The Journal of Physiology

Transcription

1 J Physiol (2012) pp Positive allosteric modulation reveals a specific role for mglu2 receptors in sensory processing in the thalamus C. S. Copeland, S. A. Neale and T. E. Salt Department of Visual Neuroscience, Institute of Ophthalmology, University College London, London, UK The Journal of Physiology Non-technicalsummary Neurones in the ventrobasal thalamic nucleus (VB) are the major source of somatosensory input to the cerebral cortex. Thorough investigation has identified neuronal circuits formed between the VB and the adjacent thalamic reticular nucleus (TRN); and Group II metabotropic glutamate (mglu) receptors located within this circuitry have been demonstrated as capable of modulating somatosensory transmission. There are two Group II mglu receptor subtypes, mglu2 and mglu3, and this study has demonstrated that there is an mglu2 component to the Group II mglu receptor effect on sensory transmission within the VB, and that the mglu2 receptor subtype may be activated physiologically upon sensory stimulation. It is believed that this circuitry is of importance in the control of sensory discriminative processes, and we propose that mglu2 functions within this circuitry to enable relevant sensory information to be discerned from background activity. Abstract Group II metabotropic glutamate receptor (mglu) modulation of sensory processing in the rat ventrobasal thalamic nucleus (VB) has been extensively studied in vivo. However, it is not yet known what the relative contributions are of the Group II mglu receptor subtypes (mglu2 and mglu3) to this modulation, nor to what extent these receptors may be activated under physiological conditions during this process. Using single-neurone recording in the rat VB in vivo with local application of the selective Group II agonist LY and the subtype selective mglu2 positive allosteric modulator (PAM) LY487379, our findings were twofold. Firstly, we found that there is an mglu2 component to the effects of LY on sensory responses in the VB. Secondly, we have demonstrated that application of the PAM alone can modulate sensory responses of single neurones in vivo. This indicates that mglu2 receptors can be activated by endogenous agonist following physiological sensory stimulation. We speculate that the mglu2 subtype could be activated under physiological stimulus-evoked conditions by glutamate spillover from synapses between excitatory sensory afferents and VB neurones that can lead to a reduction in sensory-evoked inhibition arising from the thalamic reticular nucleus (TRN). We propose that this potential mglu2 receptor modulation of inhibition could play an important role in discerning relevant information from background activity upon physiological sensory stimulation. Furthermore, this could be a site of action for mglu2 PAMs to modulate cognitive processes. (Received 29 August 2011; accepted after revision 20 December 2011; first published online 23 December 2011) Correspondingauthor C. S. Copeland: Department of Visual Neuroscience, UCL Institute of Ophthalmology, University College London, Bath Street, London EC1V 9EL, UK. c.copeland@ucl.ac.uk Abreviations mglu, metabotropic glutamate receptor; mglu1, metabotropic glutamate receptor subtype 1; mglu2, metabotropic glutamate receptor subtype 2, mglu3, metabotropic glutamate receptor subtype 3; mglu4, metabotropic glutamate receptor subtype 4; mglu5, metabotropic glutamate receptor subtype 5; mglu6, metabotropic glutamate receptor subtype 6; mglu8, metabotropic glutamate receptor subtype 8; MWE, excitation to multiple whiskers; PAM, positive allosteric modulator; PSTH, poststimulus time histogram; SWE, excitation to a single whisker; TRN, thalamic reticular nucleus; VB, ventrobasal thalamic nucleus. DOI: /jphysiol

2 938 C. S. Copeland and others J Physiol Introduction Somatosensory information ascends to the cerebral cortex via the ventrobasal thalamic nucleus (VB). Common to all thalamic nuclei, the VB possesses excitatory glutamatergic and inhibitory GABAergic synapses that contribute to the overall gating and processing of sensory information. TherodentVBcontainsonlyonemajorcelltype,the thalamocortical relay neurone (Peschanski et al. 1984; Harris, 1986; Ohara & Havton, 1994), and is largely devoid of intrinsic Golgi II inhibitory interneurons, receiving its GABAergic input almost exclusively from the adjacent thalamic reticular nucleus (TRN) (Ralston, 1983; Barbaresi et al. 1986; Harris & Hendrickson, 1987). The TRN is innervated by collaterals from VB relay neurone axons and by layer VI-originating corticothalamic fibres, both of which traverse the TRN. Thus the TRN is thought to coordinate the activation of appropriate thalamocortical circuits during selective attention (Crick, 1984; Pinault, 2004). GABAergic innervation from the TRN strongly influences how VB neurones process sensory information, making it important to understand how this inhibition is controlled. The eight known metabotropic glutamate receptor subtypes (mglu1 mglu8) are divided into three groups based upon their sequence homology, signal transduction mechanism and pharmacology (Nakanishi, 1992; Conn & Pin, 1997; Niswender & Conn, 2010). In general, the Group I receptors (mglu1, mglu5) have predominately postsynaptic actions, whereas the Group II (mglu2, mglu3) and Group III (mglu4, mglu6 8) receptors have predominately presynaptic actions. However, the relative locations of these receptor subtypes is now evidently more complex than this, with receptors from all three groups possessing post-, pre- or extrasynaptic actions (Conn & Pin, 1997; Niswender & Conn, 2010). There is considerable evidence to indicate that local application of selective Group II mglu receptor agonists in vivo can modulate somatosensory transmission within the rat VB by reducing TRN-originating inhibition (Salt & Eaton, 1995a; Salt & Turner, 1998). These findings are supported by ultra-structural studies of the rodent VB: Group II mglu receptors appear to be localised in glial processes surrounding GABAergic terminals (Liu et al. 1998; Mineff & Valtschanoff, 1999), and mglu3 can be found on TRN-originating axons (Tamaru et al. 2001). However, the relative contributions of the mglu2 and mglu3 subtypes to the observed Group II receptor effects is not known, nor to what extent these receptors may be activated under physiological conditions. The recent development of selective positive allosteric modulators (PAMs) for mglu2 (Niswender & Conn, 2010) enables the specific identification of mglu2 receptor-mediated components to responses. Such compounds typically potentiate orthosteric agonist-evoked responses whilst having relatively little direct effect on receptor activity (Ehlert, 1986; Christopoulos, 2002). Therefore, using an in vivo preparation with different sensory stimulation protocols and the PAM LY (Johnson et al. 2003) as a tool, the aim of the present study was twofold. Firstly, to identify whether there is an mglu2 component to the Group II effect on GABAergic inhibition in VB by investigating whether LY potentiates the effects of an exogenous Group II agonist, LY (Schoepp et al. 2003), on neurone responses. Secondly, to examine if mglu2 receptors are activated by endogenous neurotransmitter during the synaptic processing of sensory inputs by applying LY alone. Methods All experiments were conducted using adult male Wistar rats ( g). Animals were purchased from Harlan and were housed on a 12 h light dark cycle with unlimited access to food and water. All experimental conditions and procedures were in accordance with the UK Animals (Scientific Procedures) Act 1986 and associated guidelines, and the experiments comply with the policies and regulations of The Journal of Physiology (Drummond, 2009). Subjects (n = 47) were anaesthetised with urethane (1.2 g kg 1 I.P.) and were prepared for recording as detailed previously (Salt, 1987, 1989). Throughout the experiments electroencephalogram and electrocardiogram were monitored. Additional urethane anaesthetic was administered I.P. as required, and the experiment was terminated with an overdose of the same anaesthetic. Neuromuscular blocking agents were not used at any stage of the experiment. Seven-barrel recording and iontophoretic glass pipettes were advanced into the VB. Extracellular recordings were made from single VB neurones responsive to somatosensory input through the central barrel (filled with 4 M NaCl). Iontophoretic drug applications were performed using the outer barrels (Salt, 1987, 1989). On each occasion, one of the outer barrels was filled with 1 M NaCl for current balancing. The remaining outer barrels each contained one of the following substances: NMDA (50 mm, ph 8.0 in 150 mm NaCl), AMPA (25mM, ph 8.0 in 150 mm NaCl), (1S,2S,5R,6S)-2- aminobicyclo[3.1.0]hexane-2,6-dicarboxylic acid (LY354740; 5 mm, ph 8.0 in 75 mm NaCl), (2S)-2-amino- 2-[(1S,2S)-2-carboxycycloprop-1-yl]-3-(xanth-9-yl) propanoic acid (LY341495; 5 mm, ph 8.5 in 75 mm NaCl), as Na + salts, ejected as anions, and 2,2,2-trifluoro-N-[4-(2-methoxyphenoxy)phenyl]-N- (3-pyridinylmethyl)ethanesulfonamide hydrochloride (LY487379; 1 mm, ph 6.0, in 1% dimethyl sulfoxide (DMSO), 75 mm NaCl) and 1% DMSO (ph 6.0, 75 mm

3 J Physiol mglu2 positive allosteric modulation in the ventrobasal thalamus 939 NaCl), ejected as cations. All compounds were prevented from diffusing out of the pipette by using a retaining current (10 20 na) of opposite polarity to that of the ejection current. All compounds were obtained from Tocris. Throughout the study, extracellular single neurone action potentials were gated, timed and counted using a window discriminator, a CED1401 interface and Spike2 software, which recorded the output from the iontophoresis unit and also triggered some of the sensory stimuli sequences. Data were analysed by plotting poststimulus time histograms (PSTHs) from these recordings by counting the spikes evoked by either agonist ejection or sensory stimulation. Neurones were identified as VB neurones on the basis of stereotaxic location (Paxinos & Watson, 1998) and responses to vibrissa deflection. Vibrissa deflection was performed using fine air jets directed through 23 gauge needles mounted on micromanipulators positioned and orientated close to the vibrissa to ensure deflection of a single vibrissa was achieved. Air jets were electronically gated with solenoid valves that produced a rising air pulse at the vibrissa 8 ms after switching. Response latencies were calculated from the start of the gating pulse. Using such an approach it is possible to use air jets on adjacent vibrissae and only evoke an excitatory response from one of the vibrissa stimuli (Fig. 1), indicating the specificity of the stimulation procedure, as described previously (Salt, 1989). Prior to the beginning of each of the experimental protocols described below, the principal vibrissa (i.e. the vibrissa at the centre of the receptive field) for each neurone was identified, and responses to additional vibrissae were noted. Neurones could thus be classified as responding with excitation to a single whisker (SWE) or multiple whiskers (MWE), as described by other workers (Ito, 1988; Brecht & Sakmann, 2002). All neurones recorded from were quiescent. The effects of the Group II agonist and the mglu2 PAM on sensory responses Cycles of sensory stimulation (10 s long) were established and repeated continuously whilst recording from neurones. Cycles contained two types of sensory stimuli, which consisted of electronically gated short (10 30 ms) and long ( ms) duration air jets directed at the principal vibrissa, with 4 5 s interstimulus intervals. Depending upon the response profile of the neurone undergoing investigation, the long air jet was applied as either a continual stimulation or as a train of stimuli at a frequency of 5 7 Hz. After several control cycles displaying consistent neurone responses had been recorded, LY354740, LY and LY were iontophoretically ejected either alone or in conjunction with each other for 2 20 min as required. After cessation of LY354740, LY and/or LY ejection, sensory stimulation cycles were continued until neurone responses had returned to control levels. An inter-stimulus interval of 4 5 s was sufficient to ensure that any post-stimulus effects from either stimulus type were no longer apparent upon subsequent stimulation (Salt, 1989; Turner & Salt, 2003). This experimental protocol was also followed when the DMSO vehicle was iontophoretically ejected either alone or in conjunction with LY as appropriate. The effects of the Group II agonist and the mglu2 PAM on inhibitory sensory responses Individual VB neurones receive direct dominant input from a principal vibrissa via the trigeminal nuclei (Diamond et al. 1992; Nicolelis & Chapin, 1994; Pinault & Deschenes, 1998; Crabtree, 1999; Lavallee & Deschenes, 2004). VB neurone projections reach the cerebral cortex and the TRN, with the TRN inhibitory neurones receiving innervation from several vibrissae, which in turn make reciprocal projections back to the thalamus (Jones, 1985; Salt & Eaton, 1989; Shosaku et al. 1989; Pinault & Deschenes, 1998; Salt & Turner, 1998). Therefore, upon principal vibrissal stimulation, the responses of a VB neurone are the result of excitatory glutamatergic input and recurrent GABAergic inhibition (Salt, 1989; Shosaku et al. 1989; Lee et al. 1994; Hartings & Simons, 2000; Binns et al. 2003), whereas upon stimulation of surrounding secondary vibrissae a lateral GABAergic inhibition via the TRN is evoked (Salt, 1989; Pinault & Deschenes, 1998; Crabtree, 1999) (Fig. 1). In order to reveal these GABAergic inhibitory processes arising from the TRN, we used a condition-test protocol with two air jets each directed at an adjacent receptive field area (Salt, 1989). A conditioning stimulus presented to a secondary vibrissa preceding a test stimulus to a principal vibrissa by ms can inhibit the response of a VB neurone to stimulation of the principal vibrissa (Salt, 1989; Turner & Salt, 2003) (Fig. 1). The precise interval between presentation of the conditioning stimulus and the test stimulus was altered within the ms time window to produce the maximum inhibition achievable for each individual cell. The secondary vibrissa was normally in the same row as the principal vibrissa, but removed from the principal vibrissa by one or two positions (Lavallee & Deschenes, 2004). Several control cycles which each contained two sensory stimulations (presentation of the test stimulus alone and then the condition test stimuli, separated by an inter-stimulus interval of 4 5 s) that demonstrated the establishment of sensory inhibition were recorded, and these were then repeated during the concurrent application of LY and/or LY for 2 10 min as required. After cessation

4 940 C. S. Copeland and others J Physiol of LY and/or LY ejection, cycles were continued until neurone responses had returned to control levels. This experimental protocol was also followed when the DMSO vehicle was iontophoretically ejected either alone or in conjunction with LY as required. The degree of inhibition was quantified as a percentage as follows: Percentage inhibition = (1 T C /T) 100% where T is the total number of action potentials evoked to the test stimulus and T C is the total number of action potentials evoked to the test stimulus when preceded by the condition stimulus. Using this calculation, the maximum inhibitory response that can be achieved is 100%. Action potentials evoked ms following test stimuli were counted in analysis, as all responses occurred within 100 ms of stimulus presentation: the approximate time scale during which inhibition evoked by principal or secondary vibrissa stimulation is likely to reach its peak (Salt, 1989; Turner & Salt, 2003). It is important to note that the condition stimulus used to evoke an inhibition of a test response may in itself produce an excitatory response in some neurones. Aa Ba b b Figure 1. Circuitry underlying responses of VB neurones to vibrissa stimulation Aa, stimulation of the principal vibrissa drives excitation in the recorded VB neurone (VB), followed by recurrent inhibition via the TRN. Ab, PSTH record showing responses of the principal vibrissa when stimulated with a short-duration stimulus. Ba, stimulation of an adjacent secondary vibrissa drives lateral inhibition onto the recorded VB neurone, via the TRN. Bb, PSTH record showing responses when stimulation of the principal vibrissa is preceded by stimulation of the secondary vibrissa. Note the reduction in the response to the test stimulus (arrow). T: 20 ms short-duration test stimulation to the principal vibrissa; C: 20 ms conditioning stimulation to a secondary vibrissa; spike count: the total number of action potentials recorded per 2 ms bin over 6 trials. PSTHs were constructed using responses from the same cell (CVB050b).

5 J Physiol mglu2 positive allosteric modulation in the ventrobasal thalamus 941 The effects of the Group II agonist and the mglu2 PAM on lateral and recurrent inhibition As an alternative means of revealing functional recurrent or lateral inhibition onto VB neurones, NMDA was continuously applied to neurones whilst alternate presentations of short duration stimuli to either the principal or secondary vibrissa took place (separated by 5 s). NMDA ejections were adjusted to produce stable submaximal increases in action potential firing of neurones, typically into the range of Hz. The degree of inhibition produced when either the principal or secondary vibrissa was stimulated was quantified from cumulative PSTHs of action potential spikes before, during and after the stimulus was applied. The first 100 ms of evoked inhibition was used in the quantification analysis as this is the approximate time scale during which inhibition evoked by principal or secondary vibrissa stimulation is likely to reach its peak (Salt, 1989; Turner & Salt, 2003). Spikes counted during this period of inhibition are expressed as a percentage of the background firing rate. After several control cycles displaying consistent neurone responses had been recorded, either LY or LY was iontophoretically ejected for min as required. After cessation of LY or LY ejection, cycles were continued until neurone responses had returned to control levels. The effects of the Group II agonist and the mglu2 PAM on responses to ionotropic glutamate receptor agonist application In order to directly assess the effects of Group II mglu agents on postsynaptic responses to NMDA and AMPA receptor activation, cycles consisting of brief ejections (10 15 s) of NMDA or AMPA with intervals of s, were established and repeated continuously. NMDA and AMPA ejection parameters were adjusted to ensure that excitatory neuronal responses evoked by these ejections were submaximal. After several cycles of NMDA and AMPA responses had been recorded under control conditions, LY or LY was iontophoretically ejected during one to three cycles of NMDA/AMPA ejection as required. After cessation of LY or LY ejection, NMDA/AMPA ejection cycles were continued until neurone responses had returned to control levels. Statistical analysis Data are expressed as a percentage of control responses prior to Group II agonist and/or mglu2 PAM application (±SEM) and comparisons were made using Wilcoxon s matched-pairs test (P < 0.05). Results It has been previously demonstrated that the selective activation of Group II mglu receptors can modulate GABAergic afferent inhibition in VB (Salt & Eaton, 1995a; Salt & Turner, 1998; Turner & Salt, 2003). The results presented in this study complement these previous data whilst also providing evidence that there is an mglu2 component to this modulation. In addition, the present study has also elucidated a possible mechanism of action by which the mglu2 receptor subtype is activated in vivo during the synaptic processing of sensory inputs. The effects of the Group II agonist LY and the mglu2 PAM LY on sensory responses to short-duration vibrissa stimulation Initial investigation began by testing the effects of the Group II agonist and the mglu2 PAM on VB neurones with vibrissal receptive fields that were responsive to a short-duration stimulus directed at the principal vibrissa. Neuronal responses to short-duration stimuli were at a latency of ms (mean = 18.9 ± 0.71 ms) from stimulus onset. When this is adjusted for the time delay introduced by the air-jet stimulator (8 ms), these values correspond closely to those described by others (Ito, 1988; Brecht & Sakmann, 2002). Iontophoretic ejections of the Group II agonist alone significantly increased the responses of neurones to short-duration stimuli (128 ± 4% of control, n = 42) (Fig. 2Aa), as did application of the mglu2 PAM alone (128 ± 6% of control, n = 38) (Fig. 3A). In order to confirm the specificity of the agonist and PAM applications, the Group II antagonist LY was subsequently co-applied with these agents in a subset of the same neurones. LY reversed the effects of both the Group II agonist (LY alone: 138 ± 2% of control; LY plus LY ± 5% of control, n = 5), and the mglu2 PAM (LY alone: 127 ± 7% of control; LY plus LY ± 4% of control, n = 7) on neuronal responses to short-duration stimuli. On a subset of the neurones upon which the effects of the Group II agonist alone were tested, the mglu2 PAM was co-ejected with the Group II agonist and was found to significantly potentiate the agonist effect on neurone responses to the short-duration stimuli when compared to responses from the same neurones when the Group II agonist was applied alone (LY alone: 131 ± 6% of control; LY plus LY487379: 166 ± 12% of control, n = 22) (Fig. 2Ba). As a control for the use of DMSO vehicle, on some neurones an iontophoretic current was passed through the 1% DMSO solution that the mglu2 PAM was dissolved in at the same current as the PAM ejection on the same neurones, either alone or in conjunction with the Group II agonist as appropriate.

6 942 C. S. Copeland and others J Physiol Passing iontophoretic current through the vehicle barrel did not significantly alter responses of neurones to short-duration stimuli when applied alone (100 ± 2% of control, n = 10), nor did it significantly potentiate the responses of neurones when co-applied with the Group II agonist (LY alone: 124 ± 11% of control; LY plus 1% DMSO: 135 ± 13% of control, n = 10). Aa b Ba b Figure 2. VB neurone responses before, during and after iontophoretic application of the Group II agonist LY either alone or co-applied with the mglu2 PAM LY upon short-duration stimulation of the principal vibrissa and execution of the condition test protocol Aa, the top three PSTHs represent responses from a single neurone to short-duration stimulation of the principal vibrissa under control conditions, during application of the Group II agonist, and during recovery. The bottom bar graph represents the mean responses of a group of neurones (n = 44) to short-duration stimulation of the principal vibrissa under the same conditions. Ab, the top three PSTHs represent responses from a single neurone upon execution of the condition test protocol under control conditions, during application of the Group II agonist, and during recovery. The bottom bar graph represents the mean inhibitions of a group of neurones (n = 25) to the condition test protocol under the same conditions. Ba, the top three PSTHs represent responses from a single neurone to short-duration stimulation of the principal vibrissa under control conditions, upon co-application of the Group II agonist and the mglu2 PAM and during recovery. The bottom bar graph represents the mean responses of a group of neurones (n = 22) to short-duration stimulation of the principal vibrissa under control conditions, upon application of the Group II agonist alone, and upon co-application of the Group II agonist and the mglu2 PAM. Bb, the top three PSTHs represent responses from a single neurone to the condition test protocol under control conditions, during co-application of the Group II agonist and the mglu2 PAM, and during recovery. The bottom bar graph represents the mean reduction in sensory inhibition of a group of neurones (n = 14) compared to control to application of the Group II agonist alone, or co-applied with the mglu2 PAM upon execution of the condition test protocol. T: 20 ms short-duration test stimulation to the principal vibrissa; C: 20 ms conditioning stimulation to a secondary vibrissa; spike count: the total number of action potentials recorded per 2 ms bin over 6 trials. All PSTHs are constructed using responses from the same cell (CVB050b). Bar graph data are expressed as means ± SEM; P < 0.05.

7 J Physiol mglu2 positive allosteric modulation in the ventrobasal thalamus 943 The effects of the Group II agonist LY and the mglu2 PAM LY on inhibitory sensory responses On a subset of the VB neurones upon which the responses to a single short-duration vibrissa stimulus during Group II agonist and/or mglu2 PAM application had been examined, an additional sensory stimulation protocol was performed. Condition test protocols, which reveal GABAergic inputs from the TRN (Salt, 1989) (see Methods), were carried out under control conditions and in the presence of the Group II agonist and/or the mglu2 PAM.ForthispopulationofVBneuronesthedegreeof sensory inhibition observed under control conditions was 82± 2%. The properties of the afferent-evoked inhibition in these data are similar to those that have been previously presented (Salt, 1989; Salt & Eaton, 1995a;Salt&Turner, 1998). Iontophoretic ejections of the Group II agonist alone significantly reduced sensory inhibition (21 ± 3% reduction in sensory inhibition compared to control, n = 25) (Fig. 2Ab). The effect on sensory inhibition was reversed upon co-application of the Group II antagonist LY (LY354740: 22 ± 4% reduction in sensory inhibition compared to control; LY plus LY341495: 0.5± 3% reduction in sensory inhibition compared to control, n = 5) in a subpopulation of the same neurones. By contrast, iontophoretic ejections of the mglu2 PAM alone had no significant effect on the degree of sensory inhibition compared to control (1 ± 5% reduction in sensory inhibition, n = 7) (Fig. 3B). This result is of particular interest, as although the mglu2 PAM when applied alone had no effect on sensory inhibition, responses to short-duration sensory stimulation of the principal vibrissa were significantly potentiated in the same population of neurones (134 ± 14% of control, n = 7); a response profile to the short-duration stimulus which is reflected in the overall population of neurones upon which the mglu2 PAM was applied alone (Fig. 3A). On a subset of the neurones upon which the effects of the Group II agonist alone were tested, the mglu2 PAM was co-ejected with the Group II agonist and was found to significantly potentiate the agonist effect on sensory inhibition when compared to the reduction in sensory inhibition in the same neurones when the Group II agonist was applied alone (LY alone: 20 ± 4% reduction in sensory inhibition; LY plus LY487379: 30 ± 7% reduction in sensory inhibition, n = 14) (Fig. 2Bb). Similarly to before, vehicle control experiments in which iontophoretic current was passed through the 1% DMSO barrel at the same current as the mglu2 PAM application while ejecting the Group II agonist on the same neurones indicated that there was no significant effect of vehicle (LY alone: 23 ± 6% reduction in sensory inhibition; LY plus 1% DMSO: 20 ± 5% reduction in sensory inhibition, n = 7). The effects of the Group II agonist LY and the mglu2 PAM LY on lateral and recurrent inhibition This experimental protocol was performed in order to further investigate why application of the mglu2 PAM A Figure 3. VB neurone responses before, during and after iontophoretic application of the mglu2 PAM LY upon short-duration stimulation of the principal vibrissa and execution of the condition test protocol A, the top three PSTHs represent responses from a single neurone to short-duration stimulation of the principal vibrissa under control conditions, during application of the mglu2 PAM, and during recovery. The bottom bar graph represents the mean responses of a group of neurones (n = 39) to short-duration stimulation of the principal vibrissa under the same conditions. B, the top three PSTHs represent responses from a single neurone upon execution of the condition test protocol under control conditions, during application of the mglu2 PAM, and during recovery. The bottom bar graph represents the mean inhibitions of a group of neurones (n = 7) to the condition test protocol under the same conditions. T: 20 ms short-duration test stimulation to the principal vibrissa; C: 20 ms conditioning stimulation to a secondary vibrissa; spike count: the total number of action potentials recorded per 2ms bin over 6 trials. All PSTHs are constructed using responses from the same cell (CVB049c). Bar graph data are expressed as means ± SEM; P < B

8 944 C. S. Copeland and others J Physiol potentiated the VB neurone response to test stimulation at the principal vibrissa, but had no effect on the degree of sensory inhibition evoked by preceding test stimulation of the principal vibrissa with a conditioning stimulation A B of a secondary vibrissa. In this protocol, action potential firing rate of the neurone was elevated with a continuous ejection of NMDA while principal and secondary vibrissa stimuli were alternately delivered, and the degree of recurrent and lateral inhibition respectively was revealed by the reduction in NMDA-evoked neuronal firing (see Methods). Iontophoretic application of the Group II agonist significantly reduced both the lateral and recurrent sensory inhibition to a similar degree in the same group of neurones (Lateral: 45 ± 9% reduction in inhibition compared to control; Recurrent 45 ± 7% reduction in inhibition compared to control, n = 6) (Fig.4). In contrast, application of the mglu2 PAM significantly reduced recurrent inhibition (29 ± 5% reduction in inhibition compared to control, n = 6), but did not alter the level of lateral inhibition ( 3 ± 5% reduction in inhibition compared to control, n = 6) in the same group of neurones (Fig. 5). No significant effects of the Group II agonist or the mglu2 PAM on the neural firingevokedbynmdawereseen(seebelowforfurther details). Figure 4. VB neurone responses before, during and after iontophoretic application of the Group II agonist LY upon stimulation of either the principal or secondary vibrissa to evoke recurrent and lateral inhibition respectively A, the top three PSTHs represent responses from a single neurone during continuous NMDA application (see Methods) to short-duration stimulation of the principal vibrissa under control conditions, during application of the Group II agonist, and during recovery. The bottom bar graph represents the mean inhibitions of a group of neurones (n = 6) to short-duration stimulation of the principal vibrissa under the same conditions. B, the top three PSTHs represent responses from a single neurone during continuous NMDA application to short-duration stimulation of a secondary vibrissa under control conditions, during application of the Group II agonist, and during recovery. The bottom bar graph represents the mean inhibitions of a group of neurones (n = 6) to short-duration stimulation of a secondary vibrissa under the same conditions. T: 30 ms short-duration test stimulation to the principal vibrissa; C: 30 ms conditioning stimulation to a secondary vibrissa; spike count: the total number of action potentials recorded per 50 ms bin over 30 trials. All PSTHs are constructed using responses from the same cell (CVB060b). Bar graph data are expressed as means ± SEM; P < The effects of the Group II agonist LY and the mglu2 PAM LY on sensory responses to long-duration vibrissa stimulation As synaptic activation of mglu receptors has been often shown to be dependent upon repetitive stimulation of pathways in vitro in a number of brain regions (Ohishi et al. 1994; Fitzsimonds & Dichter, 1996; Wada et al. 1998; Vogt & Nicoll, 1999; Mitchell & Silver, 2000; Semyanov & Kullmann, 2000; Turner & Salt, 2000; Arnth-Jensen et al. 2002; Piet et al. 2003; Piet et al. 2004; Neale & Salt, 2006), we thought it important to investigate whether such activity-dependent activation occurs in vivo in response to physiological stimuli. This was done by examining the effects of Group II activation on neuronal responses to repetitive vibrissal stimulation using a long-duration stimulus directed at the principal vibrissa. Iontophoretic ejections of the Group II agonist alone significantly increased the responses of neurones to long-duration stimuli (154 ± 10% of control, n = 19) (Fig. 6A), as did application of the mglu2 PAM alone (135 ± 5% of control, n = 29) (Fig. 7). Co-ejection of the mglu2 PAM along with the Group II agonist significantly potentiated the agonist effect on neurone responses to long-duration stimuli when compared to responses from the same neurones when the Group II agonist was applied alone (LY alone: 170 ± 20% of control; LY plus LY487379: 246 ± 30% of control, n = 8) (Fig. 6B). As in previous protocols, passing iontophoretic current through the 1% DMSO barrel did not significantly alter responses of neurones when applied alone (100 ± 2% of

9 J Physiol mglu2 positive allosteric modulation in the ventrobasal thalamus 945 control, n = 10), nor did it further potentiate the responses of neurones when co-applied with the Group II agonist (LY alone: 159 ± 20% of control; LY plus 1% DMSO: 159 ± 21% of control, n = 6). A B The effects of the Group II antagonist LY on sensory responses to short- and long-duration vibrissa stimulation As the mglu2 PAM was able to potentiate neuronal responses to both the short- and long-duration A B Figure 5. VB neurone responses before, during and after iontophoretic application of the mglu2 PAM LY upon stimulation of either the principal or secondary vibrissa to evoke recurrent and lateral inhibition respectively A, the top three PSTHs represent responses from a single neurone during continuous NMDA application (see Methods) to short-duration stimulation of the principal vibrissa under control conditions, during application of the mglu2 PAM, and during recovery. The bottom bar graph represents the mean inhibitions of a group of neurones (n = 6) to short-duration stimulation of the principal vibrissa under the same conditions. B, the top three PSTHs represent responses from a single neurone during continuous NMDA application to short-duration stimulation of a secondary vibrissa under control conditions, during application of the mglu2 PAM, and during recovery. The bottom bar graph represents the mean inhibitions of a group of neurones (n = 6) to short-duration stimulation of a secondary vibrissa under the same conditions. T: 30 ms short-duration test stimulation to the principal vibrissa; C: 30 ms conditioning stimulation to a secondary vibrissa; spike count: the total number of action potentials recorded per 50 ms bin over 30 trials. All PSTHs are constructed using responses from the same cell (CVB060b). Bar graph data are expressed as means ± SEM; P < Figure 6. VB neurone responses before, during and after iontophoretic application of the Group II agonist LY either alone or co-applied with the mglu2 PAM LY upon long-duration stimulation of the principal vibrissa A, the top three PSTHs represent responses from a single neurone to long-duration stimulation of the principal vibrissa under control conditions, during application of the Group II agonist, and during recovery. The bottom bar graph represents the mean responses of a group of neurones (n = 21) to long-duration stimulation of the principal vibrissa under the same conditions. B, the top three PSTHs represent responses from a single neurone to long-duration stimulation of the principal vibrissa under control conditions, upon co-application of the Group II agonist and the mglu2 PAM, and during recovery. The bottom bar graph represents the mean responses of a group of neurones (n = 8) to long-duration stimulation of the principal vibrissa under control conditions, upon application of the Group II agonist alone and upon co-application of the Group II agonist and the mglu2 PAM. T: 1000 ms long-duration test stimulation to the principal vibrissa; spike count: the total number of action potentials recorded per 200 ms bin over 18 trials. All PSTHs are constructed using responses from the same cell (CVB038d). Bar graph data are expressed as means ± SEM; P < 0.05.

10 946 C. S. Copeland and others J Physiol stimulation of the principal vibrissa, the Group II antagonist LY was applied as an additional means to reveal the release of endogenous agonist during these sensory stimulation protocols. Iontophoretic ejections of the Group II antagonist significantly reduced the responses of neurones to both short- (91 ± 4% compared to control, n = 12) and long- (86 ± 4% compared to control, n = 7) duration stimuli of the principal vibrissa (Fig. 8). Furthermore, on a subset of the neurones upon which the effects of the mglu2 PAM alone were tested, the Group II antagonist was co-applied with the mglu2 PAM and was found to reverse the mglu2 PAM effect on neuronal responses to short- (LY alone: 127 ± 7% of control; LY plus LY341495: 104 ± 4% of control, n = 7) and long- (LY alone: 136 ± 6% of control; LY plus LY341495: 102 ± 9% of control, n = 7) duration stimuli. This verifies that the mglu2 PAM is exerting its effects via Group II receptors. A B Figure 7. VB neurone responses before, during and after iontophoretic application of the mglu2 PAM LY upon long-duration stimulation of the principal vibrissa The top three PSTHs represent responses from a single neurone to long-duration stimulation of the principal vibrissa under control conditions, during application of the mglu2 PAM, and during recovery. The bottom bar graph represents the mean responses of a group of neurones (n = 22) to long-duration stimulation of the principal vibrissa under the same conditions. T: 1000 ms long-duration test stimulation to the principal vibrissa; spike count: the total number of action potentials recorded per 100 ms bin over 18 trials. All PSTHs are constructed using responses from the same cell (CVB043a). Bar graph data are expressed as means ± SEM; P < Figure 8. VB neurone responses before, during and after iontophoretic application of the Group II antagonist LY upon short- and long-duration stimulation of the principal vibrissa A, the top three PSTHs represent responses from a single neurone to short-duration stimulation of the principal vibrissa under control conditions, during application of the Group II antagonist, and during recovery. The bottom bar graph represents the mean responses of a group of neurones (n = 12) to short-duration stimulation of the principal vibrissa under the same conditions. B, the top three PSTHs represent responses from a single neurone upon to long-duration stimulation of the principal vibrissa under control conditions, during application of the Group II antagonist, and during recovery. The bottom bar graph represents the mean inhibitions of a group of neurones (n = 7) to the condition test protocol under the same conditions. T 1 : 30 ms short-duration test stimulation to the principal vibrissa; T 2 : 500 ms long-duration test stimulation to the principal vibrissa; spike count: the total number of action potentials recorded per bin (20 ms bins in A, 100 ms bins in B) over 6 trials. All PSTHs are constructed using responses from the same cell (CVB085c). Bar graph data are expressed as means ± SEM; P < 0.05.

11 J Physiol mglu2 positive allosteric modulation in the ventrobasal thalamus 947 The effects of the Group II agonist LY and the mglu2 PAM LY on ionotropic glutamate receptor agonist responses As it is known that VB neurone responses to vibrissal stimulation are mediated via AMPA and NMDA receptors (Salt, 1986; 1987), a final experimental protocol was performed in order to investigate whether the Group II agonist and the mglu2 PAM could exert their effects on VB neurone responses via a postsynaptic interaction with NMDA and/or AMPA receptors. AMPA and NMDA were applied iontophoretically and were found to excite VB neurones in a manner concordant with that of previous work from this and other laboratories (Salt & Eaton, 1996; Binns et al. 2003). Iontophoretic ejections of the Group II agonist did not significantly alter the responses of neurones to agonist ejections (AMPA: 108 ± 7% of control; NMDA: 105 ± 7% of control, n = 23); nor did iontophoretic ejections of the mglu2 PAM (AMPA: 102 ± 8% of control; NMDA: 102 ± 6% of control, n = 13). Discussion It has been well established that the Group II mglu receptors can modulate somatosensory transmission in the rat VB (Salt & Eaton, 1995a, b; Saltet al. 1996; Salt & Turner, 1998; Turner & Salt, 2003). However, the relative contributions of the Group II mglu receptor subtypes in modulating sensory transmission are not known, nor to what extent and under what conditions these receptors may be activated upon physiological sensory stimulation. The data obtained in this study using in vivo electrophysiology and iontophoresis in the rat VB clearly demonstrate an mglu2 component to the observed Group II effect on sensory inhibition. Additionally, we have shown that mglu2 is likely to be activated during sensory afferent transmission to VB, which leads to a reduction in inhibition. We speculate this could be occurring via glutamate spillover from the sensory afferent synapse upon physiological sensory stimulation. The rat VB is well known to contain only relay neurones of a relatively homogeneous morphology (Peschanski et al. 1984; Harris, 1986; Ohara & Havton, 1994), but it is possible to categorise neurones functionally into SWE or MWE neurones (Ito, 1988; Brecht & Sakmann, 2002). In this study we recorded from both SWE (74% of cells) and MWE (26% of cells) neurones and did not find any differences in our results between these cell types. This suggests that our findings and interpretation are pertinent across different functional categories of thalamic relay neurones. LY354740, the best-studied selective Group II orthosteric agonist (Schoepp et al. 2003), has been extensively used to probe Group II mglu receptor function in behavioural (Schoeppet al. 2003; Nordquist et al. 2008) and physiological (Flor et al. 2002; Moldrich et al. 2003) responses in both the human and rodent CNS in vivo and in vitro. LY487379, a highly selective mglu2 PAM, which possesses no intrinsic agonist activity but does enhance responses to submaximal glutamate without activity at other receptors or ion channels (Johnson et al. 2003), has been used in behavioural and in vitro electrophysiological studies in the rodent CNS (Schaffhauser et al. 2003; Galici et al. 2005; Poisik et al. 2005; Harich et al. 2007; Hermes & Renaud, 2010; Nikiforuk et al. 2010). The orthosteric antagonist LY has a relatively high selectivity with a nanomolar potency for the Group II mglu receptors, and with submicromolar potencies at all other mglu receptor subtypes (Kingston et al. 1998; Schoepp et al. 1999). However, the iontophoretic parameters used for LY in this study have been demonstrated by ourselves to produce selective antagonism for the Group II mglu receptors (Cirone et al. 2002). The pharmacological specificity of our drug applications is clearly crucial to the interpretation of the results of the present study. Our finding that the Group II antagonist LY reverses the effects of both the Group II agonist and the mglu2 PAM indicates that we are applying the agents in pharmacologically appropriate quantities. Furthermore, application of either the Group II agonist or mglu2 PAM had no effect on responses evoked by NMDA or AMPA, indicating that non-specific effects are not being produced by our drug application protocols. Group II mglu receptor activation facilitates responses to sensory stimulation of the principal vibrissa A potentiation of VB neurone responses to sensory stimulation of the principal vibrissa was observed upon application of the Group II agonist LY VB neurone responses to sensory stimulation are mediated by NMDA and AMPA receptors (Salt, 1986, 1987; Salt & Eaton, 1989), but excitatory responses to these ionotropic agonists were not perturbed upon LY application. Furthermore, in vitro experiments show that LY can directly reduce TRN-evoked inhibitory postsynaptic potentials without affecting postsynaptic membrane properties of VB neurones (Turner & Salt, 2003). Thus the Group II agonist is likely to be potentiating sensory responses by reducing recurrent inhibition, therefore enabling neurones to elicit enhanced responses. Indeed, use of the same sensory stimulation protocol revealed a similar effect on VB neurone responses when GABAergic transmission originating from the TRN was reduced by the GABA antagonist bicuculline (Salt, 1989). Co-application of the mglu2 PAM further potentiated the Group II agonist effect, indicating an mglu2 component to this effect. Interestingly, the percentage increase in VB neurone response to Group II agonist application either alone or in conjunction

12 948 C. S. Copeland and others J Physiol with the mglu2 PAM was significantly greater for the long-duration stimulus compared to the short-duration stimulus. The long-duration stimulus not only provides the VB neurone with a longer duration of excitatory input than the short-duration stimulus, but also evokes a longer recurrent inhibition. When this is reduced by Group II agonist application, either alone or in conjunction with themglu2pam,thevbneuroneisabletorespondtoa greater extent to the longer duration excitatory drive. A particularly novel and important finding of our study is that application of the mglu2 PAM alone potentiated excitatory VB neurone responses to sensory stimulation, an effect that was reversed upon co-application of the Group II antagonist. Additionally, application of the Group II antagonist alone reduced neuronal responses to sensory stimulation. This indicates that the Group II receptors are normally activated in vivo by endogenous ligand upon stimulation of principal vibrissae, and that thereisanmglu2componenttothiseffect.howthismay occur and its physiological relevance are considered in detail below. Group II mglu receptor activation reduces inhibition in VB evoked by sensory stimulation of the secondary vibrissa A reduction in VB neurone inhibition evoked by stimulation of a secondary vibrissa shortly before stimulation of the principal vibrissa was observed upon application of the Group II agonist LY This finding is consistent with previous in vivo studies that have used alternative Group II agonists (e.g. (1S,3R)-ACPD, L-CCG-l, (2R,4R)-APDC) (Salt & Eaton, 1995a, b; Salt et al. 1996; Salt & Turner, 1998). In our present study, the mglu2 PAM potentiated the agonist effect on sensory inhibition, indicating an mglu2 component. mglu2 receptors may be activated by glutamate spillover from the sensory afferent synapse As application of the mglu2 PAM potentiated VB neurone responses to principal vibrissa stimulation, this indicates that mglu2 receptors are normally activated in vivo upon physiological sensory stimulation. However, when stimulation of the principal vibrissa was preceded by stimulation of a secondary vibrissa, application of the PAM alone did not alter the level of inhibition in VB. Although puzzling at first, this outcome provides a possible explanation as to how mglu2 receptors are activated in vivo. As the PAM possesses no intrinsic activity (Johnson et al. 2003), it is possible to speculate that it is potentiating the effects of endogenously released glutamate, which could lead to an increase in response to principal vibrissa stimulation. The source of this glutamate may be from excitatory afferent terminals, which could be sensory or cortical in origin. Ultrastructural studies indicate that there are no axo-axonic contacts onto GABAergic terminals, and cortical afferent terminals are not closely associated with TRN terminals in the rat VB (Ohara & Lieberman, 1993). Therefore, transmission from and receptors associated with cortico-thalamic afferent terminals are unlikely to be involved in this mechanism. However, ultrastructural studies do indicate that sensory afferent terminals are closely located to TRN afferent GABAergic terminals and glial processes on the soma or proximal dendrites of neurones in the rat VB (Ralston, 1983; Ohara & Lieberman, 1993). Glutamate may therefore be spilling over from the sensory afferent synapse to activate extrasynaptic mglu2 receptors leading to a reduction in inhibition with consequent facilitation of responses to sensory stimuli. Similar glutamate spillover has been shown to activate Group II mglu receptors in vitro (Alexander & Godwin, 2005; Chen & Bonham, 2005; Linden et al. 2005; Alexander & Godwin, 2006), making it appropriate to speculate that mglu2 receptors in VB may be activated via this mechanism in vivo. It has been demonstrated that Group II mglu receptors can be localised on glial processes surrounding GABAergic terminals (Liu et al. 1998; Mineff & Valtschanoff, 1999), and mglu3 can be found on TRN-originating GABAergic axons (Tamaru et al. 2001) in the rat VB. Group II mglu receptors have been demonstrated to regulate non-vesicular release of glutamate from glial cells via the cysteine-glutamate transporter in other brain regions (Xi et al. 2002; Moran et al. 2003; Baker et al. 2008). Such locations are therefore ideally suited for these receptors to modulate GABAergic transmission due to their close association with sensory afferent terminals on the soma and proximal dendrites of VB neurones (Ralston, 1983; Ohara & Lieberman, 1993). However, in the alternative stimulation protocol when principal vibrissa stimulation is preceded by stimulation of a secondary vibrissa, the above mechanism could not take place. In this scenario, inhibition in VB via the TRN is provided laterally upon stimulation of the secondary vibrissa (Pinault & Deschenes, 1998; Crabtree, 1999). This postsynaptic inhibition precedes the principal vibrissa excitation; therefore the subsequent release of glutamate from the sensory afferent terminal would be unable to reduce the GABAergic transmission from the TRN by acting at mglu2 receptors, as inhibition of the VB neurone has already taken place. An additional stimulation protocol was devised to further investigate the above hypothesis that mglu2 receptors may be activated by glutamate spillover from the sensory afferent synapse upon sensory stimulation. If such a mechanism does occur, then application of the mglu2 PAM should only influence inhibition if there is direct driving of the VB neurone by its principal vibrissa