Safety of ipsilesional anodal transcranial direct current stimulation in acute photothrombotic stroke: implications for early neurorehabilitation

Early rehabilitation in the acute phase of stroke, that bears unique neuroplastic properties, is the current standard to reduce disability. Anodal transcranial direct current stimulation can augment neurorehabilitation in chronic stroke. Studies in the acute phase are sparse and held back by inconclusive preclinical data pointing towards potential negative interaction of the excitability increasing tDCS modality with stroke-induced glutamate toxicity. In this present study, we aimed to evaluate structural and behavioral safety of anodal tDCS applied in the acute phase of stroke. Photothrombotic stroke including the right primary motor cortex was induced in rats. 24 h after stroke anodal tDCS was applied for 20 min ipsilesionally at one of four different current densities in freely moving animals. Effects on the infarct volume and on stroke induced neuroinflammation were assessed. Behavioral consequences were monitored. Infarct volume and the modified Neurological Severity Score were not affected by anodal tDCS. Pasta handling, a more sensitive task for sensorimotor deficits, and microglia reactivity indicated potentially harmful effects at the highest tDCS current density tested (47.8 A/m2), which is more than 60 times higher than intensities commonly used in humans. Compared to published safety limits of anodal tDCS in healthy rats, recent stroke does not increase the sensitivity of the brain to anodal tDCS, as assessed by lesion size and neuroinflammatory response. Behavioral deficits only occurred at the highest intensity, which was associated with increased neuroinflammation. When safety limits of commonly used clinical tDCS are met, augmentation of early neurorehabilitation after stroke by anodal tDCS appears to be feasible.


Safety of ipsilesional anodal transcranial direct current stimulation in acute photothrombotic stroke: implications for early neurorehabilitation
Brita Fritsch 1 , Marleen Mayer 1 , Janine Reis 1 & Anne-Kathrin Gellner 1,2* Early rehabilitation in the acute phase of stroke, that bears unique neuroplastic properties, is the current standard to reduce disability.Anodal transcranial direct current stimulation can augment neurorehabilitation in chronic stroke.Studies in the acute phase are sparse and held back by inconclusive preclinical data pointing towards potential negative interaction of the excitability increasing tDCS modality with stroke-induced glutamate toxicity.In this present study, we aimed to evaluate structural and behavioral safety of anodal tDCS applied in the acute phase of stroke.Photothrombotic stroke including the right primary motor cortex was induced in rats.24 h after stroke anodal tDCS was applied for 20 min ipsilesionally at one of four different current densities in freely moving animals.Effects on the infarct volume and on stroke induced neuroinflammation were assessed.Behavioral consequences were monitored.Infarct volume and the modified Neurological Severity Score were not affected by anodal tDCS.Pasta handling, a more sensitive task for sensorimotor deficits, and microglia reactivity indicated potentially harmful effects at the highest tDCS current density tested (47.8 A/m 2 ), which is more than 60 times higher than intensities commonly used in humans.Compared to published safety limits of anodal tDCS in healthy rats, recent stroke does not increase the sensitivity of the brain to anodal tDCS, as assessed by lesion size and neuroinflammatory response.Behavioral deficits only occurred at the highest intensity, which was associated with increased neuroinflammation.When safety limits of commonly used clinical tDCS are met, augmentation of early neurorehabilitation after stroke by anodal tDCS appears to be feasible.
Stroke is the worldwide leading cause of lasting motor impairment.The magnitude of spontaneous functional recovery is independent of external factors in the early days to weeks [1][2][3][4] .Next to spontaneous diminution of detrimental milieu factors in the stroke area (e.g., edema, inflammation) the acute brain injury opens a window for enhanced neuroplasticity.Enhanced reorganization on the network level is observable, similar to neurodevelopmental levels of plasticity [5][6][7] .Accordingly, an early start of rehabilitative treatment already in the acute setting of the stroke unit is an established procedure in stroke neurorehabilitation 8 .The definition of this early post-stroke period has not been clearly defined and varies between studies and guidelines, but is often agreed to include the first 7-14 days [9][10][11] .However, since high intensity rehabilitative activity increased infarct volume in animal studies [12][13][14] and in some cases even worsened functional outcomes 14 it is accepted that intervention intensity has to be adjusted to moderate levels.Increased infarct volume has been attributed to higher stroke related glutamate toxicity 12,13 .
Transcranial direct current stimulation (tDCS) modulates cortical excitability and synaptic plasticity across species in a polarity-dependent manner [15][16][17] .In healthy subjects and chronic stroke patients anodal tDCS www.nature.com/scientificreports/promotes neuroplastic processes such as motor learning, enhanced upper limb function, and increased gray matter volume [18][19][20][21][22] .The underlying cellular effects contributing to behavioral improvements are well understood and include enhanced BDNF-signaling and improvement of synaptic transmission [23][24][25][26] .Thus, anodal tDCS application in conjunction with early neurorehabilitation could maximize utilization of the early plastic window after stroke, especially in patients with limited physical abilities.On the other hand, application of anodal tDCS in the acute phase may negatively interact with the stroke related cerebral environment and lead to delayed neurodegeneration, e.g., by glutamate excitotoxicity, as was shown for high intensity rehabilitative activity.Current animal studies of acute ipsilesional anodal tDCS after stroke were performed under anesthesia [27][28][29][30] .Anesthesia suppresses glutamate release 31,32 and also tDCS induced cortex-wide Ca 2+ -elevations 33 and detrimental effects of high intensity tDCS such as neuroinflammation and neurodegeneration 34 .Thus, anesthesia likely masks harmful effects of acute tDCS in stroke.Moreover, with the exception of one study 27 , stimulation intensities in rodent studies were manifold above those used in clinical application.Nevertheless, the insulating structures (e.g., bone) are much thinner in rodents compared to the human skull and thus rodent studies are likely overestimating harmful effects of the stimulation.While there is also some evidence from modeling studies on how the parameters could possibly relate 35 , the optimal stimulation parameters per species may differ.Under these technical conditions with limited transferability to the human condition, only one study demonstrated an increase in infarct volume through anodal tDCS 29 while in the other studies infarct volume was not affected by tDCS 27,28,30 .
In the present study, using a rat stroke model with consistent infarct location and size (via photothrombosis) we evaluate safety aspects of acute ipsilesional anodal tDCS in a dose-response design.One day after stroke induction anodal tDCS was applied to non-anesthetized rats at one of four different stimulation intensities, including those leading to slight neuroinflammation in the non-stroke rat brain 34 .Infarct volume and microglial reactivity were investigated, as well as effects on the behavioral level measured by the modified Neurological Severity Score and the pasta handling task.We hypothesize that low intensity ipsilesional anodal tDCS can be safely applied early post stroke in rats.

Animals
Adult male Sprague Dawley rats (n = 29, age 8.5 ± 0.5 weeks, 309-400 g; Charles River, Sulzfeld, Germany), were group-housed in standard cages on a 12/12 h light/dark cycle at constant room temperature.Animals had access to water and food ad libitum, except 24 h prior to testing, when the food was weight-adjusted (10 g/100 g body weight).All animal studies were performed according to the Animal Protection Law and Directive 2010/63/EU of the European Commission.Animal protocols were approved by the Commission for Animal Experimentation of the Regional Council of Freiburg and the Commission for Animal Experimentation of the University Medical Center Freiburg.All animal experiments have been reported in compliance with the ARRIVE guidelines.
After receiving stroke surgery, rats were assigned to the four different treatment groups (sham tDCS = 0.0 A/m 2 , anodal tDCS with either 0.8 A/m 2 , 31.8 A/m 2 or 47.8 A/m 2 ) by a stratified step-by-step randomization performed by the lead investigator to maintain comparable post-stroke modified neurological severity scores per treatment arm.Stimulation intensities were based on our previous study, which found the threshold for neurodegeneration to be at or below 31.8A/m 2 in anesthetized or awake rats 34 .The experimental timeline is depicted in Fig. 1A.Recruitment proceeded until each group included at least 5 rats, extrapolated as an appropriate number of animals for analysis based on the effect sizes in a similar dose-response experiment in healthy rats 34 .The investigator performing the behavioral testing as well as data analysis was blinded for the type of tDCS applied.

Stroke induction
Ischemic stroke was induced via photothrombosis as described elsewhere 36 .In brief, under isoflurane anesthesia (4% induction, 1.5-2% maintenance), Bengal rose (Sigma Aldrich, St. Louis, MO, USA; 9.975 mg/kg; dissolved in isotonic saline) was injected into the lateral tail vein of the stereotaxically fixed rat.With the beginning of the injection (slowly applied over 2 min) a circular area targeting a cortical region including the left primary motor cortex (5 mm diameter: center relative to Bregma: anterior-posterior = + 1.5 mm, medial-lateral = 3.5 mm left, Fig. 1B) was illuminated with a cold light source (KL 1500 LCD 150 W, SCHOTT AG, Mainz, Germany) through the exposed skull and illumination continued for a total of 20 min.

Transcranial DCS
Immediately following the cold light illumination, the DCS electrode was placed during the same surgery session.A plastic cylinder was fixed with acrylic cement to two screws on the skull (inner diameter: 4 mm, center relative to Bregma: anterior-posterior = − 0.5 mm, medial-lateral = 3.5 mm left, Fig. 1B).Photothrombosis and tDCS areas had only a partial overlap, resulting in brain regions underlying the DCS electrode only (most caudal; region D in Fig. 3B), underlying both the DCS electrode and the photothrombosis (mid-section; region E in Fig. 3B), and most rostral underlying the photothrombosis only (region F in Fig. 3B).
In a separate session, tDCS was performed 24 h after stroke induction in the awake, freely moving animals.After filling the cylinder with isotonic saline solution, the tDCS electrode (Ag/Ag-Cl, contact area of 12.56 mm 2 ; World Precision Instruments, Sarasota, FL, USA and in-house production) was screwed in.A rubber counter electrode (6 cm 2 ) was covered with conductive gel and attached to the animal's chest with a vest.
Stimulation parameters were set to current densities of either 0.0, 0.8, 31.8 or 47.8 A/m 2 , and current was turned on for 20 min using a 9 V battery-driven stimulator with a one second ramp-up and -down function (inhouse production, Technical Workshop, Department of Neurology, University of Freiburg).During stimulation, the behavior of the animal and the current flow were constantly monitored.

Familiarization
Rats were familiarized to all procedures performed during alertness.This included the tDCS set-up (wearing the vest containing the chest electrode, cage), traversing a round beam (length 1 m, diameter 2.5 cm) to reach a dark box included in the modified Neurological Severity Score (mNSS) and handling the pasta after a brief 24 h weight-adjusted food restricted period (10 g/100 g body weight).

Modified neurological severity score
Neurological deficits after stroke were assessed by the mNSS with a potential maximum deficit score of 18 37 .Scores were assessed at baseline (day -1), post-stroke (day 1) and post-DCS (day 2, timeline see Fig. 1A).In brief, sensorimotor testing included placing of the affected forepaw on a surface after the vibrissae were touching it www.nature.com/scientificreports/(placing test, 0-1 point) and contraction of the forepaw while pushing it to a surface (proprioceptive test, 0-1 point).Three repetitions of the beam balance test were averaged (steady balancing to falling off, 0-6 points).For further motor testing, walking (normal walk to falling to the paretic side, 0-3 points) and the reaction to raising by the tail (1 point each for flexion of fore-or hindlimb within 60 s or turning of the head > 10° to vertical axis within 30 s), were observed.Missing corneal, pinna or startle reflexes or any seizure-like abnormalities were scored with 1 point each.

Pasta handling task
Fine motor performance was assessed in rats handling and eating a piece of pasta (length 7 cm, diameter 1.5 mm 38,39 ).Rats were placed in the experimental cage.One pasta piece was handed at a time.Testing concluded after a maximum of five pasta pieces, or after 4 valid trials.This resulted in 1-4 valid trials per rat for the final analysis.Single performances per animal were averaged.Reasons for invalid performances were no performance at all or pasta breaks, which alter the difficulty of the task 39 .Performance was video-taped for offline analysis (60 frames/s).For frame-by-frame analysis Avidemux software (Version 2.6.12,http:// fixou net.free.fr/ avide mux/) was used.General performance was assessed by the time needed to eat each piece of pasta.Asymmetries in limb use were assessed by counts of ipsi-and contralateral grasps per pasta (in relation to the stroke hemisphere).
In addition, each atypical behavior was counted as either present (1 point) or non-present (0 points).Atypical behaviors: hunched posture, angling with head tilt, paws together when long, paws apart when short, guide and grasp switch, mouth pulling.The category failure to contact was divided into 3 separate categories:-affected paw supports, no grip,-no contact with affected paw during eating (contact during adjustment excluded),-no contact with affected paw at any time.Because of the inability of most animals to perform the task 1 day after stroke induction, poststroke-data were not analyzed.

Tissue preparation
24 h after tDCS, rats were deeply anesthetized with ketamine (80 mg/kg body weight) and xylazine (12 mg/ kg body weight) intraperitoneally and, following a negative toe pinch reflex, perfused transcardially with cold phosphate-buffered saline pH 7.4 (PBS) followed by 4% paraformaldehyde dissolved in PBS (PFA).The brains were harvested, postfixed in 4% PFA at 4 °C for 24 h and cryoprotected at 4 °C in 30% sucrose for 48 h.Afterwards, the brains were cut into coronal sections of 30 μm using a freezing microtome.
Stroke volume was measured by truncated cone calculation of tagged FJ-C-stained areas (Fig. 1C), factoring in the sagittal distance between each brain section 42 .The maximum caudal stroke extension towards the region underlying the DCS electrode was assessed in relation to Bregma.
Perilesional cortical intensity of anti-Iba1-staining as a marker for microglial reactivity was measured by the mean gray values of 200 × 200 µm tiles (Matlab R2012b-Software, The MathWorks, Inc., Natick, MA, USA) and compared to the corresponding area of the contralesional hemisphere.The stroke area and artifacts were manually excluded from the analysis (Fig. 3A).The occurrence of high microglia reactivity was classified as 1.5 SD above the mean gray value of the non-lesioned hemisphere and utilized for analysis.

Statistical analysis
All statistical analyses were performed using GraphPad Prism version 9.1.1(GraphPad Software, San Diego, California USA).Normal distribution was tested by Shapiro-Wilk's test.Data were either compared using a oneway ANOVA to describe group effects (Figs.1D, 2B,F, 3C-F) or 2-way repeated measures ANOVA (Fig. 2A,C-E) www.nature.com/scientificreports/ to describe group and time effects as well as their interaction, followed by post-hoc Tukey's test corrected for multiple comparisons.Not normally distributed data were compared using the Wilcoxon test (2 groups) or the Kruskal-Wallis test (> 2 groups) with Dunn's corrected post-hoc test.For non-parametric repeated measures analysis Friedman test was used.For all statistical comparisons, the level of significance was set to p < 0.05.All data are presented as mean ± standard error of mean (SEM), additional individual data are shown as scatter plots where appropriate.

Results
Three rats were excluded from the experiment: one animal had a lack of stroke, one showed aberrant stroke location, and one could not be sufficiently familiarized with the procedures due to fearful behavior, resulting in final group sizes of 0.0 A/m 2 n = 6, 0.8 A/m 2 n = 5, 31.8A/m 2 n = 8, 47.8 A/m 2 n = 7. www.nature.com/scientificreports/

Effect of tDCS on post-stroke microglia activation
Across all analyzed brain regions (Region D + E + F, Fig. 3A,B) high microglial reactivity was similarly expressed 24 h after 20 min of the lower tDCS intensities 0.0, 0.8 and 31.8A/m 2 tDCS, that was 48 h post-stroke (0.14 ± 0.04, 0.11 ± 0.05, and 0.08 ± 0.38, respectively; Fig. 3C).The highest tDCS intensity, 47.8 A/m 2 , resulted in a significant increase of high Iba1-reactivity (0.34 ± 0.05; H(3) = 20.29,p = 0.0001; p = 0.0554 vs. 0.0 A/m 2 , p = 0.0116 vs. 0.8 A/m 2 , p < 0.0001 vs. 31.8A/m 2 ; see Supplementary Fig. S1 for representative images of low and high microglial reactivity).Assuming spatial specificity of the tDCS effect on the underlying brain, stimulation and stroke areas were chosen so that there was only a partial overlap.This led to brain areas underlying the tDCS electrode caudal to the stroked brain region, underlying the tDCS electrode and being affected by stroke, and being affected by stroke while not underlying the tDCS electrode (Fig. 3B).As can be seen in Fig. 3D-F, the proinflammatory effect of high-dose 47.8 A/m 2 tDCS occurred in all three brain regions.In region F a significant group difference was found (H(3) = 8.93, p = 0.030, p = 0.14 vs. 0.0 A/m 2 , p = 0.19 vs. 0.8 A/m 2 , p = 0.0442 vs. 31.8A/m 2 ; Fig. 3F).

Discussion
Anodal tDCS applied at low stimulation intensity to the ipsilesional forelimb motor cortex 24 h after photothrombotic stroke in awake rats, is safe as assessed by stroke lesion size, microglial reactivity, neurological deficits, and sensorimotor performance.This is the first study investigating the safety limits of acute anodal tDCS in a Vol:.(1234567890 Previous studies in rodent models were inconclusive with regard to the influence of anodal tDCS in the acute phase after stroke on stroke lesion size in rodents [27][28][29][30] .A single study reporting an increase in infarct size in mice 29 used a much earlier tDCS application (during/within 4.5 h of stroke induction) compared to the others, which had at least a 24 h delay after stroke and were performed in rats.In line with the latter studies, low to high intensity anodal tDCS applied with a delay of 24 h after stroke induction in our present study did not affect stroke volume in awake rats.Partial placement of the tDCS electrode behind the photothrombotic area allowed us to additionally assess whether tDCS influences the caudal infarct border, which might indicate harmful effects that might be otherwise masked in the overall stroke volume.In anesthetized healthy rats, neurodegeneration was detectable already at 47.8 A/m 2 (> 60 times higher than in the human application), with the extent even pronounced in the absence of anesthesia 34 .However, since in the present stroke study additional neurodegeneration to the stroke volume was not detectable at the threshold intensity in non-stroke rats 34 , it is conceivable that there is no negative interaction between stroke and tDCS.The apparently lower lesion threshold in non-stroke rats in the previous study is likely explained by the high sensitivity of the FJ-C-stain in the healthy brain, where even a single degenerating neuron can be identified, and the threshold can be set with very high sensitivity.In fact, at 47.8 A/m 2 only sparse isolated degenerating neurons were found in healthy rats.Such a finding could likely be obscured within the stroke induced neurodegeneration.We did not test higher intensities in stroke rats to find the lesion threshold under this condition as we intended to perform the stimulation in wakefulness.In our hands 47.8 A/m 2 is the highest intensity well tolerated, judged by behavior (head scratching).
On the behavioral level, the rats' neurological status improved relative to the first assessment post-stroke, regardless of the tDCS intervention.This finding is in line with previous work observing no aggravation of post-ischemic neurological or behavioral deficits after anodal tDCS to the stroke hemisphere in anesthetized rats [27][28][29] .Interestingly, despite the increased infarct volume after hyperacute ipsilesional high intensity (55.5 A/ m 2 ) anodal tDCS in mice, there was no behavioral correlate as measured by the mNSS 29 .As the mNSS covers several neurological functions (motor, sensory, reflexes and balance), there is a possibility that specific deficits in one function may be masked.Here, we used the pasta handling task to additionally assess manual skillfulness and in particular motor and sensory dysfunction with a task more sensitive to reveal even subtle deficits.Indeed, a group difference in atypical behaviors with more frequent occurrence after 47.8 A/m 2 compared to 0.0 and 0.8 A/m 2 was present.While this finding was not reflected in a time x stimulation interaction, it is one hint towards a negative effect of anodal tDCS on behavioral function at this particular stimulation intensity.Of note, drop-outs due to "no performance" only occurred at the two highest tDCS intensities.The lack of harmful effects of lower intensity ipsilesional anodal tDCS in our study supports two small monocentric clinical trials in acute stroke patients that did not find negative effects on behavioral outcome measures (structural assessment was not performed) 43,44 .
Neuroinflammation, indicated by increased microglia reactivity, occurred with high intensity anodal tDCS in healthy rats 34,45,46 .With increasing stimulation intensity microglia activation preceded neurodegeneration and might thus be a more sensitive marker than neurodegeneration to detect harmful tDCS effects 34 .Additionally, it is well known that microglia are activated in response to cerebral ischemia, which might be mediated by increased glutamate or cytokine release from affected neurons.As expected, based on our previous results in healthy rats, high intensity tDCS applied at 47.8 A/m 2 resulted in higher stroke related microglia activation in contrast to sham or lower intensity tDCS at 0.8 and 31.8A/m 2 .This effect was homogenously distributed across the differently exposed brain regions (see Fig. 3B,D-F).Since the intensity threshold for microglia activation was not reduced compared to the healthy rat exposed to the same stimulation procedure 34 , one can argue that the susceptibility of the rat brain to anodal tDCS-induced neuroinflammation is not higher in the early days after stroke.In accordance, high intensity anodal tDCS failed to induce potentiation of ischemia related microglia activation in anesthetized rats 29,30 .On the other hand, low intensity anodal tDCS may even protect from postischemic axonal degeneration in the internal capsule 27 .
While we have not assessed the effects of cathodal tDCS in our study, it is important to note that in preclinical studies cathodal tDCS seems to have neuroprotective effects after stroke 29,[47][48][49] , most notable even when applied in the hyperacute phase 29,47,48 , a finding that has been ascribed to its excitability-reducing effects and a consecutive attenuation of stroke-induced glutamate toxicity 29,47 .A recent clinical pilot study demonstrated safety and feasibility of cathodal tDCS in the hyperacute phase of stroke.Only a trend was found towards reduced infarct growths in the first 24 h, paving the grounds for future trials 50 .Additional data from a clinical trial regarding the safety, tolerability, feasibility, and potential efficacy of cathodal tDCS to salvage the penumbra in acute large vessel occlusion (application before and after endovascular therapy) are expected in 2024 (TESSERACT-BA, ClinicalTrials.govNCT04061577).However, since we and others have not found augmentation of neuroplastic processes both in human and rodent conditions by cathodal tDCS 18,23,24,[51][52][53][54] , the current work focuses on safety of anodal tDCS which may be combined with interventions augmenting learning and re-learning after stroke in early neurorehabilitation.

Implications for the application in the clinical setting
Direct translation of stimulation parameters derived from animal models to the human setting is certainly limited due to obvious reasons.Modelling data may give an idea of a scaling factor to compare current densities between species 35 .However, ipsilesional anodal tDCS in the stroke rat model appears to be safe as long as it is not applied during the hyperacute phase after stroke manifestation (within hours of stroke induction) and below a certain harmful threshold intensity.24 h post-stroke this threshold intensity does not differ from that obtained in the healthy rat brain.Thus, we were able to exclude sensitization of the rat brain to increasing doses of tDCS by a recent stroke.Accordingly, the commonly used clinical tDCS stimulation parameters considered safe in healthy subjects are likely safe in the acute phase after stroke in the early neurorehabilitation setting.

Figure 3 .
Figure 3.Effect of tDCS 24 h after photothrombotic stroke on microglia activation (A) Exemplary depiction of the 200 × 200 µm tiles utilized for analysis of high Iba1-reactivity, taken from region F in (B).Note that the non-lesioned hemisphere (left side on the figure) defined the corresponding tiles on the lesion side (right).(B) Schematic drawing of a rat brain in the sagittal plane with projection of the region underlying tDCS only (D; light grey), tDCS and photothrombosis (E; dark grey), and photothrombosis only (F; medium grey), and the location of the analyzed slices within the region (dotted lines).(C) The ratio of high Iba1-reactivity tiles across all regions was highest with 47.8 A/m 2 tDCS and comparable between the other stimulation groups (ratio of high reactivity tiles per hemisphere, contralesional subtracted from ipsilesional).(D-F) Separation of the data corresponding to the regions depicted in (B) reflects this pattern for 47.8 A/m 2 in each region, particularly in region F (photothrombosis only) vs. 31.8A/m 2 .0.0 A/m 2 (n = 6), 0.8 A/m 2 (n = 5), 31.8A/m 2 (n = 8), 47.8 A/ m 2 (n = 7).tDCS transcranial direct current stimulation.Data represent mean ± SEM, scatter in C = individual animal data points sampled across the 3 regions.t trend, *p < 0.05, ****p < 0.0001.
) -dependent manner and in freely moving rodents with a stroke.Indicators for dose-dependent harmful effects of anodal tDCS were found at stimulation intensities clearly exceeding those used in clinical application (i.e., typically 0.4-0.8A/m 2 ). dose