Address correspondence to Dr. G. Campbell Teskey, Department of Cell Biology and Anatomy, University of Calgary, Calgary, AB T2N 4N1, Canada. E-mail: firstname.lastname@example.org
Purpose: Neural reorganization and interictal behavioral anomalies have been documented in people with epilepsy and in animal seizure models. Alterations in behavior could be due to somatosensory dysfunction. This study was designed to determine whether seizures can lead to changes in somatosensory representations and whether those changes are persistent.
Methods: Twice-daily seizures were elicited by delivering 1 s of electrical stimulation through carbon fiber electrodes implanted in both the corpus callosum and sensorimotor neocortex of young adult male Long-Evans rats until a total of 20 seizures were elicited. Either 1–3 days or 3–5 weeks following the last seizure, functional magnetic resonance imaging (MRI) was used to image the brain during electrical stimulation of each forepaw independently.
Key Findings: Forepaw stimulation in control rats resulted in a focused and contralateral fMRI signal in the somatosensory neocortex. Rats that had repeated seizures had a 151% increase in the number of voxels activated in the contralateral hemisphere 1–3 days after the last seizure and a 166% increase at 3–5 weeks after the last seizure. The number of voxels activated in response to forepaw stimulation was positively correlated with the duration of the longest seizure experienced by each rat. The intensity of the activated voxels was not significantly increased at either time interval from the last seizure.
Significance: The increased area of activation in somatosensory cortex, which is persistent at 3–5 weeks, is consistent with previous observations of larger motor maps following seizures. Seizure-induced changes in the functioning of sensory cortex may also contribute to interictal behavioral anomalies.
Experimental animal models of seizures have been useful in addressing how seizures affect the functional organization of various brain structures (Pitkänen et al., 2006). Seizures can induce changes in neocortical movement representations and behavior (for a review see Teskey et al., 2008). We have shown previously that seizure activity in the neocortex can result in larger motor maps of the forelimb (Teskey et al., 2002; van Rooyen et al., 2006; Ozen et al., 2008; Young et al., 2009) as well as nonforelimb areas (Henderson et al., 2011) and can lead to alterations in forelimb performance on specific motor tasks (Henry et al., 2008; Flynn et al., 2010). Because successful behavior relies on both motor and sensory information, altered sensory processing may contribute to the behavioral deficits seen following seizures. Regarding sensory processing, it has previously been shown in feline primary auditory neocortex that neuronal firing responses to various sounds are dramatically altered following experimentally induced repeated seizures in that structure (Valentine et al., 2004, 2005). Whether changes to the cortical processing of somatosensory information can be detected in response to experimentally elicited repeated seizures has not yet been reported.
One method of evaluating somatosensory processing is by acquiring functional activation maps using functional magnetic resonance imaging (fMRI) during somatosensory stimulation. fMRI has been useful in providing insight into neural plasticity following stroke, demyelination, and spontaneous seizures (Baron et al., 2004; Hodics & Cohen, 2005; Ward, 2005; Lidzba et al., 2006; Cousin et al., 2008; Pillai, 2010). fMRI detects local increases in MR signal intensity due to decreases in the concentration of deoxyhemoglobin [blood oxygen level dependent (BOLD) signal] (Logothetis et al., 2001), which is associated with increases in neural activity and increased blood flow (Tuor, 2009). Three main questions were addressed in this study. First, we examined if 20 repeated experimentally induced seizures of relatively short duration (non–status epilepticus) result in increased volume of neocortical somatosensory activation to forepaw stimulation and if the effect is persistent. We also examined if the duration of individual seizures is correlated with the volume of activation. We employed electrical stimulation of the forepaw to provide reliable and repeatable sensory input (Wang et al., 2006; Tuor et al., 2007) for producing activation maps measurable with fMRI. We hypothesized that the volume of activation during forepaw stimulation would increase in rats who had repeated seizures both 1–3 days and 3–5 weeks following the last seizure.
Materials and Methods
Fifteen male Long-Evans hooded rats weighing 310–468 g were obtained from Charles River (Sherbrooke, QC, Canada). Rats were housed individually in clear plastic cages in a colony room that was maintained on a 12 h on/12 h off light cycle. All experimentation was conducted in the light phase. Rats were given Lab Diet #5001 (PMI Feeds Inc, St Louis, MO, U.S.A.) and had water ad libitum. The rats were handled and maintained according to the Canadian Council for Animal Care guidelines and according to experimental protocols approved by the local Animal Care Committees.
Rats (n = 15) were divided into three groups: those that did not receive any seizures (controls, n = 5); those receiving 20 seizures and imaged within 1–3 days following the last seizure (1–3 day group, n = 5); and those receiving 20 seizures and imaged 3–5 weeks following the last seizure (3–5 week group, n = 5). All rats, with the exception of two control rats, were implanted with carbon fiber electrodes in the corpus callosum and in the sensorimotor cortex to allow for the induction and recording of seizures. The two control rats without implanted electrodes and control rats with implanted electrodes were combined for analyses of the number of voxels activated. This was done because the mean number of voxels activated to forepaw stimulation in the two rats without implanted electrodes were both within one standard deviation of the control group with implanted electrodes.
Carbon fiber electrodes were produced according to Dunn et al. (2009). Briefly, electrodes were constructed from bundles of individual carbon fibers (Formosa plastics corporation, Taipei, Taiwan) to achieve a final thickness of 0.2–0.4 mm and a length of 10–20 mm. The carbon fibers were attached to the slot of a brass screw (Spaenaur, Kitchener, ON, Canada) using silver print #842 (MG Chemicals, Toronto, ON, Canada) as a conducting adhesive. Carbon fiber bundles were insulated using three coats of the nonconductive polymer polyvinylidene difluoride (PVDF) (Whitford Corporation, Frazer, PA, U.S.A). Electrodes were baked at 200°C for 10–15 min and cooled for 20 min between coats. Before implantation, electrodes were cut to achieve the desired length and to remove insulation from the tip.
Rats were anesthetized with an intramuscular injection of a cocktail of 8.5 mg/kg of 100 mg/ml ketamine (85%) and 1.5 mg/kg of 100 mg/ml xylazine (15%). Throughout the surgery, supplemental injections (0.05 ml) of the cocktail were administered when required. A subcutaneous injection of Lidocaine (2%) was administered at the incision site. One monopolar carbon fiber electrode was implanted into the corpus callosum (1.0 mm anterior to Bregma and 0.5 mm lateral to midline, 3.0 mm ventral from brain surface) and another one in the right sensorimotor neocortex (1.0 mm anterior to Bregma and 4.0 mm lateral to midline, 1.5 mm ventral from brain surface) according to the stereotaxic coordinates of Swanson (1992). The electrodes were cemented and anchored with four plastic screws to the skull with dental cement, leaving the ends of the brass screws on the electrodes exposed for stimulating and recording procedures.
An afterdischarge threshold (ADT) was determined for each animal in the seizure groups. The ADT was defined as the weakest current required for inducing an afterdischarge (AD). The current delivery commenced at 100 μA, increasing in steps of 100 μA, and was delivered at 60 s intervals until an AD of at least 4 s or longer was recorded. The electroencephalography (EEG) signals were filtered, at half amplitude, below 0.3 Hz and above 300 Hz, and then amplified 1,000 or 2,000 times (Grass Neurodata Acquisition System Model 12, Astro-Med, Inc., West Warwick, RI, U.S.A.). The stimulating current used to elicit seizures was delivered to awake, freely moving rats.
Twice daily, seizure-inducing electrical stimulation with a minimum of 2 h between stimulation was delivered through both the electrode positioned in the anterior corpus callosum and in the neocortex. Stimulation consisted of a 1 s train of 60 Hz biphasic rectangular wave pulses, 1 ms in duration and separated by 1 ms, at an intensity 100 μA greater than the ADT levels. The duration of each AD was recorded, and the longest AD duration per animal was used for further analysis. The seizure behaviors were monitored and scored according to a 5-stage scale (Racine et al., 1972).
Functional MR imaging
Functional MR images were taken during stimulation applied to a single forepaw according to a previously established protocol (Wang et al., 2006; Tuor et al., 2007). Once the data had been collected for the first forepaw the other forepaw was stimulated. Therefore, each fMRI session yielded activation scores for each forepaw independently. Quantitative comparisons of the mean number of voxels activated in response to forepaw stimulation, and their intensity, were compared between control rats, rats in the 1–3 day group, and rats in the 3–5 week group.
Rats were prepared for fMRI under isoflurane anesthesia. The femoral vein and artery were cannulated for drug and fluid administration, measuring blood pressure, and obtaining arterial blood samples. Rats were intubated and mechanically ventilated to maintain normal paO2 and paCO2. The rat was moved to an MR cradle, where the head was immobilized using ear pins and an incision bar and rectal body temperature was maintained at 37 ± 0.5°C with a circulating water blanket. Anesthesia was changed to intravenous α-chloralose with an initial dose of 80 mg/kg maintained with 20 mg/kg every 45 min or as needed. Pancuronium (2 mg/kg, intravenous) was administered to provide muscle relaxation. The left and right forepaws (individually) were electrically stimulated using two subdural silver needle electrodes and a Grass stimulator. An off-on-off-on-off stimulation time course was followed with stimulation on during images 13–26 and 39–48. A stimulation paradigm of 3 Hz monophasic rectangular wave pulses, 3 ms in duration, and at an intensity of 3 mA was used for the duration of the “on” phase of the stimulation time course eliciting a repeated and reproducible fMRI response.
Rats were scanned using a 9.4T/21 cm bore MR system with a Bruker Avance II console and a 24 mm × 18 mm elliptical surface coil positioned over the forebrain. After shimming, scout images were used to select five transverse adjacent slices of 1.5 mm thick from 1.75 mm anterior to 5.75 mm posterior to bregma containing the sensorimotor cortex. fMRI was collected using a multislice, RARE sequence [Rare factor – 64, Repetition time (TR) = 3,000 ms, effective echo time = 60 ms, field of view = 3 cm × 3 cm, matrix size = 128 × 128]. Each fMRI scan consisted of 64 sets of five slice images, each acquired in a time of approximately 6 s/image during periods of electrical stimulation of the right or left forepaws.
The fMRI scans were analyzed using local software developed by the Institute for Biodiagnostics (EvIdent) (Jarmasz & Somorjai, 2002). Voxel intensities were normalized by subtracting the median intensity value over images 1–12 of the time course (chosen as baseline) and noise was reduced using a spatial median filter (3 × 3 block). Motion was minimal due to restraint of the head with ear pins. For each fMRI scan, the entire brain within each coronal slice was selected for analysis. The intensity changes in each voxel (0.008 mm3) over time were correlated to the off-on-off-on-off stimulation time course for that scan using a Pearson’s correlation analysis with a threshold of p < 0.00001 (selected as 1/number of voxels) to correct for multiple tests (Bullmore & Suckling, 2001). “Active” voxels were those voxels correlating significantly to the off-on-off-on-off time course; therefore, the “activation response” was the number of voxels correlating to the time course. A minimum of five voxels of activation was considered necessary for a significant activation response. Two imaging trials were removed from the analysis due to lack of activation, and one imaging trial was removed from analysis due to abnormal blood pressure (below 50 mm Hg or above 120 mm Hg) as blood pressure can influence the fMRI response to forepaw stimulation (Wang et al., 2006; Tuor et al., 2007).
All data are reported as mean ± standard error of the mean (SEM). To determine if there were hemispheric differences in activation responses, the mean number of voxels activated (t25 = 1.541, p = 0.136) and the intensity of activation (t21 = 1.208, p = 0.241) were compared and found to be not significantly different. Thus two scores (right and left) were entered for each rat in the subsequent analyses. Two one-way analyses of variance (ANOVAs) were used to compare the mean number of voxels activated and intensity of the activated voxels during forepaw stimulation between the three groups (control, 1–3 day group, 3–5 week group) followed by a Newman-Keuls test for a comparison of means. Whether there was a significant correlation between the AD duration and the number of voxels activated (the higher number of voxels activated in either the left or right paw for each animal) was assessed using a Pearson’s correlation. A regression analysis was used to evaluate whether seizure duration was related to volume of activation. All statistical tests used a significance level of 0.05.
The ADT for the combined seizure groups was 435 ± 55 μA, with seizures successfully elicited each session (Fig. 1). All rats displayed an increase in seizure severity with repeated stimulation (adjusted R2 = 0.257; F1,7 = 8.257; p = 0.009). Rats tended to display stage three seizures on the first session, as indicated by unilateral forelimb clonus, subsequently progressing into stage five seizures indicated by rearing and falling (loss of postural control) and the recruitment of hind limb clonus.
Effect of repeated seizures on the somatosensory response
In all groups, voxels of activation were identified in the sensorimotor cortex in the hemisphere contralateral to the stimulated forepaw. There was no significant difference (t20 = 1.49, p = 0.15) in the number of voxels activated between the electrode implanted and nonimplanted hemispheres, indicating that the presence of the electrode did not significantly interfere with the BOLD response. In addition, there were no significant differences in the mean number of voxels activated to forepaw stimulation between control rats with implanted electrodes and the two control rats without implanted electrodes, further showing that the presence of an implanted electrode does not alter the BOLD response.
Overall, rats that had seizures had a greater number of voxels activated both in the 1–3 day group and in the 3–5 week group (Fig. 2). Two of five rats in the 1–3 day group showed a bilateral response to forepaw stimulation (Fig. 3). Control rats and rats in the 3–5 week group showed only a contralateral response to forepaw stimulation.
There was an overall difference in the number of voxels activated (Fig. 4) between the groups (F2,24 = 4.233, p = 0.027). Follow-up multiple comparisons showed that compared to controls the mean activation was higher in both the 1–3 day group (p < 0.05) and the 3–5 week group (p < 0.05). There was no significant difference in the mean number of activated voxels between the 1–3 day group and the 3–5 week group (p > 0.05), indicating that there was no significant decrement in the seizure-induced increase in responsiveness for up to 3–5 weeks.
The number of activated voxels for rats that had seizures was found to be positively correlated with the AD duration (r = 0.586, p = 0.037), indicating a linear relationship between seizure duration and the volume of activation in the somatosensory cortex following repeated seizures (Fig. 5).
There was no difference in the intensity of activated voxels between controls (6.08 ± 0.45), the 1–3 day group (6.78 ± 0.39), and the 3–5 week group (6.63 ± 0.49, F2,20) = 0.593, p = 0.562).
Three novel findings are presented in this study. This study provides the first demonstration of both a short-term (days) and persistent (weeks) enhanced MR responsiveness to somatosensory stimulation following repeated experimentally induced seizures. In addition, this study is also the first to describe a strong positive relationship between seizure duration and the volume of the activation measured with fMRI. We showed that rats that had repeated seizures induced in the sensorimotor neocortex had a greater number of activated voxels in response to electrical forepaw stimulation. This increase in voxel number was present both in the 1–3 day group and in the 3–5 week group. A bilateral response to stimulation of one forelimb was also seen in some of the rats in the 1–3 day group. In addition, the seizure-induced changes in somatosensory activation to forepaw stimulation did not show degradation in volume at 3–5 weeks following the last seizure. This is similar to the findings by Ozen et al. (2008), where they showed that larger forelimb motor maps following repeated seizures persisted and did not degrade in size for the maximum time examined: 5 weeks. The persistence of an augmented activation response also supports the idea that these seizure-induced effects are not directly related to the seizure events themselves or are transiently expressed, but are related to the long-term changes caused by repeated seizures (Dennison et al., 1995; Racine et al., 1995). Taken together, these results indicate that in addition to larger forelimb motor maps (Teskey et al., 2002; van Rooyen et al., 2006; Ozen et al., 2008; Young et al., 2009) and nonforelimb motor maps (Henderson et al., 2011) as well as behavioral alterations (Henry et al., 2008; Flynn et al., 2010), repeated seizures also affect the organization of the somatosensory neocortex.
The close spatial and temporal relationships between neural activity and cerebral blood flow form the basis of fMRI as a measure of brain activity. Larger BOLD signals are associated with local decreases in deoxyhemoglobin, which are evoked by increased oxygen consumption. This is thought to reflect increased metabolic activity that is related to both local synaptic activity (Goense & Logothetis, 2008) as well as spiking activity (Lee et al., 2010; Young et al., 2011) and results in local increases in cerebral blood flow. Seizure-induced changes to electrophysiologic measures such as enhanced evoked potentials (Teskey et al., 2002) and increased auditory unit responses to sound (Valentine et al., 2004, 2005) suggest that the increased fMRI signal to somatosensory stimulation is due, at least in part, to changes in local neural responsiveness. There was no change in the intensity of the activated voxels following seizures, indicating that changes in responsiveness to forepaw stimulation reflect a recruitment of neighboring sensory areas rather than simply an increase in blood flow to previously responsive areas. However, it is possible that vascular changes still may have occurred to allow the increased number of voxels to be activated. Factors involved with vasodilation, including cyclooxygenase-2 (COX-2) and nitric acid synthase, have been shown to be upregulated following electroconvulsive seizures (Newton et al., 2003). Other vascular changes may have occurred as well, allowing for enhanced oxygen delivery and perhaps accounting for some of the difference in responsiveness of the somatosensory cortex following seizures. The efficiency of the vasculature may also be altered, potentially contributing to the greater volume of fMRI signal. The idea that angiogenesis might occur in response to a discrepancy between metabolic demand and oxygen availability is supported by the observation that capillary density increases following hypoxic conditions (LaManna et al., 2004), wheel running in rats (Kleim et al., 2002), and sensory and motor training (Black et al., 1991; Isaacs et al., 1992; Swain et al., 2003). Similarly, following electroconvulsive seizures, angiogenic factors including vascular endothelial growth factor, fibroblast growth factor, and neuropeptide Y are also upregulated (Newton et al., 2003). In addition, electroconvulsive seizures induced in the rat hippocampus have been shown to induce the formation of new capillaries (Hellsten et al., 2005; Pitkänen et al., 2006). In people with temporal lobe epilepsy, angiogenic factors such as vascular endothelial growth factor are overexpressed (Croll et al., 2004; Rigau et al., 2007). These previous observations support the idea that seizure-induced vascular changes may also occur, allowing for enhanced oxygen delivery to the somatosensory cortex during forepaw stimulation, thus partially accounting for our observation of increased fMRI responsiveness to forepaw stimulation. To summarize, it is likely that changes to both neuronal and vascular organization may account for our observation of an increased volume of activation to forepaw stimulation.
Seizure-induced synaptic potentiation may account for the strong positive relationship between seizure duration and the volume of activated sensorimotor cortex, given that affected neurons during a seizure fire in a highly repetitive and highly synchronous pattern. This synchronous firing likely incurs the Hebbian rule that neurons that fire in temporal and spatial proximity incur stronger synaptic connections (Teskey et al., 2002; Valentine et al., 2004). This may explain why seizure duration is positively correlated to the number of voxels activated during forepaw stimulation, as longer exposure of synchronized firing would more strongly wire affected neurons together. The induction of neocortical long-term potentiation (LTP) in Long-Evans rats and rats that have been genetically selected to be seizure-prone (FAST rats) have larger movement representation areas (Flynn et al., 2004; Monfils et al., 2004). As well, seizures have been shown to induce robust synaptic potentiation in the pathways involved in seizure development (Racine et al., 1972, 1995; Teskey et al., 2002). As the duration of each seizure increases, it is likely that there is more synaptic potentiation accounting for the positive correlation between the length of individual seizures and the volume of activation in response to forepaw stimulation.
Although seizure-induced synaptic potentiation may explain the size and occurrence of bilateral activation to forepaw stimulation following seizures, it is unclear why bilateral activation does not persist for up to 5 weeks, like contralateral activation. Regular use of the forelimbs and the absence of seizures driving bilateral forelimb clonus may account for why bilateral activation was not seen 3–5 weeks following seizures. Nudo et al. (1996) has shown in squirrel monkeys that had enlarged digit motor representations following skilled digit training was reduced after cessation of training. It is possible that increased volume of activation to forepaw stimulation may also decrease in the contralateral hemisphere due to continued use of the forelimbs. Bilateral activation is likely to be pruned out earlier than contralateral activation. Most receptive fields of the cortex of mammals respond to contralateral rather than bilateral or ipsilateral input (Mountcastle, 1957; Armstrong-James & George, 1988), and would likely not be activated as much during regular forelimb use.
In summary, we have demonstrated that repeated neocortical seizures result in a greater volume of activation to somatosensory stimulation that is persistent. We have also demonstrated a strong positive relationship between seizure duration and the volume of activation to somatosensory stimulation that occurs following seizures. These results are similar to the clinical observations that some people with epilepsy have differently organized sensory maps (Uematsu et al., 1992; Urasaki et al., 1994). Possible mechanisms underlying the observed increase in neurovascular responsiveness include synaptic potentiation and vascular changes. Future studies should aim to further elucidate the mechanisms responsible for seizure-induced increases in neocortical responsiveness in the sensorimotor cortex, as well as how these changes relate to alterations in interictal behavior.
Supported by a CIHR grant to G.C.T., an NSERC grant to J.D., an AI-HS scholarship to J.D., an NSERC CGS to J.V., and an AI-HS studentship to A.K.H.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.