There are three main forms of cortical stimulation: tDCS, TMS, and MCS. tDCS is a non-invasive neuromodulation method that works through two or more scalp electrodes. Clinically, tDCS is often used as a research tool to explore the role of different brain regions and is used as a treatment for psychiatric disorders such as obsessive-compulsive disorder, depression, schizophrenia, and addiction. When tDCS is used to treat SCI, the most common protocol is to combine it with motor training to promote activity-dependent plasticity. For example, Gomes-Osman and Field-Fote (2015) showed that a meaningful improvement in hand function among SCI patients was observed when tDCS was combined with the practicing of hand functional tasks. The priming effect of tDCS is time dependent (Sriraman et al., 2014), and anodic tDCS can improve motor performance more than cathodic tDCS (Machado et al., 2019), which may be the key regulatory condition determining the therapeutic effect of tDCS. In addition, Murray et al. (2015) demonstrated that stimulation of 1 or 2 mA during tDCS improved sensory perception. Unfortunately, there are no reports that tDCS promotes recovery of lower limb motor ability after SCI, with only one pilot study (National Clinical Trial, NCT03237234) having been carried out (Raithatha et al., 2016), the results of which may inform the efficacy of tDCS in subsequent clinical trials. However, tDCS is still very popular in clinical practice due to its non-invasive nature, and low cost, as well as its good therapeutic effect on a variety of neurological diseases and recovery of upper-limb function.

TMS is a safe, non-invasive brain stimulation technique that can also be used to treat psychiatric disorders such as addiction and depression and movement disorders. The magnetic stimulator consists of a strong magnetic field that can discharge a large current through an induction coil placed on the scalp (see Figure 1). A single TMS pulse in the primary motor cortex is paired with electrical stimulation of the peripheral nerves, such that corticospinal synaptic transmission is enhanced and plasticity of the residual corticospinal projection is induced (Christiansen and Perez, 2018). There is growing evidence that exercise combined with non-invasive stimulation targeting spinal synapses further promotes functional recovery. The amplitude of corticospinal responses elicited by TMS and the magnitude of maximal voluntary contractions in targeted muscles increased on average by 40–50% after paired corticospinal–motor neuronal stimulation combined with exercise in one study (Jo and Perez, 2020). As behavioral and physiological effects were preserved 6 months after the intervention, this suggests that the stimulation helped to maintain exercise gains. TMS combined with exercise can also benefit the recovery of lower-limb motor function. High-frequency TMS combined with gait training improved the lower extremity motor score and gait velocity in individuals with SCI who were able to walk over ground (Kumru et al., 2016). With a greater understanding of its mechanism of action and improvements in the equipment used to deliver it, TMS may play an important role in clinical treatments (Alexeeva and Calancie, 2016).

MCS with extradural electrodes promotes skilled forelimb function recovery after SCI in rodents, which is attributed to compensatory conjunctive germination of retained corticospinal tract axons (Martin, 2016). A combined spinal-motor cortex neuromodulation approach has been shown to promote the partial recovery of skilled motor behavior in SCI adult rats (Zareen et al., 2017).

Cortical stimulation—either non-invasive stimulation via tDCS and TMS or invasive stimulation via MCS—activates the corticospinal tract, resulting in immediate adaptive plastic changes or inducing motor output (Chari et al., 2017). Most of these methods can be combined with motor training or spinal cord and peripheral electrical stimulation to achieve the best training effect.

DBS is commonly used to treat conditions such as Parkinson's disease, dystonia, and essential tremor. In recent years, studies have found that DBS can also play a role in motor function recovery after SCI treatment. DBS has been experimentally applied to the subcortical motor regions of rats and has led to improvements in hindlimb function in both acute and chronic injuries (Hentall and Gonzalez, 2012; Bachmann et al., 2013). Excitatory DBS of the mesencephalic locomotor region significantly improved hindlimb function in rats with incomplete SCI and even promoted step-like movement after SCI paralysis (Bachmann et al., 2013). One study has shown the synergistic effect of EES and DBS on exercise in a severe SCI contusion rat model (Bonizzato et al., 2021). This approach promoted voluntary movement and authors associated DBS with walking intention, achieving autonomous movement, and reducing the stress response. Clinical improvement of gait in patients with incomplete SCI who were treated with DBS (NCT03053791) has shown that stimulation of the mesencephalic locomotor region induces forced movement and can transcend any volitional control (Mahlknecht et al., 2015). DBS-mediated brainstem repair is simple and comprehensive, although our lack of understanding concerning its mechanism of action currently limits its further application.

tcSCS, in which electrodes are placed on the skin above the spine to stimulate the dorsal roots and activate the motor circuit, is a non-invasive treatment for SCI. This method improves upper-limb function at the C3 level in patients with chronic incomplete SCI (Inanici et al., 2018). In another study, six subjects with motor complete injury were treated with tcSCS and buspirone, which is an important and prevalent neurotransmitter system for locomotor function. As a result, mean hand strength increased by >300%, and a corresponding clinically significant improvement was observed in upper extremity motor scores and the action research arm test. Some functional improvements persisted for an extended period after the study interventions were discontinued (Freyvert et al., 2018). tcSCS can increase active motor responses in the upper and lower extremities and trunk stability and can improve function and quality of life in patients with SCI. Although tcSCS is non-invasive, inexpensive and commercially accessible, it lacks the temporal and spatial precision of EES, which can activate specific muscle groups in a step cycle. In addition, the absence of clinical trials with appropriate control groups, has thus far made it impossible to know whether the motor response obtained through spinal cord stimulation is superior to the effects of other interventions used in therapy (Megía García et al., 2020).

EES, in which electrodes are surgically implanted on the dorsal surface of the spinal cord (see Figure 1), was initially used to treat chronic pain. EES also plays an important role in the recovery of motor and sensory function after SCI. van den Brand et al. (2012) designed electrochemical neural prosthetics and robotic posture interfaces to restore autonomous motor control in paralyzed rats. Dynamic task-specific training in the presence of EES, has been used to enable patients to walk independently across the ground while maintaining hip balance. This is the first report of the realization of independent walking by task-specific training in the EES environment for people who have completely lost sensorimotor function of lower limbs due to SCI (Gill et al., 2018). After EES and treadmill training in four patients with SCI, all four patients achieved independent standing strength and physical stability, and two of them were able to walk on the ground (Angeli et al., 2018). Another research group introduced neural techniques targeting spinal cord stimulation to achieve optimized real-time neuromodulation, enabling patients with permanent motor deficits or complete paralysis to walk autonomously under spatiotemporal stimulation (Wenger et al., 2014; Wagner et al., 2018). EES has also been found to enhance seated reaching-performance of individuals with chronic SCI (Gill et al., 2020). However, EES usually needs to be combined with frequent rehabilitation training, which is not suitable for a large proportion of SCI patients. To solve this problem, Gorgey et al. (2020) proposed that combined exoskeletal-assisted walking for SCI patients may be a feasible rehabilitation method, and preliminary positive results were obtained in patients with C7 complete SCI.

EES improved not only SCI hindlimb motor function but also forelimb fine motor function after cervical SCI. Alam et al. (2017) found that cervical EES increased the success rate of reaching and grasping in rats with cervical SCI, suggesting that cervical epidural spinal cord stimulation has therapeutic potential for rehabilitation after cervical SCI. Greiner et al. (2021) combined a computational model of the cervical spinal cord with experiments in rhesus monkeys and found that the muscle response to EES was modulated during exercise. There is also clinical evidence, such as a case study of 25 participants with cervical SCI who reported improvements in grip strength and motor scores 1 week after daily epidural spinal cord stimulation (Lu et al., 2016).

In recent years, EES has been found to be helpful for sensory recovery after SCI. A study has shown that EES induced somatosensory perception in four individuals with upper-limb amputations. Because restoring somatosensory feedback in amputations is critical for improving prosthetic control, this has important implications for the development of better prosthetic limbs that rein state certain sensations (Chandrasekaran et al., 2020).

There is evidence that proprioceptive feedback may provide powerful motor regulation. In particular, excitatory signal input from the periphery can initiate motor output and reset the oscillation period, which is essential for its normal function (Akay et al., 2014; Takeoka et al., 2014). Impaired sensory function leads to a decrease in the tail hopping frequency of free-swimming juvenile zebrafish, indicating that this form of mechanical sensory feedback results in active movement (Bohm et al., 2016). However, the technical challenges of selectively positioning and manipulating sensory pathways make it difficult to explore the contribution of sensory feedback to natural movement. Using the intersectional genetic approach, Gatto et al. (2021) found that multiple excitatory neurons in the superficial dorsal horn (lamina I/II) participated in the protective scratch reflex, whereas the lamina II/III and lamina III/IV induced the paw withdrawal response and balance-related motor correction, respectively, revealing the key principle that spinal soma sensory processing is driven by different spatial complements of excitatory neurons. In addition, light stimulation of cerebrospinal fluid–contacting neurons (CSF-cNs) in zebrafish showed that GABAergic sensory neurons in the spinal cord provide strong inhibitory feedback to the escape circuit during active movement to maintain balance (Hubbard et al., 2016). A recent analysis that used functional anatomy and optogenetics-assisted mapping showed that CSF-cNs can generate synapses on the axons of reticulospinal neurons including Mauthner cells and V2a neurons, and relay their information to the higher-level centers of the hindbrain to stabilize posture and increase movement speed (Wu et al., 2021).

The ability of spinal INs to integrate the spinal cord network and adapt to changing environments has become an important target for treatments aimed at enhancing neuroplasticity and/or promoting repair. They not only play an important role in the regulation of the motor and sensory activities of the undamaged spinal cord but also contribute to plasticity after injury or disease. The development of optical and chemical technologies has gradually improved our ability to distinguish subsets of spinal INs such that we can target specific subpopulation of INs in normal and injured spinal cords, advancing our understanding of spinal INs phenotypes in SCI repair. Capitalizing on this, Alilain et al. (2008) used optogenetics to demonstrate the functional influence of spinal phrenic motor circuitry on diaphragm activity by selectively stimulating transduced motor neurons and INs immediately caudal to a high cervical SCI. Given that sensory afferents are directly received through CPGs under the commissure and proper spinal cord connection, they are projected to the middle and ventral areas of the spinal cord, allowing them to affect the muscles and limbs of multiple joints. Optogenetics regulation of the INs in the spinal cord CPG provides the opportunity to directly observe the connection between the recovery of motor function and the CPG circuit after SCI. Optogenetics can be used to directly target the activation and inactivation of spinal INs and motor neurons in a spasticity model induced by SCI in mice. In one such study, excitatory Ins, and inhibitory INs were recruited into functional circuits by sensory inputs to induce uninterrupted neural activity, revealing the operational logic of spinal cord circuits after injury (Bellardita et al., 2017). Moreover, this spasticity is caused in part by the sensory activation of V3 neurons and the corresponding CPG circuit, which is important in initiating and coordinating motor output after SCI. Optogenetics activation of V3 interneurons in the spinal cord to induce spasms (Lin et al., 2019). Bui et al. (2016) found that motor-related dI3 INs have a similar role. Although dI3 INs are not necessary for normal motor function, they are necessary for the stable recovery of local motor activity after spinal cord transection. Eliminating the glutamatergic output of dI3 INs prevented motor recovery after spinal cord transection, proving that dI3 INs are involved in spinal cord microcircuits that regulate the plasticity of the motor system. Manipulating corticospinal-INs connections by optogenetics, electrophysiological, and transgenic tools have demonstrated that the different corticospinal-INs circuits of corticospinal neurons in the motor cortex and sensory cortex control specific aspects of skilled movement, such as the expression of Chx10 and Vglut3+ (Clovis et al., 2016; Ueno et al., 2018). At present, how to obtain, stimulate and/or enhance the spinal cord network through spinal INs and how to establish the sustainable excitability of damaged and/or denervated circuits are challenges that remain for researchers in this field.

More extensive research on promoting axon regeneration in the central nervous system has emphasized the problem of local loop synaptic integration. Since optogenetics technology can directly demonstrate the synaptic integration of regenerating axons and distinguish its function from indirect relay loops and target plasticity, it provides a promising way to explore SCI. Combining the light-induced stimulation of corticospinal tract (CST) sprouted axons with single-cell recordings to evaluate the functional integration ability of CST axons stimulated by sox11 in the cervical spinal cord circuit (Jayaprakash et al., 2016). Furthermore, moderate inhibition of the pericyte-derived glial scar promotes the regeneration of CST axons and improves the recovery of sensorimotor function after SCI, whereas optogenetics stimulation confirmed that the regenerated CST axons and the local spinal cord circuit under the lesion were remodeled (Dias et al., 2018). In addition, gabapentin, a voltage-gated calcium channel α2δ2 subunit inhibitor, promotes the plasticity, and regeneration of the corticospinal structure and the recovery of upper-limb function in adulthood, and the use of optogenetics strategies combined with in vivo electrophysiological records proves that these regenerated corticospinal axons are functionally integrated into the spinal cord circuit (Sun et al., 2020). Through optogenetics stimulation of C4 axons, the muscle activation caused by the fine motor circuit can be detected, and this system has been used to verify that administration of interleukin-10 can prevent damage to the pathological electromyography signal of the affected muscle (Chen et al., 2021). In addition, optogenetics targeting of host CST axons that had regenerated following neural stem/progenitor cell (NSPC) grafts resulted in focal synaptic responses, thus proving that NSPC grafts can form a local and spontaneously active synaptic network (Ceto et al., 2020).

Through transcranial optogenetics positioning, a dynamic optogenetics movement map after SCI can be drawn to reflect the damage and recovery of longitudinal structure and function, which expands the traditional applied behavioral and histological analysis and evaluation methods (Qian et al., 2019). Light stimulation of the cuneiform nucleus in Vglut2-ChR2-EYFP mice can induce movement that makes these mice stop abruptly and make sharp turns when approaching a corner. Given that the Vglut2-positive neurons of the cuneiform nucleus are targets for increasing motor activity, it is also likely that they may improve motor function after SCI (van der Zouwen et al., 2021). In fact, optogenetics has recently been used to dissect the spinal cord circuits responsible for evoking rhythmic and stimulating limbs (Llewellyn et al., 2010; Hagglund et al., 2013). Computational model evidence has also shown that optogenetics activate axons abide by the physiological order of diameter from small to large, which is valuable for the recovery of motor function after SCI (Arlow et al., 2013).

The widespread and long-term use of neuromodulation devices in the clinic is currently limited by a number of factors, including price, availability, and the expertise required to operate certain devices. Although implantable nerve electrodes have strong advantages, the durability of implanted electrodes is a common problem. Implanted electrodes can cause tissue damage and glial growth around the implant site, resulting in inefficient recording, and stimulation (Lee et al., 2016). In contrast, tcSCS and tDCS have the potential to be rapidly and widely delivered to patients with SCI (James et al., 2018). Therefore, some researchers believe that tcSCS and tDCS are safer and more accessible treatment methods for some patients with SCI (Taccola et al., 2020).

How to achieve long-term biosignal recording is a prominent challenge across the entire field of bioelectronic medicine (Lee et al., 2016; Bouton, 2019). Recent advances in the optimization of neural electrode–tissue interfaces—including electrode materials (based on graphene or CNT fibers), electrode structures (flexible electrodes), nano-coatings and hydrogel-based neural interfaces—have helped to achieve this goal (Liu et al., 2020) (see Table 1). These optimization methods can effectively improve the long-term stability of the neural interface and chronic inflammatory response during the process of biosignal recording and regulation. Another category of devices, wearable electronic devices, could fully address the problems of implanted electrodes and make it possible to monitor physiological signals in real time (Bouton, 2019), although the effectiveness and accuracy of such devices need to be further developed.

To optimize electrode design and targeting, another major challenge is the development of an appropriate delivery system for electrode implantation. The use of stereotactic frames or frameless positioning systems during spinal surgery can lead to problems of inaccuracy due to the variability of the surface markers used (Lobel and Lee, 2014). Linking multimodal in vivo neuroimaging with neuromodulation strategies is a powerful combination that is expected to significantly advance neuromodulation technologies and provide more precise and effective treatment for refractory neurologic disorders (Edwards et al., 2017). Current advances in real-time magnetic resonance imaging coupled with frameless stereotactic approaches are enabling faster, more precise placement of DBS electrodes (Starr et al., 2010; Edwards et al., 2017).

The selection of participants and the determination of key stimulus parameters and timing of intervention are important considerations in the design of neuromodulation intervention trials for SCI. The design of the trial needs to consider whether neuromodulation intervention should be used in the early (plastic) or late (stable) stages of SCI, especially when combined with rehabilitation training (James et al., 2018). The influence of different stimulation sites and parameters on treatment outcomes is significant but is typically not addressed. Increased understanding of the mechanisms of neuromodulation would be of great help in solving these problems. It should be noted that electrical neuromodulation is a very complex technology. Except for clinical trials, patients should not rush into electrical stimulation therapy, lest more serious consequences occur (Willyard, 2019).

Despite the rapid development of neuromodulation as a treatment for SCI, we do not know much about its underlying mechanisms. In fact, this severely restricts the further development and wide application of neuromodulation. Are there certain coding rules for the electrical signal that can effectively improve the plasticity of the central nervous system? What are the effective stimulus parameters? Do the different parameters have a relatively fixed range of therapeutic effects or do they change in real time? When do stimulatory interventions have the best effect on sensorimotor recovery? Which gene expression changes in the molecular network are controlled by neural regulation? The answers to these questions will greatly promote the application of neuromodulation in clinical trials for SCI.