Corresponding author K. Funke: Department of Neurophysiology, Medical Faculty, Ruhr-University Bochum, Universitaetstrasse 150, 44780 Bochum, Germany. Email: firstname.lastname@example.org
Repetitive transcranial magnetic stimulation (rTMS) has been shown to alter cortical excitability that lasts beyond the duration of rTMS application itself. High-frequency rTMS leads primarily to facilitation, whereas low-frequency rTMS leads to inhibition of the treated cortex. However, the contribution of rTMS train duration is less clear. In this study, we investigated the effects of nine different rTMS protocols, including low and high frequencies, as well as short and long applications (1, 3 and 10 Hz applied for 1, 5 and 20 min), on visual cortex excitability in anaesthetized and paralysed cats by means of visual evoked potential (VEP) and electroencephalography (EEG) recordings. Our results show that 10 Hz rTMS applied for 1 and 5 min significantly enhanced early VEP amplitudes, while 1 and 3 Hz rTMS applied for 5 and 20 min significantly reduced them. No significant changes were found after 1 and 3 Hz rTMS applied for only 1 min, and 10 Hz rTMS applied for 20 min. EEG activity was only transiently (<20 s) affected, with increased delta activity after 1 and 3 Hz rTMS applied for 1 or 5 min. These findings indicate that the effects of rTMS on cortical excitability depend on the combination of stimulus frequency and duration (or total number of stimuli): short high-frequency trains seem to be more effective than longer trains, and low-frequency rTMS requires longer applications. Changes in the spectral composition of the EEG were not correlated to changes in VEP size.
Repetitive transcranial magnetic stimulation (rTMS) is a noninvasive and largely painless technique to stimulate the human brain. It is preferentially applied to cortical areas with the intention to either disturb ongoing processing in the targeted area (virtual lesion) or to modulate the general excitability of the cortical network. Besides transient effects, numerous studies have shown that rTMS has modulatory effects on cortical excitability that last for seconds to many minutes beyond the duration of the rTMS application itself. The primary motor cortex has been used extensively for rTMS studies because the effects of stimulation are easy to quantify by measuring the size of motor-evoked potentials (MEPs). Findings to date suggest that the modulatory effects of rTMS on cortical excitability may be inhibitory or facilitatory depending on the frequency, intensity and duration of the stimulus (see Modugno et al. 2001). A few examples: high-frequency rTMS, especially at high stimulus intensities, leads to facilitation of corticospinal activity; a 10-pulse rTMS train of 20 Hz applied at a strength of 150% resting motor threshold (RMT) caused an increase in MEP size lasting for about 3 min (Pascual-Leone et al. 1994); a 30-pulse rTMS train at 120% RMT and 15 Hz caused a shorter and smaller increase in MEP size for only 90 s (Wu et al. 2000). Stimulation intensities below RMT usually require longer trains before any lasting effect is seen. Maeda et al. (2000a,b) reported a facilitation of MEPs for 2 min after administration of 240 pulses of 20 Hz stimuli in combination with 90% RMT. Notably 10 Hz rTMS had no lasting effect on MEP size at the same strength (Maeda et al. 2000a,b).
Low-frequency rTMS usually results in suppression of corticospinal excitability. A 15 min train of 0.9 Hz applied at 115% RMT over the primary motor cortex reduced corticospinal excitability (i.e. increased the RMT and reduction of the MEP input–output curve) for at least 15 min after the end of the stimulation (Chen et al. 1997a; Muellbacher et al. 2000). Low-frequency rTMS at intensities below RMT have a much weaker effect on corticospinal excitability than supra-threshold rTMS. Fitzgerald et al. (2002) applied 15 min of 1 Hz rTMS at 85 and 115% of RMT. Both intensities produced an increase in the RMT, but only 115% stimulation reduced the size of MEP. A 240-pulse train of 1 Hz rTMS at 90% RMT reduced the MEP amplitude for 2 min (Maeda et al. 2000a). Even lower intensities, e.g. 90% active motor threshold or lower frequencies such as 0.1 Hz, had no lasting effect (Chen et al. 1997a; Gerschlager et al. 2001). However, not only strength but also duration of the rTMS application affects the duration of the after-effects. Touge et al. (2001) used 150 and 1500 pulses of 1 Hz rTMS at 95% RMT; they observed that MEP amplitudes were reduced for 2–10 min after the end of the train, with longer suppression after 1500 rTMS pulses than after 150 pulses.
Although the effects of rTMS were primarily studied in human primary motor cortex, preliminary work on occipital visual cortex suggests that the effects may be similar. Several studies have shown that 15 min of 1 Hz rTMS applied to the occipital cortex leads to increased PT (phosphene threshold, minimum stimulus intensity to induce artificial light sensations) lasting at least 10 min in control subjects. This effect was interpreted as a result of decreased cortical excitability (Boroojerdi et al. 2000; Brighina et al. 2002). So far, only a few studies have demonstrated increased visual cortex excitability after 10 Hz rTMS application to the occipital cortex (Bohotin et al. 2002; Fierro et al. 2005; Fumal et al. 2006). Using PT as an indicator for cortical excitability, Fierro et al. (2005) studied the effect of 15 min of high-frequency (10 Hz) rTMS in healthy subjects who were either in a light-adapted or in a light-deprived state. Those authors showed that 10 Hz rTMS significantly lowers PT during the light-adapted state, but does the opposite during light deprivation which had already led to a decreased PT. This elegant experiment demonstrates that the outcome of rTMS critically depends also on the actual excitability state of the cortical network. Differences in the effect of low- (1 Hz) and high-frequency (10 Hz) rTMS on human visual cortex excitability, measured with the aid of visual evoked potentials (VEPs), were also found when comparing migraineurs with healthy volunteers (Bohotin et al. 2002; Fumal et al. 2006; see also Fumal et al. 2003). While early VEP components were largely unaffected in healthy subjects following 10 Hz rTMS (900 pulses), a facilitatory effect was found in migraine patients, supposedly due to an altered cortical excitability during the interictal phases. On the other hand, 900 magnetic pulses repetitively applied at low frequency (1 Hz) reduced the amplitude of the early VEP components in both healthy subjects and migraineurs. Although the relationship between rTMS and cortical excitability has been studied quite extensively, the effect of train duration on the outcome of rTMS applied at different frequencies has not been investigated systematically, at least not for the visual cortex. Therefore, the major intention of our study was the establishment of an animal model for the investigation of rTMS effects on cortical activity. We have chosen cat visual cortex because it is one of the best-investigated cortical structures with regard to information processing and different forms of plasticity. The cat has a well-developed visual system, largely comparable to the organization of the human visual system, and, as in humans, cortical excitability can be easily accessed by electroencephalography (EEG) and VEP recordings. For a fundamental data basis of rTMS effects, we tested nine different stimulation protocols covering low and high frequencies as well as short- and long-lasting applications (1, 3 and 10 Hz applied for 1, 5 and 20 min each) in the anaesthetized and paralysed cat. EEG was recorded from two different cortical sites, from the stimulated area of the primary visual cortex (A17/18) and from the occipitotemporal area of the contralateral hemisphere to test for local or global changes in cortical activity related to rTMS.
A total of seven adult cats of both sexes (body weight 3.5–5 kg) were used in this study. To avoid interference of effects due to different TMS repetition frequencies, the individual experimental animal was treated with only one of the three stimulation frequencies (1 or 3 Hz, two cats each; or 10 Hz, three cats) but trains of different duration (1, 5 and 20 min) were tested in random order in the same animal. Intervals between rTMS applications were at least 30 min in case of 1 min trains, and more than 90 min after a 20 min train (more details follow). Trials from all cats treated with a particular rTMS frequency were handled as independent entities for the statistical computations.
Surgical procedures were in principle the same as in the previous studies (Moliadze et al. 2003, 2005). Briefly, cats were anaesthetized with a combination of ketamine (20 mg kg−1i.m.; Ketanest; Parke-Davies, Germany) and xylazine (2 mg kg−1i.m.; Rompun; Bayer, Germany) and fixed using standard stereotaxic methods. All incisions and pressure points were also locally anaesthetized by xylocaine (2%; Astra Chemicals, Germany). Craniotomies were performed over area 17 of the right hemisphere and the temporoparietal area of the left hemisphere, providing access for combined local VEP/EEG recording and global EEG recording, respectively. To enable artificial ventilation of the lungs, relaxation was achieved with alcuronium chloride (0.15 mg kg−1 h−1 Alloferin 10; Hoffmann La Roche, Germany) in 3% glucose-Ringer solution, infused via the femoral artery. Continuous anaesthesia during recording sessions was achieved by artificial respiration with N2O/O2 (70/30%) and halothane (0.6–2.5% Fluothane; ICI-Pharma, Germany). Rectal body temperature was measured and kept at about 38.5°C by the aid of a controlled heat blanket. Blood pressure, heart rate, body temperature, respiratory parameters and EEG were continuously monitored to determine the physiological state and the efficiency of anaesthesia. The optics were corrected for a viewing distance of 56 cm with contact lenses of 5–7 D (dioptre). Atropine sulphate (1% Atropin-Pos; Ursapharm, Germany) and phenylephrine hydrochloride (5%, Neosynephrin-Pos; Ursapharm) were applied topically for mydriasis and retraction of the nictitating membranes. Isoptomax (Alcon Pharma, Germany) was topically applied to the cornea to prevent infections. Recordings were continued as long as the animal could be maintained in a physiological state. At the end of each experiment, the animal was deeply anaesthetized by maximal halothane concentration (4 vol%) in combination with N2O and perfused with cold (4°C) Ringer solution followed by 4% paraformaldehyde to enable further histological studies on the brain tissue. All experimental procedures were permitted under the local government guidelines for animal welfare (Az. 50.8735.1 no. 81.6 and 105/7) also conforming to specific legal requirements in the EU, UK and US.
Recordings and visual stimulation
VEPs and EEGs were recorded with chlorided (Ag/AgCl) silver ball electrodes, which were placed above area 17 of the right hemisphere and the temporoparietal cortex of the left hemisphere on the surface of the dura mater, and stabilized by dental cement. The recordings were amplified with a Cyberamp 380 (Axon Instruments, CA, USA) (gain 2000–10000, band-pass filtered between 1 and 100 Hz) and averaged with Spike 2 analysis software (Cambridge Electronic Design, UK). EEG recordings were done from two different sites, one was in the centre of TMS in the primary visual area of the right hemisphere (stimulated area, local EEG) and the other was in the occipito-temporo-parietal area of left hemisphere (at the junction of areas 7, 21 and 22, close to suprasylvian sulcus, global EEG). The cortical area at the latter position was supposed to be not sufficiently stimulated by rTMS since it lies at the border of the peak of the magnetic field, is about 4 mm more distant to the coil and the cortical surface is out of plane of the induced electric field. Therefore, we term it the ‘nonstimulated area’ throughout the text. The analog EEG signal (amplification ×2000, band-pass filtered between 0.1 and 100 Hz) was sampled at a rate of 250 Hz, and periods corresponding to sequences of VEPs were digitally stored on hard disc.
We recorded VEPs and EEGs for episodes of 5 min duration before and after each rTMS application. Visual stimulation was achieved with high contrast (98%, [lmax−lmin]/[lmax+lmin]lmax, lmin-maximal and minimal luminance levels, mean brightness 20 cd m−2) checkerboard pattern of black and white squares that was generated by the PC-based visual stimulator Leonardo (Lohmann Research Equipment, LNC, Germany) and displayed on a 21 in (52 cm) colour monitor (liyama Vision Master 500) at a contrast-reversal rate of 1 Hz and a monitor refresh rate of 100 Hz.
Repetitive magnetic pulses were generated by a MagStim rapid equipped with two boosters (The MagStim Company, Whitland, Dyfed, UK) and applied to the occipital cortex of cats via a figure-of-eight coil (2 × 70 mm in one plane; The MagStim Company). The midpoint of the coil was centred above area 17 of the right hemisphere. Therefore, the coil was fixed at 8–10 mm above the cortical surface, with the handle of the coil pointing to the left. This way, the induced current – lateral-to-medial in the right hemisphere – would be more restricted to the occipital pole of the brain than in case of anterior–posterior orientation. A stimulus of 30% of maximal output strength was used in this study because in previous experiments performed in our laboratory, 30% of TMS strength was found to be just supra-threshold for modulating spontaneous and visually evoked activity in cat visual cortex (Moliadze et al. 2003; and see Discussion). Nine different stimulation protocols were used to test rTMS effects on visual cortex excitability, resulting from the combination of three different stimulation frequencies (1, 3 and 10 Hz) with three different train duration (1, 5 and 20 min). Further details of stimulus protocols are given in Table 1. The minimum interval between applications was chosen with regard to the duration of outlasting rTMS effects. Usually, VEP amplitudes recovered to preTMS levels within 10 min after 1 min rTMS, and within 15 and 30 min after 5 and 20 min applications, respectively (with regard to the rTMS frequency causing the longest after-effects). This was tested in three pilot cat experiments with mixed combinations of rTMS frequency and duration, while also testing a higher TMS intensity of 40% (these data were not included in the statistical analysis). Following VEP recovery measurements, a pause of at least twice the time of VEP recovery with neither TMS nor visual stimulation (only background illumination level) was interleaved between the rTMS applications to allow for recovery of the visual system. Thus, intervals between rTMS were at least 30 min in case of 1 min applications, 45 min in case of 5 min rTMS and around 90 min after 20 min of rTMS, with VEP recordings stopped 10, 15 and 30 min after 1, 5 and 20 min rTMS, respectively. This way, we tried to apply as much as possible rTMS trains at sufficiently long recovery intervals.
Table 1. Nine different rTMS protocols
1 min, 1 Hz
5 min, 1 Hz
20 min, 1 Hz
1 min, 3 Hz
5 min, 3 Hz
20 min, 3 Hz
10 trains of 40 s at intervals of 2 min
1 min, 10 Hz
5 min, 10 Hz
20 min, 10 Hz
12 trains of 1 s
60 trains of 1 s
10 blocks of 12 trains of 1 s
repeated every 5 s
repeated every 5 s
repeated every 5 s, blocks repeated every 2 min
Blocks of 300 VEP responses (5 min) were sampled before and – depending on the TMS protocol – up to 30 min after rTMS. VEPs were analysed off-line in terms of peak latencies and peak-to-peak amplitudes of the N1P1 and P1N2 components.
Power spectra of EEG recordings from the stimulated area (site of VEP recording, centre of TMS) and from the nonstimulated area (occipito-temporo-parietal cortex of the other hemisphere) were analysed separately. Five frequency bands were analysed with regard to spectral power: delta band (0.5–3 Hz), theta band (4–7 Hz) alpha band (8–13 Hz), beta band (14–30 Hz) and gamma band (30–100 Hz). EEGs were analysed in three different time windows before and after rTMS; for the 5 min blocks corresponding to the VEP analysis and on a finer time scale for a couple of 20 or 5 s time windows directly following rTMS.
Means ±s.e.m. were calculated for the N1P1 and P1N2 amplitudes (μV) and peak latencies (ms) of VEPs, and for the spectral power of the EEG (V2). For each rTMS frequency (1, 3 and 10 Hz), we computed two-way repeated measures ANOVA with ‘time’ (before and after rTMS) and ‘train duration’ (1, 5 and 20 min) as within-subject factors for the N1P1 and P1N2 VEP components, and the global and local EEG (with delta, theta, alpha, beta and gamma bands analysed separately). Post hoc comparisons of different time points were performed using Bonferroni multiple comparison tests. A P value of <0.05 was considered significant.
VEP components were clearly identified and reproducible. The VEPs consisted of three components, an initial negative wave (N1), a positive wave (P1) and a late negative wave (N2). VEP waveforms recorded in our laboratory were in accordance with the literature (Uzuka et al. 1989; Arakawa et al. 1993; Perez-Cobo et al. 1994; Padnick & Linsenmeier, 1999). However, there might be some differences in latencies, amplitudes and polarity of the components due to a wide variety of stimulation and recording systems, electrodes and type of anaesthesia used in these studies. In our study, we defined the N1 peak as the most negative point between 60 and 90 ms after the stimulus onset, P1 as the most positive point following N1 between 90 and 130 ms, and N2 as the most negative point following P1 between 140 and 170 ms (see Fig. 1A). As the baseline may vary throughout the experiment we measured peak-to-peak amplitudes N1P1 and P1N2. No significant latency differences were detected after rTMS in any case. Power spectra of EEGs (from stimulated area, ‘local’ and nonstimulated area, ‘global’) were analysed according to their frequency range (delta, theta, alpha, beta and gamma). The VEP results can be seen as a summary in Table 2.
Table 2. Summary of the results on visually evoked potential (VEP) amplitudes
No. of pulses
Time during which the effect was seen
VEP(1), first amplitude (N1P1); VEP(2), second amplitude (P1N2).
Ø, no effect; ↑, increase; ↓, decrease.
1 Hz rTMS protocol
Repeated measures ANOVA with time (before and after rTMS) and train duration (1, 5 and 20 min) as within-subject factors showed a significant main effect for train duration (F2,22= 20.20, P < 0.0001), and a significant interaction between both factors (F4,44= 2.97, P= 0.030) for the N1P1 amplitude, and a significant main effect for train duration (F2,22= 53.54, P < 0.0001) for the P1N2 amplitude, indicating that train duration is the most critical factor. Post hoc analysis (Bonferroni) revealed that 20 min of 1 Hz rTMS had significantly reduced the N1P1 amplitude (P= 0.005) between 0 and 5 min following rTMS (Fig. 1). The reduction of the P1N2 component did not reach statistical significance, as did the reduction of both components visible after 5 min rTMS. The 1 min application had no effect at all. One hertz rTMS did not affect the spectral power of the EEG, calculated from the same 5 min intervals as taken for the VEP analysis; however, within the first 5 s (P= 0.009 for 1 min; P=0.027 for 5 min 1 Hz) and 20 s (P= 0.004 for 1 min; P=0.029 for 5 min 1 Hz) following 1 and 5 min 1 Hz rTMS, local delta power strongly increased. Two-way ANOVA yielded significant train duration (F2,22= 8.43, P= 0.002 (5 s); F2,22= 5.21, P= 0.014 (20 s)), time (F2,22= 51.84, P < 0.0001 (5 s); F2,22= 51.51, P < 0.0001 (20 s)) and interaction (F4,44= 9.63, P < 0.0001 (5 s); F4,44= 8.03, P < 0.0001 (20 s)) main effects within the first 5 and 20 s window in local delta power after 1Hz rTMS protocol, respectively.
3 Hz rTMS protocol
Two-way ANOVA revealed a significant time (F1.21,10.85= 6.86, P= 0.020 (N1P1); F2,18= 15.86, P < 0.0001 (P1N2)) and train duration (F2,18= 53.71, P < 0.0001 (N1P1); F2,18= 31.84, P < 0.0001 (P1N2)) main effect for N1P1 and P1N2 amplitudes, respectively. Bonferroni comparison test showed that 5 min 3 Hz rTMS significantly decreased both VEP amplitudes between 0 and 5 min (P= 0.017 (N1P1); P= 0.001 (P1N2)) and between 5 and 10 min (P= 0.016 (N1P1); P= 0.005 (P1N2)) after rTMS application (Fig. 2). In addition, both the N1P1 (P= 0.019 and P= 0.045) and the P1N2 (P= 0.027 and P= 0.048) amplitudes were found to be decreased between 5 and 10 min and 10–15 min following 20 min 3 Hz rTMS. A significant effect of rTMS on the spectral composition of the EEG, calculated from the same 5 min interval as taken for the VEP analysis, was found only in case of the 1 min 3 Hz rTMS, resulting in an increase in local EEG delta power (P= 0.044) with a significant train duration main effect (F2,14= 24.62, P < 0.0001) and significant interaction between time and duration (F4,28= 3.296, P= 0.025, two-way ANOVA). One minute 3 Hz rTMS also significantly increased local delta power within first 5 s (P= 0.0001) and 20 s (P= 0.001). ANOVA showed significant main effects for train duration (F2,18= 18.52, P < 0.0001 (5 s); F2,18= 16.73, P < 0.0001 (20 s)), time (F2,18= 20.33, P < 0.0001 (5 s); F2,18= 14.25, P < 0.0001 (20 s)) and interaction (F4,36= 16.86, P < 0.0001 (5 s); F4,36= 11.81, P < 0.0001 (20 s)) in case of the local delta power. The increase in local delta power during the first 5 s following rTMS was about 30 times higher than that for the whole 5 min episode and about three times higher when compared with the 20 s time window, indicating that the strongest change in delta power occurred transiently directly following the rTMS application.
10 Hz rTMS protocol
Two-way repeated measures ANOVA yielded a significant train duration main effect for both N1P1 (F2,20= 123.39, P < 0.0001) and P1N2 (F1.17,7.02= 104.95, P < 0.0001) components following 10 Hz rTMS. Bonferroni test showed that the P1N2 amplitude was significantly increased during the first 5 min following a 1 min (P= 0013) or 5 min (P= 0033) application (Fig. 3). A 20 min application of 10 Hz rTMS had no effect on VEP amplitudes. None of the 10 Hz rTMS protocols affected the local or global EEG, even during the first 5 or 20 s following rTMS (Fig. 4).
To our knowledge this is the first experimental study to reveal the effects of nine different combinations of rTMS frequency and duration on visual cortex excitability in the anaesthetized and paralysed cat. The present results show that rTMS applied over the visual cortex can alter the excitability of the visual cortex for several minutes, even in the anaesthetized cat, in a manner depending on both frequency and duration of the stimulus train. Our results are largely in accordance with previous findings in human motor and visual cortex. Differences regarding distinct stimulation protocols may relate to the temporal pattern of stimulation (especially in case of 10 Hz, see below), or TMS strength.
In the cat experiments included in this study, we used a stimulus strength of 30% of maximal stimulator output, while Fumal et al. (2003) stimulated the human visual cortex at PT (67 ± 12%) or at 110% of relaxed motor threshold (58 ± 8%) which are both higher than the absolute stimulus intensity we used. Due to anaesthesia and blockade of neuromuscular transmission, we could not reliably test motor threshold in the cat. Actually, motor threshold may not be a good measure for threshold strength in the visual cortex (Gerwig et al. 2003). However, we can assume that the TMS strength used by us is comparable to PT (or maybe motor threshold). In previous experiments, we found that in case of single TMS pulses, a strength of at least 30% was needed to modulate spontaneous and visually evoked activity of single units in the visual cortex of anaesthetized cats (Moliadze et al. 2003). Usually, higher intensities (>40%) had to be used to evoke clear spiking activity in most neurons without additional visual stimulation. When comparing our findings with those obtained in human visual cortex, we can assume that moderate effects elicited with 30% stimulus strength (relatively few neurons excited) may be comparable with the threshold for eliciting weak phosphene sensations, while stronger stimuli evoke increased population background activity resulting in elevation of contrast threshold (Kammer & Nusseck, 1998; Kastner et al. 1998; Kammer, 1999). In addition, in a few pilot experiments, we also tested 40% TMS strength but found no qualitative difference to 30% strength. To keep the rTMS strength constant throughout the series of experiments, and to be in the safe range of rTMS treatment both for low and high stimulus frequencies (Chen et al. 1997b), we used the peri-threshold stimulus strength of 30%.
Effects on VEPs
At first, our results obtained in the anaesthetized cat are principally in accordance with those of Fumal et al. (2003), who demonstrated that application of low-frequency rTMS (1 Hz) led to a decrease of human VEP amplitudes, but 10 Hz rTMS applied for 15 min did not alter VEP amplitudes significantly. However, they showed that reduction of the N1P1 amplitude after 1 Hz rTMS persisted for 23 min, and for the P1N2 amplitude the effect lasted only for 3 min. In our study, we observed that only the N1P1 amplitude significantly decreased during the first 10 min following a 20 min train of 1 Hz rTMS and that the P1N2 amplitude did not change significantly. As already mentioned, this could be due to the fact that besides frequency, intensity influences the effect of electromagnetic stimulation on the underlying cortex, but also the different anatomical locations of the neural generators for the VEP components may play a role. The N1 component of human VEP is generated in the primary visual area whereas P1 is generated in extrastriate cortex which was also stimulated in the cat (A18, 21 and part of A19) and N2 is localized to a deep source in the parietal lobe (Di Russo et al. 2001). If this holds also for the cat, it could explain why the deeper generators were less affected and no significant effect on P1N2 was obtained. In this respect, our results are in good accordance with those of Schutter & Honk (2003) who demonstrated a reduction of amplitude of the very early CI component of human VEP after 1 Hz rTMS while the latency of this component was unchanged. The CI component is supposed to be generated in V1. Another reason for the relatively short-lasting effect of rTMS on the VEPs could be also a de-potentiation effect of the checkerboard flicker at 1 Hz as recently described by Teyler et al. (2005).
Furthermore, we provide novel insights by demonstrating that 5 and 20 min of 3 Hz rTMS are more effective than 5 or 20 min of 1 Hz rTMS in depressing visual cortex excitability in the anaesthetized and paralysed cat. In the literature, there is no report about the effect of 3 Hz rTMS on the visual cortex, and the motor cortex has been rarely studied with this frequency (Jennum et al. 1995; Berardelli et al. 1999; Romeo et al. 2000; Arai et al. 2005). While it has been demonstrated that 1 Hz rTMS has outlasting inhibitory effects (Pascual-Leone et al. 1994; Chen et al. 1997a; Maeda et al. 2000a; Muellbacher et al. 2000; Touge et al. 2001), and 5 Hz rTMS can have long-lasting facilitatory effects on motor (Berardelli et al. 1998; Siebner et al. 2000; Peinemann et al. 2004; but see Jennum et al. 1995 for partly inhibitory effects) and somatosensory cortex (Ragert et al. 2004), to our knowledge, 3 Hz rTMS has not been reported to have a significant outlasting effect on motor activity (mainly short-lasting facilitatory effects have been reported, Arai et al. 2005). We decided to test 3 Hz in addition to 1 and 10 Hz for the following reasons. (1) Three hertz is just in the middle between the suppressive frequency of 1 Hz and the facilitatory frequency of 5 Hz. Thus, if there is some transition between these two frequency domains, one might either expect a mixed effect or an absence of effects because suppressive and facilitatory effects cancel each other. (2) Although 3 Hz is in the middle between 1 and 5 Hz, it still belongs to the EEG delta band – as does 1 Hz – while 5 Hz belongs to the theta band. Thus, if the effect of repetitive TMS is mediated in any way via the reinforcement of distinct EEG rhythms, one might expect the 3 Hz rTMS having a similar depressive effect than a 1 Hz rTMS. This actually happened, both with regard to VEP amplitudes and for changes in EEG activity as will be discussed below. Thus, with regard to other studies demonstrating some facilitatory effects of 3 Hz rTMS during the course of stimulation, it is not absolutely clear whether 3 Hz rTMS can be defined as a high-frequency facilitative stimulation, or as a low-frequency suppressive procedure.
In case of 10 Hz rTMS, we observed a significant but short-lasting enhancement in VEP amplitudes only after short applications (1 min = 120 pulses, and 5 min = 600 pulses) but not after 20 min applications (1200 pulses). For this observation, there is no real counterpart in human studies. Fumal et al. (2003) found no signs of enhanced cortical activity after 10 Hz rTMS applied to human visual cortex. One reason could be the long-lasting application (15 min, 900 pulses). Enhancement motor responses after 10 Hz rTMS applied to human motor cortex can not be consistently evoked and show a high interindividual variability (Maeda et al. 2000a,b). For 5 Hz rTMS however, a facilitation of motor-evoked responses has been described also after long-lasting applications (1800 stimuli), while short trains (150 stimuli, Peinemann et al. 2004) had no effect. The effect of high-frequency rTMS may thus depend to some degree on the stimulus frequency but also on the kind of stimulus protocol applied, being dependent on stimulus duration and pattern. For example, in a recent in vitro study, Rosanova & Ulrich (2005) showed that synaptic LTP can be induced in cortical neurons by applying an associative pre- and postsynaptic 10 Hz protocol of sleep spindle type variation in time. Interestingly, they found that in contrast to any other kind of 10 Hz stimulation (regular, irregular, reversed) only the spindle pattern was effective and that a short train of 30 spindles (a bit less than a minute) was more effective than shorter or longer trains. The effect of 10 Hz rTMS on VEP amplitudes was stronger for the P1N2 than the N1P1 component indicating that a stimulus pattern similar to alpha oscillations possibly affects striate and extrastriate cortical areas differently.
The effect of stimulus train duration
Our study further demonstrates that the efficiency of rTMS depends on the relationship between stimulus frequency and duration or the number of stimuli applied. To be effective, rTMS applied at the rather low frequencies of 1 and 3 Hz had to be applied for at least 5 min, while the high-frequency rTMS (10 Hz) showed effects only for short applications of 1 or 5 min. This is largely in accordance with findings obtained in human motor cortex. For example, very short trains of low frequency rTMS (10–20 pulses) have no outlasting effects on the size of MEPs, but acute effects during the train: except for 1 Hz, trains of 2–15 Hz were found to gradually increase the silent period following a MEP (Berardelli et al. 1999; Romeo et al. 2000). The effect on MEP size itself is controversial, some report no effect at all (Berardelli et al. 1999; Romeo et al. 2000), while others show a gradual facilitation with 2 and 3 Hz (Arai et al. 2005), or no significant effects at frequencies of 2–3 Hz, but mixed inhibitory and facilitatory effects with higher frequencies (>5 Hz) (Pascual-Leone et al. 1994; Jennum et al. 1995). Maeda et al. (2000b) applied relatively short trains (240 pulses) of 1, 10, 15 and 20 Hz rTMS and observed a significant effect (facilitation) only for the high-frequency stimulation with 20 Hz rTMS. With a higher number of stimuli (1600), also 1 and 10 Hz stimulation yielded significant effects (inhibition versus facilitation, respectively), and also long trains (1800 stimuli) of 5 Hz rTMS lead to consistent facilitation of MEPs (Peinemann et al. 2004). Similarly, 1 Hz rTMS applied to prefrontal cortex for 15 min (900 pulses) had an inhibitory effect on auditory-evoked potentials in humans, while only 10 min (600 pulses) were inefficient (Hansenne et al. 2004). With a small number of stimuli we also did not observe any change at 1 Hz (60 pulses, 1 min) and 3 Hz rTMS (180 pulses, 1 min), but a reduction of VEP amplitudes was evident with a larger number of pulses (300 pulses, 1 Hz, 5 min; 900 pulses, 3 Hz, 5 min).
Effects on EEG
We analysed the spectral composition of the EEG, both at the site of VEP recording (local) and at a distant site in the opposite hemisphere (global), to test whether rTMS generally affects the state of the brain, and whether effects on VEPs might be related to changes in the spectral composition of the EEG. Except for an increase in delta power after 1 min 3 Hz rTMS, we did not find any significant effect of rTMS on the five different frequency bands, delta, theta, alpha, beta and gamma, when analysing the complete 5 min of VEP recording. On a first view, this finding seems to be in contrast to previous studies which demonstrated enhanced alpha synchronization either after 10 Hz rTMS (frontal and parietal cortex, Jing & Takigawa, 2000; motor cortex, Strens et al. 2002) or after 1 Hz rTMS (visual cortex, reduction of visually induced desynchronization, Thut et al. 2003). However, effects of rTMS on the spectral composition of the EEG may be also short-lasting and therefore not detectable if EEG stretches of 5 min are analysed. For example, Okamura et al. (2001) have shown that 10 Hz rTMS administered for 3 s to the left prefrontal area increased peak frequency of alpha activity for only 2 min after TMS. Interestingly, a high-frequency (9 Hz) photic stimulation for 120 s decreased human occipital alpha activity for about 1 h (Clapp et al. 2006; see also below). Even shorter effects, lasting for only a few seconds, were found by Rivadulla et al. (2004) when applying rTMS to paralysed and anaesthetized cats. rTMS applied at a low repetition frequency (0.5–1 Hz) lowered the mean EEG frequency for about 5 s, while high frequency rTMS (15 Hz) enhanced the high frequencies in the power spectrum of EEG. Taking this into consideration, we analysed the EEG also in shorter time windows of 5 and 20 s after termination of rTMS. None of the nine rTMS protocols had any significant effect on the theta, alpha, beta and gamma frequency band, but local delta activity was transiently (<20 s) increased after 1 and 3 Hz rTMS applied for 1 or 5 min. A similar but much smaller and not significant effect on the global EEG was visible after 1 min of 1 and 3 Hz rTMS. The transient change in delta power was more than 10 times stronger than the change observed during the 5 min analysis (300 s, 15 times longer than 20 s window), indicating that the weak changes found with 5 min are almost completely related to the transient change immediately after rTMS and that no longer lasting EEG changes occur.
Relation of effects on VEPs and EEG
Interestingly, rTMS applied at 1 and 3 Hz had differential effects on VEP amplitudes and EEG delta power: while VEPs declined stronger after longer rTMS trains, changes in EEG delta power were actually absent after 20 min. In addition, the variability of EEG power estimated from the standard error of the mean (s.e.m.) was usually higher after 20 min of rTMS and also increased with increasing time interval after rTMS. The reason for this may be related to spontaneous changes in EEG. Even during halothane/nitrous oxide anaesthesia, the endogenous sleep–wake rhythm of the cat goes on, leading to spontaneous transitions between high and low delta states about every 30 min (Lancel, 1993) which are not under the control of the experimenter. Thus, the likelihood that an internal state change happens at the same time as a change induced by rTMS is statistically lower for short (1 or 5 min) than for long (20 min) rTMS applications.
It has been known that changing vigilance levels affect VEPs (Brandt et al. 1991). However, in our study, we could not observe any relationship between EEG activity and VEPs. On one hand, changes in EEG delta power and changes in VEP amplitudes happened after different rTMS protocols, and on the other hand, changes in EEG occurred only in a transient fashion and only in the local EEG. Therefore, we suggest that changes observed in VEP amplitudes are more likely to be related to local effects of rTMS, e.g. via the induction of distinct activity patterns or local transmitter release, than to vigilance levels of the cat. Interestingly, Thut and colleagues (Thut et al. 2003) also found that the visually modulated alpha activity changed after 10 min of 1 Hz rTMS, but VEP amplitudes were almost unchanged. Finally, the failure to find significant changes in the high-frequency EEG bands (theta to gamma) in cats may be also related to the general reduction of these oscillations in anaesthesia.
The exact mechanisms underlying long-lasting changes in cortical excitability induced by rTMS remain unclear. It has been assumed that rTMS may modulate synaptic strength in a way comparable with long-term potentiation (LTP) or long-term depression (Wang et al. 1996). Interestingly, a recent study has demonstrated LTP-like effects on human VEP after a ‘photic tetanus’, a checkerboard rapidly presented at 9 Hz (1000 cycles; Teyler et al. 2005). Changes in VEP amplitude were restricted to the N1b component, which occurred at a latency of about 150 ms, within the second prominent VEP deflection, and may correspond to the P1 component (the second major deflection between 90 and 130 ms) of our cat VEP recordings. Thus, rTMS at 10 Hz may have a similar effect as a repetitive visual stimulation. In addition, alpha activity was reduced for about 1 h in the human study (Clapp et al. 2006). In the cat, alpha activity was not decreased after 1200 stimuli of 10 Hz rTMS but there was a trend for increasing alpha activity 10–15 min after rTMS finished. The partly different outcomes of these studies might be due to anaesthesia in the cat which promotes delta activity and dampens the higher frequencies, or could be a result of different stimulation patterns: 120 s continuous train in humans, 20 min intermittent application in cat.
In summary, we found that rTMS applied to visual cortex of the anaesthetized cats in different combinations of stimulus frequency and duration has outlasting effects on visual cortex excitability that are generally in accordance with effects observed in human motor and visual cortex, despite the diversity of rTMS effects. Our findings show that the rTMS effect depends largely on the combination of train frequency and duration, with 1 and 3 Hz rTMS being more effective after long-lasting applications (20 min with 1 Hz, and 5 or 20 min with 3 Hz), while short trains (1–5 min) were more effective for high-frequency rTMS (10 Hz). Furthermore, the temporal pattern of stimulation, given by rTMS frequency, seems to be more important than the total number of stimuli applied since, in case of 20 min rTMS, we applied the same number of stimuli (1200) for each rTMS frequency, with the result of different effects. With this study, we also demonstrate that rTMS applied to primary visual cortex (and presumably other cortical areas) of the anaesthetized cat can be seen as an appropriate model system to study rTMS effects on cortical physiology. This may offer to study the effect of different rTMS protocols on neuronal activity in combination with the application of neuro-active drugs, during states of changed balance between cortical excitation and inhibition, or in the course of re-organization processes that take place after peripheral or central lesions. This model further allows study of the changes in the expression of neuronal markers with histochemical methods after termination of the electrophysiological studies.
We are very grateful for the professional assistance of Ms Dimitrula Winkler and Ms Angelika Herker-See in the laboratory, and we would like to thank the members of the International Graduate School for Neuroscience (IGSN) for their critical comments. This study was supported by grants of the Deutsche Forschungsgemeinschaft (DFG) to K. Funke and U. Eysel (FU 256/2-1.2 + SFB509 TP C12) and by the IGSN which is funded by the Federal Ministry of Science and Education of Northrhine Westphalia.