Shoulder position and handedness differentially affect excitability and intracortical inhibition of hand muscles

Abstract

Individuals with stroke show distinct differences in hand function impairment when the shoulder is in adduction, within the workspace compared to when the shoulder is abducted, away from the body. To better understand how shoulder position affects hand control, we tested the corticomotor excitability and intracortical control of intrinsic and extrinsic hand muscles important for grasp in twelve healthy individuals. Motor evoked potentials (MEP) using single and paired-pulse transcranial magnetic stimulation were elicited in extensor digitorum communis (EDC), flexor digitorum superficialis (FDS), first dorsal interosseous (FDI), and abductor pollicis brevis (APB). The shoulder was fully supported in horizontal adduction (ADD) or abduction (ABD). Separate mixed-effect models were fit to the MEP parameters using shoulder position (or upper-extremity [UE] side) as fixed and participants as random effects. In the non-dominant UE, EDC showed significantly greater MEPs in shoulder ABD than ADD. In contrast, the dominant side EDC showed significantly greater MEPs in ADD compared to ABD; %facilitation of EDC on dominant side showed significant stimulus intensity x position interaction, EDC excitability was significantly greater in ADD at 150% of the resting threshold. Intrinsic hand muscles of the dominant UE received significantly more intracortical inhibition (SICI) when the shoulder was in ADD compared to ABD; there was no position-dependent modulation of SICI on the non-dominant side. Our findings suggest that these resting-state changes in hand muscle excitabilities reflect the natural statistics of UE movements, which in turn may arise from as well as shape the nature of shoulder-hand coupling underlying UE behaviors.

Introduction

Coordinated UE movement requires functional coupling between the shoulder and hand joints. This shoulder-hand coordination is at least partially accomplished by the projection patterns of cortical and subcortical pathways. For example, a majority of descending corticospinal and rubrospinal axons terminate on combinations of wrist-digit and shoulder/elbow muscles (Belhaj-Saif et al. 1998; Park et al. 2004, 2001). Whereas isolated joint positions are represented in the mossy fiber inputs to the cerebellar cortex (van Kan et al. 1993a), the sole output from the intermediate cerebellum (nucleus interpositus, NI), preferentially represents combinations of shoulder, elbow, and hand joints (Geed et al. 2017; van Kan et al. 1994, 1993b). These NI axons project to the magnocellular red nucleus which gives rise to the descending rubrospinal fibers that terminate on spinal interneurons or motoneurons innervating combinations of shoulder-elbow and wrist-digit muscles (Houk et al. 1988; Kennedy et al. 1986; Kuypers et al. 1962; Lawrence and Kuypers 1968); and to the primary motor cortex through the ventral lateral thalamus gives rise to the descending lateral corticospinal pathway innervating the forelimb (Kuypers 2011; Lemon et al. 2012; Morecraft et al. 2013). In summary, a large proportion of the descending motor control encodes shoulder and hand muscles as integrated functional units (Capaday 2004; Geed et al. 2017).

Consistent with the descending projection patterns, transcranial magnetic stimulation (TMS) studies show coupling between distant but functionally-related muscle groups in the UE. Neurologically intact adults show differences in hand muscle excitabilities in UE pointing compared to isolated activity of the extensor carpii radialis (Devanne et al. 2002), in sitting compared to standing (Runnalls et al. 2017), with UE loading (Runnalls et al. 2014), or with neck movement (Fujiwara et al. 2009). Individuals with stroke show greater impairment in hand control when the arm is load-bearing in shoulder flexion or abduction compared to when the arm is fully supported (Ellis et al. 2007; Hoffmann et al. 2011; Kisiel-Sajewicz et al. 2011; Lan et al. 2017; McPherson and Dewald 2019a; Miller and Dewald 2012; Twitchell 1951).

Shoulder-hand muscles are coupled not only in a task-dependent manner (Datta et al. 1989; Flament et al. 1993; Geed and Van Kan 2017; Runnalls et al. 2014) but also at rest. Change in shoulder or wrist postures modulates the excitability of muscles at distant UE joints (Dominici et al. 2005; Fujiwara et al. 2009; Ginanneschi et al. 2005, 2006; Mazzocchio et al. 2008; Mogk et al. 2014; Nuzzo et al. 2016). The resting-state changes in corticomotor excitabilty may reflect preferred activity patterns for shoulder-hand joints in the circuitry (Bizzi and Cheung 2013). A majority of studies to date have explored task- or posture-dependent changes in recruitment properties (recruitment curves) of UE muscles; however, a combination of excitatory and intracortical mechanisms of inhibition or facilitation contribute to the sum of forelimb muscle activity (Capaday et al. 2009). In particular, intracortical inhibition is crucial to shaping the local cortical output, especially for fine motor tasks (Fishell and Kepecs 2019; Kepecs and Fishell 2014). At present, there is little comprehensive data in humans on excitatory and intracortical mechanisms of inhibition or facilitation underlying the shoulder-hand coupling observed in behavioral and neurophysiological studies of the UE. Beside a neuroanatomical coupling between shoulder-hand joints, handedness (propensity to use a preferred hand) influences excitabilities of hand muscles (Daligadu et al. 2013; Souza et al. 2018); however, it is unclear if the handedness effects on corticomotor excitability are exerted uniformly throughout the UE joint space or vary in line with the natural statistics (preferred use patterns) of dominant or non-dominant UE (Kilbreath and Heard 2005; Mamolo et al. 2006; Schweighofer et al. 2015) that are clearly non-uniformly distributed in the workspace.

To better understand the neurophysiology of shoulder-hand coupling and its dysfunctional control after stroke, we specifically tested, in healthy neurologically intact adults, how change in shoulder position affects the corticomotor and intracortical control of extrinsic and intrinsic hand muscles in the dominant and non-dominant UE.

Materials and methods

Subjects

Twelve (6 male) right-handed adults participated in the study (mean ± SD age was 51.3 ± 18 years; range, 25–74 years). Handedness was determined using the Short-form Edinburgh Handedness Inventory (Oldfield 1971; Veale 2014). Study inclusion criteria required the ability to give informed consent, age between 18 and 90 years, no contraindications to TMS (Groppa et al. 2012; Rossi et al. 2011, 2009), e.g., no metal objects inside eyes or skull, history of seizures, no pacemakers, or possibility of being pregnant, no known neurological impairments, ability to perform study tasks, and no known orthopedic injuries at UE joints. All participants gave written informed consent for the protocol approved by the MedStar Health Research Institute Ethics Committee.

Study design and task

All testing was completed in a single session using a repeated measures design. Single- or paired-pulse TMS was used to elicit motor evoked potentials (MEPs) from four UE muscles: the extensor digitorum (EDC), flexor digitorum superficialis (FDS), abductor pollicis brevis (APB), and the first dorsal interosseous (FDI) on the right and left UE. TMS was elicited in two arm postures (Fig. 1); stimulus–response characteristics after single-pulse stimulation at 90, 110, 130, and 150% of the resting motor threshold, and intracortical facilitation (ICF), short intracortical inhibition (SICI), and long intracortical inhibition (LICI) protocols were used to examine corticomotor physiology in each shoulder position.

Fig. 1.

Schematic representation of the arm position during testing. UE was secured on a rigid but padded support with the elbow fixed to 90° and wrist fixed at 10° dorsiflexion (neutral) in a wrist brace with the palm facing down on the arm-support frame. The shoulder joint could be freely moved between horizontally adduction (a) to abduction (b)

Figure 1 shows a schematic of the experimental set up. Participants were familiarized with the behavioral task, sounds, and sensations of the TMS before the first testing session. During testing, participants sat in a chair with back support and feet resting on the floor. The chair’s height and distance from a table in front of the participants could be adjusted to ensure consistent positioning of UE and comfort. Participants’ testing arm was fully supported on a custom-built rigid but cushioned arm-support frame secured to the table. The arm-support frame allowed the shoulder joint to move freely in the horizontal plane from horizontal adduction (0°, testing condition “ADD”; Fig. 1a) to horizontal abduction (90°, testing condition “ABD”; Fig. 1b). The elbow was secured at 90° to the frame using elasticized fabric wrap, and participants wore a wrist brace (palm facing down on the arm-support frame) to stabilize the wrist and digits consistently throughout the testing session. Joint angles were aligned initially using a goniometer and the arm-support frame angles locked to ensure consistent placement of the UE.

Electromyography

Surface electromyography (EMG) electrodes were placed on five muscles of the right and left UE (EDC, FDS, APB, FDI, and middle Deltoid). These muscles were selected, because EDC and FDS are key extrinsic muscles of the hand necessary for a stable grasp and hand opening as in reaching to grasp. Anterior and middle Deltoid are key to shoulder elevation or abduction, which is coupled spatiotemporally with hand opening as in reaching to grasp (Geed et al. 2017; Geed and Van Kan 2017; Jeannerod 1984; Paulignan et al. 1990). Deltoid receives extensive collateral innervation through the cortical and subcortical pathways (Belhaj-Saif and Cheney 2000; Belhaj-Saif et al. 1998; Boudrias et al. 2006, 2010; Holdefer and Miller 2002) suggesting a neuroanatomical substrate for synergistic coupling with hand muscles. APB and FDI were selected, because these are key intrinsic muscles of the hand important to stabilize a precision and whole-hand grasp with the wrist stabilization provided by the EDC, extensor carpii radialis (ECR), and FDS. All muscle activities were monitored in real time; particularly Deltoid, to ensure that participants kept their shoulder relaxed throughout testing so that the arm remained unloaded. All UE muscles were located and electrode positions confirmed according to Perotto (2011). The amplified EMG signals were filtered (bandpass, 3-10 kHz), sampled at 10 kHz (CED Power 1401, Cambridge Electrical Design, Cambridge, UK), and stored on a personal computer for off-line analysis using CED Signal software (v4.09).

Transcranial magnetic stimulation

Single- or paired-pulse TMS was applied over the left (and right) motor cortex using a MagPro X100 with MagOption magnetic stimulator and the C-B60 figure of eight coil (MagVenture, Denmark). Real-time neuronavigation using a template MRI scan (BrainSight; Rogue Research, Montreal, Quebec, Canada) was used to record the coil position corresponding to the motor hotspot and ensure consistent positioning of the TMS coil. All TMS was completed by the same investigator.

The coil was held tangentially to the scalp angled approximately 45° away from the midline to induce electrical current in the cortex in the posterior-anterior direction (Brasil-Neto et al. 1992). The optimal scalp location (motor hotspot) was chosen as the location on scalp that produced the largest peak-to-peak amplitude of motor evoked potentials (MEPs) in FDI with the arm held in a standardized position placed on the table in front of participants. FDI MEPs were used to select the hotspot, because it most consistently produced maximal peak-to-peak MEPs for the four muscles. Resting motor threshold (RMT) was defined as the minimum stimulus intensity that elicited a 50-uV MEP in five out of ten trials in FDI while seated with the arm resting, fully supported, in the standardized position on the table in front of the participant. A single hotspot and RMT was used for the four muscles, unless the RMT differed between APB, EDC, FDS compared to FDI by more than 5%, in which case separate hotspots and RMTs were explored for the individual muscles. None of the participants needed this adjustment.

Single- and Paired-pulse data collection

MEPs in response to single-pulse stimulation were collected by recording the responses to 10 pulses each at 90–150% RMT intensity in increments of 20%. Order of stimulation intensities was pseudorandomized. Intracortical physiology was measured using the short intracortical inhibition (SICI), long intracortical inhibition (LICI), and intracortical facilitation (ICF) protocols described previously (Kujirai et al. 1993). Briefly, for SICI, test stimuli (TS) were delivered either alone (single-pulse) or preceded 3 ms by a subthreshold conditioning stimulus (CS; paired-pulse stimulation). To elicit SICI, CS intensity was kept constant at 80% of the RMT, and TS intensity was set to 120% of the RMT. To elicit ICF, the 80% RMT CS was followed 15 ms later by a 120% TS; for LICI, a 120% RMT CS was followed 150 ms later by a 120% TS. For each of the paired-pulse protocols, 15 single pulses were randomly interspersed with 15 paired pulses. The order of eliciting RC, SICI, ICF or LICI was randomized across participants.

Data analysis

EMG traces were inspected online during testing for the presence of appropriate stimulus–response artifact and absence of phasic muscle activity (see Supplementary Figs S1 and S2) showing raw EMG traces from single-pulse and SICI data in a single participant). Trials were examined offline for background muscle activity starting 500 ms before TMS stimulation and excluded if shoulder/hand muscles showed phasic muscle activity, coactivation, or the absence of a TMS artifact. Outlying data values for trials within and between participants were examined using the Rosner’s Many Outliers Procedure (Rosner 1983) but not removed from further processing, because there was no good reason for these trials or TMS variable values to not belong to the dataset. MEP variables were calculated using custom MATLAB scripts. Specifically, we calculated, MEP onset (ms), MEP offset (ms), MEP area (uV.ms), peak-to-peak amplitude of the average waveform (uV), and % facilitation (% increase in the mean value of the signal between MEP onset to offset compared to the mean signal 50 ms before MEP onset). Systematic interindividual differences in MEP size were normalized by setting the dominant side shoulder MEP in ABD = 1 and rescaling all within-individual values relative to 1. Amount of paired-pulse inhibition or facilitation was calculated as the % change (1- (conditioned MEP/unconditioned MEP))*100 (Motawar et al. 2012, 2016). Offline data processing was performed using custom scripts in MATLAB R2019a (MathWorks, Natick, Massachusetts).

Statistical analysis

Separate linear mixed effect models (Runnalls et al. 2017, 2014) were fit for each experimental question. A mixed model accounts for the linear regression violation of independent and identically distributed observations because of smaller intra-individual variability compared to interindividual variability in repeated measures study designs like presented here. Specifically, we asked, for each UE muscle, if (1) the mean MEP variables (MEP area, peak-to-peak value of average waveform, % facilitation) differed between shoulder positions (ADD or ABD) in the same UE (within-side analyses). Following this primary analyses, EDC showed contrasting changes in excitability in ADD compared to ABD, therefore a followup analysis comparing between-side differences was completed that asked, (2) given a shoulder position (ADD or ABD), if the MEP variables differed between dominant or non-dominant UE (between-side analysis). A more complex model using shoulder position and UE side as fixed effects was not used in a single mixed model to avoid irrational comparisons such as between shoulder adduction in the right UE compared to shoulder abduction in left UE. Separate models were estimated for each of the four muscles (EDC, FDS, FDI, and APB), because the primary interest was to test change in the same muscles’ excitability given a change in shoulder position and not intermuscular differences.

To test if excitability differs between shoulder ADD and ABD within the dominant UE, shoulder POSITION (ADD or ABD) and STIMULUS INTENSITY (90, 110, 130, 150% of RMT) were entered as the fixed effects and PARTICIPANT as random effect. An interaction term (POSITION*STIMULUS INTENSITY) was also included in the model. Thus, when examining the MEPs elicited using single-pulse stimulation at 90, 110, 130, and 150% of the RMT, the four stimulation intensities were entered as four levels of the variable “STIMULUS INTENSITY”; two shoulder positions were entered as two levels of the variable “POSITION” (ADD and ABD). To test the paired-pulse data, % change in the MEP due to the conditioning pulse was calculated (1-(conditioned MEP/unconditioned MEP))*100. To test if paired-pulse TMS variables differed between shoulder ADD and ABD within the dominant UE, shoulder POSITION was entered as the fixed effect and PARTICIPANT as the random effect. Similar models were fit for the non-dominant UE. Alpha was set to 0.05; multiple comparisons were adjusted using the Bonferroni correction. p values were estimated via t-tests using the Satterthwaite approximation to degrees of freedom.

In the follow-up analyses, to test if excitability or intracortical physiology differed between the dominant or non-dominant UE for the same shoulder positions (ADD or ABD), UE SIDE (dominant or non-dominant) and STIMULUS INTENSITY (90, 110, 130, 150% of RMT) were entered as fixed effects, PARTICIPANT as random effects. Interaction between UE SIDE*STIMULUS INTENSITY was included in the models. Similar models were fit for paired-pulse variables using % change in paired-pulse TMS variable relative to single-pulse variable as the dependent variable, UE SIDE as fixed and PARTICIPANT as random effects. Separate models were fit for shoulder ADD and for shoulder ABD. Statistical analyses were completed in Minitab® LLC., version 19.2020.1.

Results

Data from all 12 participants (6 male) is presented here. Participants’ age ranged from 25 to 74 years (mean ± SD 51.3 ± 18). All participants were right handed (laterality quotient mean ± SD = 94.3 ± 6.5), tested using the Short-Form Edinburgh Handedness Inventory (Veale 2014). Table 1 shows participant demographics and resting motor thresholds in each participant in the dominant and non-dominant UE. Significant comparisons are reported in text. A statistical summary for within-side and between-side analyses are reported in Supplementary Tables S2 and S3. Stimulus intensity showed a significant main effect in all mixed models as expected and is not reported, because we were not interested in testing the already well established “significant change in excitability with increasing stimulus intensity” hypothesis. All means are conditional fitted means (the estimates for mean response values at the fixed and random factor settings obtained from the mixed models) and standard errors (SEM).

Table 1.

Participant demographic and resting motor thresholds (RMT) on each UE

Participant ID	Age	Gender	Laterality quotient	Dominant RMT % MSO	Non-dominant RMT % MSO
P01	27	Female	100	42	40
P02	45	Male	100	51	68
P05	40	Male	100	53	54
P06	31	Male	87.5	42	38
P07	25	Female	87.5	45	48
P08	51	Male	100	52	50
P11	52	Female	100	93	87
P12	62	Female	87.5	76	68
P15	73	Male	100	48	48
P16	74	Female	87.5	55	50
P17	73	Male	87.5	56	55

Handedness quotient was computed using the Short-form Edinburgh Handedness Inventory comprised of four items (Writing, Throwing, Toothbrush, Spoon) scored on an ordinal scale (Always right (score = 100), Usually right (score = 50), Both equally (score = 0), Usually left (score = −50), Always left (score = −100). The Laterality quotient is computed by adding scores on the four items and dividing by 4; Left handers score = −100 to −61, Mixed handers = -−60 to 60, Right handers = 61 to 100

Within arm, how does moving the UE from ADD to ABD affect corticomotor drive to hand muscles?

EDC but not FDS shows significant difference in excitability with different shoulder positions

In the non-dominant UE, MEP area of EDC was significantly greater for shoulder ABD (0.593 ± 0.089 uV.ms) compared to ADD (0.407 ± 0.085 uV.ms) with a significant main effect of position: F_1,59.22 = 4.11, p = 0.04 (Fig. 2a). In contrast, FDS showed no significant difference in excitabilities in ABD (0.883 ± 0.509 uV.ms) compared to ADD (0.901 ± 0.509 uV.ms) with F _1,53.66 = 0.00, p = 0.96 (Fig. 2a). There was no significant interaction between position and stimulus intensity in the non-dominant UE.

In the dominant UE (Fig. 2b), EDC showed a significant main effect of position for MEP area, significantly smaller in ABD (0.554 ± 0.113) compared to ADD (0.793 ± 0.117) with F _1,63.26 = 7.16, p = 0.009 (Fig. 2b). On the dominant side, EDC % Facilitation also showed a significant main effect of position (F _1,63.51 = 9.40, p = 0.003), and a significant interaction effect between position x stimulus intensity (F _3,63.33 = 4.68, p = 0.009). Post-hoc testing for the interaction effect showed % facilitation was significantly greater in ADD compared to ABD only for the highest stimulation intensity of 150% (see Fig. 2c, difference between means for shoulder ABD-ADD at 150% RMT = −1.048 ± 0.025; t_62.78 = −4.18, adjusted p = 0.003). Dominant side FDS showed no significant differences in MEP area for change in shoulder position from ADD (0.673 ± 0.122) compared to ABD (0.569 ± 0.120) with F _1,59.63 = 1.41, p = 0.240. Figure 2b shows the FDS MEP areas reported here.

Similar to EDC, the dominant side APB showed significantly more % facilitation in ADD (0.894 ± 0.181) compared to ABD (0.559 ± 0.185) with F _1,63.24 = 6.07, p = 0.016, Fig. 2d. The non-dominant APB showed no significant difference between ADD (1.147 ± 0.698) compared to ABD (1.556 ± 0.700) with F _1,63.10 = 0.73, p = 0.398 (data not shown). Dominant and non-dominant side FDI showed no significant differences between MEPs whether the UE was in ADD or ABD (dominant FDI shown in Fig. 2d; statistical summary in Supplementary Tables S1 and S2).

Intrinsic, but not extrinsic hand muscles show significant differences in SICI based on shoulder position and handedness

In the dominant UE, FDI and APB showed significantly more SICI in ADD compared to ABD; a matching shoulder position-dependent influence of SICI was absent on the non-dominant side. Thus, in the dominant UE, FDI sees significantly more SICI in ADD (82.99 ± 5.95) compared to ABD (75.79 ± 6.02) with F _1,9.19 = 5.67, p = 0.04 (Fig. 3a); and APB sees significantly more SICI in ADD (71.87 ± 7.04) compared to ABD (59.28 ± 7.20) with F _1,9.28 = 5.70, p = 0.04 (Fig. 3b). The extrinsic hand muscles (EDC or FDS) did not show significant differences in SICI with change in shoulder position within the dominant UE. In the non-dominant UE, neither intrinsic (Fig. 3a, b) nor extrinsic hand muscles showed any differences in SICI with change in shoulder positions.

No significant differences in ICF or LICI in intrinsic or extrinsic hand muscles with change in shoulder position.

FDI, APB, EDC, and FDS showed no significant differences between shoulder positions of ADD or ABD in either the dominant or the non-dominant UE.

Fig. 2.

EDC, but not FDS shows position-dependent changes in excitability. A. In the Non-dominant UE, EDC MEP area (uV.ms) is significantly greater when the shoulder is in abduction (ABD) than in adduction (ADD). FDS shows no significant differences in MEP area whether the UE is in ADD or ABD. B. In the Dominant UE, EDC shows significantly greater MEP area in ADD compared to ABD. C. In the Dominant UE, EDC % Facilitation (change in the mean MEP between onset-offset relative to mean baseline signal 50 ms before the TMS pulse) shows significant interaction between Position x Stimulus intensities. D. Dominant UE, APB shows significantly more % facilitation in ADD compared to ABD. All data points are estimated means and standard errors estimated from mixed models. Thus, for a significant main effect of Position, the model estimated mean reflects the average MEP area across the four stimulation intensities at the given UE position. *shows p < 0.05, ***shows p < 0.001 Rescaled MEP variables shown, see METHODS

Fig. 3.

FDI and APB show significantly more SICI when the dominant UE is placed in ADD compared to ABD. There is no significant difference between ADD and ABD in the non-dominant UE. Estimated means and SEMs are shown. *indicates p < 0.05

Given a shoulder position (ADD or ABD), how is hand muscle excitability affected by UE side?

EDC showed a contrasting pattern of excitabilities in moving from ADD to ABD in the dominant and non-dominant UE. The non-dominant UE showed more excitability in ABD, whereas the dominant side showed more excitability in ADD compared to ABD. To followup, we tested for between-side differences in excitabilities for a given shoulder position.

In shoulder ADD, EDC MEP area was significantly more on the dominant (0.792 ± 0.11) compared to non-dominant (0.419 ± 0.108) side with F_1,59.08 = 14.05, p < 0.001; Fig. 4a. The % Facilitation in EDC also showed significant differences between sides when UE was in ADD (dominant side = 1.202 ± 0.136, compared to non-dominant side = 0.609 ± 0.132) with F_1,59.05 = 20.09, p < 0.001; data not shown). When compared in ABD, there is no between-sides difference in EDC excitabilities (dominant = 0.556 ± 0.086; non-dominant = 0.580 ± 0.092) with F_1,63.06 = 0.08, p = 0.784; Fig. 4a. FDS showed no significant between-sides difference in shoulder ADD or ABD (Fig. 4b).

Fig. 4.

Dominant EDC, compared to the non-dominant side shows significantly greater excitability in shoulder ADD. a Shows single-pulse data from EDC MEP area (uV.ms, rescaled, see METHODS) of dominant compared to non-dominant side in ADD and ABD. Mixed model estimated means and SEM are shown for all figures. ***indicates p < 0.001. b FDS shows no significant difference between sides whether the UE is positioned in ADD or ABD. Although separate models were run for ADD and ABD, notice that non-dominant EDC excitability increases in moving from ADD to ABD but there is no significant difference between dominant and non-dominant EDC excitability when either UE is in ABD

No significant difference between UE sides for SICI, in shoulder ADD or ABD.

There was no significant between-sides difference in SICI for a given shoulder position (ADD or ABD). EDC and FDI show significant between-side differences in ICF with different shoulder positions

When the shoulder is in ADD, EDC receives significantly more intracortical facilitation (ICF) on the dominant side (48.301 ± 13.82%) compared to the non-dominant side (25.384 ± 13.546%) with F_1,9.38 = 5.65, p = 0.04 (Fig. 5a). When the shoulder is in ABD in contrast, intracortical facilitation of the EDC shows no significant between-sides difference with F _1,10.04 = 2.56, p = 0.14 in the dominant (52.447 ± 12.461) compared to the non-dominant (25.959 ± 13.053) UE (Fig. 5a).

FDI shows significantly more intracortical facilitation on dominant (57.80 ± 7.827) compared to non-dominant (41.38 ± 8.064) side with F_1,9.34 = 5.38, p = 0.04, only when the shoulder is positioned in ABD (MEP area, Fig. 5b). A similar effect of position on ICF in FDI is absent when the shoulder is placed in ADD.

FDI and FDS show significant between-side differences in LICI with changing shoulder positions

Fig. 5.

LICI, ICF show significant differences between dominant and non-dominant UE when compared in ADD or in ABD. Estimated means and SEMs are shown. ***indicates p < 0.005, *indicates p < 0.05

In UE ADD, FDI shows significantly more LICI in the non-dominant UE (191.344 ± 23.444) compared to the dominant side (116.4 ± 21.846), with F_1,9.16 = 11.49, p = 0.008. This difference between sides disappears when the UE moves to ABD [dominant UE (138.001 ± 24.397), non-dominant UE (183.386 ± 23.190), with F_1,8.84 = 2.14, p = 0.178. In ABD, FDS receives significantly more LICI on the dominant UE (67.001 ± 17.629) relative to the non-dominant side (34.197 ± 16.946) with F_1,5.33 = 14.73, p = 0.011 (Fig. 5c). FDS LICI shows no difference between sides when compared in ADD.

Discussion

Our study intended to understand how change in shoulder positions affects the corticomotor excitability and inhibition of key hand muscles necessary for a successful grasp. Our findings build on current reports in the literature (Dominici et al. 2005; Ginanneschi et al. 2005, 2006; Mogk et al. 2014) by presenting data on excitatory and inhibitory mechanisms by which shoulder positions affect hand motor control in the dominant and non-dominant UE. All data come from the same group of individuals performing the same task during a single experimental session allowing us to make fair comparisons between mechanisms underlying position- and hand-preference dependent differences in hand muscle control.

This study has two key findings: Within-arm, (1) excitability of EDC (the main wrist and digit extensor muscle) is differentially affected by shoulder position and hand preference. Non-dominant EDC sees more excitability in ABD compared to ADD; the dominant EDC shows the opposite pattern: more excitability in UE ADD than in ABD. In contrast, FDS (key wrist and digit flexor) did not show significant influence of shoulder position or hand preference. (2) Intrinsic hand muscles FDI and APB of the dominant UE receive significantly more SICI when the UE is in ADD compared to ABD. The non-dominant UE in contrast, sees no differences in SICI when the UE moves from ADD into ABD, nor any differences relative to the dominant UE. We also report follow-up between-side effects that helped clarify the within-arm findings: First, the dominant EDC shows significantly more excitability than the non-dominant side only for shoulder ADD, this between-sides difference in EDC excitability disappears when the arms are positioned in ABD. Second, the dominant EDC receives significantly more ICF only in ADD relative to the non-dominant side, this between-sides difference in ICF disappears as arms move into ABD.

Shoulder-hand coupling is present even at rest and reflects the natural statistics of hand use

Shoulder positions intricately affect the excitabilities of wrist and digit muscles as our data and prior reports demonstrate (Dominici et al. 2005; Ginanneschi et al. 2005, 2006; Kouchtir-Devanne et al. 2012). These position-dependent changes may underlie the neural basis of functional coupling between shoulder-hand muscles, important for coordinated UE movements. Importantly, we found increased excitability of the EDC, a key forelimb muscle important for hand opening, with shoulder ABD in the non-dominant UE (Fig. 2a). Shoulder abduction/elevation with hand opening (EDC) is central to reaching-to-grasp behaviors in rodents, in non-human primates, and in humans. Patients with stroke show particular impairment with hand opening, especially when the shoulder is placed in ABD (Brunnström 1970; Lan et al. 2017; McPherson and Dewald 2019a; Miller and Dewald 2012; Twitchell 1951; Yiyun et al. 2014). Wrist and digit extensors preferentially receive collaterals from motor cortical neurons that innervate the shoulder abductors and elbow extensors (Belhaj-Saif et al. 1998; Boudrias et al. 2006; Park et al. 2004) in addition to receiving monosynaptic input from M1 in macaques. These descending projection patterns form a natural substrate for functional synergies spanning shoulder-hand muscles, especially for fundamental natural behaviors like reaching to grasp. The increased resting state excitability of EDC during ABD compared to ADD may reflect a state of readiness or a low-threshold solution for activation of EDC during reach, similar to the preferred activation patterns (Bizzi and Cheung 2013; Latash 2012) of shoulder and hand muscles for reaching to grasp.

The dominant UE showed greater excitability in shoulder ADD relative to shoulder ABD (Fig. 2b). Why would one expect this result on the dominant EDC, contrary to the finding in the non-dominant side? When the % facilitation of the single-pulse data on dominant side were examined, we saw a significant interaction effect between stimulus intensities and shoulder position (Fig. 2c). The dominant UE sees significantly greater %facilitation in ADD only at the higher stimulus intensities, which possibly drives the overall effect of significantly greater excitability in ADD on the dominant side. The dominant side (compared to non-dominant) EDC also receives significantly more ICF selectively in ADD (Fig. 5a), but this ICF influence disappears when arms are positioned in ABD, which might drive the increased excitability of the dominant UE in ADD mechanistically.

We believe these differences in how EDC is controlled differently in dominant compared to non-dominant UE might arise because of the natural statistics of movement (the preferred use patterns). One likely possibility is that reaching to grasp outside the workspace is typically performed with the dominant UE; dominant UE also preferentially reaches contralaterally, especially with increasing task complexity, and is preferred for precision tasks within the workspace (Bryden et al. 2011; Mamolo et al. 2006, 2005). The non-dominant UE in contrast, mainly acts ipsi-laterally for simpler reaching tasks, or in collaboration with the dominant UE within the workspace to stabilize objects (Mamolo et al. 2006, 2005). EDC contributes not only to hand opening during grasp, but also to stabilize the wrist and metacarpal joints to enable precision tasks performed by the digits that are typically performed within the workspace (in ADD) and more frequently by the dominant UE (Bryden 2016; Bryden et al. 2011; Mamolo et al. 2006, 2005). Consistent with the hypothesis, we saw an increase in % facilitation in both EDC (Figs. 2b, c, 4a, 5a) and APB (Fig. 2d) only in the dominant UE when in ADD. APB is a key thumb abductor often used to stabilize objects against which the other digits apply precisely controlled forces for precision manipulation. We believe these natural movement statistics are reflected in the excitabilities of dominant versus non-dominant EDC. In fact, when we compared between-sides excitability of EDC (Fig. 4a), the dominant-side EDC showed greater excitability than the non-dominant side, but only for ADD; the difference between sides disappears if the shoulders are placed in ABD, i.e., MEP areas increase for both sides in ABD relative to the non-dominant ADD. This pattern is consistent with the idea that the increased EDC excitability in UE ABD forms part of a primitive reach-to-grasp synergy in both arms. On top of this primitive pattern, the increased excitability of dominant EDC in ADD is likely a reflection of the preferential use of the dominant UE inside the workspace, especially on precision tasks that need wrist stabilization. Ginanneschi et al. (2006) found a similar effect of shoulder positions (as we did in EDC) on extensor carpii radialis (ECR). ECR is a synergist to EDC and contributes to stabilize and extend the wrist as when the digits are involved in precision tasks inside the workspace.

Selective facilitation of the right (dominant) but not the left side FDI and APB MEPs has been reported, and is correlated with the extent of motor learning/performance in the hand (Garry et al. 2004; Muellbacher et al. 2001). Typically, protocols examining MEPs/motor performance have been tested with the UE only in ADD, revealing a left–right hemisphere difference (similar to our between-sides difference in ADD for EDC, FDI, APB). Is this left–right hemisphere difference uniformly distributed between the left and right halves of the UE workspace? Our data suggest that the dominant UE in ADD may be a special case with selective facilitation (and SICI) of the muscles important for precision and wrist stabilization compared to the dominant UE ABD, non-dominant UE ADD and ABD conditions. The position-dependent changes we report may provide a substrate of heightened excitability or facilitation for efficient motor performance within the workspace in the dominant UE. It is worth examining if the left–right hemisphere differences in the degree of MEP facilitation during motor learning as reported previously (Garry et al. 2004; Muellbacher et al. 2001) also exist within UE, for MEP facilitation during motor learning in ADD compared to ABD. Overall, our findings suggest that these at-rest changes in muscle excitabilities may reflect the natural statistics of UE movements, which in turn may arise from as well as shape the innate functional synergies for coordinated UE behaviors.

Intrinsic muscles of the dominant UE receive more intracortical inhibition in ADD than ABD

Short intracortical inhibition, via the GABA_A type receptors (Ilić et al. 2002; Ziemann et al. 1996) fine tunes the corticomotor excitability, particularly for fractionated control that requires selective inhibition with a strong surround to minimize aberrant neuromotor coactivation. In the dominant UE, we found a significant increase in SICI selectively in the intrinsic hand muscles (FDI and APB) in ADD compared to ABD. There were no differences in SICI on the non-dominant side. Cortical disinhibition of intrinsic hand muscles like FDI and APB is an important mechanism for digit force control during precise fine motor tasks (Schieppati et al. 1996). Greater inhibition in ADD than ABD may reflect different magnitudes of gain (Hu et al. 2014) under which the circuitry works when in different positions in the workspace. Consistent disinhibition, in contrast, could potentially lead to the loss of the ability for rapid dynamic changes in excitability associated with precision in neuronal firing necessary for fine motor control (Geevasinga et al. 2014). Greater SICI in the intrinsic muscles on dominant side, in ADD compared to ABD, may reflect the neuromuscular substrate involved in precision tasks, even at rest. Notably, precision tasks are performed more frequently by the dominant UE inside the workspace than far outside the workspace. We posit that the posture-dependent coupling between groups of functionally-related muscles even when the muscles are at rest reflects the cortical mechanisms underlying preferred muscle activation patterns that may at least in part be driven by the natural statistics of movement and hand preference.

Between-side differences in UE excitabilities are dynamic and position dependent

When comparing UE muscle excitabilities between the dominant and non-dominant UE for a given shoulder position, we identified several between-side differences that remained specific to a given shoulder position. Thus, EDC showed significant differences in MEP area in ADD (dominant side > non-dominant), but this difference between sides disappears when the dominant and non-dominant sides are compared in ABD. We observed similar position and side-dependent differences in FDI, and in FDS in the long intracortical inhibition mechanism (LICI). Between-side differences in TMS-evoked potentials are widely reported (Daligadu et al. 2013; Hammond et al. 2004). While functional laterality affects corticomotor organization and the descending drive to hand muscles, our results suggest hand muscle excitabilities are intricately governed by multiple factors and are not a static representation in the neuromotor circuitry. These position- and side-dependent differences in excitabilities are particularly important when interpreting data from individuals with stroke, where the lesioned side is often compared to the non-lesioned side, or the lesioned side in individuals with stroke is compared with the non-dominant side of healthy volunteers. The functional coupling between muscles appears to arise dynamically, depending at least in part on UE positions and the natural statistics of UE use patterns.

Potential substrates of the shoulder-hand coupling effects

Skin and joint receptors are known to mediate muscle excitabilities through reflex or propriospinal circuitry (Nielsen and Pierrot-Deseilligny 1991; Nielsen 2016); however, skin contact with the UE support frame was consistent between arms and UE postures ruling out afferent receptors as sole mediators of the differences in hand muscle excitabilities in our experiment. Joint receptor afference may have contributed to differences in muscle excitabilities; however, we observed differences between the dominant and non-dominant sides for the same postures, suggesting joint afference may not be a unique influence on hand excitability. Descending monoaminergic pathways exert diffuse influence on excitability of spinal motoneurons that is amplified manifold through persistent inward currents (PICs) (Heckman et al. 2008). For instance, small joint rotations at the ankle (in cats) result in large changes in ankle muscle excitabilities (Hyngstrom et al. 2007). Although we saw excitability changes distant (hand) from the site of joint rotation (shoulder), monoaminergic PIC-driven changes may be a potential substrate for these effects. The diffuse nature of descending monoaminergic inputs may in fact provide an efficient substrate for regulating muscle excitabilities across distant joints, especially for fundamental motor behaviors like locomotion or reaching. Clinical data in individuals with stroke supports the influence of monoaminergic inputs on shoulder-hand control (McPherson et al. 2018, 2008). Individuals with stroke show uncontrolled flexion in elbow and hand when the shoulder is in abduction or loaded (Dewald et al. 1995; Lan et al. 2017; McPherson and Dewald 2019b; Yiyun et al. 2014); notably, disrupting the descending monoaminergic input (via Tizanidine, a noradrenergic α2 agonist) significantly reduces the shoulder-driven increase in uncontrolled UE flexion (McPherson et al. 2018). Alternately cervical spinal neurons (Nuzzo et al. 2016; Yaguchi et al. 2015), the C3-C4 propriospinal system (Alstermark et al. 2011; Ginanneschi et al. 2006; Pettersson 1990; Takei and Seki 2013), or intracortical mechanisms (Capaday 2004; Perez and Rothwell 2015) may be likely substrates for some of the shoulder-driven hand muscle influences we observed.

Limitations

We tested a relatively small sample size of 12 individuals ranging in age from 25 to 74. Although our sample size is comparable to similar prior studies (Ginanneschi et al. 2005, 2006; Runnalls et al. 2017, 2014), the findings ought to be reproduced in a larger sample particularly given the innate variability in TMS measures. Furthermore, intracortical inhibition shows age-related changes (Bonstrup et al. 2015; Opie et al. 2015, 2018; Opie and Semmler 2014), studying systematically stratified groups across the age spectrum will help to better understand the corticomotor mechanisms underlying hand motor control. Lastly, we studied individuals who showed a preference to be right-handed to preserve homogeneity given the other uncontrollable factors that increase the innate variability of neurophysiological data. Given the influence of hand preference in our study, future studies should to focus on systematic performance measures of hand preference (Bryden 2016; Schweighofer et al. 2015), both left and right, with TMS measures to outline the extent to which hand use patterns influence the mechanisms underlying shoulder-hand coupling.