Transcranial direct-current stimulation (tDCS) is a non-invasive form of electrical stimulation which delivers direct current of amplitudes in the range of 0.5–4 mA via electrodes placed on the scalp. While it has been shown in research to be effective for the treatment of depression (Bennabi and Haffen, 2018) and pain (O’Connell et al., 2018), it has not yet been approved as a treatment by the FDA. Direct current is delivered through scalp electrodes, which does not generate action potentials (Milighetti et al., 2020) but does lead to changes in excitability (Nitsche and Paulus, 2001). Anodal (positive) current has been shown to enhance motor cortex excitability, whereas cathodal (negative) current can reduce excitability (Nitsche et al., 2008). Important parameters include current densities, stimulation duration, electrode size, and electrode shape, which determine the safety of the brain stimulation. Most studies reported a current density between 0.029 and 0.08 mA/cm², electrode size between 25 and 35 cm² with a stimulation current of 1–3 mA for a duration of 20–30 min (Nitsche and Paulus, 2001; Fregni et al., 2005; Gandiga et al., 2006; Boggio et al., 2007; Solomons and Shanmugasundaram, 2020). Additionally, most studies utilized conductive rubber or metal electrodes that are embedded in a sodium chloride-soaked sponge (typically between 15 and 140 mM NaCl) (DaSilva et al., 2011). A comprehensive overview of the various tDCS stimulation protocols can be found elsewhere (Nitsche et al., 2008; Solomons and Shanmugasundaram, 2020).

Despite the clinical success of many electrical neurostimulation techniques, there are still challenges that require additional investigation. These challenges can be categorized into (1) charge injection limit, (2) material stability, and (3) tissue health and function. These issues are reviewed in detail by Merrill et al. (2005), Cogan (2008), and Zheng et al. (2021a). Investigation into the effects of electrical stimulation has been largely focused on neurons, with less emphasis on the roles that non-neuronal cells play. Understanding the response of non-neuronal cells to electrical stimulation will likely provide key insights into the overall biological response to electrical stimulation of the CNS, leading to more stable and effective interfaces for neural stimulation.

Neurons play the primary role in conducting electrical signals in the nervous system. Many studies have investigated the effect of various types of electrical stimulation on neurons, finding that the responses of neurons are highly modulated by the properties of the electrical field to which they are subjected (Florence et al., 2009; Aberra et al., 2018; Varoli et al., 2018; Jakobs et al., 2019). DBS and ICMS utilize pulsatile stimuli which are effective at eliciting action potentials by depolarizing neurons via the action of voltage-gated sodium channels (McIntyre and Anderson, 2016). The most likely compartment of neurons to be activated by electrical stimulation are the axon terminals due to the higher density and lower activation threshold of voltage-gated ion channels than those present in the soma (Aberra et al., 2018).

Under high-frequency (HF) DBS, neurons first exhibit altered firing patterns, with action potentials in axons becoming entrained to the stimulation frequency. DBS primarily affects axons, instead of dendrites or the cell bodies. This leads to both excitatory and inhibitory effects of DBS on neuronal activity. The net effect of these complex changes is that DBS leads to a depolarization block in neurons in the target region, preventing the regular firing of action potentials, however, due to the unstable nature of these interactions, periodic release from depolarization block and bursts of action potentials have been observed. All these factors in turn influence the behavior of the network to produce the therapeutic effect. These short-term electrophysiological effects are followed by alterations in neurotransmitter release and changes in the expression of various proteins such as c-Fos, VEGF, BDNF, GAP-43, synaptophysin and α-synuclein, some of which can take days or weeks to manifest (Delaville et al., 2011; Vedam-Mai et al., 2012; Jakobs et al., 2019).

In ICMS, a smaller region of neurons around each electrode contact is stimulated to fire action potentials. Thus, the application of programmed timing and patterning of stimulation parameters to each electrode contact individually can produce a spatially and temporally resolved sensory percept (Parker et al., 2011). A study on the safety of chronic ICMS performed in feline cortex found that damage to neurons was related to the interaction between charge density and charge per phase. Tissue was stimulated at a range of parameters, from 800 to 1,600 μC cm^–2, 52 to 104 nC ph^–1, 130 to 260 μA for 7 h. They did not find evidence of damage to glia; however, they were primarily concerned only with neurons and used Nissl and H&E staining which could limit their ability to detect changes to glia (McCreery et al., 1990). Another study in non-human primates stimulated the cortex for 4 h per day, 5 days per week, for 6 months at amplitudes of 10–100 μA, with pulse train durations of 1 or 5 s, and duty cycles of 33–100%. Based on blinded grading of the post-mortem histology, they concluded that while there was considerable damage to the cortical tissue, there was no additional detectable effect from ICMS at the stimulation parameters used in their study. These results suggest that there may be an optimal range of stimulation parameters which can induce reliable activation of neurons and is low enough that it does not lead to excessive death or damage to neurons around the site of stimulation (Rajan et al., 2015). There are, however, limitations to these studies, such as having only a single timepoint of assessment and the nearby placement of electrodes in different conditions which may confound the results if the effects of stimulation are widespread, as well as the fact that gross morphology and/or total cell count were evaluated, rather than the distribution and structure of cells surrounding stimulated electrodes. Therefore, further work on understanding the response of neurons to electrical stimulation should be conducted using new technologies such as multiphoton imaging techniques to image the response of live brain tissue in real time at cellular resolution. These techniques have been refined and their utilization provides insights into the effects of electrode implantation on many aspects of the tissue response such as microglia morphology and behavior, neuronal morphology and activity, and changes to the vasculature and the blood brain barrier (BBB) (Kozai et al., 2012, 2016; Eles et al., 2018, 2019; Michelson et al., 2018; Yang et al., 2021; Zheng et al., 2021b). These techniques have been adapted to study changes induced by electrical stimulation that are separate from those induced by the implantation of the electrode itself (Zheng et al., 2021b). More studies along these lines are needed to develop a complete picture of the way electrical stimulation influences the different cell types of the brain over time in vivo under different stimulation parameters.

The effects of tDCS on neurons have been investigated via both experimental and modeling studies (Denoyer et al., 2020). tDCS does not directly trigger action potentials but modulates neuronal excitability by shifting the resting membrane potential (Milighetti et al., 2020). In general, anodal tDCS is found to be excitatory while cathodic is found to be inhibitory, although this may vary by brain region (Varoli et al., 2018). At the cellular level, the stimulation effects depend on cell morphology and the electrical field direction (Chung et al., 2020). In mice, anodal DCS has been reported to increase spine density and induced structural synaptic plasticity in the cortex when paired with an external stimulus in a BDNF dependent manner (Gellner et al., 2020). A reduction in the number of neurons in the stimulated area of cortex was seen in mice stimulated with tDCS at a high charge density of 198 kC/m, accompanied by an increase in neurogenesis in the subventricular zone based on DCX staining (Pikhovych et al., 2016). tDCS also led to an increase in mRNA levels of proteins that play a role in synaptogenesis and axonal development and regeneration such as BDNF, Synapsin I, calcium/calmodulin-dependent protein kinase type II (CaMKII), CREB and c-FOS as measured by RT-qPCR analysis in rats stimulated daily for 7 days. In an in vitro hippocampal slice study, applying direct current during plasticity induction boosted the extent of long-term potentiation (LTP) (Kronberg et al., 2020). It also triggered alterations in postsynaptic membrane potential during endogenous synaptic activity, although it’s unclear how comparable these findings are to stimulation conditions used in human patients.

Microglia are the resident immune cells of the CNS (Kettenmann et al., 2011) and exist in the healthy CNS in an inactive and ramified state with a round cell body and extensive processes that survey their surrounding environment constantly (Nimmerjahn et al., 2005). Upon injury of the CNS, microglia are the first cells to respond and adopt an activated amoeboid morphology, followed by cell proliferation, migration, and encapsulation of the damage site (Sofroniew and Vinters, 2010; Kozai et al., 2012). Biran et al. (2005) showed that chronic implantation significantly increased the density of activated microglia as early as 24 h after implantation and that the high microglia density continued for the entire implantation period. Kozai et al. (2012) used in vivo two-photon microscopy to study microglia behavior in real-time during and immediately after the implantation of a neural probe. Upon implantation, microglial cells directly adjacent to the inserted probes extended their processes toward the probe at a rate of (1.6 ± 1.3) μm min^–1 for 30–45 min without significant cell body movement over the first 6 h after the probe implantation. Six hours after probe implantation, 50% of the microglia within 130.0 μm of the probe surface showed morphological characteristics of transitional stage (T-stage) activation, generating limited new processes that rapidly extended while most were withdrawn (Kozai et al., 2012; Robin et al., 2018; Borrachero-Conejo et al., 2019; Chen et al., 2021).

Astrocytes are a subtype of glial cells within the CNS. They have a star-shaped morphology, and their processes surround both neuronal synapses and blood vessels. They play a key role in the homeostasis of glutamate, various ions such as Ca²⁺, K⁺, and water, and defend against oxidative stress, assist in scar formation, and tissue repair (Sofroniew and Vinters, 2010). They have also been shown to have a critical role in inhibitory signal transmission (Robin et al., 2018; Chen et al., 2021). In a study of electrical stimulation by organic cell-stimulating and sensing transistors, stimulation elicited an intracellular Ca²⁺ increase in astrocytes, hinting at a more active response to electrical stimulation (Borrachero-Conejo et al., 2019). Astrocytes are also involved in the response to injury. Following the initial microglia response, they are recruited to the site of implantation and respond by increasing in size and an upregulation of GFAP intermediate filament protein, a hallmark of astrocyte activation, as well as releasing factors which contribute to promoting the ongoing foreign body response. They eventually form a dense layer of cells which surround the device by typically 4–6 weeks post implant and create a barrier between electrode sites and surrounding neurons (Vedam-Mai et al., 2011; Kozai et al., 2015; Malaga et al., 2015; Salatino et al., 2017).

Oligodendrocytes are a type of glial cell that are produced from oligodendrocyte precursor cells (OPCs). One of their significant roles is to provide axons with insulation for efficient signal transduction. Myelinating oligodendrocytes enwrap neuronal axons with myelin, providing electrical insulation which can dramatically increase the speed of nerve impulse propagation (Bradl and Lassmann, 2010). One study investigated the effect of stimulating the corpus collosum through 25 μm platinum wire microelectrodes at different stimulation frequencies on the subsequent proliferation and differentiation of oligodendrocytes. They found that stimulation at 5 Hz was more effective in promoting differentiation of OPCs into oligodendrocytes, while stimulation at 25 Hz led to increased proliferation of OPCs. Based on caspase staining, their stimulation paradigm did not induce cell death in the corpus callosum. In the same study, using slice electrophysiology, they also found that stimulation frequency influenced the quantity and timing of glutamate release at the neuron-OPC synapse, suggesting that OPCs may respond differentially to different firing patterns of neurons and by this mechanism myelination may be tuned (Nagy et al., 2017).

In a rat stroke model, tDCS resulted in accelerated functional recovery via multiple cellular mechanisms including the recruitment of OPCs. The study indicated that both anodal and cathodal tDCS accelerated functional recovery and that following cathodal tDCS delivery, the oligodendrocyte precursors were observed to migrate toward the injury site, while microglia underwent M1 polarization. Their results corroborate the findings of Miron et al. (2013), which showed that M1-macrophages dominate early after demyelination and support proliferation and migration of oligodendrocyte precursors, while a later M2-polarization can induce the terminal differentiation of the precursors into mature oligodendrocytes (Braun et al., 2016). Studying the effects of several types of stimulation on migration and proliferation of oligodendrocytes and their relationship with other glial types as well as neurons is an active field of research, which can better inform the delivery of stimulation parameters to drive desired functional outcomes. For an in-depth review of the role of OPCs and oligodendrocytes in central nervous system insult, see Wellman et al. (2018).

Endothelial cells (ECs) line the blood vessels and in the brain, ECs form tight junctions, and along with basement membrane, glial membrane, and projections of astrocytes make up the blood brain barrier. The BBB is the barrier between the cerebral capillary blood and the interstitial fluid of the brain (Wolburg et al., 2009). The vascular blood-brain barrier is comprised primarily of highly specialized brain endothelial cells (BECs) which modulate the flow of substances into and out of the brain. The BBB modulates neuroimmune communication through several complex pathways, which can be categorized as (1) modulation of BBB permeability, (2) immune regulation of BBB transporters and release of signaling molecules, (3) immune cell trafficking. These specialized brain endothelial cells also help to regulate the functions of other resident brain cells such as astrocytes, neurons, microglia, and oligodendrocytes (Erickson and Banks, 2018). BBB disruption is implicated in central nervous system disorders such as epilepsy, multiple sclerosis, PD, and Alzheimer’s disease (Farkas and Luiten, 2001; van Vliet et al., 2007; Pienaar et al., 2015; Erickson and Banks, 2018), as well as others. Understanding its function and dysregulation upon insult is of particular importance for understanding the response to brain interfacing devices, as invasive neural implants necessarily involve disruption of the BBB. It is thus critical to study the effect of therapeutic stimulation on the BBB and brain endothelial cells to understand how their function is affected and may help improve recovery upon device implantation or in specific disease states.

While the focus of this review is on the response to electrical stimulation as distinct from device implantation, we will summarize key findings related to the effects of device implantation, traumatic injury and disease as they relate to endothelial cells and the BBB here in order to highlight the importance and difficulty of disentangling these effects from the response to electrical stimulation when it comes to brain implanted devices. Specifically, it has been shown that the implantation of DBS electrodes does induce the mechanical breakdown of the BBB, which can intensify the inflammatory response and negate some of the benefits from stimulation (Kim et al., 2013). These difficulties are particular to invasive stimulating devices as in these cases stimulation will be acting on tissue injured by the device implantation itself, whereas non-invasive stimulation (e.g., tDCS) will be acting on uninjured tissue, although often under the influence of a disease state and not in a healthy brain.

Electrical stimulation has been shown to modulate cellular activity by changing downstream signaling pathways such as MAPK-ERK1/2 and PI3K/Akt which are responsible for cell migration, survival, and growth (Chen et al., 2019). Furthermore, chronic electrical stimulation also leads to an elevation in Janus kinase-signal transducer and activator of transcription (JAK/STAT) signaling, which is critical for cytokine/chemokine signaling (Zareen et al., 2018). Another study demonstrated increased c-Fos in brain regions including cingulate cortex, posterior hypothalamus, and subiculum following DBS (Gimenes et al., 2019). Microstimulation of sensorimotor cortex also leads to the expression of c-Fos and c-Jun in the striatum (Parthasarathy and Graybiel, 1997). Like DBS and microstimulation, continuous A-tDCS exposure in the cortex also increased c-Fos mRNA levels (Kim et al., 2017). STN-HFS for 3 h in a PD rodent model has been shown to modulate the expression of the tyrosine hydroxylase gene, a key enzyme in dopamine synthesis. The therapeutic effects of HFS–DBS are also supported by a downregulation of CaMKIIa and Homer1, a protein associated with glutamate neurotransmission. This downregulation could bring about decreased sensitivity to glutamate in the basal ganglia downstream of the STN (Henning et al., 2007). Another study also indicated an increase in the expression of various trophic factors such as BDNF and VEGF and the synaptic markers GAP-43, synaptophysin, and α-synuclein in the fornix following acute DBS. Interestingly, this change occurred within 2.5 h after the stimulation began, and returned to baseline levels after 5 h (Gondard et al., 2015). An A-tDCS study also reported increased mRNA expression of genes associated with synaptic plasticity such as BDNF, CREB, Synapsin I, and CaMKII following continuous stimulation (Kim et al., 2017). In an analysis of a rat model of electrical modulation therapy in spinal cord stimulation for the treatment of neuropathic pain, the authors found that anodal SCS decreased the expression of various glial-specific genes associated with inflammatory responses and microglial activation (Itgb2, Cd74, Cd68) and reactive astrocytes (Cxcl16, Tlr2) (Vallejo et al., 2020; Cedeño et al., 2021). In a study investigating SCS in a porcine model of ischemia-reperfusion triggered ventricular arrhythmia as a treatment for sudden cardiac death, the authors found that cardiac ischemia increased c-Fos expression in the spinal cord, specifically in microglia and astrocytes. Furthermore, they found that SCS treatment, which was effective at reducing cardiac arrhythmia, was accompanied by a reduction in c-Fos expression specifically in microglia and astrocytes and an increase in c-Fos expression in inhibitory interneurons in the deep dorsal laminae of the dorsal horn of the spinal cord (Howard-Quijano et al., 2021). These studies suggest that, at least in the spinal cord, electrical stimulation appears to have an anti-inflammatory and neuroprotective effect that is mediated by microglia and astrocytes. Further study into the mechanisms of these effects, as well as the effects of electrical stimulation on microglia and astrocytes in the cortex, are needed to fully understand the potential benefits this may have for devising future therapies.