The brain is one of the most complex and important organ in the body responsible for maintaining homeostasis and coordinating bodily functions. The mammalian brain is composed of about 10⁶–10⁹ neurons in a complex hierarchical network (Herculano-Houzel, 2009; Herculano-Houzel et al., 2011). These neurons communicate with each other in many ways, one of which is through neural oscillations. Neural oscillations are repetitive or rhythmic patterns of neuronal activity generated throughout the central nervous system (CNS). These neural oscillations operate at varying frequencies and can be classified into six frequency bands: slow oscillations (0.1–4 Hz), delta (1–4 Hz), theta (4–7 Hz), alpha (7– 13 Hz), beta (14–30 Hz), and gamma (>30 Hz). The synchronized oscillations originate either from the activity of single neurons or from the interactions in the neuronal network. At the single neuron level, neural oscillations reflect the fluctuations in the resting potential or the rhythmic action potentials, which then trigger neural oscillations in post-synaptic neurons. On the other hand, at the neuronal network level, the synchronized activity of neural ensembles evokes neural oscillations at various frequency bands at the macroscopic scale, which can be measured by techniques such as electroencephalography (EEG), electrocorticography (ECoG), and magnetoencephalography (MEG). Neural oscillations in the brain have been widely studied in various areas of scientific research, particularly in cellular and cognitive neuroscience. Growing evidence suggests that neural oscillations are pivotal for neural communication. The synchronous activity of neurons generates rhythms that enable different brain regions to communicate and coordinate responses to both intrinsic and extrinsic stimuli. For instance, two groups of neurons in phase lock or phase synchrony can influence each other because they are simultaneously in an excited state and have strong communication with each other. However, neurons that are desynchronized have weak communication and do not effectively orchestrate responses to stimuli. Indeed, disruption of brain rhythms and neuronal synchrony have been implicated in the pathophysiology of various neurodegenerative disorders and neurological diseases including Parkinson’s disease (PD), Alzheimer’s disease (AD), and epilepsy (Buzsáki, 2015). In patients with PD, the progressive death of dopaminergic neurons in the substantia nigra causes circuitopathy, which leads to hypersynchronization of beta oscillations in the substantia nigra and other brain structures such as the globus pallidus interna, globus pallidus externa, and subthalamic nucleus (Chan et al., 2021). Similarly, network abnormalities including hypersynchrony and altered oscillatory activity in AD (Buzsáki, 2006; Palop and Mucke, 2016) and increased neuronal hyperexcitability in epilepsy (Adaikkan et al., 2019) have also been reported. Exploring these dysfunctional brain networks raises the possibility of using neuromodulation to normalize the oscillatory relationships and restore healthy neuronal communication.

Neuromodulation is a neurostimulation technique that alters neuronal network activity by the targeted delivery of stimuli in specific regions of the body. The underlying aim of neuromodulation is to modify abnormal circuitry to restore the normal physiological state (Herrington et al., 2016). Neuromodulation has been used in a wide range of applications including altering the activity of dysfunctional neuronal networks, as a potential medical therapy, and as a research tool for studying neural functions. Most neuromodulatory stimulation methods are electricity-based, although alternative stimulation methods also include the use of magnetic fields, optogenetics, and focused ultrasound. Generally speaking, neuromodulatory stimulation can be divided into either invasive and non-invasive methods. The major difference between invasive and non-invasive neuromodulation is that the former involves the introduction of a stimulation tool or electrode into the body cavity, whereas the latter is performed outside the body. Some examples of invasive neuromodulation include deep brain stimulation (DBS), spinal cord stimulation, and stimulation methods that target specific nerves such as the vagus or sacral nerves. These invasive approaches have been reported to have several beneficial effects. For example, clinical results from four randomized control trials conducted to determine the effects of DBS in Parkinson’s disease showed that it significantly reduced the need for oral levodopa treatment while limiting dyskinesias by improving levodopa “On” time and reducing the “Off” time (Deuschl et al., 2006; Smeding et al., 2006; Weaver et al., 2009; Williams et al., 2010). The use of invasive neuromodulation as potential treatment options for chronic pain, urinary incontinence, epilepsy, and depression have also been reported. On the other hand, early non-invasive neuromodulatory approaches were unspecific. For example, electroconvulsive therapy developed in the 1930 s for treating patients with schizophrenia delivered the electrical stimulation to large areas of the brain. Subsequently, non-invasive neuromodulatory approaches have become more refined and novel approaches have been developed that can stimulate spatially distant regions of the brain. Some examples include transcranial direct current stimulation (tDCS), transcranial alternative current stimulation (tACS), and transcranial magnetic stimulation (TMS). The US Food and Drug Administration (FDA) has approved the clinical use of TMS in humans for alleviating symptoms in several mental health conditions, whereas the use of tDCS and tACS are still under investigation and awaiting approval.

There are inherent differences in the neurosurgical and stimulation risks between invasive and non-invasive neuromodulation. Depending on the specific invasive neuromodulatory technique, post-surgical complications and invasive stimulation-related adverse effects can occur, including neurological sequelae such as infection, intracranial hemorrhage, post-procedural seizures, and cognitive impairment, particularly decline in verbal fluency (John et al., 2021; Xu et al., 2021), and psychiatric side effects such as apathy, frontal disinhibition, and suicide in rare cases (Smeding et al., 2006). Although non-invasive neuromodulation does not require surgery and the stimulation is relatively painless with minimal side effects, there have been reported cases of minor adverse effects from non-invasive methods such as TMS and tDCS. These minor adverse effects, including headaches, scalp discomfort, fatigue, redness at the stimulated site, tingling, and itching sensations, were reported to be naturally resolved without the need for any medical intervention (Krishnan et al., 2015; Matsumoto and Ugawa, 2017). Nevertheless, exploring safe novel methods for altering neuronal network activity in the brain is imperative for treating diseases and improving overall brain health. Ideally, such non-invasive methods should have minimal side effects, while achieving its therapeutic goal or serving as a tool for investigative purposes. Transcorneal electrical stimulation (TES) is rapidly emerging as a safe non-invasive neuromodulatory technique that can exert beneficial effects in the brain via stimulating the eyes.

In this review, we discuss the advances in neuromodulation by TES in relation to the eye-brain connection. We summarize the conventional application of TES in the retina as a treatment for ocular disorders. We then give an in-depth explanation of the eye-brain connection and the possible mechanism of the neuromodulatory effects of TES in the brain. Lastly, we present the implications and possible future directions of TES as a tool for non-invasive brain neuromodulation. In the subsequent sections of this review, we use the term neuromodulation in the context of the alteration of neuronal network activity by TES, unless otherwise stated.

Transcorneal electrical stimulation is a non-invasive method of activating the retina and retinofugal pathways. In practice, an active electrode is placed on the corneal surface of the eye and an inactive reference electrode is carefully positioned underneath the skin in close proximity to the eye. The active electrode is connected to an electrical pulse generator that controls the delivery of electric current to the eye (Figure 1). The stimulation parameters such as current amplitude, stimulation frequency, pulse duration, and repetition periods can be adjusted and varied based on each subject and pathological condition. Several studies have shown that TES has immense neuroprotective benefits in retinal trauma and degeneration and in healthy subjects, as demonstrated in the animal studies in Table 1 (Morimoto et al., 2005, 2010; Ni et al., 2009; Tagami et al., 2009; Wang et al., 2011; Henrich-Noack et al., 2013b; Fu et al., 2018; Jassim et al., 2021) and the clinical studies in Table 2 (Inomata et al., 2007; Kurimoto et al., 2010; Oono et al., 2011; Schatz et al., 2011, 2017; Naycheva et al., 2013). These studies have provided crucial evidence for the therapeutic effectiveness of TES in the retina and have also examined the potential underlying mechanisms. Below is a summary of the conventional neuroprotective roles of TES in the retina.

Over the years, scientists have been looking for scientific evidence on the usefulness of eye research in studying the brain and its diseases. Many studies have begun exploring this eye-brain connection in terms of clinical diagnosis and therapeutic interventions for brain diseases. The concept that “our eyes are the windows to our souls” is intriguing in that our eyes provide an insight into our brain health both neurologically and optically. It is well known that the retina and the optic nerve are part of the CNS, as both structures extend from the diencephalon during embryonic development. The retina is a transparent tissue composed of specialized neurons such as RGCs arranged in layers interconnected by synapses (Figure 2A). Axons of RGCs form the optic nerve and extend to both the lateral geniculate nucleus (LGN) of the thalamus and the superior colliculus (SC) of the midbrain (Figure 2B; Dhande and Huberman, 2014). Visible light entering the eye is captured by light-sensitive photoreceptor cells in the outer retina. This process triggers a signal cascade that transduces visual information into electrical signals that move along the inner retina to the RGCs. These electrical signals are then relayed to higher visual centers responsible for proper interpretation and visual perception. It is crucial to state that although majority of the RGCs form outputs that project to the LGN and SC, there is a subgroup of specialized RGCs that also project into the suprachiasmatic nucleus of the hypothalamus. It has been noted that this region plays a key role in the sleep/wake cycle and circadian rhythms (Fernandez et al., 2016). Furthermore, other RGCs project their axons to the pretectum of the midbrain, which communicates with the motor nuclei to exert pupillary control (Levine and Schwartz, 2020).

Despite differences in the morphology of RGCs and CNS neurons, RGCs retain the typical properties of brain neurons including characteristic neuronal cell body, dendrites, and axons (Berson, 2008). The optic nerve formed by the axons of RGCs also retains fiber tracts similar to that of brain neurons. Specifically, the optic nerve is covered with a myelin sheath and enclosed in three protective meninges (Figure 2B). Similar to brain axons, trauma of the optic nerve leads to retrograde and anterograde degeneration of the injured axons. Consequently, this triggers other destructive processes including oxidative stress, myelin destruction, scar formation, inhibition of neurotrophic factors, accumulation of protein aggregates, and secondary degeneration such as the death of neighboring neurons (Levkovitch-Verbin et al., 2001, 2003; Herro and Lam, 2015; Yu-Wai-Man, 2015). It has been reported that the regeneration of damaged axons is restricted in both the brain and the optic nerve. It has also been demonstrated that the neurons in the brain and the optic nerve share similar factors that result in an internal environment that limits axonal growth following injury. Some of these factors include myelin-associated inhibitors (Nogo A), myelin debris, reactive astrocytes, and chondroitin sulphate proteoglycans (Maxwell, 2013; Pernet, 2017; Guttenplan et al., 2020; Pearson et al., 2020; Yu et al., 2021a; Fawcett and Kwok, 2022; Hammel et al., 2022). Interestingly, a significant amount of information about the axonal response to brain injury was initially discovered in studies on the optic nerve (Benowitz and Yin, 2008; Fischer, 2012; Fischer and Leibinger, 2012; Heskamp et al., 2013; Benowitz et al., 2017; Fischer et al., 2017). More recent studies have reported that signaling in complex circuitry involving the blockage of inhibitory neurotransmission of retinal amacrine cells plays a crucial role in RGC axonal regeneration after optic nerve injury (Yin et al., 2019; Sergeeva et al., 2021).

As an integral part of the CNS, the eye also interacts with the immune system. The eyes contain several immunoregulatory molecules including cytokines and other surface molecules (Figure 2C) that induce specialized immune responses similar to those of neurons in the brain and spinal cord (Streilein, 2003; Kaur et al., 2008). The anterior part of the eye contains a fluid called the aqueous humor, which is enriched in anti-inflammatory and immunoregulatory factors and is similar to the cerebrospinal fluid (CSF) that bathes the brain and spinal cord (Kaur et al., 2008; Aleci et al., 2015). Indeed, the eye is surrounded by two main blood-ocular barriers, the blood-aqueous barrier, and the blood-retinal barrier, which share similar structural characteristics and mechanisms to the blood-brain barrier. The inner regions of the blood-retinal barrier are composed of specialized endothelial cells linked by tight junctions surrounded by pericytes (Figure 2D), astrocytes, and Müller cell end-feet, which closely resembles the blood-brain barrier (Kaur et al., 2008).

As the eyes are a prominent extension of the brain, it is gradually becoming common practice to examine ocular manifestations when diagnosing brain pathologies. The eye-brain connection suggests the investigation of the eye could have clinical value in the early diagnosis of brain pathologies. Interestingly, results from ophthalmological assessments have revealed that several neurodegenerative disorders and neurological diseases such as AD, PD, and multiple sclerosis have manifestations in the retina. Similar to the pathological features in the brain of patients with AD, Aβ and p-tau can be present in the eye and are associated with ocular damage including cataract formation, retinal nerve fiber layer (RNFL) thinning, and retinal neuronal death (Goldstein et al., 2003; Ning et al., 2008; Liu et al., 2009; Gasparini et al., 2011; Chiu et al., 2012). In patients with PD, photoreceptors, and RGCs have been observed to display distinct swelling accompanied with morphological thinning of the RNFL (Moschos et al., 2011; Kirbas et al., 2013; Chang et al., 2022). In up to 50% of patients with multiple sclerosis, loss of vision has been reported as a presenting symptom, although other cases can also involve some degree of visual impairment. It is important to note that the aforementioned eye manifestations may not be specifically related to a particular disease, although their presence is a strong argument for the connection between the retina and the brain. More importantly, the direct connection to the brain via the long continuous tract of the optic nerve makes the eye ideally suited for stimulation by non-invasive approaches to exert neuromodulatory effects in the brain.

There are numerous unexplored neuronal pathways in the brain which could serve as potential targets for disease treatments and for enhancing brain function. Non-invasive neuromodulation techniques such as TES can alter brain activity and are rapidly becoming popular research tools due to their safety and not requiring surgery. Besides activating retinal neurons, TES can also modulate neuronal excitability in the brain. Preclinical studies have shown that TES can attenuate depression and anxiety-related symptoms in rodent models (Yu et al., 2019, 2021b, 2022a). Moreover, TES is adjustable depending on the disease type and disease progression. Standard anti-depressants affect entire biological systems and can lead to adverse side effects such as memory impairment, decreased concentration, and weight gain. Conversely, TES is generally safe without any serious side effects. Both short-term and prolonged TES in human subjects is generally well tolerated, although some minor and transitory side effects such as dry eyes, foreign body sensation, and corneal punctate keratopathy have been reported (Fujikado et al., 2006; Morimoto et al., 2006; Wagner et al., 2017; Jolly et al., 2020). Cases of adverse effects after TES have been rare and were possibly due to the high amplitude stimulation or mechanical pressure (Ni et al., 2009). Nonetheless, because of the simplicity of TES administration, the stimulator can be promptly turned off and the stimulating electrode removed in the rare case of adverse effects.

Besides its potential for treating depression, TES has also been explored as a potential therapy for cognitive dysfunction. In a recent study in aged mice and 5XFAD mice (AD model) treated with 3 weeks of TES, there were improvements in hippocampal-dependent learning and memory performance in both groups of mice (Yu et al., 2022b). Subsequent hippocampal immunostaining and protein expression analysis in 5XFAD mice showed that male mice had significantly reduced hippocampal plaque deposition and upregulated levels of PSD-95 protein. Although this suggests that the neuromodulatory effects of TES involve a postsynaptic mechanism, it is also interesting to note that elevated levels of PSD-95 are reported to protect synapses from amyloid-beta plaques (Shao et al., 2011; Yuki et al., 2014; Dore et al., 2021). Future studies on TES in other preclinical models of dementia are needed to investigate its beneficial effects on cognitive deficits and to delineate the underlying mechanisms.

Another future application is the use of TES as a non-invasive neuromodulatory tool for activating the brain to enhance vision-related behaviors, specifically visual attention. Previous studies involving the application of non-invasive tDCs in healthy subjects reported the modulation of spatial attention via altering the neuronal network activity ratios between the two brain hemispheres (Ni et al., 2019; Andres et al., 2020). Furthermore, electrophysiological studies reported that tDCs enhanced cognitive control and improved visual attention tasks following entrainment with alpha oscillations (Clayton et al., 2019). Similarly, prolonged TES in rd10 mice resulted in significantly increased ECoG alpha oscillations in V1 (Agadagba et al., 2021, 2022). Increased spontaneous firing rate in V1 neurons is known to be a pathological feature of rd. Alpha oscillations have also been noted to have an inhibitory role that helps the brain to focus on and process more relevant neural signals while ignoring less relevant signals. The above studies on tDCs parallel the rd10 study, which implies TES could be a beneficial tool for vision-related studies, especially studies focused on the functional rehabilitation of patients suffering from visual neglect.

This review highlights the neuromodulatory effects of TES in the brain via the unique eye-brain architecture. It is clear that TES has rapidly advanced from regulating retinal neurons to a cutting-edge research tool for modulating both visual and non-visual brain regions with beneficial outcomes in both healthy and diseased subjects. There is also ample experimental evidence from electrophysiological, behavioral, and molecular studies to support the potential use of TES as a non-invasive neuromodulatory tool for enhancing brain activity. Possible future directions of TES research include the mechanistic study of its neuromodulatory effects and further examining the cognitive and perceptual processes involved. Such mechanistic studies need to be properly designed with specific controls to reduce the generalizability of findings and to optimize the TES stimulation parameters for translation to human clinical trials.

All authors (SA, LL, and LC) contributed to the article and approved the submitted version.