The amplitude of the SSEP at popliteal region, L2-3, and at Th11-12 levels on the low extremities at each recording location is presented on Figure 1 E and C. SSEPs were not detected at Th8-9 and Cz-Fpz levels located above the SCI (Figure 1D, two uppermost traces).

Figure 2 demonstrates examples of the SEMP in m. rectus femoris (RF) and m. tibialis anterior (TA) obtained at different stimulation intensities (Figure 2A) with threshold, amplitude values, and latency of the SEMP (Figure 2B) during tSCS at Th9-10, Th10-11, Th11-12, Th12-L1, and L1-2 levels (obtained at 100 mA). The order of activation of different muscles was dependent on the rostrocaudal location of stimulating electrodes. The stimulation intensity required to reach the motor threshold was gradually decreased from Th9-10 and Th10-11 to L1-2 in proximal and in distal muscles (n = 6, p < 0.05). In both distal and proximal muscles, the maximal amplitude of SEMP was gradually increased from T9-10 level reaching the highest value at T12-L1 and then decreased at the L1-2 level (n = 6, p < 0.05) (Figure 2B). The SEMP average latencies for the distal muscles were 14.90 ± 0.50 ms for the TA, 15.28 ± 0.52 ms for the SOL, 10.53 ± 0.43 ms for the RF and 10.27 ± 0.11 ms for the MH (n = 5). The SEMP latency was compatible with the distance between the stimulation level and the muscle and was larger in TA and m. soleus (SOL) and shorter in proximal muscles

RF and medial hamstring (MH) (Figure 2B). The maximal amplitudes of SEMP for proximal muscles were significantly lower compared to distal muscles (n = 4, p < 0.05). Examples of SEMPs recorded with paired tSCS at Th11-12 level are presented on Figure 2C. It is evident that the SEMPs were depressed with paired spinal cord stimulation (see more method details in Supplementary Material), supporting the reflex nature of the observed responses (n = 6, p < 0.05) (Figure 2D).

The M-wave and the H-reflex were recorded in SOL muscle (Figure 2E). During the JM the amplitude of H-reflex increased to 106.02 ± 0.94% and 111.43 ± 1.84% from control 100% values for the right and left leg, respectively (n = 10, p < 0.05) (Figure 2F). Figure 2G demonstrates examples of changes in the amplitude of SEMP recorded from RF and TA muscles without and with the JM, tested in supine and upright positions during the spinal cord stimulation at Th12-L1 level. Amplitude of the SEMP during JM was significantly facilitated in the left TA to 153.17 ± 22.45% from control 100% values only in upright position (n = 4, p < 0.05) (Figures 2G,H, Upright). In other muscles, JM did not change SEMP for either right or left leg (Figure 2H). Amplitudes of SEMP during JM with respect to the control condition were to 103.91 ± 7.14% (supine), and to 108.14 ± 7.79% (upright); in left RF to 101.09 ± 7.90% and to 121.10 ± 13.35%; in right MH to 102.83 ± 2.73% and to 100.87 ± 1.95%; in left MH to 98.53 ± 18.39% and to 111.43 ± 14.76%; in right TA to 102.65 ± 3.04% and 104.51 ± 8.64; in left TA to 129.16 ± 14.15% (supine); in right SOL to 104.28 ± 11.09% and to 99.67 ± 1.55%; in left SOL to 108.45 ± 12.70% and to 97.46 ± 3.81%; from control 100% values. In addition, delayed motor response with great toe extension was repeatedly observed on the right and left leg during JM only in vertically suspended (Supplementary Video S1).

Figure 3A shows examples of SEMP in TA muscle, evoked by stimulation of Th11-12 in supine position (supine) and immediately following the first verticalization (upright) before (gray lines) and after 16 weeks of rehabilitation therapy (black lines). Changes in SEMP (amplitude and threshold) before and after rehabilitation therapy in the supine position presented on Figure 3C. Maximum values of monosynaptic SEMP component in distal muscles before rehabilitation therapy were significantly lower compared to the amplitudes of SEMP after rehabilitation therapy (n = 4, p < 0.05) (Figure 3C). After subject with SCI underwent rehabilitation therapy, the thresholds of SEMP significantly decreased in all muscles (n = 4, p < 0.05) (Figure 3C). Polysynaptic components of the SEMP were found in TA muscle after rehabilitation therapy particularly in upright position. The cumulative analyses of latencies of LR demonstrated that the latencies of the polysynaptic components had wider distribution after rehabilitation therapy in upright position and mostly in TA (Figure 3A, Upright, and Figure 3D).

Previous studies indicated that supraspinal influence may have different effect on motoneurons and interneurons (Sabatino et al., 1995) and, therefore, modulation of mono- and polysynaptic responses can be a sensitive assessment tool of the spinal cord circuitry functional state after SCI. Given the nature of the H-reflex and the reflex components of the SEMP, it can be expected that the JM contribute in activation of downstream effects on spinal neuronal circuit. JM (Jendrassik, 1883), was used as a reinforcement to study spinal cord evoked responses in control subjects and in subjects with SCI (Dimitrijevic et al., 1977). One of the possible mechanisms of the effect of JM on spinal cord excitability was related to reduction of segmental presynaptic inhibition (Zehr and Stein, 1999). Our results indicate that JM can alter the H-reflex in supine position and SEMP in upright positioning the subject with discomplete SCI. At the same time, SEMP were affected primary in the left TA. As opposed to the SOL H-reflex, which is a monosynaptic response in a single muscle, the SEMP evoked by tSCS is related to a complex spinal network. Thus, facilitation of SEMP during JM may reflect the results of complex intraspinal and intersegmental interaction compared to monosynaptic response related to a single motor pool during H-reflex.

The SEMP were previously studied applying tSCS in healthy subjects and in subjects with SCI, tested in various positions: supine (Minassian et al., 2007; Hofstoetter et al., 2018), upright, and during gait modulation (Minassian et al., 2015). It was demonstrated that in individuals with AIS-A and-B positional changes (supine vs. standing) can provide different modulation of SEMPs components (Sayenko et al., 2014). In this study, transition from supine to upright position facilitated the amplitude of SEMP components that could be related to changes in sensory information from mechanoreceptors affecting the spinal circuitry excitability (Harkema et al., 1997). Presynaptic inhibition of Ia afferents on the motoneuron is considered to be controlled by descending tracts and the level of presynaptic inhibition input in SCI subjects declines compared to control subjects, contributing to enhancement of spinal reflexes (Calancie et al., 1993). The body position could influence the activation of the afferent and efferent fibers by tSCS (Danner et al., 2016). In contrast to our results, Danner et al. (2016) showed that the thresholds of evoked responses in subjects with intact spinal cord were lowest in upright position and highest in the prone position (Danner et al., 2016). Variations in mono- and polysynaptic responses during tSCS can be related to motoneuronal excitability and also to complex convergence of sensory afferents on spinal reflex pathways (Schomburg, 1990; Sayenko et al., 2014). It was suggested that the afferent information can be integrated by spinal circuitry and result in elevation interneuronal excitability during standing (Harkema et al., 2011; Rejc et al., 2015). Also, Sayenko et al. (2019) demonstrated that tSCS can modulate the lumbosacral spinal networks to facilitate postural control after SCI (Sayenko et al., 2019). Therefore, the characteristics of SEMP cannot be attributed only to a certain motor pools related to the spinal cord circuitry, but rather to specific interplay of multiple peripheral sensory resources and related interneurons (Sayenko et al., 2014). These results indicate that positional changes can facilitate lumbosacral networks and increase the sensitivity of electrophysiological testing for the residual sub- functional connections after SCI.

Complete paralysis of the lower extremities and ability to stand and perform coordinated motor activity could be improved with epidural electrical stimulation (Harkema et al., 2011; Grahn et al., 2017; Gill et al., 2018; Wagner et al., 2018). Our results demonstrate that rehabilitation therapy can facilitate SEMP components, observed after 16-week rehabilitation program in supine and upright positions. It is noteworthy that facilitation of the late response in calf muscles was found in upright position. Similar to animal studies, initiation of rhythmic activity after SCI in human was associated with appearance of late responses (Minassian et al., 2004). As it has been shown earlier, the task-specific training with epidural SCS may reactivate previously silent neural circuits or promote plasticity (Harkema et al., 2011). In addition, tSCS, as well as epidural SCS, can modulate spinal circuitry in humans after SCI that enables sensory inputs to serve as a primary source of neural control of posture and balance (Sayenko et al., 2019). Decreased threshold and increased reflex excitation may be indicative for an increased spasticity. The presence of spasticity below the level of injury in patients with SCI could indicate that related motor pools are relatively preserved (Harris et al., 2007; Gorgey and Dudley, 2008), at the same time, characteristics of SEMP cannot be attributed only to the level of the spinal circuitry excitation and could be a consequence of interplay of multiple peripheral afferent signals and related interneurons (Sayenko et al., 2014). Findings of multiple studies suggest that spinal reflexes increase in patients with SCI and cannot be evaluated unambiguously. Particularly in this study, an increase of the motoneurons’ excitability was not clearly related to spasticity.