elevation of [K⁺]_o in rat hippocampal slices,²¹ in human neocortical slices²² and in submerged slices from rat neocortex.^23,24 Therefore, as observed in vivo, SD_Kt is not damaging to neurons in brain slices.

The propensity for SD_Kt and ischemic SD generation in human neocortex could be hypothesized to be a necessary consequence of higher network complexity that leads to electrical instability. However, it is important to note that SD has evolved in the insect CNS to promote survival, likely arising more than 200 million years before mammals even existed. On a gram-for-gram basis, it can be argued that the insect brain is as complex and sophisticated as a human’s brain. SD in the locust^25,26 and in Drosophila²⁷ is physiologically similar to mammals and is evoked by high temperature or by anoxia, inducing a recoverable coma as neurons depolarize to zero millivolts in a propagating fashion. Three benefits to these insects result. First, sudden electrical silence avoids uncoordinated behavior and hyperactivity that will arise from neuronal depolarization as the Na/K pump begins to fail. The only other option to induce electrical silence would be a widespread hyperpolarization, but that in itself is an energy-consuming process. Second with complete depolarization, production of reactive oxygen species is reduced because oxygen is not being consumed. But with return of oxygen, reactive oxygen species may then dramatically increase, so SD can be considered a “delay” tactic in this regard. Third, the enforced inactivity lowers the metabolic rate, so that the Na/K pump can recover. And indeed insects easily “awake” from their comatosed state after a few minutes.

Along similar lines, we have proposed that mammalian SD promotes a “lie-low” survival strategy by promoting sudden loss of consciousness following TBI whereby the chances of epileptiform activity and brain hemorrhage are reduced and metabolism is lowered.^19,28 Immediately upon head trauma from a predator, from warfare or from falling out of a tree, shut-down is a good default behavior. It is better to be quiescent when injured than to be moving in a slow or epileptiform fashion which advertises injury. Somjen²⁹ has noted that “It is easy to imagine that a blow to the head could trigger SD at once in many areas of the brain, perhaps including subcortical nuclei, rendering the victim unconscious.” A temporary shutdown of the higher gray matter with vasoconstriction would have been a reasonable default strategy to help survive head injury throughout our vertebrate evolution. In contrast, stroke and sudden cardiac arrest occur primarily in the aging population. Sudden loss of blood flow to the brain has become an issue only over the past 200 years as the human life span has increased beyond 40 years.³⁰ In these cases, our brain may be misinterpreting lost blood flow as hemorrhage. The result is ischemia-induced failure of the Na/K pump causing a propagating electrical shutdown and vasoconstriction in higher gray matter, whether the cause is focal stroke, subarachnoid hemorrhage or global ischemia.

It has been argued that migraine pain must have been beneficial for survival to be so prevalent today.³¹ But because migraine pain imparts prolonged agony and incapacitation, we think that a survival advantage is questionable. Rather, as with neuropathic pain, migraine pain can be considered a natural defense mechanism that has spiraled out of control and so lacks evolutionary benefit. There is good evidence that the CSD driving migraine aura can evoke the headache that follows.^9,10 Here we examine the regional susceptibility of the brain to generate SD_Kt and whether that susceptibility emulates the propensity of higher brain neurons to generate strong AD in comparison with lower brain gray matter.^19,20 If SD_Kt is a weaker or residual version of SD, it follows that lower brain regions being more protected from TBI by bone and higher brain (even early in vertebrate evolution) should be less susceptible to SD_Kt.

Methods

Neocortical slice preparation

Male C57 Black mice or Sprague-Dawley rats, 21–50 days old (Charles River, St. Constant, Quebec) were cared for in accordance with the principles and guidelines of the Canadian Council on Animal Care as overseen by the Queen’s University Animal Care Committee. This study complied with the ARRIVE guidelines (Animal Research: Reporting in Vivo Experiments). Rodents were housed in a controlled environment (22 ± 1℃, 12 h light:12 h dark cycle) with food (Purina rat chow) and water supplied ad libitum. Rats were placed in a rodent restrainer (DecapiCone; Braintree Scientific, Inc.) and guillotined. The brain was excised and placed in ice-cold oxygenated (95% O₂–5% CO₂) artificial cerebrospinal fluid (aCSF) with NaCl replaced by sucrose. Coronal slices (400 µm) were taken using a vibrating blade microtome (Leica VT1000S). Five to seven slices were transferred to a submerged net in a beaker containing regular aCSF gassed with O₂–CO₂. The slices were then slowly warmed over 1 h to 32℃ prior to experimentation at 33–34℃.

Experimental solutions and drugs

The aCSF contained (in mM) 120 NaCl, 3.3 KCl, 26 NaHCO₃, 1.3 MgSO₄ 7H₂O, 1.2 NaH₂PO₄, 1.8 CaCl₂, and 11D-glucose. All were dissolved in double-distilled water at pH 7.3–7.4. The aCSF was used for incubation and as a vehicle to administer experimental solutions. For high-K⁺ aCSF, 9.6, 26, or 52 mM KCl replaced equimolar NaCl. Kynurenic acid (2 mM) was added to control aCSF or to elevated-KCl ACSF as required. All drugs were from the Sigma-Aldrich. During an experiment, a slice were submerged in oxygenated aCSF flowing at a rate of 3–4 ml/min at 33–34℃.

Imaging intrinsic optical signals

Changes in light transmittance (LT) through live brain slices are generated as LT is increased by brain cell swelling or reduced by dendritic beading which increases light scatter.^32,33 Using previously described techniques,^6,23 ΔLT was monitored and recorded. Graphical and statistical analyses of data were carried out using Excel. Images were imported and figures were prepared using CorelDraw. A ΔLT value from each brain region of interest (ROI) is an average of that region (∼500 µm²). The data from the imaging experiments were analyzed such that changes in LT of a given ROI were expressed as percent changes of the T_cont for that region, taken from the control image. This normalized the graphical data across the different regions of the neocortex, which was necessary because of the variation in opacity that causes different initial LT values (T_cont). The means were calculated and reported using a paired Student’s t-test for statistical analysis.

Electrophysiology

Micropipettes were pulled from borosilicate glass (OD 1.2 mm, ID 0.68 mm; World Precision Instruments) to a resistance of 3–6 MΩ. The internal pipette solution contained (in mM) 125 K-gluconate, 10 KCl, 2 MgCl₂, 5.5 EGTA, 10 HEPES, 2 Na-ATP, and 0.1 CaCl₂ (pH was adjusted to 7.3 with KOH). A 14 mV junction potential was corrected prior to achieving whole-cell configuration. Whole-cell patched neurons were analyzed if they displayed stable resting membrane potentials and if the series resistance could be sufficiently compensated. Recordings were from neurons visualized with near-infrared illumination and Dodt gradient contrast optics (Luigs and Neumann, Ratingen, Germany) through an upright microscope (Axoscope 2FS, Zeiss) with a × 40 immersion objective lens. Neurons were recorded from neocortical layers 2–3, striatum, thalamus, the hypothalamus (supraoptic nucleus, paraventricular nucleus, suprachiasmatic nucleus), the mesencephalic nucleus (MES) of midbrain-pons, locus coeruleus (LC) of midbrain-pons, dorsal nucleus solitarius (SolN) of medulla oblongata, and dorsal motor nucleus (DMV) of medulla oblongata. All recordings were acquired in current clamp mode of an Axoclamp 2 A amplifier and a Digidata 1322 A/D converter. Clampex 10 software was used for data acquisition with subsequent analysis using Clampfit 10 software. Sampling frequency was 10 kHz and low pass filtering was with an external Bessel filter (LPF 202 a) at 2 kHz (Molecular Devices). Tabulated data of the electrophysiological properties of higher and lower neurons exposed to OGD can be found at http://hdl.handle.net/1974/7578. A subset of these neurons for this study were first exposed to elevated [K⁺]_o, and their properties are listed in Table 1. All of these neurons returned to their baseline properties following the brief K⁺ exposure. Electrophysiological data were presented as means ± standard deviation.

Table 1.

Whole-cell parameters during 2–3 min of [K⁺]_o elevated to 52 mM recorded from “higher” pyramidal neurons (PYR) in neocortex and CA1 region as well as from spiny neurons in striatum (total n = 14).

Neuron type	n	Rmp (mV)	AP ampl (mV)	Rin (MΩ)	Depol rate (mV/min)	SDKt rate (mV/s)
Neocortex PYR	5	Mean −82	78	82	NA	10 to 25
		SD ± 2.7	7.9	12.1
CA1 PYR	5	mean −73	82	97	NA	3 to 22
		SD ± 2.5	13.1	17.8
Striatum	4	mean −71	80	54	NA	12 to 30
		SD ± 4	10.5	15.2
Hypothal MNC	1	−62	94	569	15	NA
Hypothal Parvo	1	−51	70	778	18	NA
Hypothal SCN	1	−50	80	676	15	NA
Midbrain MES	1	−54	77	52	20	NA
Midbrain LC	1	−48	89	181	20	NA
Medulla SolN	1	−47	94	720	17.5	NA
Medulla DVM	1	−48	98	284	17.5	NA

These values are compared with “lower” neurons in hypothalamus and brainstem neurons (n = 7). All higher neurons undergo SD_Kt while lower neurons only slowly depolarize at a mean rate of 17.6 ± 2.0 mV/min.

Results

Imaging spreading depression

To generally assess the regional propensity of the CNS to undergo spreading depression, we exposed live coronal brain slices to elevated [K⁺]_o while imaging changes in light transmittance (ΔLT). Elevating bath [K⁺]_o to 26 mM for 3 to 5 min evokes a front of spreading depression (SD_Kt) in the mouse brain slice (n = 14 mice). The increased LT is caused by the swelling of neurons as the cells depolarize with astrocytes swelling secondarily.³⁴ The SD_Kt front (Figure 1(a), arrows) propagates in neocortex (NC) and hippocampus (Hp) but much less so in the brainstem with the exception of dorsal brainstem gray specifically, the superficial superior colliculus (Figure 1(a), SC), superficial inferior colliculi (IC) and periaqueductal gray (Figure 1(a), PAG), solitary nucleus (SolN), and tegmental nucleus (TN). All regions recover from SD_Kt-associated swelling.

Figure 1.

Elevating bath [K⁺]_o to 26 mM for 3 min triggers a front of spreading depolarization (SD_Kt) in the mouse brain slice. (a) The increased light transmittance (LT) is caused first by the swelling of neurons and then by astrocytes as they secondarily take up K⁺.^24,34 The SD_Kt front (arrows) propagates in neocortex (NC) and hippocampus (Hp) but not in the brainstem with the exception of the superficial colliculi (SC) and periaqueductal Continued.

gray (PAG). All regions recover from SD_Kt-associated swelling. (b) Changes in light transmittance (ΔLT) plotted in different brain region of interest in response to elevating [K⁺]_o to 26 mM. Each peak represents the SD_Kt front. Responses are normalized to account for regional differences in slice translucency. (c) Mean LT values plotted across three additional CNS regions, specifically cerebellar cortex (Cbell), inferior colliculus (IC), and tegmental nucleus (TN). There are no significant differences for the LT changes with kynurenic acid pre-treatment (not plotted, but see Figure 3).

SD_Kt was induced by exposing coronal mouse brain slices to 26 mM K⁺ aCSF for 3–5 min. Figure 1(b) and 1(c) illustrates several regional ΔLT time courses during simultaneous SD_Kt events. The CA1 region (46.5 ± 10.1%) and cerebellar cortex (54.0 ± 20.8%) had a significantly higher mean LT peak compared with other regions (*p < 0.05) (Figure 2(a)). The mean LT peak in SC was 34.9 ± 11.3% which was similar to the neocortex (33.9 ± 14.4%). PAG had a mean LT peak of 29.3 ± 17.6%. In addition, there was little difference in the mean LT peak among IC (32.4 ± 11.3%), TN (30.1 ± 7.3%), and SolN (32.0 ± 13.0%).

Figure 2.

K⁺-induced (26 mM) SD_Kt properties across different brain regions with or without 2 mM kynurenate pre-treatment. (a) SD_Kt–associated peak changes in light transmittance. Kynurenate blocks SD_Kt in PAG, tegmental nucleus (TN), and solitary nucleus (SolN) but does not affect peak signal strengths in the other regions. (b) K⁺-induced SD_Kt onset latency in minutes. Where SD_Kt is observed, onset latency is consistent across brain regions. Kynurenate pre-treatment does not significantly alter the time to onset. (c) The SD_Kt propagation rate (mm/min) in neocortex and the hippocampal CA1 region is significantly slowed by kynurenate pre-treatment (*p < 0.001) but not in other regions where it can initiate and propagate.

There was no significant difference in SD_Kt onset time among the responding regions (Figure 2(b)). The SD_Kt propagation rate in brainstem nuclei (Figure 2(c)) was 1.9 ± 1.2 mm/min in SC, 1.4 ± 1.5 mm/min in IC, 0.7 ± 0.2 mm/min in PAG, and 0.5 ± 0.2 mm/min in TN. These rates were significantly slower than the higher brain regions of neocortex (4.1 ± 0.8 mm/min), CA1 (3.4 ± 2.0 mm/min), and cerebellar cortex (2.6 ± 2.3 mm/min). SD_Kt spread through SC and IC at similar speeds. PAG, TN, and IC displayed significantly slower SD_Kt compared with neocortex (p < 0.05) (Figure 2(c)). Comparatively, the propagation rate was neocortex > CA1 > cerebellar cortex > SC > IC > PAG > TN. The small size of the solitary nuclei made it difficult to measure a rate.

Kynurenic acid did not significantly affect the mean peak LT in neocortex (n = 4), CA1 (n = 3), SC (n = 7), cerebellar cortex (n = 6), or IC (n = 4) (number of mice = 7) (Figure 2(a)). SD_Kt onset in the cerebellar cortex and neocortex were delayed by kynurenate compared with control (*p < 0.05). Kynurenic acid did not alter SD_Kt onset in CA1, SC, or IC (Figure 2(b)). In addition, the SD_Kt propagation rate tended to decrease in most regions including neocortex, CA1, SC, cerebellar cortex, and IC but only the propagation in neocortex and the CA1 region was significantly different (*p < 0.001) (Figure 2(c)). In some cerebellar and IC slices, we detected increases in LT but without obvious SD_Kt propagation (not shown). No SD_Kt was observed after kynurenate pre-treatment in PAG (6/6 slices) and SolN (5/5 slices). Only 1/9 TN slices displayed a signal with kynurenic acid treatment. The distinct LT increases in brainstem nuclei that emulated higher gray matter were only observed near the dorsal surface of brainstem and not in hypothalamus.

Intracellular recordings

As described in numerous previous studies, pyramidal neurons in layers II–III of the neocortex depolarize in response to elevated [K⁺]_o, leading to spreading depression (SD_Kt) that approaches zero millivolts (Figure 3(a)). In response to an initial 9.6 mM K⁺ exposure, a pyramidal cell passively depolarizes slightly and repolarizes in control aCSF. Exposure to 26 mM K⁺ elicits more depolarization with firing and then spike inactivation, before a plateau of −48 mV is reached. SD_Kt onset is aborted just as control aCSF reaches the slice (arrow). A second exposure again evokes firing and spike inactivation, reaching a plateau at −48 mV. A steep secondary depolarization signals SD_Kt onset. The membrane potential again returns to resting level after aCSF return. In contrast, hypothalamic neurons slowly depolarize without the active SD_Kt event (Figure 3(b1)). A magnocellular neuron in the supraoptic nucleus (SON) depolarizes in 52 mM K⁺ aCSF. No SD_Kt event that approaches zero millivolts is evoked. After each exposure, the cell recovers. In Figure 3(b2), a parvocellular neuron in paraventricular nucleus (PVN) depolarizes in 52 mM [K⁺]_o aCSF. Action potentials inactivate while slowly reaching a plateau but no SD_Kt is rises. With return to control aCSF, it repolarizes. Similarly, a suprachiasmatic nucleus (SCN) neuron responds to 3 min of 52 mM [K⁺]_o and depolarizes to a plateau as action potentials inactivate (Figure 3(b3)), but does not generate SDKt.

Figure 3.

Pyramidal neurons in layers II–III of the neocortex depolarize in response to elevated [K⁺]_o, leading to SD_Kt that approaches zero millivolts. In contrast, hypothalamic neurons slowly depolarize without the active SD_Kt event. (a1) In response to 5-min 9.6 mM K⁺, a pyramidal cell passively depolarizes slightly to −73 mV then repolarizes in control aCSF. Exposure to 26 mM K⁺ elicits more depolarization with action potential firing and then spike inactivation, before a plateau of −48 mV is reached. SD_Kt onset is aborted just as control aCSF reaches the slice (arrow). A second exposure to 26 mM K⁺ again evokes firing and spike inactivation, reaching a plateau at −48 mV. A steep depolarization then coincides with SD_Kt onset. Membrane potential again returns to resting level after aCSF return. (b1) A magnocellular neuron in the supraoptic nucleus (SON) depolarizes to −35 and −34 mV in response to 3 - and 4-min superfusion of 52 mM K⁺, respectively. No full-blown SD_Kt event approaching zero millivolts is evoked. After each exposure the cell recovers, albeit with reduced action potential amplitude. (b2) A parvocellular neuron in paraventricular nucleus (PVN) depolarizes in response to 52 mM [K⁺]_o exposure. Action potentials inactivate before slowly reaching a plateau at a level well below zero millivolts. No SD_Kt is generated. Following return of control aCSF, it repolarizes to near its original resting potential. (b3) An neuron in suprachiasmatic nucleus (SCN) responds to 3 min of 52 mM [K⁺]_o. As in B1 and B2, it depolarizes to a plateau as action potentials inactivate, but does not generate SD_Kt. Back in control aCSF, the neuron returns to the original resting potential. Figure modified from literature.^35,36

Exposing a neostriatal neuron to 100 s of 52 mM [K⁺]_o results in a rapid depolarization to near zero millivolts (Figure 4(a1)). A brief plateau at −30 mV (arrow) occurs before a second more rapid depolarization to near zero millivolts, typical of SD_Kt. Upon return of control aCSF, the membrane potential recovers. This contrasts to the typical response to oxygen/glucose deprivation by striatal neurons (Figure 4(a2)), where a typical “higher” neuron response is a rapid, complete, and irreversible “anoxic” depolarization (AD). In midbrain slices, a 3-min exposure to 52 mM [K⁺]_o elicited depolarization of a mesencephalic neuron (MES, n = 3) to −19 mV with loss of action potential firing (Figure 4(b1)). No SD_Kt is generated unlike in Figure 4(a1). Upon return of control aCSF, MES neurons return to their original resting membrane potential. Similarly in a locus ceruleus neuron, 3 min of 52 mM [K⁺]_o elicits depolarization and spike inactivation to −11 mV but no SD_Kt (Figure 4(b2)). Upon return of control aCSF, the LC neuron repolarizes.

Figure 4.

(a1) Striatal neuron’s response to 52 mM [K⁺]_o. The neuron rapidly depolarizes, reaching a brief plateau (arrow) before undergoing SD. After depolarizing to near-zero millivolts, return of control aCSF allowed repolarization to baseline potential. (a2) A different striatal neuron undergoes anoxic depolarization (AD) in response to oxygen/glucose deprivation (OGD). Typically, no recovery follows 10 min OGD by higher brain neurons. (b1) MES neuron response to 3 min of 52 mM [K⁺]_o. The neuron depolarizes, reaching a plateau of ∼ −20 mV as action potentials inactivate. No SD_Kt is generated. Upon return of control aCSF, the neuron returns to near-baseline. (b2) LC neuron response to 3 min of 52 mM [K⁺]_o. It plateaus but does not generate the additional depolarization of SD_Kt. Upon return to control aCSF. the neuron recovers. Figure modified from literature.^36,37

Figure 5 further contrasts a higher brain region (thalamus) with brainstem recordings from slices of medulla. A thalamic neuron responds to 120 s of 52 mM [K⁺]_o with a quick depolarization to near zero millivolts (Figure 5(a1)), typical of SD_Kt in higher neurons. The membrane potential returns to baseline upon return of control aCSF (not shown). As in Figure 4(a2), thalamic neurons also rapidly undergo AD when exposed to OGD (Figure 5(a2)). In lower neurons, a 3-min exposure to 52 mM [K⁺]_o elicits slow depolarization of a dNTS neuron with loss of action potential firing, before reaching a plateau of −30 mV (Figure 5(b1)). No SD_Kt is generated. With return of control aCSF, the SolN neuron repolarizes. Similarly, DVM neurons depolarize during 3 min of 52 mM [K⁺]_o, reaching a plateau of −29 ± 2.1 mV as firing stops (n = 8). Upon return of control aCSF, all DMV neurons repolarized to their original resting levels. The DMV neuron in Figure 5(b2) was subjected to two separate 52 mM [K⁺]_o exposures (3 min) that reached respective plateaus of −32 and −28 mV. As with the hypothalamic and brainstem neurons recorded in this study, no SD_Kt was generated (Table 1). The rate of depolarization from baseline to plateau by seven lower neurons during 2 min of 52 mM KCl exposure was 17.5 ± 2.0 mV/min. That is exceeding slow compared with the SD_Kt rate generated by higher neurons, which, once underway, ranges between 3 and 30 mV/s, as shown in Table 1.

Figure 5.

(a1) Thalamic neuron’s response to 52 mM [K⁺]_o. The neuron rapidly depolarizes, smoothly transitioning to SD_Kt. (a2) A second striatal neuron undergoes anoxic depolarization (AD) in response to oxygen/glucose deprivation (OGD). No recovery follows 10 min OGD. (b1) A brainstem neuron from the solitary nucleus (SolN) responds to 3 min of 52 mM [K⁺]_o. It depolarizes to a plateau as action potentials inactivate but SD_Kt is not generated. Upon return of control aCSF, the neuron repolarizes to near-baseline. (b2) A DMV neuron responds to two 3-min exposures of 52 mM [K⁺]_o. It gradually depolarizes, reaching a plateau as action potentials inactivate. No SD_Kt is generated. Following return to control aCSF, the neuron repolarizes and fires. During a second exposure of 52 mM [K⁺]_o, the neuron again plateaus without generating SD_Kt. Back in control aCSF, it repolarizes. Figure modified from literature.^35,36

Discussion

Recurring SD may impart some neuroprotection against further ischemic stress by preconditioning³⁸ or by promoting plasticity and regeneration as reviewed by Dreier.³ However, in general, SD has been considered a pathological process by clinicians and basic researchers alike. This is understandable because migraine aura can lead to terrible pain; ischemic SD can lead to neuronal death. On a different tack, we have proposed that the initial anoxic (ischemic) SD is protective immediately following TBI, because it causes a rapid behavioral shut-down as well as a vasoconstriction. We suggest that SD has evolved distant in our vertebrate past. Because occlusive stroke and cardiac arrest have only become prevalent recently as the human lifespan has increased,^19,20 sudden loss of blood flow in these non-traumatic cases will elicit SD and vasoconstriction (inverse neurovascular coupling) as though the brain region is hemorrhaging from TBI.²⁶ This may explain why SD during ischemia is robust in higher brain regions that are most exposed to impact. It follows that ischemic SD should be weaker and recoverable in the lower brain, and indeed that appears to be the case as reported here. Although not yet shown directly, it is likely that an ischemic depolarization in the brainstem, even though delayed and weaker than in higher brain, could disrupt respiration and prove deadly.^39,40

A less metabolically stressful version of SD is cortical spreading depression (CSD), which generates migraine aura.¹⁰ This event can be evoked by elevating [K⁺]_o, termed K⁺-triggered spreading depolarization (SD_Kt).^11,24,34 The current study shows that compared with high brain regions, SD_Kt is likewise weak or non-existent in the nuclei of hypothalamus and brainstem. This is demonstrated both by LT imaging as well as intracellular recording as discussed below.

With regard to imaging SD_Kt, all higher brain regions showed a robust front. Indeed, there are clinical correlates of SD_Kt, depending on whether it propagates in neocortex, hippocampus, striatum, or thalamus.^9,10 In contrast, hypothalamic and more ventral brainstem nuclei (including regions in which we subsequently whole-cell recorded) showed no spreading signal in slices, although LT could transiently and weakly increase during elevated [K⁺]_o. Only dorsal brainstem regions displayed a propagating LT front when [K⁺]_o was elevated, specifically in the superficial colliculi, PAG, tegmental nuclei, and solitary nuclei. From an evolutionary perspective, these dorsal regions would have been more exposed to trauma, especially prior to expansion of the cerebral hemispheres in mammals and birds. In contrast, we found that hypothalamus and ventral brainstem do not generate SD_Kt. In line with the findings here, hypothalamus and brainstem gray show only weak and recoverable AD in response to OGD.^19,20,35,37

Brain regions displayed K⁺-induced SD_Kt properties that varied in a rostral-caudal fashion. The most obvious difference was the propagation rate, which was faster in higher brain regions. In lower brain nuclei if a propagating signal was detected, the onset time was similar to higher regions. Also the strength of the peak signal at the propagating front was not significantly different among regions, except for the hippocampal CA1 region and cerebellar cortex where the higher signal likely reflected more concentrated neuronal packing. Pre-treatment with the general glutamate receptor antagonist kynurenate (2 mM) had no significant effect on the peak LT signal nor on the onset latency of the spreading front in all brain regions. However, the front’s propagation rate was consistently slowed in all regions if a spreading signal was observed.

Recording intracellularly from select neuronal types supported the imaging data. In the neocortex, striatum, and thalamus, 1–2 min of 26 or 52 mM [K⁺]_o caused an initial slow depolarization over several seconds that suddenly transitioned to a rapid depolarization that quickly approached zero millivolts i.e., classic spreading depression. Lower brain neurons did not undergo this additional rapid depolarization, despite longer exposure (3 min) to 52 mM [K⁺]_o. This is consistent with previous recordings in SON of hypothalamus and in brainstem of the adult rat^35,37 where neurons did not generate SD_Kt

The role of swelling in SD

Both astrocytes and neurons swell during AD^33,41 and SD_Kt.³⁴ Astrocytic swelling is more easily explained because their surface is studded with water channels that mediate passive water flux across their plasma membranes.⁴² As [K⁺]_o rises during SD, K⁺ and Cl⁻ are taken up followed by osmotic water. But with CNS neurons, swelling is more complicated because their plasma membrane lacks functioning aquaporins and so they do not passively swell or shrink over many minutes^33,41 across a range of osmolalities (−80 to +80 milliosmoles). Osmotic stress can acutely arise over a short period of time in a number of clinical situations.^43,44 The accumulation of water within neurons under metabolic stress occurs in at least two general ways. First, many ion and amino acid co-transporters, most energetically linked to the Na/K pump, move water across plasma membrane as they carry out their normal functions,^45–47 even against its concentration gradient.⁴⁸ Second, water is a catabolic by-product of metabolism and if the water pumping noted above fails as ischemia develops, the lack of neuronal aquaporins means accumulating water is trapped intracellularly. The result during metabolic stress is neuronal swelling because water cotransporters are disrupted so residual water produced would remain in the neuron.

Why does SD show a regional susceptibility?

The question arises as to why SD in higher gray matter is stronger and less reversible than in the lower gray matter. There is no obvious architectural reason why higher and lower regions differ in susceptibility to SD. Also there are no data indicating that higher brain regions consume more energy per unit volume than lower regions. Nor is there any evidence that a cortical neuron requires more energy than a brainstem neuron nor that neurons from higher vs. lower regions systematically differ in their ion channel composition. We think it more likely that there is a regional difference involving the molecular biology of Na⁺/K⁺ pump failure and the ensuing depolarization that results. The physiological cause of SD is pump failure or alternately, overriding the pump by raising [K⁺]_o. One approach is to look for regional variations in the expression of molecular transporters known to play important roles in maintaining normal metabolism.

For example, in situ hybridization data mined from the Allen Mouse Brain Atlas shows the regional expression pattern of the ATP1α1 pump and the glutamate transporter SLC17a7 (Figure 6). Gene expression primarily in higher brain regions is high for these two transporters or their related proteins, suggesting a potential specific interaction between them. Rose et al.⁴⁹ proposed that the Na⁺/K⁺ pump interacts through its γ-subunit because inhibiting the ATPase also partly inhibits the glutamate transporter. Such an interaction could promote extracellular glutamate accumulation during Na/K pump failure, extending ischemic injury in higher brain regions. Therefore, regional variation in acute ischemic injury may not simply involve an up- or down-regulation of metabolic pumps but also how they interact. In comparison to a wide variety of ligand and voltage-gated channels that have been proposed to malfunction in migraine and stroke (yielding little mechanistic insight), transporters have only been minimally studied.

Figure 6.

In situ hybridization data mined from the Allen Mouse Brain Atlas shows the regional expression pattern of the ATP1α1 pump and the glutamate transporter SLC17a7. Both transporters show high expression specifically in higher brain regions. The expression images highlight those cells with the highest probability of gene expression using a heat map color scale. The Expression Energy is calculated as follows: Within a given area A (voxel or structure), expression energy = (sum of intensity of expressing pixels in A)/(sum of all pixels in A). In the centre is depicted the crystal structures of a Na/K-ATPase and a glutamate transporter embedded in the cell membrane (modified from Rose and Koo⁴⁹). They proposed an interaction through the γ-subunit by showing that inhibiting the Na/K-ATPase also partly inhibits the glutamate transporter. Such an interaction could promote extracellular glutamate accumulation during Na/K pump failure, thereby increasing ischemic injury in higher brain regions where these transporters are highly expressed.

In summary, SD may contribute to survival both at behavioral and neuronal levels and so be selected during vertebrate evolution for the same reason as in insects. Both the locust^25,26 and fruit-fly²⁷ use SD to temporarily shut down their nervous system when temperature is too high or too low or when anoxia sets in. Similarly, we propose that SD arising from TBI elicits electrical silence and vasoconstriction in higher gray matter to preserve metabolic energy, to avoid epileptiform activity and to reduce hemorrhage. From a behavioral standpoint, a sudden loss of consciousness following TBI is preferable to seizure or uncoordinated movement in terms of avoiding the attention of enemies or predators. As a milder version of SD, we propose that SD_Kt is a vestigial response to local metabolic stress, and so it is generated in TBI-prone brain regions but is rare in hypothalamus and ventral brainstem.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

RDA planned and supervised experiments and wrote the manuscript. Y-TH and CDB carried out experiments, analyzed data, prepared initial figures, and contributed to early manuscript drafts as part of their theses work.