A novel ex vivo assay to define charge-balanced electrical stimulation parameters for neural precursor cell activation in vivo

Endogenous neural stem cells and their progeny (together termed neural precursor cells (NPCs)) are promising candidates to facilitate neuroregeneration. Charge-balanced biphasic monopolar stimulation (BPMP) is a clinically relevant approach that can activate NPCs both in vitro and in vivo . Herein, we established a novel ex vivo stimulation system to optimize the efficacy of BPMP electric field (EF) application in activating endogenous NPCs. Using the ex vivo system, we discerned that cathodal amplitude of 200 μ A resulted in the greatest NPC pool expansion and enhanced cathodal migration. Application of the same stimulation parameters in vivo resulted in the same NPC activation in the mouse brain. The design and implementation of the novel ex vivo model bridges the gap between in vitro and in vivo systems, enabling a moderate throughput stimulation system to explore and optimize EF parameters that can be applied to clinically relevant brain injury/disease models.


Introduction
Neurological disorders, including stroke and neurodegenerative diseases, are the leading cause of disability and the second leading cause of death globally (Avan and Hachinski, 2021). Neural stem cell (NSC)based therapies offer a promising strategy to replace neural cells lost in injuries or diseases (Kim and de Vellis, 2009). NSCs are defined by their cardinal properties of self-renewal and generation of progeny that can differentiate into mature neural phenotypes that comprise the central nervous system (CNS) (Morshead et al., 1998;Temple, 2001;Emsley et al., 2005). Within the brain, neurogenesis persists into adulthood from rare populations of NPCs that reside in well-defined neurogenic niches. The largest pool of neurogenic NPCs in the post-natal brain are spatially restricted to the subventricular zone (SVZ) lining the lateral ventricles (LV) of the forebrain (Ming and Song, 2011), where they are primarily in a quiescent state (Hamilton et al., 2009;Adams and Morshead, 2018). Injury induced activation of endogenous NPCs has proven to be insufficient for functional recovery (Moskowitz et al., 2010), however, the versatility of endogenous NSCs and their progeny, collectively referred to as neural precursor cells (NPCs), make them appealing targets for regenerative medicine strategies aimed at neurorepair.
Various auxiliary strategies have been investigated extensively to harness the potential of NPCs for neuroregeneration to aid in overcoming obstacles of low survival rate, limited mobilization from niches, and failure to functionally integrate into damaged regions (Thored et al., 2007;Yu et al., 2011;Iwasa et al., 2020). Approaches include administration of pharmacological agents such as metformin and cyclosporin A (Hunt et al., 2010;Erlandsson et al., 2011;Dadwal et al., 2015;Nusrat et al., 2018;Ruddy et al., 2019) as well as brain-derived neurotrophic factor (Yu et al., 2013;Zhang et al., 2018), epidermal growth factor (Kuhn et al., 1997) and erythropoietin (Shingo et al., 2001;Kolb et al., 2007). More recently, NPCs have been shown to be electrosensitive cells. Herein we propose the use of electric fields (EF) as an innovative approach to facilitate neural repair by activating endogenous NPCs to expand in number, migrate to sites of injury and generate new cells to replace those lost to injury or disease.
Applications of direct current EF (dcEF) have proven successful at Abbreviations: NPC, Neural precursor cells; BPMP, Biphasic monopolar; EF, Electric field; dcEF, Direct current electric field; NSC, Neural stem cells; CNS, Central nervous system; SVZ, Subventricular zone; LV, Lateral ventricles; DCX, Doublecortin. eliciting NPC responses (expansion of the NPC pool, migration towards the negative cathode) (Babona-Pilipos et al., 2011Zhao et al., 2015;Chang et al., 2016;Sefton et al., 2020). Notably, prolonged exposure to dcEF in vivo can result in adverse effects including neuroinflammation, tissue damage and electrode degradation. These adverse outcomes are primarily attributed to accumulating charges from the unidirectional ion flow highlighting the need for an alternative, clinically relevant stimulation paradigm. The use of biphasic monopolar electric fields (BPMP EF) that balance charges injected into the tissue in the cathodal phase with removal in a negative anodal phase, results in zero net charge injected thereby reducing unintended damage from the build-up of charges at the electrode-tissue interface. We have previously reported that BPMP EF can successfully elicit kinematic and behavioural alterations in NPCs both in vitro and in vivo, in line with what was observed with dcEF (Babona-Pilipos et al., 2015;Iwasa et al., 2019), including increasing the size of the NPC pool, promoting migration and enhancing neuronal differentiation (Sefton et al., 2020). Despite the potential of EF application for neural repair, and the promising results of BPMP EF, the optimal stimulation parameters for NPC activation in vivo have not been realized. Herein, we report a novel, moderate throughput ex vivo system that enables the manipulation of distinct stimulation parameters and the subsequent effects on NPC behaviour in order to optimize NPC activation prior to application to expensive, labour intensive, in vivo studies. We use biological outcome measures of NPC activation including expansion of the NPC pool, migration, and differentiation to assess both the efficacies of our novel setup and BPMP EF parameters.
We designed and implemented a stimulation chamber that allows for direct delivery of biphasic stimulation to ex vivo tissue slices that included the SVZ (NSC niche). We designed the stimulation chamber setup to ensure cell survival and tissue integrity and demonstrated that NPC activation in the ex vivo model was similar to in vivo results using the BPMP stimulation parameters previously reported (Sefton et al., 2020). Upon validation of the setup, we demonstrated that current-controlled delivery of BPMP EF can activate NPCs while delivering more consistent charges than voltage-controlled delivery, thereby enhancing the efficacy of the approach. We further tested a range of amplitudes based on the charge injection safety threshold for brain tissue (Shannon, 1992;Kuncel and Grill, 2004;Cogan et al., 2016). We observed the greatest activation of NPCs when tissue slices were stimulated with a cathodal amplitude of 200 μA, as indicated by a 3.8-fold increase in size of the NPC pool and enhanced, cathodally directed, NPC migration. Importantly, we applied the stimulation parameters derived in the ex vivo model to an in vivo mouse model and showed similar NPC activation. Together, our findings demonstrate the efficacy of the ex vivo stimulation chamber in determining a set of effective stimulation parameters for NPC activation and underscore the relevance of our system for developing EF based strategies to repair the injured brain.

Animals
All animal work was approved by the University of Toronto Animal Care Committee in accordance with institutional guidelines (protocol no. 20011279) and the federal Canadian Council on Animal Care (CCAC). Experiments were conducted with adult 7 to 11 months old C57BL/6 male mice (Charles River).

In vitro NPC colony forming assay (neurosphere assay)
The in vitro neurosphere assay is used to assess the size of the neural stem cell pool (Coles-Takabe et al., 2008) following both ex vivo and in vivo stimulations. To isolate neural stem cells, mice were anesthetized with isoflurane (Fresenius Kabi) then cervically dislocated. The neural stem cell niche (SVZ) lining the lateral ventricle was microdissected and collected as previously described (Sefton et al., 2020). The tissue was dissociated with an enzyme mix containing hyaluronidase (0.83 mg/mL, Millipore-Sigma), trypsin (1.33 mg/mL, Millipore-Sigma), and kynurenic acid (0.13 mg/mL, Millipore-Sigma) at 37 • C for 25 min. Cells were centrifuged for 5 min at 1500 revolutions per minute (RPM), resuspended and triturated in trypsin inhibitor (0.67 mg/mL, Worthington Biochemical Corporation) with a glass pipette. The suspensions were then centrifuged at 1500 RPM for 5 min followed by trituration in supplemented serum-free media (S.SFM; comprised of 10X Dulbecco's modified Eagle's medium/F12, 30 % glucose, 7.5 % NaHCO 3 , 1 M Hepes buffer, l-glutamine, hormone mix, penicillin and streptomycin, epidermal growth factor (20 ng/mL; Millipore-Sigma), basic fibroblast growth factor (20 ng/mL; Millipore-Sigma), and heparin (2 μg/mL; Millipore Sigma). Cells were then centrifuged for 3 min at 1500 RPM and resuspended in 1 ml of S.SFM. Viable cells were counted with a hemocytometer with Trypan Blue (0.4 %, Thermo Fisher) and plated in S.SFM at the density of 10 cells/µl. After 7 days, the total number of clonally derived NPC colonies > 80 µm in diameter (Coles-Takabe et al., 2008) (i. e. neurospheres) were counted. The numbers of neurospheres reflects the size of the neural stem cell pool. Trypan Blue exclusion was used to assess cell viability of tissue slices (500 μm or 1000 μm thick) after 6 h in the ex vivo stimulation chamber.

Construction of ex vivo stimulation chamber
We constructed four ex vivo stimulation chamber prototypes ( Fig. S1A-D). The chamber setup with artificial cerebral spinal fluid (aCSF; composed of 124 mM NaCl, 5 mM KCl, 1.3 mM MgCl 2 , 26 mM NaHCO 3 , 10 mM glucose, 2 mM CaCl 2 , and 1 % penicillin/streptavidin) underneath the elevated tissue sample and requiring the shortest length of electrodes was selected (Fig. S1A) based on its ability (1) to stably deliver electric stimulation with minimal interferences, (2) to maintain tissue hydration with aCSF and (3) to easily accommodate electrode adjustment and attachment. Chambers were constructed with a 35 × 10 mm Petri dishes (Corning) with 2 holes drilled on the lid of the chamber 2 mm apart with diameter of ~ 500 μm. Platinum electrodes were created as previously described for in vivo experiments (Iwasa et al., 2019;Morrison et al., 2019;Sefton et al., 2020) with uninsulated wires of 127 μm in diameter (lot #571752, 767000, A-M Systems). The cathode and anode (each were 2.0 ± 0.1 mm in length) were mounted on a stereolithography 3-D printed connector (Form 2, Formlabs), and positioned 2.0 ± 0.1 mm apart as illustrated in Fig. 1A. Electrodes were then secured on the lid of the chamber with superglue. The stimulation chambers and two filters (Millicell cell culture insert, 0.4 μm, 30 mm diameter, Millipore-Sigma) were UV sterilized for 15 min. Silicone vacuum grease (Dow Corning) was applied to the edges of the filters with a syringe. Then, 3 ml of artificial cerebral spinal fluid was added between the filters and sealed with vacuum grease. The stage was then fixed on the bottom of the chamber. The chamber was placed in the incubator at 100 % humidity, 37 • C and 5 % carbon dioxide during stimulation.

Ex vivo tissue preparation
Brains were extracted after mice were anaesthetized with 5 % isofluorane and cervically dislocated. Brains were then transferred to a cold brain cutter matrix (stainless steel, mouse, coronal, 500 μm cut, 5527 Zivic Instruments) placed on ice. Razor blades (Persona) submerged in artificial cerebrospinal fluid (aCSF; composed of 124 mM NaCl, 5 mM KCl, 1.3 mM MgCl 2 , 26 mM NaHCO 3 , 10 mM glucose, 2 mM CaCl 2 , and 1 % penicillin/streptavidin) were used to section 500 μm or 1000 μm coronal slices of the brain. Sections containing the neural stem cell niche (SVZ) were selected and were placed on the filter to keep them hydrated in the stimulation chamber. Sections were positioned under the microscope with a scale bar to ensure that the electrodes were placed in the cortex at the following coordinates: 0.2 cm from the surface of the (caption on next page) cortex, 0.7 and 2.7 cm lateral of the midline (Fig. S1E). For migration studies, to increase the chance of observing cell migration following stimulation, electrodes were positioned on the corpus callosum as NPCs have demonstrated a marked propensity to migrate along the corpus callosum (Iwasa et al., 2019).

Construction of in vitro galvanotaxis chamber
Galvanotaxis chambers were constructed as previously described (Babona-Pilipos et al., 2012). Briefly, HCl acid-treated no. 1 glass slides (22 mm × 22 mm × 0.17 mm; VWR) were attached to the bottom of 60 mm × 15 mm Petri dishes. The slides were then coated with poly-l-lysine (100 μg/ml; Sigma-Aldrich) for 2 h at room temperature followed by 4 % (v/v) Matrigel (BD Biosciences) diluted in SFM for 1 h at 37 • C. Neuropsheres were plated in S.SFM on Matrigel for 17-21 h to enable NPCs to dissociate into single cells in the incubator at 100 % humidity, 37 • C and 5 % carbon dioxide. An additional no. 1 glass slide were used to cover the cells prior to stimulation sealed with vaccumm seal (Dow Corning). Agarose bridges were constructed by filling two 15 cm PVC tubing (2.38 mm i.d., 3.97 mmo.d.; Fisher Scientific) with 1.5 % (w/v) agarose in SFM. Electrodes used were made by coiling silver wires (Alfa Aesar) and chlorinating them in bleach for at least 20 min. Time-lapse imaging for 5 h was performed using Zeiss Axiovision 4.8 software during stimulation with varying pulse widths, at 10x magnification.

Biphasic electrical stimulation
All biphasic stimulations were applied with STG4002-1.6 mA stimulator.
For voltage-controlled stimulation, a cathodal pulse with an amplitude of 500 mV and a pulse width of 500 μs was first applied, followed by a 2000 μs, 125 mV anodal pulse and a silencing period of 1000 μs. For current-controlled stimulation, charge-balanced biphasic stimulation was applied with a waveform consisting of a 500 μs cathodal pulse of 100 μA and a 2000 μs anodal pulse of 25 μA. In parameter optimization experiments, stimulation waveforms consisted of cathodal pulse amplitudes of 50 μA, 100 μA, 200 μA and 300 μA and a pulse width of 500 μs, followed by anodal pulse with amplitude and pulse width that is a quarter and four times that of the cathodal, respectively. Silencing period for all experiments was set to 1000 μs for consistency, similar to the voltage-controlled stimulation waveform. For current-controlled stimulation involving migration studies in vitro, the cathodal and anodal current amplitude was set to the stimulator maximum of 1.6 mA and 0.4 mA, respectively. Cathodal pulse width was varied from 250 μs to 2000 μs. The frequency was adjusted by altering the silencing period based on the cathodal pulse width such that the charges being injected from the cathode are the same across waveforms with different frequencies.
Simulations were performed with LTSpice to predict the voltage and current being delivered to the brain tissue during stimulation. Electric field strength perceived by NPCs was calculated using the Electric Current module (ec) in COMSOL Multiphysics, where the measured voltage delivered to the brain was defined at the electrode terminal as the boundary condition and the electric field at a given location is determined by the gradient of voltage: E = − ∇V.

Cell kinematics quantification in galvanotaxis chamber
Zeiss Axiovision software was used to track NPC migration during stimulation. 45 cells located at least one cell body apart were selected for cell tracking. Individual cell |velocity| was calculated by dividing its displacement from its position at time zero to its final position by the total stimulation duration (60 min). Directedness of a cell was determined by finding the cosine of the angle between the x-axis towards the cathode and the straight-line displacement between the initial and final position of the cell.

Surgeries for in vivo biphasic stimulation
Mice were anesthetized with 5 % isoflurane and were placed onto a stereotaxic apparatus under 2 % isofluorane. During surgery, a heating pad at 37 • C (ATC1000, World Precision Instruments, Sarasota, FL, USA) was placed under the animal and fluids including 1.5 ml of saline and meloxicam (2.0 mg/kg) were administered. An incision was made along the midline of the scalp and the skull surface was exposed and dried. A dental drill (P/N 8177 #77, 0.018 ′′ , David Kopf Instruments) was used to drill two holes that were 2 mm apart for the electrode leads (anterior/ posterior + 0.8 mm, medial/lateral -1.7 and -2.7 mm from the bregma). The electrodes were inserted into the brain and secured onto the skull with Insta-cure + cyanoacrylate glue (BSI-106, Bob Smith Industries). BPMP stimulation was delivered to the mouse brain with a STG4002 stimulator. Animals were sacrificed immediately following 1 h and 3 h of stimulation via cervical dislocation for the neurosphere assay or by cardiac perfusion for immunohistochemistry.

Immunohistochemistry
To fix brain tissues for the visualization of DCX + , ex vivo tissue slices were removed from stimulation chamber and submerged in 4 % paraformaldehyde (PFA) at 4 • C overnight; mice that received in vivo stimulation were transcardially perfused with cold 4 % PFA and brains were removed and placed in 4 % PFA at 4 • C for 4 h. Fixed tissues were then transferred to 30 % sucrose in 1x phosphate-buffered saline (PBS) for cryoprotection. Tissue was cryosectioned (HM525 NX, Thermo Scientific) and 20 μm thick sections were collected on Superfrost Plus slides ) and stored at − 20 • C until ready for staining. Slides were thawed at room temperature and rehydrated for 5 min with 1x PBS and permeabilized with 0.3 % Triton X-100 for 20 min followed by the addition of 10 % normal goat serum diluted in 1x PBS as blocking solution for an hour at room temperature. Primary antibodies of doublecortin (DCX) rabbit monoclonal IgG (1:250; Santa Cruz sc-271390) or Sox2 mouse monoclonal IgG (1:100; Abcam ab97959) diluted in blocking serum was added to the sections overnight at 4 • C to label migrating cells. Slides were then washed 3 times with 1x PBS for 5 min each and incubated for 1 h at room temperature with secondary antibody of goat anti-rabbit IgG 568 (Invitrogen, A11001; 1:500), goat anti-rabbit IgG 647 (Invitrogen, A27040; 1:500) or goat anti-mouse IgG 568 (Invitrogen, A11004; 1:500) and DAPI for nuclear staining (Vector, Fig. 1. Experimental setup and stimulation paradigm for the ex vivo system employing an acute coronal brain section. (A) Schematics of the ex vivo chamber created. Two pieces of platinum wires (127 μm-diameter), 2 mm apart, are soldered onto a machine pin that is inserted into a 6 mm by 5 mm by 3 mm stereolithography 3D printed connector. The electrodes are secured on the lid of the chamber, positioned to penetrate the tissue slice placed on the filter. (B) Experimental design to assess the expansion of NPC pool. (C) Fold change in the number of neurospheres in the ipsilateral and contralateral hemisphere at 1-hour poststimulation in both the stimulated and unstimulated slices. (n = 3 mice/group, two-way ANOVA; **p = 0.001). The ipsilateral hemisphere of the stimulated slice had significantly increase in neurospheres. (D) Fold change in the number of neurospheres from the ipsilateral hemisphere of 500um and 1000um thick sections, with and without stimulation (n = 3 mice/group, two-tailed unpaired t test; *p = 0.03, **p = 0.008). Ipsilateral hemispheres have more neurospheres, irrespective of thickness. (E) Experimental paradigm to assess NPC migration. (F) Images of DAPI + /DCX + cells in the ipsilateral hemispheres after 3 h of stimulation. Dotted line indicates wall of the SVZ. White arrows indicate farthest migrating cell from the dorsolateral corner of the SVZ. White boxes are 20x magnification of the farthest DCX + cell. Quantification reveals that the presence of the EF leads to significantly greater migration from the SVZ. Scale bar = 50 μm. Data presented as mean ± SEM. CC: corpus callosum; Str: striatum; Dlc: dorsolateral corner; SVZ: subventricular zone (n = 4 mice per group, two-tailed unpaired t test; **p = 0.002).

Statistical analysis
Data were analyzed with Origin and GraphPad Prism 9 (GraphPad Software) and reported as mean ± S.E.M. Fold change in neurospheres compared between various ex vivo stimulation configurations including thicknesses, placements, and stimulation parameters were performed using two-way ANOVA. Two-tailed unpaired t-tests were used to analyze the total numbers and NPC migration distance. Velocity and directedness of cells were compared with One-way ANOVA. For all statistical analyses, p < 0.05 was defined as statistically significant.

A moderate throughput ex vivo chamber enables investigation of EF induced NPC activation
Previously, we have demonstrated activation of endogenous NPCs in terms of increased neuronal differentiation and expansion of the NPC pool after 1 h of BPMP stimulation in vivo using the following parameters: 500 mV cathodal pulse amplitude at 500 μs and 125 mV anodal pulse amplitude at 2000 μs pulse widths, followed by a 1000 μs silencing period (Sefton et al., 2020). While these parameters successfully activated NPCs, we sought to optimize the BPMP stimlulation parameters for in vivo applications (Babona-Pilipos et al., 2015). To this end, we developed an ex vivo stimulation chamber desiged for brain tissue slices which enables us to preserve the three-dimensional anatomical structures and mimics in vivo cues that affect NPC behaviour in response to EF. The close resemblance to an in vivo environment overcomes the lack of complexity in dissociated in vitro cell culture models and is less labour intensive and time-consuming than in vivo experiments for optimizing stimulation parameters. In our design, mouse coronal brain slices containing the SVZ were collected and placed in an ex vivo chamber on the stage of a carbon dioxide, temperature and humidity-controlled incubator microscope. Platinum electrodes secured on the lid of the chamber were positioned to penetrate the tissue with the cathode positioned laterally and the anode positioned medially to focally deliver the EF (Fig. 1A).
To validate our ex vivo chamber design, we first examined the activation of endogenous NPCs in tissue slices using the BPMP parameters previously employed (Babona-Pilipos et al., 2015;Iwasa et al., 2019;Sefton et al., 2020). To examine the size of the NPC pool, we stimulated the ex vivo slice in the chamber for 1 h and performed the NSC colony forming "neurosphere" assay (Sefton et al., 2020) (Fig. 1B). We first compared stimulated and unstimulated slices from the same anatomical location along the rostro-caudal axis of the brain (Falcão et al., 2012). We observed a significant 2.1-fold increase in the number of neurospheres from the hemisphere with electrodes implanted (ipsilateral hemisphere) compared to the electrode implanted but unstimulated slice (contralateral hemisphere) (1.4 ± 0.04 vs 2.1 ± 0.05 -fold increase, contralateral vs ipsilateral, respectively) (Fig. 1C). Importantly, these findings indicate that biphasic stimulation delivered via our ex vivo set up can indeed activate resident NPCs in the 500 μm brain slice and further, the expansion of the NPC pool is similar to the reported in vivo findings demonstrating a 2.3-fold increase in neurospheres following 1 h of stimulation (Sefton et al., 2020). Given that no significant differences were observed between the unstimulated slices and the contralateral of stimulated brains, ipsilateral and contralateral hemispheres of the stimulated slices were compared in subsequent experiments assessing the size of NPC pool.
To optimize the tissue health in the ex vivo brain tissue slice, we compared the numbers of live cells in 500 μm and 1000 μm thick tissue slices and observed a significant increase in the numbers of live cells in the 1000 μm thick slices at 6 h post-plating (61.34 ± 3.22 % vs 79.91 ± 1.60 %, 500 μm vs 1000 μm) (Fig. S1F). We performed the neurosphere assay with 500 μm and 1000 μm tissue slices following 1 h of stimulation and found similar expansion of the NPC pool in the ipsilateral (stimulated) hemisphere (2.3 ± 0.30 vs 2.5 ± 0.24 -fold increase; 500 μm vs 1000 μm thickness, respectively) (Fig. 1D). Therefore, subsequent experiments were performed with 1000 μm thick acute coronal tissue slices. NPC migration in response to EFs has been demonstrated in vitro and in vivo (Babona-Pilipos et al., 2015;Iwasa et al., 2019). To examine migration in the ex vivo paradigm, electrodes were placed in the medial cortex and lateral corpus callosum (anode and cathode, respectively) of tissue slices and stimulated for 3 h, followed by immediate fixation (Fig. 1E). We predicted that endogenous NPCs would undergo cathodal migration in the presence of the applied EF. We quantified NPC migration by measuring the position of the furthest DCX + /DAPI + relative to the dorsolateral corner of the SVZ in the ipsilateral hemispheres of two adjacent tissue slices both containing the SVZ (one slice received stimulation, while the other was stored in identical environment with electrode implanted for the same duration). DCX was used to identify migrating cells which includes NPCs, neuroblasts and oligodendrocyte precursor cells (Clarke et al., 2012;Boulanger and Messier, 2017). Migrating neuroblasts are the predominant progeny of NSCs in the SVZ. Quantification revealed DCX + /DAPI + cells significantly farther from the SVZ compared to unstimulated controls (215.1 ± 34.4 μm vs 435.8 ± 51.4 μm, control vs stimulated) (Fig. 1F). Finally, NPCs as defined by Sox2 + and DCX + co-expressioin were assessed and we observed a similar extent of migration following stimulation (Fig. S1G). These finding is consistent with previous studies reporting directed cell migration in the presence of an EF (Cao et al., 2013;Feng et al., 2017;Sefton et al., 2020). Together, the neurosphere assay and migration data confirmed that the ex vivo system was able to elicit a comparable NPC behavioural response as previously demonstrated following EF application.

Current-controlled stimulation affords greater control over delivered EF while eliciting similar NPC activation
Having validated the ex vivo system for NPC activation, we next sought to optimize the mode of delivery of the electrical pulses by comparing voltage-controlled and current-controlled stimulation. To achieve this, we modeled the EF delivery of voltage-and currentcontrolled stimulation to determine how the EF is perceived in the brain tissue following charge injection. When charges are injected into the brain, the electrode and brain tissue form an equivalent Randles circuit, consisting of a series brain tissue resistance (R s ), a parallel charge transfer resistance (R ct ) and a parallel double layer capacitance (C dl ) at the electrode-tissue interface ( Fig. 2A). It is of note that the membrane capacitance and bulk tissue capacitance on the overall impedance are negligible in the circuit as ex vivo tissue slices were used for acute stimulation with short pulse widths in our setting (Butson and McIntyre, 2005;Krukiewicz, 2020). We then simulated the Randles circuitry with LTSpice using the values of R s = 2795.14 Ohm, R ct = 4.13e5 Ohm, and C dl = 0.99e-7F established using electrochemical impedance spectroscopy. The simulation results are shown in Fig. 2B-C. Under voltage-controlled stimulation with waveform (V inj .), the corresponding current across the brain (I cor. ) and voltage across the brain V brain (Fig. 2B) are influenced by the interfacial charge transfer resistance (R ct ) and double layer capacitance (C dl ). They led to unpredictable peaks and unbalanced charges delivered to the brain according to our modelling results (Fig. 2B). However, under current-controlled stimulation with waveform (I inj. ) (Fig. 2C), the influence of interfacial impedance (R ct and C dl ) on the voltage across brain (V brain ) becomes negligible as all current (I inj. ) is consistently injected into the brain (Fig. 2C) (Evers and Lowery, 2021). Subsequently, from the measured voltage (V cor. ) in response to the injected current, voltage delivered to the brain (V brain ) was calculated by measuring the instant voltage drops that represent the brain resistance (R s ).
Based on simulation and experimental results ( Fig. 2B-C), voltage delivered to the brain V brain is more consistent for current-controlled compared with voltage-controlled stimulation, making the former a better mode of stimulation.
To assess the biological impact of the current-and voltage-controlled EF delivery modes, we performed the neurosphere assay. The activation of endogenous NPCs was examined after 1 h of current controlled stimulation to ex vivo tissue slices using parameters converted from the voltage-controlled experiments with 500 mV cathodal voltage and 500 μs cathodal pulse width (Fig. 2D). We observed no significant differences in the magnitude of expansion of the NPC pool in stimulated hemispheres between the current-controlled (1.8 ± 0.13-fold increase) and voltage-controlled (2.2 ± 0.06 -fold increase) stimulation (Fig. 2D). Hence, both modes of stimulations were able to elicit regionally distinct NPC activation to a similar extent and given the consistency of delivered voltage shown in Fig. 2C, subsequent experiments were performed using current-controlled stimulation in our ex vivo paradigm.

Stimulation safety threshold determines range for parameter optimization
To optimize the parameters for stimulation, we determined a range of parameters based on the established safety threshold for brain tissue (Shannon, 1992;Merrill et al., 2005). The safety threshold considers the maximum amount of charge that can be delivered with a given electrode without eliciting any detrimental cellular response (tissue threshold) or damaging the electrodes (electrode threshold). Given the electrode diameter of 127 μm, the area of contact between the electrode and tissue can be calculated to 398,780 μm 2 with 1000 μm thick brain slice. The tissue safety threshold is 30 μC/cm 2 for electrodes exhibiting a surface area above 100,000 μm 2 with the tissue (Cogan et al., 2016). With a constant pulse width of 500 μs, the maximum current that can be injected without damaging the tissue is ~ 250 μA. With regards to the electrode safety threshold, the electrode should not exceed its maximum water reduction potential during a cathodal charge injection of − 0.6 V (Fig. 3A). Thus, we determined that the maximum current that can be applied from the platinum electrode is 311 μA, given the pulse width of 500 μs.

Cathodal amplitude of 200 μA elicits the greatest activation of NPCs
We tested various combinations of cathodal amplitudes: 50 μA, 100 μA, 200 μA, and 300 μA. The EF field strengths at the site of interest (SVZ NPC niche) were predicted by simulations using COMSOL Multiphysics software (Fig. 3B, Fig. S2A-B). The measured voltage across brain and the EF received by NPCs within the range of 50 to 500 μA cathodal current magnitude are displayed in Fig. 3B. As predicted, the peak cathodal voltage and the voltage across brain increased with increasing current amplitudes based on voltage waveforms collected during stimulation (Fig. 3C). Stimulation of 100, 200 and 300 μA resulted in NPC activation with 200 μA exhibiting the greatest expansion in the size of the NPC pool (3.8 ± 0.29 -fold increase in neurosphere numbers) (Fig. 3D). Notably, when the charge injection safety threshold was exceeded by injecting 37.5 μC/cm 2 (i.e. applying 300 μA stimulation), we observed a significantly reduced expansion in neurospheres compared to the 200 μA expansion (2.1 ± 0.98 -fold increase) (Fig. 3D). Notably, 250 μA stimulation did not result in a significantly different expansion in the NPC pool compared to 200 μA (3.8 ± 0.29 vs 4.0 ± 0.76 -fold increase, 200 μA vs 250 μA, respectively) ( Fig. S2C-D). Hence the 200 μA BPMP parameter was selected for the further experiments.
We asked if 200 μA was also sufficient to promote NPC migration after 3 h of stimulation. Again, we found that DCX + /DAPI + cells migrated significantly further from the dorsolateral corner towards the cathode after stimulation (89.3 ± 13.2 μm vs 280.5 ± 32.4 μm, control vs stimulated) (Fig. 3E). These findings suggest that with an EF at the site of interest of 78.2 mV/mm (Fig. S2B), delivered by 200 μA, can activate endogenous NPCs in terms of both the expansion of NPC pool as well as migration of NPC progeny. Data presented as mean ± SEM. CC: corpus callosum; Str: striatum; Dlc: dorsolateral corner; SVZ: subventricular zone. (n = 3 mice/group, two-tailed unpaired t test; ****p < 0.0001).

Stimulation promotes directed migration irrespective of pulse width duration
In addition to cathodal amplitude, BPMP EF delivery is also dependent on the pulse width. Since our ex vivo stimulation chamber cannot directly image EF induced changes in individual NPCs in a tissue slice, we took advantage of the in vitro galvanotaxis chamber which is amenable to examining NPC kinematics (Fig. 4A). NPCs were plated on Matrigel coated glass slides and the EF was delivered through Ag/AgCl electrodes placed in the side reservoirs connected to the petri dish with cells via agarose bridges. This setup prevents cells from being in direct contact with any chemical by-products generated from the electrodes. Different pulse widths were delivered and their impact on the migration of NPCs, namely velocity and directedness, was examined. We tested each pulse width for 1 h duration over a period of 5 h (250 μs, 500 μs, 1000 μs and 2000 μs) in 3 distinct trial paradigms (Babona-Pilipos et al., 2015) (Fig. 4A). The cathodal and anodal amplitudes, as well as total charge injected, were kept constant and the cathodal and anodal pulse widths were varied (Fig. 4B). To ensure that continuous stimulation was not causing cells to decrease velocity over time due to fatigue, we compared the velocity of migration of the same cells during the first 30 min of stimulation, with their velocity at 270 min of stimulation, and found no significant differences between the two timepoints (0.55 ± 0.04 μm/min vs 0.50 ± 0.03 μm/min; 30 mins vs 270 mins, respectively) ( Fig. 4C).
We found that all pulse widths were able to promote significant directed cathodal migration compared to the unstimulated cells (Fig. 4D). While EF resulted in increased velocity, there was no signficiant different in the magnitude of the increase in velocity (Fig. 4E). In addition, individual cell paths (Fig. 4F) demonstrated similar migratory behaviours of NPCs regardless of the cathodal pulse widths examined. Hence, pulse width alone is not a primary factor that directly impacts NPC migration.

Enhanced stimulation parameters expand the neural precursor cell pool and promote NPC migration in vivo
We predicted that the BPMP parameters established using our ex vivo system would result in a similar activation of endogenous NPCs following BPMP application in vivo. To assess NPC activation in vivo, we applied current-controlled stimulation with a cathodal amplitude of 200 μA and 500 μs cathodal pulse width (Fig. 5A) using implanted platinum electrodes in the cortex of young adult mice (Iwasa et al., 2019;Sefton et al., 2020). We delivered electrical stimulation for 1 or 3 h starting at the time of electrode implantation and sacrificed the animals immediately following stimulation. The recorded waveforms collected from the brains in vivo showed consistency in the stimulation delivered with our ex vivo current-controlled stimulation (Fig. 5B). Most important, we found that the BPMP stimulation resulted in similar activation of endogenous NPCs with a 3.4-fold increase in the numbers of neurospheres after 1 h of stimulation (Fig. 5C) which is not significantly different from our ex vivo data (Fig. S3A).
Further, we observed DCX + cells significantly farther from the dorsolateral SVZ in the stimulated hemisphere as compared to control (implanted but unstimulated mice) (149.6 ± 27.2 μm vs 272.1 ± 25.7 μm, control vs stimulated) (Fig. 5D). No significant differences in the EFenhanced migration between control and stimulated brains was observed when comparing ex vivo and in vivo outcomes (280.5 ± 32.4 μm vs 272.1 ± 25.7 μm, ex vivo vs in vivo, respectively) (Fig. S3B).
Finally, consistent with previous in vivo stimulation studies (Chang et al., 2011(Chang et al., , 2016Sefton et al., 2020), we observed an increase in the total number of DCX + /DAPI + cells following 3 h of stimulation, suggesting enhanced numbers of NPCs migrated away from their SVZ niche in response to BPMP stimulation (38.17 ± 4.40 vs 73.42 ± 3.50 cell, control vs stimulated, respectively) (Fig. 5E). Together, these results demonstrate the efficacy of the ex vivo platform in establishing stimulation parameters that activate endogenous NPCs in terms of expansion of NPC pool, and promoting migration.

Discussion
Harnessing the potential of endogenous NPCs to promote neural repair is a promising approach to treat the injured CNS. Traditional approaches to activate NPCs have focused on the delivery of cytokines, growth factors, and drugs (including Metformin (Dadwal et al., 2015;Ruddy et al., 2019) and Cyclosporin A (Hunt et al., 2010;Nusrat et al., 2018)). Herein we have taken advantage of the fact that NPC are electrosensitive cells and their behaviour can be modulated by electric field application. We built upon a previously demonstrated stimulation paradigm using BPMP stimulation to develop and test an ex vivo model that enables direct stimulation of acute brain tissue slices to examine NPC activation including expansion of the NPC pool, migration and differentiation of NPCs in brain tissue. The ex vivo model enabled the examination of distinct stimulation parameters in order to optimize the NPC activation. Finally, we demonstrated that the current-controlled BPMP parameters of cathodal amplitude of 200 μA and pulse width of 500 μs that were established in the ex vivo model were equally effective at activating NPCs when applied in vivo. Our model affords a moderate throughput tool to explore novel electrode designs and electrode placements, as well as enable further optimization of stimulation parameters with the goal of in vivo application to activate resident NPCs.
Realizing the importance of optimal electrical stimulation paradigm in regulating NPC behaviours, efforts have been made to develop various in vitro strategies including galvanotaxis chamber (Babona-Pilipos et al., 2012;Feng et al., 2012;Dong et al., 2017), culturing cells on polymer scaffolds (Tandon et al., 2011), chamber with a silicone box (Kobelt et al., 2014), 6-well electrical stimulation chamber (Leppik et al., 2019), biphasic electrical current stimulator chip with a Teflon culture dish (Chang et al., 2011) and an electrotaxis chip (Zhao et al., 2020). However, the majority of the studies focused on improving the efficacy of dcEF and on different stem cell types. Despite the shared electrosensitive properties among stem cells, their responsiveness to EF varies depending on their lineage (Petrella et al., 2018). Hence, it is important to investigate a set of optimal BPMP EF parameters tailored for NPCs in an effort to enhance the activation of endogenous cells. By employing mouse brain sections, the tool allows factors, namely CNS tissue heterogeneity (Prinz et al., 2011;Brocker and Grill, 2013;Bertucci et al., 2019), microenvironment and endogenous EFs existing in brains (Cao et al., 2013;Iwasa et al., 2019), that influences the strength of EF perceived by endogenous NPCs to be taken into account. The close resemblance to the microenvironment of adult brains in ex vivo tissue slices overcomes the lack of complexity in in vitro models while avoiding time-consuming in vivo experiments. Our model presents a moderate throughput model that was successfully validated with previously established BPMP EF parameters.
We demonstrated a comparable expansion of endogenous NPC pool as our previous study (Sefton et al., 2020) as well as cathodal migration of NPCs and their progenies away from the SVZ along the corpus callosum. Its high efficacy has proven to be an effective setup aiding in determining optimal BPMP parameters to further enhance the smaller yet significant NPC responses to EF in vivo. Of note, we did not observe an increase in the total numbers of neuroblasts and migrating cells following electrical stimulation in our ex vivo model, as would have been predicted from previous studies reporting increased neurogenesis in response to EF application (Yamada et al., 2007;Matos and Cicerone, 2010;Chang et al., 2016;Sefton et al., 2020). We postulate that our ex vivo analysis did not allow enough time for NPC differentiation, limiting our ability to assess any potential alterations in NPC fate. However, the influence of EFs on NPC differentiation can be further explored in vivo as the lifespan of the ex vivo tissue slice will no longer restrain the duration of stimulation or time post-stimulation.
clinically relevant paradigm that can effectively activate NPCs in vivo. As the first step to improve the stimulation parameters, we sought to enhance the consistency of stimulation delivered to brains. To this end, we found that the use of current-controlled stimulation permitted a more consistent delivery to the brain irrespective of the unpredictable and dynamic impedance generated at the electrode-neural tissue interface. These results are consistent with the literature where many studies have shown that current-controlled stimulation offers a more stable and consistent stimulation than voltage-controlled, thereby minimizing potential fluctuations in interfacial impedance during stimulation (Miller et al., 2016;Evers and Lowery, 2021). The advantage of the currentcontrolled approach has made it more clinically relevant in recent DBS device designs, as companies begin to offer more current-controlled stimulation protocols (Lettieri et al., 2015). In this study, we also demonstrated its ability to elicit similar extent of expansion of NPC pool in the tissue following stimulation. Hence, current-controlled stimulation appears to be more suitable particularly when applied in a physiological environment. EF strength can significantly alter cellular behaviours regardless of the type of stimulation applied. We previously established in vitro and in vivo that EF strength of 250 mV/mm can elicit migration, cell proliferation, and neurogenesis in NPCs (Babona-Pilipos et al., 2012;Iwasa et al., 2019;Sefton et al., 2020). Thus, we manipulated BPMP EF delivered to the SVZ by primarily changing cathodal amplitudes to optimize the stimulation and adjusted corresponding anodal amplitudes to balance the charges injected to the tissue. Cathodal amplitude of 200 μA with an EF of 180 mV/mm at the SVZ demonstrated the greatest expansion of the endogenous NPC pool while retaining the ability to promote cathodal migration. Notably, when exceeding the charge injected to the tissue slice (37.5 μC/cm 2 ) by stimulating with 300 μA, we observed a decrease in NPC activation. This finding suggests that unintended tissue damage and microenvironment changes as a result of the stimulation could have exerted a negative impact on NPCs.
One of the limitations of our ex vivo stimulation setup is tissue health over prolonged periods of culturing. Contributing factors to the reduction of acute tissue lifespan include the exponential bacterial growth, despite the addition of antibiotics to the media, as well as the release of excitatory amino acids from the damaged cells following tissue slice preparation (Hinrichsen, 1980;Buskila et al., 2015). While we were able to examine the NPC expansion and migration, extended time in culture would be needed to examine NPC differentiation. This limitation could be addressed by employing organotypic slice cultures, which can be maintained for days to week, with the knowledge that these require more extensive preparation (Kim et al., 2013;Li et al., 2016;Linsley et al., 2019). Due to the thickness of our en bloc slices, we are unable to examine the migration of prelabelled NPCs over time in the presence of an EF. The use of 2-photon imaging may circumvent this issue in the ex vivo system, allowing for imaging penetration depth up to 1 mm which is appropriate in our setup (Theer et al., 2003;Aziz et al., 2019;Costantini et al., 2019).

Conclusions
In this study, we established a high efficacy ex vivo system primarily for testing and comparing the efficiency of various parameters. Along with COMSOL Multiphysics modeling, we were able to identify a set of current-controlled BPMP stimulation parameters that enhances NPCs activation. This study demonstrated the practicality of employing an ex vivo model as an alternative to time-consuming in vivo experiments. The novel ex vivo model also supports other usages including, but not limited to, testing of novel electrode materials, investigating brain properties, and examining effects of changing electrode configurations towards the goal of harnessing the therapeutic potential of well-characterized electrosensitive NPCs.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.