Author response: Postictal behavioural impairments are due to a severe prolonged hypoperfusion/hypoxia event that is COX-2 dependent Article Swipe

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Seizures are often followed by sensory, cognitive or motor impairments during the postictal phase that show striking similarity to transient hypoxic/ischemic attacks. Here we show that seizures result in a severe hypoxic attack confined to the postictal period. We measured brain oxygenation in localized areas from freely-moving rodents and discovered a severe hypoxic event (pO2 < 10 mmHg) after the termination of seizures. This event lasted over an hour, is mediated by hypoperfusion, generalizes to people with epilepsy, and is attenuated by inhibiting cyclooxygenase-2 or L-type calcium channels. Using inhibitors of these targets we separated the seizure from the resulting severe hypoxia and show that structure specific postictal memory and behavioral impairments are the consequence of this severe hypoperfusion/hypoxic event. Thus, epilepsy is much more than a disease hallmarked by seizures, since the occurrence of postictal hypoperfusion/hypoxia results in a separate set of neurological consequences that are currently not being treated and are preventable. https://doi.org/10.7554/eLife.19352.001 eLife digest It has long been known that after an epileptic seizure, individuals often experience an extended period of impairments that affect how the brain works. Because the brain is organized so that specific tasks happen in particular areas, seizures that affect areas of the brain that control movement are often followed by muscle weakness. Likewise, amnesia may follow a seizure that affects brain areas involved in memory. While these events reduce quality of life in people with epilepsy, they have gone untreated because we did not understand what occurs in the brain after seizures. It has been observed that the impairments that follow seizures are similar to those that follow strokes, where for a period of time blood flow to certain areas of the brain is restricted and these areas are starved of oxygen. Following a seizure, is there a local stroke-like event that is responsible for the behavioural and memory impairments? To address this question, Farrell et al. studied blood flow in the brains of mice, rats and human volunteers with epilepsy. The experiments show that after an epileptic seizure, blood vessels become narrower, which reduces blood supply to the areas of the brain involved in the seizure and dramatically reduces oxygen levels in those same areas. Using drugs to block the activity of an enzyme called cyclooxygenase-2 or other proteins called L-type calcium channels prevented both the oxygen shortage and the behavioural impairments that follow seizures. Thus, people with epilepsy are experiencing stroke-like events after seizures that they should be able to avoid with simple medical treatments. In the future, clinical research should determine how effective these treatments are in people with epilepsy. Other experiments should reveal if a shortage of oxygen after a seizure causes noticeable brain damage and the long-term behavioural problems that are often associated with epilepsy. https://doi.org/10.7554/eLife.19352.002 Introduction For proper neuronal functioning, brain tissue oxygen levels are normally maintained through exquisite regulation of blood supply (Erecińska and Silver, 2001). However, during neurological events when brain tissue oxygen levels fall below the severe hypoxic threshold (pO2 < 10 mmHg), neuronal dysfunction and behavioural disturbances are observed (Farrar, 1991; van den Brink et al., 2000; Maloney-Wilensky et al., 2009). Following termination of a seizure, behavioral impairments related to the specific brain structures that participated in the seizure are expressed (Leung et al., 2000; Gallmetzer et al., 2004). Todd’s Paresis, for example, specifically refers to moderate to severe motor weakness following seizures and typically subsides within a few hours (Todd, 1849). The symptoms are so similar to ischemic stroke that Todd’s paresis is often misdiagnosed (Mathews et al., 2008; Masterson et al., 2009). The occurrence of these behavioral impairments have been attributed to the affected cortex being ‘exhausted’ (Franck and Pitres, 1878) or silenced due to increased inhibition (Gowers, 1901), but these conjectures are not supported. Despite being first described over 160 years ago, the cause of postictal behavioral disturbances has not been determined. A few case studies of persons experiencing Todd’s paresis demonstrated that the affected cortex was inadequately perfused during this period (Mathews et al., 2008; Yarnell and Paralysis, 1975; Rupprecht et al., 2010). Thus, we hypothesized that following termination of a seizure, a long-lasting and severe hypoxic episode would occur and manifest as behavioral dysfunction. This simple idea is intriguing, testable and provides a critical new insight into this broad reaching neurological disease by explaining some of the detrimental consequences associated with epilepsy. But to date, a continuous and detailed examination of local tissue oxygenation and blood flow following seizures has not been carried out. Most investigations of the hemodynamics associated with seizures have concentrated on changes before or during the seizure and not during the postictal period. It is well established that there is a dramatic, transient increase in blood flow during seizures (Gibbs et al., 1934; Penfield et al., 1939) with a corresponding decrease in oxygenation that quickly recovers (Bahar et al., 2006; Suh et al., 2006). The few studies that investigated changes in blood flow during the postictal period have not yielded consistent observations with reports of local hypoperfusion (Rowe et al., 1991; Newton et al., 1992; Leonhardt et al., 2005) and hyperperfusion (Fong et al., 2000; Tatlidil, 2000; Hassan et al., 2012). These inconsistencies were likely due to the variable time-points after the seizure when blood flow was measured. To circumvent this problem we systematically investigated local oxygen levels and blood flow following evoked and spontaneous seizures in rats and mice. We then relied on those results to guide our clinical study and used arterial spin labeling (ASL) MRI to measure postictal hypoperfusion following spontaneous seizures in people with epilepsy. We further hypothesized that the enzyme cyclooxygenase-2 (COX-2), which is primarily responsible for catalyzing arachidonic acid to the PGH2-derived prostanoids (Hla and Neilson, 1992) and plays a major role in neurovascular coupling (Niwa et al., 2000; Lecrux et al., 2011; Lacroix et al., 2015), would mediate postictal hypoperfusion/hypoxia. We reasoned that electrographic seizures would engage this system and lead to a pronounced vasoconstriction and subsequent severe hypoxia. Downstream of this, we anticipated that elevated free calcium in vascular smooth muscle, particularly calcium conducted by L-type channels (Putney and McKay, 1999), would sustain this vasoconstriction. Here we demonstrate that blocking these pathways prevents postictal severe hypoperfusion/hypoxia and spotlights its contribution to postictal behavioral impairments. Results Long-lasting severe hypoxia follows spontaneous and evoked seizures A bipolar electrode and oxygen-sensing probe (optode) were chronically implanted into CA1 and CA3, respectively, of the dorsal hippocampus (Figure 1L). The optode continuously recorded the partial pressure of oxygen (pO2 in mmHg) in awake, freely-moving rats before, during and, most importantly, after brief electrographic seizures over several weeks. During typical behavioral states, hippocampal pO2 levels vary within the normoxia range between 18 and 30 mmHg (Figure 1A), as previously reported (Erecińska and Silver, 2001). 10 mmHg oxygen was chosen as the threshold for defining severe hypoxia since several independent studies have demonstrated that pO2 levels at or below 10 mmHg cause significant changes to cellular physiology and brain injury. Hypoxia-dependent gene expression via Hypoxia-Inducible Factor 1 transcriptional activation occurs as an exponential function with respect to decreasing pO2 (Jiang et al., 1996). In a HeLa cell culture system, 10–15 mmHg was determined to be the half-maximal response and since the steep portion of the exponential curve is at lower pO2 levels, most gene expression occurs below this threshold. Restricting middle cerebral artery blood flow in the cat led to cell death at a threshold of 7–8 mmHg (Farrar, 1991). Furthermore, both the depth and duration of severe hypoxia following traumatic brain injury are good predictors for clinical outcome (van den Brink et al., 2000; Maloney-Wilensky et al., 2009). 10 min of pO2 below 10 mmHg was associated with increased risk of death, and further increases with lower pO2 and longer duration of severe hypoxia (van den Brink et al., 2000). A systematic review of clinical traumatic brain injury found that 10 mmHg was a suitable threshold for defining severe hypoxia associated with worse outcomes (Maloney-Wilensky et al., 2009). In defining tumor hypoxia, which is thought to exacerbate cancer pathophysiology, 8–10 mmHg is the suggested threshold (Höckel and Vaupel, 2001). Thus, 10 mmHg is a reasonable threshold for defining severe hypoxia and integrating the entire area below 10mmHg combines the depth and duration to reveal the total hypoxic burden. Figure 1 with 1 supplement see all Download asset Open asset Seizures induce severe postictal hypoxia. (A) Local tissue oxygenation in the hippocampus of an awake, freely-moving rat (blue). Green denotes normoxia while red denotes severe hypoxia. (B) Representative oxygen profile before, during, and after a spontaneous seizure. The inset expands the time-scale during the seizure with the green and red lines denoting the beginning and end of an 80 s seizure. This inset corresponds to the white vertical block near the beginning of the full oxygen recording. (C) Representative oxygen profile before, during, and after a 106 s electrically kindled seizure. (D) Scatterplot of the relationship between the duration of kindled seizures in the dorsal hippocampus (primary afterdischarge) and the degree of severe hypoxia expressed as the total area below the severe hypoxic threshold (10.0 mmHg) by time (min). Symbols with both the same colour and shape are from the same animal (n = 14). The line of best fit (y = 11.85x+7.57) is indicated. R square = 0.55, p<0.0001. (E–J) Hypoxyprobe immunohistochemistry. (E) Representative image from control rat with close-ups of CA3 (H) and hilus (G). (F) Representative image following a seizure with close-ups of CA3 (J) and hilus (I). Scale bar for (E,F) = 400 µm. Scale bar for (G–I) = 150 µm. (K) Densely stained neurons in the hilus and CA3 were quantified. There are significantly more stained cells in the hilus and CA3 following seizures. Data are mean ± SEM. *p<0.05 (t-test). (L–N) Comparing hippocampus and neocortex. (L) Location of chronic hippocampal implants. (M) Location of chronic neocortical implants. Bipolar electrodes were used for stimulating and recording seizure and O2 sensors for continuous oxygen recordings. C.C. is corpus callosum, L5 is layer 5. (N) Inset displays mean seizure duration ±SEM (n = 5). Hippocampus had significantly longer seizures. *p<0.05. The mean pO2 (opaque) ±SEM (transparent) over time recorded in motor neocortex dorsal hippocampus. (O) Quantification of (N). Hypoxia was more severe in the hippocampus relative to neocortex as assessed by the area below 10mmHg. *p<0.05 (t-test). https://doi.org/10.7554/eLife.19352.003 Oxygen levels were then monitored in the intrahippocampal kainate model of temporal lobe epilepsy (Rattka et al., 2013) before, during, and after a spontaneous seizure. We measured a brief and small drop in hippocampal oxygenation at seizure onset and a subsequent increase in oxygenation during the electrographic seizure. However, oxygen levels dropped precipitously to below the severe hypoxic level (pO2 < 10 mmHg) following seizure termination (Figure 1B). Even more striking was that the severe hypoxia was sustained for an hour. We then systematically examined this phenomenon in a higher throughput model of evoked focal seizures: electrical kindling. Using electrical kindling we elicited an afterdischarge (i.e. electrographic seizure) by applying brief stimulation (1 ms pulse widths at 60 Hz for 1 s) through the chronically implanted bipolar electrode. A brief dip in oxygenation at seizure onset and subsequent increase was observed (Figure 1C). Importantly, the severe hypoxic event following a kindled seizure was markedly similar in magnitude and duration compared to spontaneous seizures (Figure 1C). Since no substantive differences in postictal hypoxia were observed between the two models, we chose to evoke afterdischarges with kindling stimulation as our principal model for subsequent experiments to maximize data collection efforts. Daily repeated stimulation of the dorsal hippocampus gave rise to a wide range of primary afterdischarge durations across rats (10–50 s) and allowed us to determine that there is a strong positive and linear relationship between the duration of the hippocampal seizures and the severity of postictal hypoxia, as revealed by integrating the area below 10 mmHg (Figure 1D). To provide convergent evidence for this postictal hypoxia phenomenon, we also used an immunohistological staining technique based on pimonidazole (HypoxyprobeTM-1), which selectively labels hypoxic cells (Chapman et al., 1981; Höckel and Vaupel, 2001). Rats were injected with pimonidazole and afterdischarges elicited. Since postictal severe hypoxia in the rat hippocampus lasted a mean of 73.1 ± 7.8 min (Figure 1N), we examined brains one hour post-seizure. We observed significantly increased numbers of pimonidazole-labeled cells in the dentate hilus and CA3 regions (Figure 1E–K) relative to non-seizure controls. Thus, two different and unrelated techniques identified a severe hypoxic period following brief seizures. We then asked the question whether the phenomenon could be observed in a different structure. Rats were chronically implanted with an electrode in the corpus callosum to elicit neocortical afterdischarges and an optode was implanted into layer V of motor neocortex (Figure 1M). Postictal severe hypoxia was also present in the motor cortex where the typically shorter neocortical afterdischarges gave rise to higher minimum pO2 and a more rapid return to baseline oxygen levels, relative to hippocampus (Figure 1N). Quantification of severe hypoxia in these two structures revealed significantly worse severe hypoxia in the hippocampus (Figure 1O), likely due to longer seizure durations. To further ensure that postictal hypoxia was not specific to the method of seizure induction we also tested two other experimentally-induced seizure models. Maximal electroconvulsive shock (MES) which models generalized convulsions (Young et al., 2006) and 3 Hz electrical stimulation, which drives epileptiform activity and behavioral seizures (Teskey and Racine, 1993) both induced a long-lasting hypoxic event in the hippocampus (Figure 1—figure supplement 1) similar to those observed following spontaneous or electrical kindled seizures. We conclude that a postictal severe hypoxic event follows epileptiform seizures regardless of the eliciting technique. Hypoperfusion mediates postictal severe hypoxia Severe hypoxic events are often the consequence of inadequate blood flow mediated by vasoconstriction and determining the contribution of reduced blood flow to hypoxia can identify new routes for intervention. We first determined the relationship between local tissue hypoxia and blood flow using an implantable laser Doppler flowmetry (LDF) probe to measure relative changes in hippocampal blood flow while simultaneously monitoring pO2 levels in the CA3 region of the rat dorsal hippocampus (Figure 2A). Figure 2B displays concurrent measures of mean tissue pO2 with blood flow. Blood flow increased during afterdischarges, which was consistent with previous observations (Metzger, 1977). Importantly, the postictal severe hypoxia is accompanied by reduced blood flow to 53.0 ± 8.3% relative to baseline between 40 and 60 min post-seizure. Furthermore, blood flow recovered on a similar time-scale to tissue oxygenation indicating that the postictal severe hypoxia is, in part, related to hypoperfusion. Figure 2 with 1 supplement see all Download asset Open asset Seizures cause postictal vessel constriction in an in vitro preparation and reduced blood flow (hypoperfusion) in vivo. (A) Location of implants for simultaneous blood flow and pO2 recordings. These two probes were placed at opposing angles to leave room for cable attachment. (B) The simultaneous measurement of mean blood flow and mean pO2 in the hippocampus following brief seizures (n = 5). (C) Nifedipine pre-treatment (15 mg/kg) caused an elevation of baseline pO2, inhibited severe hypoxia, and increased the rate of recovery (n = 5). Inset reveals no difference in seizure duration. (D) Quantification of (C). Nifedipine pre-treatment reduced the amount of severe hypoxia (area below 10mmHg). *p<0.05 (within-subject ANOVA). (E) Validation of 3 Hz stimulation in young rats (P28–P35). Mean oxygen profiles following standard kindling and 3 Hz stimulation (n = 6). (F) Quantification of (E). No significant differences were found in the amount of severe of hypoxia (area below 10 mmHg). (G) Acute hippocampal in vitro slice preparation for measuring postictal vessel constriction. 3 Hz stimulation was applied to the Schaeffer collaterals and imaging was captured in stratum radiatum of CA1. (H,I) Representative images of CA1 arteriole pre-stimulation (H) and post-stimulation (I). Scale bar for (H) and (I) is 15 µm. (J) Lumen diameter over time in stimulated and sham controls. Mean lumen diameter is reduced following 3 Hz stimulation (n = 11) relative to sham (n = 8) between 60 and 90 min post-stim. Data displayed as mean ± SEM. **p=0.001 (t-test). (K) 30 min following nifedipine (50 µM; n = 7) or vehicle (n = 5) application slices were stimulated. Mean lumen diameter was significantly different between 60 and 90 min post-stim. **p<0.01. https://doi.org/10.7554/eLife.19352.005 If the postictal hypoxia is due to a vascular event then blocking calcium conductance should reduce vasoconstriction and prevent hypoxia. Pre-treatment with the L-type calcium channel antagonist nifedipine 30 min prior to seizure elicitation did not alter the afterdischarge duration (Figure 2C) indicating that nifedipine effects can be attributed to a neurovascular mechanism. We observed higher pO2 levels at baseline (+9.0 mmHg, p<0.01, paired ANOVA), throughout the postictal period, and an earlier return to baseline relative to vehicle (Figure 2C). Moreover, nifedipine pre-treatment significantly attenuated the area below the severe hypoxic threshold (pO2 < 10 mmHg) (Figure 2D). We also observed that nifedipine was effective at reducing postictal severe hypoxia when administered immediately following termination of the afterdischarge (Figure 2—figure supplement 1A,B). This is not surprising since L-type calcium channels play a key role in the tonic phase of vascular smooth muscle contraction (Putney and McKay, 1999). These results provide key evidence that hypoxia is largely mediated by hypoperfusion and that L-type calcium channel activity plays a role in sustaining this response. While laser Doppler flowmetry provides inferential evidence that blood flow was reduced postictally, we sought a direct measure of changes in blood vessel diameter following epileptiform-like activity in the hippocampus. In young (P25-P40) freely moving rats, we first determined that 2 min of 3 Hz stimulation resulted in a remarkably similar profile of severe hypoxia compared to the standard 1 s stimulation evoked afterdischarges (Figure 2E,F). We then moved the 3 Hz stimulation protocol to an acute hippocampal slice preparation (Figure 2G) and imaged hippocampal arterioles in the synaptic layer of CA1 (stratum radiatum) before, during, and after Schaffer collateral stimulation (Figure 2H,I). We observed sustained 13.8 ± 4.0% arteriole constriction following stimulation (60–90 min post-stim, Figure 2J). Poiseuille’s law dictates that a 13.8% reduction in diameter would lead to blood flow that is 55.2% of baseline, which correlates well with our laser Doppler flowmetry measurements (53.0 ± 8.3% at 40–60 min, Figure 2B). Pretreatment with nifedipine (50 µM) caused significant vessel dilation (108.4 ± 2.9% relative to baseline, p<0.05, paired t-test) and prevented 3 Hz stimulation-induced vessel constriction (Figure 2K). Our data obtained using laser Doppler flowmetry, nifedipine treatment, and arteriole constriction in slice all provided convergent evidence that severe hypoxia following seizures is mediated by vasoconstriction and hypoperfusion. Postictal hypoperfusion in clinical epilepsy Based on our time-course data generated from model systems, we asked whether there was postictal hypoperfusion within one hour of a spontaneous ictal event in epilepsy patients. All patients had intractable focal epilepsy and the mean age was 32 years (range 20–42 years). Two patients were excluded from the study due to excessive motion during the acquisition of postictal ASL images, leaving 10 patients for analysis (Table 1). The mean age at seizure onset was 12.3 years (range one month-29 years). Four patients had MRI had of one had an one had changes and one had changes as a result of previous blood flow was in postictal and baseline a seizure) ASL Postictal ASL images were from baseline to reveal areas of hypoperfusion. Maximal changes in blood flow were in and the brain area where this was compared to ictal to determine between local hypoperfusion and seizure activity (Table 1 of patients to ASL at MRI in the temporal of changes in the temporal 2 of postictal ASL hypoperfusion with brain areas involved in the seizure. of ictal activity was determined by an based on denotes when a generalized seizure of hypoperfusion were identified by a and of were those areas and for (Figure 2—figure supplement 1A,B). was if ictal activity with the area of hypoperfusion. of seizure of hypoperfusion for ASL temporal Maximal areas over temporal and to Maximal postictal cerebral blood flow of at 10 were in of 10 (Table at the region of for all 10 we observed a mean baseline of ± and mean of ± following seizures, for a difference of ± or ± from baseline (Figure supplement Furthermore, the magnitude of hypoperfusion was with seizure duration (Figure 2—figure supplement as we with the magnitude of hypoxia and seizure duration in the rat (Figure 1D). In the two hypoperfusion of the seizure durations were shorter and may for the of hypoperfusion. these two of ± relative to baseline was observed in the postictal While we observed a in the rat (53.0 ± 8.3% relative to the for ASL also of which would decrease the reported and the relative of should be This of allowed us to the between the brain areas of hypoperfusion and areas of seizure activity to determine the of this Two are in Figure 3 and provide of this in and temporal lobe epilepsy. In all where hypoperfusion was we that the area of hypoperfusion with a brain structure to seizure activity (Table Thus, hypoperfusion is specifically localized to those areas involved in the seizure and is a local Figure 3 with 1 supplement see all Download asset Open asset Representative local postictal hypoperfusion in clinical epilepsy. (A) with focal epilepsy. recording on a bipolar with seizure onset and to the Green seizure the electrodes to seizure Scale bar = 1 and also to (B) and recovery images area of with white over the region R and and and also to (C) and (C) the image indicating areas of hypoperfusion reduction compared (D) with intractable epilepsy. recording on a bipolar with temporal seizure onset and to the region (E) was (F) hypoperfusion in temporal Postictal severe hypoxia is mediated by activity during seizures is expressed throughout the rat brain (Figure with particularly expression in neurons of the hippocampus (Figure This expression profile and the role in neurovascular coupling (Niwa et al., 2000; Lecrux et al., 2011; Lacroix et al., a likely responsible for postictal hypoperfusion/hypoxia. We also hypothesized that would not play a role since is involved in the tonic control of brain blood flow through et al., inhibitors for these two of have been and we or to and had significant on afterdischarge duration and as (Figure but not (Figure supplement prevented postictal severe hypoxia. Figure with 1 supplement see all Download asset Open asset activity during a seizure is for postictal severe hypoxia. (A) expression in inset is displayed Scale bar = 2 of inset from neurons (B) in and hilus (C). (D) displays of and Scale bar = µm. (E) pre-treatment mg/kg) inhibited severe hypoxia during the postictal period (n = 5). Inset reveals no difference in seizure duration. (F) caused a significant reduction in the area below 10mmHg. *p<0.05 ANOVA). (G) inhibited postictal severe hypoxia (n = 5). are on inset Inset reveals no difference in seizure duration. (H) 150 and of significantly the area below 10mmHg. *p<0.05 ANOVA). (I) Lumen diameter over time following application