Probing phase- and frequency-dependent characteristics of cortical interneurons using combined transcranial alternating current stimulation and transcranial magnetic stimulation

Abstract

Paired-pulse transcranial magnetic stimulation (TMS) and peripheral stimulation combined with TMS can be used to study cortical interneuronal circuitry. By combining these procedures with concurrent transcranial alternating current stimulation (tACS), Guerra and colleagues recently showed that different cortical interneuronal populations are differentially modulated by the phase and frequency of tACS-imposed oscillations (Guerra A, Pogosyan A, Nowak M, Tan H, Ferreri F, Di Lazzaro V, Brown P. Cerebral Cortex 26: 3977–2990, 2016). This work suggests that different cortical interneuronal populations can be characterized by their phase and frequency dependency. Here we discuss how combining TMS and tACS can reveal the frequency at which cortical interneuronal populations oscillate, the neuronal origins of behaviorally relevant cortical oscillations, and how entraining cortical oscillations could potentially treat brain disorders.

brain oscillations are produced by rhythmic electrical activity of neurons, occur in distinct frequency bands, and are thought to mediate information flow between distant cortical regions (Buzsáki 2006). The emergence of a new technique, transcranial alternating current stimulation (tACS), has spurred renewed scientific interest in brain oscillations as an important intermediate to single neuron physiology and behavior. tACS involves applying alternating current to the scalp through surface electrodes and can entrain underlying neuronal networks so that they begin to oscillate at the stimulation frequency, close to the stimulation frequency, or at a subharmonic of the stimulation frequency (Hermann et al. 2013; Reato et al. 2013). Recent work has combined tACS with neuroimaging and other brain stimulation modalities to better understand the neural mechanisms mediating motor functions and plasticity processes (Goldsworthy et al. 2016; Moisa et al. 2016). These multimodal approaches hold substantial promise for determining the characteristics of neuronal populations that generate cortical oscillations and mediate different behavioral states.

In a recent study, Guerra and colleagues (2016) tested multiple motor cortical electrophysiological measures using transcranial magnetic stimulation (TMS) during concurrent tACS. They aimed to determine which cortical interneuronal circuits were affected by the tACS-imposed oscillatory activity and whether the relationship between the imposed oscillatory activity and cortical interneuronal activity depended on the phase and frequency of the imposed activity. Specifically, the authors measured corticospinal excitability (CSE), short-latency intracortical inhibition (SICI), long-latency intracortical inhibition (LICI), intracortical facilitation (ICF), and short-latency afferent inhibition (SAI) during tACS at two different frequencies: one that mimics endogenous motor cortical oscillatory activity (20 Hz) and one that does not (7 Hz). Measurements were made at four phases of the tACS-imposed oscillation: rising, peak, falling, and trough phases. Importantly, each TMS measurement tested is thought to reflect different populations of cortical interneurons synapsing onto the pyramidal cells that comprise the corticospinal tract, with SICI, LICI, and SAI reflecting inhibitory interneuronal circuitry and ICF reflecting excitatory interneuronal circuitry. The results indicate that SAI, which is related to cholinergic inhibition, was modified in a frequency-dependent manner. Specifically, SAI was abolished during 20 Hz but not 7 Hz tACS. With respect to phase-dependent effects, CSE, ICF, and SICI were all greatest when TMS was applied at the trough of the tACS-imposed oscillation, meaning that both excitatory and inhibitory interneuronal activity were highest at this time. Additionally, for SICI, this effect was frequency dependent, meaning that phase-dependent modulation of SICI was stronger during 20 Hz tACS than during 7 Hz tACS. To further investigate the phase and frequency dependency of inhibitory interneuronal populations that contribute to SICI, the authors divided participants into two groups depending on whether each participant’s peak beta frequency at rest was closer to or further from 20 Hz. This post hoc analysis showed that participants with individual peak beta frequencies closer to 20 Hz exhibited stronger phase-dependent modulation of SICI during 20 Hz tACS than those with peak beta frequencies further from 20 Hz. As a whole, the results of this work point to previously unrecognized phase- and frequency-dependent properties of specific cortical interneuronal populations, raising the exciting possibility that human cortical interneuronal circuits could be classified based on their phase and frequency dependency.

Guerra and colleagues (2016) specifically demonstrated that cortical interneuronal populations respond to tACS at a particular frequency and phase in four different ways and can therefore be categorized as 1) neurons that exhibit both phase- and frequency-dependent modulation, 2) neurons that exhibit phase- but not frequency-dependent modulation, 3) neurons that exhibit frequency- but not phase-dependent modulation, and 4) neurons that exhibit neither phase- nor frequency-dependent modulation. Importantly, the interneuronal populations that comprise each category depend on the frequency of tACS used when assessing phase and frequency dependency. In humans, tACS can only be applied at low current densities, so effective stimulation is thought to be related to interactions between endogenous oscillatory activity and the tACS-imposed oscillatory activity. Thus entrainment of brain oscillatory activity only occurs when the stimulation frequency closely matches the frequency of endogenous oscillations, or when the frequency of endogenous oscillations is a subharmonic of the stimulation frequency (Hermann et al. 2013; Reato et al. 2013). For example, if a given population of cortical neurons is known to oscillate near 20 Hz, that population can only be entrained if the imposed oscillation is close to 20 Hz or is a multiple of 20 Hz. This principle also extends to entrainment with rhythmic TMS; Romei et al. (2016) recently demonstrated that entrainment of human motor cortical activity is most evident when the frequency of rhythmic TMS closely matches the individual peak beta frequency. Based on this principle, if a given frequency of exogenously imposed oscillatory activity entrains a specific neuronal population, it can be inferred that the entrained neurons normally oscillate at a frequency that is close to the exogenously imposed frequency or a subharmonic of the imposed frequency, even when no exogenous frequency is being applied. Because Guerra et al. (2016) showed that SICI and SAI were significantly modified during 20 but not 7 Hz tACS, it is likely that inhibitory interneuronal populations contributing to SICI or SAI oscillate at ~20 Hz or its subharmonic, 10 Hz, and may therefore generate 10 and/or 20 Hz oscillatory activity. In contrast, because CSE, ICF, and LICI did not exhibit frequency-dependent characteristics at either frequency tested, the cortical interneuronal populations tested using these measures likely oscillate at frequencies other than 20 or 7 Hz (or their subharmonics, 10 and 3.5 Hz). Using this logic, future studies could systematically apply tACS at different frequencies during various measurements of cortical physiology to identify the approximate frequencies at which different cortical interneuronal populations oscillate, and how these frequencies might vary across individuals. This approach would also help determine whether specific cortical interneuronal populations might generate sensorimotor oscillations at various frequencies.

Studies in nonhuman primates have shown that the oscillatory characteristics of cortical interneuronal populations change with behavior (Murthy and Fetz 1996). Therefore, one question regarding the findings of Guerra et al. (2016) relates to the generalizability of the results to different behavioral and/or cognitive states. Specifically, are the phase- and frequency-dependent characteristics of the different neuronal populations probed by Guerra and colleagues the same immediately before, during, and immediately after movement? To address this, it would be beneficial to explore the phase and frequency dependency of different cortical interneuronal populations during multiple behavioral states. This could be tested by comparing phase- and frequency-dependent characteristics during rest, premovement, movement, and postmovement periods. Because oscillatory characteristics of cortical interneuronal populations differ depending on the motor state (Murthy and Fetz 1996), individual peak beta frequencies may also vary depending on whether the individual is resting quietly, preparing a movement, moving, or has just completed a movement. These different peak frequencies could be interpreted as separate physiological phenomena produced by distinct cortical oscillatory generators, suggesting that different neuronal populations might generate the different individual peak beta frequencies that are observed during different motor states. This could be tested by applying 20 and 7 Hz tACS to primary motor cortex and comparing how different TMS measurements are modulated by tACS phase and frequency during continuous rest as well as different stages of movement. If all TMS measurements are similarly affected by the tACS-imposed oscillatory activity during the different motor states, this would suggest that all beta oscillation variants (occurring during continuous rest, premovement, movement, and postmovement periods) share a common generator and are mechanistically related. In contrast, if the TMS measurements are affected differently by the phase and frequency of tACS-imposed oscillatory activity during each motor state, this might suggest that the beta oscillation variants have different mechanistic origins. Another possibility is that the same neuronal population oscillates at multiple frequencies depending on the behavioral state and therefore generates all variants of beta oscillatory activity (Hutcheon and Yarom 2000). This would be supported if a given TMS measurement was modified in a phase- and/or frequency-dependent manner during the application of multiple tACS frequencies. Testing how interactions between cortical excitability measures and tACS are modified by different behavioral states may therefore deepen our understanding of how and where movement-related cortical oscillatory activity originates in human motor cortex.

The concepts presented above are not inherently limited to primary motor cortex. However, to use combined TMS and tACS to classify cortical interneuronal populations in other areas, it must be possible to reliably quantify the excitability of nonmotor cortical regions. Recent work has begun to address this using concurrent TMS-EEG. In these studies, TMS pulses are applied and the cortical reactivity to these pulses is measured via TMS-evoked EEG potentials (Ilmoniemi et al. 1997). However, this approach requires substantial development before specific cortical interneuronal populations and their functions can be definitively identified. Once this challenge is overcome, similar experiments to the one performed by Guerra and colleagues (2016) could be performed in nonmotor brain areas. Then, instead of testing how different motor states affect the interactions between tACS-imposed oscillatory activity and cortical excitability measures, how different mental and cognitive states modify the interaction between tACS and TMS-evoked EEG potentials could be addressed. Combining tACS and TMS might also be useful for identifying neuronal populations that generate abnormal oscillatory patterns in certain neurological and psychiatric disorders. For example, patients with focal hand dystonia exhibit decreased movement-related beta desynchronization (Kristeva et al. 2005) and impaired cortical inhibition (Hallett 2011). To determine whether the neurons mediating aberrant cortical inhibition are the same neurons that undergo beta desynchronization during movement, modulation of cortical inhibition during tACS at the individual peak beta frequency could be examined during rest and movement. Furthermore, studies should also test whether entraining brain regions that exhibit abnormal oscillatory patterns to relevant frequencies will have any significant behavioral impact.

Overall, the work by Guerra and colleagues (2016) demonstrates that concurrent tACS and TMS presents new opportunities for understanding the neuronal generators, physiological mechanisms, and behavioral relevance of different brain oscillations. This multimodal approach will likely produce novel insights into the phase- and frequency-dependent characteristics of cortical interneuronal populations in humans, potentially leading to the development of neuromodulation approaches that can selectively target specific neurons based on their phase and frequency dependency.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

S.J.H. and N.T. conceived and designed research; S.J.H. and N.T. interpreted results of experiments; S.J.H. drafted manuscript; S.J.H. and N.T. edited and revised manuscript; S.J.H. and N.T. approved final version of manuscript.