INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The dynamics of synaptic transmission largely depend on the mechanisms of short-term neuronal plasticity. During the repetitive activity of a presynaptic cell, the strength of synaptic responses in the target neuron can be either depressed or facilitated. The characteristics of short-term synaptic plasticity of a given synapse mainly depend on presynaptic mechanisms of neurotransmitter release (Rozov et al. 2001a; see Zucker 1999 for a review), but postsynaptic mechanisms may also be implicated (Jones and Westbrook 1996; Rozov and Burnashev 1999; Rozov et al. 2001b; Thomson 1997).

During postnatal development, characteristics of short-term synaptic plasticity studied at unitary connections can be modified probably due to changes of the presynaptic release properties (Angulo et al. 1999a; Bolshakov and Siegelbaum 1995; Pouzat and Hestrin 1997; Reyes and Sakmann 1999). In the postnatal (Simons and Land 1987) as well as in the adult neocortex (Fox 2002), the topographic representation of sensory stimuli show experience-dependent modifications and changes in short-term synaptic plasticity appear as a possible mechanism controlling the consecutive cortical map reorganization (Finnerty et al. 1999; see Buonomano and Merzenich 1998 for a review). A precise description of synaptic interactions between different types of neocortical neurons should help to understand how adult local circuits may govern such phenomena. We reported the paired-pulse characteristics of the unitary excitatory connections formed by pyramidal cells onto a prominent type of GABAergic interneurons, called fast-spiking (FS) cells (Angulo et al. 1999a). FS cells constitute a relatively homogeneous population of interneurons that fire fast action potentials at high nonaccommodating frequencies (Angulo et al. 1999a; Cauli et al. 1997, 2000; McCormick et al. 1985), express the calcium-binding protein parvalbumin (Cauli et al. 1997, 2000; Kawaguchi and Kubota 1993), and show multiple axo-axonic (chandelier cells) or axo-somatic (basket cells) contacts on pyramidal cells (Kawaguchi and Kubota 1993, 1998). We showed that, in 5-wk-old rats, two excitatory inputs from different pyramidal cells onto a single FS interneuron exhibited paired-pulse responses displaying either facilitation (PPF) or depression (PPD). This observation suggests that the mechanisms regulating the paired-pulse characteristics at pyramidal-FS connections are mainly presynaptic and led us to hypothesize that distinct types of pyramidal cells form synapses with FS interneurons that differ in their paired-pulse behaviors. Indeed, different classes of pyramidal cells acquire the morphological and electrophysiological characteristics of adult animals after the third postnatal week (Franceschetti et al. 1998; Kasper et al. 1994; Petit et al. 1988). In spite of this, the modulation of the synaptic efficacy between specific pyramidal cell classes and inhibitory interneurons is still poorly understood in intracortical circuits of mature animals.

In the present work, we focused our attention on possible functional and structural differences of layer V pyramidal cells of the motor cortex displaying either PPD or PPF at their synapses with FS interneurons. Our results showed that in 28- to 52-day-old rats, connections displaying PPD always showed multiple functional release sites, whereas those displaying PPF had either single or multiple release sites. In addition, two types of morphologically different pyramidal cells, distinguished mainly by the arborization of their apical dendritic tree, had different paired-pulse characteristics at pyramidal-FS connections. Our data indicate that two distinct types of pyramidal cells contact FS interneurons and form connections with different short-term synaptic characteristics in neocortical circuits of young adult rats.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Paired recordings

Parasagittal sections (300 µm thick) were prepared from the motor cortex of Wistar rats (34 ± 6 postnatal day-old; range, 28-52) as previously described (Angulo et al. 1999a) and in accordance with French regulations on animal care. For recordings, slices were transferred to a chamber and perfused with a physiological extracellular saline solution containing (in mM) 121.0 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 3 CaCl₂, 1 MgCl₂, 26 NaHCO₃, 20 glucose, and 5 pyruvate and oxygenated with a mixture of 95% O₂-5% CO₂ at 30-33°C.

Pyramidal-FS connections were examined by paired recordings as previously described (Angulo et al. 1999a). Briefly, postsynaptic FS interneurons in layer V of the motor cortex were recorded with patch pipettes (resistance, 3-5 M) filled with an internal solution containing (in mM) 134 K-gluconate, 10 phosphocreatine, 10 HEPES, 2 Na₂-ATP, 2 Na-GTP, and 4 mg/ml biocytin (pH 7.2-7.4, 300 mosmol). To avoid the effect of polyamines on calcium permeable AMPA receptors during short-term synaptic plasticity, we did not include spermine into the patch pipette (see Rozov and Burnashev 1999). We previously reported, however, that both PPD and PPF could be recorded at pyramidal-FS connections of young adult rats when spermine was included in the internal solution (Angulo et al. 1999a). Postsynaptic FS cells were initially visualized using videomicroscopy with Nomarski optics under infrared illumination (Stuart et al. 1993) and later identified by the characteristics of their action potential discharges according to the procedure reported by Cauli et al. (1997). Six recorded FS interneurons were successfully stained by biocytin injection and further reconstructed for morphological analyses (see RESULTS). Only FS neurons with membrane potentials more negative than -60 mV were included in the sample. All membrane potentials were corrected for a junction potential of -12 mV.

After whole cell recordings were established, presynaptic neurons were impaled with a sharp microelectrode filled with 1.5 M KCl and 2% biocytin (resistance, 60-120 M). Pyramidal cells were initially identified by their characteristic action potential firing induced by depolarizing current pulses (Cauli et al. 1997; Connors and Gutnick 1990; McCormick et al. 1985). The identification of the presynaptic cells was further confirmed by their characteristic features obtained by biocytin labeling (see Morphology). The impaled presynaptic cells were iontophoretically injected with biocytin by applying depolarizing current pulses at the end of the experiment (2 nA, 5-15 min).

Whole cell voltage-clamp recordings of postsynaptic FS cells were obtained using a patch-clamp amplifier (Axopatch 200A, Axon Instruments) and filtered at 2 kHz. Series resistances between 10 and 25 M were monitored throughout the experiments, but they were not compensated. Intracellular current-clamp recordings of presynaptic pyramidal cells were obtained using an intracellular amplifier (Neuro Data, Instruments). Digitized data were acquired and analyzed off-line using Acquis1 software (Gérard Sadoc, CNRS, Gif-sur-Yvette, France).

The firing properties of the presynaptic and postsynaptic cells were studied as previously described (Cauli et al. 1997, 2000). In particular, the frequency adaptation parameters of pyramidal cells were measured on discharges elicited by application of 300-ms depolarizing current pulses. The instantaneous discharge frequencies between the first two spikes (f_initial), 20 ms after the beginning of the discharge (f₂₀), and at the end of the stimulation (f_final) were determined for the voltage trace showing five to eight action potentials. Early and late phases of the frequency adaptation were calculated according to (f_initial  f₂₀)/f_initial and (f₂₀  f_final)/f_initial, respectively.

Unitary EPSCs were induced in FS cells by triggering action potentials in presynaptic pyramidal cells with depolarizing pulses (3 ms, 1.5 nA). The stability of the recordings was first assessed during the time course of the experiment by plotting the amplitudes of individual synaptic responses against time. We then averaged responses by groups of 20 consecutive sweeps and only connections showing <10% of variation in the amplitude of these mean EPSCs were further considered for analysis. In our experimental conditions, a run down of the synaptic responses was rarely observed and the mean peak amplitude and the probability of response of the EPSCs could remain stable for 3 h of recording.

To examine the paired-pulse response characteristics of the EPSCs, pairs of action potentials were elicited in presynaptic cells by applying two short depolarizing current pulses separated by 50 ms at a stimulation rate of 0.2 Hz. Means of elicited synaptic responses were obtained by averaging the traces after presynaptic action potentials were aligned using automatic peak detection. The mean peak amplitudes of the first EPSC (EPSC1) and second EPSC (EPSC2) including failures were measured from the baseline preceding each EPSC, and the paired-pulse responses were determined by calculating the amplitude ratio of EPSC2/EPSC1. Assuming an error of 5% in the mean EPSC amplitude measurements, the connections showing neither depression nor facilitation had a paired-pulse ratio between 0.95 and 1.05.

At each connection, the amplitude of individual responses was estimated by measuring the average current of a 0.5- to 1-ms window at the peak of the EPSC. The noise level was estimated using the same method in portions of sweeps devoid of synaptic responses. The number of failures was determined from the amplitude distribution histograms in cases in which there was no overlap between the distributions of the EPSCs and of the noise. When a minimal overlap between the two distributions occurred, we used the method described by Liao et al. (1995) to estimate the number of failures. Briefly, we doubled the number of sweeps in which the amplitude of the response was >0 pA to provide the number of failures.

The quantal size of connections showing multiple functional release sites was obtained by decreasing the release probability of presynaptic cells either by decreasing [Ca²⁺]_e from 3 to 0.5 mM or by applying 20-50 µM Cd²⁺ in the bath. In these experiments, the paired-pulse responses were first characterized with a stimulation rate of 0.2 Hz, and the effect of Cd²⁺ or low [Ca²⁺]_e was then studied on paired-pulse responses elicited at a stimulation rate of 1 Hz to sample more responses.

The statistical significance of the difference between means was computed with the nonparametrical Mann-Whitney U test for unpaired samples and with a Wilcoxon U test for paired samples. For data presenting too many ties, the statistical significance of the means was analyzed with a Student's t-test. Statistical data are given as means ± SD, unless otherwise indicated.

Morphology

Morphological analyses were performed as previously described (Angulo et al. 1999a). Briefly, slices containing biocytin-filled cells were fixed overnight in 4% paraformaldehyde and 0.2% glutaraldehyde in 0.1 M phosphate buffer (PB) at 4°C. Then, they were rinsed extensively with PB, including an intermediate step with 1% H₂O₂ (in PB) to block endogenous peroxidase activity. Next they were incubated in a cryoprotectant (25% saccharose, 10% glycerol in 0.01 M PB) for 1 h. Then, the slices were freeze-thawed three times over liquid nitrogen. After three rinses in PB, the slices were incubated overnight with ABC (1:200, Vector) at 4°C. Thereafter, 1 mg/ml 3, 3' diaminobenzidine (DAB, Sigma) was preincubated for 10 min and then the peroxidase was revealed by starting the reaction with 0.01% H₂O₂. The reaction was stopped by rinsing with PB. The slices were intensified with 1% OsO₄ (in PB) for 1 h and dehydrated in an ascending series of ethanol (including contrasting with 1% uranyl acetate in 70% ethanol for 45 min). After immersion in propylene oxide, the sections were flat-embedded in Durcupan ACM (Fluka, Buchs, Switzerland) and cured for 16 h at 60°C.

The slices were examined with a light microscope (NIKON Eclipse 800; Ratingen, Germany) connected to the computerized reconstruction system Neurolucida (Microbrightfield, Colchester, VT). The soma and the dendrites of pyramidal cells which appeared to be sufficiently well labeled were three-dimensionally and quantitatively reconstructed with a ×40 objective. Numeric analyses of the reconstructions were performed with Neuroexplorer (Microbrightfield, Colchester, VT), and data collected in a spread sheet were statistically tested for differences between pyramidal neurons displaying PPD and PPF on their target interneurons. Parameters suitable to characterize the morphological complexity of the reconstructed neurons were calculated and statistically evaluated (see Table 1). These included, among others, the number of primary dendrites, the total number of branches and the highest order (indicating the topological branching complexity of the dendritic trees) as well as the volume taken by the dendroplasm (indicating the overall thickness of the dendrites in relation to a certain length). Some neurons were digitally photographed by using the capability of Neurolucida to acquire Z stacks. These stacks were collapsed into an "extended focus-view" by the minimum projection of Metaview 4.0 (Universal Imaging; West Chester, PA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We studied the paired-pulse characteristics of 52 pyramidal cell to FS interneuron connections in layer V of the motor cortex by using paired recordings in acute neocortical slices of young adult rats. Presynaptic pyramidal cells were impaled with sharp intracellular electrodes, and unitary EPSCs were recorded in FS interneurons with a patch pipette.

Identification of pyramidal cells and FS interneurons in neocortical slices

The identity of the neurons was first determined by the characteristic firing pattern of these two different neocortical cell types as previously described (Angulo et al. 1999a,b). Pyramidal cells were mainly characterized by a strong firing accommodation during depolarizing current pulses (Cauli et al. 1997, 2000; Connors and Gutnick 1990) (see also Fig. 5 and the morphological analysis of the recorded pyramidal neurons at the end of RESULTS). FS interneurons were characterized by a low input resistance (Fig. 1A1) (Angulo et al. 1999a) and by their ability to discharge clusters of nonaccommodating action potentials in response to near-threshold depolarizing current pulses (Fig. 1A1) (Cauli et al. 1997). At higher stimulation intensities, they exhibited continuous discharges with no frequency adaptation, even at rates >100 Hz (Fig. 1A1). Although only a limited number of FS cells were labeled during the present study (n = 6), their morphological characteristics resembled those of FS cells described in our previous reports (Ali et al. 2001; Angulo et al. 1999a). Briefly, these cells were characterized by a round or an ovoid somata from which four to seven smooth, partially beaded primary dendrites originated (Fig. 1, A2 and B). The axon was emitted toward the pia and branched soon and repeatedly, providing a dense field of boutons, typical of extended local plexus cells that are regarded to be putative basket cells (Fig. 1, A2 and B) (Wang et al. 2002). In four of six cases, the axon clearly showed typical basket-like features forming a pericellular basket around unstained cells (Fig. 1, A2, inset, and B). These observations confirmed previous findings indicating that layer V FS cells are interneurons (basket or chandelier cells) targeting mainly the somatic region of their targets (Ali et al. 2001; Angulo et al. 1999a; Kawaguchi and Kubota 1993; Tamas et al. 1997). Despite the large diversity of neocortical interneurons, FS cells seem to constitute a relatively homogenous population of interneurons that co-express the enzyme glutamic acid decarboxylase and the calcium-binding protein parvalbumin (Cauli et al. 1997, 2000; Kawaguchi and Kubota 1993), establish electrical networks exclusively among each others through gap-junctions (Galarreta and Hestrin 1999; Gibson et al. 1999), and control the output of the neocortex by inhibiting pyramidal neurons (Ali et al. 2001; Galarreta and Hestrin 1998, 1999; Gibson et al. 1999; Tamas et al. 1997; Thomson et al. 1996).

View larger version (59K): [in this window] [in a new window]   Fig. 1. Firing properties and morphology of neocortical FS interneurons. A1: current-clamp recordings obtained in fast-spiking (FS) interneurons from a 31-day-old rat during injection of current pulses (bottom). The input resistance of this cell was 160 M, calculated from the response to a hyperpolarizing pulse of 200 pA. Note the silent periods in the response to a near-threshold stimulation and the high-frequency discharge at a higher depolarization. A2: high-power reconstruction of a small multipolar interneuron (soma and dendrites in gray) with an extensive local axonal arbor (in black). The small rectangle marks the location of the cell shown in inset. Inset: 5 boutons (3 marked by arrowheads) in close apposition to a slightly background-stained pyramidal cell soma (arrow), in accord with the classification of putative basket cell. The electrophysiology of this interneuron is shown in A1. Bars, 10 and 100 µm. B: minimum-intensity projection (covering 60 µm in z) of another small multipolar interneuron with thin beaded dendrites in layer V. Parts of the axon form pericellular arrangements around somata (asterisks). Bars, 25 µm.

Paired-pulse characteristics of pyramidal-FS connections in the neocortex of young adult rats

The paired-pulse characteristics of pyramidal-FS connections were analyzed by recording unitary EPSCs in FS interneurons when two presynaptic action potentials separated by 50 ms were elicited in pyramidal cells at a stimulation rate of 0.2 Hz (Fig. 2, A and B). The paired-pulse behavior of unitary EPSCs was determined by calculating the amplitude ratio EPSC2/EPSC1 from an average of 40 individual responses, including failures. Figure 2, A and B, illustrates two pyramidal-FS connections with different paired-pulse behaviors, one displaying PPF with a paired-pulse ratio of 1.4 (Fig. 2A; the morphology of the FS cell is shown in Fig. 1B) and the other displaying PPD with a paired-pulse ratio of 0.7 (Fig. 2B). We also averaged groups of 20 consecutive responses to calculate the paired-pulse ratio at different time points of the experiment and did not observe any switch from facilitation to depression or vice versa (data not shown).

View larger version (27K): [in this window] [in a new window]   Fig. 2. Paired-pulse response characteristics at neocortical pyramidal-FS connections of young adult rats. A and B: paired-pulse responses elicited with a presynaptic interspike interval of 50 ms (top) at a stimulation rate of 0.2 Hz in FS interneurons from 28 (A)- and 33 (B)-day-old rats. Bottom: traces correspond to the mean amplitudes, including failures, of the postsynaptic currents. The response probabilities of the 1st and 2nd excitatory postsynaptic currents (EPSC1 and EPSC2) were, respectively, 0.73 and 0.84 for the connection in A and 1 for the connection in B. Note that in A the connection showed a paired-pusle faciliation (PPF), whereas in B the connection showed a paired-pulse depression (PPD). Traces are an average of 80 responses. The morphology of the FS cell shown in A is presented in Fig. 1B. C: plot of the paired-pulse ratios of all individual connections as a function of age. Note the large variability of these ratios. D: average of the amplitudes, including and excluding failures, and of the probability of response at the 1st presynaptic action potential for facilitating (), depressing () and nondepressing, nonfacilitating () connections (n = 52). Statistical differences between these three types of connections are shown by lines (*: P < 0.05; **: P < 0.02, and ***: P < 0.01; Mann-Whitney test). PPR means paired-pulse ratio. Error bars correspond to SE.

In our sample, the paired-pulse ratio obtained at connections from 28- to 36-day-old rats was widely distributed with values ranging from 0.42 to 2.34 (Fig. 2C). Because we previously observed that facilitating pyramidal-FS connections appeared only after the third postnatal week (Angulo et al. 1999a), the wide range of the paired-pulse ratios observed in the present study could be explained by the fact that the fifth postnatal week still corresponds to a transitional developmental phase of the synaptic properties. We thus tested the paired-pulse responses of seven connections obtained from 46- to 52 day-old rats and observed both facilitating and depressing synapses. The amplitude EPSC ratio varied from 0.84 to 1.34 in these older animals (Fig. 2C), indicating that at this later stage of the postnatal development, pyramidal-FS connections can still display different paired-pulse response characteristics.

Most of the pyramidal-FS connections clearly displayed either PPF or PPD (Fig. 2C). In our sample, 48% of the connections displayed a PPF with a mean paired-pulse ratio of 1.37 ± 0.29 (n = 25) and 38.5% displayed a PPD with a mean paired-pulse ratio of 0.76 ± 0.12 (n = 20). Four pairs from 28- to 36-day-old rats and three pairs from 46- to 52-day-old rats (13.5%) had a paired-pulse ratio between 0.95 and 1.05 and were considered as nondepressing, nonfacilitating connections (mean paired-pulse ratio of 1 ± 0.04; see METHODS).

The amplitude including failures, the amplitude excluding failures (potency) and the probability of response of EPSC1 at these nondepressing, nonfacilitating connections were equivalent to those at connections displaying PPF (Fig. 2D). On the contrary, the amplitude, including failures, and the probability of response at nondepressing, nonfacilitating connections were significantly lower than those at connections displaying PPD (Fig. 2D), suggesting that functional properties differ at least between these two types of connections. We did not observe significant differences of the amplitude, the potency, and the probability of response of EPSC2 between the three types of connections (mean values: 46.5 ± 50, 60 ± 48.7, 0.86 ± 0.19, respectively; n = 52). Considering the low number of nondepressing, nonfacilitating connections in our sample, only connections displaying either facilitation or depression were characterized hereafter (but see DISCUSSION).

The amplitude including failures, the amplitude excluding failures as well as the probability of response of EPSC1 at connections displaying PPD were significantly higher than those at connections displaying PPF (Fig. 2D). These data suggest that the efficacy of synaptic transmission is differentially regulated at these two main types of pyramidal-FS connections and led us to hypothesize that, beyond their short-term synaptic behaviors, facilitating and depressing connections exhibit probably different functional and structural properties.

Single and multiple release sites at pyramidal-FS connections

To investigate further the differences between the two main types of pyramidal-FS connections, we determined if they consisted of one or of several functional release sites by comparing the amplitude distributions of EPSC1 and EPSC2. Figure 3 illustrates a pyramidal-FS connection that displayed a PPF. At this connection, the response probabilities at the first and second EPSCs were 0.55 and 0.84, respectively, but there was no significant difference between the cumulative amplitude distributions of the two EPSCs (Fig. 3C; P > 0.05, Mann-Whitney test). This result indicates that the same amount of transmitter was released each time there was a response and therefore suggests that this connection had a single functional release site. In such a case, the mean amplitude of EPSCs excluding failures (potency) corresponds to the quantal size, which had a value of -25 pA at a holding membrane potential of -72 mV (Fig. 3C). We observed that 20% of pyramidal-FS connections displaying PPF consisted of a single functional release site with an average quantal size of -21.9 ± 7.5 pA (n = 5). The release probabilities at these single release site connections had average values of 0.49 ± 0.19 and 0.73 ± 0.21 at the first and second action potential, respectively (n = 5). These values correspond to release probabilities reported at other cortical excitatory connections (Bolshakov and Siegelbaum 1995; Dobrunz and Stevens 1997; Gulyas et al. 1993).

View larger version (18K): [in this window] [in a new window]   Fig. 3. A single quantum of transmitter released at a facilitating pyramidal-FS connection from a 32-day-old rat. A, bottom: the mean amplitude, including failures, of the postsynaptic currents (EPSC1 and EPSC2) recorded in a FS interneuron in response to pairs of action potentials (top) elicited in a presynaptic pyramidal cell with an interspike interval of 50 ms at a stimulation rate of 0.2 Hz. The response probabilities of EPSC1 and EPSC2 were, respectively, 0.55 and 0.84. Note that this connection displayed a PPF with a ratio of 1.6. Traces are average of 183 responses. B: amplitude distribution of EPSC1, including failures, obtained in standard recording conditions. C: cumulative plots of EPSC1 and EPSC2, excluding failure, showing their similar distribution (P > 0.05; Mann-Whitney test). This indicates that a single quantum was released when there was a response. An apparent quantal size of -25 pA at a holding potential of -72 mV could be estimated from the current corresponding to 50% of the cumulative distribution.

For 19 connections displaying PPF and all of the 20 connections displaying PPD, we observed a difference between the cumulative amplitude distributions of the first and second EPSC. This indicates that different amounts of transmitter were released at each presynaptic action potential and thus suggests that these connections had several functional release sites. To determine the quantal size at connections consisting of multiple functional release sites, we decreased the release probability of presynaptic pyramidal neurons by applying 20-50 µM Cd²⁺ in the bath (n = 11) or by lowering the [Ca²⁺]_e from 3 to 0.5 mM (n = 5; see METHODS). Figure 4 illustrates the effects of 50 µM Cd²⁺ at a depressing pyramidal-FS connection. In control conditions, the amplitude distributions of EPSC1 and EPSC2 were significantly different (Fig. 4A2, inset; P < 0.0001; Mann-Whitney test). Indeed, the histogram of EPSC1 was centered around larger values (Fig. 4A2), indicating that more vesicles were released at the first than at the second presynaptic action potential. The application of Cd²⁺ decreased the probabilities of response from 1 to 0.31 for EPSC1 and from 0.99 to 0.22 for EPSC2. In these conditions, the amplitude distributions of the first and the second EPSCs did not differ significantly (Fig. 4B2 and inset; P > 0.05; Mann-Whitney test). This result indicates that only one quantum was released, on average, when 50 µM Cd²⁺ was applied in the bath. The mean amplitude of both EPSCs, excluding failures, corresponded to a mean quantal size of -19 pA at this connection (Fig. 4B2, inset).

View larger version (27K): [in this window] [in a new window]   Fig. 4. Multiple quanta of transmitter released at a pyramidal-FS connection displaying PPD at a stimulation rate of 0.2 Hz from a 33-day-old rat. A1 and B1, bottom: the mean amplitude, including failures, of the postsynaptic currents (EPSC1 and EPSC2) recorded in a FS interneuron in response to pairs of action potentials (top) elicited in a presynaptic pyramidal cell with an interspike interval of 50 ms at a stimulation rate of 1 Hz in control conditions (A1) and when 50 µM Cd2+ was applied in the bath (B1). At a stimulation rate of 1 Hz, this connection displayed a PPD with ratios of 0.58 in control conditions and 0.76 in presence of Cd2+. The response probabilities of EPSC1 and EPSC2 were 1 in control conditions and 0.31 and 0.22, respectively, in 50 µM Cd2+. Traces are averages of 82 and 558 responses for A1 and B1, respectively. A2 and B2: amplitude distributions of EPSC1 and EPSC2, excluding failures, obtained in control conditions (A2) and in the presence of Cd2+ (B2). The histograms and cumulative plots (A2, inset) showed that the distribution of EPSC1 and EPSC2 were significantly different in control conditions (P < 0.0001; Mann-Whitney test), suggesting that multiple quanta were released at this connection. This was confirmed by the reduction in amplitude of both responses in 50 µM Cd2+ (B2, inset). In presence of Cd2+, the amplitude distribution of EPSC1 and EPSC2 were identical (P > 0.05; Mann-Whitney test), indicating that, under these conditions of low release probability, a single quantum was released when there was a response. An apparent quantal size of -19 pA could be estimated from the current corresponding to 50% of the cumulative distribution (B2).

At all of the connections for which we decreased the probability of release of the presynaptic cell by adding external Cd²⁺, the paired-pulse ratio was only slightly affected (Fig. 4, A1 and B1). A lack of effect of Cd²⁺ on short-term synaptic plasticity has already been described in cultured hippocampal neurons (Brody and Yue 2000). However, the paired-pulse response of connections for which we lowered the [Ca²⁺]_e switched from PPD to PPF or displayed an enhanced PPF as expected from previous reports (data not shown). In both experimental conditions, the response probabilities of synaptic currents decreased significantly compared with those in control conditions (P < 0.01; Wilcoxon test). For 8 of the 16 multiple site connections for which we decreased the probability of release, the amplitude distribution of EPSC1 and EPSC2 were not statistically different (P > 0.05; Mann-Whitney test), although the response probabilities differed between the two responses. These observations suggested that, at these connections, a single vesicle was released in the presence of either external Cd²⁺ or in low [Ca²⁺]_e. The average quantal sizes at connections displaying PPD and PPF were -15.3 ± 2.5 pA (n = 4) and 23.9 ± 9.8 pA (n = 4), respectively, and did not differ significantly (mean: -20 ± 7 pA; P > 0.05; Mann-Whitney test). In addition, we did not observe any significant difference when we compared these values with the mean quantal size of the five facilitating connections showing a single functional release site in control conditions (P > 0.05; Mann-Whitney test). Therefore in our recording conditions, pyramidal-FS connections had a similar mean quantal size, which had a value of -20.5 ± 7.7 pA, at a holding potential of -72 mV (n = 13).

Assuming that the quantal sizes at the different release sites of a connection are equivalent, we used this average value of the quantal size to estimate the average quantal content in control conditions at pyramidal-FS connections with multiple release sites. We calculated the ratio of the mean EPSC1 amplitude, including failures, over the mean quantal size of -20.5 pA. The estimated quantal content at depressing connections was 4.1 ± 3.9 (n = 18), significantly different of that at facilitating connections with multiple release sites (1.9 ± 1.5, n = 18; P < 0.04; Mann-Whitney test). We also estimated the total number of release sites of these connections by dividing the amplitude of the largest EPSCs (obtained by averaging the responses corresponding to the upper 10% of the amplitude histograms) by the mean quantal size. Facilitating connections had on average 5.1 ± 3 release sites (n = 18) while depressing synapses had 7.8 ± 5.4 sites (n = 18) and these values did not differ significantly (P > 0.05, Mann-Whitney test).

These data indicate that pyramidal-FS connections displaying PPF have either single or multiple release sites with a low quantal content. In contrast, connections showing PPD showed always multiple functional release sites and a larger estimated quantal content.

Two distinct morphological types of pyramidal cells display different types of inputs onto FS interneurons

Pyramidal neurons with distinct electrophysiological and morphological characteristics have already been described in layer V of the sensory and motor cortices (see DISCUSSION). We therefore analyzed the action potential discharge and used biocytin labeling to characterize both the firing properties and the morphology of pyramidal cells that differed in their paired-pulse behaviors at their connections with FS cells.

First, we studied the intrinsic physiological characteristics of presynaptic pyramidal cells by injecting depolarizing current pulses and selected the more suitable recordings for analysis (Fig. 5, A and B; n = 19). All of recorded presynaptic cells were regular spiking (RS) neurons. These cells discharged action potentials at a relatively rapid rate within the first 40 ms, followed by a strong spike frequency adaptation. As already demonstrated (Franceschetti et al. 1998), during the phase of low-frequency discharge, pyramidal cells fire at a constant rate (Fig. 5A1; RS1 cells) or continue to adapt throughout the current pulse (Fig. 5B1; RS2 cells). We used the plots of the instantaneous discharge frequency during responses to depolarizing current pulses to quantify the degree of adaptation during the late phase of these responses (Fig. 5, A2 and B2; see METHODS). In our sample, the late accommodation was widely distributed as a continuum with values ranging from -8 to 53% (mean value of 15 ± 16%; Fig. 5C). It is worth noting that such a variability has also been reported for regular spiking pyramidal cells in rat auditory cortex (Hefti and Smith 2000). No correlation between the late accommodation and the paired-pulse responses was obtained, probably because of the variable degree of accommodation of RS pyramidal cells (mean value of 11 ± 15% for PPF and 22 ± 16% for PPD; Fig. 5C).

View larger version (20K): [in this window] [in a new window]   Fig. 5. Electrophysiological characterization of presynaptic pyramidal cells at pyramidal-FS connections. A1 and B1: current-clamp recordings of 2 presynaptic pyramidal cells on injection of depolarizing current pulses. Note the difference of frequency adaptation during the late phase of the responses between the RS1 cell (A) and the RS2 cell (B). A2 and B2: analysis of the instantaneous firing frequency of the action potential discharges elicited in the same pyramidal cells by current pulses of increasing intensities (400, 600, 800, 1,000, and 1,200 pA). C: plot of the late accommodation as a function of the paired-pulse ratio.

We then searched for possible morphological differences between pyramidal neurons involved in connections showing PPF and PPD. A presynaptic pyramidal cell was labeled by biocytin in 42 of 47 cases (Fig. 6). In the remaining five slices, the pyramidal neuron was either absent or extremely weakly filled. All were located in layer V except one located in layer VI.

View larger version (84K): [in this window] [in a new window]   Fig. 6. Morphology of layer V pyramidal cells displaying PPF or PPD at their synapses with fast-spiking interneurons. A-C: large pyramidal cells (showing PPF) with a complex apical dendritic tree consisting of many side branches and a rich terminal tuft which reached the pial surface. B: large pyramidal cell where the apical dendrite possessed especially many side branches. Left: an "extended-focus view" of the cell (ca. 200 µm depth); right: full reconstruction. Roman numerals indicate cortical layers. C: largest pyramidal cell of the sample demonstrating a comparably elaborate apical dendritic tree like cell in A. D-F: pyramidal cells (showing PPD) with a simple apical dendritic tree that reached the pial surface. D: one of the smallest pyramidal cells of our sample. E: "extended-focus view" of a pyramidal to FS cell pair (ca. 150 µm depth). Note the pyramidal cell (top right) whose main axon gives off a very fine collateral (arrowheads) which approaches the multipolar interneuron and forms 2 contacts (white arrows). The thin black arrow points to axon initial segment (AIS) of the pyramidal cell. F: full somato-dendritic reconstruction of the pyramidal cell (showing PPD) imaged in E. Similar to the cell in D, it possessed a simple dendritic tree unlike the complex ones in A-C. Scale bars: A-D, and F, 100 µm; E, 25 µm.

To objectively correlate the morphological appearance of presynaptic neurons with the facilitating and depressing nature of the connections, the most completely recovered cells were three-dimensionally and quantitatively reconstructed (n = 16). Technical limitations like weak or absent filling or a significant dendritic tree cut during the slice preparation precluded a classification of all presynaptic cells. We compared 11 different morphological parameters between presynaptic neurons showing distinct paired-pulse characteristics (Table 1). The parameters measured at basal dendrites were indistinguishable between facilitating and depressing connections. In contrast, the number of side branches and total branches as well as the volume of the apical dendrite was significantly higher at pyramidal cells displaying PPF than at those displaying PPD (Table 1). Indeed, six of the eight stained pyramidal cells displaying PPF showed a "complex" dendritic tree arborization (Fig. 6, A-C). In two cases, pyramidal cells showed PPF although the morphology was clearly "simple." In spite of their variability, the majority of stained pyramidal neurons displaying PPF had a "complex" morphological pattern. On the contrary, all of the stained pyramidal cells displaying PPD presented a "simple" dendritic morphology (n = 8; Fig. 6, B and D). These differences were strengthened by the fact that the number of side branches of pyramidal neurons and the paired-pulse ratio obtained in pyramidal-FS connections were correlated with a linear coefficient of regression of 0.53 (P < 0.05). It is worth noting that no obvious morphological differences emerged between pyramidal cells involved in facilitating connections with a single release site (n = 4) and those with multiple release sites (n = 4). Finally, in two cases, the filling appeared to be sufficiently complete in the presynaptic pyramidal cell as well as in the postsynaptic interneuron. These pairs, which showed PPD, consisted of pyramidal cells with a simple dendritic tree arborization and small, multipolar interneurons (Fig. 6, E and F). Although only electron microscopic analysis can truly provide a reliable estimation of the anatomical number of contacts, we observed with light microscopical "extended-focus" reconstructions that, in both cases, the interneuron received two putative axonal contacts on primary dendrites close to the soma from the presynaptic pyramidal cell (Fig. 6E).

Our results indicate that two layer V pyramidal cell types, best distinguished by the complexity of their apical dendritic tree arborization, provide inputs onto FS interneurons that differ in their short-term synaptic characteristics.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We previously reported that pyramidal-FS connections in neocortical slices of young adult rats differ in their short-term synaptic plasticity characteristics (Angulo et al. 1999a,b). The present study shows that connections displaying paired-pulse facilitation consist of one or several functional release sites and involve mainly presynaptic pyramidal cells with a densely ramified apical dendritic tree and a prominent terminal tuft. In contrast, depressing connections always show multiple functional release sites, and the apical dendrite of the presynaptic pyramidal cells has a limited number of side branches and a restricted terminal tuft. Our results thus indicate that different types of pyramidal cells form distinct local circuits with FS interneurons that differ in their structural and functional properties.

Origin of different paired-pulse characteristics at pyramidal-FS connections

The efficacy of synaptic transmission during repetitive activity largely depends on presynaptic mechanisms (see Zucker 1999 for a review; Bellingham and Walmsley 1999; Brody and Yue 2000; Kraushaar and Jonas 2000; Thomson and Bannister 1999). It has been recently reported, however, that an activity-dependent relief of polyamine block at postsynaptic calcium-permeable AMPA receptors, probably lacking GluR2 subunits, can also contribute to the frequency-dependent facilitation of synaptic responses at cortical synapses (Rozov and Burnashev 1999). Neocortical FS interneurons express calcium-permeable AMPA receptors sensitive to polyamines (Angulo et al. 1997, 1999b; Lambolez et al. 1996). However, in the present work, we did not include spermine into the patch pipette to avoid a possible postsynaptic contribution on short-term plasticity due to the effect of polyamines on these receptors. Moreover, in our previous studies of pyramidal-FS connections, we never observed differences between facilitating and depressing synapses either in the kinetics or in the rectification properties of AMPA receptors activated during unitary EPSCs (Angulo et al. 1997, 1999b). Therefore a difference in the subunit composition of postsynaptic AMPA receptors is unlikely to explain the difference in short-term synaptic dynamics of pyramidal-FS connections in the mature neocortex. Other postsynaptic differences are also unlikely if we consider that both facilitating and depressing inputs from different pyramidal cells can be observed in the same FS interneuron of adult rats (Angulo et al. 1999a). In addition, we restricted our analysis to highly stereotyped nonaccommodating FS cells from layer V that have been shown previously to consist mostly of basket cells and express the calcium-binding protein parvalbumin (Angulo et al. 1999a; Cauli et al. 1997, 2000; Kawaguchi and Kubota 1993). In the present study, the six FS interneurons that were sufficiently labeled to be analyzed, showed a typical basket-cell morphology in their somato-dendritic organization, four of them even possessing axonal arbors with perisomatic boutons. Four of them (2 of which showing basket-like formations) received depressing inputs from pyramidal cells, one (with basket-like formations, Fig. 1B) received a facilitating input and the remainder one (with basket-like formations) formed a nondepressing, nonfacilitating connection with its presynaptic pyramidal cell. Although these observations do not exclude completely that a certain morphological or biochemical postsynaptic heterogeneity is related with the two paired-pulse behaviors, it seems that the differences at pyramidal-FS connections are mainly, if not exclusively, determined by different presynaptic mechanisms.

We observed that, among multiple release site synapses, the quantal content of depressing connections was larger than that of facilitating connections. However, the number of functional release sites was not significantly different between these two types of connections. Together, these results suggest a difference in the probability of release between facilitating and depressing connections. If we assume a simple binomial model of synaptic transmission, we can use the quantal size (q) and the number of release sites (n) to calculate the probability of release of connections with multiple release sites. This gives a lower value of p at facilitating synapses (0.36 ± 0.13; n = 18) than at depressing synapses (0.48 ± 0.13, n = 18; P < 0.04, Mann-Whitney test). If true, this difference would favor the classical view that facilitation relies on presynaptic residual calcium and depression on depletion of vesicles (see Zucker 1989, 1999 for reviews). It is worth noting, however, that this parameter is difficult to estimate for multiple release site synapses because we do not know if p is uniform at all release sites (Dobrunz and Stevens 1997; Hessler et al. 1993; Murthy et al. 1997; Rosenmund et al. 1993).

Changes in the quantal parameters or the characteristics of frequency-dependent short-term plasticity have been already reported at central synapses during the postnatal development (see for instance Bolshakov and Siegelbaum 1995; Pouzat and Hestrin 1997; Reyes and Sakmann 1999). In a previous work, we described a developmental maturation switch of the short-term synaptic plasticity characteristics at layer V neocortical pyramidal-FS connections. We showed that in 14- to 20-day-old rats these connections always display depression, whereas in 27- to 36-day-old rats both depressing and facilitating pyramidal-FS connections co-exist (Angulo et al. 1999a). In the present report, we analyzed in more details the functional and structural properties of these connections in young adult animals and observe that the two types of synapses persist even in older animals (at least in 7-wk-old rats). Within our sample, pyramidal-FS connections displayed mainly facilitation (48%) and depression (38.5%) but also, in a minor degree, neither depression nor facilitation (13.5%). Whether this picture represents the final mature state or still a developmental step would be difficult to test experimentally because of the difficulty to perform whole cell recordings of visually identified interneurons in truly adult rats. We cannot exclude, for instance, that depressing pyramidal-FS connections recorded in 28- to 52-postnatal-day-old rats still correspond to immature synapses that finally will become fully mature and facilitating connections later on.

Two morphological classes of layer V pyramidal cells provide different inputs onto FS interneurons

Distinct types of pyramidal cells may be distinguished by their somatodendritic morphology (see Amitai and Connors 1995 for a review). At least two major classes of layer V pyramidal cells, with apical dendrites reaching layer I, co-exist in several cortical areas of rodents (premotor: Yang et al. 1996; motor: this study; Franceschetti et al. 1998; somatosensory: Chagnac-Amitai et al. 1990; Schubert et al. 2001; auditory: Hefti and Smith 2000; visual: Larkman and Mason 1990). Quantitative descriptions of the dendritic branching patterns in these studies have shown that the morphological difference between these two cell types is most prominently related to the complexity of their apical dendritic arborization. The present report provides evidence that these two types of morphologically different layer V pyramidal cells differ in their short-term synaptic behaviors at their synapses onto FS interneurons. Indeed, most of pyramidal cells with a "simple" dendritic tree displayed PPD whereas all of those with "complex" apical dendrites displayed PPF on their target FS cell. It is worth noting that, after the third postnatal week, both "simple" and "complex" pyramidal neurons may show the firing pattern of RS neurons in the motor cortex of the rat (Franceschetti et al. 1998).

Interestingly, the axon collateral arborization of major layer V pyramidal cell types in the sensory and motor cortices expands differently in neocortical layers. The axon collaterals of pyramidal neurons with "complex" apical dendrites project horizontally and are largely limited to layer V/VI, whereas those of cells with "simple" dendritic patterns ramify extensively in supragranular layers and reach layer I in a more or less columnar fashion (Chagnac-Amitai et al. 1990; Franceschetti et al. 1998; Gottlieb and Keller 1997; Hefti and Smith 2000; see also Tseng and Prince 1993). The different patterns of axonal arborization of these two types of pyramidal cells, in addition to their cell type specific input pattern (Schubert et al. 2001), suggest that they are engaged in different intracortical circuits. Although the axonal labeling of presynaptic cells was not sufficient to permit a quantitative reconstruction, a close inspection of the course of the axons suggests that "simple" pyramidal cells possessed a local axonal tree that was preferentially vertically orientated, whereas "complex" pyramidal cells showed a more horizontally organized axonal arbor. This observation together with the functional differences at connections between the two types of pyramidal cells and FS interneurons suggest that facilitating and depressing pyramidal-FS connections subserve distinct and specific roles in the neocortical network. Further studies will be necessary to establish if "complex" and "simple" layer V pyramidal cells, synaptically connected to FS interneurons, also differ by their extra-cortical projections (see for instance Hallman et al. 1988; Kasper et al. 1994).

Physiological implications

The intrinsic membrane properties and the fast kinetics of the EPSCs of FS interneurons make these cells poor temporal integrators that require a synchronous activation of several presynaptic pyramidal cells to reach their action potential threshold (Angulo et al. 1999b). The time window during which this synchronous activity must occur is narrower at depressing than at facilitating connections where the increase of release probabilities during repetitive activity provides some temporal integration. Therefore FS interneurons play a strict role of coincidence detectors when receiving depressing inputs from "simple" pyramidal cells but may integrate the facilitating inputs of "complex" pyramidal cells over longer periods of time. Considering the major role played by FS cells in controlling the activity of cortical outputs (Buzsaki 1997; Cobb et al. 1995; Whittington et al. 1995; see also Angulo et al. 1999b; McBain and Fisahn 2001), the balance of depressing and facilitating connections in these intracortical networks may have important functional implications because it will determine the inhibitory impact of these interneurons on their target cell population and during layer-specific information processing.