Transcranial alternating current stimulation modulates spontaneous low frequency fluctuations as measured with fMRI

Highlights

• The effects of tACS on spontaneous low frequency fluctuations were evaluated using simultaneous fMRI-tACS.

• Electric field simulations were used to predict and evaluate tACS effects relative to directly stimulated regions.

• In brain regions exhibiting frequency-dependent effects of tACS, 10 and 40 Hz tACS generally induced opposite effects.

• The left fronto-parietal control resting-state network appears to be especially vulnerable to tACS.

• Most tACS effects were observed as modulation of functional connectivity between networks.

Abstract

Transcranial alternating current stimulation (tACS) is a promising tool for modulating brain oscillations. Combining tACS with functional magnetic resonance imaging (fMRI), we recently showed that tACS applied over the occipital cortex did not exert its strongest effect on regions below the electrodes, but mainly on more distant fronto-parietal regions. Theoretically, this effect could be explained by tACS-induced modulation of functional connectivity between directly stimulated areas and more distant but anatomically and functionally connected regions. In the present study, we aimed to characterize the effect of tACS on low frequency fMRI signal fluctuations. We employed simultaneous fMRI-tACS in 20 subjects during resting state (eyes open with central fixation for ~ 8 min). Subjects received tACS at different frequencies (10, 16, 40 Hz) and with different electrode montages (Cz-Oz, P5–P6) previously used in behavioral studies. Electric field simulations showed that tACS over Cz-Oz directly stimulates occipital cortex, while tACS over P5–P6 primarily targets parietal cortices. Group-level simulation-based functional connectivity maps for Cz-Oz and P5–P6 resembled the visual and fronto-parietal control resting-state networks, respectively. The effects of tACS were frequency and partly electrode montage dependent. In regions where frequency-dependent effects of tACS were observed, 10 and 40 Hz tACS generally induced opposite effects. Most tACS effects on functional connectivity were observed between, as opposed to within, resting-state networks. The left fronto-parietal control network showed the most extensive frequency-dependent modulation in functional connectivity, mainly with occipito-parietal regions, where 10 Hz tACS increased and 40 Hz tACS decreased correlation values. Taken together, our results show that tACS modulates local spontaneous low frequency fluctuations and their correlations with more distant regions, which should be taken into account when interpreting tACS effects on brain function.

Introduction

Transcranial alternating current stimulation (tACS) is a non-invasive brain stimulation technique that has proven to be effective in modulating brain oscillations (Ali et al., 2013, Helfrich et al., 2014a, Ozen et al., 2010, Zaehle et al., 2010, Reato et al., 2013). A growing body of literature exists documenting frequency-dependent effects of tACS on different brain functions in health (Cabral-Calderin et al., 2015, Feurra et al., 2013, Feurra et al., 2011a, Helfrich et al., 2014b, Joundi et al., 2012, Kanai et al., 2008, Kar and Krekelberg, 2014, Laczo et al., 2012, Polania et al., 2012, Santarnecchi et al., 2013, Wach et al., 2013, Wang et al., 2015) and disease (Brittain et al., 2013, Fedorov et al., 2010). For instance, effects of tACS on visual perception have been reported when tACS was applied over occipital and parietal cortices at alpha or gamma frequencies (Helfrich et al., 2014a, Cabral-Calderin et al., 2015, Laczo et al., 2012, Wang et al., 2015). Although these effects have been mainly attributed to increasing the power or synchrony of alpha/gamma oscillations in occipito-parietal areas, recent studies indicate that underlying mechanisms of tACS effects are far more complicated. With the exception of a few studies combining tACS with electroencephalographic (EEG) recordings (Helfrich et al., 2014a, Zaehle et al., 2010) or functional magnetic resonance imaging (fMRI) (Alekseichuk et al., 2015, Cabral-Calderin et al., 2016, Vosskuhl et al., 2015), most of our knowledge about tACS is derived from behavioral studies. Technical challenges inherent to concurrent measurements of brain activity when applying tACS, such as artifacts in EEG signals or the need for specialized MRI-safe equipment, leave only several published studies of combined brain imaging and tACS, making it difficult to interpret the behavioral results in terms of the underlying brain activity (Zaehle et al., 2010, Helfrich et al., 2014b, Alekseichuk et al., 2015, Cabral-Calderin et al., 2016, Vosskuhl et al., 2015, Witkowski et al., 2015, Neuling et al., 2015).

In a recent study combining tACS with fMRI (Cabral-Calderin et al., 2016) we showed that the effect of tACS on brain activity, as inferred from the blood-oxygenation-level-dependent (BOLD) signal, is rather complex and cannot entirely be predicted based on electrode positions alone. In general, the effects of tACS over Cz-Oz were frequency-, region-, and task-dependent. Specifically, BOLD activity changes were stronger with lower tACS frequencies; that is, alpha (10 Hz) and beta (16 Hz) frequencies were more effective than gamma (60, 80 Hz). Additionally, effects were mainly observed in regions not activated by the task and that were distant from the electrodes (i.e., fronto-parietal but not occipital regions) (Cabral-Calderin et al., 2016). The fact that the above-mentioned study reported tACS effects mainly in areas distant from the electrodes (i.e., fronto-parietal) could indicate that the effects of tACS are transmitted from directly stimulated brain regions (i.e., occipital) to more distant but anatomically and functionally connected regions. In fact, it has been proposed that the effects of phase entrainment (a possible mechanism of tACS) in a given brain area may propagate through a larger network due to inter-regional phase locking (Canolty and Knight, 2010). It is thus reasonable to assume that tACS also interferes with the dynamics of brain networks. A recent study combining tACS (applied over occipito-parietal areas) with EEG has reported tACS-induced modulation of interhemispheric functional connectivity in occipito-parietal cortices in a phase-specific manner (Helfrich et al., 2014b). The authors reported that in-phase tACS in the gamma range (40 Hz) enhanced interhemispheric synchrony while anti-phase tACS impaired it, providing evidence of the possibility of modifying functional coupling between brain regions with tACS.

The dynamics of intrinsically functionally connected networks have been largely studied with resting-state fMRI (rs-fMRI), in which subjects are not engaged in an active task (e.g., awake with eyes closed or eyes open, with or without eye fixation (Van Dijk et al., 2010)). Over the last two decades many studies have documented the presence of resting-state networks (RSNs) arising from regionally distributed but temporally correlated low frequency BOLD signal fluctuations (< 0.1 Hz) in the resting brain (Fox et al., 2005, Fox and Raichle, 2007, Biswal et al., 1995, Beckmann et al., 2005, Raichle, 2015). These RSNs are constrained by, but not limited to, structural connectivity (Park and Friston, 2013) and resemble several networks of functionally connected regions modulated by active tasks such as the dorsal attention network (DAN), lateralized right and left fronto-parietal control networks (r-FP and l-FP, respectively), executive network, motor network, visual network, and default mode network (DMN). It has been suggested that spontaneous BOLD signal fluctuations have neurophysiological origins (Raichle, 2010, Shmuel and Leopold, 2008, Tagliazucchi et al., 2012). Previous studies attempted to correlate the power of specific EEG rhythms with resting-state measurements and showed that RSNs are related not only to one brain rhythm, but to combinations of several. In terms of the BOLD signal amplitude in resting-state networks, global (i.e., average over all EEG channels), as well as particularly occipital, alpha (8–13 Hz) power has been positively correlated with DMN and negatively correlated with DAN, r-FP, l-FP, and visual networks (Mantini et al., 2007, Mo et al., 2013, Zhan et al., 2014). Global beta (13–30 Hz) power has been positively correlated with DMN and negatively correlated with DAN, visual and motor networks (Mantini et al., 2007). Global gamma (> 30 Hz) power has been positively correlated with global resting-state activity and, in particular, with regions from the executive network (e.g., medial-ventral prefrontal cortex and anterior cingulate (Mantini et al., 2007)). In terms of functional connectivity (expressed as temporal correlations), special attention has been paid to occipital alpha power, which negatively correlates with functional connectivity within visual cortices and between visual and more frontal and subcortical regions (Scheeringa et al., 2012). An inverse relationship has also been reported between central alpha and beta power spectra and changes in functional connectivity, mostly between subcortical regions and association cortices. In addition, a positive relationship has been found between gamma power and functional connectivity changes between subcortical, association and primary cortices (central and occipital gamma power) and within these regions (frontal gamma power) (Tagliazucchi et al., 2012). These links between specific EEG frequencies and RSNs motivate the possibility of probing RSNs by modulating oscillatory activity with tACS.

The aim of the present study was to evaluate whether tACS applied over occipital and parietal cortices induces changes in spontaneous activity measured with rs-fMRI. To this end, we applied tACS at three different frequencies in the alpha, beta, and gamma range (10, 16, 40 Hz, respectively), and with two different electrode montages (Cz-Oz and P5–P6). The tACS frequencies and electrode montages were chosen based on previous studies showing modulatory effects of tACS at alpha or gamma frequencies on visual perception (Helfrich et al., 2014a, Laczo et al., 2012, Wang et al., 2015, Cabral-Calderin et al., 2016).

Given that our previous fMRI study indicated strongest tACS effects in regions that were distant from the electrodes (i.e., fronto-parietal but not occipital regions) (Cabral-Calderin et al., 2016), (I) we hypothesized that tACS would change functional connectivity between directly stimulated regions and more distant areas. Based on previous EEG/fMRI and tACS/fMRI studies, (II) we expected to observe frequency-dependent modulation of RSNs with tACS. Given the previously reported negative correlations between the power of alpha/beta rhythms and functional connectivity, as well as BOLD signal amplitude in most RSNs (Tagliazucchi et al., 2012, Mantini et al., 2007, Scheeringa et al., 2012), (III) we hypothesized that tACS at alpha and beta frequencies should reduce the amplitude of spontaneous low frequency fluctuations and functional connectivity in most RSNs (e.g., DAN and visual network). An exception should be observed in the DMN, where the power of alpha and beta rhythms have been found to positively correlate with that network's BOLD signal amplitude, therefore, (IV) alpha and beta tACS should increase the amplitude of spontaneous low frequency fluctuations in the DMN (Mantini et al., 2007, Mo et al., 2013). Since gamma power has been positively associated with the amplitude of BOLD signal fluctuations and their correlations (Tagliazucchi et al., 2012, Mantini et al., 2007), (V) tACS in the gamma range should increase the amplitude of spontaneous low frequency fluctuations and functional connectivity. According to our previous study (Cabral-Calderin et al., 2016) we hypothesized that (VI) the strongest modulation of spontaneous low frequency fluctuations would be observed with alpha/beta tACS mainly in fronto-parietal networks (i.e., r-FP and l-FP). In addition, since different electrode positions induce different electric field distributions (Neuling et al., 2012, Windhoff et al., 2013), (VII) we expected to observe a difference in the low frequency fluctuation changes associated with each electrode montage. Based on a previous study that simulated the electric field distribution induced by transcranial electric stimulation (Neuling et al., 2012), (VIII) we expect tACS over Cz-Oz to stimulate mainly occipital cortex and, as a result, to modulate the amplitude of low frequency fluctuations and functional connectivity relative to the visual network. Following the same reasoning, (IX) tACS over P5–P6 should stimulate mainly parietal regions, translating into modulation of the amplitude of low frequency fluctuations in parietal cortices and functional connectivity of RSNs involving parietal regions such as r-FP, l-FP and DAN.