|Home | About | Journals | Submit | Contact Us | Français|
Spinal hyperexcitability and hyperreflexia gradually develop in the majority of stroke patients. These pathologies develop as a result of reduced cortical modulation of spinal reflexes, mediated largely indirectly via relays in the brainstem and other subcortical structures. Cortical control of spinal reflexes is markedly different in small animals, such as rodents, while in some larger species, such as cats, it is more comparable to that in humans. In this study, we developed a novel model of stroke in the cat, with controllable and reversible inhibition of cortical neuronal activity appearing approximately 1h after initiation of low-frequency electrical stimulation in the frontal cerebral cortex, evidenced by a large increase in the alpha frequency band (7–14Hz) of the frontal electrocorticographic signal. Hyperreflexia of the urinary bladder developed 3h or more after induction of reversible cortical inactivation with optimized stimulation parameters (frequency of 1–2Hz, amplitude of 10mA, applied for 30min). The bladder hyperreflexia persisted for at least 8h, and disappeared within 24h. At the S2 level of the spinal cord, where neural circuits mediating micturition and other pelvic reflexes reside, we have recorded an increase in neuronal activity correlated with the development of hyperreflexia. The low-frequency stimulation-induced reversible cortical inactivation model of stroke is highly reproducible and allows evaluation of spinal hyperexcitability and hyperreflexia using within-animal comparisons across experimental conditions, which can be of great value in examination of mechanisms of spinal hyperreflexia following stroke or brain trauma, and for developing more effective treatments for these conditions.
Stroke is the third leading cause of death in the United States (Wolf et al., 1992) and is the main cause of long-term disability among adults, producing more than $30 billion yearly in health care costs (Gorelick, 1995). About 30% of stroke victims are permanently disabled with a variety of impairments, including memory loss; loss of vision, speech, and motor control; voiding problems; and sexual dysfunction. Voiding dysfunction is an important factor in the long-term morbidity of stroke patients. A stroke is reported to produce incontinence in, respectively, 45–60% and 18–23% of acute and chronic stroke survivors (Wade and Hewer, 1985; Borrie et al., 1986; Wade and Hewer, 1987; Taub et al., 1994; Nakayama et al., 1997; Brittain et al., 2000; Patel et al., 2001). Post-stroke bladder dysfunction is generally a result of bladder hyperreflexia (Khan et al., 1981; Sakakibara et al., 1996), which is present in 48% of patients within 3 days after stroke (Burney et al., 1996), and in 70% of long-term stroke survivors (Siroky, 2003). Patients sustaining a stroke in the frontoparietal cortex or the internal capsule are even more susceptible to development of bladder hyperreflexia (Burney et al., 1996). Conservative management of bladder dysfunction after stroke involves catheterization and anticholinergic drug treatment, but both of these treatments have a significant risk of side effects, such as urinary tract infections and urinary retention (Corujo et al., 1999). Development of novel therapies is impeded by a lack of reproducible animal models of bladder hyperreflexia. While it is possible to induce bladder hyperreflexia by experimental cerebral infarction (Yokoyama et al., 1997, 2000) or cerebral ischemia (Yotsuyanagi et al., 2006) in an animal model, the extent of the cerebral lesion is quite variable (Aspey et al., 1998), and the lesion is irreversible, necessitating an inter-animal experimental design. Rodents have become the most widely used animals for experimental stroke studies. However, the anatomical and functional organization and connectivity pattern of the rodent frontal cortex are quite different from those found in humans and primates (Uylings et al., 2003; Povysheva et al., 2007). Specifically, in rodents, there are multiple direct projections from the frontal cortex to regions in the medulla and spinal cord involved in autonomic control (Hurley et al., 1991; Holstege et al., 1996; Gabbott et al., 2005). Due to this direct autonomic control, infarction of the rodent frontal cortex results in immediate (within 30min) development of bladder hyperreflexia (Yokoyama et al., 1997, 2000), in contrast with the several hours (less than 72h in the majority of patients) required for development of stroke-induced bladder hyperreflexia in humans (Burney et al., 1996). Animals with more evolved frontal cortices, and the cat in particular, may provide a better animal model of stroke-induced bladder hyperreflexia. The pre-cruciate gyrus in the cat is well developed and homologous with the frontal cortex in humans and nonhuman primates (Hassler and Muhs-Clement, 1964; Brutkowski, 1965), and has been implicated in the cortical control of voiding (Strom and Uvnas, 1950; Gjone and Setekleiv, 1963; Neafsey, 1990).
Stroke-like temporary and localized cortical inactivation can be achieved in large animals with more human-like organization of the cerebral cortex using cooling and pharmacological agents (Lomber, 1999; Martin and Ghez, 1999). This can significantly improve reproducibility of the “lesion” while minimizing animal use and inter-animal variability. Low-frequency electrical stimulation (LFES) has recently emerged as a more practical and efficient method of inducing reversible cortical inactivation (RCI). Unlike cooling, there is no neuronal hyperexcitability at the periphery of the inactivation site (Brooks, 1982), and LFES can be done with a chronically-implanted device and thus repeatedly induce RCI in the same animal. Unlike pharmacological treatment, the effects of LFES can be localized to a small region of the cortex beneath the implanted electrodes. In cortical slices, LFES applied at 1–2Hz was shown to produce long-term depression of neuronal activity, while stimulation at 4Hz had no significant effect, and high-frequency stimulation at 40–100Hz actually potentiated synaptic transmission (Kirkwood and Bear, 1994; Chen et al., 1996). In vivo, repetitive daily 15-min sessions of LFES at 1 and 10Hz gradually induced cortical depression that reached a maximum over the course of 1 week (Froc et al., 2000; Werk et al., 2006; Teskey et al., 2007). One study indirectly evaluated an effect of LFES on spinal reflexes. When single electrical pulses were applied through subdural electrodes on the precruciate frontal cortex, they produced inhibition of corticospinal excitability (Hanajima et al., 2002), which represents a combination of intracortical, subcortical, and spinal effects (Nielsen et al., 1999; Talelli et al., 2006).
In this study, we conducted the first experimental evaluation of LFES-induced RCI on bladder reflexes and spinal neuronal activity. We hypothesized that LFES can induce reversible spinal hyperreflexia and potentially be used as an animal model of stroke-induced spinal hyperreflexia.
Eight male cats, 1–2 years in age, weighing 2.5–4kg, were purchased from Liberty Research Inc. (Waverly, NY). The animal studies were conducted according to National Institutes of Health guidelines and were approved by the Huntington Medical Research Institutes Animal Care & Use Committee. Six animals were used in the experimental group and two animals served as controls. Animals in the experimental group were anesthetized with isoflurane and nitrous oxide and placed in a stereotaxic apparatus. A mid-line scalp incision was made, and the muscle and the periosteum were reflected to expose the skull. Craniectomies were performed using a Hall drill. Two oval-shaped platinum electrodes, 6×10mm in diameter and 25μm thick were insulated on their outer surface with epoxy and bent to conform to the surface of the frontal cortex. They were placed subdurally over the pre-cruciate gyri of the frontal cortex of both hemispheres, and the dura was closed with 7-0 sutures. The cables, extending caudally from the electrodes, were sutured to the dura and fixed to the edge of the craniectomy with medical-grade cyanoacrylate cement to hold the electrodes in place. The stimulating electrodes were referenced to an indifferent platinum ground electrode, uninsulated on both surfaces, which was placed between the muscular fascia of the scalp and the cranial surface at the parieto-occipital cranial junction.
Both frontal sinuses were packed with gelfoam and sealed with bone cement. The exposed cortices were covered with fascia harvested through an incision over the perispinal muscles and overlaid with sterile compressed sponge sheets. The craniectomies were sealed with several layers of bone cement. A percutaneous connector with the cables from all electrodes was fixed to the occipital area of the skull using stainless screws and cranioplasty cement. The scalp was closed in layers.
For spinal cord recording, an array with 12 microelectrodes was implanted in one animal. The electrode cable and attached microelectrode array were routed subcutaneously from the percutaneous connector mounted on the skull. A dorsal laminectomy was performed at the L3–L6 vertebral levels. To locate the junction of the S1 and S2 segments, the perigenital skin was stimulated using a pair of needle electrodes inserted approximately 20mm apart, while the evoked responses were recorded at several rostro-caudal locations along the dorsal surface of the sacral cord. The array was implanted 5mm caudal to the maximum of the second component of the evoked response (the dorsal cord potential) (McCreery et al., 2004). Later, at autopsy, the position of the array relative to the spinal cord level was validated by a complete dissection of the sacral spinal roots. For array implantation, a longitudinal incision was made through the dura at the S2 level, and the dorsal roots were retracted laterally. The array was inserted into the cord using a custom-made inserter tool, at a velocity of approximately 1m/sec (McCreery et al., 2004). A stabilizing pad of polyester mesh, attached to the cable 10mm from the array's superstructure, was tamped onto the dura to prevent the transmission of torque and longitudinal forces from the cable to the array. The dura was then closed over the array. The cable was secured to the dura and to the L5 vertebral process. The indifferent electrode for the spinal cord array was placed on top of the spinal cord, between two pieces of sterile compressed sponge. The muscle and skin over the laminectomy were closed in layers.
For the first 2 weeks after surgery, the animals were housed in a room without perches to reduce spinal mobility and allow the array superstructure and cable to become stabilized by connective tissue. No neurological problems were encountered that were related to implantation of the cortical surface electrodes and intraspinal microelectrode array. To prevent infection, the area around the percutaneous connector was cleaned daily in each animal. If an animal did develop an infection, as manifested by elevated body temperature and discharge of exudate around the percutaneous connector, it was treated with oral antibiotics. In the rare cases when an acute inflammation developed along the subcutaneous leads, the inflamed tissue was surgically removed, while the surrounding tissue was thoroughly treated with topical antibiotics before closing the skin. In one animal, despite these procedures a brain abscess developed and the animal had to be euthanized. Data from this animal were excluded from analysis due to a possible influence of cortical inflammation on the RCI induction by LFES. In the remaining five animals, immuhohistochemical analysis of the frontal cortical tissue indicated a lack of effect of chronic surface electrode implantation and stimulation on neuronal and astroglial survival and morphology (data not shown).
Beginning at 3 weeks after implantation of the electrodes, the cats in the experimental group were tested weekly for induction of RCI by LFES, while the control animals were tested only once. Light propofol anesthesia (5–7mg/kg/h intravenously [IV]) was maintained with the animal breathing on its own and responding to strong sensory stimulation. In the anesthetized animals, the baseline electrocorticographic signal and intravesical pressure (IVP) were collected for 1h, and LFES was applied in their frontal cortex using biphasic square current-controlled pulses. These pulses were 2, 6, or 10mA in amplitude, had stimulus pulse duration of 0.4ms and a frequency of 0.5, 1, 2, 4, or 8Hz, and were applied for 10, 30, 60, or 120min in order to produce the stroke-like condition of RCI. The order of applied frequencies, amplitudes, and durations of the LFES was randomized to avoid the possibility of systematic carry-over stimulation effects. Biphasic stimulation of the same intensity but opposite polarity was applied simultaneously to each cortical electrode in a bipolar configuration with a common indifferent electrode placed epicranially. To prevent dehydration during prolonged anesthesia (up to 12h), an IV drip of saline was administered (10mL/kg/h). LFES was not associated with pain or distress, thus no analgesics or tranquilizers were administered.
For recordings of electrocorticograms (ECoGs), the signals from the frontal cortical electrodes were amplified using a differential amplifier (Model FA8D; Multi Channel Systems, Reutingen, Germany), digitized at 100Hz using a 12-bit data acquisition board (Model PCI-6070E; National Instruments, Austin, TX), and displayed and stored on a computer using custom software written in Visual Basic (Microsoft Corporation, Redmond, WA) and Measurement Studio ActiveX components (National Instruments, Austin, TX). Time-frequency distribution of ECoG power was assessed using the zero-interval subtraction (ZIS) algorithm of estimating power spectrum (Marchenko and Rogers, 2006). The ZIS algorithm, generously provided by D. Marchenko in the form of Matlab (MathWorks, Natick, MA) M-code, uses a sliding zeroed segment to generate a series of fast Fourier transforms (FFTs) (using Matlab function pwelch), and then to calculate their difference spectra for adjacent time intervals. The obtained time-frequency distribution was visualized using Matlab's isocontour plots. The ZIS algorithm was chosen based on its speed and its ability to more accurately calculate time-varying power at lower frequencies compared with parametric FFTs. Values of mean spectral frequency and total spectral power were derived directly from the ZIS time-frequency distribution.
For recordings of the IVP, a double-lumen transurethral catheter was inserted via the urethral orifice. The animals were catheterized using aseptic procedures and checked by animal personnel on a weekly basis for urinary tract infections and other infections and given prophylactic antibiotics; consequently, none of the animals developed infections during the study period. Once catheterized, the bladder was filled with sterile saline through one of the lumens of the catheter until equilibrium was reached with the hydrostatic pressure in the open-top reservoir, 7–9mm Hg, which is below the threshold for reflex bladder contractions (20–25mm Hg), and reflects the storage phase of the micturition cycle. The connection to the reservoir remained open for the duration of the experiment in order to maintain constant pressure inside the bladder. Monitoring of dynamic bladder pressure waves while connected to the constant pressure reservoir was done in order to remove confounding influences of slow pressure drifts due to changes in arterial pressure, respiration, endocrine function, or pelvic floor motility. Bladder pressure changes were measured via a pressure transducer (Model 041500503A; Maxxim Medical, Athens, TX), connected to another lumen of the catheter. The diameter of the lumen was large enough (1.3mm) and short enough (10cm) to convey dynamic pressure changes up to 15Hz without phase or amplitude distortions (Li et al., 1976), which we found to be an acceptable level of bladder change. Connection with the open-top reservoir allowed reliable measurement of bladder pressure changes down to a frequency of 0.5Hz. The signal from the pressure transducer was amplified using a physiological monitor (Model 78534A; Hewlett-Packard, Palo Alto, CA), digitized at 10Hz (34% of the data) or 100Hz (and averaged to 10Hz, 66% of the data), and stored using the same data acquisition board and software used to record the ECoG. Thus bladder activity was effectively measured in the frequency band of 0.5–5Hz. Dynamic bladder activity was calculated in 10-min epochs as the standard deviation of this activity. Onset of bladder hyperreflexia was defined as the beginning of a 10-min epoch during which average dynamic bladder activity was at least 50% higher than the baseline level recorded before stimulation.
An array of 12 intraspinal Parylene®-coated iridium microelectrodes, 75μm in diameter, was chronically implanted at the S2 level of the spinal cord in one animal, according to a previously-described technique (McCreery et al., 2004). Neuronal activity was recorded in several LFES-induced RCI experiments. The recorded signals were amplified and bandpass filtered at 100–10,000Hz (FA16I; Multi Channel Systems), digitized at 25kHz, and stored using the same data acquisition board and software used for ECoG. In the subsequent off-line analysis, extracellularly-recorded action potentials (“spikes”) were detected in the signal using a simple two-threshold detector algorithm, adapted from the open-source public-domain software PowerNAP (neuroshare.sourceforge.net). Average neuronal activity was calculated for each record and expressed as the number of spikes per second. Spike sorting was not performed, and thus spikes detected during multiple experiments in the same animal may reflect the activity of more than one neuron. In each RCI experiment, the first 120-sec record was collected before the start of LFES, and the second 120-sec record was collected after the onset of bladder hyperreflexia (6–10h after initiation of LFES).
The animals were sacrificed by deep anesthesia (IV injection of pentobarbital, 50mg/kg) and transcardial perfusion with a pre-wash solution consisting of phosphate-buffered saline and 0.05% procaine HCl for 30sec, followed by 4% paraformaldehyde. The spinal cord and frontal cortices were removed, and transverse sections were cut at a thickness of 7μm and immunostained with antibodies for the neuronal marker NeuN (MAB377, 1:2000; Chemicon, Temecula, CA) and for astrocytic marker GFAP (Z0334, 1:10000; Dako Corp., Carpinteria, CA). Tissue sections were photographed using a digital microscope camera (Spot RT; Diagnostic Instruments Inc., Sterling Heights, MI). Cortical tissue sections under the surface electrodes were compared to cortical sections away from the electrodes to identify possible differences in neuronal and astrocytic appearance and density. Intraspinal microelectrode tip locations were traced along with the outline of the spinal cord, boundaries between the gray and white matter, and location of the central canal. By combining the drawings of spinal sections through the individual electrode tip sites, a composite outline of the spinal cord was generated.
Values of data sets are expressed as mean±standard deviation. Throughout the study, we used the General Linear Model (SPSS; SPSS Inc., Chicago, IL), followed by Bonferroni post-hoc tests for the experimental conditions relative to baseline, and linear and non-linear regression analyses (SigmaPlot; Systat Software Inc., San Jose, CA). p Values and coefficients of determination are as indicated in the figure legends.
RCI experiments were performed 45±19 (mean±SD) times in each of five animals with at least weekly intervals between experiments. RCI was induced by a train of biphasic charge-balanced LFES applied to the surface subdural electrodes, located bilaterally in the frontal cortices. Immunohistochemical evaluation of cortical tissue under the electrodes and adjacent tissue indicated a lack of damage to cortical neurons and astrocytes from chronic implantation and electrical stimulation. Figure 1 (top panel) shows an ECoG record of a typical RCI experiment, induced by LFES. Frontal cortical LFES was applied for 30min at amplitude of 10mA and at frequency of 1Hz. Characteristically, the ECoG signal increased soon after the frontal cortical stimulation was discontinued, and persisted through the end of the experiment (for ~7–8h). Examination of the time-frequency distribution indicated an increase in ECoG power at frequencies from 7–14Hz, with a mean frequency of 10–11Hz. Evaluation of the ECoG recordings from 18 experiments in three animals (Fig. 2) indicated that the mean spectral frequency of the ECoG signal was not significantly altered, and the total spectral power of the ECoG increased 100% or more following RCI induction at all stimulation frequencies (0.5–8Hz).
Bladder activity was measured while the bladder was coupled to a saline reservoir at 7–9mm Hg, which allowed monitoring of dynamic pressure changes in the range from 0.5–5Hz, and this dynamic activity was averaged in 10-min intervals as the standard deviation. Averaged dynamic bladder activity increased transiently during LFES of the frontal cortex, and then again after a considerable delay (Fig. 3). The initial transient increase persisted for 1–2h, and this was a universal finding among all animals, and there was no effect of different frequencies or duration of stimulation. The delayed post-stimulation bladder hyperactivity developed gradually, beginning at 3–4h after the initiation of LFES at 1Hz (Fig. 3A), and at 6h after the initiation of LFES at 8Hz (Fig. 3B), and persisted for the remainder of the experiment. This delayed post-stimulation bladder hyperactivity appeared to be somewhat bistable, with a low-variability state during the initial gradual increase, and a high-variability state for the remainder of the experiment. However, no quantification to support the possibility of bistability was performed. Bladder hyperreflexia, defined as a delayed increase in mean dynamic bladder activity (50% or more above the baseline) in one or more 10-min intervals, was evident at all frequencies of frontal cortical LFES (Fig. 4). Examination of the time delay after the initiation of LFES until the onset of bladder hyperactivity indicated a strong positive correlation of this delay with the frequency of LFES (Fig. 5) (linear function, adjusted R2=0.94). Bladder hyperreflexia, induced by stimulation at 8Hz, developed considerably slower compared to stimulation at 0.5, 1, and 2Hz. Bladder hyperreflexia was frequently observed until the end of an experiment, 8–10h after initiation of LFES. Concerns about animal welfare prevented us from keeping the animals under anesthesia for longer periods of time. In several experiments, the animals were re-anesthetized 24h after initiation of LFES, and at that time their cortical and bladder activities had returned to baseline levels. Possible effects of prolonged propofol anesthesia on bladder pressure activity were evaluated in two control animals. The animals remained anesthetized for 7–8h and their dynamic bladder pressure remained unchanged at 0.4–0.5mm Hg (data not shown).
The amplitude of cortical LFES was also an important factor in inducing bladder hyperreflexia. While cortical stimulation at 2mA induced no significant bladder hyperreflexia, stimulation at 6 and 10mA had a similarly strong effect on dynamic bladder activity (Fig. 6).
In addition to stimulation frequency and amplitude, we have examined the effect of the duration of the cortical LFES. Barely detectable bladder hyperreflexia could be induced by 10min of frontal cortical LFES at an amplitude of 10mA and frequencies of 1–4Hz, while prominent bladder hyperreflexia was evident after 30min of stimulation (Fig. 3). No additional increase in the amount of bladder hyperreflexia was seen following cortical stimulation for 60, 90, or 180min (data not shown), possibly due to a ceiling effect.
In one animal, we evaluated the effect of LFES-induced RCI on neuronal activity in the S2 spinal cord, which is predominantly involved in processing of the bladder voiding reflex and other pelvic reflexes. The animal was chronically implanted with an array of intraspinal microelectrodes. The tips of three microelectrodes were located in the gray matter (Fig. 7), and of these, the tip of electrode 7 was the closest to the dorsal gray commissure region near the central canal, which is intimately involved in bladder voiding reflexes (Pikov et al., 2007). Multi-unit neuronal activity was recorded before LFES and during bladder hyperreflexia induced by LFES-induced RCI at multiple frequencies (Fig. 7). Statistical analysis of mean neuronal activity recorded from all three electrodes indicated neuronal hyperactivity during frontal cortical stimulation at 1 and 2Hz, with the greatest increase seen at electrode 7. The time course of development of neuronal hyperactivity was not evaluated since the recordings were done later in the recording sessions, 6–10h after the initiation of LFES.
We have developed a novel model of stroke-induced spinal hyperexcitability and bladder hyperreflexia. Unlike the models in which a cortical infarct is actually inflicted, in this model the stroke-like effects on the spinal cord activity are temporary and reversible. RCI was reliably induced by LFES, applied at a range of frequencies from 0.5–8Hz, and was manifested as emergence of synchronized cortical activity in the alpha frequency band (7–14Hz). After a considerable delay, the duration of which was dependent upon the frequency of LFES, bladder hyperreflexia emerged and persisted for the duration of the experiment (at least 8h). The optimal parameters of frontal cortical stimulation for inducing spinal neuronal hyperactivity and bladder hyperreflexia were: an amplitude of ±10mA, a frequency of 0.5–8Hz, and a duration of 30 or 60min.
The RCI-related increase in frontal ECoG signal in the alpha frequency band (7–14Hz) in this study was similar to increases in ECoG/EEG signals seen during anesthesia (Sonn and Mayevsky, 2006), slow-wave sleep (Karasinski et al., 1994; Duckrow and Zaveri, 2005), and chemical cortical inactivation with a GABAA-receptor agonist (Vyazovskiy et al., 2007). In several experimental paradigms, it was demonstrated that an increased EEG signal in the alpha band reflects the inhibition of asynchronous cortical neuronal activity and increased influence of thalamocortical activity in the alpha band (Steriade et al., 1993; Pfurtscheller, 2001, 2003; Hughes and Crunelli, 2005). The synchronized neuronal activity in the alpha band (7–14Hz) is generated by thalamic reticular neurons, which receive topographically organized connections from the cortex (Guillery and Harting, 2003; Pinault, 2004), and are transmitted to the cortex via thalamocortical relay neurons (Yousif and Denham, 2005; Huguenard and McCormick, 2007). Due to this topographic organization of the thalamocortical input, the alpha-band activity can be localized in the occipital cortex as the occipital alpha rhythm, in parietal and frontal cortices as the mu rhythm, or in temporal cortex as the third rhythm (Hughes and Crunelli, 2005).
We believe that LFES-induced RCI resulted from temporarily reduced neuronal excitability in the stimulated area of the frontal cortex. Support for this hypothesis comes from four separate lines of study. First, a cortical slice study directly evaluated the effect of LFES on neuronal excitability and indicated that the maximal inhibitory effect of LFES on layer III neurons was achieved at 1–2Hz (Kirkwood and Bear, 1994). Second, in-vivo animal studies of cortical excitability evaluated the velocity of propagation of cortical spreading depression in response to epidural LFES, and demonstrated decreased velocity at 1Hz, but not at 20Hz (Fregni et al., 2005; Fregni et al., 2007). Third, a human study evaluated the effect of LFES on suppression of spontaneous seizures and found that most seizures were transiently suppressed by 0.5-Hz subdural stimulation (Schrader et al., 2006). And fourth, several human studies used a related form of cortical stimulation, repetitive transcranial magnetic stimulation (rTMS), applied at 1Hz, to produce a long-lasting decrease (at least 30–60min) in cortical excitability, measured as reduced duration of the cortical silent period (Stefan et al., 2000; Munchau et al., 2002; Fitzgerald et al., 2004). Several studies indicate that decreased cortical excitability in rTMS was due to induced long-term depression of cortical neurons (Chen et al., 1997; Heide et al., 2006).
In a real ischemic stroke there is considerable loss of neurons within the zone of infarct. In addition, decreased intracortical excitability was documented in the peri-infarct region, using electrophysiological recordings in the affected cortical hemisphere in rats with experimental transient middle cerebral artery occlusion (Neumann-Haefelin and Witte, 2000). Similarly, in stroke patients, a prolonged decrease in the excitability of cortical neurons was inferred from a reduced duration of the cortical silent period following rTMS, and the amount of decrease positively correlated with the severity of their chronic motor deficits and spasticity (Catano et al., 1997). Reduced long-term excitability of neurons in the peri-infarct region of a cortical stroke is likely attributable to ischemia- and hypoxia-induced secondary excitotoxic injury, affecting glutamate receptor expression in survived neurons (Lo et al., 2003). LFES-induced RCI may thus provide an animal model exhibiting decreased cortical excitability similar to that seen in actual stroke; however, is not presently known whether similar mechanisms for downregulation of cortical excitability are involved in RCI and actual stroke.
To our knowledge, this study provides the first evidence of a direct influence of frontal cortical LFES-induced RCI on spinal excitability and spinal reflexes. The evidence is provided for frequency-dependent effects of LFES-induced RCI on neuronal activity in the S2 ventral spinal cord and on bladder reflexive activity. The effect of LFES on the excitability of spinal neurons was strongest at 1 and 2Hz and weakest at 8Hz. One human study assessed the effect of rTMS on the monosynaptic H-reflex, a specific measure of spinal segmental excitability, and demonstrated the development of hyperreflexia at 1Hz versus development of hyporeflexia at 20Hz (Valero-Cabre and Pascual-Leone, 2005). In contrast, long-term electrical stimulation at 25Hz and 10% duty cycle (1sec on/9sec off) in sensorimotor cortex of the rat produced a lasting increase rather than decrease in the amplitude of the H-reflex (Chen et al., 2007). These differences in H-reflex plasticity following cortical stimulation may be a result of a differential pattern of corticospinal projections in rodents and primates: primates have multiple direct cortical projections to spinal motoneurons (Lemon et al., 2004), while rodents lack these (Yang and Lemon, 2003), and instead have well-developed monosynaptic rubro-motoneuronal projections (Kuchler et al., 2002). In addition to direct or indirect effects on motoneurons, changes in cortical excitability may lead to hyperreflexia via a reduced supraspinal modulation of spinal inhibitory interneurons, mediating presynaptic inhibition of Ia afferents or reciprocal Ia inhibition (Jankowska and Hammar, 2002). These as well as other possible modes of supraspinal control over segmental spinal reflexes may play a role in plasticity of spinal excitability following stroke, and the animal model of LFES-induced RCI can facilitate their detailed examination.
A surprising finding of this study is a considerable delay in development of bladder hyperreflexia following frontal cortical stimulation. At all frequencies of stimulation, bladder hyperreflexia developed more than 3h after the initiation of LFES, and this delay was correlated with the frequency of LFES. This can be contrasted with the time course of development of RCI. As can be seen in Figure 1, increased ECoG activity in the alpha band, reflecting decreased cortical activity, emerged within 1h after initiation of LFES. This suggests that bladder hyperreflexia did not result directly from RCI. This is plausible, considering the paucity of direct descending tracts from the frontal cortex to the spinal cord. In fact, most supraspinal centers involved in the control of bladder voiding are located in the medulla, pons, and hypothalamus, with only sparse transneuronal labeling seen in the frontal cortex (Nadelhaft et al., 1992; Vizzard et al., 1995; Marson, 1997; Sugaya et al., 1997). We hypothesize that, similarly to an actual stroke, in which development of somatic and autonomic hyperreflexia is slow and gradual, in this RCI model, bladder hyperreflexia developed due to changes in reticulospinal modulation induced by altered frontal cortical control of brainstem autonomic centers.
In summary, our animal model of LFES-induced RCI is highly reproducible and allows study of the mechanisms of spinal hyperexcitability and hyperreflexia following stroke or brain trauma. This model can also be used for development of treatment strategies that overcome this pathological condition.
This work was funded by National Institutes of Health grant R01-EB000518 to V.P. The authors express their appreciation to Yelena Smirnova for fabricating the surface cortical electrodes and spinal microelectrode arrays, and for her assistance during LFES-induced RCI experiments, Edna Smith for assistance with the surgical and perfusion procedures, and Clarence Graham and Jesus Chavez for their histological and immunohistochemical work.
No conflicting financial interests exist.