Experimental studies emphasize the importance of homeostatic plasticity as a mean of stabilizing the properties of neural circuits. In the present work we combined two techniques able to produce short-term (5-Hz repetitive transcranial magnetic stimulation, rTMS) and long-term (transcranial direct current stimulation, tDCS) effects on corticospinal excitability to evaluate whether and how the effects of 5-Hz rTMS can be tuned by tDCS preconditioning. Twelve healthy subjects participated in the study. Brief trains of 5-Hz rTMS were applied to the primary motor cortex at an intensity of 120% of the resting motor threshold, with recording of the electromyograph traces evoked by each stimulus of the train from the contralateral abductor pollicis brevis muscle. This interventional protocol was preconditioned by 15 min of anodal or cathodal tDCS delivered at 1.5 mA intensity. Our results showed that motor-evoked potentials (MEPs) increased significantly in size during trains of 5-Hz rTMS in the absence of tDCS preconditioning. After facilitatory preconditioning with anodal tDCS, 5-Hz rTMS failed to produce progressive MEP facilitation. Conversely, when 5-Hz rTMS was preceded by inhibitory cathodal tDCS, MEP facilitation was not abolished. These findings may give insight into the mechanisms of homeostatic plasticity in the human cerebral cortex, suggesting also more suitable applications of tDCS in a clinical setting.
Transcranial direct current stimulation (tDCS) and repetitive transcranial magnetic stimulation (rTMS) are non-invasive techniques to induce changes in the activity of cortical neurons that outlast the end of stimulation (Hallett et al., 1999; Nitsche & Paulus, 2000; Siebner & Rothwell, 2003). Although the exact mechanisms mediating the after-effects are not fully clarified, there is evidence that changes in synaptic strength play a role (Liebetanz et al., 2002; Iyer et al., 2003). Several data suggest that rTMS-induced cortical plastic changes depend on the functional state of the motor cortex and may be shifted by tDCS preconditioning (Lang et al., 2004; Siebner et al., 2004; Ziemann & Siebner, 2008). Recent studies exploring the preconditioning effects of tDCS on the after-effect of rTMS on corticospinal excitability showed that the facilitatory effects of high-frequency rTMS can be reversed by ‘facilitatory preconditioning’ with anodal tDCS (Lang et al., 2004), as well as ‘inhibitory preconditioning’ with cathodal tDCS can reverse the inhibitory effects of low-frequency rTMS (Siebner et al., 2004). These findings have been interpreted in the context of homeostatic plasticity in the human motor cortex, and demonstrate that preconditioning sessions of tDCS can be used to shape the effects of subsequent rTMS on the primary motor cortex (Malenka, 2003; Turrigiano & Nelson, 2004; Cooke & Bliss, 2006; Hoogendam et al., 2010). It is known that long periods of tDCS (e.g. 10 min or more) can alter cortical excitability for up to 1 h, depending on the intensity of the current and the duration of the stimulation. Anodal tDCS applied to the motor cortex increases cortical excitability and activity, whilst cathodal tDCS results in the reverse effect (Nitsche & Paulus, 2001; Nitsche et al., 2003b). Pharmacological studies suggest that the cortical excitability shifts induced by tDCS depend on membrane polarization, thus, modulating the conductance of sodium and calcium channels. Moreover, they suggest that the after-effects involve modulation of N-methyl-d-aspartate (NMDA) receptors efficacy resembling those of long-term potentiation (LTP) and long-term depression (LTD; Nitsche et al., 2003a,b, 2004; Antal et al., 2006). In the present study we evaluated in normal subjects how changes in the state of human motor cortical excitability induced by tDCS may affect motor cortical response to brief trains of 5-Hz rTMS. There is evidence that short trains of suprathreshold 5-Hz rTMS produce a progressive facilitation of the motor-evoked potentials (MEPs; Berardelli et al., 1998; Inghilleri et al., 2005; Pascual-Leone et al., 1994), which is likely due to a calcium-dependent increase in presynaptic glutamate release (Fisher et al., 1997; Zucker & Regehr, 2002; Ziemann et al., 2008). MEP potentiation shortly outlasts the end of brief trains of high-frequency rTMS decaying along a time scale of seconds. LTP and LTD are unsuitable to explain this rapid and short-lasting MEP potentiation, thus candidates are presynaptic mechanisms of short-term synaptic enhancement (Berardelli et al., 1998; Zucker & Regehr, 2002). On this basis, we aimed to gain insight into the mechanisms by which pre- and postsynaptic mechanisms of synaptic plasticity could interact in the human motor cortex by combining high-frequency rTMS and tDCS.
Materials and methods
Twelve healthy volunteers (five males/seven females, mean age 27.1 ± 2.71 SD) participated in the study. All subjects were consistent right-handers (Oldfield, 1971). Four of the subjects have been exposed to TMS or tDCS in the past, while the other eight subjects were completely naïve to the experimental procedures. None of the subjects had a history of neurological disease or was on CNS-active drugs at the time of the experiments. Before participation, all subjects were checked for contraindications for TMS and tDCS (Keel et al., 2001), and gave their written informed consent for the study. The experimental procedures were approved by the Local Ethical Committee.
All subjects were comfortably seated in a chair and instructed to be as relaxed as possible. They wore a tight-fitting plastic swimmer’s cap to mark the optimum site of stimulation and coil placement. Electromyographic (EMG) signals were recorded from the right abductor pollicis brevis (APB) muscle using 0.9-cm diameter Ag–AgCl surface electrodes placed 3 cm apart over the belly and tendon of the muscle. The EMG activity was sampled at 5 kHz A/D rate with a band-pass between 10 and 1000 Hz, and a display gain ranging from 50 to 1000 μV/cm. EMG signals were collected, averaged and analysed off-line. Focal TMS was applied over the hand motor cortex of the left hemisphere by using a figure-of-eight coil connected to a monophasic Cadwell High Speed Magnetic Stimulator (Cadwell Laboratories, Kennewick, WA, USA). The stimulating coil was placed over the optimal site for eliciting responses in the contralateral target muscle with posteroanterior orientation (André-Obadia et al., 2008). The resting motor threshold (RMT) for eliciting responses in the relaxed APB muscle was defined as the minimum intensity of stimulation needed to produce responses of 50 μV in at least 50% of trials. Subjects were given audiovisual feedback of EMG activity to assist in maintaining complete relaxation. Constant coil position was continuously monitored during the experiment. Stimulation was performed following safety guidelines (Rossi et al., 2009).
Continuous tDCS was delivered through a pair of electrodes in a 5 × 7 cm water-soaked synthetic sponge using a battery-driven constant current stimulator (Magstim DC Stimulator). The first electrode was positioned over the motor hotspot of the right ABP muscle, as revealed by TMS. The second electrode was placed above the contralateral orbit. tDCS polarity refers to the electrode over the left primary motor cortex. Currents were given for 15 min with an intensity of 1.5 mA, and were ramped up or down over the first and last 8 s of stimulation. For the sham condition, the intensity was set to 1.5 mA as for anodal and cathodal conditions, but the DC stimulator was only switched on for 8 s at the beginning of the sham session and then turned off. This produced a short-lasting skin sensation comparable to real tDCS.
Experimental procedures and measurements
The main experiment was designed to explore the preconditioning effect of anodal or cathodal tDCS on the corticospinal response to brief trains of high-frequency and suprathreshold intensity rTMS. All subjects underwent four separated experimental sessions, performed at least 1 week apart from each other. The order of interventions was pseudorandomized and balanced across participants.
In two sessions we performed six rTMS trains of 10 stimuli at 5 Hz frequency to the left primary motor hand area (baseline conditions). The motor-cortical response to the rTMS trains was tested twice in distinct sessions to assess whether intra-individual variability in the rTMS response performed in baseline could occur (Maeda et al., 2000). In the two other sessions the six rTMS trains were performed immediately after 15-min tDCS preconditioning at 1.5 mA intensity in order to probe the after-effects of tDCS on motor cortical response to 5-Hz rTMS trains. The two sessions differed only for tDCS polarity (i.e. anodal vs. cathodal tDCS). In all sessions rTMS trains were applied in subjects at rest with a 2-min inter-train interval at an intensity of 120% of the RMT.
In order to evaluate MEP facilitation, for each subject MEP sizes were calculated peak-to-peak from single traces of the six trains and then averaged according to their position in the train.
Three groups of seven subjects within the 12 enrolled in the study were randomly selected to participate each in one of three control experiments. The first one was designed to test whether the observed variations in size of the MEPs during the trains performed after anodal and cathodal tDCS preconditioning could affect the motor-cortical response to the following stimuli of the trains. It consisted of two sessions in which the rTMS trains were delivered after the tDCS preconditioning (anodal and cathodal) at an intensity of the stimulator output adjusted to evoke MEPs of about 400–600 μV amplitude (i.e. the range of the averaged first MEP size in the rTMS trains performed at 120% RMT intensity in baseline). Thus, the intensity of the 5-Hz rTMS trains was reduced or increased, respectively, after anodal and cathodal tDCS preconditioning, so that the amplitude of the first MEP evoked by the trains matched the MEP amplitude at baseline. The second control experiment was performed to rule out possible unspecific effects of tDCS preconditioning. It consisted of one session in which six 5-Hz rTMS trains were given after sham tDCS. Finally, we investigated in the third control experiment the interaction effect between anodal tDCS and 5-Hz rTMS trains at a 30-min delay between the two protocols. The need to perform this control experiment arises from evidence that different physiological mechanisms could underlie the changes in motor-cortical response observed after (short-term effects) and at a delay (long-term effects) from the tDCS (Nitsche et al., 2003a). The different control experimental sessions of one subject were at least 1 week apart from each other. The order of sessions was counterbalanced across subjects.
To assess statistical significance in comparing changes in MEP amplitude during the 5-Hz rTMS trains (from first to 10th MEP) in the main experiment, a two-way repeated-measures analysis of variance (anova) was performed with ‘Condition’ (three levels: baseline, post-anodal tDCS, post-cathodal tDCS) and ‘Number of stimuli’ in the train (10 levels) as within-subjects factors. Separated two-way anova was performed for the three control experiments. For the first one anova was performed with ‘Condition’ (three levels: baseline, post-anodal and post-cathodal tDCS with adjusted MEP amplitude) and ‘Number of stimuli’ (10 levels) in the train. anova with baseline and post-sham tDCS as ‘Condition’ and ‘Number of stimuli’ (10 levels) in the train as within-subjects factors was performed for the second control experiment. Finally, we used anova with ‘Condition’ (three levels: baseline, post-anodal at 0 min delay, post-anodal at 30 min delay) and ‘Number of stimuli’ as within-subjects factors for the third control experiment. In each condition anova included the MEP values obtained by averaging MEP amplitudes of the six trains according to their position in the train (i.e. first, second, ..., 10th averaged MEP value). To compare the two baseline sessions we performed anova with ‘Condition’ (two levels: baseline 1 and 2) and ‘Number of stimuli’ (10 levels) as within-subjects factors. Because no differences were found (interaction between ‘Condition’ and ‘Number of stimuli’: F9,99 = 0.67, P = 0.73; ‘Condition’: F1,11 = 0.0013, P = 0.97; ‘Number of stimuli’: F9,99 = 12.7, P = 0.00000), MEP amplitudes obtained during the two baseline sessions were averaged and the mean values were used for the analyses (Fig. 1). The sphericity assumption was checked by using Mauchly’s test (Mauchly, 1940), and Huynh–Feldt’s correction (Huynh & Feldt, 1976) was adopted, if necessary, for the degrees of freedom. Duncan’s test was used for post hoc analysis. For all analyses the statistical significance was set at P-values lower than 0.05.
All enrolled subjects completed the planned cortical excitability measurements. The experimental procedures were well tolerated, and none of the participants experienced any adverse effect during or after tDCS or rTMS. Trains of 5-Hz rTMS preconditioned by anodal tDCS failed to determine the normal MEP facilitation when given both immediately after and at a 30-min delay from the end of tDCS preconditioning. Conversely, cathodal tDCS preconditioning did not abolish the increase in motor-cortical response during the trains of stimuli.
anova performed to evaluate changes in MEP size during the trains of stimuli performed in baseline condition and after anodal and cathodal tDCS preconditioning showed a significant effect for factor ‘Number of stimuli’ (F2,22 = 2.91, P = 0.00000), and a significant interaction between factors ‘Condition’ and ‘Number of stimuli’ (F18,198 = 5.32, P = 0.00000). Although a trend towards reduced mean MEP amplitudes was observed after both anodal and cathodal tDCS preconditioning, we did not find a significant effect for factor ‘Condition’ (F2,22 = 2.91, P = 0.07). According to post hoc analysis, MEP amplitudes increased significantly over the course of the trains from the fourth to the last response compared with the first one both in baseline condition (P < 0.01) and after cathodal tDCS preconditioning (P < 0.05). Oppositely, after anodal tDCS preconditioning MEP amplitudes reduced significantly in size from the second to the fourth response (P < 0.05) during the rTMS trains. As shown by post hoc analysis, the first MEP amplitude in the trains was significantly increased and reduced, respectively, for post-anodal (P < 0.01) and post-cathodal (P < 0.05) tDCS compared with the baseline condition (Fig. 2).
Changes in the MEP size during 5-Hz rTMS after anodal and cathodal tDCS preconditioning with adjusted MEP amplitude
Similarly to that observed in the main experiment, when anova was performed to evaluate changes in MEP amplitudes over the course of the trains given at adjusted intensity of stimulation, we observed a significant effect for factor ‘Number of stimuli’ (F9,54 = 7.81, P = 0.00000) and a significant interaction between factors ‘Condition’ and ‘Number of stimuli’ (F18,108 = 3.76, P = 0.00001). Post hoc analysis showed that MEP increased significantly in size from the third to the last response compared with the first one in the baseline condition (P < 0.05), and from the fourth to the last response of the trains after cathodal tDCS preconditioning (P < 0.02). After anodal tDCS preconditioning, post hoc analysis did not show significant variations in MEP amplitude over the course of the trains (Fig. 3).
Changes in the MEP size during 5-Hz rTMS after sham tDCS preconditioning
Sham tDCS preconditioning did not interfere with the rTMS-induced changes in MEP amplitudes observed in the baseline condition. anova showed that both factor ‘Condition’ and interaction between factors ‘Condition’ and ‘Number of stimuli’ were not significant, while a significant effect for factor ‘Number of stimuli’ was found (F9,54 = 9.87, P = 0.00000). Post hoc analysis showed that MEP amplitudes during the rTMS trains increased in a similar extent, from the sixth to the last response, both before and after the sham tDCS preconditioning (P < 0.05; Fig. 4).
Changes in the MEP size during 5-Hz rTMS given at 30 min after anodal tDCS preconditioning
In the group of subjects that underwent the third control experiment, we observed that MEPs remained unchanged in size throughout the course of the trains when the rTMS was given at 30 min after the end of anodal tDCS preconditioning. anova showed a significant effect for factor ‘Number of stimuli’ (F9,54 = 3.2409, P = 0.00325) and a significant interaction between factors ‘Condition’ and ‘Number of stimuli’ (F18,108 = 1.7208, P = 0.04624). According to post hoc analysis, the ability of anodal tDCS to abolish the normal facilitatory effect of 5-Hz rTMS was still evident at 30 min after the end of the preconditioning. Indeed, MEP amplitudes during the rTMS trains significantly increased from the fifth to the last response in baseline (P < 0.03), whilst after anodal tDCS preconditioning we did not observe significant variations in MEP amplitudes over the course of the trains performed at both 0 and 30 min delay. A trend toward reduced MEP amplitudes was observed during the trains given at 0 min delay. Post hoc analysis showed no significant differences in the first MEP amplitudes of the trains between the three conditions (baseline, post-anodal at 0 min delay, post-anodal at 30 min delay; Fig. 5).
In the present work we evaluated in healthy subjects the motor cortical response to brief trains of 5-Hz rTMS after modulating the level of cortical excitability by tDCS preconditioning. Our aim was to study how presynaptic mechanisms of glutamatergic neurotransmission could be affected by tDCS-induced LTP or LTD-like mechanisms. The rTMS protocol used in the study determines in normal subjects a progressive and short-lasting MEP facilitation likely due to an increase in presynaptic Ca2+ channels activity and glutamate release (Inghilleri et al., 2005, 2006; Conte et al., 2010; Brighina et al., 2011). This is in line with studies showing that short-term synaptic plasticity mainly depends on presynaptic mechanisms of glutamatergic neurotransmission (Berardelli et al., 1998; Di Lazzaro et al., 2002; Moulder et al., 2003, 2006; Richards, 2010). In addition, it should be noted that MEP facilitation likely takes place at cortical rather than spinal level, as suggested by two lines of evidence: (i) TMS tends to activate corticospinal neurons via cortico-cortical synapses (Rothwell et al., 1991, 1994; Berardelli et al., 1998); (ii) H-reflexes after trains of high-frequency rTMS are not enhanced (Di Lazzaro et al., 2002; Inghilleri et al., 2005, 2006). The critical new finding of the study refers to the capacity of anodal tDCS to turn into inhibition the ‘normal’ facilitatory effect of 5-Hz rTMS. Which molecular mechanisms are involved cannot be derived directly from our results. We speculate that anodal tDCS preconditioning could reduce the threshold for activation of inhibitory mechanisms of glutamate release, which depend on the modulation of presynaptic calcium channels activity (Borst & Sakmann, 1998; Catterall & Few, 2008; Catterall, 2010). This hypothesis agrees with the increasing evidence that glutamate synapses may be possible sites of homeostatic changes (Mochida et al., 2008; Cohen & Segal, 2009). Furthermore, it is in line with the study by Rango et al. (2008) showing that tDCS can affect metabolic pathways involved in neuronal excitability and synaptic neurotransmission. Particularly, the possible role of Ca2+-dependent mechanisms in the inhibitory response observed after anodal tDCS preconditioning during the rTMS trains can be supported by two lines of evidence: (i) chronic excitation leads to specific depression of presynaptic Ca2+ current through multiple subtypes of HVA Ca2+ channels (Moulder et al., 2004, 2006; Zhao et al., 2011); (ii) it has been shown that acute hypercalcaemia decreases MEP facilitation elicited by 5-Hz rTMS (Iacovelli et al., 2011). Alternatively, if we consider recent findings by other neurophysiological studies in humans suggesting the involvement of the postsynaptic L-type voltage-gated Ca2+ channels in short-term homeostatic plasticity (Wankerl et al., 2010; Fricke et al., 2011), the role of postsynaptic mechanisms could be taken into account. However, in our study the role of inhibitory mechanisms at a postsynaptic level seems unlikely given that anodal tDCS increases NMDA receptor activity and sensitivity of the postsynaptic neuron to glutamate. Moreover, the rapid occurrence of the decrease in MEP amplitude during the rTMS trains given in the baseline condition mainly resembles presynaptic mechanisms of short-term depression observed in animal experiments (Catterall & Few, 2008). Finally, an hypothetical role of γ-aminobutyric acid (GABA)ergic mechanisms can be hardly invoked to explain the results given the evidence showing that anodal tDCS reduces intracortical inhibition (Nitsche et al., 2005; Adams et al., 2010; Antal et al., 2010).
Considering all of the above, our results suggest the existence of interactions between mechanisms of short-term facilitation and long-term plasticity in the human motor cortex, as recently shown also by Iezzi et al. (2011). Therefore, our findings could be interpreted in the context of the homeostatic plasticity of the human motor cortex, which makes sure that activity-dependent synaptic plastic changes, required for modification of network properties, only occur within a physiologically useful range and allow for network stability (Turrigiano & Nelson, 2004). On that account, previous observations by Lang et al., (2004) and Nitsche et al. (2007) showed that protocols of non-invasive cortical stimulation known to increase cortical excitability can induce an opposite inhibitory effect when combined with another facilitatory protocol of stimulation. In these studies, however, the effects of two combined facilitatory protocols of stimulation were evaluated by single-pulse TMS after the end of stimulation. Instead, in the present work the after-effects of tDCS were studied during the following 5-Hz rTMS trains, thus allowing us to study mainly the presynaptic mechanisms of glutamate release.
Concerning the after-effects of cathodal tDCS on motor-cortical response to the rTMS trains, we observed a reduction in the first MEP size, in line with the assumption that cathodal tDCS reduces intracortical activity by inducing LTD-like mechanisms (Nitsche & Paulus, 2000; Liebetanz et al., 2002). MEP following the first, conversely, showed a progressive increase in amplitude. This suggests that inhibitory cathodal tDCS does not interfere with the facilitatory mechanisms of short-term plasticity elicited by high-frequency rTMS trains. According to the rules of homeostatic plasticity, one would expect that after inducing motor cortical inhibition by cathodal tDCS, MEP facilitation during the trains could be increased with respect to the baseline. Notwithstanding, the reduction in the conductance of sodium and calcium channels induced by cathodal tDCS, which in turn can be responsible for the reduction in glutamate release during the trains of 5-Hz rTMS, could explain the present findings (Adams et al., 2010).
In conclusion, the current results indicate that we can achieve important information about mechanisms of cortical homeostatic plasticity by combining different neurophysiological techniques such as rTMS and tDCS. This stimulation approach could be helpful in interpreting some conflicting results from studies in pathological conditions in which an impairment of homeostatic plasticity has been supposed (Brighina et al., 2011). Moreover, it could suggest possible and more suitable applications of neurophysiological techniques as a treatment for these brain disorders (Fritschy, 2008; Azdad et al., 2009; Dickman & Davis, 2009; Kang et al., 2011). Indeed, consecutive sessions of rTMS or tDCS could represent a safe therapeutic tool due to their ability to affect cerebral cortex excitability and mechanisms of homeostatic plasticity.