HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 92/3/1445    most recent 00079.2004v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (10) Citing Articles via Google Scholar Google Scholar Articles by Manunta, Y. Articles by Edeline, J.-M. Search for Related Content PubMed PubMed Citation Articles by Manunta, Y. Articles by Edeline, J.-M.

Noradrenergic Induction of Selective Plasticity in the Frequency Tuning of Auditory Cortex Neurons

Laboratoire de Neurobiologie de l'Apprentissage, de la Mémoire et de la Communication, Unité Mixte de Recherche 8620, Centre National de la Recherche Scientifique, 91405 Orsay, France

Submitted 26 January 2004; accepted in final form 9 April 2004

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Neuromodulators have long been viewed as permissive factors in experience-induced cortical plasticity, both during development and in adulthood. Experiments performed over the last two decades have reported the potency of acetylcholine to promote changes in functional properties of cortical cells in the auditory, visual, and somatosensory modality. In contrast, very few attempts were made with the monoaminergic systems. The present study evaluates how repeated presentation of brief pulses of noradrenaline (NA) concomitant with presentation of a particular tone frequency changes the frequency tuning curves of auditory cortex neurons determined at 20 dB above threshold. After 100 trials of NA-tone pairing, 28% of the cells (19/67) exhibited selective tuning modifications for the frequency paired with NA. All the selective effects were obtained when the paired frequency was within 1/4 of an octave from the initial best frequency. For these cells, selective decreases were prominent (15/19 cases), and these effects lasted 15 min after pairing. No selective effects were observed under various control conditions: tone alone (n = 10 cells), NA alone (n = 11 cells), pairing with ascorbic acid (n = 6 cells), or with GABA (n = 20 cells). Selective effects were observed when the NA-tone pairing was performed in the presence of propranolol (4/10 cells) but not when it was performed in the presence phentolamine (0/13 cells), suggesting that the effects were mediated by alpha receptors. These results indicate that brief increases in noradrenaline concentration can trigger selective modifications in the tuning curves of cortical neurons that, in most of the cases, go in opposite direction compared with those usually reported with acetylcholine.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Over the last two decades, a large field of research has investigated how neuromodulators regulate or trigger neuronal plasticity in sensory cortex. Historically, developmental plasticity in visual cortex has been the first model where neuromodulators were considered as permissive factors for expression of functional plasticity. After several years of controversy (see Daw et al. 1985), it was established that both acetylcholine (ACh) and noradrenaline (NA) impact visual cortex plasticity and that only combined lesions of these two neuromodulatory systems could reliably block experience-dependent plasticity (Bear and Singer 1986). Subsequent studies have amply documented that neuromodulators such as ACh also promote neuronal plasticity in adult sensory cortices. Both local applications of ACh, or nucleus basalis (NB) stimulation, can induce long-lasting facilitation of evoked responses when paired with presentation of a sensory stimulus (see for reviews Dykes 1997; Gu 2002; Rasmusson 2000; Weinberger 1998). In auditory cortex, selective changes in the neurons' frequency tuning were found for the frequency associated with ACh application or with NB activation (Bakin and Weinberger 1996; Bjordahl et al. 1998; Metherate and Weinberger 1989, 1990). Also, prolonged periods of pairing between a particular sound frequency and NB activation produced cortical map reorganizations in favor of that frequency (Kilgard and Merzenich 1998).

Investigations performed in vitro have also contributed to promote neuromodulators as permissive factors in cortical plasticity. Earlier studies showed that ACh and NA act synergistically to increase the probability that individual cells display long-term potentiation (LTP) (Bröcher et al. 1992; but see Nowicky et al. 1992). In visual cortex slices, both cholinergic agonists (carbachol) and NA can trigger long-term depression (LTD) in conditions of stimulation that do not normally trigger LTD (Kirkwood et al. 1999).

Previous studies examining the noradrenergic modulation of cortical processing have only investigated the immediate consequences of an increased NA concentration. During continuous iontophoretic NA applications, several studies have reported that NA increases the signal-to-noise ratio (Foote et al. 1975; Waterhouse and Woodward 1980; Waterhouse et al. 1981, 1988) and the neuronal selectivity for some dimensions of the stimulus (Manunta and Edeline 1997, 1999; McLean and Waterhouse 1994; but see Ego-Stengel et al. 2002). Using stimulation of the locus coeruleus (LC), some studies found effects similar to those observed with iontophoretic NA application (in particular for the S/N ratio) (see Waterhouse et al. 1998), whereas others revealed a decrease in latency and in latency variability of cortical responses (Lecas 2001, 2004).

To the best of our knowledge, no study has ever tried to determine whether brief, but repeated, associations between a sensory stimulus and an increased NA concentration could promote cortical plasticity. More fundamentally, it has never been tested whether the effects produced by NA application (or by LC stimulation) were selective for the stimulus associated with NA or were general for any stimulus discharging the recorded cortical cell. The present study aimed at determining whether pairing between a particular tone frequency and an increase in NA concentration induces selective plasticity inauditory cortex. The frequency tuning of auditory cortex neurons was determined at suprathreshold intensity before, then several times after 100 pairing trials between a tone and a brief pulse of NA ejection. Controls included cells recorded after 100 trials of tone- or NA-alone presentation and comparisons with the effects obtained in the presence of noradrenergic antagonist and the effects induced by 100 pairing trials with GABA application instead of NA application.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Animal preparation

Experiments were performed on 76 adult Sprague-Dawley rats (300–550 g) anesthetized by an initial injection of urethan (1.5g/kg ip), systematically supplemented with lower doses (0.5 g/kg ip) when vibrissae movements were detected, when reflex movements were observed after pinching the hindpaw , or when the electroencephalogram (EEG) tended to be less synchronized. The body temperature was maintained between 36 and 37°C by a heating pad throughout all the experiment. A local anesthetic (xylocaine, 2%) was infiltrated in the opening made in the scalp. A large opening was made in the temporal bone (from –3.0 to –6.0 posterior to Bregma and 3–4 mm dorsoventral). The dura matter was removed under microscopic control, and the location of the auditory cortex was estimated based on the pattern of vasculature observed in previous studies (Manunta and Edeline 1997, 1998). Three silver balls (400 µm) were placed between the bone and the dura. One was used as reference during the recording session; the two others were used to record the EEG with a large fronto-parietal derivation. A pedestal in dental acrylic cement including two cylindrical threaded tubes was built to allow fixation of the animal's head during the recording session. The stereotaxic frame supporting the animal was placed in a sound-attenuating chamber (IAC, model AC2), and the animal's head was tilted 45° to facilitate electrode penetrations perpendicularly to the cortical surface. The trachea was cannulated, and the cisterna magna was drained to improve the stability of the recordings; the bone opening was filled with warm agar (2% in saline) to prevent pulsations. A rough mapping of the cortical surface was made to localize the primary auditory cortex. For this, low-impedance (<3M) glass pipettes were lowered in the cortex, and neuronal clusters were recorded until a progression from low to high frequency was observed in the caudorostral direction (Doron et al. 2002; Sally and Kelly 1988).

Electrodes and ionophoresis

The recording electrode was a glass pipette (5–15 M) filled with 3 M NaCl solution and glued to a multibarrel ionophoretic electrode (the recording electrode protruding 10–40 µm from the tip of the multibarrel electrode). One barrel was always filled with 1 M NaCl for automatic current balance. The other barrels were filled with the following solutions: NA (arterenol bitartrate, dissolved in ascorbic acid, 0.1 M, pH = 4.5, Sigma); propranolol-HCl (dissolved in ascorbic acid, 0.2 M, pH = 4.5 Sigma); phentolamine mesylate (0.1 M, pH = 3.0–4.0, RBI); and GABA (0.5 M, pH = 3.8, RBI). Drugs were ejected with positive currents using a ionophoretic system (Bionic, France) that included automatic current compensation. The retaining currents were –5/–10 nA depending of the impedance of the ionophoretic electrodes (25–80 M), and the ejection currents were between +40 and +80 nA. While searching for cells, the impedance of the recording pipette and that of the iontophoretic barrels was checked at regular intervals.

Recordings procedure

The signal from the electrode was amplified (gain: 5,000, band-pass: 0.3–10 kHz) and multiplexed in an audio monitor and a voltage window discriminator. The waveform of the action potentials and the corresponding TTL pulses generated by the discriminator were displayed on a digital oscilloscope and digitized (50-kHz sampling rate, Superscope, GW Instruments). The pulses were sent to the acquisition board (PClab, PCL 720) of a laboratory microcomputer, which registered them with a 50-µs resolution. The EEG (band-pass: 1–90 Hz) was sent to a polygraph (Grass model 79D) and was monitored during the neuronal recordings. Successive recording sites were separated by 100 µm in depth.

Auditory stimulus generation

Pure-tone frequencies were generated by a remotely controlled wave analyzer (Hewlett-Packard model HP 8903B) and attenuated by a passive programmable attenuator (Wavetek, P557, maximal attenuation 127dB), both controlled via an IEEE bus. Stimuli were delivered through calibrated earphone (Beyer DT48) mounted in a small (25 mm OD) stainless steel container filled with foam. Calibration of the system was done using a sound-level calibrator, a condenser microphone/preamplifier (Bruel & Kjaer models 4133 and 2639T), and a standard reference tone (1 kHz at 94 dB re 20 µPa) generated by the calibrator (B&K model 4230). A table of frequency versus maximal intensity was stored on the computer, and ascending sequences of 11 isointensity tones were generated from this table. Between 100 Hz and 1.5 kHz, the stepping frequency was 100 Hz; between 1 and 20 kHz, the step was usually 1 kHz (sometimes of 500 Hz); the step was 2 kHz above 20 kHz. This sound-delivery system (the HP 8903B, the attenuators, and the speaker) can delivered tones of 80 dB 20 kHz and of 70 dB 35 kHz. Harmonic distortion products were measured to be down 50 dB from the fundamental.

Experimental protocol

The electrode was lowered perpendicularly to the cortical surface using pure tones as search stimuli. Before starting the protocol, the waveform of the AP was collected for a few minutes to check for its stability, then the frequency range to which the cell responded was evaluated (Fig. 1) .

View larger version (11K): [in this window] [in a new window]   FIG. 1. Schematic representation of the experimental protocol. During phases 1, 3, and 5, the neuron frequency tuning curve was determined at 20 dB above threshold. During phase 2, tones (100-ms duration, 30 trials) of a particular frequency were presented with an intertrial interval (ITI) of 20 s. During phase 4, the tones were presented at the end of a noradrenaline (NA) ejection (1 s). The ITI was also of 20 s. During phase 5, the tuning was tested at regular intervals from immediatly after completion of the pairing 15 min postpairing. Statistical analyses and criteria defined in METHODS were used to determine whether the pairing protocol has induced selective or general modifications of the tuning curve.

PHASE 2. One frequency was arbitrarily selected for the following phases of the protocol. Thirty trials of presentation of this frequency were delivered (tone: 100 ms in duration; intertrial interval: 20 s). To increase the specificity of the activated afferences, the tone intensity was 10 dB below the one used to test the tuning curves. On-line rasters and histograms were displayed by blocks of 10 trials. As our initial goal was to assess if pairing a particular tone frequency with NA application could promote selective re-tuning of cortical cells, the selected frequency was often a frequency adjacent to the one providing the strongest excitatory responses (called the BF). However, to make sure we did not introduce a systematic bias, we decided to select the BF as paired frequency in some experiments (n= 11).

PHASE 3. The frequency tuning was re-tested a second time (control 2) as in phase 1

PHASE 4. One hundred pairing trials between the tone and a drug application were delivered. Each trial lasted 6 s: it involved a pre-ejection period of 4 s, a 1-s period of drug ejection and a 1-s post-ejection period. The tone (100 ms) was presented just at the end of the drug application. As in phase 2, the ITI was 20 s. When the pairing protocol was carried out with a noradrenergic antagonist (phentolamine or propranolol), the ejection of the antagonist started and ended at the beginning and at the end of each trial. This phase lasted 35 min.

For control cells, this phase was replaced by 100 presentations of either a tone alone or a pulse of NA alone with an ITI of 20 s.

PHASE 5. The frequency tuning was re-tested immediately after the end of the 100 pairing trials. Successive tests were then carried out at fixed intervals: 3, 6, 10, and 15 min after the end of the pairing (when 20 repetitions of the tone sequences were used to determine the tuning, the postpairing tests at 3 and 10 min could not be delivered).

Usually, only one pairing protocol was carried out for each recorded cell. However, for 16 cells, a second protocol was performed to test another pharmacological agent (e.g., GABA or ascorbic acid) or to try to block the effect of NA by a noradrenergic antagonist. In 15 cases, when no apparent effect was observed after the first pairing protocol with NA, a second protocol was attempted with a higher current of ejection. For cells tested under control conditions (pairing with GABA, with ascorbic acid, or with NA- or tone-alone presentation), the paired frequency was often a frequency adjacent to the BF, but it was also the BF in one to six cases per conditions.

As the lack of effects after the NA-tone pairing protocol could come from various technical reasons, we decided after completion of this protocol to test the tuning of the cell under continuous (2 min) ejection of NA to make sure that the drug, the multibarrel pipette, and the recording conditions allowed modulation of neuronal responses as previously described (Manunta and Edeline 1997–1999). These tests were performed each time the cells were recorded over extensive periods of time (n = 41 cells).

Data analysis

EFFECTS OBSERVED DURING THE PAIRING TRIALS. Spontaneous activity was analyzed over the 1-s period during which NA was ejected (i.e., before the tone presentation). The evoked response was quantified over the whole tone duration (100 ms). For each cell, a repeated-measures ANOVA was performed (by 10-trials block) to determine if there was an effect induced by NA application over the pairing protocol. When differences were observed (P < 0.05), paired t-tests were used to compare the values obtained during the 30 trials of tone-alone presentation (phase 2) with those obtained during the pairing trials (phase 4). The 30 control trials were compared with the first 30 pairing trials, with the 30 following trials, and with the last 30 pairing trials (trials 61–70 were skipped for this analysis). According to these comparisons, spontaneous and evoked activity were assigned to three categories: decrease (DEC), increase (INC), or no change (NO). Changes classified as increases or decreases included effects that were significant throughout the whole session, transient effects (occurring at the beginning of the pairing), or late effects that only emerged on the last block of 30 pairing trials. We never observed biphasic effects (e.g., decreases at the beginning of the pairing protocol followed by increases at the end). The relative proportions of cells exhibiting increases, decreases, and no change were compared between different pharmacological agents using ² tests.

As in previous iontophoretic experiments (Manunta and Edeline 1997–2000), we tried to classify the cells in three categories: regular spiking, bursting, and fast spiking. This classification is not totally satisfactory because these categories refer to cells behavior in response to intracellularly injected current pulses and in vivo spontaneous activity is often made of a mixture of bursts and of single spikes. "Regular-spiking" cells were defined as cells that almost exclusively fired single action potentials, "bursting" cells were defined as cells that mainly fired bursts of two to five action potentials, and "thin spike" cells were cells with action potentials <0.5 ms. Some cells were not assigned to any category as their spontaneous firing pattern contained equivalent proportions of single spikes and bursts. Interspike interval (ISI) histograms were built for each block of 10 trials. Differences in the ISI distributions between control trials and pairing trials were evaluated using Kolmokorov-Smirnov tests (P < 0.05).

EFFECTS INDUCED BY THE PAIRING IN THE FREQUENCY TUNING. The values of spontaneous and evoked activity were computed from the poststimulus time histograms (PSTHs) obtained in phases 1, 3, and 5. Spontaneous activity was quantified over the 200 ms preceding each tone, and evoked activity was defined by a temporal window that was function of the type of evoked response. For phasic responses, the window was either 0–25 or 0–50 ms after tone onset; it was 0–100 ms for tonic responses.

For each cell, "difference curves" were computed by subtracting the tuning curve obtained at each postpairing test from the one obtained during control 2. The difference curve between controls 1 and 2 was also computed. These difference curves were normalized by dividing each value by the largest one then multiplying by 100. The values of these normalized difference curves were between +100 and –100. A value of +100 indicated that the largest increase occurred at a given frequency; a value of –100 indicated that the largest decrease occurred at a given frequency.

These normalized values were used for a global analysis over the whole population of recorded cells. The percentage of changes obtained at the paired frequency (PF) before pairing (control 1 – control 2) were compared with the percentage of changes obtained after pairing (immediately postpairing – control 2). The percentage of changes obtained immediately postpairing at the PF were also compared with the percentages obtained at all the other frequencies (the nonpaired frequencies).

Individual statistics were performed for each cell. First a global comparison (on the 11 tested frequencies) was performed to determine if the responses obtained in control 2 differ from those obtained in the postpairing frequency tuning curves (using paired t-test and P < 0.05). Second, individual statistics (paired t-test) were made on responses obtained from individual frequency, e.g., at the PF and at the adjacent frequencies. When there was a significant difference for the PF and not for more than two frequencies on each side of the PF, the cell was classified as "frequency-specific effect" (FS effects). When there was a global effect across the 11 tested frequencies or when the analyses revealed significant effect on more than six frequencies, the cell was classified as "general effect." When the cell met neither the criteria for FS effect nor the criteria for general effect, it was classified as "no change."

In all cases, the classification of the effects obtained after the pairing protocol were done blind, i.e., without the knowledge of the effects observed during the pairing trials. The distribution of the different proportions of cells exhibiting FS, general, or no changes were compared between pharmacological agents using ² (P < 0.05).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Long-lasting (>90 min) single-unit recordings were collected between 300 and 1,600 µm below pia; the majority (61% of them) were obtained between 600 and 1,000 µm. For 110 cells, the pairing protocol was completed and the postpairing tuning curves were tested. Most of the cells that could be assigned to a category (see METHODS) were regular-spiking cells (95/104; bursting cells: n = 6; thin spike cells: n = 3). Additional cells were tested before and after 100 presentations of tone alone (n = 10) or after 100 presentations NA alone (n = 11). Spontaneous firing rate ranged from 0.01 to 7.5 spikes/s (mean: 1.33 ± 0.57; mean ± SD); the responses at the BF ranged from 12.2 to 46.5 spikes/s (mean: 20.5 ± 1.7) and the value of the BF ranged from 0.5 to 35 kHz

In presenting the results, we will focus on the changes detected in the frequency tuning curves after the pairing protocol, first based a population analysis and second based on the classification in terms of FS versus general effects. We will then describe the effects observed during the pairing trials to look for potential relationships between the effects detected during and the effects detected after pairing. Next, the changes obtained when the pairing with NA was performed in the presence of antagonists of or adrenoceptors will be presented. Finally, we will describe results obtained when only pulses of NA, or only tones, were presented for 100 trials as well as effects detected when GABA was paired with a tone frequency.

Pairing with NA preferentially attenuated the responses at the PF

POPULATION ANALYSIS. For 67 cells, the pairing protocol was performed with iontophoretic applications of NA. As shown in Fig. 2A, the distributions of percentages of changes at the PF before pairing (control 1 – control 2) and after pairing (immediately post-control 2) significantly differed (P < 0.01). The distributions of changes obtained between controls 1 and 2 did not significantly differ from a normal distribution (² = 1.25, NS), whereas the distributions of the changes obtained after pairing did (P < 0.05). As illustrated by these curves, this difference mainly comes from the high proportion of cells exhibiting strong decreases at the PF and, in particular, from cells exhibiting the largest decreased responses at the PF. After pairing, the distribution of changes at the nonpaired frequencies differed from the one obtained at the PF (P < 0.001), but it did not differ from the one obtained at the PF before pairing (control 1 – control 2, Fig. 2B, ² < 1, NS). Looking at the changes obtained at the PF versus the mean change at the non-PF frequencies (Fig. 2C) revealed that for all but three cells, the decrease in evoked responses was larger for the PF than for the other frequencies: these cells are represented by the dots above the diagonal line in the left part of the figure. Plotting such a scattergram with the changes occurring at the two frequencies adjacent to the PF provided a similar picture (Fig. 2D): among the cells exhibiting decreased evoked responses at the PF, only two showed a stronger decrease at the frequencies adjacent to the PF than at the PF itself.

View larger version (36K): [in this window] [in a new window]   FIG. 2. Distribution of response changes before and after pairing with NA. A: comparison of the cumulative distributions of response changes obtained between controls 2 and 1 and between control 2 and immediately postpairing for the paired frequency. The two distributions significantly differed and the difference mainly came from the high proportion of cells with important decreases at the paired frequency. B: after pairing, the cumulative distribution of changes at the nonpaired frequency did not differ from the distribution of change obtained for the paired frequency between controls 2 and 1. After pairing, the distribution of changes at the nonpaired frequency significantly differed from the distribution of changes obtained for the paired frequency. C: scattergram representing the percentage of changes at the paired frequency (PF) vs. the percentage of changes at nonpaired frequencies (value averaged over all the other frequencies). Note that in the cases of negative values, all but 3 points are above the diagonal line, indicating that in all but 3 cases the decreases at the PF is larger than the decreases at the nonpaired frequencies. D: scattergram representing the percentage of changes at the PF vs. the percentage of changes at the 2 adjacent frequencies. Note that in the cases of negative values, all but 2 points are above the diagonal line indicating except in 2 cases, the decreases at the PF was larger than the decreases at the adjacent frequencies. Note also that in the cases of increases there were twice more points below the diagonal line than above; this indicates that in many cases the increases were larger at the PF than at the adjacent frequencies.

FS EFFECTS WERE MAINLY DECREASES AFTER NA-TONE PAIRING. As indicated in Table 1, 28 cells exhibited decreased responses at the PF, 21 exhibited increased responses at the PF, and 18 showed no effects. Of 28 cells exhibiting significant decreased responses after the pairing protocol, 15 (53%) were classified as FS: they showed the largest decreased responses at the PF without significant decreases at more than two adjacent frequencies. Figure 3 gives a first example of such effects. For this cell, a marked depression of the phasic evoked responses was present in all the postpairing PSTHs at 12 kHz, the frequency paired with NA application. Another illustration of this effect is displayed in Fig. 4B. For this cell, after 100 pairing trials, a selective and long-lasting depression of the evoked responses was detected at 16 kHz in all the postpairing tuning curves. In this case, the effect lasted 45 min postpairing.

View larger version (21K): [in this window] [in a new window]   FIG. 3. Selective decrease in evoked response after pairing. For this cell recorded at 730 µm below pia, evoked responses were observed between 8 and 15 kHz when testing the tuning at 50 dB. Controls 1 and 2 gave virtually identical histograms, and only control 2 is presented here. The cell's best frequency (eliciting the strongest excitatory response) was 11 kHz, and the frequency of 12 kHz (surrounded by a rectangle) was used as PF. Immediately postpairing, the responses at 12 kHz were largely attenuated, whereas the responses at adjacent frequencies were preserved (at 11 kHz) or even facilitated (e.g., at 8, 13, and 15 kHz). The decrease in response at 12 kHz was present in all the postpairing tuning curves, i.e., 15 min postpairing. During the pairing trials, NA (30 nA) slightly attenuated the evoked responses but this effect never reached significance (data not shown). The histograms (5-ms bin) and rasters display the cell's activity during 10 repetitions of a sequence of 11 ascending frequencies. The short bars below the x axis denote the 100-ms tone duration, and only the 100 ms immediately preceding each tone is presented. The numbers below the short bars are the values of the frequencies in kilohertz. Each inset displays the action potential waveform at the different stages of the protocol (30 sweeps, 50-kHz sampling rate. Scale bars: 0.5 mV, 0.5 ms).

View larger version (26K): [in this window] [in a new window]   FIG. 4. Selective decreases occurring at a non-best frequency (A) or at the best frequency (B). For these 2 cells, the graphs compared the frequency tuning curves obtained at different phases of the one protocol with the 2nd control tuning curve (control 2). , the PF. Values are means ± SE. A: this cell is the one the histograms and rasters of which are presented on Fig. 3. The selective decrease at the PF (12 kHz) can be observed in all the postpairing tuning curves. B: during the 2 control tuning curves performed at 40 dB, this regular spiking cell recorded at 930 µm below pia exhibited responses from 13 to 20 kHz. The frequency associated with NA was 16 kHz, which was in fact the neuron's best frequency. In all the frequency tuning curves obtained postpairing, the responses at 16 kHz were largely attenuated, whereas no effects were observed on many frequencies. This effect maintained 45 min postpairing. There was significant decreases at 14 kHz at 3 min postpairing and at 15 kHz at 10–15 min postpairing, but in no case there was a decrease at 2 frequencies adjacent from the PF (16 kHz). During the pairing trials with NA (40 nA), the evoked responses never exhibited significant changes compared with the control responses (data not shown).

View larger version (51K): [in this window] [in a new window]   FIG. 5. Group difference curves for the cells displaying decreased responses after pairing. For each cell, difference tuning curves were computed by substracting the curve obtained at each phase of the protocol from the control 2 tuning curve. These differences were expressed as percentage of the maximal change occurring in the tuning curves and were plotted according to the octave distance from the PF (see METHODS). These percentages of changes were averaged across all the cells exhibiting decreased responses at the PF. , the PF. Values are means ± SE. A: all cells exhibiting decreased responses: no particular effects were detected between controls 1 and 2, but after pairing a decrease centered on the PF was observed. On average, the magnitude of the changes at the PF remained the same over time and the effect was broader around the PF 15 min posttraining compared with immediately postpairing. B: cells exhibiting selective decrease responses: after completion of pairing, there was a prominent decrease centered on the frequency paired with NA application. On average, the magnitude of the changes at the PF remained the same over time, whereas the selectivity of the decrease was greater immediately posttraining compared with 15 min posttraining. Note that although a point adjacent to the PF was significantly decreased 15 min postpairing, the effects did not spread on more that 2 frequencies adjacent from the paired 1. C: cells exhibiting general decrease responses: after completion of pairing, there were decreased responses extending on >1 octave around the frequency paired with NA application. Unexpectedly, the responses tended to recover over time on the low-frequency side, whereas the effect maintained on the high-frequency side.

View larger version (31K): [in this window] [in a new window]   FIG. 6. Examples of selective increases in evoked response. These 2 examples illustrate the effects classified as frequency-specific (FS) increases. For each cell (A and B), the graphs compared the frequency tuning curves obtained after pairing with the control 2 tuning curves. , the PF. Values are means ± SE. A: for this regular spiking cell recorded 840 µm below pia, evoked responses were elicited between 7 and 13 kHz when testing the tuning at 50 dB during control 1. From control 1 to control 2, an attenuation in evoked responses was observed. Immediately postpairing and 6 and 15 min postpairing there was a large and selective increase in evoked responses at the PF (9 kHz). During pairing with NA (40 nA), significant increases in evoked responses were detected from the first (trials 1–30) to the last (trials 71–100) blocks of pairing trials (data not shown). B: for this regular spiking cell recorded 490 µm below pia, evoked responses were elicited between 1 and 11 kHz when testing the tuning at 70 dB during controls 1 and 2. After 100 pairing trials between a 5-kHz tone and NA (60–80 nA), the responses at 5 kHz were the only facilitated responses in the neuron tuning curve. In the following tests (6 and 15 min postpairing) decreased evoked responses were observed for most of the frequencies except at the PF where increases were detected, which suggests that this FS effect developped on a background of attenuated responses (which might be caused by a decreased in cell excitability). These effects were observed 22 min postpairing. During pairing with NA, significant increases in evoked responses were detected at the 1st (trial 1–30) and at the 2nd block of pairing trials (trials 31–71, data not shown). The results obtained for the 2 other cells exhibiting FS increases were similar to those obtained in Fig. 6B, i.e., the increased responses at the PF tended to develop on a background of attenuated responses.

View larger version (21K): [in this window] [in a new window]   FIG. 7. Group difference curves for the cells displaying increases immediately after pairing. A: when all the cell exhibiting significant increased responses were averaged together, a broad and nonselective effect was obtained. B: when only the 4 cells that meet the criterai to be classifed as FS effect were pooled together, the selectivity of the effects emerged. The largest increase was at the PF, and the effects did not extend on more than 1/4 of an octave on both side of the PF. For 2 cells, these effects maintained until the last tests (see legend of Fig. 6), whereas for 2 other cells the effects regressed over time. C: when the cells exhibiting general increased responses were averaged together, a broad and nonselective effect was obtained. On average, these effects were still present 15 min postpairing. Conventions as in Fig. 5. Values are means ± SE. , the PF.

Are there relationships between changes observed during pairing and the postpairing effects?

On average, for the 67 cells tested with NA, there was no significant change in spontaneous or in evoked activity during the pairing trials (paired t-test <1, NS). The mean spontaneous firing rate was 1.3 spikes/s during control trials and was 1.2, 1.5, and 1.4 spike/s at the beginning, middle, and end, respectively, of the pairing protocol. The mean evoked responses at the PF was 10.2 spikes/s during control trials and was 12.5, 11.1, and 10.4 at the beginning, middle, and end, respectively, of the pairing protocol

However, statistical analyses (ANOVA) performed individually for each cell revealed that evoked activity was significantly decreased for 23 cells, was significantly increased for 20 cells, and was unchanged for 24 cells. Several analyses were performed to look for relationships between changes in evoked activity during pairing and the effects detected after pairing in the neurons frequency tuning.

First knowing the effect on the evoked response during pairing (decrease, increase or no change), what was the probability of obtaining a selective or a general effect after pairing? As displayed in Table 2, when evoked activity was increased during the pairing treatment, only a minority of cells (3/20, 15%) exhibited FS effects, and most of them showed general effects (14/20 cells, 70%). When evoked activity was decreased during the pairing trials, an equal number of cells showed FS effects and general effects (10/23 cells, 43.5% in both cases). Second, considering the effects obtained after pairing what was the effect during pairing? As shown in Fig. 8, for cells exhibiting either FS decreases (Fig. 8C) or FS increases (Fig. 8D), the evoked responses did not change during pairing (respectively, P = 0.22 and P = 0.42 for the last blocks of pairing trials). In contrast, for the general effects, the changes during pairing predicted the direction of the postpairing effects: cells exhibiting general decreases (Fig. 8E) exhibited decreases during pairing (P = 0.01 for the last 30 trials); cells exhibiting general increases (Fig. 8F) exhibited significant increases during pairing (P = 0.04 and P = 0.01 for the last 2 blocks of pairing).

View larger version (51K): [in this window] [in a new window]   FIG. 8. Effects observed during pairing as a function of the effects observed postpairing. The histograms represent the average response evoked at the PF during the 30 trials of tone-alone presentation then for 3 blocks of trials during the pairing protocol with NA. On the whole population of cells (n = 67, A), no effect was detected across the pairing trials. There was also no effects for the cells exhibiting a stability of their tuning after pairing (n = 18, B). There was nonsignificant tendencies for cells exhibiting selective decreases (n = 15, C) or selective increases (n = 4; D). In contrast, for cells that exhibited general effects, the direction of the effects observed during pairing predicted the direction of the postpairing effects: cells exhibiting general decreases in their postpairing tuning curves showed decreased responses during the pairing trials (n = 13, E), whereas the cells exhibiting general increases in their postpairing tuning curves showed increased responses during the pairing trials (n = 13, F). *P < 0.05. The scattergrams presented in G and H indicate that there was a significant relationship between the effects obtained during pairing and the effects obtained after pairing for the cells exhibiting decreases (n = 28) or increases (n = 21). However, no relationship were found for cells developping FS effects (increases: P = 0.21; decreases: P = 0.18).

Alpha, not beta, antagonists blocked the FS effects triggered by the pairing protocol

For 13 cells, pairing with NA was performed in presence of phentolamine. No selective FS effects were detected after this protocol (Table 1). For 12/13 cells, the pairing protocol led to general changes in the frequency tuning curves: increases (6/12 cases) and decreases (6/12 cases) were present in the same proportion. For 4/13 of these cells, a second pairing protocol was performed using NA application alone and in one case this led to a FS decrease (2 cells developed general decreases and 1 no change). The distributions of the postpairing effects (FS, general, no effect) observed with NA alone and with NA+phentolamine significantly differed (² = 7.18, P < 0.02).

For 10 cells, pairing with NA was performed in presence of propranolol. After pairing, FS effects were detected in 4/10 cells; in 3/4 cells these effects were FS decreases (Table 1). General changes were obtained for 5/10 cells with a preponderance of decreases (4/5 cells). No effect was obtained for the 10th cell. For 4/10 of these cells, a second pairing was performed using NA application alone, and in one case, this led to a FS decrease (2 other cells developed general effects and 1 no change).

There was no significant differences between the distribution of changes (FS, general, no change) observed under NA alone and under NA+propranolol (² = 2.84, P = 0.24). In contrast, the distribution of changes observed with NA+phentolamine and the one observed with NA+propranolol significantly differed (² = 6.60; P = 0.03). For two cells, successive pairing protocols were carried out: one with NA+propranolol and a second with NA+phentolamine. In the case illustrated on Fig. 9, the first pairing with NA+phentolamine did not produce any particular effect. In contrast, the second pairing with NA+propranolol produced a clear FS decrease in the cell's tuning curve. In the other case, a general decrease was observed after pairing with NA+propranolol, whereas no change was observed after the second pairing with NA+phentolamine.

View larger version (34K): [in this window] [in a new window]   FIG. 9. Illustration of the pharmacological profile for the selective decreases. This regular-spiking cell was recorded at 940 µm below pia and was successively submitted to two different pairing protocols: a first pairing protocol was performed with NA in the presence of phentolamine, then a second pairing protocol was performed with NA in the presence of propanolol. The tuning curve was tested at 50 dB between 10 and 20 kHz. A: after the pairing protocol in presence of phentolamine, no particular effect was observed at the pairing frequency. During this pairing (NA 40 nA/Phentolamine 80 nA), the evoked responses never exhibited significant changes compared with the control responses (data not shown). B: after the pairing protocol in presence of propranolol, selective decreases in evoked responses were obtained at the PF and maintained 15 min postpairing. Note that the decreased evoked responses developped at the PF whereas facilitation of evoked responses were observed for 2 adjacent frequencies, illustrating that selective depression remained specific for the PF. During this pairing (NA 40 nA/propranolol 80 nA), the evoked responses were significantly increased compared with the control responses (data not shown). Values are means ± SE. , the PF.

Effects obtained after NA-alone presentations, tone-alone presentations, or pairing with ascorbic acid

For 11 cells, 100 pulses of NA were delivered alone (ITI 20 s), and the tuning was tested immediately, post 3, 6, and 10 min. Two cells showed general alterations in their tuning (1 a decrease, 1 an increase), the other nine cells did not exhibit any tuning changes. For 10 cells, 100 tone-alone presentations were delivered after control 2, and the tuning was tested at regular interval after this treatment. Applying the criteria used to detect FS effects did not reveal any single case of selective change: two cells exhibited broad (>1 octave) decreased responsiveness around the selected frequency; the other eight cells showed a good stability of their tuning.

As NA was dissolved in a solution of ascorbic acid, this antioxydative agent was always co-ejected with NA during pairing. It was of importance in testing its potential effect because it was shown to produce effects on sensory processing (Mouly et al. 1990). Pairing protocols between a particular frequency and ascorbic acid were completed in six cells: one cell showed a general increase in response during the postpairing tests, whereas the other five cells did not develop any significant modifications of their tuning curves (neither FS nor general effects).

Effects induced by pairing with GABA

As attenuated evoked responses were prominent after pairing with NA, we wondered if such effects could also be obtained after pairing with GABA. For 20 cells, pairing was performed between a tone and GABA ejection. For 11/20 cells, no change was detected in the postpairing tuning curve (Table 1). Surprisingly, general increases were obtained postpairing for 7/20 cells: these effects dissipated for only two cells whereas they lasted 10 min postpairing for the other cells. An example of such an effect is presented in Fig. 10A. Two cases of general decreases were also observed after pairing with GABA.

View larger version (35K): [in this window] [in a new window]   FIG. 10. Effects induced by pairing with GABA. A: individual example: This cell, recorded at 1,010 µm below pia, was tested at 50 dB between 5 and 15 kHz. Initially (control 1), this cell showed robust responses between 8 and 11 kHz (with more modest responses at 14–15 kHz); these responses were slightly increased at the 2nd control tuning curve (control 2, A1). During the entire pairing protocol with GABA (30 nA), the evoked responses were largely suppressed compared with the control responses (A2). However, when the tuning curve was re-tested after the pairing, the evoked responses were largely increased, an effect that lasted 6 min postpairing (A3) ** P < 0.001 and *** P < 0.0001. B: group data: over the whole population of cells tested with GABA, the evoked responses were significantly decreased across the entire protocol (B1). For the 7 cells that exhibited general increases in evoked responses in their tuning curves after pairing, the suppression of the responses by GABA was large and occurred as early as the 1st 30 pairing trials (B2). For cells exhibiting no modification of their tuning curves after pairing with GABA, the effects were only present at the last 2 blocks of pairing trials and the decrease was modest (B3). * P < 0.05, Values are means ± SE. , the PF.

During the pairing trials, the dominant effect of GABA was a decrease in tone-evoked response. Over the 20 cells, there was a significant decrease in evoked activity across the blocks of pairing trials (Fig. 10B1). However, for the 11 cells exhibiting no change postpairing, the inhibitory effect of GABA during pairing were modest, being significant only at the second and third block of pairing trials (Fig. 10B3). In contrast, for the seven cells exhibiting general increases in their postpairing tuning curves, the decrease in evoked responses induced by GABA during pairing was strong and significant at all pairing blocks (Fig. 10B2).

To summarize, after pairing between a tone and GABA ejection, no FS effects were detected. In 35% of the cases, general increases were detected, and, surprisingly, these effects corresponded to cells where GABA was apparently very efficient in decreasing evoked responses during pairing.

Factors contributing to the occurrence of selective effects

A first factor that can explain the occurrence of selective effects is the magnitude of changes during pairing. One can logically assume that the better was the control of the postsynaptic activity during the pairing trials, the higher was the probability of obtaining selective effects. But, as detailed in the preceding text (Are there relationships between changes observed during pairing and the postpairing effects?), FS effects were from cells exhibiting increases, decreases, or no changes during pairing (Table 2), and there was no relationship between the magnitude of the changes during pairing and after pairing. It is unlikely that the selective effects corresponded to neurons from particular layers because FS effects, general effects, and no changes were obtained at similar cortical depth (unpaired t-test < 1 for all comparisons). Also, there was no difference between the depth at which increases (FS and general altogether) and decreases (FS and general altogether) were observed after the pairing protocol. Finally, there was also no difference in the levels of current ejection between FS and general effects, and no difference between increases and decreases (unpaired t-test, t < 1 NS).

We also considered the possibility that the FS effects corresponded to cells for which changes in interspike interval distribution occurred during the pairing trials. Analyzing the interspike interval distribution (see METHODS) revealed that during pairing the firing pattern was significantly changed for only 6/67 cells. Four cells developed general effects after pairing (3 decreases and 1 increase), one developed a FS increase and one no changes. Therefore in most of the cases, NA did not affect the cells firing pattern (see also Manunta and Edeline 2000), and when it did, this did not favor the occurrence of FS effects in the postpairing tests.

Also there was no relationship between the magnitude and type of effects obtained after the pairing protocol and the modulation detected during continuous (2 min) application of NA: in most of the cases (27/41) evoked responses were significantly decreased (increases: n = 3; no effect: n = 11).

Last, we investigated the possibility that the probability of observing a particular effect was a function of the distance between the initial BF and the frequency paired with NA application. Indeed, such a relationship was present (Fig. 11). FS changes occurred only when the PF was at the BF or very close to it (<1/4 of an octave; see examples in Figs. 3, 4, and 6). The highest proportion of FS effects occurred when the BF was selected as PF: 9/11 cases of FS effects were obtained and 6 of them were FS decreases (see in Fig. 11, A and B). General effects and no changes were observed for various distances between PF and initial BF. When quantified using as a metric the 1/16th of an octave, the distance between the PF and the BF was significantly smaller for the FS effects than for the general effects (1.3 vs. 7.9, unpaired t-test P < 0.01) and than for the no changes (1.3 vs. 3.8; P < 0.01). This result was corroborated by computing the ratio between the responses strength at the PF and at the initial BF. This ratio was higher for the FS effects (0.86) than for the general effects (0.71) and for the no changes (0.65; unpaired t-test, P < 0.02 for both comparisons with the FS effects). This indicates that FS effects were obtained when the strength of the responses at the PF almost reached the level of responses at the BF. As FS effects were mostly decreases (15/19 cases), it was difficult to evaluate whether the direction of the FS effects could be predicted. However, one can note that the few cases of FS increases were found over the same range of distance than the FS decreases (Fig. 11B). The distribution of these two effects did not differ (² = 1.75, P = 0.62).

View larger version (24K): [in this window] [in a new window]   FIG. 11. Types of effects observed according to the distance between the PF and the initial BF. For each cell, the distance between the PF and the initial BF was calculated using as metric the 1/16th of an octave. A: the distribution showed that all the cases of FS effects were obtained when the PF was <4/16th of an octave (i.e., 1/4th of an octave) from the initial BF. Within this distance, the FS effects represent 50% (19/38) of the cases. General and no effects were obtained for all distances between the PF and the initial BF. B: within the distance of 1/4 of an octave where FS effects emerged, decreases and increases were observed at similar distances.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Repeated pairing between a particular tone frequency and a brief pulse of NA produced selective modifications in the tuning curves of 28% of auditory cortex neurons and general modifications for 45% of the cells. In most of the cases (79%), the selective modifications were specific decreases which lasted 15 min after pairing (45 min in some cases). These selective effects, blocked by phentolamine, were observed only when the distance between the PF and the initial BF was 1/4 of an octave.

Methodological considerations

Several technical aspects need to be discussed before considering the functional implications of our results. First one can wonder whether brief iontophoretic pulses of NA are efficient to produce phasic increases in NA concentration at the vicinity of the recorded cell. The use of a constant "retaining current" (5/10 nA) could deplete the tip of the pipette in NA or can delay the time required for NA to be ejected in the extracellular space (Armstrong-James and Fox 1983). However, several factors have probably counteracted this possibility. First, as in most of the in vivo iontophoretic experiments, the impedances were measured at regular intervals (20 min) to verify that the pipettes were not clogged, and such tests required to stop the retaining currents for a few seconds. Second, after pairing, many cells were tested with continuous (2 min) NA ejections to make sure that the recording conditions were appropriate to modulate the cells' tuning curve. Finally, a validation of our procedure to reliably eject a drug in the extracellular space came from the results obtained with GABA ejections: during pairing decreased responses were observed in 17/20 cells.

Second, because selective effects were observed for only 28% of the cases, one can wonder whether such effects might have been observed by chance. The answer comes from the lack of any FS effects in various situations: of 59 protocols run with tone alone presentation (n = 10), NA alone presentation (n = 11), pairing with an alpha antagonist (n = 13), ascorbic acid (n = 5), or GABA application (n = 20), no case of FS effects was detected. Actually, two factors might have contributed to lower the number of selective cases reported in the present study. First, a change was classified as FS if the effects did not extend on more than two frequencies adjacent to the PF; this could have been too restrictive and could led to classify as general effect some potential FS effects. Second, and more importantly, the distance between the PF and the initial BF was crucial for the occurrence of FS effects: they emerged only if the PF-BF distance was <1/4 of an octave. Therefore one can envision that FS effects and general effects stem from a continuum of effects that do, or do not, meet the criteria to be classified as FS.

Last, even if care was taken during each recording sessions to collect the data in similar states of synchronized EEGs, one can suspect that subtle changes in neuronal excitability have generated global modifications in frequency tuning after pairing with NA (45% of the cases). However, the probability of observing general effects was lower after pairing with ascorbic acid (20% of the cases), tone-alone, or NA alone protocols (20 and 18% of the case, respectively). The lower proportion of general effects in these conditions suggests that part of the general effects observed after NA pairing are physiological consequences of repeated tone-NA pairing which did not met the criteria to be classified as FS.

Comparisons with selective effects obtained with other neuromodulators during pairing protocols

Over the last 15 years, experiments have reported that repeated pairing between neuromodulators and sensory stimuli can produce functional plasticity in sensory cortex. In visual cortex Greuel et al. (1988) evaluated whether pairing performed with neuromodulators (acetylcholine and/or NA) induced effects similar to those triggered by applications of excitatory amino acids (EAAs). Neuromodulators were as efficient as EAAs (47 vs. 54%) to modulate response properties; co-applications of EAAs and neuromodulators only slightly increased the percentage of changes in orientation selectivity (62%). Subsequent studies have documented that ACh can trigger selective changes in sensory processing. In auditory cortex, pairing between a tone and ACh led to FS effects in 29% of the cases (Metherate and Weinberger 1989, 1990). In the barrel cortex, the tuning to the temporal frequency of whisker deflection was selectively modified in favor of the frequency associated with ACh for 33% of the neurons (Ego-Stengel et al. 2001; Shulz et al. 2000). Note that the direction of the effects differed between these studies: decreased evoked responses dominated in auditory cortex; facilitations dominated in the barrel cortex. That the postpairing tests were performed either in absence (Metherate and Weinberger 1989, 1990) or in presence of ACh (Shulz et al. 2000) might explain this discrepancy.

Considering the effects induced by repeated pairing between a particular stimulus and activation of the source nuclei of neuromodulatory systems, there are both similarities and discrepancies with the present data. When a sound frequency was associated with activation of the cholinergic or the dopaminergic system, selective map expansions were obtained in favor of the PF (Bao et al. 2001; Kilgard and Merzenich 1998). In other words, the similarity with our data are the input specificity. However, in contrast with the present results, map expansions most likely rely on increased responses for the PF as it has been demonstrated when the tuning curves were tested after pairing involving activation of NB neurons (Bakin and Weinberger 1996; Bjordahl et al. 1998; Dimyan and Weinberger 1999).

To the best of our knowledge, the consequences of repeated pairing between NA and a particular sensory stimulus have not yet been investigated. Up until now, all the studies performed with NA have evaluated the functional properties of cortical cells under continuous applications of NA, i.e., without explicit pairing between a particular stimulus and NA applications.¹The situations the closest to our study, are those where LC stimulation was concomitant with presentation of a stimulus. With this protocol, LC activation facilitated both the excitatory phasic component and the postexcitation inhibitory phase of the response (Lecas 2001, 2004; Waterhouse et al. 1998). As only a single stimulus was tested in these experiments, the consequences of such pairing on receptive field properties remained unexplored.

Selective effects could not be predicted by the changes during pairing trials

In Hebbian models of neuronal plasticity, changes in temporal correlation between pre- and postsynaptic activity is the key element to trigger the activation of intra-cellular events. In turn, these intra-cellular events promote increase or decrease in synaptic transmission. According to this scheme, any treatment acting on the temporal correlation between pre- and postsynaptic activity can affect the efficacy of synaptic transmission. Initially, testing this hypothesis in sensory cortex has been achieved by paired presentations between a sensory stimulus and an increase (or a decrease) in postsynaptic activity (Frégnac et al. 1988). Applied in adult sensory cortex, this protocol produced durable changes in the direction predicted by Hebbian rules for 30% of the tested cells (Cruikshank and Weinberger 1996a; McLean and Palmer 1998; Shulz and Frégnac 1992).

Here and in other studies (Metherate and Weinberger 1990; Shulz et al. 2004), no relationship was found between the effects observed during the pairing trials and the effects observed after pairing. Several factors can explain this lack of relationship. For example, it is difficult to estimate to which extent the extracellular application of a pharmacological agent really acts on the temporal correlation between pre and postsynaptic activity. Obviously, the effects observed during pairing reflect the drug action on the postsynaptic cell, the presynaptic terminals, as well as on the local network, including on interneurons which, in some cortical areas, are strongly depolarized by NA (Kawaguchi and Shindou 1998; Sessler et al. 1995). Thus the net effect observed during pairing most likely results from a mixture of direct and indirect physiological effects from which it is unwise to infer the actual level of correlation between pre- and postsynaptic activity.

In the present data, relationships were found after NA application only in the case of general modifications of the neurons' frequency tuning (Fig. 8). However, the results obtained with GABA contrast with these relations: when GABA produced the strongest decreases in evoked responses during pairing, general increases in evoked responses were observed after pairing.²Although one can view this result as a clue that anti-Hebbian rules of plasticity (see for review Bell 2002) operate in auditory cortex, it might simply denote that even if the dominant effect of NA in auditory cortex is inhibitory (Manunta and Edeline 1997–1999), the noradrenergic modulation cannot be reduced to effects mediated by GABAergic interneurons (see also Scheiderer et al. 2004). Finally, one might suspect that the induction of neuronal plasticity during a pairing protocol between a sensory stimulus and neuromodulators relies on more complex cellular mechanisms than simple changes in temporal correlation. An alternative view could be a direct action of neuromodulators on the cascade of intra-cellular events triggered by the changes of temporal correlation (Ahissar et al. 1996; Cruikshank and Weinberger 1996b; Shulz et al. 2004).

Mechanisms of FS effects: a reflection of the pharmacological profile of NA in auditory cortex?

In several cortical areas, NA decreases synaptic transmission, affecting both the NMDA and non-NMDA component of excitatory postsynaptic potentials by acting on alpha1 (Law-Tho et al. 1993) or on alpha2 adrenoceptors (Pralong and Magistretti 1994, 1995). This weakened synaptic transmission can explain the decreased responses described in the visual (Ego-Stengel et al. 2002) and auditory cortex (Manunta and Edeline 1997, 1999) during iontophoretic application of NA. But does this effect explain the plasticity occurring after repeated pairing with NA?

Several in vitro studies have investigated the potency of NA to promote durable changes in synaptic transmission, particularly in hippocampus, but also in sensory cortices. Using protocols aimed at triggering LTP, NA favored the probability of LTP induction in visual cortex (Bröcher et al. 1992). More recently, NA was found to promote the occurrence of LTD in conditions of paired-pulse stimulation that do not normally promote LTD (Kirkwood et al. 1999). This result shares some properties with the present findings: in both cases, NA produces depression of evoked responses in conditions that do not normally produce significant changes, and in both cases, the effects are mediated by alpha receptors. Therefore the FS effects observed in our experiment might result from changes in synaptic efficacy that operate at thalamo-cortical synapses. An unresolved question concerns the fact that both FS decreases and FS increases were obtained in our experiment. Several explanations can be proposed. The direction of the effect could depend on the local NA concentration and/or on the level of membrane potential during pairing. For example, it was shown that, although dopamine (DA) normally favors the occurrence of LTD over LTP in frontal cortex (Law-Tho et al. 1995; Otani et al. 1998, 1999), a brief application of DA performed 40 min before tetanic stimulation leads to the opposite results, i.e., to the systematic occurrence of LTP (Blond et al. 2002). Alternatively, the direction of the changes might be a function of the ratio between alpha and beta receptors on the postsynaptic cells or on the ratio between pre- and postsynaptic receptors.

That FS effects occurred only when the PF was close to the initial BF indicates that NA acts differentially on strong versus weak inputs: if a strong input is activated while NA is present, this input is selectively affected by NA, but if a weak input is activated, no change or general change are observed. It is also possible that selective effects developed only when a particular configuration and/or distance was present between the locus of NA ejection and the locus of activated inputs on the dendriti