Sensorimotor integration in cranial muscles tested by short-and long-latency afferent inhibition

(cid:1) Short-(SAI) and long-latency (LAI) afferent inhibition result from cortical integration of sensory inputs and motor output. (cid:1) This work shows that SAI and LAI depend on type of sensory receptors, muscle function and cortical motor area. (cid:1) This ﬁnding has pathophysiological implications in motor disorders involving speciﬁc muscular districts, such as focal dystonia.


Introduction
The integration of sensory inputs with motor output plays a crucial role in the physiology of fine movement control and motor learning (Asanuma and Pavlides, 1997).Such sensorimotor interaction can be tested with paired pulse paradigms, where conditioning electrical stimuli given to a peripheral nerve are followed by transcranial magnetic stimulation (TMS) of the contralateral primary motor cortex (M1) at short intervals (short-afferent inhibition, SAI) (Classen et al., 2000;Tokimura et al., 2000;Pilurzi et al., 2013Pilurzi et al., , 2020) ) or longer intervals (long-afferent inhibition, LAI) (Chen et al., 1999;Classen et al., 2000;Pilurzi et al., 2020).
Although these protocols have been widely used to explore sensorimotor integration in physiological and neurological conditions, the neural pathways traversed by the afferent volley leading to SAI and LAI remain unclear (Turco et al., 2018).It has been suggested that SAI may result from the activation of direct thalamocortical projections to M1 via the cholinergic paramedian thalamic nuclei; alternatively, it has been proposed that afferents responsible for SAI relay through somatosensory cortex (S1), activate neuronal projections from S1 to M1, and recruit M1 interneurons that inhibit layer V pyramidal motor neurons (Turco et al., 2018).According to this view, a modulatory pathway for SAI resides in thalamocortical projections carry afferent input onto spiny stellate cells in layer IV of S1.These spiny stellate cells have excitatory synaptic connections on the apical dendrites of pyramidal neurons located in S1 layers II/III which project onto I3-wave generating interneurons in M1 layers II/III.The pyramidal neurons in S1 also project to layers V/VI in M1, potentially onto basket cells which are also located in these layers (Turco et al., 2018).Basket cells are GABAergic neurons that synapse onto the perisomatic region of pyramidal cells to hyperpolarize and shunt depolarization.Basket cells are modulated by cholinergic input.Therefore, SAI modulation on M1 output is the result of inhibitory shunting of the pyramidal neurons via basket cells and it is likely that SAI comes from direct thalamocortical projections to M1, potentially onto these basket cells, to regulate shunting of the pyramidal neuron and resultant afferent inhibition (Turco et al., 2018).
In contrast, LAI may be related to the widespread activation of various somatosensory cortical areas (Turco et al., 2018).Although, the pathway responsible for LAI remains unclear, several studies demonstrated that LAI has a cortical origin, since the peripheral stimulation has no effect on spinal reflexes at the timing corresponding to LAI (Chen et al., 1999;Turco et al., 2018).In particular, it has been showed that the contralateral S1, the posterior parietal cortex (PPC), and secondary somatosensory cortex (S2) cortices are activated during the first 200 ms after the peripheral stimulation (Allison et al., 1989(Allison et al., , 1992)).These cortical areas project to M1 and may thus be involved in mediating inhibition within M1 (Turco et al., 2018).However, LAI may also traverse the basal gangliathalamocortical loop as suggested by studies performed in patients affected by Parkinson diseases (Sailer et al., 2003).
Several factors influence SAI and LAI, including the intensities of nerve stimulation and of TMS, nerve composition and recorded muscle (Turco et al., 2018).Regarding the influence exerted by the muscle district on SAI and LAI, it has been shown that toe stimulation is able to facilitate rather than inhibit the motor output to both foot and upper arm muscles (Bikmullina et al., 2009).Another peculiar case is that of lower face muscles, namely the depressor anguli oris (DAO).Here, SAI depends on the stimulation of the trigeminal sensory afferents, while LAI depends on stimulation of the facial motor nerve (Pilurzi et al., 2013;2020).This evidence led to the assumption that SAI depends on the activity of low threshold, presumably cutaneous, afferents, whereas LAI requires activity in sensory afferents, activated by the muscle twitch evoked by electrical stimulation of the motor nerve (Pilurzi et al., 2013;2020).According to this view, cutaneous inputs may exert a paucisynaptic inhibitory effect (SAI) on face M1, while proprioceptive information is likely to target inhibitory polysynaptic circuits involved in LAI.However, it is unclear if these results are due to the peculiar features of proprioception in face muscles, which are devoid of conventional muscle spindles, or to the mixed trigeminal-facial innervation of face, or to a different role played by cutaneous and proprioceptive afferents in the origin of SAI and LAI.
In light of this background, it seems a worthy endeavor to investigate in detail sensory-motor integration in the cranial district, which differs from that in hand muscles in many respects, both anatomically and physiologically (Cattaneo and Pavesi, 2014;Ginatempo et al., 2019;2021;2022).To this aim we investigated SAI and LAI in two cranial muscles: the DAO and the anterior digastric muscle (AD), which have in common the peculiar feature of being devoid or poor in muscle spindles (Lennartsson, 1979;Cattaneo and Pavesi, 2014), but have different functions (frowning of the mouth and jaw opening, respectively) and a different innervation.The sensory and motor innervation of the AD is via the trigeminal nerve (Standring, 2008;Kim and Loukas, 2019), while the motor innervation of the DAO is provided by the facial nerve while its mechanosensory input is relayed by the facial nerve in the trigeminal system through a complex nervous anastomosis (Hwang et al., 2007;Cattaneo and Pavesi, 2014).This last feature of face muscles and the dependence of face LAI exclusively on the stimulation of facial nerve raises the question whether LAI depends mainly on the stimulation of muscle mechanosensors rather than cutaneous afferents.To test this hypothesis, we also investigated SAI and LAI in the upper trapezius (UT) muscle which has a distinct sensory and motor innervation, with the former travelling in spinal nerves (Ourieff et al., 2022) and the latter provided by the accessory nerve.

Participants
Experiments were conducted in twenty healthy volunteers (11 females and 9 males; mean age 28.69 ± 4.84 years), all righthanded according to the Oldfield inventory scale.All subjects gave their informed written consent to participate in the study, which was approved by the local ethical committee (Bioethics Committee of ASL.n. 1 -Sassari, ID 2075/CE/2014) and conducted in accordance with the Helsinki declaration.None of the subjects had a history of neurological or psychiatric diseases.Subjects sat in a comfortable chair and were asked to stay relaxed but alert during the experiments.

EMG
EMG was recorded, in separate experimental sessions, from the left DAO, AD and UT, using 9 mm diameter Ag-AgCl surface electrodes.For the DAO EMG recordings, the active electrode was placed at the midpoint between the angle of the mouth and the lower border of the mandible, with the reference electrode over the mandible border, 1 cm below the active electrode, and the ground electrode over the right forehead (Pilurzi et al., 2020).For the AD EMG recording, the active electrode was positioned over the anterior part of the muscle belly and the reference electrode about 2 cm anterior to it, adjacent to the inferior border of the mandible (Nordstrom et al., 1999).For the UT EMG recording, the active and reference electrodes were placed 3 cm apart over the UT and the ground on the sternum manubrium (Matthews et al., 2013).Unrectified EMG signals were recorded (D360 amplifier, Digitimer Ltd, Welwyn Garden City, UK), amplified (x1000), filtered (bandpass 3-3000 Hz), and sampled (5 kHz per channel; window frame length: 500 ms) using a 1401 power analog-todigital converter (Cambridge Electronic Design, Cambridge, UK) and Signal 5 software on a computer and stored for off-line analysis.

Electrical stimulation (ES)
Electrical stimulation (square-wave pulses of 0.2 ms duration) was applied in different experimental sessions to the left mentalis branch of the trigeminal nerve, to the left marginal branch of the facial nerve and to the left accessory nerve through a pair of cup electrodes (cathode distal) connected to a constant current stimulator (model DS7; Digitimer, Welwyn-Garden City, Herts, UK) (Fig. 1).Perceptual thresholds (PT) were calculated for each nerve tested.The intensity of the electrical stimulus was set as follows: in the AD experiment, at a value three times the PT of the subject for the trigeminal and facial nerve stimulation; in the DAO and UT experiments, at a value able to evoke a small stable compound muscle action potential (cMAP) in the respective left muscles.In a subgroup of subjects, facial and accessory nerves were also stimulated at subthreshold intensity for eliciting a cMAP in the respective muscles, to avoid the muscle twitch and thus ensure a cutaneous stimulation only.For the AD muscle, it was not possible to evoke a cMAP since its motor nerve originates from the mylohyoid nerve, that is a branch of the inferior alveolar nerve, which requires an invasive approach to be stimulated.

TMS
TMS was performed using a figure-of-eight shaped coil with external loop diameter of 7 cm connected to a Magstim 200 stimulator (Magstim Co., Whitland, Dyfed, UK).The optimal stimulation site in the right hemisphere, was carefully searched and then marked with a soft tip pen over the scalp, to maintain the same coil position throughout the experiments.For AD and DAO, the handle of the coil pointed posteriorly and laterally at approximately 30-45°to the interhemispheric line (Nordstrom et al., 1999;Pilurzi et al., 2013;2020).For UT the coil pointed backwards and laterally at 45°away from the midline (Matthews et al., 2013).The resting motor threshold (RMT) was taken as the lowest TMS intensity, expressed as percentage of the maximum stimulator output (MSO), that elicited, in the relaxed muscle, MEPs of 0.05 mV in at least 5 out of 10 consecutive trials (Rossini et al., 2015).The intensity of the test stimulus (TS) was 120 % of RMT.Peak-to-peak MEP amplitude was used as variable for the analysis.Signal 5 software was used for storing the traces and for off-line analysis of MEPs.

Experimental design
Three experiments were performed in separated sessions where SAI and LAI protocols were done.In the SAI protocol, stimulation of the peripheral nerve preceded the TMS pulse by 15, 20 and 25 ms.In the LAI protocol, stimulation of the peripheral nerve preceded the TMS pulse by 100, 180 and 200 m.Therefore, in the SAI blocks, TS alone and 15, 20 and 25 ms ISIs were tested and in the LAI blocks TS alone as well as 100, 180 and 200 ms ISIs were tested.ISIs for SAI and LAI were chosen based on previous studies that demonstrated the precence of SAI and LAI in the DAO muscle at these intervals (Pilurzi et al., 2013;2020).
The four blocks and all states (TS alone and ISIs) were randomized in each subject.Fifteen unconditioned MEPs and 15 conditioned responses for each ISI were recorded from the muscle of interest at rest.

Experiment 1. Effects of trigeminal and facial nerve stimulation on AD MEP in the SAI and LAI protocols
In all twenty subjects, the effects of trigeminal and facial nerve stimulation at 3xPT intensity on AD MEPs were compared in the SAI and LAI paradigms.Single pulse TMS of the right masticatory M1 was preceded by ES of the left trigeminal or facial nerves at various ISIs.The experiment was divided up into four blocks: trigeminal-SAI (tSAI), facial-SAI (fSAI), trigeminal-LAI (tLAI) and facial-LAI (fLAI).
3 Fig. 1.Position of the electrodes for electromyography (EMG) recordings from the depressor anguli oris (DAO), anterior digastric muscle (AD) and trapezius muscles (UT) and for the electrical stimulation of the trigeminal, facial, and accessory nerves.Panel A shows the position of EMG recording electrodes over the DAO: the active electrode (black) is placed at the midpoint between the angle of the mouth and the lower border of the mandible, the reference electrode (red) over the mandible border, 1 cm below the active electrode.Electrodes for stimulation of trigeminal and facial nerves are depicted as orange and violet circles, respectively.Panel B shows the position of EMG recording electrodes over the AD: the active electrode (black) was over the anterior part of the muscle belly and the reference electrode (red) about 2 cm anterior to it, adjacent to the inferior border of the mandible.Electrodes for stimulation of the trigeminal and facial nerve are depicted as orange and violet circles, respectively.For both DAO and AD EMG recordings, the ground electrode was placed over the right forehead.Panel C shows the position of EMG recording electrodes over the UT: the active (black) and reference (red) electrodes were placed 3 cm apart over the UT belly and the ground electrode (green) on the sternum manubrium.
F. Ginatempo, N. Loi, J.C. Rothwell et al. Clinical Neurophysiology 157 (2024) 15-24 2.5.2.Experiment 2. Effects of trigeminal and facial nerve stimulation on DAO MEP in the SAI and LAI protocols Sixteen out of the 20 subjects enrolled in experiment 1 participated in experiment 2. The effects of trigeminal and facial nerve stimulation on DAO MEPs were compared in the SAI and LAI paradigms.Single pulse TMS of the contralateral right face M1 was preceded by ES of the left trigeminal or facial nerves at various ISIs.The experiment was divided up into two four blocks: tSAI, fSAI, tLAI and fLAI.
In a subgroup of 10 subjects, SAI and LAI were tested using an intensity of facial nerve stimulation subthreshold for evoking a cMAP in the DAO.

Experiment 3. Effects of the accessory nerve stimulation on UT MEP in the SAI and LAI protocols
Thirteen out of the 20 subjects enrolled in experiment 1 participated in experiment 3. The effects of left accessory nerve stimulation at an intensity able to evoke a small stable cMAP in the left UT was investigated in the SAI and LAI paradigms.Single pulse TMS of the contralateral right M1 was preceded by ES of the left accessory nerves at various ISIs.The experiment was divided up into two blocks: accessory-SAI (aSAI) and accessory-LAI (aLAI).
In a subgroup of ten subjects, the aSAI and aLAI were tested using an intensity of accessory nerve stimulation subthreshold to evoke a cMAP in the left UT.

Statistical analysis
Statistical analysis was performed with SPSS 20 software (SPSS Inc, Chicago, IL, USA).Student's paired t-test, repeated measures (RM) analysis of variance (ANOVA) and planned post hoc t-test with Bonferroni correction for multiple comparison were used.Compound symmetry was evaluated with the Mauchly's test and the Greenhouse-Geisser correction was used when required.Significance was set for p values 0.05.Data are expressed as means ± standard deviation (SD).In all experiments amplitude of conditioned and unconditioned MEPs were analyzed.Raw amplitude and ratio were used as variables.
Experiments 1, 2 and 3: one-way RM-ANOVAs, using raw amplitude as variable with ISI as within subject factors was used separately for SAI and LAI protocols and for trigeminal and facial nerve stimulation (SAI: TS, 15, 20 and 25 ms; LAI: TS, 100, 180, 200 ms).In experiment 4 two separated one-way RM-ANOVAs were also performed based on the different ES intensities used (threshold and subthreshold intensity for cMAP).
Experiments 1 and 2: To compare the effect between muscles and nerve stimulations within the cranio-facial district, two-way RM-ANOVA was performed separately for each protocol (SAI and LAI) using ratio and the significant ISI, as variable, with muscle (AD and DAO) and nerve (trigeminal and facial) as within subject factors.
Experiments 2 and 3: To compare the effect of the intensity of the stimulation, two-way RM-ANOVA was performed separately for each protocol (SAI and LAI) using ratio as variable, with ISI (SAI: 15, 20 and 25 ms; LAI: 100, 180 and 200 ms) and intensity (subthreshold and suprathreshold for cMAP) as within subject factors.

Results
Neurophysiological parameters and measures collected in each experiment are reported in Table 1.

Experiment 1. Effects of trigeminal and facial nerve stimulation on AD MEP in the SAI and LAI protocol
In the AD, a clear SAI was detected following trigeminal but not facial nerve stimulation (Fig. 2A).In particular, the one-way RM-ANOVA of tSAI protocol showed a significant effect of ISI (F 1,19 = 4.629; p = 0.007) and the post-hoc analysis revealed a significant inhibition at 20 ms ISI (p = 0.006).On the other hand, significant effect of ISI (F 1,19 = 3.478; p = 0.040) was observed in the fSAI protocol, but the post-hoc test failed to detect any significant difference between the unconditioned and conditioned MEPs.
In the AD, a clear LAI was detected following trigeminal but not facial nerve stimulation (Fig. 2A).In particular, the one-way RM-ANOVA of tLAI protocol showed a significant effect of ISI (F 1,19 = 7.368; p = 0.007) and the post-hoc analysis revealed a significant inhibition at 100 ms ISI (p = 0.032).The one-way RM-ANOVA of the fLAI protocol showed a significant effect of ISI (F 1,19 = 4.088; p = 0.018), but the post-hoc failed to detect any significant differences between the unconditioned and conditioned MEPs.

Experiment 2. Effects of trigeminal and facial nerve stimulation on DAO MEP in the SAI and LAI protocols
In the DAO, a clear SAI was detected following trigeminal but not facial nerve stimulation (Fig. 2B).In particular, the one-way RM-ANOVA of tSAI protocol showed a significant effect of ISI (F 1,19 = 3.036; p = 0.039) and the post-hoc analysis revealed a significant inhibition at 20 ms ISI (p = 0.020).By contrast, the one-way RM-ANOVA of fSAI protocol showed a non-significant effect of ISI (F 1,19 = 0.796; p = 0.472).
In the DAO, a clear LAI was detected following facial but not trigeminal nerve stimulation (Fig. 2B).In particular, the one-way RM-ANOVA of tLAI protocol showed a non-significant effect of ISI (F 1,19 = 0.767; p = 0.493).By contrast, the one-way RM-ANOVA of fLAI protocol showed a significant effect of ISI (F 1,19 = 4.268; p = 0.020) and the post-hoc analysis revealed a significant inhibition at 100 ms ISI (p = 0.032).
When the fSAI and fLAI protocols were tested in the DAO using an intensity of facial nerve stimulation unable to elicit a clear cMAP in this muscle, no significant effects were detected (Fig. 3).In par-ticular, the one-way RM-ANOVA showed a non-significant effect of ISI in the fSAI protocol (F 1,9 = 0.306; p = 0.705) and in the fLAI protocol (F 1,9 = 1.076; p = 0.355).
About LAI, the statistical analysis demonstrated that it depended on the intensity of facial nerve stimulation used.In particular, the two-way RM-ANOVA showed a significant effect of intensity (F 1,9 = 6.179; p = 0.035) but a non-significant effect of ISI (F 2,18 = 1.287; p = 0.297) and interaction between factors (F 2,18 = 1.747; p = 0.216).

Comparison of SAI and LAI protocols in AD and DAO muscles following trigeminal and facial nerve stimulations
The statistical analysis was performed using the significant ISIs for SAI and LAI detected in the Experiments 1-2.In particular, SAI was observed at an ISI of 20 ms and LAI at an ISI of 100 ms in both AD and DAO muscles.
The statistical analysis of SAI demonstrated that in both muscles SAI depended only on trigeminal nerve stimulation.In particular, the two-way RM-ANOVA showed a significant effect of nerve (F 1,15 = 7.490; p = 0.015) but a non-significant effect of muscle (F 1,15 = 0.081; p = 0.781) and interaction between factors (F 1,15 = 0.083; p = 0.778).
In regard to LAI, the statistical analysis demonstrated that it depended on trigeminal nerve stimulation in the AD, and on facial nerve stimulation in the DAO.In particular, the two-way RM-ANOVA showed a non-significant effect of nerve (F 1,15 = 0.465; p = 0.506) and muscle (F 1,15 = 1.370; p = 0.260) but a significant interaction between factors (F 1,15 = 9.440; p = 0.008).The Bonferroni post-hoc analysis of interactions showed a significant difference between muscles following trigeminal nerve stimulation (p = 0.05).The boxplots report the raw amplitudes of the test MEP, obtained with a single pulse intensity of 120 % resting motor threshold, and of the conditioned MEPs obtained following the coupling of the accessory nerve and TMS stimulations at various interstimulus intervals (ISI).The accessory nerve was stimulated with an intensity able to evoke a small stable compound muscle action potential (cMAP) in the UT, and, in a separate session, at subthreshold intensity for eliciting a cMAP in the UT.The amplitude of UT MEPs was significantly reduced by accessory nerve stimulation only in the LAI protocol (100 ms ISI) and only with a stimulation intensity able to evoke a cMAP in the UT.The continuous line in the boxplot represents the median value, while the 'Â' symbol represents the mean value of the group.*p < 0.05.

Experiment 3. Effects of the accessory nerve stimulation on UT MEP in the SAI and LAI protocols
The stimulation of the accessory nerve at suprathreshold intensity to evoke a clear and stable cMAP in the ipsilateral UT produced no significant effects in the SAI paradigm, while a clear MEP inhibition was detected in the LAI paradigm (Fig. 4).In particular, the one-way RM-ANOVA of aSAI protocol showed a non-significant effect of ISI (F 1,12 = 0.451; p = 0.641).By contrast, the one-way RM-ANOVA of the aLAI protocol showed a significant effect of ISI (F 1,12 = 4.058; p = 0.05) and the post-hoc analysis revealed a significant inhibition at 100 ms ISI (p = 0.022).
When the aSAI and aLAI protocols were tested using an intensity of accessory nerve stimulation unable to elicit a clear cMAP in the UT, no significant effects were detected (Fig. 3) either in the aSAI and aLAI protocols.In particular, the one-way RM-ANOVA showed a non-significant effect of ISI in the aSAI protocol (F 1,9 = 0.428; p = 0.677).In the aLAI protocol, the one-way RM-ANOVA showed a significant effect of ISI (F 1,9 = 8.970; p = 0.002) but the post-hoc failed to detect any significant differences between unconditioned and conditioned MEPs.
In regard to LAI, the statistical analysis demonstrated that it depended on the intensity of accessory nerve stimulation used.In particular, the two-way RM-ANOVA showed a significant effect of intensity (F 1,9 = 6.515; p = 0.031) and ISI (F 2,18 = 16.634;p = 0.003) but a non-significant interaction between factors (F 2,18 = 0.187; p = 0.825).Post-Hoc analysis of the ISI effect showed a stronger inhibition at 100 ms versus 200 ms (p = 0.002) and 180 ms (p = 0.021).

Discussion
In the present study, sensorimotor integration was investigated in two cranial muscles, the AD and DAO, which belong to the stomatognathic and facial systems, respectively, and in UT, used as a model of cranio-cervical muscle.In the AD muscle, both SAI and LAI were observed following stimulation of trigeminal afferents.According to a previous study (Pilurzi et al 2020), in the DAO muscle, SAI was detected exclusively with the stimulation of trigeminal afferents, while LAI depended solely on the stimulation of the motor facial nerve.As novel finding, here we demonstrated that the latter was evocable only if the facial nerve was stimulated at an intensity suprathreshold for cMAP, suggesting that cutaneous afferents are not crucial for LAI.In the UT LAI, but not SAI, was detected following stimulation of the accessory nerve, again only at an intensity suprathreshold for a cMAP, confirming what observed in the DAO.

SAI and LAI in the AD
AD showed clear SAI and LAI following trigeminal nerve stimulation at an intensity of 3 Â PT.The digastric muscle is an important muscle in the neck, which pulls the mandible downward to open the jaw and elevates the hyoid bone for stabilization during swallowing, and it is also involved in chewing and speech (Sowman et al., 2009;Kim and Loukas 2019).The digastric muscle consists of an anterior and a posterior belly.The former is attached to the digastric fossa on the base of the mandible close to the midline and runs toward the hyoid bone; the latter is attached to the notch of the mastoid process of the temporal bone and also runs toward the hyoid bone (Standring, 2008).The anterior belly of the digastric muscle, or AD, is innervated by the mylohyoid nerve, a branch of the inferior alveolar nerve, which arises from the mandibular branch of the trigeminal nerve.The posterior belly of the digastric muscle is innervated by the facial nerve (Standring, 2008;Kim and Loukas, 2019).In contrast, the posterior belly of the digastric muscle and the DAO, the anterior belly of the digastric muscle both receive their sensory and motor supply from the trigeminal nerve, which could explain why in AD both SAI and LAI depend on trigeminal stimulation, but not on facial nerve stimulation.In addition, differently from the face muscles, AD has a bony insertion but similarly to them is devoid of visual feedback of movement trajectories.Another feature of the AD is the rarity of the absolute number of muscle spindles (Saverino et al., 2014), which is similar to that of the face muscles, where proprioception is the prerogative of Ruffini-like mechanoreceptors, whose afferents travel in the facial nerve, then cross into the trigeminal nerve at many distal anastomoses (Hwang et al., 2007;Cattaneo and Pavesi, 2014).
It is well known that nerve composition influences the range of ISIs over which SAI and LAI are evoked.In hand muscles, stimulation of the D2 digital nerve (DN) evokes SAI in the first dorsal interosseus muscle (FDI) over ISIs from 20 to 50 ms, whereas stimulation of a mixed nerve at the wrist produces SAI over a smaller range of ISIs from 18 to 28 ms (Turco et al., 2018;Tamburin et al., 2005;2002).Despite the different ISI ranges both cutaneous and mixed nerves produce the same depth of SAI (Turco et al., 2018;Tamburin et al., 2005).Similarly, LAI is evoked in the FDI by mixed nerve stimulation at the wrist at ISIs from 200 to 1000 ms, and following DN stimulation at ISIs from 200 to 600 ms, with no differences in amplitude between stimulus types (Chen et al., 1999;Turco et al., 2017;2018).In line with data from FDI, our results in AD indicate that activation of cutaneous afferents alone is sufficient to evoke both SAI and LAI.This is suggested not only by the low intensity of the stimulus used (3xPT), which did not elicit any muscle twitch in the AD, but also because muscle spindles are rare or absent in AD (Saverino et al., 2014), making a contribution of proprioceptive afferents to SAI and LAI unlikely.

SAI and LAI in the DAO muscle
In line with previous papers, SAI in the DAO was evoked only following trigeminal nerve stimulation, while LAI was observed only after facial nerve stimulation at intensity able to elicit a cMAP in the DAO (Pilurzi et al., 2013;2020).Pilurzi and colleagues (2020) suggested that stimulation of the facial nerve could generate afferent activity by evoking muscle twitches that were detected by mechanoreceptors in the overlying skin (Edin and Johansson, 1995) and/or by activation of Ruffini-like receptors (Cobo et al., 2017).They also suggested that a possible explanation for this result could be the different threshold of the stimulated fibers involved in the SAI and LAI.In particular, the afferents responsible for the SAI induced by stimulation of trigeminal sensory afferents may have a low threshold and are readily activated by the peripheral nerve stimulus.In contrast, mechanosensitive receptors in the skin and in the face muscles, activated by a facial muscle twitch, may have a higher threshold, and therefore are not activated by electrical stimulation alone (Pilurzi et al., 2020).According to this hypothesis, stimulation of the trigeminal nerve at 3 Â PT, sufficient to activate low threshold cutaneous receptors, is capable of evok-ing SAI but not LAI, for which the activation of muscle mechanoreceptors is necessary.This is compatible with the finding that LAI was detectable only when the intensity of facial nerve stimulation was able to evoke a stable cMAP and muscle twitch in the DAO.In contrast, LAI disappeared when the facial nerve was stimulated at an intensity subthreshold for eliciting a cMAP in DAO.Based on these results, it is reasonable to suggest that in the DAO, SAI is primarily due to activity in cutaneous afferents rather than proprioceptive or pure muscle afferents whereas LAI requires muscle afferents.

SAI and LAI in the UT muscle
In the UT, LAI was detected following stimulation of the accessory nerve at an intensity able to evoke a stable cMAP in the muscle.As in DAO, LAI was only present when motor nerve fibers were stimulated at an intensity able to induce a muscle twitch.
In the UT, we failed to detect clear SAI with any of the stimulation intensities used.This could be because activation of cutaneous afferents from the skin overlying the UT is necessary for SAI and such afferents would not have been activated by the stimuli we used.In fact, effective stimulation of the sensory supply from the UT territory, requires stimulation of the ventral rami of the third (C3) and fourth (C4) cervical nerves (Ourieff et al., 2022).
The lack of SAI in the UT could be explained by the observation that it is weak or absent in proximal muscles compared to distal muscles, as suggested by the distal to proximal attenuation of SAI following digital stimulation (Bikmullina et al., 2009).In addition, it has been reported that the depth of afferent inhibition depends on the proximity of the nerve stimulated to the muscle from which SAI/LAI is recorded (Turco et al., 2018).
The lack of SAI in the UT and the presence of LAI only with stimuli capable of inducing a muscle twitch could be also interpreted in light of the functional role of this muscle, which is mainly involved in postural rather than manipulative tasks, which require more proprioceptive than cutaneous inputs, respectively.This hypothesis is compatible with previous work showing that DN stimulation causes segmental SAI in upper limb muscles, while toe stimulation facilitates motor output both in foot and upper arm muscles (Bikmullina et al., 2009) implying that the cutaneo-motor pathways in arms and legs are organized in a functionally different way (Bikmullina et al., 2009).It is reasonable to think that cutaneous finger or hand stimuli have a different biological meaning than an external stimulus applied to the foot.Indeed, the former are fundamental for manipulation of tools, while the latter are more important in the context of postural control.Along these same lines, activation of proximal arm muscles and of cervical muscles could serve postural control by stabilizing the trunk and the head.In this light as for leg muscles, UT proprioceptive stimuli may be functionally more important than cutaneous stimuli or, at least, the sensory-motor pathway might be more sensitive for the former stimuli.

SAI and LAI in the cranio-cervical muscles
The results of the present study suggest that SAI and LAI probably do not depend only on the pathways activated at the different ISIs and by the nerve composition (Tamburin et al., 2002;2005;Turco et al., 2018) but they are probably influenced also by the functional role of the target muscle.
Data obtained from the AD muscle are in line with those obtained in previous studies on the hand model Chen et al., 1999;Tamburin et al., 2005;Turco et al., 2017;2018).Although, the neural pathway traversed by the afferent volley leading to SAI remains unclear, it has been hypothesized that SAI may result from direct thalamocortical projections to M1 via cholinergic paramedian thalamic nuclei (Di Lazzaro et al., 2000).Alternatively, SAI may arise via a relay through S1 with neuronal projections from S1 to M1, which recruit M1 interneurons that in turn inhibit pyramidal neurons in layer V (Di Lazzaro et al., 2005a;2005b;2007;Turco et al., 2018).Also, the neural pathway mediating LAI remains unclear (Chen et al., 1999;Turco et al., 2018).It appears to be cortical in origin and occurs at long intervals between the nerve stimulus and TMS pulse, indicating the opportunity for widespread activation of various somatosensory cortical areas responsive to afferent sensory input (Allison et al., 1989;1992;Forss et al., 1994;Huang et al., 2000;Turco et al., 2018).The contralateral S1, PPC, and S2 cortices are activated during the first 200 ms after mixed nerve stimulation, while digital nerve stimulation leads to the activation of S2 after $100 ms (Hari et al., 1984).Since LAI was observed in the AD with the shortest ISI and a stimulation intensity of 3 Â PT, it could be suggested that cutaneous afferents produced initial activation of S1, mediating SAI, followed by activation of the S2 cortex, producing LAI.
Data obtained from all the three muscles recorded (AD, DAO and UT) showed SAI and LAI at ISIs of 20 ms for SAI and 100 ms for LAI.In regard to SAI, several studies showed that the perfect ISIs should be chosen after the recording of somatosensory evoked potential (SEP) in order to measure the latency of the SEP arrival in somatosensory cortex at which it should be added 2 ms to accommodate the cortical transmission (Turco et al., 2018).Although the recording procedure of trigeminal somatosensory evoked potential is still debated in the literature, several works (Hong et al., 2019;Karre et al., 2023;Baba et al., 2021) demonstrated that the minimum time needed to reach the S1 is around 13-15 ms.Therefore, considering that to elicit SAI it should be added 2/5 ms (Turco et al., 2018) to accommodate the cortical transmission we reach 20 ms that is the ISI significant for the cranial-cervical district.On the other hand, LAI was detected in the cranial cervical district only at an ISI of 100 ms (Pilurzi et al. 2013;2020), differently from the hand muscle district, where LAI was observed at an ISI of 180-200 ms (Chen et al., 1999;Turco et al., 2018).It has been observed that the contralateral S1, PPC, and S2 cortices are activated during the first 200 ms after mixed nerve stimulation, while digital nerve stimulation leads to the activation of S2 after $100 ms (Hari et al., 1984).Furthermore, it has been suggested that in the DAO, LAI could mainly depend on mechanoreceptor activity or proprioceptive facial inputs (Pilurzi et al., 2020), which activate inhibitory circuits involving contralateral SI (Friedman and Jones, 1981;Allison et al., 1992) and bilaterally the SII and the posterior parietal cortex PPC.These findings led to speculate that S2 may be mainly involved in the LAI mechanism, and this could explain why in the DAO LAI was observed at an ISI of 100 ms.
In the LAI paradigm results are similar to those obtained in the DAO, in that both muscles require activation of muscular afferents.In contrast, in the AD cutaneous afferents alone were able to evoke robust SAI and LAI, as in the hand model (Turco et al., 2017;2018).This evidence suggests that sensorimotor integration is not the same in all muscles probably because of differences in muscle function.The AD plays a crucial role in mastication (Morquette et al., 2012), where sensory inputs are not necessary to generate basic masticatory movements although they are essential to adapt movements to the hardness of the food and to compensate for unexpected perturbations (Rossignol et al., 1988, Morquette et al., 2012).While periodontal receptors are crucial in providing feedback on food hardness and bolus, muscle spindle afferents from jaw closing muscles are important for feedback about the position of the jaw (Appenteng et al., 1982).The movement is greatest around the corners of the mouth, an area highly prone to skin distortion during jaw-movement (Lund, 1991) and in this context cutaneous afferents provided by trigeminal sensory afferents play a crucial role in food manipulation in the oral cavity and in the control of fine movements of the mouth.In this sense, the AD resembles to hand muscles when they are engaged in object manipulation and grasping.
In the whole, our data suggest that cortical sensorimotor integration processes depend on the type of receptor, the function of the muscle involved, and the cortical area engaged.The results obtained in craniofacial and cervical muscles could explain data recently shown in patients with focal dystonia in whom it has been demonstrated that alterations in cortical excitability and sensorimotor integration present a specific topography (Ginatempo et al., 2023).

Limitations of the study
We acknowledge that this study has some limitations.First, we recorded from a jaw opening muscle, i.e., AD, but we did not compare data with those obtained in a jaw-closing muscle, which are not devoid of muscle spindles, such as masseter muscle.The reason for this choice is that the TMS-induced MEPs in AD are elicitable in the resting condition, as opposed to the masseter muscles where pre-innervation is necessary to elicit MEPs (Ortu et al., 2008).Second, due to technical limitations we did not investigate the SAI protocol in UT following the stimulation of rami of the third (C3) and fourth (C4) cervical nerves.Third, further studies are needed to investigate the effect of ISIs between 25 ms and 100 ms on SAI and LAI.Fourth, the effect of handedness was not investigated in the present study since cortical motor projection to cranial muscles is bilateral.However, all the participants were right-handed according to the Oldfield inventory scale, which minimize possible contamination of the results by a non-homogeneous sample in respect to handedness.Finally, we used the perceptual threshold for defining electrical stimulation intensity to tailor it on the individual pain subthreshold intensity, according to previous studies (Turco et al., 2017;2018;Pilurzi et al., 2013;2020).

Conclusions
In conclusion, results show that the cortical integration of sensory inputs and motor output in the cranio-cervical district varies according to the type of afferent fibers (cutaneous or muscular) and to and to the specific function of the target muscle.These findings should be taken into account when studying the pathophysiology of focal motor disorders, where specific muscular districts are involved, but also in those cases where the afferent fibers are damaged such as in the trigeminal neuralgia, which could lead to an altered sensorimotor integration in both face and masticatory system.

Fig. 2 .
Fig.2.Effects of trigeminal and facial nerve stimulation on motor evoked potentials (MEP) of the anterior digastric muscle (AD) and depressor anguli oris muscle (DAO) in the short afferent inhibition (SAI) and long afferent inhibition (LAI) protocols.The boxplots report the raw amplitudes of the test MEP, obtained with a single pulse intensity of 120 % resting motor threshold, and of the conditioned MEPs obtained following the coupling of peripheral nerve (intensity 3x perceptual threshold) and TMS stimulations at various interstimulus intervals (ISI).In the SAI protocol (15-25 ms, ISIs), the amplitude of AD and DAO MEPs was significantly reduced by trigeminal stimulation at 20 ms ISI, while it appeared to be unaffected by facial nerve stimulation.In the LAI protocol (100-200 ms ISIs), the amplitude of AD MEPs was significantly reduced by trigeminal stimulation at 100 ms ISI, while DAO MEPs showed a significant inhibition at 100 ms ISI after facial nerve stimulation.The continuous line in the boxplot represents the median value, while the 'Â' symbol represents the mean value of the group.*p < 0.05.

Fig. 3 .
Fig. 3. Effects of facial nerve stimulation on motor evoked potentials (MEP) of the depressor anguli oris muscle (DAO) in the short afferent inhibition (SAI) and long afferent inhibition (LAI) following facial nerve stimulation at different intensities.The boxplots report the raw amplitudes of the test MEP, obtained with a single pulse intensity of 120 % resting motor threshold, and of the conditioned MEPs obtained following the coupling of facial nerve and TMS stimulations at various interstimulus intervals (ISI).Facial nerve was stimulated with an intensity able to evoke a stable compound muscle action potential (cMAP) in the DAO and, in a separate session, at subthreshold intensity for eliciting a cMAP in the DAO.The amplitude of DAO MEPs was significantly reduced by facial nerve stimulation only in the LAI protocol (100 ms ISI) and only with a cMAP suprathreshold intensity.The continuous line in the boxplot represents the median value, while the 'Â' symbol represents the mean value of the group.*p < 0.05.

Fig. 4 .
Fig. 4. Effects of accessory nerve stimulation on motor evoked potentials (MEP) of the upper trapezius (UT) in the short afferent inhibition (SAI) and long afferent inhibition (LAI) protocols at different electrical stimulus intensities of the accessory nerve.The boxplots report the raw amplitudes of the test MEP, obtained with a single pulse intensity of 120 % resting motor threshold, and of the conditioned MEPs obtained following the coupling of the accessory nerve and TMS stimulations at various interstimulus intervals (ISI).The accessory nerve was stimulated with an intensity able to evoke a small stable compound muscle action potential (cMAP) in the UT, and, in a separate session, at subthreshold intensity for eliciting a cMAP in the UT.The amplitude of UT MEPs was significantly reduced by accessory nerve stimulation only in the LAI protocol (100 ms ISI) and only with a stimulation intensity able to evoke a cMAP in the UT.The continuous line in the boxplot represents the median value, while the 'Â' symbol represents the mean value of the group.*p < 0.05.
RMT, Resting Motor Threshold; MSO, maximum stimulator output; AD, anterior digastric muscle, DAO, depressor anguli oris muscle, UT, upper trapezius muscle, cMAP, compound motor action potential, mA, milliampere.The table reports mean ± standard mean error (SEM) and minimum and maximum value range.