Neuroscience 278 (2014) 93–104

NEUROSCIENCE FOREFRONT REVIEW THE ROLE OF INHIBITION IN HUMAN MOTOR CORTICAL PLASTICITY V. BACHTIAR AND C. J. STAGG *

Can GABA decreases be related directly to learning of a motor task? Do NIBS protocols that modulate GABA levels also modulate learning rates? GABAergic synapses as a site for homeostatic plasticity A role for GABA in long-term plasticity? Future directions Is GABA modulation necessary for short-term plasticity induction? A role for GABA modulation in the chronic stage of stroke recovery? Functionally-relevant modulation of GABA Exploring the physiological underpinnings of the MRS-assessed GABA signal – the necessity for multimodal approaches M1 as part of a wider functional network Conclusions Acknowledgments References

Oxford Centre for Functional MRI of the Brain (FMRIB), Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK

Abstract—Over recent years evidence from animal studies strongly suggests that a decrease in local inhibitory signaling is necessary for synaptic plasticity to occur. However, the role of GABAergic modulation in human motor plasticity is less well understood. Here, we summarize the techniques available to quantify GABA in humans, before reviewing the existing evidence for the role of inhibitory signaling in human motor plasticity. We discuss a number of important outstanding questions that remain before the role of GABAergic modulation in long-term plasticity in humans, such as that underpinning recovery after stroke, can be established. Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved.

98 98 99 100 101 101 101 101

101 102 102 102 102

Key words: GABA, motor, plasticity, stroke recovery, human. Contents Introduction Quantifying GABA in humans in vivo A brief overview of GABA metabolism and signaling in the adult human brain Measurement of GABA in the resting brain MRS TMS Measures of GABAergic activity during movement Changes in GABAergic signaling with aging GABA modulation is demonstrated during plasticity in healthy adults GABA decreases occur in response to a wide range of plasticity-induction protocols NIBS

INTRODUCTION 93 94

Plasticity can be loosely defined as the ability of the brain to adapt to respond to new challenges. In humans, plasticity occurs largely through modification of the strength of synaptic connections. This flexibility in behavior can be demonstrated in the learning of a new motor task or recovery of motor function after a stroke, but can also be induced non-invasively via transcranial stimulation techniques such as transcranial direct current stimulation (tDCS). Synaptic plasticity was described as follows in Hebb’s seminal 1949 work: ‘‘When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells, such that A’s efficiency, as one of the cells firing B, is increased’’ (Hebb, 1949). Synaptic plasticity encompasses a number of pre-synaptic and post-synaptic changes, the most ubiquitous of which is Long-Term Potentiation (LTP)-like plasticity. LTP-like plasticity is demonstrated as the rapid and sustained increase of strength of glutamatergic synapses. The details of glutamatergic LTP and LTPlike plasticity are increasingly well understood, but their complexity falls outside the scope of this review (see for example, Feldman, 2009 for a review). There is increasing evidence emerging from both animal models

94 95 95 96 97 97 97 97 98

*Corresponding author. Address: FMRIB, John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK. Tel: +44-(0)1865-222736; fax: +44-(0)1865-222717. E-mail address: [email protected] (C. J. Stagg). Abbreviations: BCM, Bienenstock–Cooper–Munro; CS, conditioned stimulus; cTBS, continuous theta-burst stimulation; EMG, electromyography; fMRI, functional MRI; GABA, c-aminobutyric acid; GAD, glutamic acid decarboxylase; ISI, inter-stimulus interval; iTBS, intermittent theta-burst stimulation; LICI, Long-Interval Intracortical Inhibition; LTD, Long-Term Depression; LTP, Long-Term Potentiation; M1, primary motor cortex; MEG, magnetoencephalography; MRS, Magnetic Resonance Spectroscopy; NIBS, non-invasive brain stimulation; ppTMS, paired-pulse TMS; SICI, short-interval intracortical inhibition; TBS, theta-burst stimulation; tDCS, transcranial direct current stimulation; TMS, Transcranial Magnetic Stimulation. http://dx.doi.org/10.1016/j.neuroscience.2014.07.059 0306-4522/Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved. 93

94

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

and human studies that an initial decrease in local c-aminobutyric acid (GABA), the major inhibitory neurotransmitter in the adult mammalian brain, is of vital importance for allowing LTP-like plasticity to occur (Castro-Alamancos et al., 1995; Trepel and Racine, 2000; Clarkson et al., 2010). Here, we review the evidence for GABA modulation in human motor plasticity, both during the learning of new motor skills in healthy controls and in the recovery of motor function after stroke, two processes that have many similarities (Krakauer, 2006). We focus on changes occurring within the primary motor cortex (M1) during the learning of motor skills as this is the region that has been studied in most depth. We first review the methods by which inhibitory signaling can be quantified in humans, and the relative advantages and disadvantages of each of these techniques. We then go on to discuss the emerging literature highlighting the role of GABAergic modulation in plasticity in healthy subjects. Finally, we review the current evidence for the role of GABA in recovery of function after a brain injury such as a stroke.

QUANTIFYING GABA IN HUMANS IN VIVO The role of GABA in the induction of LTP-like processes that underpin neuroplastic mechanisms in animal models is well established (Castro-Alamancos et al., 1995; Sanes and Donoghue, 2000; Trepel and Racine, 2000; Clarkson et al., 2010). However, in the adult human brain, GABA is present at millimolar levels, which has historically made it difficult to quantify. Nevertheless, recent technical and methodological advances in Magnetic Resonance Spectroscopy (MRS) and non-invasive brain stimulation (NIBS) approaches have made reliable quantification of GABA more achievable. In turn, the ability to accurately measure inhibitory signaling has sparked increased interest in understanding the role of GABA both in motor learning in healthy controls and in recovery of motor function after stroke. In addition to MRS, other imaging methods have been shown to be able to evaluate inhibitory activity. Positron Emission Tomography (PET) is increasingly being used to quantify GABAergic activity, with specific ligands being developed for specific GABA receptors (see Andersson and Halldin, 2013 for a full review and comprehensive summary of available compounds). Other techniques are also available to assess inhibitory activity, albeit indirectly. For example, magnetoencephalography (MEG) allows for the non-invasive measurement of neuronal network oscillations in the beta-frequency range (15–30 Hz), and gamma-frequency range (30–100 Hz), both of which are thought to be under the direct control of GABAergic modulation (Bartos et al., 2007). Others have shown the potential effect of excitation-inhibition changes on hemodynamic responses as measured by BOLD functional MRI (fMRI) (Logothetis, 2008). However, a full discussion of these methods is beyond the scope of this review.

A brief overview of GABA metabolism and signaling in the adult human brain GABA is the primary inhibitory neurotransmitter in the central nervous system and is primarily metabolized from glutamate via the enzyme glutamic acid decarboxylase (GAD), found within GABAergic neurons. In humans, GAD is found in two distinct forms with molecular weights of 65 and 67 kDa (given the nomenclature GAD65 and GAD67 respectively). GABA is found in two distinct functional pools within inhibitory neurons, each of which appear to be regulated by a distinct form of GAD. The majority of GABA is found in the cytosol and is regulated by the tonically active GAD67 (Martin and Rimvall, 1993) and does not appear to contribute significantly to GABAergic neurotransmission (Wei et al., 2004; Wei and Wu, 2008). GABA is also found at high concentrations in a vesicular pool in the presynaptic boutons, the production of which is regulated by the phasically active GAD65 (Martin and Barke, 1998). Vesicular GABA has a key role in inhibitory synaptic neurotransmission (Martin and Rimvall, 1993) and GAD67 has been shown to have an activity-dependent role in GABA synthesis such that decreased network activity leads to decreased expression of GAD67, which in turn leads to lower vesicular filling of the transmitter and thus GABA levels (Lau and Murthy, 2012). After release from the GABAergic neuron, the vast majority of GABA is metabolized to succinic acid semialdehyde in the GABAergic interneurons and surrounding astrocytes by GABA transaminase (GABAT) (Petroff and Rothman, 1998). As a neurotransmitter GABA acts on two major families of post-synaptic GABA receptors. GABAA

Fig. 1. Typical GABA-edited MRS spectrum. (A) A typical voxel location for a 2  2  2 cm left M1 voxel. (B) A typical GABA-edited MEGA-PRESS spectrum acquired from the voxel illustrated in A. Representative peaks from GABA, Glx (a composite measure of glutamate and glutamine) and NAA are visible.

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

95

receptors are fast-acting ionotropic receptors which gate chloride channels. GABAB receptors, which are sloweracting metabotropic receptors linked to potassium channel receptors, are also present post-synaptically although additionally found pre-synaptically on GABA neurons and glutamatergic neurons. In addition to binding to post-synaptic GABA receptors, GABA also acts as a neuromodulator, by binding to non-synaptic GABAA receptors (Martin and Rimvall, 1993). This extra-synaptic GABAergic inhibition has been termed ‘GABAergic tone’, and appears to allow for more widespread, persistent inhibitory control of cellular firing than the point-to-point inhibition allowed by GABAergic synapses (Belelli et al., 2009). Importantly, decreasing this tonic inhibition is an important mechanism in the induction of plasticity within the motor cortex (Clarkson et al., 2010). Measurement of GABA in the resting brain MRS. Since the pioneering work of Rothman and colleagues (Rothman et al., 1993) it is now possible to measure the concentration of GABA in vivo in the human brain using MRS. At its simplest, MRS allows for the accurate quantification of endogenous metabolites within a single pre-defined region (or voxel) of tissue. A typical voxel placement for M1 is shown in Fig. 1A. A detailed explanation of the physical principles underlying MRS is outside the scope of this review (see for example Stagg and Rothman, 2014). In essence, the MRS signal is derived from differences in the molecular structure of metabolites. If the MR signal from the voxel of interest is plotted as a function of its frequency, characteristic patterns of peaks that relate to individual molecular structure, can be seen (Fig. 1B). The magnitude of these peaks reflect the concentration of the neurochemicals from which they arise. The spectrum can thus be used to quantify the concentrations of individual neurochemicals within the region of interest. A spectrum with sufficient signal-to-noise to allow accurate quantification can normally be acquired in an acquisition lasting in the order of a few minutes. In the human brain, quantifying GABA is less straightforward than for some other neurochemicals due to its low concentration and the overlap between its signal and those of other molecules found in higher concentrations such as creatine, glutamate and glutamine. Spectral editing methods are a common way to reduce signal overlap in order to reveal the GABA signal. One of the most widely applied editing methods to measure GABA is the MEscher-Garwood-Point RESolved Spectroscopy (MEGA-PRESS) sequence (Mescher et al., 1998). A disadvantage of this approach, as with all editing methods, is its reliance upon the subtraction of signals to remove strong overlapping frequencies from the spectrum in order to reveal GABA (Puts and Edden, 2012). This subtraction approach can be sensitive to any experimental noise such as subject movement or instrumental factors, which can then obscure the edited GABA signal (Waddell et al., 2007). Furthermore, some protein-containing macromolecules (Behar et al., 1994) resonate at a frequency coupled to a resonance near

Fig. 2. Relationship between MRS-assessed GABA and 1-ms SICI. (A) A significant relationship was demonstrated between MRSassessed GABA levels and 1-ms ISI SICI, suggesting that MRS GABA signal reflects extra-synaptic GABA tone (from Fig. 3, Stagg et al. (2011a)). (B) A decrease in M1 GABA levels was observed after anodal tDCS, but not after sham stimulation. No concomitant change in Glx (a composite measure of glutamate and glutamine) was seen (figure adapted from Fig. 1, Stagg et al. (2009)).

the GABA peak. Thus, particularly at lower field strengths, the edited GABA peak can include contributions from those macromolecules. However, the variation of macromolecules is likely to be low between subjects when compared with the variation of GABA, and therefore it is unlikely that this contamination provides a practical difficulty when interpreting MRS results (Kreis et al., 2005). GABA MRS is a promising technique, however, there are several limitations. The limited topographical resolution allows measurement from a relatively large area, typically 2  2  2 cm voxel of interest, making it difficult for more precise measurements. In addition, the limited temporal resolution – in the order of a few minutes – only allows for a resting state evaluation, thus making it difficult to do event-related, functional measurements. In designing GABA MRS studies there is a need to balance maximizing SNR and measurement precision with minimizing experiment time to increase subject comfort and therefore reduce motion artifacts to give better data quality. Although the spectral, temporal, and frequency resolution of MRS data can be improved with increased field strength (Puts and Edden, 2012) the number of sites with ultra-high field scanners are currently limited. One important limitation of MRS is that it quantifies the total signal arising from a particular neurochemical within

96

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

Fig. 3. ppTMS can be used to assess GABAergic inhibition during movement preparation. The degree of GABAA activity is significantly decreased during the movement preparation period. Data are presented for three age groups (young adults (aged 20–40 years; dark blue), middle aged subjects (40–60 years; blue) and older adults (60–90 years; light blue). GABAA-ergic inhibition can be seen immediately after the visual cue ( 5) in the younger two age groups, but this is significantly reduced immediately prior to movement (0). The dotted line represents no effect of the SICI conditioning pulse (i.e. no GABAA activity). Figure taken from Heise et al. (2013).

the voxel of interest. In the case of GABA, this means that it is not possible to differentiate between the GABA in the two neuronal pools and that found extracellularly. It is therefore unclear how the MRS-assessed GABA concentration relates to local synaptic activity. Indeed, a recent study demonstrated no significant relationship between MRS-assessed GABA concentration and Transcranial Magnetic Stimulation (TMS)-measures of GABAA and GABAB synaptic activity (see the Section ‘Measures of GABAergic activity during movement’ for more details) (Stagg et al., 2011b). However, a significant relationship was observed between MRS-assessed GABA and a TMS measure proposed to reflect extrasynaptic GABAergic tone (Stagg et al., 2011b) (Fig. 2). It is important to note, however, that this relationship was demonstrated between trait measures of MRSassessed GABA and TMS measures of GABAergic activity. It should not be inferred from this result that the changes in GABA levels due to plasticity induction protocols are caused solely by changes in extrasynaptic GABAergic tone. Rather, it is likely that induced changes in MRS-assessed GABA levels are due to modulation of both synaptic and extrasynaptic activity (Stagg, 2014), a hypothesis supported by animal studies showing different effects on GAD65 and GAD67 following rTMS protocols (Funke and Benali, 2011; Volz et al., 2013). TMS. TMS can also be used to quantify GABAergic activity in humans. TMS is a non-invasive neurostimulation technique that induces a flow of current in the brain via the principles of electromagnetic induction (Hallett, 2000). A magnetic field is generated via a brief current flowing through a coil of wire that is placed above the scalp. This in turn induces an electric current in the underlying brain tissue. When TMS is

applied to M1 it elicits a peripheral response and the effects of TMS can be measured via the study of electromyographic (EMG) recordings of the targeted muscle (Walsh and Rushworth, 1999). Of particular interest here are paired-pulse TMS (ppTMS) protocols, which have been shown to reflect distinct aspects of inhibitory processing within M1 with a relatively high degree of pharmacological specificity. As the name suggests, ppTMS approaches involve two TMS pulses being applied to the same region of cortex with a very short inter-stimulus interval (ISI). Shortinterval intra-cortical inhibition (SICI) utilizes a subthreshold conditioning stimulus (CS) followed a few milliseconds later by a supra-threshold test stimulus (TS), which elicits an MEP. The ratio of the amplitude of the conditioned to the unconditioned MEP is then calculated. ISI of 2–4 ms leads to a distinct inhibition of the conditioned test stimulus compared with an unconditioned stimulus (Kujirai et al., 1993) and has been shown to reflect GABAA-synaptic mediated inhibition (Ziemann et al., 1996; Ilic´ et al., 2002; Di Lazzaro et al., 2006). Inhibition of the test stimulus is also seen when using a SICI protocol with an ISI of 1 ms. It is less clear exactly what underlying cortical processes this inhibition reflects, but it is known to be a GABAergic phenomenon (Ni et al., 2007) with a distinct mechanism from the 2–4 ms ISI SICI (Fisher et al., 2002; Roshan et al., 2003) and has been postulated to reflect extrasynaptic GABAergic tone, mediated by extrasynaptic GABAA receptors (Stagg et al., 2011b). In addition to studying GABAA receptor activity, it is also possible to study GABAB synaptic activity using ppTMS. Long-Interval Intracortical Inhibition (LICI), a marker of GABAB activity, can be demonstrated by increasing the ISI to 100–200 ms and using a

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

suprathreshold CS (Valls-Sole´ et al., 1992; Werhahn et al., 1999; McDonnell et al., 2006). Although ppTMS is a powerful tool for investigating inhibitory cortical circuits, the ppTMS protocols mentioned above are not entirely specific to the receptor subtypes mentioned. For example, it has been shown that SICI and LICI interact with each other and it is unclear whether the same population of neurons mediates both these measures, or whether they are mediated by closely interacting interneuronal circuits (Sanger et al., 2001; Chen, 2004). Clarifying the interactions between distinct inhibitory ppTMS paradigms can give insights into the neurophysiological basis of the inhibitory effects they measure (Reis et al., 2008). Measures of GABAergic activity during movement Using ppTMS it is also possible to quantify GABAergic inhibition in the pre-movement period. Using a visuallycued movement task it is possible to measure GABAA activity over a series of times between the appearance of the visual cue and the onset of movement, thereby giving an insight into the role of GABAergic inhibition in movement preparation. In young healthy subjects, GABAA activity is at normal levels immediately after the appearance of the visual cue, but significantly decreased late in pre-movement period (i.e. shortly before movement occurs) (Heise et al., 2013) (see Fig. 3). Indeed, in older controls at the timepoints close to movement onset, the SICI protocol, which usually produces significant inhibition instead becomes facilitatory (Hummel et al., 2009). The excellent temporal resolution of the ppTMS approach has further highlighted important pathological changes in conditions such as stroke, where there is abnormal intracortical inhibition during movement preparation (Hummel et al., 2009). Similar patterns have also been shown in patients with Gilles de la Tourette syndrome (Heise et al., 2010) and dystonia (Gilio et al., 2003). Changes in GABAergic signaling with aging A number of studies have investigated the effects of aging on GABAergic activity using both ppTMS and MRS. A small MRS study investigating neurochemical changes in early adulthood demonstrated higher GABA levels in the orbitofrontal cortex, and, less markedly, the sensorimotor cortex in the 25–31-year old group compared with the 19–24-year old group (Grachev and Apkarian, 2000). A subsequent study showed a significant decrease in GABA levels in the dorsolateral prefrontal cortex in a group of subjects aged 40–52 years compared with the 19–31-year-old cohort (Grachev and Apkarian, 2001). A more recent, technically more robust and much larger study in 100 subjects supports this early evidence for a decrease in GABA with aging, showing a broadly linear decrease in GABA levels in the frontal and parietal lobes between the ages of 20 and 76 years (Gao et al., 2013). In line with this suggestion that GABA decreases with age, some ppTMS studies have shown an age-related

97

decrease in resting SICI (Peinemann et al., 2001; Heise et al., 2013). In line with this, a decrease in the Cortical Silent Period (CSP), a putative marker of GABAB activity, has also been observed with aging (Oliviero et al., 2006). However, another study showed no correlation between ppTMS measures and age across a large sample of subjects aged 18–76 (Wassermann, 2002), suggesting that age-related effects may be subtle. In addition to investigating changes in GABAergic signaling at rest with aging, some studies have shown age-related changes in pre-movement measures of SICI (see Section ‘Measures of GABAergic activity during movement’). A study of a group of older adults (aged 63.8 ± 3.0 years) showed a slight inhibition in response to the SICI protocol, which turned into significant facilitation in the immediate pre-movement period (Hummel et al., 2009). However, a larger study showed similarly reduced GABAergic activity early in the premovement period, but this was not modulated by movement preparation (Heise et al., 2013) (Fig. 3). Taken together, these results suggest that aging is probably associated with a decrease in GABAergic inhibition, though these effects may be subtle and their precise functional relevance is still not clear. However, it is important to remember when interpreting changes in GABA in stroke patients, that age-related changes in GABAergic signaling are present, and therefore an agematched control group is of prime importance in these studies.

GABA MODULATION IS DEMONSTRATED DURING PLASTICITY IN HEALTHY ADULTS Changes in muscle representations within M1 caused by short-term motor learning have been readily demonstrated in both animal and human studies (Pascual-Leone et al., 1994, 1995; Karni et al., 1995; Kleim et al., 1998). The size of cortical representation of a digit within M1 increases in response to the increased use of that digit in a motor task within minutes (Classen et al., 1998), and decrease in the use of a muscle group due to paralysis or extra-cortical infarct leads to a rapid concomitant decrease in cortical representation (Liepert et al., 1995; Nudo and Milliken, 1996). Horizontal fibers arising from the pyramidal neurons allow for the co-activation of adjacent cortical columns as well as modulating inhibitory activity via the activation of GABAergic interneurons (Jones, 1993). The observed shifts in cortical representations seen during learning appear to be facilitated by modulation of local inhibitory signaling within these horizontal connections (Jacobs and Donoghue, 1991; Castro-Alamancos et al., 1995; Rioult-Pedotti et al., 1998; Trepel and Racine, 2000; Nudo, 2013), thus giving a putative anatomical site for the role of inhibitory signaling in LTP-like plasticity. GABA decreases occur in response to a wide range of plasticity-induction protocols In healthy people, GABA has been shown to be involved in motor plasticity induced in variety of different ways. The first study to show a modulation of GABA levels within the

98

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

primary motor cortex (M1) in humans demonstrated a significant decrease in GABA in response to a forearm ischemic nerve block (Levy et al., 2002), an intervention known to increase cortical excitability and induce shortterm reorganization of M1 (Brasil-Neto et al., 1993; Ziemann et al., 1998). A significant decrease in GABA within M1 was also demonstrated during learning of a visuo-motor tracking task (Floyer-Lea et al., 2006). This decrease in GABA was specific to subjects who learnt the task - no change in M1 GABA was seen in response to a similar task without a learning component or if the subjects lay at rest, suggesting that the observed decrease in GABA was specific to motor learning and not simply as a result of motor performance. These findings are in line with those from TMS studies suggesting that the early phases of motor learning decreased SICI, the TMS measure of GABAA activity (Liepert et al., 1998, 2004; Perez et al., 2004; Rosenkranz et al., 2007). NIBS. Cortical excitability and LTP-like plasticity can also be modulated using NIBS techniques such as tDCS. tDCS relies on the passage of a very small electric current, in the order of 1–2 mA, through the brain most commonly via two large electrodes placed on the scalp (Nitsche and Paulus, 2000, 2001). The effects of tDCS on cortical excitability are polarity specific; such that anodal stimulation, where the anode is placed over the cortical region of interest is facilitatory, whereas cathodal stimulation, where the current direction is reversed, decreases excitability to the underlying cortex (Nitsche and Paulus, 2000). Anodal tDCS is thought to induce LTP-like plasticity within the stimulated cortex (Stagg and Nitsche, 2011), and has been shown to decrease GABA levels within the stimulated M1 (Stagg et al., 2009, 2011a). A decrease in GABA of approximately 10% has been observed after anodal tDCS (Fig. 2B), and a modulation of the same order of magnitude is observed after motor learning. Consistently, anodal tDCS has also been shown to decrease ppTMS measures of GABAA activity by approximately the same degree (Nitsche et al., 2005).

Can GABA decreases be related directly to learning of a motor task? From the data presented above, it would seem that a diverse range of plasticity induction protocols modulate GABA in healthy adults. However, it is not clear from these data whether or not the GABA decrease is necessary for plasticity to occur, as might be hypothesized from the animal literature, where a reduction of inhibition is a necessary first step in the induction of LTP-like plasticity in horizontal fiber pathways in M1 (Jacobs and Donoghue, 1991). Preventing this GABAergic decrease results in an inability to induce LTP-like plasticity within the cortex (CastroAlamancos et al., 1995), whereas reducing GABA inhibition facilitates LTP-like plasticity induction (Trepel and Racine, 2000). To begin to address this question we performed a study investigating the link between motor learning and decreases in inhibitory activity on a subject-by-subject basis (Stagg et al., 2011a). This study used anodal tDCS to M1 to investigate the responsiveness of the GABAergic system, by acquiring M1 GABA levels before and after 10 min of 1-mA stimulation. On a different day, subjects performed an explicit, visually-cued sequence-learning task during fMRI. A significant relationship was demonstrated between the responsiveness of the GABAergic system and the degree to which the subjects learnt the task, such that subjects who showed greater learning of the task were also those who had a greater decrease in their GABA levels in response to tDCS (Fig. 4). By analyzing the relationship between the learning-related fMRI signal and the tDCS-induced GABA change, this relationship could be localized to the left M1 (Stagg et al., 2011a). While these approaches cannot demonstrate causality, the relationship between the responsiveness of the GABAergic system and the ability of a subject to learn a task, as well as the anatomical specificity of this relationship, strongly supports the hypothesis that GABA decreases are essential for motor cortical plasticity to occur. Do NIBS protocols that modulate GABA levels also modulate learning rates?

Fig. 4. Link between tDCS-induced change in GABA and learning rates in a motor task. A significant relationship was demonstrated between GABA change induced by anodal tDCS on one day and the rate of learning of a motor task on another day. Figure adapted with permission from Stagg et al. (2011a).

If a local decrease in GABA levels is necessary for learning to occur, then NIBS protocols which are known to decrease GABA should also improve learning of a task performed during stimulation. Anodal tDCS has been shown to both decrease GABA levels within M1 and lead to significant improvements in learning of a variety of motor tasks (Nitsche et al., 2003; Galea and Celnik, 2009; Reis et al., 2009; Stagg et al., 2011c). However, the relative timing of the two interventions is critical – if the motor task takes place during stimulation then learning is facilitated, but when the motor task is preceded by anodal tDCS stimulation, learning is either not modulated by the stimulation (Kuo et al., 2008) or actually slowed (Stagg et al., 2011c). The time-dependent effects of anodal tDCS on motor learning are suggestive of a homeostatic interaction between the two interventions. Homeostatic plasticity

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

99

can be thought of as the ‘‘plasticity of plasticity’’ and acts within the brain to regulate on-going activity and to maintain cortical firing within a useful range. The most influential model of homeostatic plasticity was described by Bienenstock, Cooper and Munro in 1982 and known eponymously as the Bienenstock–Cooper–Munro (BCM) model (Bienenstock et al., 1982). The BCM model proposes that there is a level of post-synaptic activity, termed the modification threshold (hm), above which LTP is induced and below which long-term depression (LTD) is induced (Bienenstock et al., 1982). The modification threshold is a dynamic entity, and can be raised or lowered based on the previous time-averaged level of postsynaptic cell firing. In practical terms, the BCM model suggests that the application of one stimulus that induces LTP-like plasticity (in this example anodal tDCS) will cause a subsequent LTP-like inducing stimulus (in this example motor learning), to induce LTD-like plasticity. Conversely, a priming stimulus that induces LTD will increase the LTP-like effects seen after a subsequent LTP-inducing stimulus, compared with when that LTP-inducing stimulus is applied without a prior priming stimulus. GABAergic synapses as a site for homeostatic plasticity According to the BCM model, homeostatic plasticity occurs via modifications in synaptic mechanisms and has been well characterized for glutamatergic synapses (Lee et al., 2010; Pozo and Goda, 2010). The study of the contribution of GABA to such homeostatic processes has long focused on its ability to indirectly shift hm via increasing the NMDA response and intracellular Ca2+ concentration (Abraham and Tate, 1997), or by autoregulatory inhibition of GABA release as a result of postsynaptic excitation of neighboring terminals (Chevaleyre and Castillo, 2004). However, recent work in animals suggests that specific inhibitory interneuronal subtypes show homeostatic plasticity in Aplysia (Fischer et al., 1997), as well as in the mammalian spinal dorsal horn (Miletic et al., 2001); hippocampus (Karmarkar and Buonomano, 2006; Bukalo et al., 2007; Edwards et al., 2008); and neocortex (Bartley et al., 2008). In addition, a recent study in rats has demonstrated GABAergic involvement in homeostatic plasticity arising from repeated trains of intermittent TBS (iTBS) and continuous TBS (cTBS) (Volz et al., 2013). The evidence for the role of GABAergic synapses in homeostatic plasticity in humans is sparse. The first paper to investigate homeostatic plasticity in humans demonstrated a homeostatic interaction between tDCS priming and subsequent low-intensity 1-Hz repetitive TMS (rTMS), a NIBS protocol that would normally not be expected to modulate plasticity in terms of overall cortical excitability changes (Siebner et al., 2004) (Fig. 5A). However, there was no clear homeostatic effect on ppTMS measures of GABAergic signaling. Two subsequent studies have investigated potential homeostatic interactions between the two trains of theta-burst TMS (TBS). TBS is a TMS protocol designed to induce LTP-like or LTD-like effects through the application of groups of three pulses at 50 Hz,

Fig. 5. Homeostatic plasticity. (A) Homeostatic interactions between tDCS and 1-Hz rTMS. Anodal tDCS increases cortical excitability (as demonstrated by an increase in MEP amplitude; filled upward triangles). Cathodal tDCS decreases cortical excitability (as demonstrated by a decrease in MEP amplitude; filled downward triangles). Sham stimulation (circles) had no effect on cortical excitability. 1-Hz rTMS led to no change in cortical excitability after sham tDCS. However, when 1-Hz rTMS was primed by anodal tDCS, it led to cortical inhibition, in accordance with the BCM model of homeostatic plasticity. 1-Hz rTMS primed by cathodal tDCS led to an increase in cortical excitability (figure adapted from Siebner et al. (2004)). (B) Theta-Burst Stimulation (TBS) protocols. (C) SICI changes after TBS priming. iTBS, a facilitatory protocol, led to no significant change in SICI measures when applied alone. However, when primed by a prior iTBS train (green lines) a significant decrease in SICI was seen (figure adapted from Murakami et al. (2012)).

repeated every 200 ms. If given as a continuous stream for 40 s (600 pulses; cTBS), LTD-like plasticity results, whereas if 2 s of stimulation is given followed by 8 s of rest (iTBS), 600 TMS pulses induce an LTP-like effect within the stimulated cortex (Huang et al., 2005) (Fig. 5B). The first paper to study the effects of two applications of TBS trains demonstrated significant homeostatic interactions between the trains in terms of overall cortical excitability, similar to those seen by Siebner and colleagues described above, but did not demonstrate any evidence of homeostatic effects at GABAergic synapses (Doeltgen and Ridding, 2011).

100

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

A more recent study also investigated homeostatic interactions between TBS trains, and demonstrated similar homeostatic interactions in terms of overall effects on cortical excitability to those seen by Doeltgen and Ridding (Murakami et al., 2012). Like Doeltgen and Ridding, Murakami and colleagues also went on to investigate the role of GABAA synapses in these homeostatic interactions using ppTMS SICI with a 2-ms ISI. Neither iTBS nor cTBS when applied alone had any effects on SICI measures. However, significant changes in GABAA-ergic activity were elicited in response to TBS priming. Recall that iTBS typically leads to LTP-like effects, whereas cTBS typically produces LTD-like effects (Huang et al., 2005). When primed by a preceding iTBS train, iTBS led to a decrease in SICI (i.e., less inhibition) (Fig. 5C), whereas cTBS preceded by a train of cTBS led to an increase in SICI (i.e., more inhibition). iTBS priming by cTBS and cTBS priming by iTBS elicited no significant effects on GABAergic signaling. The results of this study are particularly interesting as they suggest for the first time that GABAA synapses play a role in homeostatic plasticity in humans. However, further studies are needed to understand whether similar homeostatic interactions are seen between plasticityinduction protocols that differ in nature but nevertheless act on the same synapses or micro-circuits. Perhaps of particular interest is the physiological basis of the interaction between anodal tDCS and motor learning. Not only does the dependence of the behavioral effect on the relative timing of the two interventions have important implications for the use of anodal tDCS as a putative therapeutic tool in stroke rehabilitation, but the demonstrated homeostatic interaction between anodal tDCS and motor learning suggests that the two interventions modulate the same synapses. Evidence for a role of GABAergic synapses in this homeostatic interaction would therefore add significant weight to the hypothesis that anodal tDCS facilitates concurrent motor learning via modulation of GABAergic activity.

A ROLE FOR GABA IN LONG-TERM PLASTICITY? As detailed above, there is growing evidence for a decrease in GABA activity to be necessary for glutamatergic LTP-like plasticity to be induced. What is less clear from the literature is whether long-term maintenance of synaptic plasticity is sustained by GABAergic modulation. In animal models, significant remodeling of motor cortical representations occurs after an experimental lesion (Nudo et al., 1996). In humans, the most robust example of long-term motor cortical remodeling and plasticity is in the recovery of function after a stroke. There is increasing evidence that GABAergic modulation plays a part in this long-term remodeling. Administration of the GABAA agonist Midazolam to well-recovered stroke patients has been shown to result in a re-emergence of their former symptoms (Lazar et al., 2010). Importantly, focal motor deficits were re-induced only in patients with prior hemiparesis, and aphasia re-induced only in patients with former language

impairment. The specificity of these results indicates that the administration of the GABAA agonist temporarily disrupted the plastic changes that had led to functional recovery, supporting the hypothesis that modulation of GABAergic signaling is necessary for the maintenance of functional improvements following stroke. Other studies have used TMS measures of GABA to investigate physiological changes during stroke recovery. At rest, SICI, a marker of GABAA synaptic activity, is significantly reduced in the ipsilesional hemisphere in the acute and sub-acute phases of recovery after stroke (Liepert et al., 2000; Swayne et al., 2008). There is some evidence that indicates this decrease in GABAA signaling normalizes with recovery (Talelli et al., 2006), although more recent studies have suggested that it is maintained into the chronic phase of recovery after stroke (Blicher et al., 2009; Hummel et al., 2009). Pre-movement SICI in the acute phase is also a predictor of recovery, such that patients who show greater disinhibition are also those who show greater recovery, independent of the initial level of impairment (Liuzzi et al., 2014). Blicher and colleagues reported lowered SICI in the ipsilesional M1 in patients in the chronic phase (>6 months) of recovery after a subcortical stroke. Little or no spontaneous improvement in function would be expected in these patients, so this finding suggests that intracortical disinhibition is important both for functional recovery during the acute and sub-acute recovery period, and in maintaining motor function in the chronic stage. Blicher and colleagues went on to investigate GABAergic changes in response to a short-term motor learning task. In age-matched healthy controls inhibitory activity decreased for up to 30 min after training. However, no change in SICI was found in the stroke group after training (Blicher et al., 2009). As the authors discuss, these results are somewhat difficult to interpret for technical reasons, however they do suggest that GABAergic processing is altered after long-term major cortical reorganization, as seen in the chronic stages of stroke recovery. In addition to studying GABAergic activity at rest, important information about the functional relevance of abnormalities in inhibitory processing can also be quantified using TMS measures of GABAA synaptic activity. As discussed in the Section ‘Measures of GABAergic activity during movement’, in healthy people, GABAA activity within M1 decreases markedly in the milliseconds preceding cued movement – resulting in the SICI protocol, producing markedly less inhibition immediately before movement onset (Heise et al., 2010). In stroke patients, even though GABAA inhibitory activity is lower than in healthy controls at rest (Blicher et al., 2009), a decrease in SICI in the pre-movement period is not evident in the ipsilesional M1 when patients move their stroke-affected hand (Hummel et al., 2009). Taken together, these findings suggest that although GABAA activity may be abnormally reduced at rest, consistent with long-term cortical remodeling, an inability of GABAergic signaling to respond appropriately to cortical

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

101

demands may still be responsible for some of the difficulties experienced by patients when moving.

GABAergic signaling is necessary for LTP-like plasticity to occur in humans.

FUTURE DIRECTIONS

A role for GABA modulation in the chronic stage of stroke recovery?

In this review we have discussed the evidence that GABAergic processing is important in plasticity induction in humans. Understanding how the brain learns new skills and recovers motor function after stroke is of vital importance when developing new rehabilitation strategies to optimize patient outcome. However, a number of questions remain as to the role of inhibitory processing in post-stroke recovery, which currently limit attempts to reach this goal. Is GABA modulation necessary for short-term plasticity induction? As discussed in the Section ‘GABA modulation is demonstrated during plasticity in healthy adults’, an increasing body of evidence from both MRS and TMS studies suggest that a decrease in GABAergic inhibition occurs during short-term learning in healthy controls. There is a close relationship between the degree of GABA modulation by anodal tDCS and the rate of learning of a motor task. In addition, an intervention known to decrease GABA, anodal tDCS, has also been shown to improve motor learning. However, it is difficult to ascertain causality in vivo. A potential method to investigate whether a decrease in GABA is necessary for motor learning is to use a pharmacological manipulation. Unfortunately nonselective GABAA agonists have sedative effects that make the interpretation of the results of these studies difficult, and GABA antagonists reduce the seizure threshold, meaning that this approach has a limited utility in humans. More recently, highly selective agonists for specific subtypes of GABAA receptors have been developed, which may provide a fruitful avenue for this research in the future. A more indirect method of ascertaining whether GABA is necessary in motor learning in humans is to study homeostatic plasticity. As discussed in detail in the Section ‘GABAergic synapses as a site for homeostatic plasticity’, homeostatic interactions only occur between two plasticity-inducing protocols when they affect the same synapses or microcircuits within the cortex. The behavioral effects of anodal tDCS on a motor learning task appear to follow the rules of homeostatic plasticity, although to date the evidence for GABAA synapses being a site of homeostatic plasticity in the human is limited. Interestingly, a recent study has suggested that homeostatic interactions between NIBS protocols may occur at GABAA synapses (Murakami et al., 2012). If the homeostatic interaction between anodal tDCS and motor learning could be similarly demonstrated to occur, at least in part, at GABAA synapses then this would greatly strengthen the hypothesis that the modulation of inhibitory signaling is a commonality to both anodal tDCS and motor learning. This would therefore greatly strengthen the hypothesis that the modulation of

The role of inhibitory signaling in long-term plasticity underlying stroke recovery is less well studied than its role in short-term plasticity induction. Although there is relatively strong evidence that a decrease in inhibitory processing occurs in the acute and sub-acute stages of recovery after stroke, there is less evidence to support the hypothesis that modulation of inhibitory processing is important in the maintenance of long-term plasticity as seen in the chronic stages of recovery after stroke. Establishing the role of GABAergic modulation in the maintenance of cortical remodeling is of vital importance both when designing putative new adjunct therapies and when considering the use of GABAergic agonists, such as Baclofen, in the treatment of spasticity in this patient group. Functionally-relevant modulation of GABA As discussed in the Section ‘A role for GABA in long-term plasticity?’, GABAA synaptic activity is decreased in patients in the chronic stages of recovery after stroke compared with age-matched healthy controls (Blicher et al., 2009), a finding in line with the hypothesis that cortical remodeling following stroke is accompanied by local decreases in inhibitory signaling. However, when healthy people are asked to move their hand, this movement is accompanied by a significant reduction in GABAA activity within that muscle representation in the few milliseconds preceding movement. This decrease in inhibition is not seen in patients in the chronic stage of recovery after stroke (Hummel et al., 2009). These initially contradictory findings do not mean that studies of motor cortical inhibition at rest are not valuable, but rather are a reminder of the complexity of inhibitory signaling within the cortex, and highlight the importance of studying GABAergic modulation not only at rest but also in the context of movement. Exploring the physiological underpinnings of the MRS-assessed GABA signal – the necessity for multimodal approaches As discussed in detail in the Section ‘MRS’, MRS gives a sensitive measure of GABA concentration within a specific region of tissue, but cannot differentiate between GABA in different pools within the cortex, nor between different types of inhibitory interneurons. Recent studies investigating the effects of rTMS on excitatory and inhibitory processing have highlighted that different classes of GABAergic interneurons are differentially modulated by rTMS (see Funke and Benali, 2011 for a detailed review). Future work, either combining MRS with detailed physiological studies in animal models, or using MEG or EEG as surrogate measures of specific interneuronal pools, will be necessary to fully elucidate the effects of transcranial stimulation on different GABAergic interneurons.

102

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

M1 as part of a wider functional network The studies reviewed here have investigated the role of local inhibitory processing within M1 in motor plasticity, and indeed a discussion of the plasticity within distant regions is not within the scope of this review (see for example, Stagg and Johansen-Berg, 2013 for a review). However, the ipsilesional M1 does not act in isolation during recovery after stroke – a wealth of evidence suggests that significant modulation occurs in other closely connected motor regions, for example the contralateral M1, the premotor cortices bilaterally, and the supplementary motor area (see for example, Grefkes and Fink, 2014 for a review). It is not clear, however, how local inhibitory processing is related to these network-level functional connectivity changes, although evidence is emerging that behavioral improvements induced by tDCS applied to the contralesional M1 can be related to changes in local inhibitory signaling (Zimerman et al., 2012). A recent study by our group demonstrated a relationship in healthy subjects between inhibitory activity within M1 and the strength of functional connectivity within the wider motor network, such that lower MRS-assessed GABA levels were associated with greater functional connectivity (Stagg et al., 2014). A similar finding has been shown between the posteromedial cortex and the default mode network (Kapogiannis et al., 2013). The relationship between the well-described changes in network-wide connectivity after stroke and local inhibitory processing within major network nodes such as the ipsilesional M1 needs to be explored.

CONCLUSIONS This review aimed to summarize the techniques by which GABA can be quantified in vivo in humans, before discussing the current evidence concerning the role of inhibitory processing in plasticity, both in the context of short-term motor learning and in recovery after stroke. Although there is an increasing body of evidence that a decrease in local inhibitory signaling is important in human motor plasticity, a number of questions remain to be addressed before therapies targeted at reducing GABA levels can be optimized as potential adjunct rehabilitative strategies. If these can be answered, it is to be hoped that the role of GABAergic processing in LTP-like motor cortical plasticity could be a fruitful avenue to explore in the development of novel rehabilitative therapies. Acknowledgments—V.B. is funded by the Clarendon Fund Scholarship. C.J.S. holds a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (Grant Number 102584/Z/13/Z).

REFERENCES Abraham WC, Tate WP (1997) Metaplasticity: a new vista across the field of synaptic plasticity. Prog Neurobiol 52:303–323. Andersson JD, Halldin C (2013) PET radioligands targeting the brain GABA A/benzodiazepine receptor complex [Dolle´ F, ed.]. J Label Compd Radiopharm 56:196–206.

Bartley AF, Huang ZJ, Huber KM, Gibson JR (2008) Differential activity-dependent, homeostatic plasticity of two neocortical inhibitory circuits. J Neurophysiol 100:1983–1994. Bartos M, Vida I, Jonas P (2007) Synaptic mechanisms of synchronized gamma oscillations in inhibitory interneuron networks. Nat Rev Neurosci 8:45–56. Behar K, Rothman D, Spencer D, Petroff O (1994) Analysis of macromolecule resonances in 1H NMR spectra of human brain. Magn Reson Med 32:294–302. Belelli D, Harrison NL, Maguire J, Macdonald RL, Walker MC, Cope DW (2009) Extrasynaptic GABAA receptors: form, pharmacology, and function. J Neurosci 29:12757–12763. Bienenstock E, Cooper L, Munro P (1982) Theory for the development of neuron selectivity: orientation specificity and binocular interaction in visual cortex. J Neurosci 2:32–48. Blicher JU, Jakobsen J, Andersen G, Nielsen JF (2009) Cortical excitability in chronic stroke and modulation by training: a TMS study. Neurorehabil Neural Repair 23:486–493. Brasil-Neto J, Valls-Sole J, Pascual-Leone A, Cammarota A, Amassian V, Cracco R, Maccabee P, Cracco J, Hallett M, Cohen L (1993) Rapid modulation of human cortical motor outputs following ischaemic nerve block. Brain 116(Pt 3):511–525. Bukalo O, Schachner M, Dityatev A (2007) Hippocampal metaplasticity induced by deficiency in the extracellular matrix glycoprotein tenascin-R. J Neurosci 27:6019–6028. Castro-Alamancos MA, Donoghue JP, Connors BW (1995) Different forms of synaptic plasticity in somatosensory and motor areas of the neocortex. J Neurosci 15:5324–5333. Chen R (2004) Interactions between inhibitory and excitatory circuits in the human motor cortex. Exp Brain Res 154:1–10. Chevaleyre V, Castillo PE (2004) Endocannabinoid-mediated metaplasticity in the hippocampus. Neuron 43:871–881. Clarkson AN, Huang BS, Macisaac SE, Mody I, Carmichael ST (2010) Reducing excessive GABA-mediated tonic inhibition promotes functional recovery after stroke. Nature 468:305–309. Classen J, Liepert J, Wise SP, Hallett M, Cohen L (1998) Rapid plasticity of human cortical movement representation induced by practice. J Neurophysiol 79:1117–1123. Di Lazzaro V, Pilato F, Dileone M, Ranieri F, Ricci V, Profice P, Bria P, Tonali PA, Ziemann U (2006) GABAA receptor subtype specific enhancement of inhibition in human motor cortex. J Physiol (Lond) 575:721–726. Doeltgen SH, Ridding MC (2011) Modulation of cortical motor networks following primed h burst transcranial magnetic stimulation. Exp Brain Res 215:199–206. Edwards DA, Zhang L, Alger BE (2008) Metaplastic control of the endocannabinoid system at inhibitory synapses in hippocampus. Proc Natl Acad Sci 105:8142–8147. Feldman DE (2009) Synaptic mechanisms for plasticity in neocortex. Annu Rev Neurosci 32:33–55. Fischer TM, Blazis DE, Priver NA, Carew TJ (1997) Metaplasticity at identified inhibitory synapses in Aplysia. Nature 389:860–865. Fisher RJ, Nakamura Y, Bestmann S, Rothwell JC, Bostock H (2002) Two phases of intracortical inhibition revealed by transcranial magnetic threshold tracking. Exp Brain Res 143:240–248. Floyer-Lea A, Wylezinska M, Kincses T, Matthews PM (2006) Rapid modulation of GABA concentration in human sensorimotor cortex during motor learning. J Neurophysiol 95:1639–1644. Funke K, Benali A (2011) Modulation of cortical inhibition by rTMS – findings obtained from animal models. J Physiol (Lond) 589:4423–4435. Galea JM, Celnik P (2009) Brain polarization enhances the formation and retention of motor memories. J Neurophysiol 102:294–301. Gao F, Edden RAE, Li M, Puts NAJ, Wang G, Liu C, Zhao Bin, Wang H, Bai X, Zhao C, Wang X, Barker PB (2013) Edited magnetic resonance spectroscopy detects an age-related decline in brain GABA levels. NeuroImage 78:75–82. Gilio F, Curra` A, Inghilleri M, Lorenzano C, Suppa A, Manfredi M, Berardelli A (2003) Abnormalities of motor cortex excitability preceding movement in patients with dystonia. Brain 126:1745–1754.

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104 Grachev ID, Apkarian AV (2000) Chemical heterogeneity of the living human brain: a proton MR spectroscopy study on the effects of sex, age, and brain region. NeuroImage 11:554–563. Grachev ID, Apkarian AV (2001) Aging alters regional multichemical profile of the human brain: an in vivo 1H-MRS study of young versus middle-aged subjects. J Neurochem 76:582–593. Grefkes C, Fink GR (2014) Connectivity-based approaches in stroke and recovery of function. Lancet Neurol 13:206–216. Hallett M (2000) Transcranial magnetic stimulation and the human brain. Nature 406:147–150. Hebb D (1949) The organization of behaviour. Heise KF, Steven B, Liuzzi G, Thomalla G, Jonas M (2010) Altered modulation of intracortical excitability during movement preparation in Gilles de la Tourette syndrome. Brain 133:580–590. Heise KF, Zimerman M, Hoppe J, Gerloff C, Wegscheider K, Hummel FC (2013) The aging motor system as a model for plastic changes of GABA-mediated intracortical inhibition and their behavioral relevance. J Neurosci 33:9039–9049. Huang Y, Edwards M, Rounis E, Bhatia K, Rothwell JC (2005) Theta burst stimulation of the human motor cortex. Neuron 45:201–206. Hummel FC, Steven B, Hoppe J, Heise K, Thomalla G, Cohen LG, Gerloff C (2009) Deficient intracortical inhibition (SICI) during movement preparation after chronic stroke. Neurology 72:1766–1772. Ilic´ TV, Meintzschel F, Cleff U, Ruge D, Kessler KR, Ziemann U (2002) Short-interval paired-pulse inhibition and facilitation of human motor cortex: the dimension of stimulus intensity. J Physiol (Lond) 545:153–167. Jacobs K, Donoghue J (1991) Reshaping the cortical motor map by unmasking latent intracortical connections. Science 251:944–947. Jones EG (1993) GABAergic neurons and their role in cortical plasticity in primates. Cereb Cortex 3:361–372. Kapogiannis D, Reiter DA, Willette AA, Mattson MP (2013) Posteromedial cortex glutamate and GABA predict intrinsic functional connectivity of the default mode network. NeuroImage 64:112–119. Karmarkar UR, Buonomano DV (2006) Different forms of homeostatic plasticity are engaged with distinct temporal profiles. Eur J Neurosci 23:1575–1584. Karni A, Meyer G, Jezzard P, Adams MM, Turner R, Ungerleider LG (1995) Functional MRI evidence for adult motor cortex plasticity during motor skill learning. Nature 377:155–158. Kleim J, Barbay S, Nudo R (1998) Functional reorganization of the rat motor cortex following motor skill learning. J Neurophysiol 80:3321–3325. Krakauer JW (2006) Motor learning: its relevance to stroke recovery and neurorehabilitation. Curr Opin Neurol 19:84–90. Kreis R, Slotboom J, Hofmann L, Boesch C (2005) Integrated data acquisition and processing to determine metabolite contents, relaxation times, and macromolecule baseline in single examinations of individual subjects. Magn Reson Med 54:761–768. Kujirai T, Caramia MD, Rothwell JC, Day BL, Thompson PD, Ferbert A, Wroe S, Asselman P, Marsden CD (1993) Corticocortical inhibition in human motor cortex. J Physiol (Lond) 471:501–519. Kuo M-F, Unger M, Liebetanz D, Lang N, Tergau F, Paulus W, Nitsche MA (2008) Limited impact of homeostatic plasticity on motor learning in humans. Neuropsychologia 46:2122–2128. Lau CG, Murthy VN (2012) Activity-dependent regulation of inhibition via GAD67. J Neurosci 32:8521–8531. Lazar RM, Berman MF, Festa JR, Geller AE, Matejovsky TG, Marshall RS (2010) GABAergic but not anti-cholinergic agents reinduce clinical deficits after stroke. J Neurol Sci 292:72–76. Lee M-C, Lee M-C, Yasuda R, Yasuda R, Ehlers MD, Ehlers MD (2010) Metaplasticity at single glutamatergic synapses. Neuron 66:859–870. Levy LM, Ziemann U, Chen R, Cohen LG (2002) Rapid modulation of GABA in sensorimotor cortex induced by acute deafferentation. Ann Neurol 52:755–761.

103

Liepert J, Classen J, Cohen LG, Hallett M (1998) Task-dependent changes of intracortical inhibition. Exp Brain Res 118:421–426. Liepert J, Storch P, Fritsch A, Weiller C (2000) Motor cortex disinhibition in acute stroke. Clin Neurophysiol 111:671–676. Liepert J, Tegenthoff M, Malin J (1995) Changes of cortical motor area size during immobilization. Electroencephalogr Clin Neurophysiol 97:382–386. Liepert J, Weiss T, Meissner W, Steinru¨cke K, Weiller C (2004) Exercise-induced changes of motor excitability with and without sensory block. Brain Res 1003:68–76. Liuzzi G, Horniss V, Lechner P, Hoppe J, Heise K, Zimerman M, Gerloff C, Hummel FC (2014) Development of movement-related intracortical inhibition in acute to chronic subcortical stroke. Neurology 82:198–205. Logothetis NK (2008) What we can do and what we cannot do with fMRI. Nature 453:869–878. Martin DL, Barke KE (1998) Are GAD65 and GAD67 associated with specific pools of GABA in brain? Perspect Dev Neurobiol 5:119–129. Martin DL, Rimvall K (1993) Regulation of gamma-aminobutyric acid synthesis in the brain. J Neurochem 60:395–407. McDonnell MN, Orekhov Y, Ziemann U (2006) The role of GABA(B) receptors in intracortical inhibition in the human motor cortex. Exp Brain Res 173:86–93. Mescher M, Merkle H, Kirsch J, Garwood M, Gruetter R (1998) Simultaneous in vivo spectral editing and water suppression. NMR Biomed 11:266–272. Miletic G, Miletic G, Miletic V, Miletic V (2001) Contribution of GABAA receptors to metaplasticity in the spinal dorsal horn. Pain 90:157–162. Murakami T, Mu¨ller-Dahlhaus F, Lu M-K, Ziemann U (2012) Homeostatic metaplasticity of corticospinal excitatory and intracortical inhibitory neural circuits in human motor cortex. J Physiol (Lond) 590:5765–5781. Ni Z, Gunraj C, Chen R (2007) Short interval intracortical inhibition and facilitation during the silent period in human. J Physiol (Lond) 583:971–982. Nitsche M, Paulus W (2000) Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. J Physiol 527:633–639. Nitsche M, Schauenburg A, Lang N, Liebetanz D, Exner C, Paulus W, Tergau F (2003) Facilitation of implicit motor learning by weak transcranial direct current stimulation of the primary motor cortex in the human. J Cogn Neurosci 15:619–626. Nitsche MA, Paulus W (2001) Sustained excitability elevations induced by transcranial DC motor cortex stimulation in humans. Neurology 57:1899–1901. Nitsche MA, Seeber A, Frommann K, Klein CC, Rochford C, Nitsche MS, Fricke K, Liebetanz D, Lang N, Antal A, Paulus W, Tergau F (2005) Modulating parameters of excitability during and after transcranial direct current stimulation of the human motor cortex. J Physiol (Lond) 568:291–303. Nudo R, Milliken G (1996) Reorganization of movement representations in primary motor cortex following focal ischemic infarcts in adult squirrel monkeys. J Neurophysiol 75:2144–2149. Nudo R, Wise B, SiFuentes F, Milliken G (1996) Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct. Science 272:1791–1794. Nudo RJ (2013) Recovery after brain injury: mechanisms and principles. Front Hum Neurosci 7:887. Oliviero A, Profice P, Tonali PA, Pilato F, Saturno E, Dileone M, Ranieri F, Di Lazzaro V (2006) Effects of aging on motor cortex excitability. Neurosci Res 55:74–77. Pascual-Leone A, Grafman J, Hallett M (1994) Modulation of cortical motor output maps during development of implicit and explicit knowledge. Science 263:1287–1289. Pascual-Leone A, Nguyet D, Cohen LG, Brasil-Neto J, Cammarota A, Hallett M (1995) Modulation of muscle responses evoked by transcranial magnetic stimulation during the acquisition of new fine motor skills. J Neurophysiol 74:1037–1045.

104

V. Bachtiar, C. J. Stagg / Neuroscience 278 (2014) 93–104

Peinemann A, Lehner C, Conrad B, Siebner HR (2001) Age-related decrease in paired-pulse intracortical inhibition in the human primary motor cortex. Neurosci Lett 313:33–36. Perez MA, Lungholt BKS, Nyborg K, Nielsen JB (2004) Motor skill training induces changes in the excitability of the leg cortical area in healthy humans. Exp Brain Res 159:197–205. Petroff OA, Rothman DL (1998) Measuring human brain GABA in vivo: effects of GABA-transaminase inhibition with vigabatrin. Mol Neurobiol 16:97–121. Pozo K, Goda Y (2010) Unraveling mechanisms of homeostatic synaptic plasticity. Neuron 66:337–351. Puts NAJ, Edden RAE (2012) In vivo magnetic resonance spectroscopy of GABA: a methodological review. Prog Nucl Magn Reson Spectrosc 60:29–41. Reis J, Schambra H, Cohen LG, Buch ER, Fritsch B, Zarahn E, Celnik P, Krakauer J (2009) Noninvasive cortical stimulation enhances motor skill acquisition over multiple days through an effect on consolidation. Proc Natl Acad Sci 106: 1590–1595. Reis J, Swayne OB, Vandermeeren Y, Camus M, Dimyan MA, HarrisLove M, Perez MA, Ragert P, Rothwell JC, Cohen LG (2008) Contribution of transcranial magnetic stimulation to the understanding of cortical mechanisms involved in motor control. J Physiol (Lond) 586:325–351. Rioult-Pedotti M, Friedman D, Hess G, Donoghue J (1998) Strengthening of horizontal cortical connections following skill learning. Nat Neurosci 1:230–234. Rosenkranz K, Kacar A, Rothwell J (2007) Differential modulation of motor cortical plasticity and excitability in early and late phases of human motor learning. J Neurosci 27:12058–12066. Roshan L, Paradiso GO, Chen R (2003) Two phases of short-interval intracortical inhibition. Exp Brain Res 151:330–337. Rothman D, Petroff O, Behar K, Mattson R (1993) Localized 1H NMR measurements of gamma-aminobutyric acid in human brain in vivo. Proc Natl Acad Sci U S A 90:5662–5666. Sanes JN, Donoghue JP (2000) Plasticity and primary motor cortex. Annu Rev Neurosci 23:393–415. Sanger TD, Garg RR, Chen R (2001) Interactions between two different inhibitory systems in the human motor cortex. J Physiol (Lond) 530:307–317. Siebner HR, Lang N, Rizzo V, Nitsche MA, Paulus W, Lemon RN, Rothwell JC (2004) Preconditioning of low-frequency repetitive transcranial magnetic stimulation with transcranial direct current stimulation: evidence for homeostatic plasticity in the human motor cortex. J Neurosci 24:3379–3385. Stagg C, Rothman D (2014) Magnetic resonance spectroscopy. Elsevier Inc. Stagg CJ (2014) Magnetic resonance spectroscopy as a tool to study the role of GABA in motor-cortical plasticity. NeuroImage 86:19–27. Stagg CJ, Bachtiar V, Amadi U, Gudberg CA, Ilie AS, SampaioBaptista C, et al (2014) Local GABA concentration is related to network-level resting functional connectivity. eLife 3:e01465. Stagg CJ, Bachtiar V, Johansen-Berg H (2011a) The role of GABA in human motor learning. Curr Biol 21:480–484. Stagg CJ, Best JG, Stephenson MC, O’Shea J, Wylezinska M, Kincses ZT, Morris PG, Matthews PM, Johansen-Berg H (2009) Polarity-sensitive modulation of cortical neurotransmitters by transcranial stimulation. J Neurosci 29:5202–5206.

Stagg CJ, Bestmann S, Constantinescu AO, Moreno LM, Allman C, Mekle R, Woolrich M, Near J, Johansen-Berg H, Rothwell JC (2011b) Relationship between physiological measures of excitability and levels of glutamate and GABA in the human motor cortex. J Physiol (Lond) 589:5845–5855. Stagg CJ, Jayaram G, Pastor D, Kincses ZT, Matthews PM, Johansen-Berg H (2011c) Polarity and timing-dependent effects of transcranial direct current stimulation in explicit motor learning. Neuropsychologia 49:800–804. Stagg CJ, Johansen-Berg H (2013) Studying the effects of transcranial direct-current stimulation in stroke recovery using magnetic resonance imaging. Front Hum Neurosci 7:1–8. Stagg CJ, Nitsche MA (2011) Physiological basis of transcranial direct current stimulation. The Neuroscientist 17:37–53. Swayne OB, Rothwell JC, Ward NS, Greenwood RJ (2008) Stages of motor output reorganization after hemispheric stroke suggested by longitudinal studies of cortical physiology. Cereb Cortex 18:1909–1922. Talelli P, Greenwood RJ, Rothwell JC (2006) Arm function after stroke: neurophysiological correlates and recovery mechanisms assessed by transcranial magnetic stimulation. Clin Neurophysiol 117:1641–1659. Trepel C, Racine RJ (2000) GABAergic modulation of neocortical longterm potentiation in the freely moving rat. Synapse 35:120–128. Valls-Sole´ J, Pascual-Leone A, Wassermann EM, Hallett M (1992) Human motor evoked responses to paired transcranial magnetic stimuli. Electroencephalogr Clin Neurophysiol 85:355–364. Volz LJ, Benali A, Mix A, Neubacher U, Funke K (2013) Dosedependence of changes in cortical protein expression induced with repeated transcranial magnetic theta-burst stimulation in the rat. Brain Stimul 6:598–606. Waddell KW, Avison MJ, Joers JM, Gore JC (2007) A practical guide to robust detection of GABA in human brain by J-difference spectroscopy at 3 T using a standard volume coil. Magn Reson Imaging 25:1032–1038. Walsh V, Rushworth M (1999) A primer of magnetic stimulation as a tool for neuropsychology. Neuropsychologia 37:125–135. Wassermann EM (2002) Variation in the response to transcranial magnetic brain stimulation in the general population. Clin Neurophysiol 113:1165–1171. Wei J, Wu J-Y (2008) Post-translational regulation of L-glutamic acid decarboxylase in the brain. Neurochem Res 33:1459–1465. Wei J, Zhang M, Zhu Y, Wang J-H (2004) Ca(2+)-calmodulin signalling pathway up-regulates GABA synaptic transmission through cytoskeleton-mediated mechanisms. Neuroscience 127:637–647. Werhahn KJ, Kunesch E, Noachtar S, Benecke R, Classen J (1999) Differential effects on motorcortical inhibition induced by blockade of GABA uptake in humans. J Physiol (Lond) 517(Pt 2):591–597. Ziemann U, Hallett M, Cohen LG (1998) Mechanisms of deafferentation-induced plasticity in human motor cortex. J Neurosci 18:7000–7007. Ziemann U, Lo¨nnecker S, Steinhoff BJ, Paulus W (1996) The effect of lorazepam on the motor cortical excitability in man. Exp Brain Res 109:127–135. Zimerman M, Heise KF, Hoppe J, Cohen LG, Gerloff C, Hummel FC (2012) Modulation of training by single-session transcranial direct current stimulation to the intact motor cortex enhances motor skill acquisition of the paretic hand. Stroke 43:2185–2191.

(Accepted 24 July 2014) (Available online 1 August 2014)