Transcranial direct current stimulation over the sensory‐motor regions inhibits gamma synchrony

Abstract

Transcranial direct current stimulation (tDCS) is a noninvasive brain stimulation technique able to induce plasticity phenomena. Although tDCS application has been spreading over a variety of neuroscience domains, the mechanisms by which the stimulation acts are largely unknown. We investigated tDCS effects on cortical gamma synchrony, which is a crucial player in cortical function. We performed a randomized, sham‐controlled, double‐blind study on healthy subjects, combining tDCS and magnetoencephalography. By driving brain activity via 40 Hz auditory stimulation during magnetoencephalography, we experimentally tuned cortical gamma synchrony and measured it before and after bilateral tDCS of the primary sensory‐motor hand regions (anode left, cathode right). We demonstrated that the stimulation induces a remarkable decrease of gamma synchrony (13 out of 15 subjects), as measured by gamma phase at 40 Hz. tDCS has strong remote effects, as the cortical region mostly affected was located far away from the stimulation site and covered a large area of the right centro‐temporal cortex. No significant differences between stimulations were found for baseline gamma synchrony, as well as early transient auditory responses. This suggests a specific tDCS effect on externally driven gamma synchronization. This study sheds new light on the effect of tDCS on cortical function showing that the net effect of the stimulation on cortical gamma synchronization is an inhibition.

1. INTRODUCTION

Transcranial direct current stimulation (tDCS) is an easy to use, safe and low‐cost noninvasive neurostimulation technique. Its applications as neuromodulatory tool have been increasing in both neuroscientific and clinical settings (Assenza et al., 2016; Brunoni et al., 2012; Di Lazzaro et al., 2014; Di Pino et al., 2014; Di Pino, Pellegrino, Capone, & Di Lazzaro, 2014; Poreisz, Boros, Antal, & Paulus, 2007; Ruff, Ugazio, & Fehr, 2013). tDCS consists of a weak constant current delivered over the scalp with two electrodes. The continuous electric field depolarizes neurons by changing their membrane potential and, hereby, their excitability and spontaneous firing rate (Ardolino, Bossi, Barbieri, & Priori, 2005; Zaghi, Acar, Hultgren, Boggio, & Fregni, 2010). This effect has been widely investigated in pericentral regions corresponding to the hand sensory‐motor cortex, where tDCS induces excitability changes and triggers synaptic plasticity‐like mechanisms (Di Lazzaro et al., 2013; Lang et al., 2011; Nitsche et al., 2008; Nitsche & Paulus, 2000; Ziemann et al., 2008). The hand sensory‐motor cortex under the anode shows a local enhancement of excitability (LTP‐like plasticity), whereas the sensory‐motor cortex under the cathode shows a decrease of excitability (LTD‐like plasticity). However, tDCS‐cerebral cortex interaction is far more complex for at least the four following reasons. Firstly, in addition to the well‐known effect on cortical excitability, tDCS has an impact on several domains of brain function, including oscillatory activity, brain synchrony and hemodynamic processes (Herrmann, Rach, Neuling, & Strüber, 2013; Hoy, Bailey, Arnold, & Fitzgerald, 2015; Lang et al., 2005; Notturno, Marzetti, Pizzella, Uncini, & Zappasodi, 2014; Pellegrino et al., 2018; Stagg & Nitsche, 2011). Secondly, tDCS has remote effects in areas distant from the cortex underneath the stimulating electrodes. These effects are not specific to tDCS, but are also elicited by other brain stimulation techniques and can be investigated by combining brain stimulation with neuroimaging (Bergmann, Karabanov, Hartwigsen, Thielscher, & Siebner, 2016; Shafi, Westover, Fox, & Pascual‐Leone, 2012; Siebner et al., 2009). Remote effects have been attributed to interactions between the targeted brain area and connected areas forming a functional network (Fischer et al., 2017; Fox et al., 2014; Giambattistelli et al., 2014; Lindenberg, Nachtigall, Meinzer, Sieg, & Flöel, 2013; Peña‐Gómez et al., 2012). tDCS‐related remote effects have been demonstrated with a number of different read‐outs of regional brain function, including BOLD signal (Polanía, Paulus, & Nitsche, 2012), measures of cortical excitability (Boros, Poreisz, Münchau, Paulus, & Nitsche, 2008; Lang, Nitsche, Paulus, Rothwell, & Lemon, 2004) and signatures of cortical activity and rhythms measured by electroencephalography and magnetoencephalography (Notturno et al., 2014; Pellegrino, Maran, et al., 2018). Thirdly, tDCS stimulation is a bipolar technique and its effects depend on the activity and position of both anode and cathode. Finally, the effect of tDCS is dependent upon neuron orientation and on the spatial folding of the cortex (Miranda, Lomarev, & Hallett, 2006).

In this study we focused on investigating the effect of bi‐hemispheric bipolar tDCS of the primary sensory‐motor hand regions on gamma synchronization in cortical areas distant from the stimulation sites. The bipolar bi‐hemispheric montage of this study is particularly reliable and is becoming very popular for clinical applications (Assenza et al., 2016; Di Lazzaro et al., 2014). The opposite effect of cathode and anode, together with the interhemispheric reciprocal inhibition, result in an overall improved stimulation (Assenza et al., 2016; Di Lazzaro et al., 2014; Lindenberg et al., 2013; Mahmoudi et al., 2011; Vines, Cerruti, & Schlaug, 2008).

Our interest in gamma synchrony is related to the role that it plays in a wide spectrum of brain functions, ranging from cognition to neuropsychiatric conditions such as schizophrenia and stroke plasticity (Fries, 2009; Kwon et al., 1999; Pellegrino et al., 2012). Previous studies on the effect of tDCS on gamma synchronization are not univocal. For instance, it has been reported that anodal tDCS increases (Hoy et al., 2015; Wilson, McDermott, Mills, Coolidge, & Heinrichs‐Graham, 2017), reduces (Hanley, Singh, & McGonigle, 2016) or does not affect (Marshall, Esterer, Herring, Bergmann, & Jensen, 2016) visual gamma synchronization.

A reliable way to experimentally investigate gamma synchrony is by listening to a sound at regular gamma frequency, typically at 40 Hz. Brain activity becomes progressively aligned in phase with the sound and reaches a stable power increase at 40 Hz (40 Hz auditory entrainment or Auditory Steady State Responses–ASSR). Forty hertz entrainment occurs maximally in the primary auditory cortices (Roß, Borgmann, Draganova, Roberts, & Pantev, 2000; Roß, Picton, & Pantev, 2002; Santarelli & Conti, 1999), is right hemisphere dominant (Ross, Draganova, Picton, & Pantev, 2003; Ross, Herdman, & Pantev, 2005) and provides a reliable estimate of cerebral cortex ability to become synchronized in gamma band (Brenner et al., 2009; Thut, Schyns, & Gross, 2011).

Given that auditory cortex and hand sensory‐motor areas are interconnected (Fischer et al., 2017; Zatorre, Chen, & Penhune, 2007), we expected bilateral sensory‐motor tDCS to influence gamma synchronization of the temporal regions. To test this hypothesis, we applied bi‐hemispheric tDCS via two skin electrodes placed over the right and left sensory‐motor cortices and recorded brain activity with magnetencephalography after tDCS to probe the effects on auditory steady state responses (ASSR) evoked by external 40 Hz auditory entrainment. Although 40 Hz auditory entrainment is maximally expressed in the auditory areas, we sampled and investigated the effect of tDCS over the entire cortex.

2. MATERIALS AND METHODS

2.1. Participants and experimental design

The study was approved by the Ethic Committee of the Province of Venice and all participants provided a written informed consent. All the procedures were performed in agreement with the 1964 Helsinki declaration and its later amendments. We recruited subjects aged between 20 and 50 years. Exclusion criteria were (a) hearing deficiency, (b) any medication acting on the central nervous system, (c) neuropsychiatric disorders or other medical conditions that may affect brain function.

Fifteen right‐handed healthy subjects (mean age = 28.8 ± 3 (2 SE); 12 F; Oldfield's Edinburg Inventory (91.13 ± 6.8) (Oldfield, 1971)) took part in a paired, randomized, sham‐controlled, double‐blind study (Figure 1). Apart from one individual, all participants were naïve to tDCS. Each participant underwent bi‐hemispheric tDCS stimulation twice (sham and real), in two different sessions performed at least 20 hr far apart. The order of the tDCS stimulation (sham and real) was counterbalanced across subjects. All tDCS procedures were performed in accordance to current guidelines (Antal et al., 2017). Forty hertz auditory entrainment was measured in the MEG scanner soon before and after each bi‐hemispheric tDCS. Since gamma synchrony can be influenced by sleepiness, all participants were required to keep their regular wake/sleep cycle before participation. The Stanford sleepiness scale (SSS; Hoddes, Zarcone, Smythe, Phillips, & Dement, 1973) was used to monitor the vigilance and was administered at four different time points all along the session (before starting the experiment, before tDCS, after tDCS, at the end of the procedures). A high‐resolution anatomical MRI scan was acquired for each subject to build the head model for MEG data analysis. The MRI was performed after the last MEG scan to prevent MEG magnetization artifacts.

Figure 1.

Experimental design. Panel (a) auditory stimulus. An amplitude modulated tone was generated with the following parameters: Carrier frequency = 1,000 Hz, Amplitude modulating frequency = 40 Hz. Panel (b) the paradigm employed to induce 40 Hz synchrony consisted of 180 repetitions lasting 2 s each. Each trial included 1 s of pause followed by a 40 Hz amplitude‐modulated (AM) tone, so that the stimulus onset asynchrony was 2 s. Panel (c) each participant underwent bipolar bi‐hemispheric tDCS applied to the left and right pericentral regions twice (Real and Sham), in two different sessions at least 20 hr far apart. Cortical gamma synchrony was measured in MEG before and after tDCS stimulation. For the purpose of source imaging, an anatomical MRI was also acquired for each participant [Color figure can be viewed at http://wileyonlinelibrary.com]

2.2. MEG data acquisition and ASSR paradigm

MEG scans were performed at the MEGLab of the IRCCS San Camillo Hospital in Venice (https://sites.google.com/site/meglabsc/) in a quiet shielded room with a CTF‐MEG‐system (MISL, Vancouver, Canada) equipped with 275 gradiometers. Additional bipolar electrodes were placed to detect eye movements and to record the cardiac electric activity (EKG). The sampling rate was set to 1,200 Hz. Head position within the dewar was continuously monitored with three localization coils placed on anatomical landmarks (nasion, left and right ear canals) and tracked by the CTF Continuous Head Localization system. Subjects were lying down in supine position, with their eyes closed and were asked to relax, without paying too much attention to the stimulation.

The 40 Hz sound was delivered during magnetoencephalography binaurally, via in‐ear headphones connected with a plastic tube to the CTF audio delivery system. The sound pressure level assessed with a Sound Level Meter at each in‐ear headphone was 85 dB. All subjects were able to hear the sound above the hearing threshold, balanced between left and right ears, and none of them reported any significant discomfort. The overall duration of auditory ASSR stimulation was 360 s (6 min), consisting of 180 trials of 2 s each. Each trial consisted of 1 s pause followed by 1 s of a 40 Hz amplitude‐modulated tone, so that the interval between the onset of two consecutive trials was 2 s. Forty hertz amplitude‐modulated tones were designed on a carrier frequency of 1,000 Hz, with a 6 ms fade‐in and fade‐out periods, and were normalized so to prevent clipping. The carrier frequency was within the range of those producing a reliable 40 Hz cortical entrainment (Roß et al., 2000; Ross et al., 2003). The 40 Hz sound was generated in MATLAB (The Mathworks) according to the formula:

$$A=\sin \left(2\pi f_{c}t\right)*\left(1+m*\cos \left(2\pi f_{m}t\right)\right)$$

where A is the amplitude, f _c is the carrier frequency set to 1,000, m is the modulation depth, set to 1, f _m is the frequency of modulation, set to 40 and t is the vector of time points for 1 s of stimulus, at a sampling rate of 44,100 Hz. The MATLAB code is freely available online (https://sites.google.com/site/meglabsc/utilities). The entire task was programmed and delivered with the PsychoPy toolbox (http://www.psychopy.org/; Peirce, 2007; Peirce, 2008). To perform the analysis time‐locked to the sound, we recorded digital markers sent by Psychopy to the CTF acquisition system. Additionally, an analogical channel of the MEG system recorded the onset of the actual sound sent to earplugs. The signal from this analogical channel was used to adjust the digital markers and to better account for the random time jitters that normally occur when delivering stimuli via software.

2.3. tDCS

tDCS stimulation (both real and sham) was applied bilaterally. The cathode (−), or excitability reducing electrode, was positioned on the scalp upon the right hemisphere and centered over C4. The anode (+), or excitability enhancing electrode, was upon the left hemisphere and centered over C3 (C4 and C3 are standard positions according to the 10/20 international EEG system). tDCS was delivered with two saline‐soaked sponge electrodes having a surface of 35 cm² each. The electrodes were connected to a battery‐powered stimulator. Real stimulation lasted 20 min with 20 s of fade‐in and fade‐out. The intensity was set to 2 mA and the current density was 0.057 mA/cm². Sham stimulation was performed with the same settings, but electrical current was delivered only for 20 s at the beginning and at the end of the tDCS session. This strategy provided the slight tingling sensation that many subjects report at the beginning and end of stimulation, while producing no effects on brain function.

2.4. MEG data preprocessing

MEG data analysis was performed in MATLAB version 2016b, using in‐home code and the freely available Brainstorm toolbox (Tadel, Baillet, Mosher, Pantazis, & Leahy, 2011). Data preprocessing included: third‐order spatial gradient noise cancellation, downsampling to 600 Hz, signal space‐separation (SSP; Pellegrino et al., 2018). Heartbeat and eye‐movement artifacts were cleaned with the SSP approach (Taulu & Simola, 2006; Tesche et al., 1995). MEG data was cut in 3 s epochs ranging from −1.5 s to +1.5 s around the 40 Hz sound onset. Epochs were visually inspected and those affected by artifacts were manually rejected.

2.5. MRI acquisition and analysis, MEG‐MRI co‐registration and source imaging

The brain MRI was performed for each participant with a 3T Ingenia CX Philips scanner (Philips Medical Systems, Best, the Netherlands). A 3‐dimensional sagittal T1‐weighted‐3D‐TFE scan was acquired with the following parameters: repetition time [TR] = 8.3 ms, echo time [TE] = 4.1 ms, flip angle = 8°, acquired matrix resolution [MR] = 288 × 288, slice thickness [ST] = 0.87 mm. The Freesurfer software (Dale, Fischl, & Sereno, 1999) was employed to segment the brain MRI and to reconstruct the cortical mesh of the “mid” cortical layer, equidistant from white/gray matter interface and pia mater. This surface was tasseled into 15,000 vertices, imported in the Brainstorm toolbox and down‐sampled to 8,000 vertices. The reconstruction of the skull surface was performed from the original MRI and the co‐registration between functional MEG data and anatomical MRI data was calculated with the Brainstorm toolbox (Tadel et al., 2011). The Continuous Head Localization system allowed to track the position of the head in the MEG scanner and to ensure a shift of the head position below 0.5 cm (Pellegrino et al., 2016). Once MEG functional and anatomical MRI data were co‐registered, the personalized head model was built with the boundary element method approach, considering only one cortical layer, with a conductivity of 0.33 S/m. The freely available OpenMEEG toolbox was used for this purpose (Gramfort, Papadopoulo, Olivi, & Clerc, 2010). The inverse problem was solved with the whitened and depth‐weighted linear L2‐minimum norm estimate, with the dipole orientation constrained to be normal to the cortical surface. A common imaging kernel was computed and then applied to obtain single epoch cortical reconstructions. Noise covariance for source reconstruction was calculated from a short resting state of about 6 s acquired at the beginning of each MEG scan.

2.6. ASSR data analysis

Time‐frequency (TF) decomposition was computed applying a Morlet wavelet transformation in the 39–41 Hz gamma band for each epoch, for its entire duration, for each vertex of the cortex of the individual source space. We extracted two measures:

1. POWER: time‐frequency magnitude at 40 Hz, averaged across epochs, so to keep both evoked and induced responses;

2. ITPC: Inter‐Trial Phase Consistency at 40 Hz. Inter‐Trial Phase Consistency was computed as follows:

$$\mathit{\text{ITPC}}\mathit{=}\left|{\mathit{n}}^{\mathit{-}\mathbf{1}}\sum _{\mathit{r}\mathit{=}\mathbf{1}}^{\mathit{n}}{\mathit{e}}^{{\mathit{ik}}_{\mathit{tr}}}\right|$$

where n is the number of trials, e ^ik is the complex polar representation of phase angle k on trial r, at the timepoint t. It is a measure of phase consistency estimated across epochs. The phase at 40 Hz was estimated from the Morlet wavelet coefficients. ITPC can take values between 0 and 1. The higher is ITPC, the higher is the 40 Hz gamma synchronization (Makeig et al., 2002).

Both measures were estimated for the entire epoch. The investigation of baseline differences was performed focusing on the [−1 s to 0 s] time window. Once ruled out baseline differences, both measures were z‐normalized over the signal between −500 and −200 ms (Pfurtscheller & Da Silva, 1999). Baseline time‐window was selected to avoid proximity to the sound and edge effects. Both z‐transformed measures were averaged in the 300–700 ms time‐window, as brain activity reaches a steady state after about 200 ms of 40 Hz entrainment (Larsen et al., 2017; Roß et al., 2000, 2002; Santarelli & Conti, 1999). This entire procedure allowed to estimate one single cortical map for each measure.

To assess the impact of tDCS over the entire cortex, Power and ITPC maps were projected onto a template cortical surface extracted from the standard MRI (ICBM) (Mazziotta et al., 2001) and spatially smoothed with Full Width at Half Maximum of 10 mm (Worsley et al., 2009). After verifying the lack of significant difference between the Pre‐Sham and Pre‐Real sessions, the map resulting from Post minus Pre sham/real tDCS effects were firstly computed and then the Real versus Sham stimulation were contrasted using a cluster‐based permutation approach (Maris & Oostenveld, 2007). The null‐hypothesis corresponded to no difference between the strength of Power (or ITPC) clusters obtained from Real and Sham tDCS stimulation, after having shuffled them through 5,000 permutations. The strength of each cluster was defined as the sum of the absolute t‐values of all the vertices in the cluster, allowing considering both the size of the cluster and its statistical significance. Note that this procedure accounts for multiple comparisons by design, and, since it is quite conservative, some clusters may not reach significance.

To investigate the effect of tDCS on early and transient brain responses, we ran a whole‐cortex cluster‐based premutation analysis on event‐related responses using all the timepoints in the 0–150 ms post stimulus interval. As for the ITPC and Power, Real and Sham stimulations were compared after calculating Post‐minus Pre‐maps.

Finally, the SSS scores were modeled using a two‐way repeated measure ANOVA, with factor time (four levels) and stimulation (two levels). The alpha was set to 5% for all analyses. Statistical analysis was performed with the Brainstorm and Fieldtrip (Oostenveld, Fries, Maris, & Schoffelen, 2011) toolboxes in MATLAB (Mathworks) environment and with the software IBM SPSS Statistics (Ver. 24).

3. RESULTS

No participant reported any side effect or discomfort due to the experimental procedure. The analysis of sleepiness revealed a significant main factor time (F(3,42) = 6.596, p = 0.001) because of a sleepiness increase between the beginning and the end of the MEG scan. The size of the effect was negligible (about 1 point), with average scores probing that subjects were awake for the entire experiment (<2.5 at all time‐points). Neither main factor Stimulation, nor Time by Stimulation interaction were significant (p > 0.200 consistently), thus ruling out that results could be driven by an increase of sleepiness.

3.1. ASSR cortical mapping

As there was no significant difference between Sham and Real Pre‐tDCS sessions for both Power and ITPC, ASSR mapping was performed pooling together data of both sessions as reported in Figure 2. Cortical gamma synchronization as measured by ITPC is shown in the upper panel. Normalized ITPC (expressed as Z‐score over the baseline) was never negative, suggesting that at the time of 40 Hz auditory sound no cortical region displayed a decrease of gamma synchronization. Forty hertz auditory stimulation mapped to a large region, covering the entire temporal lobe, insula, inferior part of primary sensory‐motor regions, up to the inferior frontal regions and beyond and with maximal amplitude ‐as expected‐ in bilateral auditory cortices. To be noted, we confirmed a right hemispheric dominance.

Figure 2.

40 Hz auditory mapping. Panel (a) time‐frequency plots of the ASSR within the 1–60 Hz range from four regions of interest corresponding to site of stimulations (left and right sensory‐motor cortex) and to the left and right primary auditory regions (A1 and A2) exhibiting the highest ASSR response. Left and right sensory‐motor regions were manually drawn on the individual cortical surface based on anatomical landmarks (Raffin, Pellegrino, Di Lazzaro, Thielscher, & Siebner, 2015). Each region covered 10 cm² of cortical surface and included the rostral section of the precentral gyrus, central sulcus and post‐central gyrus around the hand knob. Left and right auditory regions were defined as the cortical area of 10 cm² with stronger ASSR‐related gamma synchronization (ITPC). In the time‐frequency response a cut‐off can be seen (ie, white area without data in the color plots) due to the length of the wavelet. Panel (b) the upper row shows the ITPC mapping. ITPC is expressed as Z‐score with respect to the baseline. ITPC is a measure of entrainment and shows that 40 Hz auditory entrainment maps much further than the primary auditory cortex, covering the entire temporal lobe, the insula and the inferior portion of the sensori‐motor regions. Note that no cortical regions exhibited negative values, suggesting that at the time of 40 Hz auditory stimulation no cortical region even very far from the auditory cortex decreased its gamma synchrony. The cortical mapping also showed an expected right hemispheric dominance. The inferior row shows the power mapping. Power is expressed as Z‐score with respect to the baseline. The cortical distribution of power mapping was very similar to ITPC, although more restricted and with lower magnitude. Cortical surfaces have been inflated (50% inflation) to better show the bottom of the sulci. A = anterior; P = posterior [Color figure can be viewed at http://wileyonlinelibrary.com]

The inferior panel depicts the normalized (Z‐score) Power. No cortical regions showed negative Z‐scores. The spatial distribution of cortical power increase resembled the one described above for ITPC, although with a less widespread distribution and lower magnitude.

3.2. tDCS effect

Real versus Sham statistical comparison is reported in Figure 3. There is a decrease of gamma synchronization over the right superior temporal lobe, insula, inferior part of primary sensory‐motor regions. The effects that survived statistical thresholding by means of cluster‐based permutation analysis is represented in blue over the cortical surface. No other significant clusters were found. The violin plot extracted for the significant cluster shows that tDCS effect was very consistent across subjects, with a decrease of ITPC in 13 out of 15 participants (87%). The investigation of power and ITPC in the baseline (−1 to 0) time window did not reveal any significant difference across conditions. The cluster‐based premutation analysis of early evoked responses did not reveal significant differences across conditions (Supporting Information Figure S1).

Figure 3.

tDCS effects. Sham versus real statistical differences were computed after estimating the effect of tDCS for each experimental condition as post minus pre. The upper panel shows significant clusters surviving after cluster‐based permutation statistics. No significant differences were found for 40 Hz power (left), whereas only a negative cluster in the right hemisphere was found of ITPC. The zoom on the significant cluster shows the time‐course of the tDCS effect expressed as real minus sham (T‐values). The shaded region corresponds to the standard error across subjects. The violin plot shows that tDCS effect was very consistent, with a decrease of ITPC in 13 out of 15 participants (87%) [Color figure can be viewed at http://wileyonlinelibrary.com]

4. DISCUSSION

This study unveiled that bilateral tDCS over primary sensory‐motor hand regions induces a remarkable decrease of gamma synchrony and entrainment. This effect is stronger where the gamma synchrony is most represented, namely the acoustic areas of the right hemisphere. No significant effects beneath the electrodes of stimulation were found.

4.1. tDCS inhibits auditory gamma entrainment

Previous studies on healthy subjects and patients have reported not‐univocal results about the effects of tDCS on gamma activity and synchronization (Hanley et al., 2016; Hoy et al., 2015; Marshall et al., 2016; Wilson et al., 2017). The study of the entire cortical surface allowed to demonstrate that the strongest and most reliable tDCS effect was an inhibition of gamma synchrony and entrainment in the right hemisphere (Figure 3). Such an effect: (a) had a very high magnitude with maximum Z‐score about equal to 5, (b) was very consistent across subjects (87%), (c) was essentially localized where 40 Hz entrainment is stronger, namely in a cortical region including the superior temporal gyrus, the posterior insula and the inferior part of the primary sensory‐motor regions. This is the first evidence of the remote effect of tDCS on gamma synchrony.

We were expecting a remote effect of sensory‐motor stimulation over the temporal cortex as the latter is closely connected to the hand sensory‐motor areas (Fischer et al., 2017; Zatorre et al., 2007). However, it should be noted that bi‐polar inter‐hemispheric tDCS of pericentral cortex has also a polarizing effect of the midsagittal cortex in the interhemispheric fissure. Anodal left pericentral tDCS produces a cathodal effect in the midline supplementary motor area (SMA). Likewise, right pericentral cathodal tDCS produces an anodal current in the SMA. In other words, midline cortex undergoes an opposite polarization to the cortical surface close to the tDCS electrodes. Thus, we cannot rule out that the effects observed in this study might be due to effective polarization of midline mesial cortex.

4.2. Selective effect on temporal regions and ITPC

We did not find a significant change of gamma synchrony in hand sensory‐motor regions underneath the electrodes of stimulation. tDCS effects seem to be only unveiled where the externally driven gamma synchrony is stronger, namely right temporal regions. As previously reported, this area is closely connected to the hand sensory‐motor cortices (Fischer et al., 2017; Zatorre et al., 2007).

Although the 40 Hz sound induced a consistent 40 Hz power and ITPC increase in all conditions (Figure 2), the whole cortex analysis did not unveil any significant effect for Power, but only for ITPC (Figure 3). ITPC is known to provide a reliable measure of the degree of gamma synchronization externally driven and is much more sensitive than Power (Larsen et al., 2017). While it is possible that the cluster‐permutation statistics applied in this study was too conservative with a chance of false negative, our results suggest that tDCS has a selective effect on gamma synchronization as measured by ITPC. This effect only becomes evident when the mechanism sustaining gamma synchronization is strongly challenged, for instance during externally driven entrainment at 40 Hz. Accordingly, the tDCS effect on ITPC was restricted to the right hemisphere, and neither MEG activity at rest (during the ISIs) nor early auditory evoked responses were modified by tDCS. The selectivity of the tDCS effect, only affecting ITPC, indicates that externally driven gamma synchrony and transient auditory responses are driven by different neural mechanisms (Gandal, Edgar, Klook, & Siegel, 2012; Zhang, Peng, Zhang, & Hu, 2013). For example, patients with schizophrenia usually suffer from impaired auditory‐driven gamma synchrony at 40 Hz, yet they often exhibit an increased baseline gamma synchrony (Gandal et al., 2012; Hong et al., 2008; Winterer et al., 2004).

4.3. Mechanisms and circuits potentially involved in the effect of tDCS on ASSR

We purposely designed a study combining tDCS and 40 Hz auditory entrainment because they act on the same intracortical circuitry and neurotransmitters.

tDCS of the sensory‐motor regions largely depends on the modulation of fast‐spiking neurons of Layers 2 and 3 projecting onto pyramidal neurons (Di Lazzaro et al., 2013; Di Lazzaro & Rothwell, 2014; Lang et al., 2011) and on the modulation of glutamatergic and GABA‐ergic pathways (Di Lazzaro et al., 2013; Di Lazzaro & Rothwell, 2014; Lang et al., 2011; Liebetanz, Nitsche, Tergau, & Paulus, 2002; Nitsche et al., 2003; Nitsche et al., 2005; Stagg, Bachtiar, & Johansen‐Berg, 2011). Anodal stimulation decreases GABA concentration and transmission (Antal, Terney, Kühnl, & Paulus, 2010; Cirillo et al., 2017; Hummel et al., 2005; Kidgell et al., 2013; Nitsche et al., 2005; Stagg et al., 2009; Zhang et al., 2014), whereas cathodal tDCS reduces the concentration of both GABA and Glutamate (Clark, Coffman, Trumbo, & Gasparovic, 2011; Stagg et al., 2009).

Forty Hz auditory synchrony also depends on inhibitory GABA interneurons firing in gamma band (chandelier and basket cells, parvalbumin positive, located in the 2 and 3 cortical layers) and activated by glutamate (for more information please see Brenner et al., 2009; Javitt & Sweet, 2015; Traub, Whittington, Stanford, & Jefferys, 1996). NMDA blockage depresses the generation of gamma activity (O'Donnell et al., 2013; Plourde, Baribeau, & Bonhomme, 1997; for review see (Bartos, Vida, & Jonas, 2007)), whereas GABA activation improves gamma synchronization (Traub et al., 2003; Whittington, Traub, & Jefferys, 1995).

tDCS may therefore inhibit gamma synchrony because of the combined effect of GABAergic depression and glutamatergic blockade (Antal et al., 2010; Clark et al., 2011; Hummel et al., 2005; Kidgell et al., 2013; Nitsche et al., 2005; Stagg et al., 2009; Zhang et al., 2014). Such an interpretation of our results is based on the critical analysis of previous literature. Newer data modeling techniques, such as dynamical causal modeling (DCM), may provide more direct insight into the mechanisms of tDCS‐related gamma inhibition (Symmonds et al., 2018).

4.4. Translational value

Over the last 15 years, it has become clearer that many neuropsychiatric conditions are characterized by an impairment of gamma synchronization. Gamma oscillations have been proposed as a marker of neuropsychiatric conditions such as major depression (Fitzgerald & Watson, 2018; Larsen et al., 2017), but the topographical distribution of gamma impairment and its direction (synchronization increase vs. decrease) remain to be clarified (Herrmann & Demiralp, 2005; Uhlhaas & Singer, 2006). For instance, gamma synchrony is increased in schizoaffective bipolar disorder (Brealy et al., 2015), Alzheimer's disease (Başar, Femir, Emek‐Savaş, Güntekin, & Yener, 2017), in first episode psychosis (Flynn et al., 2008) and after stroke (Pellegrino et al., 2012). Conversely, patients with schizophrenia exhibit reduced gamma synchronization (Krishnan et al., 2009; Thuné, Recasens, & Uhlhaas, 2016) and so do patients affected by focal epilepsy (Pellegrino et al., 2018; for review Herrmann & Demiralp, 2005; Krishnan et al., 2009; Thuné et al., 2016; Uhlhaas & Singer, 2006). tDCS has not been implemented in clinical practice yet, but tDCS has been applied in more or less all neuropsychiatric conditions (for review and guidelines on this topic please see Lefaucheur et al., 2017). This study demonstrates that bilateral (left anode, right cathode) sensory‐motor tDCS inhibits gamma synchrony in remote regions. The inhibitory effect of tDCS on gamma oscillations may be exploited whenever gamma synchronization is increased. On the other hand, the inhibitory effect may be undesirable in conditions that are associated with reduced gamma synchrony.

4.5. Methodological considerations

The first step of our study was the accurate mapping of gamma synchrony driven by a 40 Hz sound. To this purpose, we took advantage of MEG. This technique allows measuring cortical activity noninvasively, with high temporal resolution, with better spatial resolution than EEG (Chowdhury et al., 2018; Hedrich, Pellegrino, Kobayashi, Lina, & Grova, 2017; Larsen et al., 2017; Pellegrino et al., 2016; Pellegrino, Tomasevic, Herz, Larsen, & Siebner, 2018; von Ellenrieder et al., 2016). This technique is also especially suitable in the context of off‐line tDCS as it does not require the application of electrodes on the scalp and has no impact on tDCS montage (Pellegrino et al., 2018). Forty hertz sounds synchronized gamma activity over a very large area spanning from frontal to central, parietal and temporal regions reaching cortical surface far away from the primary auditory cortex, up to the sensory‐motor regions (Figure 2).

In this study we chose to place the tDCS electrodes over the sensory‐motor regions rather than over the temporal regions. Brain stimulation techniques may impact the function of remote cortical regions which were not the primary target. For instance, an excitability enhancing technique designed for motor cortex, can lead to inhibition when applied to the sensory cortex (Poreisz et al., 2008). This is because tDCS effect depends on the interaction between the induced tissue current, the geometry of the neural structures and the cortical folding pattern. While such effects are well known and predictable for the sensory‐motor cortex, they are largely unknown for stimulation of other brain areas (Landi et al., 2015; Baltus, Wagner, Wolters, & Herrmann, 2018; Jacobson, Koslowsky, & Lavidor, 2012; Joos, De Ridder, Van de Heyning, & Vanneste, 2014; Lorenz, Müller, Schlee, Langguth, & Weisz, 2010; Prete, D'Anselmo, Tommasi, & Brancucci, 2017).

5. CONCLUSIONS

This study demonstrates that bilateral tDCS (left anode, right cathode) affects gamma activity. Our findings shed new light on the effects of tDCS on brain oscillations and synchronization and on the circuitry where it acts. Beyond its effect on cortical excitability, tDCS might be applied as useful tool to tune gamma synchrony.

CONFLICT OF INTERESTS

Hartwig R. Siebner has received honoraria as speaker from Novartis Denmark and Sanofi‐Genzyme, Denmark, has received honoraria as consultant from Sanofi‐Genzyme, Denmark, has received honoraria as editor from Elsevier Publishers, Amsterdam, the Netherlands and Springer Publishing, Stuttgart, Germany. The other authors have no conflict of interests to disclose.

Supporting information

Supplementary Figure 1 Event related responses.

MEG data were low‐pass filtered at 40 Hz. The upper panel shows the source imaging at 40 msec, 100 msec and 400 ms. Source data were z‐transformed. Threshold was set at 1.5 z, with brighter colors indicating higher z‐values. As no differences were found across conditions, Real Pre, Real Post, Sham Pre and Sham Post were averaged to obtain the figure.

Pellegrino G, Arcara G, Di Pino G, et al. Transcranial direct current stimulation over the sensory‐motor regions inhibits gamma synchrony. Hum Brain Mapp. 2019;40:2736–2746. 10.1002/hbm.24556