Connecting the dots: Harnessing dual-site transcranial magnetic stimulation to assess the causal influence of medial frontal areas on the motor cortex

Dual-site transcranial magnetic stimulation (TMS) has been widely employed to investigate the influence of cortical structures on the primary motor cortex (M1). Here, we leveraged this technique to probe the causal influence of two key areas of the medial frontal cortex, namely the supplementary motor area (SMA) and the medial orbitofrontal cortex (mOFC), on M1. We show that SMA stimulation facilitates M1 activity across short (6 and 8 ms) and long (12 ms) inter-stimulation intervals, putatively recruiting cortico-cortical and cortico-subcortico-cortical circuits, respectively. Crucially, magnetic resonance imaging revealed that this facilitatory effect depended on a key morphometric feature of SMA: individuals with larger SMA volumes exhibited more facilitation from SMA to M1. Notably, we also provide evidence that the facilitatory effect of SMA stimulation at short intervals did not arise from spinal interactions of volleys descending simultaneously from SMA and M1. On the other hand, mOFC stimulation moderately suppressed M1 activity at both short and long intervals, irrespective of mOFC volume. These results suggest that dual-site TMS is an interesting tool to study the differential influence of SMA and mOFC on M1 activity, paving the way for the multi-modal assessment of these fronto-motor circuits in health and disease. Key points Dual-site TMS has been widely employed to investigate effective connectivity between cortical structures and the primary motor cortex (M1). Here, we probed the causal influence of the supplementary motor area (SMA) and the medial orbitofrontal cortex (mOFC) on M1 activity. SMA stimulation facilitates M1 activity at both short and long inter-stimulation intervals; this facilitatory effect is related to SMA volume. mOFC stimulation moderately suppresses M1 activity, independent of mOFC volume. The findings pave the way for multi-modal assessment of fronto-motor circuits in health and disease.

conditioning stimulation is used to pre-activate the targeted area, while a second, test 96 stimulation is applied over M1 with another coil to elicit a MEP and assess the nature of the 97 influence (i.e., facilitatory or suppressive) of the pre-activated area on corticospinal 98 excitability. Potentiation of conditioned MEP amplitudes (i.e., relative to unconditioned 99 MEPs) reflects a facilitatory influence of the pre-activated area, whereas a reduction of 100 conditioned MEPs reflects a suppressive effect. Interestingly, varying the inter-stimulation 101 interval allows to probe different circuits, with short inter-stimulation intervals (e.g., between 102 4 and 8 ms) recruiting cortico-cortical circuits preferentially, and longer ones (e.g., higher 103 than 10 ms) recruiting more indirect circuits presumably funneling through subcortical 104 structures (Neubert et al., 2010). 105 Over the past two decades, dual-site ppTMS has been widely used in humans, with 106 studies probing the causal influence of several areas of the premotor cortex (Koch et al., 2006; 107 Davare et al., 2008), of the lateral prefrontal cortex (Neubert et  inter-stimulation intervals (i.e., 10 to 15 ms). In fact, the SMA projects to M1 through 117 multiple cortico-subcortico-cortical circuits (Nachev et al., 2008), some of which exert a net 118 facilitatory influence on motor activity (e.g., the direct basal ganglia pathway) and some of 119 which play a suppressive role (e.g., the indirect and hyperdirect pathways). Interestingly, 120 ppTMS studies focusing on the preSMAi.e., another key area of the medial frontal cortex -121 with intervals of 12 ms, reported a potentiation of conditioned MEP amplitudes, which 122 strongly covaried with white matter density in preSMA-basal ganglia-M1 circuits (Mars et al., 123 2009;Neubert et al., 2010). Taken together, these two sets of findings suggest that, when 124 applied over preSMA at such intervals, ppTMS recruits circuits that have a facilitatory 125 influence on M1 and funnel through the basal ganglia. A first goal of the present study is to 126 shed light on to the influence of SMA stimulation on M1 at such intervals, by testing the idea 127 that SMA-originating circuits bear a similar facilitatory influence on motor activity. 128 As such, ppTMS studies targeting SMA with short inter-stimulation intervals (6 to 8 ms) 129 have reported a potentiation of MEP amplitudes, which has been assumed to reflect the 130 operation of cortico-cortical, facilitatory circuits from SMA to M1 (Arai et al., 2011(Arai et al., , 2012 Importantly though, other observations challenge the validity of this assumption. Indeed, like 136 M1, SMA has pyramidal cells that project to the spine (Dum & Strick 1996) and, in certain 137 contexts, unique stimulation of SMA with single-pulse TMS can evoke MEPs (Spieser et al., 138 2013;Entakli et al., 2014), suggesting that these pyramidal cells can also recruit motoneurons. 139 Thus, it is possible that the MEP potentiation reported in ppTMS studies using short intervals 140 reflects the summation of volleys descending from SMA and M1 and converging at close 141 times on motoneurons. In other words, it is currently unclear whether this potentiation can be 142 taken as a pure measure of effective connectivity between SMA and M1 or not. Addressing 143 this issue is fundamental for any investigation targeting motor areas (e.g., the dorsal or the 144 ventral premotor cortex; Davare et al., 2009, Koch et al., 2006, which, for the most part, 145 present corticospinal projections (Dum & Strick, 1991). This is the second goal of the current 146 study. Specifically, we tested the effect of SMA conditioning on MEP amplitudes using a 147 very short inter-stimulation intervals of 1 ms. The rationale here is that a 1 ms interval would 148 be too short for a MEP potentiation to result from the recruitment of cortico-cortical circuits 149 (Aizawa & Tanji, 1994;Tokuno & Nambu, 2000); any potentiation occurring at this interval 150 would instead provide evidence for a summation of neural inputs occurring at the spinal level. 151 As mentioned above, mOFC is another major area of the medial frontal cortex (Amodio 152 & Frith, 2006). Strikingly though, ppTMS has never been used to probe the influence of this 153 area on M1, potentially because of the presumed difficulty of reaching it with magnetic fields. 154 Hence, it is currently unknown whether TMS could be exploited to probe effective 155 connectivity between mOFC and M1 and, if so, what would be the nature of the influence of 156 the recruited circuits on motor activity. In fact, while former investigations on caudal areas of 157 the medial frontal cortex (i.e., SMA and preSMA) generally reported a facilitatory influence 158 on M1, ppTMS studies on more rostral areas of the frontal lobe, such as the dorsolateral PFC, 159 revealed the operation of suppressive circuits (Wang et al., 2020 such as cortical thickness, gray matter volume, and microstructural properties of cerebral 168 white matter, can significantly influence cortical excitability (Klöppel et al., 2008;List et al., 169 2013). Therefore, in the current study, we exploited structural MRI to characterize the 170 relationship between the morphometric features of SMA and mOFC, particularly their cortical 171 volume, and the facilitatory/suppressive effect of their stimulation on M1 activity. 172 Overall, the current study addresses four main goals. The first one is to provide insight 173 into the influence of SMA stimulation on M1 activity with long inter-stimulation intervals, 174 which are thought to recruit cortico-subcortico-cortical circuits. As a second objective, we 175 aim to clarify whether the MEP potentiation reported in studies targeting SMA and M1 with 176 short intervals can be taken as a pure measure of cortico-cortical connectivity between these 177 areas or if it could in part reflect the summation of volleys descending from those on 178 motoneurons. Moreover, we seek to test the feasibility of exploiting ppTMS to probe effective 179 connectivity between mOFC and M1 and, relatedly, to determine the influence of mOFC 180 stimulation on motor activity. As a last goal, we also investigated whether these TMS-based The MNI coordinates exploited to Initially localize SMA and mOFC were x = -8, y = -9, 257 z = 77 and x = − 7, y = 71, z = -4, respectively (Codol et al., 2020). These two locations were 258 then slightly adjusted for each subject using the Visor software, so that they corresponded to 259 the point where the scalp-to-cortex distance was minimal. Following this procedure, the MNI 260 coordinates for SMA and mOFC locations were x = -7.9 ± 0.3, y = -7.4 ± 0.8, z = 82.9 ± 1 and 261 x = -9.4 ± 0.7, y = 72.6 ± 0.5, z = 8.2 ± 1.7, respectively (mean ± standard error (SE) of the 262 group; see Figure 1 and Table 1   exploited to initially localize SMA and mOFC were x = -8, y = -9, z = 77 and x = − 7, y = 71, z = -4, 272 respectively. These two locations were then slightly adjusted for each subject using the Visor 273 software, so that they corresponded to the point where the scalp-to-cortex distance was minimal.

275
For both SMA and mOFC stimulation, the center of the coil was placed over the 276 corresponding target location (see Figure 1). In SMA blocks, the coil was held tangential to 277 the scalp with the handle pointing at a -100° angle away from the midsagittal line (i.e., in the 278 counter-clockwise direction), resulting in a medio-lateral current flow within the cortex 279 ( Figure 1). This coil position was chosen based on a previous experiment showing that it 280 allows the most optimal recruitment of SMA neurons (Arai et al., 2012). In mOFC blocks, the 281 coil was held tangential to the forefront with the handle directed upward and parallel to the 282 midsagittal line (Codol et al., 2020), resulting in a downward current flow at the cortical level. 283

Stimulation intensities 284
Once the real and the adjusted hotspots were found (see above), we determined the 285 resting motor threshold (rMT) for both locations. The rMT was defined as the lowest 286 stimulation intensity (expressed in percentage of maximal stimulator output (%MSO)) 287 required to evoke MEPs of 50 μV amplitude on 5 out of 10 consecutive trials in the relaxed 288 FDI muscle (Rossini et al., 1994(Rossini et al., , 2015. The rMTs for the real and the adjusted hotspots were 289 39.9 ± 1.5 % and 44.6 ± 1.9 %MSO, respectively (see Table 2

294
These rMT values were exploited to determine the stimulation intensities to be used for 295 the rest of the experiment. In SMA blocks, the M1 coil was positioned over the adjusted 296 hotspot. Hence, we based on the rMT obtained at this location to define the stimulation 297 intensity for M1; we stimulated M1 at 120 % of this rMT (Derosiere et al., 2017a(Derosiere et al., , 2017b. 298 Conversely, in mOFC blocks, the M1 coil could be easily positioned over the real hotspot and 299 we thus stimulated at 120 % of the rMT obtained for the real hotspot. Finally, CS intensity 300 was set at 120 % of the rMT obtained for the real hotspot, both in SMA and in mOFC blocks 301 (Brown et al., 2019). 302 303

Inter-stimulation intervals and blocks 304
As mentioned in the Introduction section, the goal of the present study was fourfold. 305 First, we aimed to test the influence of SMA stimulation on M1 activity with long inter-306 stimulation intervals, presumed to recruit cortico-subcortico-cortical circuits. To this aim, we 307 provide evidence for a summation of neural inputs occurring at the spinal level. We also 317 included other short inter-stimulation intervals of 4, 6 and 8 ms in the experiment to be able to 318 compare the effect obtained on MEP amplitudes when using the 1 ms interval vs. when 319 exploiting more classical intervals. Finally, we aimed to test the feasibility of exploiting 320 ppTMS to probe effective connectivity between mOFC and M1 and, relatedly, to determine 321 the influence of mOFC stimulation on motor activity. Given the current lack of data regarding 322 the latter issue, we exploited all of the intervals mentioned above in mOFC blocks too. Note 323 that, contrary to SMA, mOFC does not present corticospinal projections. Hence, the use of a 1 324 ms interval in mOFC blocks allowed us to verify that any effect of SMA stimulation on MEPs 325 using this interval was specific to SMA. Altogether, the experiment involved 6 inter-326 stimulation intervals, both in SMA and in mOFC blocks: 1, 4, 6, 8, 12 and 15 ms. 327 The experiment was divided into 10 blocks of 42 trials (i.e., 5 SMA blocks and 5 mOFC 328 blocks). Each block comprised trials with single-pulse (i.e., TS only) and paired-pulse TMS 329 (i.e., CS+TS with the 6 intervals mentioned above), occurring in a randomized order. As such, 330 within each block, a total of 6 trials was recorded for each of the 7 conditions (i.e., single- .025; BF10 = 2.42). Of note, these effects did not reach statistical significance anymore if 474 corrected for multiple testing by a conservative Bonferroni correction (i.e., p-value threshold 475 = .05 / 6 = .008). However, as highlighted above, BF analyses revealed a moderate to strong 476 evidence for the presence of a facilitatory effect at these three inter-stimulation intervals 477 (average BF10 = 2.76 ± 0.81), indicating that the data were 2.76 times more likely to show 478 facilitation than no difference from the constant value of 1. As such, 75 %, 70 % and again 75 479 % of subjects had a MEP ratio above 1 at 6, 8 and 12 ms, respectively, indicating that SMA 480 stimulation potentiated MEP amplitudes for most of the subjects at these intervals (see Figure  481 2.C). Finally, another interesting finding revealed by the t-tests was the presence of a 482 significant suppressive influence of SMA stimulation on MEP amplitudes for the 1 ms

554
Over the past two decades, dual-site ppTMS has been widely used in humans, with 555 studies probing the causal influence of multiple frontal and parietal areas on M1. However, 556 several important questions currently remain unanswered, notably regarding the application of 557 this approach to key areas of the medial frontal cortexincluding SMA and mOFC. The 558 present study directly addressed three of these issues. First, we aimed to provide insight into 559 the influence of SMA stimulation on M1 activity with long inter-stimulation intervals (12 to 560 15 ms), which are thought to recruit cortico-subcortico-cortical circuits (Neubert et al., 2010). 561 Our data reveal that SMA stimulation significantly potentiates MEP amplitudes with a 12 ms 562 interval, indicating the recruitment of circuits that bear a facilitatory influence on M1. Second, 563 we sought to clarify whether the MEP potentiation reported in studies targeting SMA and M1 564 with short intervals (6 to 8 ms) can be taken as a pure measure of cortico-cortical connectivity 565 between these areas or whether it might in part reflect the summation of descending volleys 566 on motoneurons at the spinal level. Here, we were able to replicate the MEP potentiation 567 previously observed for such intervals. More importantly, our data show that this facilitation 568 does not occur when using a very short interval of 1 ms, assumed to recruit spinal circuits. 569 Rather, we found a suppressive influence of SMA stimulation on MEP amplitudes at this 570 interval. Finally, we tested the feasibility of exploiting ppTMS to probe effective connectivity 571 between mOFC and M1 and, relatedly, we determined the influence of mOFC stimulation on 572 motor activity. We found that mOFC stimulation induced a moderate suppressive effect on 573 MEP amplitudes with both short and long inter-stimulation intervals. Interestingly, mOFC 574 stimulation did not alter MEP amplitudes with the 1 ms interval, suggesting that the 575 suppressive effect observed with this interval was specific to SMA conditioning. 576 As mentioned above, SMA stimulation induced a significant potentiation of MEP 577 amplitudes with the 12 ms interval. This finding is not trivial because SMA projects to M1 578 through multiple cortico-subcortico-cortical circuits (Nachev et al., 2008;Accolla et al., 2016;579 Oswal et al., 2021), some of which exert a net facilitatory influence on motor activity (e.g., 580 the direct pathway of the basal ganglia) and others of which play a suppressive role (e.g., the 581 indirect and hyperdirect pathways). One possibility is that SMA stimulation preferentially 582 recruits the direct pathway of the basal ganglia. In this pathway, areas of the frontal cortex 583 (including SMA) rely on their projections to the striatum to inhibit the internal segment of the 584 globus pallidus, which in turn suppresses neural activity in the subthalamic nucleus 585 (Alexander & Crutcher, 1990;Aron et al., 2007;Calabresi et al., 2014;Niranjan et al., 2018). 586 Since the latter structure exerts a suppressive influence on the motor system (Frank, 2006;587 Aron et al., 2016;Quartarone et al., 2020), the recruitment of this whole circuit ultimately 588 leads to a disinhibition of M1, putatively explaining the MEP potentiation observed at 12 ms 589 interval. Interestingly, the potentiation did not reach statistical significance when a 15 ms 590 interval was used, suggesting that the facilitatory effect uncovered here is interval-dependent, 591 with an optimal temporal window of about 12 ms. The latter observation is relevant for future 592 ppTMS studies aimed at investigating these circuits. 593 Of note, we were able to replicate the MEP potentiation previously observed for intervals 594 However, cortical volume alone does not account for all aspects of this inter-individual 649 variability. Notably, we did not find any association between mOFC cortical volume and its 650 suppressive influence on M1 activity. In fact, it is likely that the suppressive influence of 651 mOFC stimulation on M1 occurs indirectly through polysynaptic connections (Derosiere & 652 Duque, 2020), thereby reducing the contribution of mOFC cortical volume to inter-individual 653 differences in effective connectivity. Future studies should explore the potential contribution 654 of other anatomical features to inter-individual variability in effective connectivity (Neubert 655 et al., 2010). issue warrants further studies, as the application of a high intensity stimulation to this 668 orbitofrontal location may be particularly uncomfortable for the subjects. A second, critical 669 aspect to consider is the coil placement. This is especially true when stimulating SMA due to 670 its spatial proximity with M1. Here, we used a neuronavigation system to target the MNI 671 coordinates of the SMA based on individual MRI images. In the majority of the subjects (n = 672 12 / 20), we had to slightly adjust the M1 stimulation site (i.e., the hotspot) and its 673 corresponding rMT. To the best of our knowledge, this type of adjustment has never been 674 reported before, despite the existence of several ppTMS studies on areas lying close to M1. 675 For the sake of transparency, we believe that future studies should systematically report any 676 adjustment of coil locations. 677 Overall, the current study shows that SMA and mOFC conditioning exerts interval-and 678 region-specific facilitatory and suppressive influences on motor activity. Our findings pave 679 the way for both fundamental and clinical investigations aimed at understanding the causal 680 role of these areas in the modulation of motor activity, as may occur in motor planning, 681 decision-making, and inhibitory control, in which SMA and mOFC play a central role.