Anodal transcranial direct current stimulation with monopolar pulses improves limb use after stroke by enhancing inter‐hemispheric coherence

Post‐stroke neurological deficits, such as sensorimotor impairments, are often permanent and a leading cause of disability. Stroke is also associated with changes in neuronal synchrony among different brain areas. Multiple studies demonstrated that non‐invasive brain stimulation, such as transcranial direct current stimulation (tDCS), enhances the efficacy of existing rehabilitative therapies. We hypothesized that the therapeutic effects of tDCS could be due to its influence on neuronal synchrony. To study this, we recorded local field potentials in rats treated with anodal tDCS (a‐tDCS) after unilateral ischemic motor cortex lesion. To enhance the effect of a‐tDCS on neuronal synchrony, we added monopolar pulses (a‐tDCSmp) during a treatment. We found that ischemic lesions reduced interhemispheric coherence in the low gamma frequency range. By contrast, a‐tDCSmp treatment increased interhemispheric coherence along with motor improvement in a skilled reaching task. These observations indicate that increased neuronal coherence is a likely mechanism by which tDCS improves stroke recovery. Moreover, this work adds to previous evidence that measures of brain coherence could be used as a biomarker of stroke recovery, which may help in the design of more effective tDCS protocols for stroke rehabilitation.


INTRODUCTION
Stroke is characterized by reduced blood flow to the brain.It represents the leading cause of chronic adult disability and the third leading cause of death in North America (Benjamin et al., 2017).Although considerable spontaneous rearrangement of connections after stroke occurs (Murphy and Corbett, 2009;Lim et al., 2014), spontaneous recovery is limited and mainly mediated by compensational adjustments (Buckner et al., 1996;Metz et al., 2005;Alaverdashvili and Whishaw, 2010).Post-stroke neuroplasticity includes regeneration of damaged pathways, axonal and dendritic sprouting, and remapping of functional representations (Adkins et al., 2002;Winship and Murphy, 2009;Westlake and Nagarajan, 2011).Studies using resting state functional magnetic resonance imaging (fMRI) reported that post-stroke loss and recovery of sensorimotor function is associated with acute deterioration and subsequent retrieval of interhemispheric functional connectivity within the sensorimotor system (van Meer et al., 2010;2012;Westlake and Nagarajan, 2011).Similarly, large-scale phase synchrony in electroencephalography (EEG) was shown to correlate with post-stroke motor assessment (Kawano et al., 2017;2018).Clinical rehabilitative strategies take advantage of this information, for example by using bilateral arm training (Coupar et al., 2010;Choo et al., 2015), and improvements in stroke rehabilitation correlated with increases in inter-hemispheric coupling in EEG (Pellegrino et al., 2012).
Studies using non-invasive brain stimulation have shown promising results in enhancing effectiveness of stroke therapies.Transcranial direct current stimulation (tDCS) has been shown to improve recovery both in animal models and patients with ischemic brain injury (for review, see Floel and Cohen, 2010;Gomez Palacio Schjetnan et al., 2013;Gall et al., 2015;Lefebvre and Liew, 2017;Boonzaier et al., 2018).tDCS was also shown to affect cognitive and motor functions in intact animals/humans; however, its effect crucially depends on stimulation parameters (e.g., electrode polarity, stimulation site, current intensity) (Nitsche et al., 2007;Jacobson et al., 2012).This method is a noninvasive, low-cost and easy-to-use technique that can be applied to modify cerebral excitability.This is achieved by using weak direct currents to shift the resting potential of cortical neurons (Floel and Cohen, 2010).The direct currents are applied by attaching two electrodes (usually one anode and one cathode) to distinct areas of the scalp (Utz et al., 2010;Lefebvre and Liew, 2017).The degree of motor improvement for tDCS therapy depends on the interval between the infarct and the time of initiation of therapy, as well as, the type of stimulation (monopolar or bipolar), frequency of pulses, and duration of the stimulation (Reis and Fritsch, 2011).In addition to improving stroke outcomes, tDCS has been shown to promote a variety of motor and cognitive skills in healthy subjects (Ukueberuwa and Wassermann, 2010), demonstrating the broad applicability of tDCS for the enhancement of brain function (Fox, 2011;Fox et al., 2014).
Despite the demonstrated beneficial effects of tDCS, little is known about the underlying mechanisms.It has been proposed that tDCS provides a sub-threshold stimulus to modulate the likelihood that neurons will fire (Nitsche et al., 2003).The prolonged sensory, motor, and cognitive effects of tDCS have been attributed to a persistent bidirectional modification of post-synaptic connections, similar to long-term potentiation and long-term depression (Teskey et al., 2003;Fritsch et al., 2010;Koo et al., 2016).The N-methyl-D-aspartate (NMDA) antagonist dextromethorphan suppresses both anodal and cathodal tDCS effects, strongly suggesting an involvement of NMDA receptors in both types of DC-induced neuroplasticity (Liebetanz et al., 2002).Similarly, a study demonstrated that neuroplastic changes elicited by tDCS depend on brain-derived neurotrophic factor (BDNF)-TrkB interactions (Fritsch et al., 2010).Plastic modifications elicited by tDCS have also been attributed to the actions of glial cells (Gellner et al., 2016).In a previous study conducted by our group, we show that bilateral somatosensory stimulation enhances skilled movement quality and neural density in intact behaving rats (Faraji et al., 2013).Nev-ertheless, the underlying mechanisms and neuronal correlates mediating the benefit of tDCS require further exploration to serve as a viable technique in the clinical context.
Here, we investigated the interactions between cortical networks in rats with focal cerebral ischemia wherein the rehabilitation process was aided by anodal tDCS with added monopolar pulses (a-tDCSmp).Our findings show that animals with larger coherent activity between neuronal populations exhibit greater recovery from ischemia, as measured by skilled motor performance.In addition, we conducted intracortical local field potentials (LFP) recordings in anesthetized rats, which demonstrated that application of a-tDCSmp induced an increase in synchronization between hemispheres.All those results together suggest that increased interhemispheric coherence in the ~10-40Hz band represents a mechanism by which tDCS promotes sensorimotor recovery and enhances the efficacy of rehabilitation training after stroke.

METHODS
All procedures were performed in accordance with the guidelines set by the Canadian Council for Animal Care and approved by the University of Lethbridge animal welfare committee.

Skilled reaching task
Fifteen male Long-Evans hooded rats received from Charles River (Ontario, Canada), aged 70-80 days (400 g -500 g), were habituated to the experimenter for seven days.After habituation, all rats were trained in the skilled reaching task, which is a reach-to-eat task designed to quantify and qualify skilled forelimb use (Whishaw and Pellis, 1990;Metz et al., 2005).The skilled reaching apparatus was a rectangular Plexiglas box, 40 cm long × 45 cm high × 13 cm wide with a 1 cm slot at one end, and a shelf fixed to the outside of the box which was accessible through the slot (Fig. 3A).The shelf had two symmetric indentations which were 1.5 mm deep, 1.5 cm from the slot, parallel to each edge of the slot to provide stability and a consistent position for the food pellets to be placed.Small 45 mg food pellets were used (Bioserve, USA; Product # F0021).
Once trained to asymptote success levels with their preferred paw, rats received 20 pellets per day for five consecutive days in the reaching task to establish baseline reaching success (RS) values.In each trial, animals were required to walk to the rear end of the box prior to reaching for a pellet to readjust their body position (Metz and Whishaw, 2000).A successful reach was defined as an animal reaching for the pellet, grasping it and placing it in its mouth on the first attempt.RS for each rat and each session were calculated using the following formula: RS=number of successful reaches / 20 × 100%.The data were scored by group-blind observers from recorded videos.
Before inducing lesions and implanting stimulating electrodes (see next section), animals were assigned to experimental groups to match paw preference and RS between groups.Six days following surgery, animals were tested in skilled reaching for three weeks.After a four-week break, rats were re-tested for four consecutive days to evaluate the possible long-term effects of a-tDCSmp.

Photothrombosis and electrode implantation
Focal photothrombosis was induced in the forelimb area of primary motor cortex contralateral to the forelimb preferred in the reaching task (Metz et al., 2005).Animals were anesthetized using 4% isoflurane in a mixture of 1.5% oxygen and secured in a stereotaxic frame (David Kopf, Germany).The skull over the motor cortex was thinned using a fine dental burr in a rectangular shape from Bregma -1.0 to 4.0 anterior/posterior and Bregma 1.0 to 4.0 medial/lateral.A cold light source (Schott KL 1500, Germany) with an aperture of the same size and shape as the partial craniotomy was positioned over the skull.The skull was illuminated at maximum light settings for 20 min.During the first 2 min of illumination, Bengal Rose dye solution was injected through a tail vein (20 mg/kg, 10% solution in 0.9% saline).When the illumination period was completed, animals were implanted with stainless steel screws used for electrical stimulation (1 mm diameter; Small Parts, USA).Pilot holes were drilled in the skull bilaterally at coordinates: +4.5 A/P, ±1.0 M/L (Paxinos and Watson, 1998) (Fig. 1A).A dental acrylic skull cap was fashioned on the exposed skull, engulfing the electrodes, connecting wires, and connecting plug (Ginder Scientific, ON, Canada).Once the acrylic was hardened, animals were removed from the stereotaxic frame, placed on a heating pad, and monitored until they were awake.

Lesion analysis
DAPI (4',6-diamidino-2-phenylindole; Vector Labs Inc., CA) stained coronal sections (40 µm thickness), cut on a freezing microtome and mounted on microscope slides, were digitally scanned at 40x magnification (Nanozoomer, Hammatsu Photonics, Japan).The images were transferred to Image J software (NIH, USA) and the lesion volumes were quantified.Volumes were measured by tracing lesion borders, and then multiplying the lesion area by section thickness and number of sections in the series.Lesion area was determined by superimposing a straight-line connecting the lesion boundaries, as previously reported by our group (Metz et al., 2005;Knieling et al., 2009).

Transcranial direct current stimulation
The a-tDCSmp stimulation protocol consisted of 10 min of an anodal 65 µA direct current with superimposed 65 µA monopolar pulses of 30 ms duration, applied every 5 seconds during a-tDCS (Fig. 1C).The pulses were included in the protocol in an attempt to increase synchrony in interhemispheric activity.The stimulation was applied only during the first week of rehabilitation, which occurred concurrently with the skilled reaching task, 6 days after stroke and electrode implantation.The time point of stimulation was selected in order to: 1) allow a recovery period after surgical implantation as requested by the by the Canadian Council for Animal Care; 2) allow a time course of ischemic stroke to stabilize (Karl et al., 2010); and 3) evaluate the electrical stimulation effects in conjunction with motor rehabilitation as performed in many rehabilitation protocols that use this strategy (Alsharidah et al., 2018).Rats in all groups were connected to the stimulator during all reaching trials, but the stimulator was switched on only for the Stroke + Stim group.

Acute surgery and recordings
Detailed descriptions of the surgery and recording procedures have been published in (Luczak et al., 2007;2009;Schjetnan and Luczak, 2011;Luczak and Barthó, 2012).Briefly, after the last session of re-testing, the animals were anaesthetized with urethane (1.5 g/kg), injected i.p.Then, rats were placed in a stereotaxic frame, and two 3 mm 2 craniotomies were prepared over the forelimb somatosensory cortex of each hemisphere (AP: +2.7 to -0.3; ML: ±2.5 to 5) (Paxinos and Watson, 1998).One rat from the Stroke + Stim group died during the electrophysiological recordings, and data from this experiment were excluded from analyses.Extracellular signals were recorded with silicon probes (NeuroNexus Technologies, Ann Arbor, MI) consisting of eight shanks (Fig. 1A), and each shank was separated by 200 µm.The location of the recording sites was determined to be layer V by histological reconstruction of the electrode tracks and electrode depth.Local field potentials were referenced to a titanium screw implanted over the cerebellar cortex and sampled at 32 kHz and were band-pass filtered between 0.1 -8000 Hz (Neuralynx, Inc., Bozeman, Montana, USA).Ten min of spontaneous activity was recorded, followed by 20 min of electrical stimu-lation with the same parameters (65 µA direct current with an additional 30 ms duration and 65 µA pulses applied every 5 seconds), as in training (Fig. 1B).This was followed by 10 min of recording of spontaneous activity.Stimulation during the acute electrophysiological re- cordings was applied to all experimental groups and the duration increased (to 20 min) in order to have a direct assessment of the effects of stimulation in LFP coherence.Following acute electrophysiological recordings, animals were deeply anaesthetized with pentobarbital and perfused.Histological analysis on the extracted brains was performed using DAPI mounting medium.

Coherence analysis
Coherence was calculated as the magnitude squared coherence estimate using Welch's averaged, modified periodogram method (Rabiner and Gold, 1975) implemented in Matlab (MathWorks, Inc.; mscohere function).The magnitude squared coherence estimate is

Decrease local field potential (LFP) coherence after stroke
To investigate how interactions between sensory areas change after stroke and how this is modified by the application of electrical stimulation, three groups of rats were studied: Control (n=5), Stroke (n=5), and a group with stroke treated with direct current stimulation (Stroke+Stim, n=5).All rats were implanted bilaterally with stimulating electrodes anterior to the cortical motor areas and one reference electrode over the cerebellum (Fig. 1A), but stimulation was applied only in the Stroke+Stim group (Fig. 1B; Methods).During the same surgery, focal cortical ischemia was induced by photothrombosis in the forelimb area of motor cortex in the hemisphere contralateral to the preferred paw in the Stroke + Stim and Stroke groups.Ischemia was largely constrained to cortical areas (Fig. 1C).Average infarct volumes (V Stroke =5.42 mm³ ± 2.01; V Stroke+Stim =4.92 mm³ ± 1.80) did not statistically differ between groups (p=0.9, t-test).The stroke affected primarily the area corresponding to M1, which was verified with a histological examination.
Persistent effects of stroke and electrical stimulation on cortical activity were measured 8 weeks following the infarct.Animals were anesthetized with urethane and LFPs were recorded in the deep cortical layers.Two electrodes with 8 recording shanks each were inserted in sensory areas in both hemispheres (Fig. 1A).In all experimental groups, the LFP power spectra was highest around 0.5-2 Hz, with a systematic decrease in the higher frequencies (Fig. 2A).There were no significant differences between experimental groups (p contr-strok >0.2; p contr-strok+stim >0.2; p Stroke-Strok+Stim >0.1; Kolmogorov-Smirnov test), or between lesion and intact hemispheres (p>0.2;Kolmogorov-Smirnov test) at any frequency band between 0.5-40 Hz.
Despite the lack of differences in LFP power spectra, LFP coherence showed notable differences between groups (Fig. 2B) which were most pronounced in the low gamma frequency range (20-40Hz).Inter-hemispheric coherence was significantly lower in the Stroke group as compared to the Control and Stroke+Stim groups (p Con- trol-Strok =0.044; p Stroke+Stim-Strok =0.032; Kolmogorov-Smirnov test).This effect was consistent across different pairs of electrodes, although the values of coherence systematically changed with distance.For example, coherence between hemispheres was usually the highest between homologous areas and decreased with spatial offset between areas (Fig. 2C,D  group, the correlation was not significant, most likely due to insufficient coherence to detect this effect.Similarly, within-hemisphere coherence decreased significantly as a function of distance between electrodes (Fig. 2E; R Contr =-0.93; p<0.001;R Stroke+Stim =-0.92; p<0.001;R Stroke =-0.96; p<0.0001).There were no systematic differences between the intact and lesion hemispheres concerning the relation between coherence and distance.Altogether, these results suggest that focal ischemia reduces the coherence between cortical areas, and electrical stimulation restores coherence in the gamma band.

Coherence in low gamma band correlates with behavior
A skilled forelimb reaching task was used to provide a behavioral measure of motor recovery after stroke (Metz et al., 2005) (Fig. 3A).On the last day of testing (one day prior to electrophysiological recordings), RS was calculated for each animal (RS Control = 43.76+ 11.05, RS Stroke = 20.92+ 6.31, RS Stroke+Stim = 40.81+ 8.3).These results are consistent with previous studies showing decreased RS in rats after focal ischemia (Metz et al., 2005; Alaver- dashvili and Whishaw, 2010) and improvements due to anodal electrical stimulation (Adkins et al., 2008;Ukueberuwa and Wassermann, 2010;Faraji et al., 2013).RS and coherence in the gamma band were strongly correlated (Fig. 3B; R=0.78; p=0.001).Notably, even within individual experimental groups, correlations tended to be positive, suggesting a general positive relationship between coherence and skilled movements (R Contr =0.83; p=0.08;R Stroke+Stim =0.75; p=0.24;R Stroke =0.18; p=0.77).These findings suggest that low gamma band coherence may be important for competency in motor tasks.

Anodal stimulation increases LFP coherence
The data above shows that the Stroke+Stim group had higher gamma band coherence than the Stroke group.To investigate if enhanced gamma band coherence is causally linked to electrical stimulation, we examined the effect of electrical stimulation on brain activity.During the acute LFP recordings in anesthetized animals, we applied 20 min of anodal stimulation in the animals from all of our experimental groups.This was aimed to compare brain activity directly before and after electrical stimulation.The stimulation significantly increased LFP power in the 10-40 Hz band (p=0.016;t-test).Of note, for each animal, the power spectra calculated in 2 min window before stimulation was subtracted from the power spectra calculated in a 2 min window after stimulation.Next, the average difference in the 10-40 Hz band was calculated, and we tested the null hypothesis that the results come from a normal distribution with a mean equal to zero, meaning no change in power.This effect was consistent across all experimental groups (Fig. 4A).Moreover, LFP power increased similarly in the intact and lesion hemispheres (Fig. 4B; p=0.34; paired t-test between the intact and stroke hemispheres in the 10-40 Hz band; of note, results were consistent when instead using the non-parametric sign test).
Importantly, the observed increased LFP power after electrical stimulation was associated with increased coherence in the corresponding frequency band (Fig. 4C; p=0.038; t-test; 10-40 Hz).Again, this increase was consistent within the intact and lesion hemispheres and between hemispheres (Fig. 4C).This effect was similar across experimental groups (Fig. 4D), suggesting that anodal stimulation directly caused a general increase in coherence among cortical areas.

DISCUSSION
In this study, we performed electrophysiological recordings to investigate interactions between corti-cal networks in rats treated with a-tDCSmp after a focal ischemic motor cortex lesion.We observed a widespread decrease in low gamma band coherence between cortical areas following ischemia.We also showed that coherence in the low gamma band range was positively correlated with performance on a skilled motor task.Importantly, we demonstrated that anodal stimulation enhanced gamma band coherence between cortical areas, which may explain the reported therapeutic effect of tDCS on stroke recovery.Our results suggest that maximizing synchronization among cortical areas with electrical stimulation could facilitate neuroplastic mechanisms and translate to more effective stroke therapies.
Although very few studies investigated changes in low gamma band coherence after stroke, we hypothesize that its decrease relates to a general reduction in connectivity among brain areas following stroke.Consistent with our results are reports of a decrease in interhemispheric functional connectivity in a rat model of stroke (van Meer et al., 2010;2012).Crofts and Higham (2009), using data from diffusion magnetic resonance imaging, compared connectivity between 56 brain areas in stroke patients and healthy controls, and showed decreased communication among a number of brain regions in stroke patients.Similar results were obtained by Muehlschlegel and colleagues who, using near-infrared spectroscopy, showed that the interhemispheric correlation coefficient at frequency bands 0.15-3 Hz was smaller in stroke patients (Muehlschlegel et al., 2009).Analysis of functional connectivity with EEG also revealed reduced interhemispheric coherence in stroke patients (Strens et al., 2004;Gerloff et al., 2006).An interesting interpretation of this decline in interactions among brain areas after stroke was described by Wang et al. (Wang et al., 2010).Based on analysis of the topological configuration of the resting-state networks, Wang and coauthors found that brain networks shifted towards a more random and less optimized mode of function after stroke (Wang et al., 2010).Thus, our finding of decreased coherence in the low gamma band could be another manifestation of reduced efficiency of communication among brain networks after stroke.
Cortical gamma activity has been implicated in a variety of functions including attention, working memory, object recognition, feature binding, information processing, and sensorimotor performance (for review, see Herrmann et al., 2009).In general, 20-40 Hz activity is associated with top-down attentional processing by which the brain may be controlling the flow of sensorimotor information (Wróbel et al., 1994;Debener et al., 2003).Our results correspond well with this general framework.It should be noted that our electrophysiological data were recorded under anesthesia, which means that observed changes reflect long-lasting prop-erties of cortical circuitry rather than task-related changes in neuronal synchronization.Thus, increased 20-40 Hz activity observed in rats with better skilled movements could indicate that cortical networks in those animals may be able to synchronize more readily, which may contribute to better control over motor tasks.Such interpretation is supported by experiments demonstrating that higher baseline levels of gamma activity correspond to shorter response times (Gonzalez Andino et al., 2005;Womelsdorf et al., 2006;Schoffelen et al., 2011) and greater task performance efficiency (Kaiser et al., 2009;Joundi et al., 2012).Although the mechanisms of tDCS are not well understood, we hypothesize that a long-term increase in low gamma band coherence could be the result of an increase in strength of neuronal connections.It has also been shown that direct current stimulation facilitates LTP (Kronberg et al., 2017).Thus, application of a-tDCSmp concurrently with a reaching task may promote a strengthening of connections between active motor cortex neurons.Previous studies have shown that, following stroke, reduced gamma band coherence relates to a general reduction in connectivity among brain areas (van Meer et al., 2010;Crofts and Higham, 2009;Muehlschlegel et al., 2009).Therefore, tDCS facilitation of LTP may strengthen motor cortex connectivity and may thus lead to a long-lasting increase in gamma coherence.
In this study, we show that anodal stimulation enhances gamma activity and coherence.This result is of interest because the electrophysiological effect of direct current stimulation on neuronal activity in vivo is poorly understood.Reato et al. previously reported that hippocampal gamma activity can be enhanced with weak electrical currents (Reato et al., 2014).Here, we report that direct current stimulation increases gamma power immediately after stimulation and has long-lasting modifications in gamma coherence.Interestingly, the increase in power above 10 Hz also resembles characteristics of an activated state (Luczak et al., 2013), which is known to enhance neuronal plasticity (Bermudez Contreras et al., 2013) and increase information transfer rate by generating more neuronal packets (Luczak et al., 2015).For example, during the attentive state, brain activity decreases in the low frequencies (<10 Hz) with a simultaneous increase in gamma range, which characterizes the activated state and is also linked to attention (Harris and Thiele, 2011).Thus, one might argue that increasing gamma activity with anodal stimulation may emulate the effect of attention.This idea is consistent with other studies in which the application of anodal tDCS had a striking resemblance to the effects of increased attention (Gladwin et al., 2012;Roe et al., 2016).For example, anodal tDCS applied over the prefrontal cortex can improve working mem-ory and visual attention (Javadi and Walsh, 2011;Roy et al., 2015), and tDCS applied over the temporal lobe has been shown to facilitate problem solving (Chi and Snyder, 2011).Further, motor performance is improved when tDCS is applied over the motor cortex (Reis and Fritsch, 2011;Tanaka et al., 2011).Although the proposed idea linking the effect of tDCS to enhanced attention is speculative, we believe that it is important to state it here because it could lead to better conceptual understanding of tDCS mechanisms and possibly to better treatment strategies.It is also worth noting that stroke patients show attention deficits (Barker-Collo et al., 2011).Thus, one may argue that the therapeutic effects of tDCS when combined with physical rehabilitation may work by helping motor networks to "pay attention" to an exercise, thereby enhancing its efficacy.
From a translational research perspective, our data suggest that identifying the parameters of electrical stimulation that maximize 20-40 Hz coherence could lead to more effective stroke therapies.In this study, we used constant anodal stimulation at 65 µA with additional 30 ms long 65 µA pulses every 5 seconds, applied over both hemispheres.This stimulus was chosen because we hypothesized that anodal currents will excite motor cortices and additional short pulses may assist synchronization between the hemispheres.However, other types of stimuli with more translational implications and adherence to recent safety guidelines could be designed to activate and synchronize cortical networks more effectively, potentially leading to better stroke therapy and assure tDCS application (Gomez Palacio Schjetnan et al., 2013;Aparício et al., 2016;Bikson et al., 2016;Jackson et al., 2017).Studies in humans using EEG and fMRI suggest an increase in neural synchronization and functional connectivity induced by tDCS (Suppa and Berardelli, 2011;Sehm et al., 2013;Roy et al., 2014).A recent clinical study measuring resting state functional connectivity supports the present findings of increased connectivity after stroke rehabilitation in relevant motor control areas (Lefebvre et al., 2017).
In summary, our findings add to evidence that improved coherent activity between neuronal populations facilitates recovery after ischemic lesion.Importantly, our study suggests that a-tDCS mediates sensorimotor recovery by increasing the synchronization between cortical networks.This research was supported by Alberta Innovates-Health Solutions (GM, AL), the Canadian Institutes of Health Research (GM), and the Natural Sciences and Engineering Research Council of Canada (GM, AL).

Fig. 1 .
Fig. 1.Experimental procedures.(A) Timeline of experimental stages.(B) Illustration of the position of the stimulation electrodes (black dots, anode '+', cathode '-'), location of the photothrombotic lesion (dark rectangle), areas with inserted silicon probes (boxes with dots representing electrode locations), and schematic of the silicon probe with 8 shanks (right side).(C) Schematic illustration of stimulation parameters used for a-tDCSmp.(D) Representative coronal sections (40 µm) of the brain of an animal with ischemic stroke.Localization of the stroke area (gray blobs) corresponds to the primary motor cortex (M1).Top section shows silicon probe traces in the deeper layers of the limb somatosensory cortex in both hemispheres.AP=Anteroposterior coordinates(Paxinos and Watson, 1998).

Fig. 2 .
Fig. 2. Changes in LFP after stroke and electrical stimulation.(A) Power spectra in sensory cortex in the Control group (light gray), the Stroke group (dark gray), and Stroke+Stim group (black).Dashed lines show standard error.For each animal, data were averaged over both hemispheres.(B) Coherence between hemispheres for the same experimental groups that are shown in panel A. Error bars denote standard error.Note: higher coherence in the 20-40 Hz band for Control and Stroke + Stim groups.(C) Illustration of the definition of offset between homologous areas between hemispheres.Blue circles represent recording sites.(D) Coherence as a function of distance between homologous areas.Each point represents an average coherence between 20-40 Hz across animals, for each pair of shanks.Symbols represent experimental groups: cross=Stroke + Stim; triangle=Stroke group; circles=Control group.Lines show linear fit for each group.For Control and Stroke+Stim groups, the coherence decreased with offset between areas.(E) Coherence as a function of distance between shanks between hemisphere.Each cross represents an average coherence between 20-40 Hz across animals, for each pair of shanks.Symbols represent the same experimental groups that are shown in panel D. Lines show linear fit for each group.For all groups, the coherence decreased with distance between recording shanks.LFP=Local field potential.

Fig. 3 .
Fig. 3. (A) Picture of a rat performing a skilled reaching movement.(B) Relationship between coherence in the low gamma range and reaching success (circles=Control group; triangles=Stroke group; crosses=Stroke + Stim group).Dashed line represents the least-squares fit for the points combined from all groups.

Fig. 4 .
Fig. 4. Changes in brain activity after electrical stimulation.(A) Difference in LFP power spectra (subtracting power in the periods immediately before stimulation from power in periods immediately after stimulation).The dashed trace shows average and standard error (shaded area), for all animals.Note an increase in power after stimulation in the 10-40 Hz band.Light gray, dark gray, and black lines show the average for Control, Stroke, and Stroke + Stim groups, respectively.(B) A change in LFP power spectra after electrical stimulation in intact (black) and stroke (gray) hemispheres.(C) Increase in LFP coherence after electrical stimulation in the 10-40 Hz band.The dashed line shows average change in coherence across all recording sites for all animals, and the shaded area shows standard error.Light gray, dark gray, and black lines represent an average change in coherence between hemispheres, within the stroke hemisphere, and within the intact hemisphere, respectively.(D) Change in LFP coherence in the stroke hemisphere (dark gray=Stroke group; black=Stroke+Stim), and in control group in the left hemisphere (light gray).Differences between groups were not significant in the 10-40 Hz band (p>0.8 for all between groups comparisons; t-test).LFP=Local field potential.