Interactions between descending and spinal circuits on motor output

Abstract

Introduction:

Supraspinal influence on spinal networks remains incompletely understood yet is crucial for restoring motor function after neurological insults, including spinal cord injury or stroke. Neuromodulation interventions have been employed with varying success to aid recovery and can be utilized to investigate the relationship between supraspinal and spinal networks.

Material and Methods:

We elicited hindlimb muscle responses by motor cortex stimulation paired with either epidural or transcutaneous spinal stimulation in neurologically intact Long-Evans rats and Yucatan miniature pigs.

Results:

Our findings indicate that modulation of sensorimotor networks using the two stimulation modalities varies with the intensity of spinal stimulation. Specifically, spinal stimulation at near-motor-threshold levels modulates the magnitude of the weak descending volleys, with pronounced increases in compound motor evoked potential magnitude of up to 400–500%. As spinal stimulation intensity increased, we observed a transition from modulated cortically evoked motor responses toward modulated spinally evoked motor responses. However, when spinal stimulation was supra-motor-threshold, the conditioned responses were abolished. We also examined the effects of timing between paired cortical and spinal stimulation and found that the highest modulation occurred when delivering spinal stimulation using a latency approximately equal to the central conduction time of cortical stimulation.

Conclusion:

The capacity of cortical stimulation to modulate the effects of spinal modulation can be described as a convergence of supraspinal and spinal networks on the motor pathway. Overall, our results suggest potential stimulation strategies that capitalize on supraspinal-spinal interactions without the need for targeting individual motor pools with focal spinal stimulation.

Introduction

Supraspinal command centers and rhythmogenic central pattern generators within the spinal cord work together to modulate motor function [1] by engaging in ongoing and precise real-time decision-making from exceedingly intricate inputs [2, 3]. Furthermore, experiments manipulating supraspinal influence on the spinal cord, such as those conducted in spinal cord transection or decerebrate models [4–6], have demonstrated that spinal neuronal networks disconnected from supraspinal structures maintain remarkable automaticity in expressing rhythmic activity, as well as in processing autonomic and sensory inputs that can modulate functional outcomes [7]. However, the interaction and regulation between these supraspinal and spinal networks is complex and is not entirely understood.

As both supraspinal and spinal networks are key components in functional movements, they are frequently targeted for neuromodulation interventions aimed at restoring functional outcomes following spinal cord or brain injuries. Modulation of these neural networks can be achieved using epidural spinal stimulation (ESS), transcutaneous spinal stimulation (TSS), and motor cortex stimulation (MCS). However, the mechanisms of spinal cord neuromodulation for the recovery of motor functions are still unclear and only working hypothesis have been provided so far [8]. Computational [9, 10] and electrophysiological [11–13] studies have demonstrated that spinal stimulation can produce motor responses via Ia afferents synapsing on α-motoneurons. However, other studies have suggested that spinal stimulation can interact with other neural components within the spinal cord, including interneurons, ascending sensory fibers in the dorsal columns, descending motor tracts, and other polysynaptic pathways [14–17]. In this context, it is worth noting that heterosynaptic inputs from Ia afferents and corticospinal neurons can interact at the motoneuron pool to facilitate cortically induced motor evoked potentials (CMEPs) and spinally evoked motor potentials (SEMPs) [15, 18, 19]. However, transient alterations in the excitability of polysynaptic spinal pathways, such as presynaptic inhibition [20–22] and reciprocal inhibition [23], may also play a role in this interaction. When engaged using tonic spinal stimulation, the interaction between the descending inputs and spinal networks has been shown to increase the volitional drive to motoneurons and promote movement [24–26]. However, our understanding of the intricate interplay between spinal stimulation and descending inputs from supraspinal centers remains limited. The current work addresses a crucial, yet unresolved, scientific question: How does spinal cord stimulation enhance motor output? Evidence suggests that relatively diffused electrical stimulation across a large portion of the dorsal surface of the spinal cord effectively target the spinal interneuronal networks responsible for locomotion [27] and a widespread increase in spinal network excitability make spinal neurons more responsive to descending and afferent inputs [25, 28] Moreover, paired corticospinal-motoneuronal stimulation (PCMS) is a protocol involving synchronized stimulation specifically targeting the corticospinal pathway at the spinal cord level [29–32]. In particular, the timing of repeated paired TMS and peripheral nerve stimuli can either potentiate or depress corticospinal transmission [29]. In intact human subjects, CMEPs and voluntary muscle activity were depressed when postsynaptic input preceded presynaptic timing by approximately 15 ms, and similarly depressed when the presynaptic potential occurred approximately 20 ms before the postsynaptic input. Conversely, when the presynaptic potential arrived approximately 1 ms before the postsynaptic input, both responses and volitional muscle contractions were significantly facilitated for at least 30 minutes in neurologically intact subjects [29], and enhancing muscle function and manual dexterity for up to two hours after chronic incomplete cervical SCI [30]. Thus, the induction of synaptic plastic changes dependent on the timing of paired pulses at residual corticospinal-motoneuronal synapses may provide a mechanism for strengthening residual descending connections to enhance volitional motor control with functional benefits after incomplete spinal cord injury [30]. Interestingly, prolonged facilitation and more consistent outcomes were achieved when the number of stimuli in the PCMS was doubled [31].

In the current study, we aimed to investigate three main hypotheses: First, that subthreshold ESS can enhance weak cortical drive. Second, that the strength and timing of cortical and spinal pulses interact in tailoring the resulting motor output. Finally, we examined whether increasing the intensity of spinal stimulation beyond the motor threshold can dominate and suppress the cortical signals. To test these hypotheses, we employed a combination of MCS to activate descending motor systems and ESS/TSS to activate spinal lumbosacral motoneurons. This approach allows us to evaluate the interactions between supraspinal centers and spinal circuitry.

Our experimental design involved the use of acute fully anesthetized, neurologically intact Long-Evans rats with epidural electrodes implanted along the lumbar spinal cord, electrodes positioned about the motor cortex, and electromyography (EMG) recordings from the limb muscles. Following promising results from the rat model, these findings were replicated in a large-animal translational model using Yucatan miniature pigs, further supporting the clinical relevance of our study. We administered weak cortical stimulation at near-motor threshold (MT) intensity to replicate the diminished descending inputs caused by incomplete or discomplete spinal cord injury (SCI) [26, 33, 34]. We found that low-intensity ESS facilitated the expression of full muscle responses to weak cortical pulses. As ESS intensity approached the MT, SEMPs were enhanced by cortical inputs. However, when ESS exceeded the MT, CMEPs became distorted, likely due to the dominance of spinal reflexes in the motor response. Finally, strong cortical MEPs were unaffected by under-MT ESS. Similar effects of timing and intensity modulation occurred when TSS was delivered instead of ESS.

In summary, modulatory interactions occur between the two neuromodulation sites with cortical stimulation influencing ESS modulation of spinal networks and vice versa. Although the precise neural pathways responsible for interactions between supraspinal and spinal motor networks remain to be fully explored, our findings reveal a peculiar functional interplay between the two. The evidence also suggests that motor output can be predicted based on the relative influence of cortical and spinal inputs, offering potential for improved control of motor function through spinal stimulation and volitional intent.

Methods

Animal Safety

The Institutional Animal Care and Use Committee (IACUC) of the Houston Methodist Research Institute (Houston, TX) approved the experimental methods and procedures, which were carried out in compliance with the ARRIVE standards (Pig study ID: IS00006754; Rat study ID: IS00006709) and the National Institutes of Health (NIH) Guide for the care and use of laboratory animals. Additionally, animal work was done in accordance with the European Union directive overseeing animal experimentation (2010/63/EU). Furthermore, the housing, feeding, and general care of the animals were conducted in accordance with the guidelines set forth by the US Department of Agriculture and the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). A total of 11 Long Evans rats and 4 Yucatan minipigs were used for the described experiments.

Surgical Implantation of ESS and MCS

For both rats and minipigs, animals were anesthetized for the duration of the experiment and were euthanized after collecting electrophysiological recordings. While under anesthesia, the animal’s temperature was continuously monitored and maintained at 37° C using warm water heating pads, and additionally, for minipigs, Bair Hugger warming blankets were used. Pinch reflexes were periodically assessed to ensure animals were deeply anesthetized, and if reflexes were observed, anesthesia dosage was adjusted. Specifically, eleven adult female Long Evans rats (300 to 350 g body weight; Charles River Long-Evans, Houston, TX, USA) were anesthetized by intraperitoneal (IP) administration of a ketamine (100mg/kg) and xylazine (5mg/kg) mix. At the end of the electrophysiological recordings, euthanasia was induced by CO₂ administration followed by thoracotomy. Four female Yucatan minipigs (P001, P002, P003, P004) ages 13-, 8-, 4-, and 4 months old and weighing 51, 38.4, 20.6, and 21 kg, respectively, were pre-anesthetized with an intramuscular injection of midazolam (0.3 mg/kg), ketamine (20 mg/kg), and hydromorphone (0.15 mg/kg), then endotracheally intubated, ventilated, and maintained in the prone position with inhaled isoflurane (2–4%) for the implantation of stimulation equipment. As isoflurane can diminish electrophysiological recording, gas anesthesia was then replaced with total intravenous anesthesia (TIVA) using propofol (8 mg/kg/hr), ketamine (5 mg/kg/hr), and dexmedetomidine (2 μg/kg/hr). Following a 20-minute washout period of gas anesthesia, terminal electrophysiological recordings were collected. Throughout the entire procedure, a veterinary professional with expertise in performing research using porcine models was present.

In rats, the electrode implantation procedure has been previously described and verified by Taccola et al [35]. Briefly, to place the epidural stimulating electrode, a partial laminectomy was done at T12/L1 and the complete removal of T13 vertebrae was performed to expose the spinal cord. Teflon-coated stainless-steel wires (AS 632, Cooner Wire Co, Chatsworth, CA, USA) were advanced through the partial laminectomy site up to T12. A small notch (0.5 – 1.0 mm) to each wire was made to remove the Teflon coating, thereby exposing the conductor which serves as the electrode. These notched wires were then aligned over each side of the dorsal cord approximately 0.5 mm lateral from the midline. The two wires were used as cathode and anode for bipolar stimulation. In minipigs, the electrode implantation procedure has been previously described and verified by Steele et al. [36]. Briefly, a laminectomy of the sixth lumbar vertebra was performed, to place a 4-column, 32-contact epidural spinal stimulator array (CoverEdge X; size: 67mm × 10mm, Boston Scientific, USA) in the epidural space corresponding to the L5 and L6 vertebrae. After the initial implantation of the epidural spinal stimulator, C-arm fluoroscopy and electrophysiological tests were performed to verify proper placement. First, fluoroscopy images were used to establish that the array corresponded to the approximate anatomical midline. Then, to ensure adequate coverage of the lumbosacral enlargement, electrical stimulation was delivered to both the rostral and caudal regions of the array. The goal was to preferentially target and differentiate activation between proximal and distal hindlimb muscles. Once the electrodes were positioned in the desired location, they were sutured in place, and the incision site was closed. Following the conclusion of the electrophysiological recordings, a 360° three-dimensional (3D) rotating fluoroscopy (Siemens AXIOM-Artis, Munich, Germany; slice thickness: 0.49 mm) scan was taken to validate the final epidural array positioning.

To provide MCS, in rats, a 1 × 3 mm horizontal slit was created at the craniotomy site on the left side (infraorbital line = −1.3 cm, interaural line = 0.8 cm). This exposed the motor cortex at bregma −1 mm. A monopolar needle electrode was then inserted through the slit, extending 1 mm in caudal direction to define the site of cortical stimulation at bregma −2 mm that corresponds to the hindlimb area [37]. In minipigs, two sterilized stainless-steel screws (length: 15 mm; diameter: 1.1 mm) were advanced into the skull about 5 mm, 2–3 cm bilaterally from the plane formed by the midline of the ears and snout and placed approximately 3–4 cm above the eyeline to [38].

Electromyography

Electromyography (EMG) was performed with subdermal needle electrodes (needle length: 20 mm; Rhythmlink Columbia, SC, USA) attached bilaterally to lower limb muscles. In rats, EMG was recorded from biceps brachii (Bic) tibialis anterior (TA), and gastrocnemius (GM) (Fig. 1D, E). A common ground electrode was placed subcutaneously in the distal portion of the left forelimb. In minipigs, EMG was recorded for vastus lateralis (VL), semitendinosus (SEM), tibialis cranialis (TC), and gastrocnemius lateralis (GL) (Fig. 2D, E) with an additional subdermal needle electrode placed over the right calcaneal tuber as the amplifier’s reference. The EMG signals in both animal models were amplified using a differential amplifier (Octal Bio Amp; ADInstruments, Australia; gain: 100, range: ± 200 μV, resolution: 100 nV, common-mode rejection ratio: > 60 dB, input impedance: > 1 GOhm, < 100 pF). Measurements were sampled at 10,000 Hz utilizing a PowerLab data acquisition device (ADInstruments) and recorded in LabChart (ADInstruments version 8.1.24).

Figure 1. Determining cortically induced motor evoked potential (CMEP) and spinally evoked motor potentials (SEMP) motor threshold (MT) in rats.

A train of three 200 μs biphasic pulses with a 2 ms interstimulus interval was used to administer (A) motor cortex stimulation (MCS). A single biphasic 100 μs pulse was used for delivering epidural spinal stimulation (ESS) (B) at the T12/L1 vertebral level. The central conduction time (CCT) was defined as the approximate time for the MCS potential to reach the lumbar enlargement and was calculated by subtracting the latency of the SEMP from the CMEP. The (C) condition test interval (CTI) is defined as the time between the MCS train’s initial pulse and the ESS pulse. The intensities of (D) MCS and (E) ESS were gradually increased separately to identify the MT (which is denoted using color), using bilateral electromyography (EMG) recordings from two hindlimb muscles. The data presented are from an exemplary animal.

Figure 2. Determining CMEP and SEMP MT in minipigs.

MCS was delivered (A) using a train of five 500 μs monophasic pulses with an interstimulus interval of 1 ms. ESS was applied with a paddle array placed (B) between the L5 and L6 vertebrae, using a single monophasic 200 μs pulse. CCT was defined as the approximate time for the MCS potential to reach the lumbar enlargement and was determined by subtracting the latency of the SEMP from the latency of the CMEP. The CTI (C) is defined as the time between the first pulse of the MCS train and the ESS pulse. The intensities of (D) MCS and (E) ESS were gradually increased separately to identify the MT, using bilateral EMG recordings from the hindlimb muscles. Colored responses denote MT for the given muscle. The data shown is from a representative animal (P002).

Motor Threshold (MT)

For both MCS and ESS in rats, the MT was defined as the lowest intensity required to elicit consistent EMG responses from at least one of the recorded hindlimb muscles, indicated by a signal deflection three times greater than the standard deviation (SD) of the baseline.

For minipigs, MT was determined for each measured muscle and was defined as the lowest intensity that evoked consistent responses from that muscle. An evoked response was similarly defined as one which exceeded three times the SD of the baseline.

Motor Cortex Stimulation

MCS was administered by a constant current stimulator for both rats and minipigs (DS8R, Digitimer Ltd., UK). In rats, CMEP in hindlimb muscles (Fig. 1D) was elicited using a train of three biphasic pulses (each pulse lasting 200 μs) delivered at intensities ranging from 0.2 to 0.8 mA, and an interstimulus interval of 2–3 ms (Fig. 1A) [39]. Trains were delivered with a minimum interval of 1 second between repetitions. Biceps brachii EMGs were monitored to ensure MCS targeted hindlimb muscles. In minipigs, CMEPs in the hindlimb muscles were elicited using a train of five monophasic square-wave pulses with a pulse width of 500 μs and an interstimulus interval of 1 ms [38, 40], as seen in Fig. 2D. Stimulation intensity began at 30 mA and was increased in 10 mA increments to 200 mA or until the CMEP magnitude plateaued for two consecutive amplitude increases. Each train of five pluses was repeated three times per amplitude, with a minimum of five seconds between repetitions. Examples of the averaged CMEP as the stimulation amplitude are shown in Fig. 2D. For both rat and minipig experiments, in some cases, stimulation was intentionally limited before reaching a true plateau to avoid excessive muscle contractions and to ensure animal safety under anesthesia.

Epidural Spinal Stimulation

To deliver ESS, a constant current stimulator (DS8R, Digitimer Ltd., UK) was connected to the implanted electrode array. In rats, negative-first symmetrical 100 μs rectangular biphasic pulses at 2 Hz were delivered to generate SEMPs for the TA and GM (Fig. 1B). In minipigs, SEMPs were induced employing 200 μs monophasic square pulses at 2 Hz with an anode that consisted of the rostral four columns and a cathode comprised of the caudal four columns, as illustrated in Fig. 2B. Recruitment curves were generated using 0.1 to 0.5 mA increments of ESS; however, stimulation was limited in some cases before reaching a true plateau to prevent excessive muscle contractions and ensure animal safety. Examples of the averaged SEMP waveforms resulting from ESS in relation to the increase in stimulus amplitude are shown in Figs. 1E and 2E.

Transcutaneous Spinal Stimulation

TSS was only performed in one minipig (P004) and was applied to the skin rostrally from an approximate region of the lumbosacral spinal enlargement. Stimulation was delivered using an electronically controlled constant-current stimulator (DS8R, Digitimer Ltd., UK). A self-adhesive cathode (3.2 cm in diameter; PALS, Axelgaard Manufacturing Co. Ltd., USA) was positioned along the spine’s midline between the L4 and L5 spinous processes to provide stimulation. An oval anode (7.5 cm × 13 cm; PALS, Axelgaard Manufacturing Co. Ltd., USA) was attached to the lower part of the abdomen. TSS was delivered using a single 500 μs monophasic square-wave pulse. Recruitment curves of SEMPs were generated using 5 mA increments, starting at 20 mA and continuing up to 100 mA.

Combination of MCS and ESS (TSS)

Before the start of the combined MCS + ESS intervention, MCS – consisting of three pulses in rats or five pulses in minipigs – was adjusted to 0.8 and 1 MT, aiming to elicit responses in at least half of the monitored muscles. Individual MTs (in mA) for all muscles identified during MCS are presented in Table 1 for rats and Table 3 for minipigs. MCS was then delivered continuously at 1 Hz in both species. After a minimum of five seconds, ESS was delivered at 30 Hz [25, 41, 42] concurrently with the 1 Hz MCS so that the condition test interval (CTI), or the time between the first pulse in the MCS train and the ESS pulse, increased by 0.5 ms each time MCS was delivered (Figs. 1C and 2C). In total, 66 MCS trains were delivered during each conditioning sequence, with the CTI incrementally increasing by 0.5 ms across trials. ESS intensity was systematically varied between 0.2 and 4 mA (corresponding to 0.2 – 2.0 MT) (Table 2 for rats and Table 4 in minipigs). During the intervention, compound motor evoked potentials (MEPs) were recorded to investigate the modulatory effects of the combined MCS and ESS. Post-intervention CMEPs were collected up to 40 seconds after ESS was stopped to investigate the transient effects of combined MCS + ESS. Additionally, this process was repeated with one minipig using TSS in place of ESS (P004). In this case, TSS intensity ranged from 20 to 60 mA (0.3 and 1.8 MT) (Table 5).

Table 1.

Individual motor thresholds (in mA) for all muscles tested during ESS in rats

	LTA	LGAST	RTA	RGAST
R1	1.1	1.1	0.5	0.5
R2	0.5	0.4	0.5	0.4
R3	0.4	0.4	0.4	0.4
R4	0.4	0.4	0.4	0.4
R5	0.4	0.4	0.4	0.4
R6	0.2	0.2	0.4	0.4
R7	0.4	0.4	0.3	0.3
R8	0.5	0.5	0.6	0.6
R9	1.0	1.0	1.0	1.0
R10	0.3	0.2	0.5	0.5
R11	0.9	1.0	0.8	0.7

Table 3.

Individual motor thresholds (in mA) for all muscles tested during MCS in minipigs

	LVL	LSEM	LTC	LGAST	RVL	RSEM	RTC	RGAST
P001	110	100	100	100	90	90	90	90
P002	160	160	170	170	180	180	180	180
P003	170	170	170	170	180	180	180	180
P004	130	120	130	120	130	130	130	130

Table 2.

Individual motor thresholds (in mA) for all muscles tested during MCS in rats

	LTA	LGAST	RTA	RGAST
R1	0.5	0.6	1.0	0.8
R2	0.4	0.4	X	0.4
R3	0.8	0.8	0.8	0.8
R4	0.3	0.3	0.2	0.3
R5	0.4	0.4	0.4	0.4
R6	0.2	0.2	X	X
R7	0.3	0.2	0.4	0.4
R8	0.4	0.4	0.4	0.4
R9	X	X	0.4	0.4
R10	0.3	0.2	0.4	0.4
R11	0.3	0.3	0.3	0.3

X – missing data

Table 4.

Individual motor thresholds (in mA) for all muscles tested during ESS in minipigs

	LVL	LSEM	LTC	LGAST	RVL	RSEM	RTC	RGAST
P001	2.1	2.1	2.2	2.2	2.5	2.5	2.6	2.6
P002	1.6	1.7	1.7	1.7	2	2.1	2.1	2.1
P003	1.2	1.4	1.2	1.4	1.5	1.5	1.7	1.7
P004	0.4	0.6	0.4	0.6	0.8	1.0	1.0	1.0

Table 5.

Individual motor thresholds (in mA) for all muscles tested during TSS in minipigs

	LVL	LSEM	LTC	LGAST	RVL	RSEM	RTC	RGAST
P001	60	55	55	45	55	55	55	45

Data Processing and Statistical Analysis

Data were processed using LabChart Pro 8.1.13 software (ADInstruments, Australia). The average peak-to-peak amplitude of the evoked response was calculated for each muscle. Central conduction time (CCT), or the approximate time the cortically evoked motor potential reaches the lumbar enlargement, was determined for each animal at MT intensities. CCT was calculated by subtracting the average latency of SEMP across recorded hindlimb muscles from the average latency of CMEP [43–45]. To compare CTI across animals, the values were scaled to the CCT for the individual animal. Response amplitudes were scaled to the MT for the target muscle. Data are expressed as mean ± SD and the coefficient of variation (CV), an indicator of response consistency, was determined by the ratio between standard deviation and mean value [46]. All statistical analyses were carried out using OriginPro 2024 (Origin Lab Corporation, Northampton, MA, USA). A Shapiro-Wilk test was used to determine normality. For rats, the data were normally distributed as assessed by the Kolmogorov–Smirnov normality test, and a paired t-test was used. For pigs, the data were not normally distributed and therefore, they were examined using the non-parametric Kruskal-Wallis test. A Dunn’s multiple comparison test was then used for post-hoc analysis to determine which groups (i.e., for compound MEP: CTI/CCT, intensity/MT, and to determine transient effects: baseline CMEP control vs. post-intervention CMEP) were different. A p-value < 0.05 was considered significant (*p < 0.05, **p < 0.01, and ***p < 0.001).

Results

Rats

Cortical and spinal stimulations evoke motor responses on their own

In fully anesthetized adult rats, a train of three biphasic pulses with an interstimulus interval of 2 ms applied at 1 Hz to the epidural space of the left cortical hemisphere elicited CMEPs. CMEPs appeared with a stimulation intensity of 0.37 ± 0.16 mA (n = 11), which were first observed from contralateral hindlimb muscles in 6 animals, from ipsilateral muscles in 2 animals and from both legs simultaneously in 3 animals. CMEPs showed a mean peak amplitude of 0.21 ± 0.20 mV from TA (n = 6) and 0.37 ± 0.66 mV from GM (n = 5), while the mean latency was 7.85 ± 0.73 ms on TA (n = 6) and 7.22 ± 1.79 ms on GM (n = 5). Results were very consistent among animals as demonstrated by the low variability of latency values (CV_TA = 0.04 ± 0.02, n = 6; CV_GM = 0.04 ± 0.01, n = 5).

Then, ESS was applied to segments T12/L1 at 30 Hz. SMEPs appeared with a stimulation intensity of 0.41 ± 0.17 mA (n = 11) and were first recorded from GM muscles in 5 animals, and from both GM and TA muscles simultaneously in 6 animals. In the absence of any cortical stimulation, ESS at MT intensity induced SEMPs from the TAs and GMs of both sides, with a mean peak amplitude of 0.07 ± 0.05 mV (TA, n = 5) and 0.20 ± 0.27 mV (GM, n = 6), respectively. The latency of each SEMP was 6.82 ± 1.25 ms for TAs (n = 5) and 6.61 ± 0.81 ms for GMs (n = 6). Results were very consistent across animals, with a latency CV_TA = 0.07 ± 0.08 (n = 5) and CV_GM 0.08 ± 0.07 (n = 6). The CCT in rats was 1.66 ± 0.93 ms for TAs (n = 5) and 2.3 ± 1.01 ms for GMs (n = 5).

Combined cortical and spinal stimulation facilitates the motor output with distinct features

MCS at 1 Hz and ESS at 30 Hz were then combined in rats. After collecting baseline CMEPs, ESS was applied so that the first spinal pulse occurred at the same time as the first cortical stimulus. Due to the different frequencies of the two trains of stimulation (1 Hz cortical vs. 30 Hz spinal stimulation), the lag between a cortical stimulus and the closest following spinal pulse increased by 0.5 ms with each cortical pulse. When cortical and spinal stimulations were combined, compound MEP in the hindlimb muscles were facilitated (Fig. 3 A, B, G, and H). Modulation of leg motor output during combined stimulation followed two distinct trends based on the intensity of ESS as described below.

Fig 3. Combined stimulation generates three distinct outputs based on different stimulation intensity in rats.

Five consecutive traces from three sample rats (A, D, G) summarize the three different combinations of MCS and ESS of various intensity. For each experiment, time courses of MEP amplitude are plotted for TA and GM, respectively (B, E, H). All MEPs induced during combined stimulation are superimposed to display the consistency of the responses’ onset (C, F, I). The mean increase in MEP amplitude during combined stimulation, compared to values of pre-ESS cortical stimulation, is binned in 5 ms intervals between MCS and ESS pulses (n = 7, J). In another rat, ESS of increasing intensity is added to cortical stimulation (K). Responses from left GM during each phase of the protocol are superimposed to show the variability of MEPs onset (L).

Combined ESS + cortical stimulation at sub-motor threshold intensities facilitates cortically-induced motor responses

In five out of the eleven rats, MCS combined with an ESS intensity that was not sufficient to induce motor responses on its own, facilitated the amplitude of the resulting MEPs. The magnitude of modulation appeared to be inversely related to the intensity of MCS. Indeed, weaker cortical pulses of 0.38 ± 0.14 mA (n = 5), which per se induced only barely appreciable CMEPs (0.02 ± 0.02 mV, n = 5), were strongly facilitated by the presence of ESS. Combined stimulation induced higher MEPs from TA and GM, which reached an average value of 554 ± 268 % (p = 0.016, paired t-test; n = 5) of pre-ESS control. An example of traces in response to single or combined sub-MT pulses applied to both, the motor cortex and spinal cord, is illustrated in Fig. 3 A. Time courses for left TA and GM of the same rat show that, despite the lack of CMEPs during sub-MT MCS alone, compound MEPs emerged once sub-MT spinal stimuli were added (Fig. 3 B). Superimposed sample traces report the consistency of latencies, yet with some amplitude variability, of motor responses evoked by all consecutive pulses (Fig. 3 C).

Moreover, in a subgroup of four animals, MCS at MT intensity, viz. the ones that alone induced CMEPs, were not modulated by sub-MT ESS as confirmed by the mean amplitude of MEPs that was 116 ± 7 % of pre-ESS control (p = 0.234, paired t-test; n = 4) throughout the entire protocol of MCS + ESS. Specifically, when sub-MT ESS was combined with sub-MT or 1 MT MCS, the latency of MEPs, calculated from the first pulse in the triplet during MCS, was 7.57 ± 0.85 ms for TA (n = 6) and 8.41 ± 1.01 for GM (n = 5). Consistency of latencies was confirmed by low CVs, which were 0.04 ± 0.02 (CV_TA, n = 6) and 0.05 ± 0.02 (CV_GM, n = 5).

An example of the dynamic of CMEP, SEMP, and compound MEP during MT MCS, sub-MT ESS, and their combination, respectively, is reported in Fig. 3 D. Time courses for left TA and GM muscles of the same animal show that the amplitude of CMEPs was slightly potentiated by concurrent sub-MT ESS (Fig. 3 E). Superimposed sample traces report the consistent latencies of MEPs throughout combined stimulation, yet with a distinct amplitude variability of motor responses evoked by all consecutive pulses (Fig. 3 E).

Combined threshold ESS + cortical stimulation facilitates spinally-induced motor responses

In all animals, facilitation of compound MEPs during combined MCS and ESS occurred when the intensity of the latter was raised to the MT level. The mean amplitude of MEPs showed a trend toward increase to 260 ± 71 % (TA; p = 0.050, paired t-test; n = 5) and 247 ± 166 % (GM; p = 0.090, paired t-test; n = 6) of pre-ESS control for the entire delivery of combined stimuli. However, unlike sub-MT ESS, a stronger ESS modified the latency of facilitated MEPs, which became independent of cortical stimulation, as evidenced by the increased variability in latencies from the preceding cortical pulses, i.e., 0.29 ± 0.08 (CV_TA, n = 5) and 0.28 ± 0.08 (CV_GM, n = 6). Indeed, the facilitated MEPs followed the pattern of ESS, maintaining a consistent latency from the preceding spinal pulses of 5.98 ± 0.79 ms (TA, n = 5) and 5.57 ± 0.96 ms (GM, n = 6).

Sample traces for cortical pulses combined with spinal pulses at threshold intensity are displayed in Fig. 3 G. Time courses for left TA and GM muscles of the same rat indicate that the amplitude of CMEPs was potentiated by the concurrent threshold ESS (Fig. 3 H). Superimposed sample traces demonstrate that compound MEPs become staggered as they move rightward, to follow the increasing lag between cortical pulses and each consecutive spinal stimulus during continuous combined stimulation (Fig. 3 I).

Comparison of facilitated MEPs during combined stimulation with ESS at both sub-MT vs. 1 MT intensities (Fig. 3 C and I) revealed that weaker ESS resulted in facilitated MEPs that resembled CMEPs, while stronger ESS generated responses similar to SEMPs. These results suggest a distinct role for ESS intensity in determining the nature of responses facilitated by combined stimulation. To corroborate this observation, in another rat, cortical stimulation was combined with serial increases in ESS intensity (Fig. 3 K). The MCS with triplets of pulses at 0.2 mA (0.50 MT) evoked CMEPs from left TA of 0.006 ± 0.010 mV mean amplitude and 8.49 ± 0.40 ms mean latency (CV = 0.05; Fig. 3 L). Adding ESS at 0.1 mA (0.25 MT) did not affect the mean amplitude of MEPs (0.006 ± 0.009 mV) nor their mean latency (9.33 ± 0.59 ms, CV = 0.06), which remained close to pre-ESS CMEPs. However, increasing ESS to 0.2 mA (0.50 MT) or 0.3 mA (0.75 MT) resulted in a pronounced facilitation of the compound MEPs to 296 % and 426 %, respectively. A further increase of spinal stimulation to 0.4 mA (1 MT), elicited 10 higher MEPs (483 %) out of the 85 combined pulses supplied. Potentiated MEPs were scattered within a range that were paired with the preceding spinal pulses followed by cortical triplets and have been quantified by the wider distribution of CMEP latencies (12.84 ± 2.36 ms, CV = 0.18). However, decreasing the stimulation intensity to 0.3 mA (0.75 MT) consistently reduced MEP latencies to a mean value (10.32 ± 0.70 ms, CV = 0.06), which is close to the CMEPs elicited in the same rat by MCS alone, in the absence of ESS. In this experiment, the combined stimulation protocol was concluded by turning off ESS and continuing MCS alone for 4.8 min. Post-ESS CMEPs consistently occurred at 9.67 ± 0.48 ms from cortical stimulation (CV = 0.05). Interestingly, their amplitude was increased to 383 % on average when compared to initial cortical stimulation before ESS, showing postponed potentiating effects after the protocol of combined stimulation, that were recorded until 42 sec from the end of the combined stimulation protocol.

To ascertain whether the facilitation of MEP amplitude relies on an optimal timing between the two combined inputs, data were pooled from seven animals in which sub-MT ESS combined with both sub-MT and 1 MT MCS, respectively, was prolonged for up to 60 s. The resulting average peak-to-peak amplitude of MEPs was then plotted against the lag between cortical and spinal stimuli. While a zero value on the x-axis corresponds to the synchronous delivery of two pulses, positive values state a delay of ESS, and negative values refer to spinal pulses occurring before MCS (Fig. 3 J). The highest increases in MEP amplitude were obtained when ESS followed MCS at an interval of 4 to 5 ms and resulted in a facilitation of 621 ± 78 % (p = 0.010, Friedman repeated measures analysis of variance on ranks followed by multiple comparisons versus control group with Dunn’s method; n = 7) compared to the control pre-ESS CMEPs.

Supra-motor threshold ESS + cortical stimulation strongly reduces evoked muscle responses that reappear higher than control at the end of ESS

In a subgroup of four animals, combined stimulation revealed that high intensity ESS (0.54 ± 0.38 mA) largely reduced the MEPs amplitude, with complete abolishment of MEPs in one rat. In a sample experiment, the continuous recording and related time course of MEP amplitude exemplify the effect of combined stimulation with high-intensity ESS (Fig. 4 A – B). Before the combined stimulation protocol, MCS alone at threshold intensity (0.6 mA) elicited CMEPs from the left TA muscle, which were slightly increased in amplitude during the addition of sub-MT ESS pulses at 0.1 mA (0.5 MT; 109 %) and 0.2 mA (1 MT; 111%). The further increase of ESS intensity to 0.3 mA (1.5 MT) maintained MEPs amplitude unchanged (104 %) at first, but then was accompanied by a suppression of the MEPs amplitude to 14 % of pre-ESS control after 23 s of continuous combined stimulation.

Figure 4. Above-Threshold ESS largely suppresses MEP amplitude in rats.

A continuous EMG recording from the left TA exemplifies the test where MCS is combined with ESS of increasing intensity (A). The time course of MEP amplitude visualizes MEP facilitation by ESS both at sub-MT and 1 MT intensities, while the compound MEPs are reduced when the ESS intensity reaches 1.5 MT (B). Note the persistent facilitation of post-intervention CMEPs after ESS is stopped.

Interestingly, following a three-min long combined stimulation with increasing ESS intensity, MCS without ESS generated higher CMEPs (172 %) compared to those at the baseline. The average amplitude of CMEPs at the end of combined stimulation was 193 ± 106 % of pre-ESS control (p = 0.016, paired t-test; n = 7) and it was recorded until 2.38 ± 1.63 min after the end of combined stimulation.

These observations demonstrate that the intensity of ESS plays a critical role in determining not only the latency, but also the amplitude of motor responses evoked in the presence of combined stimulation.

Yucatan minipigs

From a translational perspective, protocols initially developed in rats were successfully replicated in a larger animal model with similarities to the human anatomical organization [47], utilizing a clinical-grade paddle for ESS. In anesthetized minipigs, MCS produced CMEPs in all recorded hindlimb muscles, with the initial response occurring at the average intensity of 93 mA ± 14.8 mA (n = 4). ESS produced SEMP at the average intensity of 1.58 ± 0.60 mA (n = 4). Using the measured CMEP and SEMP latencies, the CCT (n = 4) was determined to be 18.3 ms ± 3.7 ms for the proximal muscles (VL, SEM) and 18.7 ± 3.5 ms for the distal muscles (TC, GL).

Combined effects of MCS + ESS

The facilitatory effects of combined MCS + ESS on the compound MEPs over a selected time series are depicted in Fig. 5 for a representative pig, demonstrating that the facilitation of MEP is dependent on both the ESS intensity and timing (e.g., CTI) in relation to MCS (Fig 5, left). Due to the relatively extensive rostrocaudal distribution of motor pools innervating proximal and distal hindlimb muscles within the lumbosacral enlargement of the pig spinal cord, it was not feasible to activate all motor pools simultaneously at a single motor threshold intensity. As a result, some muscles reached motor threshold while others remained subthreshold. To account for this, we titrated the stimulation intensity such that most muscles were engaged, though some may have been stimulated slightly above their individual thresholds (e.g., one muscle at 1.0 MT, another at 1.2 MT; see Fig. 5) (Table 4). This approach allowed us to systematically examine the effects of combined MCS + ESS conditioning across a physiologically relevant range. It can also be observed that post-intervention CMEPs are more pronounced in some muscles compared to the baseline CMEPs.

Figure 5. Time-series data in minipigs.

Data from left hindlimb muscles demonstrate the effects of combined MCS + ESS as both the CTI and ESS intensity are increased. Facilitation can be seen in all muscles as the ESS intensity increases to the MT and above. Once ESS intensity reached the 1 MT level (right), facilitation of compound MEPs occurred, with the largest responses seen at the 18 ms CTI. Arrows indicate the timing when baseline and post-intervention MCS at sub- or near-MT intensity were delivered.

The faciliatory effects of MCS + ESS for two different ESS intensities is illustrated in Fig 6. The compound MEPs are shown using the assigned muscle color when the CTI is approximately equal to the CCT (Fig. 6B, D). During the experiment, both the CTI and the intensity of ESS were incrementally increased to investigate the effects of time and ESS intensity on the compound MEPs. When ESS was delivered at an intensity of 0.4 mA, mostly at a sub-MT level of activation of underlying motoneurons (Fig. 6A), the modulating effects became evident as shown in Fig. 6B. When the ESS intensity was increased to 0.6 mA, approaching the MT level of motoneuron activation (Fig. 6C), the compound MEPs were facilitated to a greater degree, as demonstrated in Fig. 6D. As intensity increased, both the variance of the CMEP latency and the amount of facilitation observed increased (n = 4), as shown in Fig. 6E–F.

Figure 6. Effects of combined MCS + ESS on the compound MEPs in minipigs.

(A) With ESS delivered at an intensity of 0.4 mA, which was sub-threshold for most muscles, (B) the facilitation of compound MEPs became more pronounced as the CTI approached the CCT. (C) When ESS was increased to 0.6 mA, near the MT for most muscles, (D) the facilitation of compound MEPs was even greater (note the difference in scaling). As ESS intensity increased from sub-MT to near-MT, the (E) variance of the response also increased as did the (F) percent of facilitation with respect to control CMEP, where the left VL is not shown due to the level of facilitation of 270x and 790x for 0.4 and 0.6 mA, respectively. The data shown is from a representative animal (P002). The dot plots demonstrate (G) the effect of CTI on the compound MEP for each pig (n = 4). The dashed vertical lines mark the average CCT for the four animals. To limit comparisons, the CTI were scaled to CCT so that comparisons could be made across animals and (H) binned into seven different ranges and comparisons were made to baseline CMEPs. Similarly, to determine the effect of ESS intensity, the responses were binned into up to six groups (I) and compared to the baseline CMEPs (significance: *p < 0.05, **p < 0.01, and ***p < 0.001, n = 4). Panels (B) and (D) display CMEP waveforms recorded at different CTIs – with increasing gradients of gray represent progressively longer CTIs, while the waveform corresponding to the CCT is shown in color.

Figure 6G demonstrates the changes in the facilitation of the compound MEPs with respect to CTI for the grouped data (n = 4) by individual muscle. Because the data are not scaled to CCT, the CTI that resulted in the largest facilitation varied between animals, as did the level of facilitation observed. To determine the timing effect with respect to CCT, CTI were then scaled by CCT and binned into seven different ranges as shown in Fig. 6H. We determined that the amplitude of the baseline CMEP and compound MEP were different in the recorded muscles (LVL X²(7) = 139.2, LSEM X²(7) = 124.8, LTC X²(7) = 155.8, LGL X²(7) = 131.1, RVL X²(7) = 70.6, RSEM X²(7) = 95.4, RTC X²(7) = 79.9, RGL X²(7) = 83.7; Dunn’s test, p < 0.001, n = 4). Once scaled, the data demonstrate that when using a CTI approximately equal to CCT, facilitation occurs across muscles (Dunn’s test, p < 0.001, n = 4).

Further, we systematically explored the effect of ESS intensity on the facilitation of CMEP (Fig. 6I). ESS intensity was normalized to the MT and binned into six different groups. The normalized ESS intensity was found to affect the compound MEP when compared to the baseline CMEP for all muscles (LVL X²(5) = 44.7, LSEM X²(5) = 96.5, LTC X²(5) = 49.0, LGL X²(5) = 67.4, RVL X²(5) = 44.3, RSEM X²(5) = 80.1, RTC X²(4) = 68.9, RGL X²(4) = 15.1; Dunn’s test, p < 0.001, n = 4). The greatest facilitation was observed when ESS intensity was within 0.8x to 1x MT range, showing a trend in LVL (p <0.05), and clearer differences in LTC, RVL, RGL (Dunn’s test, p < 0.01) and LSEM, LGL (Dunn’s test, p < 0.001, n = 4), as well as in the 0.6x to 0.8x MT range for RSEM and RTC (Dunn’s test, p < 0.001, n = 4).

The transient effects of combined MCS + ESS on the post-intervention CMEP across muscles are depicted in Fig. 7. to allow comparison of post-intervention CMEPs across animals with varying response durations. Interestingly, when comparing the post-intervention CMEP (i.e., during MCS only) to the baseline control, a lasting potentiation of CMEPs was observed after the ESS stopped (LVL X²(6) = 64.3, LSEM X²(6) = 60.8, LTC X²(6) = 67.5, LGL X²(6) = 60.9, RVL X²(6) = 56.5, RSEM X²(6) = 59.7, RTC X²(6) = 60.1, RGL X²(6) = 63.6; Dunn’s test, p < 0.001, n = 4).

Figure 7. Transient effects of the combined MCS + ESS on the post-intervention CMEPs in minipigs.

Post-intervention CMEP were normalized to illustrate the lasting potentiation following the combined MCS + ESS intervention. Time windows (1–7, 8–14, 15–21, 22–28, 29–36, and 37–39 seconds) are grouped and displayed using different color gradients. Scatter plots with individual symbols represent values for each animal and each muscle at each time point, and mean values are overlaid to indicate group trends.

Combined effects of MCS + TSS

An overview of the MCS + TSS intervention in one pig (P004) is shown in Fig. 8A–D. Similar to ESS, the administration of TSS during MCS resulted in faciliatory effects on the compound MEP, as shown in Fig. 8C using exemplary traces. Changes in facilitation as a function of CTI and amplitude were also examined independently (Fig. 8E–G). Individual MTs (in mA) for all muscles identified during TSS are shown in Table 5.

Figure 8. Effects of combined MCS + TSS on the compound MEPs in one minipig (P004).

TSS was applied (A) to the skin over the L4 and L5 vertebrae using a single monophasic 500 μs pulse. Prior to the MCS + TSS intervention, TSS intensity was systematically increased (B) to determine the MT using SEMPs from four hindlimb muscles of each leg. During the MCS + TSS combination, (C) facilitation of the compound MEP was pronounced as the timing between MCS and TSS (i.e., CTI) approached the CCT. TSS intensity (D) was set to 40 mA, which was sub-MT for all recorded muscles. The dot plots demonstrate (E) the effect of CTI on the CMEP at different TSS intensities (see individual traces). The dashed vertical lines mark the CCT. To limit comparisons, the CTI were scaled to CCT so that comparisons could be made across test trials and (F) binned into seven different ranges. Statistical comparisons were made to the baseline CMEP. To determine the effect of the TSS intensity, the responses were binned into five groups (G) and compared to the baseline CMEP (significance: *p < 0.05, **p < 0.01, and ***p < 0.001, n = 1, comparisons = 4 per limb).

We determined that TSS resulted in facilitation of the compound MEPs across all recorded muscles depending on the CTI (LVL X²(7) = 61.0, LSEM X²(7) = 116.4, LTC X²(7) = 119.5, LGL X²(7) = 85.2, RVL X²(7) = 57.8, RSEM X²(7) = 60.6, RTC X²(7) = 101.6, RGL X²(7) = 61.0; Dunn’s test, p < 0.001, n = 1) (Fig. 8E). The greatest facilitation of the compound MEP was observed across muscles when CTI was close to CCT, with a range from 0.8–1.2 CTI/CCT (Dunn’s test, p < 0.001) (Fig. 8F). Although observed in a single animal, these findings suggest that TSS has a larger temporal range of faciliatory effects (SEM/GL: 0.4 to >1.6) on the CMEPs, when compared to the ESS experiments (0.6–1.4 CTI/CCT).

To assess the effects of TSS intensity on the compound MEP during MCS + TSS, data were binned into five intensity ranges, as seen in Fig. 8G. We found that TSS intensity influenced the CMEP across recorded muscles (LVL X²(3) = 60.4, LSEM X²(3) = 44.8, LTC X²(3) = 28.2, LGL X²(4) = 89.7, RVL X²(3) = 36.4, RSEM X²(4) = 128.2, RTC X²(4) = 29.9, RGL X²(4) = 102.4, Dunn’s test, p < 0.001, n = 1). With respect to the number of muscles that were facilitated, TSS of 0.8–1 MT intensity facilitated the compound MEP in every muscle apart from RVL (LVL, RTC: p < 0.01; LSEM, LTC, LGL, RSEM, RGL: Dunn’s test, p < 0.001, n = 1).

Discussion

The present study investigated the interaction between electrically induced descending volleys and spinal inputs to hindlimb muscles in fully anesthetized intact rats and Yucatan minipigs. While other groups have investigated similar paired stimulation paradigms [43, 44], we have confirmed these findings and extended these results to include the combination of lumbar ESS/TSS with cortical stimulation to facilitate MEPs from hindlimb muscles. Furthermore, we have validated a large animal model, allowing greater translation of the data available from neuromodulation studies in rats to minipigs, where clinical-grade systems can be utilized and evaluated [36]. Our findings elucidate a unique interplay in the timing and intensity of paired stimulation, showcasing a peculiar interplay between supraspinal and spinal sensorimotor networks.

Combined cortical and spinal input

Interestingly, we observed a dynamic interplay and hierarchical order between cortical stimulation first and spinal stimulation after, to facilitate hindlimb motor function. The present work identified novel findings on the dependence of compound MEP magnitude on the intensity of modulation. While this intensity effect may be less prominent for MCS, varying the intensity of ESS can introduce inconsistencies in the degree of facilitation of resultant MEPs, as different spinal neurons and networks are activated to varying extent. In the spinal cord, different stimulation intensities appear to engage distinct circuits: lower intensities may preferentially facilitate afferent inputs, whereas higher intensities have a direct effect on premotor interneurons, motoneurons, or motor axons [48]. Consequently, compound MEP may vary depending on the activation of these different neural structures. We observed changes in the compound MEP onset, indicating a “transition” from modulated CMEPs in the presence of weak ESS toward what could be determined as modulated SEMP, when ESS was supplied at MT intensity. Although the potentiation of compound MEPs by paired stimulation with sub-MT ESS can be explained as the result of an increased motor pool excitability [43], this view is challenged by experiments with stronger intensity ESS, as the one adopted in the present study. We propose that the dynamics of combined cortical and spinal stimulations entail a balanced “competition” between two sources of motor pool activation. A hierarchical privilege of a higher-level input directing the tempo of compound MEPs is confirmed by the augmentation of CMEPs in the presence of weak ESS. Furthermore, during cortical pulses with an intensity too weak to elicit any CMEPs, the addition of sub-MT or near-MT ESS resulted in pronounced facilitation of the CMEPs. This mechanism might explain the recovery of volitional motor control during lumbar sub-MT ESS in a person with motor-complete SCI [25, 49]. Similarly, this mechanism can explain the effects of enhanced descending control in the presence of sub-MT or near-MT ESS applied to the cervical spinal cord, evident in individuals with chronic cerebral stroke [50]. However, the balance in this hierarchical relationship can be altered when spinal networks are further enhanced by depolarizing spinal stimulation at MT intensity. In such conditions, the increase in SEMPs due to the modulation provided by cortical inputs indicates the dominant role of spinal stimuli in the interplay between supraspinal and spinal networks. This mechanism may underlie the volitional conditioning of spinal reflexes, as described by Wolpaw et al. [51]. It is worth noting that the lack of modulatory effects when spinal stimulation is applied at low intensities relative to the MT may help explain the discrepancies observed in clinical outcomes of studies aimed at harnessing the modulation effects of spinal stimulation, particularly concerning TSS [52–56]. Higher TSS intensities can sometimes be intolerable for patients, and not all individuals can withstand TSS delivered at near-MT levels. In contrast, abolition of compound MEPs in the presence of supra-MT ESS may prompt further investigation into the therapeutic effectiveness of certain spinal stimulation techniques aimed at promoting voluntary motor control [57]. If spinal stimulation intensity becomes excessively high, leading to supra-MT SEMPs, it could potentially affect the desired motor responses and limit the benefits of these therapies. These observations are critical in understanding descending pathways’ functioning following their disruption during various neurological conditions, as they highlight the amplification of the residual pathway’s connection during spinal stimulation interventions. Furthermore, this insight highlights the importance of not viewing spinal stimulation as a standalone method for enhancing motor output. Instead, it should be integrated with other neuromodulation techniques and therapeutic approaches that target different levels of the neuroaxis. Combining spinal stimulation with physical rehabilitation and/or cortical stimulation can create a more holistic and effective intervention strategy.

Temporal modulation of compound responses during combinatory stimulation of motor cortex and spinal cord

Temporal facilitation of the compound MEP was observed in all recorded hindlimb muscles, with remarkable increases in magnitude of over 500% in rats and up to 400% in minipigs. This modulation can be interpreted in two ways. Firstly, the facilitation could result from increased CMEPs in the presence of sub-MT spinal stimulation, as the onset of the CMEPs remained unchanged. At the same time, their magnitude varied based on the timing between cortical and spinal pulses. A second alternative explanation is that the increase could be due to SEMP enhancement during MT spinal stimulation in the presence of MCS, as evidenced by the SEMPs’ onset shift. Consequently, depending on the scenario, the compound response waveforms more closely resemble CMEPs or SEMPs. It has been suggested previously that corticospinal excitation may induce a distinct level of depolarization of motoneurons, enabling them to reach the firing threshold during spinal stimulation [18]. However, evidence of monosynaptic corticomotoneuronal connections is largely absent in rats and minipigs [47], with significant direct projections considered to be unique to primates and humans [58]. Nevertheless, descending pathways originating from the motor cortex, with relay neurons in the basal ganglia [59], brainstem [60], or cerebellum [61], can facilitate the pairing and temporal summation of cortical and spinal pulses. Therefore, the key to temporal facilitation of the compound potentials should lie in the synchronous arrival of supraspinal and spinal inputs at the motoneuron pools. Assuming monosynaptic corticomotoneuronal projections [62], the delay between cortical and spinal pulses leading to peak facilitation of the compound MEP is expected to correspond to the CCT. However, in minipigs, facilitation of the compound MEPs occurred within an extensive window ranging between 0.6x to 1.4x CCT while in rats around 2x CCT. This wide time window can be attributed to several factors. Firstly, the decay time of motoneuron membrane depolarization may contribute to this phenomenon. However, afferent inputs are likely to increase motoneuron membrane voltage for only a brief period, typically lasting a few milliseconds [18, 63]. In contrast, corticospinal volleys may depolarize motoneurons for a longer duration, as indicated in human studies, where sustained facilitation has been observed over intervals of 13 to 30 ms [18, 63–65]. Additionally, spinal interneuronal network connections could be depolarized for extended periods when engaged by spinal stimulation [66, 67]. However, at the spinal level, the complex interaction of agonist and antagonist motoneurons, which can be summarized as the function of central pattern generators, suggests that more complex patterns of activation within various motor pools would be expected during the depolarization of the interneuronal networks. Despite this complexity, the consistent timing and simultaneous facilitation of both agonist and antagonist motoneurons make the hypothesis of facilitation at this level less likely. Lastly, the observed wide window of the prolonged effects of paired supraspinal and spinal stimulation may be attributed to polysynaptic descending pathways synapsing on spinal motoneurons. Our experimental protocol does not allow us to distinguish between the proposed mechanisms, rule out polysynaptic transmission, or even narrow down the acting descending transmission to purely corticospinal pathways. Given the timing of the facilitatory effects, the ascending influences from spinal stimulation reaching the brain and potentiating supraspinal sensorimotor networks seem unlikely. Nevertheless, this cannot be ruled out as a factor in the late modulation of compound MEPs. These considerations underscore the need for further research to elucidate the precise mechanisms underlying the observed phenomenon. Interestingly, the facilitation of MEPs lasts even when ESS is turned off, suggesting temporary activity-associated plasticity after combined stimulation, possibly due to the accumulation of neurotransmitter/molecular factors released by prolonged combined stimulation [68, 69].

Enhanced motor outcomes with ESS and TSS in clinical applications

Combined brain and spinal neuromodulation has been explored to enhance descending signals from the brain and brainstem, with the goal of restoring neurological function in individuals with SCI [70]. A randomized clinical trial evaluated paired noninvasive stimulation of the brain and spinal cord, where TSS and TMS, delivered during assisted stepping in a robotic gait orthosis, increased phase-dependent motor output in specific leg muscles in ten individuals with chronic incomplete or complete SCI [71, 72].

The current study addressed the mechanisms by which ESS and TSS facilitate motor output, without making immediate implications for existing clinical practices in spinal stimulation. Our data were collected from an intact central nervous system in a fully anesthetized animal model, which retains the anatomical and neurophysiological differences between quadrupeds and humans at both cortical and spinal levels. Nonetheless, the selective facilitation of weak descending input provides valuable insights into the effects of ESS and TSS on the under-threshold descending drive, which is designed to mimic discomplete spinal cord injuries [26, 33, 34]. In such injuries, anatomically intact descending fibers spared by the lesion are unable to activate downstream spinal targets, hindering volitional muscle recruitment. Our study confirms that spinal cord stimulation should be applied at near-MT to MT intensities to optimally enable motor functions. Regardless of the magnitude of cortical input, spinal stimulation at significantly higher than SEMPs MT would likely interfere with ongoing descending motor control.

Translating our findings to clinical settings presents challenges. In clinical environments, cortical and spinal stimulations are not paired during ESS and TSS, as remaining descending drive and afferent inputs are constantly and randomly influencing the background activity. However, our study suggests that descending inputs originating from supraspinal sources, even when limited due to neurological injury, may play a critical, if not the most critical, role in enhancing motor responses during spinal stimulation. This modulatory strategy can enable supraspinal-spinal networks to generate coordinated and complex upper or lower limb movements even in the presence of constantly changing volleys of afferent inputs. In this context, it seems plausible to harness the supraspinal-spinal interaction without excessively focal spinal stimulation projected to individual motor pools. Manipulating the intensity of spinal stimulation to achieve near-MT activation of selected motor pools may be sufficient to enable volitional motor control after neurological disturbance.

Limitations of the study

Some potential limitations of our study should be noted. Firstly, it is important to acknowledge that all our experiments were conducted on animals under anesthesia, which can affect corticospinal excitability [73]. However, it is important to emphasize that continuous intraoperative monitoring of CMEPs under propofol anesthesia is a standard practice during spinal surgeries [73]. Secondly, the use of intact spinal cords in this study limits direct comparability to injured conditions. Future studies with larger sample sizes are needed to determine whether these spinal-cortical interactions are preserved in awake animals or following neurological injuries. Here, the selection of neurologically intact animals was driven by the necessity to rely on a consistent descending input to selectively titrate the recruitment of fibers from weak to full CMEPs, systematically testing the interaction between subthreshold cortical stimuli and graded spinal inputs. Additionally, our study did not specifically attempt to identify the recruitment of direct corticospinal pathways, such as through histological tracing [74], direct supraspinal recordings [75], or the selective activation of pyramidal motoneurons [76]. Our primary goal was to investigate the interactions of broad descending inputs, as recovery following SCI is primarily attributed to the complex reorganization of various descending tracts, rather than the selective activation of direct corticospinal pathways. Overall, while this acute laboratory model serves as a controlled physiological study with inherent limitations in direct clinical application, it provides valuable insights for future research focused on developing stimulation protocols for rehabilitation. Facilitation effects may depend on the test stimulus amplitude and could account for the variability observed in the degree of facilitation across muscles and animals

Summary

In summary, we report a significant increase in compound MEP, encompassing both CMEPs and SEMPs, when cortical stimulation was combined with spinal stimulation. The latencies of these compound potentials suggest reciprocal effects between cortical stimulation and spinally evoked potentials. This potentiation occurred over a wide range of time intervals between cortical and spinal pulses, surpassing calculated CCT, indicating a complex interaction between descending and spinal pathways. The magnitude of potentiated compound MEPs was greatest when spinal stimulation was delivered within a specific intensity range between sub-MT and MT, whereas ESS at too high an intensity nullified the effects of cortical inputs. Our findings reflect the concept of functional interactions between supraspinal and spinal networks. Specifically, we demonstrate that cortical motor stimulation can modulate the effects of spinal stimulation, with the intensity and timing of descending and spinal inputs being critical parameters in shaping the resulting motor output.

• Sub-motor motor cortex stimulation enhances hindlimb response to spinal stimulation

• Spinal stimulation intensity and motor facilitation showed an inverse relationship

• Facilitation is highest when spinal stimulation and central conduction time match

• Cortical and spinal stimulation cause a lasting facilitation of descending input

• Leveraging supraspinal-spinal interactions may replace targeted spinal stimulation

Data Availability

The data that support the findings of this study are available upon reasonable request.