One of the several types of ionic currents present in central neurons is the low-threshold, or T-type calcium current (LTCC). This current is responsible for the firing of low-threshold calcium spikes (LTS) and participates in the generation of rhythmic activity and bursts of action potentials in several brain nuclei. We have studied the distribution and properties of pyramidal neurons recorded from the guinea-pig medial frontal cortex that have this calcium current. Pyramidal neurons were recorded in an in vitro slice preparation using either current clamp or single-electrode voltage-clamp recording. Pyramidal neurons that generated LTS or had the LTCC were found only between 500 mm from the pial surface and the white matter (approximately layers V/VI) and were absent in more superficial layers. All pyramidal neurons that fired LTS or had the LTCC were characterized as regular spiking and had some important morphological and physiological differences from the rest of the pyramidal neurons studied. This group of neurons had shorter and less complex apical dendritic arbors, fired action potentials of lower amplitude and longer duration, and were the only type of pyramidal neurons able to generate bursts of action potentials. In addition, the inhibitory synaptic potentials elicited by stimulation of layer I were more powerful in this group of neurons. This research provides new evidence for the presence of the LTCC in subsets of cortical pyramidal neurons, which have specific and well defined morphological and physiological properties.
- calcium currents
- cerebral cortex
- pyramidal neurons
- synaptic potentials
- calcium spikes
- frontal cortex
- intracellular recording
- single-electrode voltage-clamp
Neurons differ in the types and distribution of specific ion channels present on their somata and dendrites. These intrinsic membrane differences are manifest in the shape of individual action potentials, lead to distinctive temporal patterns of repetitive firing, and determine to a large extent the way individual neurons transform synaptic input into spike output (Llinás, 1988;McCormick, 1990). In the cerebral cortex, neurons have been classified according to their intrinsic electrophysiological responses, and these neuronal types have been correlated with distinct morphological types (Connors et al., 1982; McCormick et al., 1985; Chagnac-Amitai et al., 1990) (for review, see Connors and Gutnick, 1990).
The type and distribution of calcium currents in the somato-dendritic neuronal membrane are important determinants of their intrinsic electrical properties. In spite of their presumed importance, and in marked contrast to other neuronal cell types (Hess, 1990;McCleskey, 1994), the properties of voltage-gated calcium currents in neurons of the mammalian cortex have not been fully characterized. This perhaps is not surprising, because the cellular heterogeneity of the cortex and the difficulties associated with identifying specific cell types have limited the study of the intrinsic membrane properties of cortical neurons. Nevertheless, the presence of low-threshold and several types of high-threshold calcium currents has been reported in acutely isolated pyramidal neurons (Sayer et al., 1990; Brown et al., 1993; Sayer et al., 1993; Brown et al., 1994) and in intact pyramidal neurons recorded from cortical slices (Franz et al., 1986; Sutor and Zieglgänsberger, 1987). One of these calcium currents, the low-threshold (or T type) calcium current (LTCC), plays an important role in generating bursts of action potentials and slow rhythmic activity in structures such as the thalamus and inferior olive (for review, see Steriade and Llinás, 1988). LTCC is present only in specific neuronal types in the mammalian CNS (Llinás, 1988). An issue of importance is to study the presence of this calcium current in areas of the CNS such as the cerebral cortex, where multiple neuronal types are present, and to determine whether it is present in all neurons or whether it is restricted to some particular layers or neuronal groups.
This study was undertaken to examine the presence of LTCCs in pyramidal neurons of guinea-pig medial frontal cortex (MFC) and to determine the laminar distribution and the physiological and morphological characteristics of those in which it was present.
Part of this work has been presented in abstract form (Geijo-Barrientos and de la Peña, 1995).
MATERIALS AND METHODS
The experiments were performed on slices of frontal cortex prepared from young adult guinea-pigs (150–350 gm; 3–6 weeks) following methods described elsewhere (Geijo-Barrientos and Pastore, 1995). In brief, the animals were anesthetized with sodium pentobarbital (30–35 mg/kg, i.p.), decapitated, and their brains quickly excised and submerged in ice-cold saline [composition (in mm): 124 NaCl, 5 KCl, 1.2 KH2PO4, 2.4 CaCl2, 1.3 MgSO4, 26 NaHCO3, 10 glucose, pH 7.4, when saturated with 95% O2 + 5% CO2]. A block of tissue, which included the frontal pole of the brain, was dissected out, and 3–4 slices 250–350 μm thick were cut from it in a coronal plane using a Vibratome (Pelco, Redding, CA). The slices were stored in small glass vials where the tissue was submerged in saline continuously bubbled with 95% O2 + 5% CO2 at room temperature. The slices were incubated in these vials for at least 1.5 hr, allowing the tissue to stabilize before recordings were performed.
Intracellular recordings and single-electrode voltage-clamp
A single slice was placed in the recording chamber where it was kept submerged in saline flowing at a rate of 1–2 ml/min at 33–35°C. The recording chamber was mounted on the stage of an upright microscope fitted with a calibrated eyepiece; under a total magnification of 100×, it was possible to measure the position of the microelectrode’s tip relative to the pial surface and the white matter. Intracellular recordings were performed with glass microelectrodes pulled from thick wall borosilicate glass (1.0 mm outer diameter, 0.5 mm wall thickness) on a Sutter P-87 puller (Sutter Instruments, Novato, CA) and filled with 3 m K-acetate (80–150 MΩ); in some voltage-clamp experiments, we used electrodes filled with 2 m CsCl. Intracellular signals were recorded with an Axoclamp 2A (Axon Instruments, Foster City, CA) operating either in the bridge mode for current-clamp recording or in the discontinuous single-electrode voltage-clamp mode (dSEVC); in the latter case, we adjusted the amplifier settings following the procedure suggested by Finkel and Redman (1985). During dSEVC recordings, the output signal of the amplifier headstage was continuously monitored on a separate oscilloscope, and the switching frequency, capacitance neutralization, phase, and gain were adjusted for optimal settlement of the electrode voltage transients. After reducing to a minimum the electrode stray capacitance, the switching frequency was set at 3–4 kHz, and the gain was increased up to 0.8–2.5 nA/mV. The recordings obtained under dSEVC were accepted only if the voltage transient in the electrode settled entirely after each current-injecting cycle and if the membrane potential did not deviate >3–4 mV from the command potential at the peak of the inward transient current described in Results; this deviation was considered acceptable in relation to the small amplitude of these currents (∼0.6 nA). Although the use of the above criteria resulted in a relatively small number of cells recorded under dSEVC, all of them were located at all depths within the cortex, and we did not find any layer-specific bias in the quality of the voltage clamp.
Synaptic potentials were elicited using bipolar stimulation electrodes made of Teflon-coated platinum wire (150 μm diameter) positioned on the pial surface of the slice or directly over layer I. The duration of the stimulus was fixed at either 0.2 or 0.5 msec, and the strength was adjusted to elicit synaptic potentials of maximum amplitude.
The drugs used in this work were tetrodotoxin (TTX) and tetraethylammonium chloride (TEA) from Sigma (St. Louis, MO), nimodipine, ω-conotoxin GVIA, and ω-conotoxin MVIIC from RBI (Natick, MA), and ω-agatoxin IVA from Alomone Labs (Jerusalem, Israel). Nimodipine was made up as a 50 mm stock solution in dimethylsulfoxide and dissolved in extracellular solution to its final concentration. TTX, TEA, and nimodipine were applied dissolved in extracellular solution at the concentrations shown. Stocks of ω-conotoxin GVIA and MVIIC (100 μm) and ω-agatoxin (1 μm) were prepared in saline, and 0.1% cytochromec was added to the stock of ω-agatoxin to prevent its absorption on the walls of the container. A solution containing a mixture of 20 μm ω-conotoxin GVIA and MVIIC and 200 nm ω-agatoxin was made in extracellular solution from the above stocks; this blocking cocktail was applied for at least 3 min by a puffer pipette (10–20 μm of tip diameter) placed over the recording site close to the slice surface; the efflux from the pipette was oriented toward the pial surface and in parallel to the flow of the bulk of the extracellular solution in the recording chamber. The composition of the Ca2+-free extracellular solution was the same as the extracellular solution given above, except that CaCl2 was omitted, the concentration of MgSO4 was increased to 3.7 mm and, in some experiments, 1 mm EGTA was added. Ni2+ was added to the extracellular solution as NiCl2.
The recorded signals were stored on digital audiotape (Biological, Claix, France) and subsequently digitized using an analog-to-digital converter and commercial software (Cambridge Electronic Design, Cambridge, UK); the recordings obtained under voltage-clamp were digitized on-line using computer software that controlled data acquisition and the generation of voltage pulses. Numerical values are given as mean ± SEM (number of cases).
Intracellular staining with Neurobiotin
Neurobiotin was used to identify the morphology of recorded cells (Vector Laboratories, Burlingame, CA). Microelectrode tips were filled with 2–4% Neurobiotin in 1 m KCl, and the tracer was injected using depolarizing current pulses (0.2–0.4 nA, 450 msec, 1 Hz for 3–15 min). Slices were fixed by overnight submersion in 4% paraformaldehyde PBS (0.1 m, pH 7.4). After H2O2 (0.3%) and Triton X-100 (0.6%) pretreatment, the slices were then processed by incubation in a 1:100 dilution of ABC complex (Vector) and by a 0.03% solution of 3,3-diaminobenzidine and 0.005% H2O2. The stained neurons were drawn using a video camera coupled to the microscope and suitable software (Neurograph program from Microptic, Barcelona, Spain).
General electrophysiological properties
The results presented here are from stable intracellular recordings obtained from guinea-pig MFC neurons. Figure1 A shows diagrammatically the cortical area of a guinea-pig frontal slice where the recordings were obtained; this cortical area was agranular (Fig.1 B) and very similar to rat anterior cingulate cortex and infralimbic cortex (Vogt and Peters, 1981; Zilles and Wree, 1985). The 136 cells recorded had resting membrane potentials of −58 ± 0.54 mV and input resistances (measured from their response to small hyperpolarizing current pulses) of 53 ± 2.5 MΩ. The action potential amplitudes were 75 ± 0.84 mV with a duration at half amplitude of 1.5 ± 0.21 msec. Their general electrophysiological properties, studied from their responses to the injection of depolarizing and hyperpolarizing current pulses, are illustrated in Figure 2. The neurons had electrical responses similar to the regular spiking (RS) class of cells described previously in guinea-pig and rat neocortex (Connors et al., 1982;McCormick et al., 1985; Chagnac-Amitai et al., 1990; Connors and Gutnick, 1990). The electrophysiological hallmark of this cell class is tonic firing with frequency adaptation in response to long suprathreshold current pulses (Fig. 2 A1). Of the cells recorded, only 2% had responses similar to fast spiking (FS) or intrinsically bursting (IB) cells; these cells were excluded from the analysis. Although none of the studied neurons produced bursts of action potentials from the resting membrane potential, in many cells we found a response generated from negative membrane potentials that was very similar to the “rebound” response generated by the presence of low-threshold calcium spikes (LTS) in cells from the thalamus or inferior olive (Llinás and Yarom, 1981; Jahnsen and Llinás, 1984). This response consisted of one or two action potentials firing on top of a slow depolarization and appeared only when the membrane potential was held negative to −65 or −70 mV for >200–300 msec (Fig. 2 A2,B1). In the presence of 1 μmTTX, the fast sodium spikes were blocked, and a clear LTS was revealed (Fig. 2 B2). This LTS was similar to responses described in other areas of guinea-pig neocortex (Friedman and Gutnick, 1987; Sutor and Zieglgänsberger, 1987; Montoro et al., 1988) and, although less powerful, similar to the LTS described in thalamus and inferior olive.
Presence of LTS in MFC pyramidal neurons
We could separate two groups of cells within guinea-pig MFC based on the presence of this LTS. Those cells with LTS (LTS+; Fig.3 A) generated a powerful rebound response at the break of a hyperpolarizing current pulse of 450 msec (or longer) applied from a membrane potential of −55 to −60 mV (Fig.3 A1); after the application of TTX (1 μm), the LTS was revealed as a slow spike of 15–25 mV of amplitude (Fig.3 A2) following a current pulse that hyperpolarized the cell below −75 mV. Those cells without LTS (LTS−; Fig. 3 B) had a passive response at the break of the current pulse in control extracellular solution and in the presence of TTX (Fig.3 B1,B2). In response to depolarizing current pulses in the presence of TTX (1 μm) and TEA (5 mm), both groups of neurons generated high-threshold calcium spikes (Fig.3 A3,B3), which were easily differentiated from the LTS given their differences in threshold, amplitude, and time course. Approximately one-half of the recorded neurons had LTS (53% LTS+ vs 47% LTS−); however, the distribution of LTS+ and LTS− neurons within MFC was very different (see below).
LTS is generated by the LTCC
The current underlying the LTS was recorded with dSEVC. In the presence of TTX (1 μm) and TEA (10 mm) used to block voltage-dependent sodium and potassium currents, depolarizing pulses of 100–150 msec were applied from a holding potential of −80 to −90 mV. Figure 4 illustrates representative voltage-clamp recordings taken from an LTS+ neuron (Fig.4 A) and from an LTS− neuron (Fig. 4 B) in response to stepped voltage pulses from −85 to −38 mV; the LTS+ neuron showed a transient inward current that was totally absent in the LTS− neuron. This difference is clear when we compare the averagedI–V plots of nine LTS+ cells and seven LTS− cells (Fig.4 C). In the LTS− neurons, there were no voltage-dependent currents between −60 and −40 mV, whereas in LTS+ neurons there was a large and clear inward current activated in this voltage range. In LTS− neurons, depolarizations to potentials more positive than −35 or −40 mV evoked high-threshold noninactivating (or slowly inactivating) calcium currents (see Fig. 4 C, open circles) that were not studied further. The properties of transient inward currents recorded in LTS+ cells are illustrated in Figure 5. The current started to activate at membrane potentials close to −60 mV (Fig. 5 B) and inactivated completely after 40–60 msec (Fig.5 A,C); this inactivation was voltage-dependent and occurred in a membrane potential range of −65 to −90 mV (Fig.5 D,E). The pharmacological properties of this current are shown in Figure 6. It was blocked after the perfusion of the slice with calcium-free extracellular solution (n = 5; Fig. 6 A) or by low concentrations (50–100 μm) of extracellular Ni2+ (n = 7; Fig. 6 B). Drugs that block high-voltage-activated Ca2+ currents were also tested. Nimodipine applied at concentrations of 3–10 μm (n = 4; Fig.6 C) produced no effect or a slight decrease of the transient current (<15% of the peak value). Finally, a mixture of ω-conotoxin GVIA (20 mm), ω-conotoxin MVIIC (20 μm), and ω-agatoxin IVA (200 nm) did not affect the transient current (n = 3; Fig. 6 D). All these properties are typical of the LTCC recorded in thalamic relay neurons (Coulter et al., 1989; Crunelli et al., 1989), isolated neocortical pyramidal neurons (Sayer et al., 1990, 1993), and isolated immature pyramidal neurons from hippocampus (Thompson and Wong, 1991), including the partial sensitivity to dihydropyridines that, although not a characteristic of LTCCs, has been reported in other mammalian central neurons, including cortical neurons (Akaike et al., 1989; Sayer et al., 1990). The electrophysiological properties of this current were also consistent with the characteristics of the LTS recorded in current clamp.
Distribution of LTS+ and LTS− neurons within MFC
The distribution of LTS+ and LTS− cells within MFC is shown in Figure 7. LTS+ neurons were found only in middle and deep layers of the cortex (starting at 500 μm from the pial surface), being most frequent 700-1200 μm from the pial surface and almost totally absent in a cortical band of 400 μm thickness lying just below layer I (Fig. 7 B). Only 2.5% of the recorded LTS+ neurons were located between 250 and 600 μm (approximately layers II/III), and this was 5% of the recorded cells in these layers, whereas 97.5% were located between 600 and 1300 μm (layers V/VI), which was 68% of the cells recorded in these layers. In contrast, LTS− neurons were found at all depths within MFC, most frequently in superficial layers (Fig. 7 A); note that 50% of LTS− neurons were between 250 and 600 μm and 50% were between 600 and 1300 μm. The distributions of LTS+ and LTS− neurons (Figs.7 A,B) were statistically different (p < 0.0001, Mann–Whitney rank sum test). Figure 7, C and D, shows the distribution of a sample of 36 neurons recorded in voltage clamp and separated into two groups according to the presence or absence of the LTCC (see theinsets in Fig. 7 C,D). These distributions were statistically different (p < 0.002, Mann–Whitney rank sum test) and very similar to the distributions of the LTS+ and LTS− neurons. The distributions of LTS− neurons and neurons without LTCC (Fig. 7 A,C) were not statistically different (compared with the Mann–Whitney rank sum test) and their medians were similar (490 and 570 μm, respectively). The distributions of LTS+ neurons and neurons with T calcium current also were not different (Fig. 7 B,D; medians 815 and 800 μm, respectively).
The distribution of LTS+ and LTS− cells might be distorted by several factors, such as the relative undersampling of neurons in deep layer VI (deeper than 1200–1300 μm) not only because of the lower density of cells and the larger amount of fibers in this area, but also because of the variation in thickness of the MFC in different slices, ranging from 1100 to 1300 μm. Another possible source of error is our assumption that the measured position of the recording electrode through the calibrated eyepiece is an accurate estimate of the position of the neuronal soma. We believe our assumption is correct because the position of the electrode, measured at the time of recording, always matched the position of the soma of Neurobiotin-stained neurons.
Morphological characteristics of LTS− and LTS+ neurons
To study neuronal morphology, 41 cells located at different depths in MFC were intracellularly stained with Neurobiotin. All of them had electrophysiology of the RS type and were pyramidal in shape with clear apical dendrites that reached layer I (Fig. 8). The stained neurons, the cell bodies of which were no deeper than 600 μm from the pial surface (n = 7), had short apical dendrites that branched extensively in layer I. None of them had LTS in the presence of TTX (Fig. 8 A). Those neurons whose cell bodies were located deeper in the cortex also had a clear pyramidal shape, and the general morphological appearance of those with (n = 21) and without (n = 13) LTS was similar (Fig. 8 B,C). However, some important differences, mostly related to the size and morphology of the apical dendrite, were found when measuring morphological parameters. The overall shape and size of the dendritic tree in LTS+ and LTS− neurons was clearly different (Fig. 8 B,C). To quantify these differences, we used several parameters (Table 1); as an estimate of the dendritic size, we measured the total number of branches, the total dendritic length, and the diameter of the main shaft. The values of these parameters showed that apical dendritic trees were shorter and much thinner in LTS+ neurons (Table 1). Furthermore, the apical dendritic tree of LTS+ neurons was less ramified as shown by the parameters used to quantify the degree of branching (number of terminal branches and number of branching points; Table 1 and Fig. 8). These differences were restricted to the apical dendritic tree, because the degree of branching of the basal dendrites was similar in both classes of cells (Table 1). The cells shown in Figure 8, B andC, are representative of the LTS+ and LTS− cells, respectively, because their values of the parameters of dendritic complexity are very similar to the average values of the samples of LTS+ and LTS− cells given in Table 1. In particular, the apical dendrite of the cell in Figure 8 B has 10 terminal branches, 9 branching points, 17 branches, and a total length of 1976 μm, and the values for the apical dendrite of the cell in Figure8 C are 16, 15, 31, and 2697 μm, respectively.
Despite the differences in dendritic structure, the somatic size was similar in both neuronal types: the maximum somatic diameter and somatic area averaged 17.6 ± 0.66 μm and 183.8 ± 11.2 μm2 in LTS+ (n = 21) and 18.4 ± 1.22 μm and 196.9 ± 23.6 μm2 in LTS− neurons (n = 13).
Electrophysiological differences between LTS+ and LTS− neurons
To study other possible differences associated with the presence of a low-threshold spike, we compared the general electrophysiological properties and synaptic responses of LTS+ and LTS− neurons. For this comparison, we selected a sample of 26 LTS+ neurons and 18 LTS− neurons, the cell bodies of which were located at approximately the same depth in the cortex (between 700 and 900 μm from the pial surface, estimated from the position of the recording electrode). Resting membrane potential and input resistance were similar in both groups of neurons, but the action potential characteristics were different: LTS+ neurons fired action potentials of lower amplitude and longer duration than LTS− neurons (Table2).
We also studied the synaptic responses elicited by the stimulation of layer I (subpial region) in the radial column where the impaled cell was located. Figure 9 illustrates the complex sequence of synaptic potentials elicited by stimulation of layer I, with a brief EPSP truncated by a slow IPSP that lasted 300–400 msec (Fig.9 A,B). The IPSP had two components: an early component that reversed between −55 and −60 mV and a late and much more prolonged component that reversed between −70 and −80 mV (Fig. 9 C). This sequence of synaptic potentials is very similar to the response evoked by stimulation of the white matter (Higashi et al., 1991) and intracortical layers (Sutor and Hablitz, 1989). The early and late IPSPs probably were associated with the activation of GABAAand GABAB receptors, respectively (Connors et al., 1988). This synaptic response was very similar in both LTS+ and LTS− neurons, but the amplitude of the late IPSP (measured at −55 mV) was much larger in LTS+ neurons, whereas the amplitude of the early component was similar in both groups (Fig. 9 D). In spite of its amplitude being larger in LTS+ neurons, the late component was always too small to deactivate the LTCC and generate a rebound response. Stimulation of the white matter underlying the site of recording (n = 3) also generated a very similar sequence of early and late IPSPs, but the late component also was unable to generate rebound responses in LTS+ neurons.
The results presented here show that some pyramidal neurons in guinea-pig MFC were able to fire LTS and that these spikes were generated by a voltage-dependent current similar to the LTCC described in peripheral (Carbone and Lux, 1987; Fox et al., 1987) and central neurons (Coulter et al., 1989; Crunelli et al., 1989). Furthermore, the pyramidal neurons able to generate these LTS (and therefore endowed with the LTCC) were very precisely located within MFC, rarely found in layers II/III and very frequently found in layers V/VI. As well as this localization, LTS+ neurons recorded in layers V/VI had some important morphological and physiological differences from those unable to generate LTS. These demonstrate that LTCCs are present in subsets of pyramidal neurons of guinea-pig MFC. The restricted localization of the LTCC appears not to be a general property of calcium currents because other types, such as the L-type, are present in all callosal and colliculus-projecting neurons of the visual cortex (Giffin et al., 1991). The firing of LTS and their generation by LTCC is well documented in other areas of the brain such as the thalamus (Crunelli et al., 1989) and, apart from the guinea-pig MFC described here, the presence of LTS has been reported with no reference to the localization of the cells in human neocortex (Foehring and Waters, 1991), rat neocortex (Sutor and Zieglgänsberger, 1987) guinea-pig parietal neocortex (Friedman and Gutnick, 1987), and guinea-pig visual cortex (Montoro et al., 1988).
The presence of LTCC in the middle and deep layers of the cortex may be implicated in several physiological functions. First, it may be related to the site of origin of some forms of rhythmic activity in the cerebral cortex; thus, in rat neocortex, oscillatory activity in the range of 5 to 10 Hz is originated in layer V pyramidal neurons (Silva et al., 1991). Second, the presence of LTS in some cortical neurons may be related to the thalamocortical pathway, contributing to the synchronization of cortex and thalamus during the generation of slow oscillations (spindles) in slow-wave sleep (for review, see Steriade and Llinás, 1988). In fact, it has been shown that low voltage-activated calcium current is present in different proportions of pyramidal neurons isolated from rat visual cortex, depending on their projection site (Giffin et al., 1991). Finally, the location of pyramidal neurons with LTCC may have a developmental significance because those cells that do not have the current are superficial cells, which differentiate and migrate later in development (for review, seeRakic, 1995).
Electrophysiological properties and generation of bursts
Pyramidal neurons recorded in isocortical areas have been shown to differ in the patterns of action potential firing that they generate in response to the injection of current pulses (Connors et al., 1982;McCormick et al., 1985; Connors and Gutnick, 1990). Most pyramidal neurons (RS neurons) generate a single spike when stimulated with a just-threshold current pulse; as the stimulus amplitude increases, RS cells fire tonically with frequency adaptation. The degree of frequency adaptation is variable among different subsets of RS cells (Chagnac-Amitai and Connors, 1989; Agmon and Connors, 1992). A small population of pyramidal cells (IB cells) generates a burst of action potentials in a stereotyped pattern that consists of three or more fast spikes arising from a slow afterdepolarization. IB cells are rare, found only in certain cortical laminae, and have slight morphological differences from regular spiking cells (Chagnac-Amitai et al., 1990;Mason and Larkman, 1990). The neurons included in our sample had electrophysiological properties homogeneous and consistent with those of the regular spiking type based on their firing pattern (Fig. 2), duration of their action potentials (Table 2), and the absence of bursts generated from the resting membrane potential. Only two or three typical bursting cells were recorded in a sample of more than 250 cells, and the only kind of bursting-like response that we have found in guinea-pig MFC is that associated with the presence of LTS (see Figs. 2 B1, 3 A1). This finding, along with the fact that the distribution of LTS+ cells in MFC resembles that of IB cells in neocortex (Connors and Gutnick, 1990), raises the possibility that in guinea-pig MFC (and in other allocortical areas), the generation of bursts depends only on the LTCC, whereas in cortical areas with isocortical structure, bursting depends on different and perhaps more complex ionic mechanisms. In agreement with this hypothesis, IB cells recorded in layer V of guinea-pig isocortex (sensorimotor cortex) did not have the LTCC (E. de la Peña and E. Geijo-Barrientos, unpublished results).
Morphological and physiological differences between LTS+ and LTS− neurons
The morphological and physiological properties of LTS+ neurons suggest that they form a separate group within the general class of regular spiking cells. In our study, LTS+ cells had apical dendrites thinner, shorter, and less ramified than LTS− cells (Table 1, Fig. 8) and had action potentials of smaller amplitude and longer duration (Table 2) than LTS− cells. These characteristics make LTS+ cells very different from the other neuronal type able to generate bursting, the IB cells, which have almost the opposite characteristics. IB cells have extensive apical and basal dendritic trees, and the thickness of their main apical trunk is almost twice that of RS cells (Chagnac-Amitai et al., 1990). In spite of their different shapes, the input resistance of LTS+ and LTS− neurons was similar. In rat frontal cortex, nonpyramidal neurons that fire LTS have higher input resistance that those that do not (Kawaguchi, 1993), although in this case the difference in input resistance is associated with differences in overall neuronal shape. According to Kawaguchi (1993), pyramidal neurons may also be divided into high and low input resistance groups, but none of them generate LTS or bursts of action potentials.
Differences in IPSPs
Besides the generation of bursts of action potentials, LTCC determines the possibility of rebound firing at the end of inhibitory synaptic potentials. This has been clearly demonstrated in thalamic relay cells, which are able to fire at the end of inhibitory synaptic potentials generated by thalamic reticularis neurons [for example, seeHuguenard and Prince (1994), their Fig. 8]. We never observed this rebound response in LTS+ neurons from guinea-pig MFC. Although in these neurons the late component of the IPSP was larger than in LTS− neurons (see Fig. 9), it was never able to initiate a rebound firing of the LTS. The reason for this may be that the IPSP amplitude was not enough to deinactivate the LTCC or that the repolarization phase of the IPSP was too slow to activate it, given that the degree of activation of LTCC is very sensitive to the slope of the depolarization (Crunelli et al., 1989). Still, it is important to note that the conditions of extracellular stimulation in vitro used in these experiments may not be fully representative of the situation in vivo, where the synchronous activation of a large number of inhibitory fibers may generate IPSPs capable of initiating rebound firing in LTS+ neurons. The difference in inhibitory synaptic potentials correlated to different intrinsic properties is supported by several studies. For example, activation of inhibitory interneurons by focal activation of acetylcholine induces a larger inhibitory effect on RS than on IB cells (McCormick and Prince, 1986). Also, there is some morphological evidence that different classes of pyramidal neurons receive different ratios of excitatory and inhibitory contacts (Hersch and White, 1981).
This work has clearly demonstrated that LTCC is present in a subset of pyramidal neurons in the guinea-pig MFC. These neurons are localized in middle and deep layers of the cortex and have well defined morphological and physiological properties, different from those of other pyramidal neurons in that area. These properties are the size and the degree of branching in the apical dendritic tree, the characteristics of the action potentials and the amplitude of the late component of the inhibitory synaptic potentials evoked by stimulation of layer I.
This work was supported by Grant PB92-0347 from the DGICYT (Spain). We are grateful to Drs. R. Gallego and J. V. Sánchez-Andrés for helpful comments on this manuscript and to Drs. P. Berbel and A. Fairén for advice on morphology. We also thank Messrs. S. Moya and A. Pérez-Vegara for their expert technical assistance.
Correspondence should be addressed to Dr. Emilio Geijo-Barrientos, Departamento de Fisiologı́a, Universidad de Alicante, Apartado 374, E-03080 Alicante, Spain.