Cortical sensory suppression during arousal is due to the activity-dependent depression of thalamocortical synapses
Manuel A. Castro-Alamancos,
Department of Neurology and Neurosurgery, Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada H3A 2B4
Corresponding author M. A. Castro-Alamancos: Department of Neurology and Neurosurgery, Room WB210, Montreal Neurological Institute, 3801 University Street, Montreal, Quebec, Canada H3A 2B4. Email: email@example.com
The thalamus serves as a gate that regulates the flow of sensory inputs to the neocortex, and this gate is controlled by neuromodulators from the brainstem reticular formation that are released during arousal. Here we show in rats that sensory-evoked responses were suppressed in the neocortex by activating the brainstem reticular formation and during natural arousal. Sensory suppression occurred at the thalamocortical connection and was a consequence of the activity-dependent depression of thalamocortical synapses caused by increased thalamocortical tonic firing during arousal. Thalamocortical suppression may serve as a mechanism to focus sensory inputs to their appropriate representations in neocortex, which is helpful for the spatial processing of sensory information.
The thalamus serves as a gate that regulates the flow of sensory inputs to the neocortex, and this gate is controlled by neuromodulators from the brainstem reticular formation that are released during arousal (Steriade et al. 1969, 1997; Singer, 1977; Sherman & Koch, 1986; Castro-Alamancos, 2002a,b). Among others, cholinergic and noradrenergic fibres project to the thalamus (Hallanger et al. 1990; Simpson et al. 1997). Neurons from these neuromodulatory systems discharge vigorously during behavioural arousal (Buzsaki et al. 1988; Aston-Jones et al. 1991), and the transmitters they release depolarize thalamocortical neurons and enhance their firing rates (McCormick, 1992). Thus, during aroused states of the brain, thalamocortical neurons display significantly enhanced spontaneous firing rates. Synapses are sensitive to activity and, in particular, thalamocortical synapses display robust depression when stimulated at high rates (Castro-Alamancos, 1997; Gil et al. 1997). These properties suggest that differences in the tonic firing rates of thalamocortical neurons between quiescent and aroused states can change the gain of thalamocortical synapses and significantly affect the mode of sensory transmission at the thalamocortical connection.
The present study investigates how the primary thalamocortical pathway changes during aroused states. We show that sensory responses evoked in the barrel cortex by whisker stimulation are suppressed during aroused states. Sensory suppression in the barrel cortex is mainly a consequence of the activity-dependent depression of thalamocortical synapses caused by increased thalamocortical tonic firing in VPM neurons during arousal. Thalamocortical suppression during aroused states of the brain may serve as a mechanism to focus sensory inputs to their appropriate representations (barrels) in neocortex, which is helpful for the spatial processing of sensory information.
Adult Sprague-Dawley rats (300 g) were anaesthetized with urethane (1.5 g kg−1i.p.) and placed in a stereotaxic frame. Lidocaine (2 %) was injected at incision sites and at points of contact of the skin with the frame. A unilateral craniotomy extended over a large area of the parietal cortex. Small incisions were made in the dura as necessary and the cortical surface was covered with artificial cerebrospinal fluid (ACSF) containing (mm): NaCl 126; KCl 3; NaH2PO4 1.25; NaHCO3 26; MgSO4.7H2O 1.3; dextrose 10; CaCl2.2H2O 2.5. Body temperature was automatically maintained constant with a heating pad. The level of anaesthesia was monitored with field recordings and limb-withdrawal reflexes and kept constant at about stage III/3 using supplemental doses of urethane (Friedberg et al. 1999). At the end of the experiments the animals were killed with an overdose of sodium pentobarbitone (i.p.). The Animal Care Committee of McGill University, Canada, approved protocols for all experiments.
Extracellular recordings were performed using electrodes (5-10 mΩ) filled with ACSF; single units and field potentials were recorded simultaneously via the same electrodes located in the VPM thalamus and the primary somatosensory neocortex (barrel cortex). When field potentials were recorded alone the electrode was placed at 800-1000 μm from the surface. Field potential polarity is displayed as negative down. Coordinates (in mm, from bregma and the dura; Paxinos & Watson, 1982) for the VPM thalamus recording electrode were anterior-posterior =−3.5, lateral = 3, depth = 5-6. Coordinates (mm) for stimulating the laterodorsal tegmentum (brainstem reticular formation; 100 Hz, 1 s) were posterior = 9, lateral = 0.7, depth = 5-6. The thalamic radiation was stimulated at approximately the following coordinates (mm): posterior = 3, lateral = 4, depth = 5. The medial lemniscus was stimulated at approximately the following coordinates (mm): posterior = 5.5, lateral = 1.5, depth = 7.5. Electrical stimuli consisted of 200 μs pulses of < 200 μA and were evoked using a concentric stimulating electrode.
To apply drugs into the neocortex during recordings a microdialysis probe (250 μm diameter, 2 mm long) was placed in the neocortex 0.5-1 mm medial from the recording electrode, as previously described (Castro-Alamancos, 2000). ACSF was continuously infused through the probe at 2-4 μl min−1. Drugs were prepared fresh, and protected from light and from oxidation (40 μM ascorbic acid in the ACSF) as required. Scopolamine, hexamethodide, phentolamine and propanolol were applied at 1-5 mm each, and CGP35348 (Novartis) was applied at 10 mm in ACSF. To apply TTX (2 μM in ACSF) into the VPM thalamus a microdialysis probe (250 μm diameter, 2 mm long) was inserted at the following coordinates (mm): posterior = 3, lateral = 2-3, depth = 4-6.
The sensory stimulation consisted of deflecting large caudal whiskers (one to four), which reliably discharged (> 80 % of trials at 0.1 Hz) the neurons recorded in VPM and barrel neocortex with short latencies (3-7 ms in VPM and 5-12 ms in neocortex). The selected whiskers were inserted into a glass micropipette (1 mm diameter) that was glued to the membrane of a miniature speaker. Application of a 1 ms square current pulse to the speaker deflected the micropipette and the whiskers inside ≈400 μm. Whisker stimulation was applied between 0.5 and 10 s after the reticular formation (RF) stimulation.
Current source-density analysis
A 16-channel linear silicon probe (CNCT, University of Michigan, USA) was inserted into the barrel cortex perpendicular to the pial surface. This required insertion of the silicon probe at a 45 deg angle (in the coronal plane) at 5.5-6 mm lateral from the midline. Field potential recordings were obtained simultaneously from the 16 sites on the probe and from a VPM electrode that served to monitor multiunit activity. Band-pass filter settings were selected for field potential (1 Hz to 3 kHz) or for multiunit recordings (300 Hz to 3 kHz). A current source-density analysis (CSD) was derived from the 16-channel cortical recordings, as previously described (Castro-Alamancos, 2000).
Adult Sprague-Dawley rats (300 g) were anaesthetized with sodium pentobarbitone (50 mg kg−1i.p.) and placed in a stereotaxic frame. Lidocaine (2 %) was injected at incision sites and at points of contact of the skin with the frame. Recording electrodes were placed in the barrel cortex and stimulating electrodes were placed in the thalamic radiation. An insulated stainless-steel bipolar recording electrode was placed in the whisker pad to record EMG signals. All electrodes and connectors were held in place using mini-screws and dental cement. During recovery from surgery the animals were given Buprinorphine (0.02 mg kg−1s.c.). Animals were allowed 5-7 days before testing and were recorded for several days up to a maximum of 15 days after surgery. During recovery after surgery, animals were closely monitored for any sign of distress or complications arising from the procedure. Electrophysiological recordings were performed as in anaesthetized animals, but JFET-operational amplifiers were attached to the recording electrodes at the animal's head connector. During the recording sessions the animal was placed in an open field containing photobeams that detected movements performed by the animal. The field potential activity in the neocortex and the motor activity detected with photobeams allowed us to differentiate periods of active exploration from periods of slow-wave sleep. For the population analysis, peak amplitudes of 20 randomly selected thalamic radiation-evoked responses were measured per animal (n= 10) and per condition (active vs. sleep). At the end of the experiments the animals were killed with an overdose of sodium pentobarbitone.
Thalamocortical suppression during activation
Single-unit recordings were obtained simultaneously from thalamocortical neurons of the VPM and from neurons in layers III-IV of the primary somatosensory barrel neocortex (Fig. 1) of urethane-anaesthetized rats. Application of a train of electrical stimulation (100 Hz, 1 s) to the brainstem reticular formation (RF stimulation) produced a strong effect typical of aroused states (Fig. 1A) called activation, which is characterized by an electrographic sign of low amplitude fast activity (Moruzzi & Magoun, 1949). At the single-neuron level, during activation the firing rate of all VPM thalamocortical neurons recorded increased (n= 55 of 55; 100 %), while the firing rate of the neocortical neurons recorded either decreased (n= 49 of 65; 75 %) or increased (n= 16 of 65; 25 %). VPM neurons increased their tonic firing after RF stimulation to 33 ± 4 Hz (mean ±s.d.) for several seconds.
In urethane-anaesthetized rats, whisker displacements using a mechanical stimulator produced successive sensory responses in the VPM and barrel cortex (Fig. 1B). Single units in VPM and in layer IV of barrel cortex responded with short latency (3-7 and 5-12 ms, respectively) and high fidelity (> 80 % probability of firing at short latency) to whisker stimulation delivered at low frequencies (0.1 Hz). When brain activation was induced by RF stimulation, the probability of firing to whisker stimulation at short latency (3-7 ms) in the VPM increased from 82 to 100 % (18 ± 3 % increase; P < 0.0001, Student's t test; n= 55 units). In contrast, in the barrel cortex the probability of firing to whisker stimulation at short latency (5-12 ms) decreased from 82 to 19 % (Fig. 1B and Fig. 3A) (63 ± 7 % reduction; P < 0.0001, t test; n= 65 units). Cortical single units that enhanced or reduced their spontaneous firing in response to RF stimulation both showed a suppressed sensory-evoked response during activation. Thus, cortical neurons that enhanced their tonic firing as a consequence of RF stimulation depressed their response to whisker stimulation (Fig. 1C). The decrease in responsiveness to whisker stimulation during activation was also reflected in the suppression of the sensory-evoked field potential response recorded in the cortex (57 ± 8 % reduction in amplitude; P < 0.0001, t test; n= 15; Fig. 1B and Fig. 3B). The field potential response was recorded via the same electrode as the single units and reflects the subthreshold synaptic activity of a population of neurons surrounding the electrode. It is noteworthy that the single-unit responses evoked by whisker stimulation showed on average a stronger suppression than the field potential responses (Fig. 3A and B). This was a consequence of the fact that some neurons such as the one in Fig. 1B almost entirely stopped responding to the sensory stimulus. It is likely that some of these neurons still produced a subthreshold response but we would not have been able to detect these responses using unit recordings. Other neurons may simply not respond at all after RF stimulation because they are only driven polysynaptically by the thalamic input and thus are entirely dependent on the firing of other cortical neurons, which may have been suppressed. This seems to be the case because cortical neurons responding with very short latency (5-8 ms) to whisker stimulation showed less suppression than the whole population of cortical neurons. Thus, the probability of firing to whisker stimulation for these short latency neurons was 88 % during control conditions and 40 % during activation induced by RF stimulation (48 ± 3 % reduction; P < 0.0001, t test; n= 12 units). The 48 % reduction of very short latency cells is significantly less than the 63 % reduction observed for the whole population of cells that includes cells with longer latencies. In summary, during activation induced by RF stimulation the sensory response recorded in the barrel cortex is suppressed, while the sensory response recorded in the VPM thalamus is not suppressed.
Suppression occurs at the thalamocortical connection
The barrel cortex is a complex structure that receives afferents from the VPM thalamus in both layers IV and VI, from where activity is distributed to other layers. To test which parts of this thalamocortical network are being suppressed by the RF stimulation we used a linear silicon probe containing 16 recording sites at 100 μm intervals to record voltage throughout the layers of neocortex (Fig. 2A) and derive a CSD in response to whisker stimulation (Bragin et al. 2000; Castro-Alamancos, 2000). The current flow in the barrel cortex revealed by the CSD (Fig. 2B) showed that the sensory-evoked response corresponded to short latency current sinks in upper layer VI and layer IV, which spread horizontally within those layers and vertically to layer III. Application of RF stimulation strongly depressed the sensory response in the barrel cortex, but not the sensory response in the VPM (Fig. 2B). Current flow in the neocortex was depressed beginning with the earliest (monosynaptic) sinks in layers VI and IV. As a consequence, the spread of activity within these layers and to layer III was also strongly suppressed. On average the peak amplitude of the short latency current sinks in layers IV and VI and the longer latency sink in layer III were significantly depressed by 51.6 ± 7, 59.8 ± 7 and 54.7 ± 8 %, respectively (n= 3 experiments; P < 0.0001, t test; Fig. 3C). We also found that the response evoked in the barrel cortex by stimulating thalamocortical fibres in the thalamic radiation was suppressed by RF stimulation (Fig. 2C and Fig. 3B; see below). In contrast, the response evoked in VPM by stimulating the primary sensory fibres in the medial lemniscus was not suppressed by RF stimulation. The thalamic response evoked by medial lemniscus stimulation has been characterized previously (Mishima, 1992). It consists of a very short latency and fast component (arrow in Fig. 2C) that is blocked by glutamate receptor antagonists (not shown) followed by a slower and longer latency component (asterisk in Fig. 2C) which is the recurrent corticothalamic response, as demonstrated by inactivating the neocortex (Mishima, 1992). RF stimulation did not significantly affect the initial fast response (Fig. 2C and Fig. 3B; n= 5 experiments; the peak amplitude of the medial lemniscus-evoked response was 1.1 ± 0.08 mV before and 1.19 ± 0.1 mV after RF stimulation; not significant, t test), but always abolished the long latency corticothalamic response that followed. Taken together, the results indicate that sensory suppression is occurring at the interface between the thalamus and the neocortex, at the thalamocortical connection.
Thalamocortical suppression occurs during behavioural arousal
The experiments presented thus far were performed in anaesthetized animals, and activation was induced artificially by RF stimulation. Although RF stimulation triggers wakefulness in sleeping animals (Lucas, 1975) and mimics many of the features of the aroused brain (Moruzzi & Magoun, 1949), the question remains whether suppression of the thalamocortical input actually occurs in behaving animals during activated states. To test this directly we chronically implanted recording and stimulating electrodes in the barrel cortex and thalamic radiation, respectively. The animals were placed in an open field (43 cm × 43 cm) and motor activity was monitored using photobeams and an EMG electrode in the whisker pad. We found that indeed during behaviourally activated states the thalamocortical response evoked by stimulating the thalamic radiation was suppressed. Figure 4 shows recordings from a rat during two distinct behavioural states: sleep and waking. During slow-wave sleep, as indicated by the enhanced fast Fourier transform (FFT) power of the spontaneous cortical activity at low frequencies (< 2 Hz; Fig. 4A), the thalamocortical-evoked response was at its greatest level. As the animal awoke, the thalamocortical-evoked response was strongly reduced, and was maintained at this reduced level during the vigourously active period of exploration that followed. During waking the FFT power showed an enhancement at 4-5 Hz (Fig. 4A). This is probably theta activity picked up by volume conduction from the cortical electrode because the FFT analysis of the spontaneous activity did not distinguish between negative and positive components. Sensory suppression occurred when waking occurred spontaneously or was triggered in a sleeping rat by the investigator. Based on recordings from several behaving animals (n= 10), the amplitude of the field potential thalamocortical response evoked by VPM stimulation was suppressed on average by 42 ± 7 % (P < 0.0001, t test; n= 10) between slow-wave sleep and active exploration. In conclusion, similar to the events that occur after RF stimulation, during behaviourally activated states the thalamocortical response is suppressed and therefore RF stimulation as used in the present study mimics this aspect of natural arousal.
Mechanisms of thalamocortical suppression
How does thalamocortical sensory suppression induced by RF stimulation occur? Thalamocortical synapses are sensitive to activity and display pronounced activity-dependent depression at frequencies above 1 Hz (Castro-Alamancos, 1997; Gil et al. 1997). Since RF stimulation produces a strong activating effect in thalamocortical neurons, which increases their firing rate, we reasoned that increased thalamocortical activity caused by RF stimulation could be depressing thalamocortical synapses and reducing the efficacy of the thalamocortical connection. If RF stimulation is depressing thalamocortical synapses by increasing thalamocortical activity, then blocking thalamocortical activity by inactivating the VPM thalamus should eliminate the suppressive effect of RF stimulation. VPM inactivation was produced with the sodium channel blocker tetrodotoxin (TTX), and was confirmed when whisker-evoked responses were completely absent in the neocortex (Fig. 5A). To test the effect of RF stimulation on the thalamocortical pathway before and after thalamic inactivation we stimulated the thalamic radiation. When the thalamus was intact, the response evoked in the barrel cortex by stimulating the thalamic radiation was suppressed by RF stimulation (Fig. 5B). However, when the VPM thalamus was inactivated with TTX, RF stimulation no longer suppressed the thalamic radiation-evoked response (Fig. 5B). This indicates that sensory suppression induced by RF stimulation is a consequence of increased thalamocortical firing in VPM. This experiment was performed several times (n= 6 rats) with similar results. On average the suppression of the thalamic radiation response was 55 ± 7 % before (P < 0.0001, t test; n= 6) and 6 ± 4 % after (P > 0.1, t test; n= 6) TTX application, i.e. there was a 90 % block of the effect of RF stimulation with thalamic inactivation.
Thalamocortical synapses depress in response to activity and also in response to application of certain neuromodulators in vitro (Gil et al. 1997; Hsieh et al. 2000) and in vivo (Oldford et al. 2000). To distinguish between the two possibilities, an activity-dependent depression of thalamocortical synapses or a neuromodulator-mediated depression of thalamocortical synapses, we tested whether TTX application in the VPM thalamus was affecting the cortical activating effects of RF stimulation. This was accomplished by comparing the power spectrums of cortical activity in the presence and absence of thalamic TTX (Fig. 5A). The results revealed that the activating effects of RF stimulation in the barrel cortex were not significantly different before and during thalamic TTX application (n= 6; t test for the power between 0.5-15 Hz, P > 0.1). This would be expected if the activating effect of RF stimulation in neocortex was mainly mediated by the basal forebrain (Jones, 1993). Since the modulation of cortical neurons caused by RF stimulation was still present during VPM inactivation with TTX, we reasoned that RF stimulation is not depressing the thalamocortical connection by releasing neuromodulator(s) in the cortex. Thus, sensory suppression and cortical activation induced by RF stimulation are independent processes. Conversely, we propose that increased thalamocortical activity during activated states produces the depression of thalamocortical synapses and consequently suppresses sensory-evoked responses in the neocortex. If this is the case, activity in thalamocortical synapses should be able to mimic the effect of RF stimulation. Indeed, similar to the effects of RF stimulation, repetitive stimulation at 10 Hz using sensory or thalamic radiation stimulation robustly suppressed thalamocortical responses (Castro-Alamancos & Connors, 1996) to a similar extent as RF stimulation (Fig. 5C). To further test the potential for a cholinergic, noradrenergic or GABAergic modulation of thalamocortical synapses in the barrel neocortex, we applied simultaneously cholinergic (scopolamine and hexamethodide), noradrenergic (phentolamine and propanolol) and GABAB (CGP35348) receptor antagonists via a microdialysis probe in the barrel neocortex. Application of this drug combination in the cortex via microdialysis (Fig. 6) significantly enhanced the amplitude of the whisker-evoked response and made the response broader (1.4 ± 0.2 mV before vs. 2 ± 0.3 mV after the drug combination; n= 3 rats; P < 0.0001, t test). However, application of this drug combination did not block the sensory suppression induced by RF stimulation (Fig. 6; n= 3, suppression by RF was 59 ± 6 % before and 75 ± 5 % after the drug combination).
The principal conclusion of the present study is that during aroused states the transmission efficacy of the thalamocortical connection is reduced leading to the suppression of sensory responses in the neocortex. This is a consequence of the activity-dependent depression of thalamocortical synapses caused by increased tonic firing of thalamic neurons. Importantly, this finding obtained initially using brainstem RF stimulation was validated in behaving animals. This indicates that the RF stimulation used in the present study mimics the cortical sensory suppression that occurs during natural aroused states.
As a sensory input travels upward from the periphery it is not depressed by RF stimulation until it reaches the neocortex. In fact, at the level of the thalamus, sensory responses are enhanced by RF stimulation (Steriade et al. 1969; Singer, 1977; Castro-Alamancos, 2002a,b). CSD revealed that the earliest current sinks in the thalamocortical recipient layers (IV and VI) of neocortex are suppressed by RF stimulation. The activity flow revealed by the CSD closely agrees with morphological studies, which have shown that thalamocortical fibres from VPM project to layers IV-III leaving collaterals in layer VI of the barrel neocortex (Bernardo & Woolsey, 1987; Jensen & Killackey, 1987), and with electrophysiological studies that mapped the laminar spread of whisker-evoked activity within the neocortex using single-unit recordings (Armstrong-James, 1995; Simons, 1995) and field potentials (Di et al. 1990). The CSDs are also similar to those obtained in primary somatosensory cortex using electrical stimulation of the ventroposterior lateral thalamus (VPL) (Castro-Alamancos & Connors, 1996; Kandel & Buzsaki, 1997). Since the thalamic output is enhanced and the earliest current sinks in the thalamocortical recipient layers (IV and VI) are suppressed, this indicates that sensory suppression occurs at the thalamocortical connection. This conclusion is supported by the observation that cortical cells which enhanced or reduced their tonic firing to RF stimulation both displayed a reduced sensory response during arousal, indicating that a change in cortical cell excitability cannot explain the suppression of sensory responses. Taken together the results indicate that cortical sensory suppression during arousal occurs at thalamocortical synapses.
Previous work has shown that sensory responses are reduced in the neocortex, thalamus and also brainstem sensory nuclei during behaviourally aroused states and movement in rodents, monkeys and humans (Chapin & Woodward, 1981; Nelson, 1984; Cohen & Starr, 1987; Shin & Chapin, 1989, 1990; Fanselow & Nicolelis, 1999). These investigators have proposed that a central modulatory process must account for sensory suppression since it occurs away from the periphery and in the absence of actual motor activity. The present study shows that one mechanism which contributes to the suppression observed at the cortical level is the significantly increased thalamocortical unit firing during arousal, which leads to activity-dependent depression of thalamocortical synapses. However, it is important to note that other factors that are not recruited by the RF stimulation used in the present study may also contribute to the changes previously detected at the level of the thalamus and brainstem sensory nuclei. The present study focused only on the thalamocortical pathway because this is what we found to be modified by RF stimulation. Accordingly, our behavioural experiments monitored only the thalamocortical sensory pathway and not the sensory pathways to the thalamus or brainstem. It is likely that additional modulatory systems, which are activated during arousal or movement, produce further effects at the thalamic and brainstem levels. It seems also clear that the modulations that occur during the waking state may be different depending on what the animal is actually doing (Chapin & Woodward, 1981; Fanselow & Nicolelis, 1999). Although this was not explored in detail in the present study, there are indications in Fig. 4 that this is the case because the amount of thalamocortical suppression varied during arousal. The present study emphasizes that during active behavioural states, such as exploration, thalamocortical suppression is prevalent and that RF stimulation simulates this effect in anaesthetized animals.
The present study proposes that increased thalamocortical unit firing produces activity-dependent depression of thalamocortical synapses, which leads to sensory suppression of neocortical responses. This conclusion is based on several findings. First, thalamocortical synapses depress with activity (Castro-Alamancos, 1997), and neuronal tonic firing increases in thalamocortical neurons during arousal. After RF stimulation the firing rate of all thalamocortical neurons increased to ≈33 Hz. This discharge rate is characteristic of VPM neurons in awake behaving rats (Nicolelis et al. 1993; Fanselow & Nicolelis, 1999). The analogous response of all VPM neurons to RF stimulation was expected because thalamocortical neurons represent a homogeneous population in their response to neuromodulators (McCormick & Prince, 1987; McCormick, 1992). In addition, the differential effect of RF stimulation on the firing of neocortical neurons was also expected because in vitro studies have shown distinct actions of neuromodulators depending on the cortical neuronal type (McCormick & Prince, 1985; McCormick, 1992; Xiang et al. 1998).
Second, blocking the firing of thalamocortical neurons in the VPM with TTX is sufficient to eliminate the thalamocortical sensory suppression induced by RF stimulation by about 90 %. A possible interpretation of this result is that it resulted from the block by TTX of the cortical activation mediated by the intralaminar nuclei of the thalamus (Steriade et al. 1997) (e.g. by spread of TTX to the intralaminar nuclei). However, this is unlikely for two reasons. (1) We found that cortical activation induced by RF stimulation is not different when the VPM thalamus is blocked with TTX. Thus, the cortical modulation induced by RF stimulation is still present during application of TTX in VPM, although the thalamocortical suppression induced by RF stimulation is blocked. (2) It is unlikely that the intralaminar nuclei were affected by the TTX because the distance between the microdialysis probe and the intralaminar nuclei is the same as that between the probe and the thalamic radiation. If TTX was spreading this distance the thalamic radiation-evoked responses should have been affected, which was not the case. Another important consideration with this experiment is that due to the need to inactivate VPM using TTX we had to use electrical stimulation of the thalamic radiation to stimulate thalamocortical fibres. However, electrical stimulation of the thalamic radiation also evokes corticothalamic responses (Castro-Alamancos & Calcagnotto, 2001), which means that layer VI corticothalamic neurons are being antidromically activated. There are two consequences of this. (1) The thalamus is recurrently stimulated by corticothalamic synapses. However, this is not a problem because under these same experimental conditions corticothalamic responses to low frequency stimulation are very small and only corticothalamic stimulation above 5 Hz produces a strong thalamic response due to facilitation (Castro-Alamancos & Calcagnotto, 2001). Moreover, the lack of involvement of the corticothalamic connection in the cortical responses evoked by thalamic radiation stimulation is demonstrated by the fact that the amplitude of the cortical responses to single stimuli of the thalamic radiation was not significantly different before and after application of TTX in VPM (Fig. 5). (2) The other consequence of the antidromic activation of layer VI corticothalamic neurons is that the cortex is recurrently stimulated via intracortical collaterals from these neurons that reach the upper layers (Zhang & Deschenes, 1997). Interestingly, these intracortical collaterals behave much the same way as corticothalamic synapses, producing strong facilitation (Stratford et al. 1996) and long latency responses because these small diameter fibres conduct much more slowly than the larger thalamocortical fibres (Ferster & Lindstrom, 1985; Swadlow, 1989). Consequently, we expected to observe a long latency response that could be attributed to these intracortical fibres when we stimulated at high frequencies. Indeed, as shown in Fig. 5C (asterisk), what we found was a very small, long latency response that followed the initial thalamocortical response. The amplitude of this facilitated response to repetitive stimulation was about 5-10 % of the amplitude of the response we measured to single stimuli, and it was expected to be even smaller to single stimuli because of the absence of facilitation. Thus, this leads to the conclusion that an intracortical component originating from axon collaterals of corticothalamic cells would not be present in the single stimuli responses we measured or it would be very small (< 5 % of the response) and have a long latency. Therefore, the response we measure in the cortex using single stimuli of the thalamic radiation is mostly (> 90 %), if not entirely, due to stimulation of thalamocortical fibres.
Finally, neuromodulators that may be released in the neocortex by RF stimulation are known to affect thalamocortical synapses when applied in vivo (Oldford et al. 2000) and in vitro (Gil et al. 1997; Hsieh et al. 2000). However, we found that cholinergic, noradrenergic and GABAB receptor antagonists applied together in the neocortex did not reduce sensory suppression induced by RF stimulation (in fact, they slightly enhanced the suppression), which demonstrates that these major neurotransmitter systems do not contribute to thalamocortical sensory suppression induced by RF stimulation.
Taken together, the results of the present study lead to the conclusion that increased thalamocortical activity in the VPM produces thalamocortical sensory suppression during arousal. Our results do not rule out the effects of a potential neuromodulator released in the neocortex by VPM thalamocortical activity, which could depress thalamocortical synapses. Alternatively, thalamocortical depression may be a consequence of an activity-dependent depletion of the synaptic machinery (Thomson, 2000). Both of these mechanisms would be blocked by application of TTX in the VPM thalamus. In vitro preparations are best suited to investigate these issues.
Interestingly, a very recent study has reached similar conclusions to ours (Swadlow & Gusev, 2001). They found that the efficacy of the connection between thalamic and cortical units was doubled immediately after silent periods of thalamic firing. Thus, in agreement with our results, thalamocortical efficacy is suppressed during periods of enhanced thalamic firing. We also found in a recent study that the corticothalamic connection is suppressed during arousal under the same experimental conditions (Castro-Alamancos & Calcagnotto, 2001). This was also manifest in the present study by the RF-induced suppression of the long latency component of the field potential response evoked in the VPM by medial lemniscus stimulation, which is a feedback corticothalamic response (Mishima, 1992). This means that during arousal the thalamo-cortico-thalamic recurrent loop is suppressed compared with quiescent states. The enhanced loop during sleep may serve to facilitate the propagation of slow oscillations, which are prominent in the thalamocortical system during that state. The suppressed loop during aroused states may impede the flow of low frequency signals and selectively allow the flow of high-frequency activity, as shown for the corticothalamic pathway (Castro-Alamancos & Calcagnotto, 2001).
What is the functional value of a suppressed thalamocortical connection during arousal? Thalamocortical suppression may be functionally useful as a gain regulator of activity reaching the neocortex (Abbott et al. 1997; Tsodyks & Markram, 1997). Increased thalamic tonic firing during activation will reduce the strength of the thalamocortical connection. By reducing the impact of thalamocortical inputs sensory representations become focused in the neocortex. This is important because in studies of sensory representation mapping in anaesthetized animals the area of neocortex that responds to a focal peripheral stimulus is extremely large. For instance, several barrels respond in the neocortex to deflection of a single whisker in anaesthetized rodents (Simons, 1978; Armstrong-James et al. 1992; Masino et al. 1993; Ghazanfar & Nicolelis, 1999; Moore et al. 1999; Petersen & Diamond, 2000). In contrast, because of thalamocortical sensory suppression during arousal, sensory inputs (i.e. whiskers) may become significantly focused in the neocortex to their appropriate representations (i.e. barrels). This could be particularly helpful for spatial processing, such as stimulus location, because the topographic arrangement at the morphological level is maintained at the physiological level.
We thank W. Sossin and B. Jones for comments on the manuscript, and Novartis for providing CGP35348. Multichannel silicon probes were provided by the University of Michigan Center for Neural Communication Technology sponsored by NIH NCRR. The Medical Research Council of Canada, Natural Sciences and Engineering Council of Canada, Fonds de la Recherche en Santé du Quebec, Canadian Foundation for Innovation and Savoy Foundation supported this research.