Differential modulation of motor network connectivity during movements of the upper and lower limbs

YNIMG-12341; No. of pages: 10; 4C: 2, 5, 6, 7 NeuroImage xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

3Q2

Lukas J. Volz a,b,c, Simon B. Eickhoff d,e, Eva-Maria Pool c,d, Gereon R. Fink a,d, Christian Grefkes a,c,d,⁎

4 5 6 7 8

a

9

a r t i c l e

10 11 12 13

Article history: Received 6 March 2015 Accepted 28 May 2015 Available online xxxx

Department of Neurology, University of Cologne, Germany Department of Psychological and Brain Sciences, University of California at Santa Barbara, USA Neuromodulation & Neurorehabilitation, Max Planck Institute for Neurological Research, Cologne, Germany d Institute of Neuroscience and Medicine (INM-1, INM-3), Juelich Research Centre, Germany e Institute for Clinical Neuroscience and Medical Psychology, Heinrich-Heine University, Düsseldorf, Germany

O

F

2

Differential modulation of motor network connectivity during movements of the upper and lower limbs

b

a b s t r a c t

P

i n f o

R O

c

Voluntary movements depend on a well-regulated interplay between the primary motor cortex (M1) and premotor areas. While to date the neural underpinnings of hand movements are relatively well understood, we only have rather limited knowledge on the cortical control of lower-limb movements. Given that our hands and feet have different roles for activities of daily living, with hand movements being more frequently used in a lateralized fashion, we hypothesized that such behavioral differences also impact onto network dynamics underlying upper and lower limb movements. We, therefore, used functional magnetic resonance imaging (fMRI) and dynamic causal modeling (DCM) to investigate differences in effective connectivity underlying isolated movements of the hands or feet in 16 healthy subjects. The connectivity analyses revealed that both movements of the hand and feet were accompanied by strong facilitatory coupling of the respective contralateral M1 representations with premotor areas of both hemispheres. However, excitatory influences were significantly lower for movements of the feet compared to hand movements. During hand movements, the M1hand representation ipsilateral to the movement was strongly inhibited by premotor regions and the contralateral M1 homologue. In contrast, interhemispheric inhibition was absent between the M1foot representations during foot movements. Furthermore, M1foot ipsilateral to the moving foot exerted promoting influences onto contralateral M1foot. In conclusion, the generally stronger and more lateralized coupling pattern associated with hand movements suggests distinct fine-tuning of cortical control to underlie voluntary movements with the upper compared to the lower limb. © 2015 Published by Elsevier Inc.

D

1Q1

R

E

C

T

E

14

32

R

36 34 33 35

Introduction

38

Within the primary motor cortex (M1), body parts are somatotopically distributed with locally distinct representations of arm and leg muscles (Penfield, 1937). Planning, preparing and executing motor actions involve a number of brain areas beyond M1 (Boudrias et al., 2012; Eickhoff and Grefkes, 2011). For example, simple isolated movements of the right or left hand typically activate a distributed network comprising bihemispheric motor regions including the primary motor cortex (M1), lateral premotor cortex, supplementary motor area (SMA), somatosensory cortex, basal ganglia, and the cerebellum. These motor areas are also involved in movements of the lower limb (Grafton et al., 1993). However, brain activation patterns differ between upper and lower limb movements. Luft and colleagues

41 42 43 44 45 46 47 48 49

U

39 40

N C O

37

⁎ Corresponding author at: Institute of Medicine and Neuroscience – Cognitive Neuroscience (INM-3), Juelich Research Center, 52425 Juelich, Germany. Fax: + 49 221 478 87698. E-mail address: [email protected] (C. Grefkes).

reported neural activity within “inactive” M1 (i.e. M1 ipsilateral to the moving limb) to be significantly stronger for lower compared to upper limb movements (Luft et al., 2002). Likewise, Kapreli and colleagues found that finger movements featured a stronger lateralization of neural activity in M1 and SMA as compared to lower limb movements (Kapreli et al., 2006). This raises the question, whether interhemispheric differences in neural activation for unilateral movements of the hands and feet result from limb-specific differences in interhemispheric processing underlying a (lateralized) motor action. The vast majority of studies investigating cortical interactions underlying voluntary movements evaluated upper limb actions whereas lower limb movements have not been assessed prior to this study. Experiments using transcranial magnetic stimulation (TMS) to assess interhemispheric inhibition within the motor system reported inhibitory interactions between bilateral M1hand at rest, which were modulated during the preparation and execution of unilateral hand movements (Ferbert et al., 1992; Liang et al., 2014; Murase et al., 2004). Such hemisphere-specific effects seem to facilitate unilateral movements of the “active” hand and a concurrent inhibition of the “inactive” motor cortex, thereby preventing mirror

http://dx.doi.org/10.1016/j.neuroimage.2015.05.101 1053-8119/© 2015 Published by Elsevier Inc.

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68

106 107

114

16 healthy, right-handed subjects (4 male, all right handed, mean age 26 years ± 4 years SD) without any history of neurological or psychiatric problems participated in the study after giving written informed consent in accordance with the Declaration of Helsinki. Handedness was assessed using the Edinburgh Handedness Inventory (Oldfield, 1971), which also comprises an item addressing footedness (“Which foot do you prefer to kick with?”) that showed strong right foot preference for all subjects. The study had been approved by the local ethics committee at the University of Cologne.

115

Experimental design

116

Leg movements tend to translate into spine and head displacement, which typically causes stronger head movement artifacts in fMRI time series compared to hand movements (Seto et al., 2001; Weiss et al., 2013). We thus employed an event-related design using a “sparse sampling” protocol (Fig. 1). The aim of sparse sampling lies in decoupling actual movements from image acquisition, thereby minimizing head movement artifacts in the imaging data. Especially movements occurring during the acquisition of an EPI volume are highly problematic, as they interfere with the correct readouts of the respective slices, leading to slice displacement and wrong excitation order (some slices get excited twice, some are omitted). These effects cannot be corrected by any post-hoc realignment procedure (while simple between-volume displacements are easy to correct for). Therefore, in order to compare brain activity (and derived connectivity estimates) between movements

94 95 96 97 98 99 100 101

108 109 110 111 112 113

117 118 119 120 121 122 123 124 125 126 127 128 129

C

92 93

E

90 91

R

88 89

R

86 87

O

84 85

C

82 83

N

80 81

U

78 79

Inst. Move1s ment 2s

4s 5s

Inst. Move1s ment 2s

3s

EPI 2s

2s

EPI 2s

1s

Inst. Move1s ment 2s Inst. Move1s ment 2s

EPI 2s

3s 4s 5s

EPI 2s EPI 2s

TR = 11

sparse sampling design

F

Subjects

76 77

Inst. Move1s ment 2s

EPI 2s

3s

EPI 2s

2s

Fig. 1. Sparse sampling design: image acquisition is performed de-coupled from movements, with systematic variation of the time lag (2–5 s) between movement execution and recording of fMRI images resulting in sampling of the hemodynamic response.

O

105

75

Inst. Move1s ment 2s

Inst. Move1s ment 2s

4s

R O

Material & methods

73 74

EPI 2s

2s

of the hand and foot, it appears necessary to account for the expected differences in head displacements occurring during movement acquisition. Therefore, in sparse sampling, EPI acquisition is performed after movement execution is finished, i.e., during rest. Sparse sampling makes use of the relatively long delay between a neural event and the evoked hemodynamic response. In the present study, images were acquired 2–5 s after a block of movements, leaving enough time for residual movements of the body to settle and allow the participants to lie as still as possible during actual image acquisition. Sampling of taskinduced neural activation is granted by the fact that maximum of the hemodynamic response function (HRF) is delayed by approximately 5 s (value assumed within the HRF implemented in SPM). By varying the time between movements and image acquisition we sampled the movement-induced hemodynamic response at different time points, accounting for regional differences in HRF peaks. A disadvantage of this method compared to a classical block design lies in its decreased statistical power due to the lower number of images per condition, which limits its application depending on the type of task used to induce neural activation. However, simple motor tasks as used here typically result in highly robust BOLD-signal changes compared to more complex, e.g., cognitive tasks and are therefore especially suited for sparse sampling designs. In the scanner, subjects were first informed which limb (left hand, right hand, left foot, right foot) to move in the upcoming trial (instruction displayed for 1 s). A red blinking circle cued synchronous movements of the respective limb (wrist flexions or ankle flexions) at a rate of 1.5 Hertz for 2 s (equivalent to 3 movements). The end of a given trial was indicated by a centered white fixation-cross on a black background. A temporal jitter of 2–5 s was interleaved between the movement blocks and EPI acquisition (Fig. 1) in order to enable a better sampling of the hemodynamic response (Amaro et al., 2002; Dresel et al., 2005). The entire duration of a block (consisting of instruction, movements, EPI-acquisition and jitters) was 11 s. Each condition was repeated 22 times, and 22 “null events” were included as a “resting baseline” (black screen), adding up to a total number of 110 trials (total scanning time: 20 min). Of note, subjects were explicitly instructed to remain as motionless as possible during “null events” to exclude movement related activity within this baseline condition to bias our results. The sequence of the conditions and null events was randomized across subjects. Prior to the fMRI recording, subjects were familiarized with the task outside the scanner. All subjects were able to perform the requested movements at the defined frequency after a few practice trials. Inside the MR scanner, the subject's forearms were placed onto a pillow in a neutral position between supination and pronation. The wrists were free so that the subjects could easily perform the flexion task. The legs were positioned on a cubic cushion, resulting in a comfortable posture

P

104

71 72

Inst. Move1s ment 2s

T

102 103

movements or even interference (Duque et al., 2007; Hinder et al., 2010; Hinder, 2012). Similar results were obtained from connectivity studies based on fMRI motor activation studies (Grefkes et al., 2008a; Pool et al., 2013). In, addition, these studies also demonstrated strong coupling of contralateral M1hand with premotor areas in both hemispheres (Grefkes et al., 2008a; Pool et al., 2013). Such dynamic modulations of connectivity are strongly linked to motor performance in both healthy subjects (Pool et al., 2013) and stroke patients (Grefkes et al., 2008b; Volz et al., 2015). In contrast, neural dynamics underlying lower limb movement are far less understood. Even for primates—which are frequently used for studying motor system physiology—data on movements of the lower limbs (i.e., hindlimbs) are relatively scarce (Hudson et al., 2015). Furthermore, in humans, to date studies on cortical connectivity underlying movements of the legs are missing. Contrasting ample studies on hand movements, the network dynamics underlying lower limb movements have not been addressed yet. We, therefore, used fMRI and analyses of effective connectivity to address the question of whether similar cortical motor network interactions underlie unilateral movements of the upper and lower limb, or if network dynamics differ substantially, possibly reflecting distinct daily utilization. To this end we assessed the task-dependent modulation of motor network connectivity in 16 healthy, right-handed subjects performing a simple unilateral motor task consisting of similar movement types for the hands and feet. Dynamic causal modeling (DCM) was used to estimate effective connectivity within a cortical motor network, including bilateral M1 representations of the hand and foot (M1hand, M1foot), ventral premotor cortex (vPMC) and supplementary motor area (SMA). We hypothesized that connectivity patterns underlying hand movements differ from connectivity patterns during foot movements. Specifically, we expected differential inhibition of the supposedly inactive, ipsilateral M1 exerted by premotor areas and contralateral M1 with stronger interhemispheric inhibition occurring during unilateral hand movements compared to foot movements possibly reflecting stronger specialization of unilateral utilization of the hands compared to the feet.

D

69 70

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

E

2

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

202

Image preprocessing

203 204

228 229

For image preprocessing, statistical analysis, and dynamic causal modeling, we used the SPM software package (SPM8 update revision number 3042, featuring DCM8; Wellcome Department of Imaging Neuroscience, London, UK, http://www.fil.ion.ucl.ac.uk). The first two volumes of the fMRI time series (dummy images) were discarded from further analysis. After realignment of the remaining 110 EPI volumes and co-registration with the anatomical 3D image, all volumes were spatially normalized to the standard template of the Montreal Neurological Institute (MNI, Canada) using the unified segmentation approach (Ashburner and Friston, 2005). An isotropic smoothing kernel of 8 mm full width half maximum (FWHM) was applied to the EPI images to suppress noise and effects due to residual differences in functional and gyral anatomy. Following spatial normalization and smoothing, statistical analysis was performed. The time series in each voxel were high-pass filtered at 1/128 Hz to remove low frequency drifts. Movement parameters as assessed by the realignment algorithm were treated as covariates to exclude movement related variance from the image time series. The parameter estimates for all four movement conditions as obtained from the general linear model at the single subject level were subsequently compared between subjects in an analysis of variance (ANOVA). The t-statistics for the linear contrasts vs. “rest” and between condition-specific activations (e.g., comparing upper and lower limb movements) were then interpreted by referring to the probabilistic behavior of Gaussian random fields. Voxels were identified as significant if their t-values passed a height threshold of t N 5.11, corresponding to p b 0.05 (family-wise error (FWE) corrected at the voxel level for multiple comparisons).

230

Dynamic causal modeling

231

Effective connectivity was computed for key motor areas activated by the fMRI motor tasks using dynamic causal modeling (DCM; Friston et al., 2003). DCM estimates interregional interactions at the neuronal level including (A) endogenous (movement-independent) coupling, (B) condition-specific coupling, and (C) the direct experimental input to the system that drives regional activity. The coupling

207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227

232 233 234 235 236

E

T

205 206

C

198 199

E

196 197

R

194 195

R

192 193

N C O

190 191

U

188 189

F

200 201

FMRI data were recorded on a Siemens Trio 3.0 T scanner (Siemens Medical Solutions, Erlangen, Germany) using a gradient echo planar imaging (EPI) sequence recorded continuously in an ascending order with the following parameters: TR = 11 s, TA = 2 s, TE = 30 ms, FoV = 192 mm, voxel size: 3.0 × 3.0 × 3.0 mm3, slices = 30, distance factor = 15%, volumes: 110 (plus 2 “dummy images”). Of note, due to the sparse sampling design the TR denoting the time between consecutive image volume acquisitions was relatively long (11 s), whereas each EPI volume was acquired in 2 s (TA = 2 s) with jittered presentation times to allow a sufficient sampling of the HRF. The whole brain from the vertex to lower parts of the cerebellum was covered and. Instructions were presented to the subject in the scanner by a shielded TFTscreen, visible via a mirror mounted to the MR head-coil. In a separate session, high-resolution T1-weighted structural images were acquired (TR = 2250 ms, TE = 3.93 ms, FoV = 256 mm, 176 sagittal slices, voxel size = 1.0 mm3).

O

186 187

183

R O

Functional magnetic resonance imaging

181 182

P

185

179 180

parameters estimated in DCM denote the rate (or speed) of change in 237 neuronal activity that one area exerts over another. DCM can also han- 238 dle sparse-imaging data with discrete temporal timings by using sam- 239 pling functions bridging the gap between the measured data and the 240 predicted neuronal dynamics which are modeled on a continuous 241 time scale (Kiebel et al., 2007; Kumar et al., 2007). 242 As DCMs are computed at the single subject level, we extracted the 243 first eigenvariate of the BOLD time series, adjusted for effects of interest, 244 from 8 regions-of-interest (ROIs) at subject specific coordinates. ROIs 245 were defined as spheres (radius: 4 mm) centered upon individual acti- 246 vation maxima based on individually normalized SPMs. The ROIs 247 consisted of the representations of the hand (M1hand) and foot 248 (M1foot) within the primary motor cortex, the supplementary motor 249 area (SMA), and the ventrolateral premotor cortex (vPMC), i.e., core re- 250 gions of the motor system typically engaged in isolated movements of 251 the upper and lower limbs (Grefkes et al., 2008a; Kapreli et al., 2006; 252 Luft et al., 2002). ROIs for M1 representations of the hand (M1hand) 253 and foot (M1foot) in the left hemisphere were identified using the con- 254 trasts “right hand vs. rest” and “right foot vs. rest”, while M1 ROIs in 255 the right hemisphere were localized using the contrasts “left hand vs. 256 rest” and “left foot vs. rest”, simple unilateral limb movements typically 257 result in unilateral M1 activation in the contralateral hemisphere, ren- 258 dering a reliable detection of M1 ipsilateral to the moving limb extreme- 259 ly difficult. In order to ensure that premotor ROIs were active during 260 both upper and lower limb movements, i.e., the modulated region was 261 significantly engaged in all tasks, premotor ROIs were extracted from 262 a conjunction analyses on the respective contrasts (e.g., right hand vs. 263 rest ∩ right foot vs. rest). 264 All ROIs were defined by functional and anatomical criteria: First re- 265 gion specific activation maxima were located at the group-level using a- 266 priori defined anatomical constraints: (i) M1hand on the rostral wall of 267 the central sulcus at the “hand knob” formation (Yousry et al., 1997), 268 (ii) M1foot at the paracentral lobule (Lotze et al., 2000), (iii) SMA on 269 the medial wall within the interhemispheric fissure between the 270 paracentral lobule (posterior landmark) and the coronal plane running 271 through the anterior commissure (Picard and Strick, 2001), and 272 (iv) vPMC close to the inferior precentral gyrus and pars opercularis of 273 the inferior frontal gyrus (Rizzolatti et al., 2002). Then, individual activa- 274 tion maxima were identified in each and every subject in a search sphere 275 (radius 2 cm) centered around the group coordinate in the respective 276 SPMs, superimposed on the corresponding structural T1-volume. Final- 277 ly, we assumed task-induced activity within the cortical motor system to 278 be driven by premotor areas (vPMC, SMA), which were accordingly de- 279 fined as DCM input regions (DCM-C) (Goldman-Rakic et al., 1992; Wang 280 et al., 2011). The individual coordinates for all ROIs are given in Table 1. 281

D

184

with about 45° flexion of both knees. The feet were free so that the subjects could perform the ankle flexion movements without touching the scanner bed or cushion. Task execution was monitored and documented by an experimenter standing next to the subjects inside the scanner room and period and in the rare incidence of errors (e.g., movement of the wrong side or limb) according trials were discarded prior to subsequent analyses. Note that in each subject baseline scans (“null events”) were free from any detectable movement.

177 178

3

Connectivity models

282

Endogenous connections between the motor ROIs were based on published data on anatomical connectivity in macaque monkeys. Of note, we hypothesized similar endogenous connectivity patterns between premotor areas and M1 representation of the hand (M1hand) and foot (M1foot). We assumed endogenous connections between all areas: SMA and ipsilateral and contralateral M1 (Rouiller et al., 1994), between SMA and ipsilateral (Luppino et al., 1993) as well as contralateral vPMC (Boussaoud et al., 2005), between vPMC and both ipsiand contralateral M1 (Rouiller et al., 1994), as well as homotopic transcallosal connections among M1-M1 (Rouiller et al., 1994), SMASMA (McGuire et al., 1991), and vPMC-vPMC (Boussaoud et al., 2005). As the task-specific modulation of interregional coupling may not necessarily affect all possible anatomical connections, a total of 39 alternative connectivity models representing biologically plausible hypotheses on interregional coupling were constructed. We assumed the motor tasks to directly impact on the activity of all premotor regions (bilateral SMA and vPMC), which were accordingly defined as input regions (DCM-C) in all models (Wang et al., 2011). Therefore, all alternative

283

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300

4

Table 1 Individual fMRI activation maxima used as ROIs for DCM (MNI coordinates; M1H: M1hand, M1F: M1foot).

301

56 46 68 46 74 70 72 52 70 68 72 70 54 70 36 70 62 12

8 6 4 12 14 8 2 18 4 1 4 6 10 20 12 14 9 5

vPMC L

−6 4 −4 2 −6 2 −6 −10 0 −2 −6 −8 −10 −6 −10 2 −4 5

50 50 70 46 72 66 46 72 52 58 50 74 52 72 54 70 60 10

−56 −48 −48 −52 −58 −54 −56 −52 −56 −56 −62 −46 −58 −56 −58 −14 −52 11

vPMC R 6 −8 2 0 −4 4 0 2 −4 8 −16 −2 8 2 −24 0 −2 8

40 44 38 48 44 38 40 42 38 40 28 44 6 40 30 70 39 12

52 52 60 64 62 50 56 34 58 60 48 50 54 58 60 54 55 7

M1H L

6 10 −14 −20 14 8 6 0 4 14 8 2 12 8 −26 8 3 12

318

Statistical analysis of coupling parameters

319 320

Individual endogenous (DCM-A) and task-related (DCM-B) coupling parameters of the most likely, i.e., the “winning”, model were first tested for statistical significance at the group level by means of one-sample t-tests (p b 0.05 false discovery rate (FDR)-corrected for multiple comparisons, Benjamini and Hochberg, 1995). Statistical group-level differences between significant connections were assessed by means of analyses of variance (ANOVAs) and post-hoc t-tests within each experimental session. We were specifically interested whether interactions of M1 representations of the hand and foot area differ dependent on which limb is moved. Therefore, we computed a 3-factorial ANOVA including the factors: LIMB (levels: hand, foot), SIDE (left, right), and CONNECTIONS. To address our hypotheses based on the winner model (model 39, see below), a total number of 12 CONNECTIONS from the DCM-B matrix were entered into the ANOVA: (i) bi-directional connections between bilateral “active” M1 and “inactive” M1 (n = 4) and (ii) connections from bilateral premotor areas onto bilateral M1 representation of the moving limb (n = 8). Please note that “active” refers to the M1 representation contralateral to the moving limb, while “inactive” denotes M1 ipsilateral to the moving limb. Hence, connections onto M1hand during hand movements were compared to respective connections targeting M1foot during foot movements. Importantly, “inactive” as used here in order to simplify terminology does not imply true neuronal inactivity, but rather denotes a relative difference between the activities of the respective M1 regions.

321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343

C

E

315 316 Q3

R

314

R

312 313

O

310 311

C

308 309

N

306 307

U

304 305

M1H R −18 −22 −24 −18 −24 −10 −12 −14 −24 −22 −20 −14 −24 −24 −24 −8 −19 5

Results BOLD activation

62 68 62 62 64 60 64 66 58 64 62 64 62 64 66 64 63 2

36 42 40 34 28 30 −40 32 36 32 42 40 24 40 32 38 30 19

M1F L −26 −20 −22 −20 −12 −18 −6 −18 −18 −22 −18 −20 −16 −12 −20 −16 −18 5

58 56 64 60 70 68 60 64 50 70 54 60 74 62 68 64 63 6

−2 −6 −8 −8 −2 −8 −4 −2 −14 −8 −6 −6 −6 −8 −6 −8 −6 3

M1F R −20 −10 −36 −24 −24 −28 −20 −16 −20 −14 −26 −16 −28 −18 −26 −36 −23 7

64 66 70 62 74 74 74 64 72 78 76 72 76 78 74 56 71 6

6 10 4 10 10 12 6 8 10 10 12 6 10 10 10 10 9 2

−26 −10 −26 −46 −28 −20 −14 −12 −14 −18 −16 −16 −18 −14 −38 −10 −20 10

68 72 70 62 74 72 72 72 72 72 76 72 74 76 66 70 71 3

344 345

The fMRI group analysis (Fig. 2) showed that compared to rest visually paced movements of the right hand were associated with increased BOLD activity in a left-lateralized network comprising left M1hand, SMA, bilateral vPMC, bilateral striate and extrastriate cortex, as well as subcortical regions including left thalamus, left putamen, and right anterior cerebellum (p b 0.05, FWE corrected). Movements of the left hand yielded a similar, yet mirror-reversed network of activity. Unilateral foot movements induced a similar pattern of BOLD signal in visual, premotor, and subcortical areas as suggested by a conjunction analysis including all 4 conditions (Fig. 2, blue-colored voxels). Of note, each condition resulted in significant activation of both vPMC and SMA, confirming that all premotor ROIs used in the present DCM analyses were activated by all conditions. Contrasting hand and foot movements showed limb-movement specific activation in lateral portions of the pre- and postcentral gyrus for hand movements (Fig. 2, red-colored voxels) and medial parts of the precentral gyrus for foot movements (Fig. 2, green-colored voxels, i.e., in accordance with the somatotopic representation of the lower limbs M1foot).

346

Bayesian model selection

364

According to random-effects Bayesian model selection, the “fully modulated” model (assuming connectivity between all ROIs) showed the highest exceedance probability of all tested models for the entire group. It was hence considered the most likely generative model of the data (Fig. 3). Of note, the coupling parameters estimated by the winning model were highly similar to those of all other models with N 3% exceedance probability (i.e., models 21, 22, 23, 24, 29, 30, 31, 32, see Suppl. Fig. 1 for further details on alternative models) as indicated by highly significant correlations between all matrices estimated by these model (all r N 0.9, p b 0.001). Hence, coupling parameters estimated by the winning model were used for all subsequent analyses. With respect to the divergence between prior and posterior parameter distributions, we computed total mean variance explained. On average the winning model explained 60.94% ± SD: 19.71% of the variance, indicating a good fit of predicted and observed responses. Of note, the percentage of variance explained by the winning model was not used for model comparison, which was rather based on the negative free energy to account for both model fit and complexity (please see Penny, 2012 for further details).

365 366

T

317

models needed to include some connection from premotor areas to M1 contralateral to the moving limb, to enable propagation of induced activity to result in descending activation towards the spinal level. Furthermore, as reflected by our hypothesis, we were particularly interested in interactions between different M1 representations, constituting the “minimal” requirements for our most sparse model (model 1, please see Suppl. Fig. 1). The models systematically varied in complexity and number of connections modulated, ranging from sparsely to a fully modulated model with 56 connections (Suppl. Fig. 1). After estimation of all 39 models, we used Bayesian model selection (Penny et al., 2004) to identify the model yielding the highest evidence given the data using a random-effects approach based on negative free energy estimates for given models (Stephan et al., 2009; Penny, 2012). To compute the total mean variance explained by this model we used the spm_dcm_fmri_check.m script provided in the SPM helpline by Karl Friston (2012; https://www.jiscmail.ac.uk/cgi-bin/webadmin? A2=spm;bebd494.1203).

302 303

−38 −34 −36 −32 −36 −38 −30 −28 −40 −36 −40 −30 −38 −36 −32 −34 −35 4

40 18 26 18 22 34 38 62 34 26 44 50 10 30 26 38 32 13

F

SMA R −8 8 2 0 −12 −6 −10 −10 −4 4 −2 2 −12 −2 −8 0 −4 6

O

−8 −4 −8 −8 −14 −10 −4 −2 −4 −8 −10 −8 −4 −10 −8 −14 −8 3

R O

SMA L

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Mean SD

P

Subject

t1:4 t1:5 t1:6 t1:7 t1:8 t1:9 t1:10 t1:11 t1:12 t1:13 t1:14 t1:15 t1:16 t1:17 t1:18 t1:19 t1:20 t1:21

D

t1:3

E

t1:1 t1:2

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363

367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

Regions of Interest

5

Right hand

Left hand

Right foot

Left foot

vPMC SMA

M1F

F

M1H

R O

O

HANDS > FEET HANDS < FEET HANDS FEET

P

Fig. 2. fMRI activation during unilateral limb movements: Conjunction analyses (left image, including all 4 conditions) showed that all conditions resulted in similar activation of bilateral premotor regions of interest (blue clusters, contrast: conjunction of hand and foot movements). During hand movements, M1hand representations were significantly stronger activated than M1foot representations (red clusters, contrast: hand movement N foot movements) and vice versa (green clusters, contrast: hand movement N foot movements). Activation patterns were highly similar yet mirror reversed for movements performed with the left compared to the right hand/foot (p b 0.05, FWE corrected for multiple comparisons; M1H: M1hand, M1F: M1foot).

Endogenous connectivity

Task specific model input

405

385

Endogenous connectivity (DCM-A matrix) refers to the coupling of areas that remained constant across all conditions (i.e., movementindependent coupling). The DCM-A matrix revealed a symmetric network of mostly excitatory couplings, which were significant across our cohort (Fig. 4; p b 0.05, FDR corrected for multiple comparisons). Here, premotor areas featured strong bilateral excitatory input onto each other as well as on both M1hand and M1foot. In contrast, almost all intra- and inter-hemispheric endogenous M1-interactions were of inhibitory nature, except one: M1foot regions exerted facilitatory influences onto each other. T-tests confirmed that for both interhemispheric directions (left → right, right → left) endogenous coupling between the M1foot–M1foot was significantly more positive compared to the corresponding M1hand–M1hand connections (p b 0.001, FDR-corrected for multiple comparisons). No significant difference in couplings between premotor areas and M1 representations of the hand compared to the foot were evident (p N 0.1, FDR corrected for multiple comparisons). Therefore, the most prominent difference in the endogenous motor network for the upper and lower limbs was at the level of interhemispheric processing with excitatory coupling among the foot representations and inhibitory between the M1hand representations.

We assumed the motor tasks to directly impact on the activity of all premotor regions (bilateral SMA and vPMC), which were accordingly defined as input regions (DCM C) in all models (Wang et al., 2011). The task-specific input values across our cohort were: movement of the right hand: left SMA: 0.20 ± 0.19 Hz (mean ± SD), right SMA: 0.15 ± 0.20 Hz, left vPMC: 0.09 ± 0.09 Hz, right vPMC: 0.14 ± 0.16 Hz; movement of the left hand: left SMA: 0.10 ± 0.12 Hz, right SMA: 0.26 ± 0.25 Hz, left vPMC: 0.07 ± 0.07 Hz, right vPMC: 0.17 ± 0.15 Hz; movement of the right foot: left SMA: 0.19 ± 0.20 Hz, right SMA: 0.16 ± 0.17 Hz, left vPMC: 0.08 ± 0.09 Hz, right vPMC: 0.16 ± 0.17 Hz; movement of the left foot: left SMA: 0.15 ± 0.16 Hz, right SMA: 0.25 ± 0.24 Hz, left vPMC: 0.07 ± 0.10 Hz, right vPMC: 0.16 ± 0.16 Hz.

406

Hand-movement specific connectivity

419

Task-related connectivity, i.e., specific changes in neural coupling induced by unilateral movements of a hand or foot, is represented in the DCM-B matrix (Fig. 5; p b 0.05, FDR corrected for multiple comparisons).

420 421

397 398 399 400 401 402 403 404

E

T

C

E

395 396

R

394

R

392 393

N C O

390 391

Bayesian model selection

0.30

U

388 389

Exeedance probability

386 387

D

384

0.25 0.20 0.15 0.10 0.05 0.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

Models Fig. 3. Bayesian model selection: random-effects Bayesian model selection identified model 39 (the “fully connected” model, assuming connectivity between all ROIs) to feature the highest exceedance probability of all tested models for the entire group. On average the winning model explained 60.94% ± SD: 19.71% of the variance, indicating a very good fit of the data by our model.

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

407 408 409 410 411 412 413 414 415 416 417 418

422 423

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

Endogenous connectivity

.01

.02.01

SMA

.01 .01

.02

.02 .03 .03 .01 1 .0

.01

1 .0

.0 1

vPMC

.01 .04

.08

.10 .05

-.02 -.02 M1H -.04 .08 -.03 .08 .06 .05 .03 -.02 -.02 -.02 -.03-.04

M1H

L

Foot-movement specific connectivity

438

O

M1F .01 .02 M1F

R

P < 0.05 (FDR-corrected)

SMA

C

Condition 1: Right hand

T

E

Fig. 4. Endogenous connectivity: The DCM-A matrix revealed a symmetric network of mostly excitatory couplings (Fig. 4; p b 0.05, FDR corrected for multiple comparisons). Of note, interhemispheric processing between M1hand homologues significantly differed from that of M1foot homologues, with inhibitory couplings between bilateral M1hand but excitatory interactions between bilateral M1foot. Note that connections that are not shown did not feature significant coupling parameters across participants (M1H: M1hand, M1F: M1foot).

Movements of the right foot were associated with a similar coupling structure as observed for equivalent connections during hand movements. Accordingly, enhanced excitatory influences on left M1foot were exerted by premotor areas of both hemispheres while activity in the M1hand areas was inhibited by premotor areas and M1foot. Of note, and in contrast to movements of the hands, the M1foot representation ipsilateral to the moving foot exerted an excitatory influence onto the “active” contralateral M1foot representation. Movements of the left foot resulted in a very similar, yet mirror-reverse network of significant interactions.

R O

vPMC

SMA

P

.02 .03 .04 .0 .01 1

424 425

D

.03

.02.02

Movement of the right hand was accompanied by bilaterally enhanced excitatory input from premotor areas (vPMC, SMA) onto left M1hand. In contrast, all premotor areas exerted significant inhibitory influences onto the right M1hand. Notably, the inter-hemispheric influence of the left hand motor area onto the right hand motor area was negative. Likewise, both M1foot received inhibitory input from all premotor areas during movements of the hands (Fig. 5). Movements of the left hand resulted in a similar, yet mirror-reversed pattern of connectivity (Fig. 5), albeit fewer connections survived correction for multiple comparisons. In summary, unilateral hand movements were accompanied by excitatory input from premotor areas onto the “active” M1hand (i.e., M1hand of the active hemisphere), whereas all other primary motor areas (the “inactive” M1hand, as well as bilateral M1foot) were strongly inhibited.

F

6

SMA

vPMC

.001

E

vPMC .13 .10

-.02 -.01 -.02

R

.08

.11

-.02

-.01 -.01 -.02 -.03 -.04 -.03 -.02 -.04 -.02

.08 -.03 -.02 -.01

M1F

C

N U L

M1F

Condition 4: Left foot SMA

vPMC

-.02 -.01

-.004 .09 .09 .09

M1F

-.01

.002

R

SMA

-.03

.06

.02 -.01

-.02

-.02

vPMC

-.02

.01

M1H

-.003

-.04

-.01

M1F

R L

vPMC

-.02 .07 .11 M1H .10 .05 -.005 -.02

-.003

M1F

SMA

-.02 -.04

-.02 -.03 -.02

M1H

.002

-.001

-.03

-.02

M1H -.04

R L

-.03 -.03

M1H

.12

.07

M1H

M1H

.001

-.02

vPMC

M1F

SMA

vPMC

.001

.13

Condition 3: Right foot .001

SMA

vPMC

-.02

O

M1F

L

SMA

R

M1H

Condition 2: Left hand

.01

M1F

R

Fig. 5. Condition specific modulation motor network connectivity: significant modulation of cortical connectivity during unilateral hand movements, as represented by the DCM-B matrix (p b 0.05, FDR corrected for multiple comparisons). All conditions resulted in strong excitatory inputs onto the M1 representation contralateral to the moving respective limb. Also note differences in interhemispheric M1 interactions (M1H: M1hand, M1F: M1foot).

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

426 427 428 429 430 431 432 433 434 435 436 437

439 440 441 442 443 444 445 446 447 448

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

467 468 469 Q8 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486

494 495

Control of lower and upper limb movements

507

Cortical motor networks governing movements of the upper and lower limbs are—at least in principle—similarly organized. Higher motor areas like the premotor cortex (PMC) and the supplementary motor area (SMA) contain somatotopic representations of all extremities (Godschalk et al., 1995; Mitz and Wise, 1987). Dense fiber connections link premotor areas with the respective somatotopic targets within M1 (Rouiller et al., 1994). This similarity in the structural organization is also reflected in the functional organization of the motor system. Previous neuroimaging studies reported analog networks to be activated by comparable movements of the upper and lower limbs (Dettmers et al., 1995; Grafton et al., 1993). For example, isolated movements of the upper and lower limb activate a number of primary and non-primary motor areas such as SMA, PMC, primary sensory cortex (S1), and M1 (Fink et al., 1997; Luft et al., 2002; Miyai et al., 2001).

508

F

We found that motor network dynamics significantly differ between unilateral hand and foot movements. The key finding was that unilateral hand movements are associated with a more pronounced lateralization of connectivity, with a stronger excitatory drive onto the “active” (contralateral) M1hand exerted by premotor areas and pronounced inhibition of the “inactive” (ipsilateral) M1hand compared to movements of the feet. In stark contrast, during unilateral foot movements, the “inactive” M1foot was not inhibited by its homologue or by premotor areas. Rather the “inactive” M1foot exerted a significant excitatory influence onto the “active” M1foot. These differences in cortical motor network dynamics well reflect the differential activation profiles of the motor system for movements of the upper and lower limbs, and may thus underlie effector specific functional specialization.

O

465 466

493

R O

463 464

C

461 462

E

459 460

R

457 458

Discussion

Limb specific differences

R

455 456

Right Hand vs. Foot

N C O

454

SMA

Δ.021 .021

L

Δ.022 .022

M1 H/F

Δ-.011 -.011

Left Hand vs. Foot SMA

SMA

vPMC

U

452 453

P

We next tested whether interactions of the primary motor cortex representations of the hand and foot area differ dependent on which limb is moved. Therefore, we computed a 3-factorial ANOVA including the factors: LIMB (levels: hand, foot), SIDE (left, right), and CONNECTIONS (interhemispheric M1-interactions (n = 4) and premotor–M1 connections (n = 8) see Methods for further details). The ANOVA did not reveal a significant main effect for SIDE (F(1,15) = 0.045, p = 0.834) nor interaction effects for the factor SIDE (SIDE × LIMB: F(1,15) = 0.047, p = 0.831; SIDE × CONNECTIONS: F(11,35) = 1.242, p = 0.305; SIDE × LIMB × CONNECTIONS: F(11,63) = 0.277, p = 0.899, Greenhouse–Geisser corrected), indicating no hemispheric difference in motor system connectivity. Likewise, no significant main effect was evident for LIMB (F(1,15) = 0.006, p = 0.938). In contrast, the interaction LIMB × CONNECTIONS was significant (F(3,39) = 6.881, p = 0.001) suggesting a differential modulation of M1 connectivity depending on which limb (hand or foot) was moved. To identify which connections drove the significant interaction, Fisher's PLSD post-hoc tests were computed, comparing network dynamics during movement of the right hand with interactions underlying movement of the right foot (Fig. 6A) and movements of the left upper limb with left lower limb movements (Fig. 6B). For both leftand right-sided movements we found a significantly stronger interhemispheric inhibition of the “inactive” M1 during hand compared to foot movements (Fig. 6, movement of right hand/foot: p = 0.008; movement of the left hand/foot: p = 0.035). Thus, the “active” M1hand (contralateral to the moving hand) exerted inhibition over the “inactive” M1hand (ipsilateral to the moving hand), whereas no such interhemispheric inhibition was evident during movements of the foot. Finally, a significant interhemispheric excitatory influence from the “inactive” M1 onto the “active” M1 was evident during movements of the foot but not of the hand (Fig. 6, movement of the right hand/ foot: p = 0.050; movement of the left hand/foot: p = 0.003; twosided t-tests). Regarding premotor input onto M1, post-hoc tests revealed excitatory influences from SMA to M1 within the “active” hemisphere to be similar for upper and lower limb movements during both left- and right-sided movements (right hand/foot: p = 0.258; left hand/foot:

487 488

D

450 451

p = 0.285). In contrast, positive coupling of other premotor areas (vPMC within the active and “inactive” hemisphere) with M1 was significantly stronger for right hand compared to foot movements (Fig. 6, p b 0.01). Furthermore, the inhibition of the inactive M1 by premotor areas was stronger for hand compared to foot movements during left and right-sided movements (p = 0.001).

E

Coupling differences between hand and foot movements

T

449

7

vPMC

vPMC

Δ-.013 -.013

vPMC Δ-.022 -.022

Δ-.014 -.014 Δ-.007 -.007

SMA

Δ-.007 -.007

M1 H/F

M1 H/F

R L

Δ-.007 -.007 Δ-.015 -.015

M1 H/F

R

excitation hand > foot inhibition hand > foot excitation foot > hand

Fig. 6. Limb specific differences in cortical connectivity: Comparing the modulation of motor network connectivity between hand and foot movements resulted in highly symmetrical coupling differences for movements of the right compared to the left limbs. Hand movements were accompanied by significantly stronger excitatory input from premotor areas onto the “active” (i.e., contralateral to the moving limb) M1 (orange arrows) and significantly stronger inhibition of the “inactive” (i.e., ipsilateral to the moving limb) M1 by its contralateral homologue (purple arrows) compared to foot movements. During foot movements, the “inactive” M1 exerted a significantly stronger excitatory input onto the “active” M1 compared to hand movements (blue arrows).

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

489 490 491 492

496 497 498 499 500 501 502 503 504 505 506

509 510 511 512 513 514 515 516 517 518 519 520 521

539 540 541 542 543 544 545 Q4 546 547 548 549 550 551 552 553 554 555 556 557 558

The role of premotor input onto M1

560

Premotor areas feature cortical representations of both the upper and lower limb, which are highly interconnected with M1 as well as spinal motor neurons (Burman et al., 2014; Dum and Strick, 2005; Shimazu et al., 2004). Hence, from an anatomical perspective, premotor areas are likely to be involved in the coordination of movements of the hands and feet alike. Electrical stimulation of SMA in macaques evokes both arm and leg movements, with less spatially localized hindlimb responses (Macpherson et al., 1982). In contrast to SMA, electrical stimulation of vPMC neurons primarily results in hand movements, which is paralleled by a larger fraction of vPMC neurons connected to the M1 hand motor area (Dum and Strick, 2005). In the present study we found that SMA and vPMC are both activated during hand and foot movements, with a strong spatial overlap of activation in the conjunction of both tasks (Fig. 2), suggesting that fMRI at the current spatial resolution led to the activation of highly similarly located parts of vPMC and SMA for hand and foot movements alike. Of note, our fMRI analysis did not reveal significantly stronger activation of SMA or vPMC for one of the effectors, implying similar premotor processes to underlie voluntary movements (Fig. 2). However, the functional implications of these signal changes might considerably differ from a network level perspective. In fact, the DCM analysis of the present study demonstrated the influence of premotor regions onto M1 to differ for hand compared to foot movements in two ways. In line with the predominance of vPMC for hand motor function, we found significantly stronger coupling of vPMC with M1hand

569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584

O

C

567 568

N

565 566

U

563 564

R

559

561 562

Interhemispheric M1-interactions

605

Besides neural input from premotor regions, the dynamic adjustment of interhemispheric M1-interactions during movements constitutes a key mechanism underlying the control of isolated voluntary movements. Primary motor representations within the precentral gyrus feature strong structural connections to the contralateral hemisphere via the corpus callosum. Interhemispheric fibers are most dense between homotopic areas (e.g., bilateral M1hand) in both primates and humans (Doron and Gazzaniga, 2008; Wahl et al., 2007). These structural properties of the motor system are mirrored by functional resting-state fMRI data, showing that sub-regions of M1 also show highest levels of functional connectivity with homotopic regions in the contralateral hemisphere (van den Heuvel and Hulshoff Pol, 2010). We found that task-independent interhemispheric M1-connectivity significantly differed for upper and lower limb representations: M1hand inhibited each other, whereas significant facilitatory couplings were evident between bilateral M1foot (Fig. 4). Support for an inhibitory influence between both M1hand stems from studies using double-pulse TMS paradigms, showing a suppression of motor-evoked potentials (MEPs) in one hemisphere upon preceding stimulation of M1hand in the other hemisphere in resting subjects (Ferbert et al., 1992). Interestingly, during movement preparation interhemispheric inhibition from the “inactive” M1hand to the “active” M1hand is reduced (disinhibited) in order to release the planned action whereas inhibition of the “inactive” M1hand is increased to prevent unwanted mirror activity (Duque et al., 2007; Hinder, 2012; Hinder et al., 2010). Similar effects can be also observed in the present study (Fig. 5). Furthermore, our data may also help to explain disinhibition phenomena encountered in patients suffering from brain lesions. Many patients with motor stroke suffer from “mirror movements”, i.e., involuntary co-activation of the unaffected hand during movements of the affected hand. Such effects have been interpreted to be caused by insufficient interhemispheric inhibition targeting the unaffected M1hand (Takeuchi et al., 2012). Interestingly, mirror movements of the lower limb are only rarely observed after stroke (Luft et al., 2005). These differential effects of brain lesions on movement suppression are well in line with the findings of the present connectivity study showing that interhemispheric inhibition was only found between M1hand, while there seems to be no inhibitory influences between M1foot that could be potentially “disinhibited” after a brain lesion. In summary, we observed a complex modulation of cortical interactions causing stronger lateralization of facilitatory drive during hand compared to foot movements. Of note, the total number of significant connections slightly differed comparing movements of the left and

606

F

537 538

O

535 536

R O

533 534

P

531 532

585 586

D

529 530

during hand movements compared to coupling with M1foot during movements performed with a foot (Fig. 6). Second, in accordance with previous DCM-studies on hand movements, strong inhibitory influences were observed from SMA and vPMC onto the “inactive” M1hand (Grefkes et al., 2008a; Pool et al., 2013). However, no significant inhibition of “inactive” M1foot by premotor areas was observed (Fig. 5). A possible explanation for these effector-specific differences in interhemispheric inhibition might lie in a stronger impact of spinal cord circuits for movements of the lower limb. For example, Danner and colleagues recently reported that in paraplegic patients with chronic spinal cord injury (i.e., with deafferented motor cortex) epidural stimulation of the lumbar spinal cord leads to a wide range of coordinated movements patterns of the legs, implying a strong influence of spinal neurons for the motor repertoire of the lower limbs (Danner et al., 2015). Although we did not study locomotor activity but rather voluntary isolated movements, it seems plausible to assume that a stronger impact of spinal sources on lower limb function might be associated with a weaker control at the cortical level, which would be in line with the connectivity findings of the present study.

T

528

C

526 527

E

524 525

Accordingly, we here found overlapping BOLD activity for both hand and foot movements in a number of sensorimotor areas such as SMA, vPMC, posterior parietal cortex, and dorsolateral prefrontal cortex (Fig. 2). At the same time, differences were evident between networks controlling movements of the hands and feet which were most obvious for the clear spatial separation of the M1 representations of the hand and foot (Fig. 2) (Penfield, 1937). At the structural level, the composition of fiber tracts projecting from the cortex to the spinal level differs for the upper and lower extremity. Using diffusion tensor imaging (DTI), Yeo and colleagues reported that portions of the corticospinal tract (CST) projecting to the lower limbs include a significantly higher amount of fibers originating bilaterally from M1foot compared to hand-related CST fiber tracts. In turn, the latter mostly originate from M1hand contralateral to the respective hand (Yeo and Jang, 2012). Of note, influences from uncrossed (ipsilateral) CST fibers are basically absent for hand motor function, at least in non-human primates (Soteropoulos et al., 2011). The differences in structural anatomy discussed above raise the question whether the control of upper and lower limb movements is differentially implemented at the subcortical to spinal level. Indeed, animal studies frequently reported simple periodic unilateral and alternating bi-pedal lower limb movements, e.g., during stepping, to be primarily generated and coordinated on the spinal level (e.g., Drew et al., 1996; Orlovsky et al., 1999). For example, the experimental inactivation of the motor cortex does not impact on the generation of simple locomotion in even terrain in cats (Beloozerova and Sirota, 1993), indicating alternating lower limb movements to be predominantly controlled at the subcortical and spinal level (Zelenin et al., 2011). However, in humans, movements of the legs also strongly depend on cortical influences (Kumral et al., 2002). Here, unilateral movements of the foot are associated with bilateral M1 activity (Miyai et al., 2001; Stippich et al., 2007), which is compatible with a bilateral influence of the M1CST on spinal neurons (Yeo and Jang, 2012). The present study extends current knowledge on the limb motor control physiology by showing that movements of the hand and feet are also mediated by differential interactions between premotor regions and M1, as well as interhemispheric M1 couplings as discussed in more detail below.

R

522 523

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

E

8

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604

607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

Conclusion

665

678

Our data revealed differences in cortical motor network dynamics suggesting the cortical control of upper limb movements to be more lateralized compared to foot movements. Despite similar recruitment of connections, movements of the hand were associated with stronger coupling with premotor areas. Hence, our data are in line with the assumption that hand movements are subjected to a stronger modulation of cortical control, most likely reflecting patterns of daily use, i.e., finetuned hand and finger movements versus locomotion and gait, with the latter also drawing on spinal influences. Such differences in functional network configurations underlying hand and foot movements may also impact on the mechanisms engaged in post-stroke recovery, which needs to be addressed in future studies. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.neuroimage.2015.05.101.

679

References

680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718

Amaro, E., Williams, S.C., Shergill, S.S., Fu, C.H., MacSweeney, M., Picchioni, M.M., Brammer, M.J., McGuire, P.K., 2002. Acoustic noise and functional magnetic resonance imaging: current strategies and future prospects. J. Magn. Reson. Imaging 16, 497–510. Ashburner, J., Friston, K.J., 2005. Unified segmentation. NeuroImage 26, 839–851. Beloozerova, I.N., Sirota, M.G., 1993. The role of the motor cortex in the control of accuracy of locomotor movements in the cat. J. Physiol. 461, 1–25. Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate: a practical and powerfull approach to multiple testing. J. R. Stat. Soc. B 57, 289–300. Boudrias, M.H., Goncalves, C.S., Penny, W.D., Park, C.H., Rossiter, H.E., Talelli, P., Ward, N.S., 2012. Age-related changes in causal interactions between cortical motor regions during hand grip. NeuroImage 59, 3398–3405. Boussaoud, D., Tanne-Gariepy, J., Wannier, T., Rouiller, E.M., 2005. Callosal connections of dorsal versus ventral premotor areas in the macaque monkey: a multiple retrograde tracing study. BMC Neurosci. 6, 67. Burman, K.J., Bakola, S., Richardson, K.E., Reser, D.H., Rosa, M.G., 2014. Patterns of cortical input to the primary motor area in the marmoset monkey. J. Comp. Neurol. 522, 811–843. Danner, S.M., Hofstoetter, U.S., Freundl, B., Binder, H., Mayr, W., Rattay, F., Minassian, K., 2015. Human spinal locomotor control is based on flexibly organized burst generators. Brain 138, 577–588. Dettmers, C., Fink, G.R., Lemon, R.N., Stephan, K.M., Passingham, R.E., Silbersweig, D., Holmes, A., Ridding, M.C., Brooks, D.J., Frackowiak, R.S., 1995. Relation between cerebral activity and force in the motor areas of the human brain. J. Neurophysiol. 74, 802–815. Doron, K.W., Gazzaniga, M.S., 2008. Neuroimaging techniques offer new perspectives on callosal transfer and interhemispheric communication. Cortex 44, 1023–1029. Dresel, C., Castrop, F., Haslinger, B., Wohlschlaeger, A.M., Hennenlotter, A., CeballosBaumann, A.O., 2005. The functional neuroanatomy of coordinated orofacial movements: sparse sampling fMRI of whistling. NeuroImage 28, 588–597. Drew, T., Jiang, W., Kably, B., Lavoie, S., 1996. Role of the motor cortex in the control of visually triggered gait modifications. Can. J. Physiol. Pharmacol. 74, 426–442. Dum, R.P., Strick, P.L., 2005. Frontal lobe inputs to the digit representations of the motor areas on the lateral surface of the hemisphere. J. Neurosci. 25, 1375–1386. Duque, J., Murase, N., Celnik, P., Hummel, F., Harris-Love, M., Mazzocchio, R., Olivier, E., Cohen, L.G., 2007. Intermanual differences in movement-related interhemispheric inhibition. J. Cogn. Neurosci. 19, 204–213. Eickhoff, S.B., Grefkes, C., 2011. Approaches for the integrated analysis of structure, function and connectivity of the human brain. Clin. EEG Neurosci. 42, 107–121.

674 675 676 677

R O

P

672 673

D

670 671

T

668 669

C

666 667

E

660 661

R

658 659

R

656 657

N C O

655

U

653 654

F

664

651 652

Ferbert, A., Priori, A., Rothwell, J.C., Day, B.L., Colebatch, J.G., Marsden, C.D., 1992. Interhemispheric inhibition of the human motor cortex. J. Physiol. 453, 525–546. Fink, G.R., Frackowiak, R.S., Pietrzyk, U., Passingham, R.E., 1997. Multiple nonprimary motor areas in the human cortex. J. Neurophysiol. 77, 2164–2174. Friston, K.J., Harrison, L., Penny, W., 2003. Dynamic causal modelling. NeuroImage 19, 1273–1302. Godschalk, M., Mitz, A.R., van Duin, B., van der Burg, H., 1995. Somatotopy of monkey premotor cortex examined with microstimulation. Neurosci. Res. 23, 269–279. Goldman-Rakic, P.S., Bates, J.F., Chafee, M.V., 1992. The prefrontal cortex and internally generated motor acts. Curr. Opin. Neurobiol. 2, 830–835. Grafton, S.T., Woods, R.P., Mazziotta, J.C., 1993. Within-arm somatotopy in human motor areas determined by positron emission tomography imaging of cerebral blood flow. Exp. Brain Res. 95, 172–176. Grefkes, C., Eickhoff, S.B., Nowak, D.A., Dafotakis, M., Fink, G.R., 2008a. Dynamic intra- and interhemispheric interactions during unilateral and bilateral hand movements assessed with fMRI and DCM. NeuroImage 41, 1382–1394. Grefkes, C., Nowak, D.A., Eickhoff, S.B., Dafotakis, M., Kust, J., Karbe, H., Fink, G.R., 2008b. Cortical connectivity after subcortical stroke assessed with functional magnetic resonance imaging. Ann. Neurol. 63, 236–246. Hinder, M.R., 2012. Interhemispheric connectivity between distinct motor regions as a window into bimanual coordination. J. Neurophysiol. 107, 1791–1794. Hinder, M.R., Schmidt, M.W., Garry, M.I., Summers, J.J., 2010. Unilateral contractions modulate interhemispheric inhibition most strongly and most adaptively in the homologous muscle of the contralateral limb. Exp. Brain Res. 205, 423–433. Hudson, H.M., Griffin, D.M., Belhaj-Saif, A., Cheney, P.D., 2015. Properties of primary motor cortex output to hindlimb muscles in the macaque monkey. J. Neurophysiol. 113, 937–949. Kapreli, E., Athanasopoulos, S., Papathanasiou, M., Van Hecke, P., Strimpakos, N., Gouliamos, A., Peeters, R., Sunaert, S., 2006. Lateralization of brain activity during lower limb joints movement. An fMRI study. NeuroImage 32, 1709–1721. Kiebel, S.J., Kloppel, S., Weiskopf, N., Friston, K.J., 2007. Dynamic causal modeling: a generative model of slice timing in fMRI. NeuroImage 34, 1487–1496. Kumar, S., Stephan, K.E., Warren, J.D., Friston, K.J., Griffiths, T.D., 2007. Hierarchical processing of auditory objects in humans. PLoS Comput. Biol. 3, e100. Kumral, E., Bayulkem, G., Evyapan, D., Yunten, N., 2002. Spectrum of anterior cerebral artery territory infarction: clinical and MRI findings. Eur. J. Neurol. 9, 615–624. Liang, N., Funase, K., Takahashi, M., Matsukawa, K., Kasai, T., 2014. Unilateral imagined movement increases interhemispheric inhibition from the contralateral to ipsilateral motor cortex. Exp. Brain Res. 232, 1823–1832. Lotze, M., Erb, M., Flor, H., Huelsmann, E., Godde, B., Grodd, W., 2000. fMRI evaluation of somatotopic representation in human primary motor cortex. NeuroImage 11, 473–481. Luft, A.R., Smith, G.V., Forrester, L., Whitall, J., Macko, R.F., Hauser, T.K., Goldberg, A.P., Hanley, D.F., 2002. Comparing brain activation associated with isolated upper and lower limb movement across corresponding joints. Hum. Brain Mapp. 17, 131–140. Luft, A.R., Forrester, L., Macko, R.F., McCombe-Waller, S., Whitall, J., Villagra, F., Hanley, D.F., 2005. Brain activation of lower extremity movement in chronically impaired stroke survivors. NeuroImage 26, 184–194. Luppino, G., Matelli, M., Camarda, R., Rizzolatti, G., 1993. Corticocortical connections of area F3 (SMA-proper) and area F6 (pre-SMA) in the macaque monkey. J. Comp. Neurol. 338, 114–140. Macpherson, J.M., Marangoz, C., Miles, T.S., Wiesendanger, M., 1982. Microstimulation of the supplementary motor area (SMA) in the awake monkey. Exp. Brain Res. 45, 410–416. McGuire, P.K., Bates, J.F., Goldman-Rakic, P.S., 1991. Interhemispheric integration: I. Symmetry and convergence of the corticocortical connections of the left and the right principal sulcus (PS) and the left and the right supplementary motor area (SMA) in the rhesus monkey. Cereb. Cortex 1, 390–407. Mitz, A.R., Wise, S.P., 1987. The somatotopic organization of the supplementary motor area: intracortical microstimulation mapping. J. Neurosci. 7, 1010–1021. Miyai, I., Tanabe, H.C., Sase, I., Eda, H., Oda, I., Konishi, I., Tsunazawa, Y., Suzuki, T., Yanagida, T., Kubota, K., 2001. Cortical mapping of gait in humans: a near-infrared spectroscopic topography study. NeuroImage 14, 1186–1192. Murase, N., Duque, J., Mazzocchio, R., Cohen, L.G., 2004. Influence of interhemispheric interactions on motor function in chronic stroke. Ann. Neurol. 55, 400–409. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97–113. Orlovsky, G.D., TG, Grillner, S., 1999. Neuronal control of locomotion. From Mollusc to Man. Oxford U, New York. Penfield, W.B.E., 1937. Somatic motor and sensory representation in the cerebral cortex of man as studied by electrical stimulation. Brain 60, 389–443. Penny, W.D., 2012. Comparing dynamic causal models using AIC, BIC and free energy. NeuroImage 59, 319–330. Penny, W.D., Stephan, K.E., Mechelli, A., Friston, K.J., 2004. Comparing dynamic causal models. NeuroImage 22, 1157–1172. Picard, N., Strick, P.L., 2001. Imaging the premotor areas. Curr. Opin. Neurobiol. 11, 663–672. Pool, E.M., Rehme, A.K., Fink, G.R., Eickhoff, S.B., Grefkes, C., 2013. Network dynamics engaged in the modulation of motor behavior in healthy subjects. NeuroImage 82, 68–76. Pool, E.M., Rehme, A.K., Fink, G.R., Eickhoff, S.B., Grefkes, C., 2014. Handedness and effective connectivity of the motor system. NeuroImage 99, 451–460. Rathelot, J.A., Strick, P.L., 2009. Subdivisions of primary motor cortex based on corticomotoneuronal cells. Proc. Natl. Acad. Sci. 106, 918–923. Rizzolatti, G., Fogassi, L., Gallese, V., 2002. Motor and cognitive functions of the ventral premotor cortex. Curr. Opin. Neurobiol. 12, 149–154.

O

662 663

right hand with a higher number of significantly modulated connections observed during movements of the dominant hand (Fig. 5). This finding is perfectly in line with previous data, which indeed showed that handedness and hand dominance impact on the modulation of connectivity (Pool et al., 2014). From an evolutionary perspective skillful unilateral hand movement reflect a relatively new mammal skill. For example, although cebus and squirrel monkeys live in the same ecological niche and have biomechanically similar hands, only cebus monkeys perform relatively independent finger movements to pick up small objects and manipulate tools (Rathelot and Strick, 2009). Hence, the more complex modulation of cortical networks during hand movements might represent an evolution away from more simple patterns of cortical interactions, allowing the generation of fine-tuned specific muscle activation primarily observed during hand movements.

E

649 650

9

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 Q7 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804

Rouiller, E.M., Babalian, A., Kazennikov, O., Moret, V., Yu, X.H., Wiesendanger, M., 1994. Transcallosal connections of the distal forelimb representations of the primary and supplementary motor cortical areas in macaque monkeys. Exp. Brain Res. 102, 227–243. Seto, E., Sela, G., McIlroy, W.E., Black, S.E., Staines, W.R., Bronskill, M.J., McIntosh, A.R., Graham, S.J., 2001. Quantifying head motion associated with motor tasks used in fMRI. NeuroImage 14, 284–297. Shimazu, H., Maier, M.A., Cerri, G., Kirkwood, P.A., Lemon, R.N., 2004. Macaque ventral premotor cortex exerts powerful facilitation of motor cortex outputs to upper limb motoneurons. J. Neurosci. 24, 1200–1211. Soteropoulos, D.S., Edgley, S.A., Baker, S.N., 2011. Lack of evidence for direct corticospinal contributions to control of the ipsilateral forelimb in monkey. J. Neurosci. 31, 11208–11219. Stephan, K.E., Penny, W.D., Daunizeau, J., Moran, R.J., Friston, K.J., 2009. Bayesian model selection for group studies. NeuroImage 46, 1004–1017. Stippich, C., Blatow, M., Durst, A., Dreyhaupt, J., Sartor, K., 2007. Global activation of primary motor cortex during voluntary movements in man. NeuroImage 34, 1227–1237. Takeuchi, N., Oouchida, Y., Izumi, S., 2012. Motor control and neural plasticity through interhemispheric interactions. Neural Plast. 2012, 823285. van den Heuvel, M.P., Hulshoff Pol, H.E., 2010. Specific somatotopic organization of functional connections of the primary motor network during resting state. Hum. Brain Mapp. 31, 631–644.

Volz, L.J., Sarfeld, A.S., Diekhoff, S., Rehme, A.K., Pool, E.M., Eickhoff, S.B., Fink, G.R., Grefkes, C., 2015. Motor cortex excitability and connectivity in chronic stroke: a multimodal model of functional reorganization. Brain Struct. Funct. 220, 1093–1107. Wahl, M., Lauterbach-Soon, B., Hattingen, E., Jung, P., Singer, O., Volz, S., Klein, J.C., Steinmetz, H., Ziemann, U., 2007. Human motor corpus callosum: topography, somatotopy, and link between microstructure and function. J. Neurosci. 27, 12132–12138. Wang, L.E., Fink, G.R., Diekhoff, S., Rehme, A.K., Eickhoff, S.B., Grefkes, C., 2011. Noradrenergic enhancement improves motor network connectivity in stroke patients. Ann. Neurol. 69, 375–388. Weiss, C., Nettekoven, C., Rehme, A.K., Neuschmelting, V., Eisenbeis, A., Goldbrunner, R., Grefkes, C., 2013. Mapping the hand, foot and face representations in the primary motor cortex—retest reliability of neuronavigated TMS versus functional MRI. NeuroImage 66, 531–542. Yeo, S.S., Jang, S.H., 2012. Differences between the somatotopic corticospinal tract for the fingers and toes in the human brain. NeuroRehabilitation 31, 395–399. Yousry, T.A., Schmid, U.D., Alkadhi, H., Schmidt, D., Peraud, A., Buettner, A., Winkler, P., 1997. Localization of the motor hand area to a knob on the precentral gyrus. A new landmark. Brain 120 (Pt 1), 141–157. Zelenin, P.V., Deliagina, T.G., Orlovsky, G.N., Karayannidou, A., Dasgupta, N.M., Sirota, M.G., Beloozerova, I.N., 2011. Contribution of different limb controllers to modulation of motor cortex neurons during locomotion. J. Neurosci. 31, 4636–4649.

F

805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826

L.J. Volz et al. / NeuroImage xxx (2015) xxx–xxx

O

10

U

N

C

O

R

R

E

C

T

E

D

P

R O

849

Please cite this article as: Volz, L.J., et al., Differential modulation of motor network connectivity during movements of the upper and lower limbs, NeuroImage (2015), http://dx.doi.org/10.1016/j.neuroimage.2015.05.101

827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848