An Excitatory GABAergic Plexus in Developing ... - CiteSeerX

An Excitatory GABAergic Plexus in Developing Neocortical Layer 1 R. S. DAMMERMAN, A. C. FLINT, S. NOCTOR, AND A. R. KRIEGSTEIN Department of Neurology and the Center for Neurobiology and Behavior, Columbia University College of Physicians and Surgeons, New York, New York 10032 Received 8 February 2000; accepted in final form 30 March 2000

Dammerman, R. S., A. C. Flint, S. Noctor, and A. R. Kriegstein. An excitatory GABAergic plexus in developing neocortical layer 1. J Neurophysiol 84: 428 – 434, 2000. Layer 1 of the developing rodent somatosensory cortex contains a dense, transient GABAergic fiber plexus. Axons arising from the zona incerta (ZI) of the ventral thalamus contribute to this plexus, as do axons of intrinsic GABAergic cells of layer 1. The function of this early-appearing fiber plexus is not known, but these fibers are positioned to contact the apical dendrites of most postmigratory neurons. Here we show that electrical stimulation of layer 1 results in a GABAA-mediated postsynaptic current (PSC) in pyramidal neurons. Gramicidin perforated patch recording demonstrates that the GABAergic layer 1 synapse is excitatory and can trigger action potentials in cortical neurons. In contrast to electrical stimulation, activation of intrinsic layer 1 neurons with a glutamate agonist fails to produce PSCs in pyramidal cells. In addition, responses can be evoked by stimulation of layer 1 at relatively large distances from the recording site. These findings are consistent with a contribution of the widely projecting incertocortical pathway, the only described GABAergic projection to neonatal cortex. Recording of identified neonatal incertocortical neurons reveals a population of active cells that exhibit high frequencies of spontaneous action potentials and are capable of robustly activating neonatal cortical neurons. Because the fiber plexus is confined to layer 1, this pathway provides a spatially restricted excitatory GABAergic innervation of the distal apical dendrites of pyramidal neurons during the peak period of cortical synaptogenesis.

A prominent feature of the early postnatal rodent somatosensory cortex is a transient, dense GABAergic fiber plexus confined to layer 1 (Del Rio et al. 1992; Lauder et al. 1986). Neurites of intrinsic GABAergic cortical neurons, including a subset of CajalRetzius (CR) cells and neurogliaform cells potentially contribute to this plexus (Cobas et al. 1991; Hestrin and Armstrong 1996; Imamoto et al.1994; Marin-Padilla 1998). The only demonstrated extrinsic source of GABA, however, is a lamina-specific projection from the zona incerta (ZI) (Castro-Alamancos and Connors 1997; Lin et al. 1990). Incertocortical fibers ramify within the upper half of layer 1 where they make contact with the apical dendrites of pyramidal neurons (Lin et al. 1997). This incertocortical pathway develops at the same time as the major thalamocortical pathway from the dorsal thalamus (Catalano et al. 1991; Nicolelis et al. 1995) and both pathways are driven by sensory input from the trigeminal system (Carstens et al. 1990). The layer 1 GABAergic plexus is positioned to provide GABAergic acti-

vation during early stages of synaptogenesis. Whether, or to what degree, these fibers make functional synapses with developing cortical neurons is unknown. Additionally, while the incertocortical pathway contributes fibers to this plexus, the physiological properties of the cells providing this projection have not been examined. Unlike the action of GABA in more mature cortex, GABA released from fibers in layer 1 is likely to have an excitatory effect on developing cortical neurons. Gramicidin perforated patch recordings have demonstrated that GABAA receptor activation produces depolarization of neonatal neurons (Owens et al. 1996). This is due to a high intracellular chloride concentration in immature neurons maintained by an active chloride transport mechanism (Clayton et al. 1998; Rivera et al. 1999). Because of the relatively high intracellular chloride concentration in immature neurons, GABAergic activation in layer 1 would be predicted to depolarize the apical dendrites of pyramidal neurons. Studies of the developing neocortex have indicated a role for membrane depolarization as a necessary step in associative processes including paradigms involving synaptic plasticity and the conversion of “silent” synapses to functional synapses (Feldman et al. 1998; Isaac et al. 1997). Given the presumed paucity of AMPA mediated excitatory synapses in neonatal cortex (Isaac et al. 1997; Kim et al. 1995), the synaptic source for depolarization at this age is unclear. A widespread excitatory GABAergic projection within layer 1 could provide the membrane depolarization necessary to induce long-term changes in synaptic efficacy. Here we characterize the postsynaptic current resulting from selective stimulation of layer 1. We find, in all neonatal pyramidal cells examined, a GABAA-mediated depolarizing response that can often initiate action potential firing. This GABAA-mediated postsynaptic current (PSC) could be elicited by electrical but not chemical stimulation of layer 1 and could be elicited by electrical stimulation at relatively large distances from the recorded cells. These observations suggest that the presynaptic source furnishing the excitatory GABAergic response is provided by fibers of passage rather than intrinsic layer 1 cells. The only described extrinsic GABAergic innervation of developing rodent somatosensory cortex is provided by the ZI. We therefore identified incertocortical projection neurons which could contribute to the presynaptic source of the evoked response, and found them to be physiologically mature at neonatal stages and to exhibit high rates of spontaneous

Address for reprint requests: A. Kriegstein, Dept. of Neurology, Columbia University College of Physicians and Surgeons, 630 West 168th St., Rm. 4-408, New York, NY 10032 (E-mail: [email protected]).

The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

INTRODUCTION

428

0022-3077/00 $5.00 Copyright © 2000 The American Physiological Society

www.jn.physiology.org

GABAERGIC PROJECTION TO DEVELOPING LAYER 1

activity. The GABAergic plexus in layer 1 therefore provides robust, spatially restricted excitation of pyramidal cells at early stages of development. METHODS

Whole cell recording Sprague-Dawley rats were used for all electrophysiological experiments, and the day of birth was defined as postnatal day 0 (P0). Standard methods were used for preparation of vibratome slices as previously described (Blanton et al. 1989). A sapphire blade (Delaware Diamond Knives, Wilmington, DE) was used to maintain the fragile layer 1 in early postnatal slices. The extracellular artificial cerebrospinal fluid (ACSF) solution contained (in mM): 124 NaCl, 2.5 KCl, 1.25 NaH2PO4, 1 MgSO4, 2 CaCl2, 25 NaHCO3, and 20 glucose, pH 7.4 at 25°C, and was bubbled with 95% O2-5% CO2. Borosilicate glass patch clamp electrodes (4 – 8 M⍀) were filled with (in mM): 130 KCl, 5 NaCl, 10 HEPES, 0.4 CaCl2, 1 MgCl2, and 1.1 EGTA, pH 7.3 at 25°C. In some experiments the sodium channel antagonist QX-314 (2-[Diethlyamino]-N-[2,6-dimethlyphenyl]acetamide) (0.1 mM) was included in the intracellular solution. Other experiments were performed using CsCl (130 mM) in place of KCl to block voltage-gated potassium channels. In experiments to test different concentrations of [Cl⫺]pipette, gluconate was substituted for Cl⫺ in the filling solution. Patch clamp recordings were obtained under infrared differential interference contrast videomicroscopy (IR-DIC) using an Olympus BX50WI microscope (Melville, NY). Voltage and current clamp experiments were performed using an EPC-9 patch-clamp amplifier (Heka Electronic, Lambrecht, Germany) controlled by a Macintosh computer running Pulse v. 8.0 software (Heka Electronic). Liquid junction potentials were determined and corrected for using the method of Neher (Neher 1992). Unless otherwise indicated, pyramidal cells were recorded as assessed by the presence of the following under IR-DIC: 1) pyramid-shaped soma, 2) apical dendrite extending to the pial surface, and 3) one or more basal dendrites. Synaptic events were evoked by 200 ␮s stimulation at intensities of 100 – 400 ␮A applied using a concentric bipolar stimulating electrode (World Precision Instruments, Sarasota, FL) or by focal application (pipette diameter ⬇25–35 ␮m) of 200 ␮M L-glutamic acid (Tocris, Ballwin, MO) for 0.5 s. The following drugs (Research Biochemicals International, Natick, MA) were used at the indicated concentrations: 50 ␮M bicuculline, 50 ␮M D(-)-2-amino-5-phosphonovalerate (D-AP5), 1 mM kynurenate, and 10 ␮M 6,7-dinitroquinoxaline-2,3-dione (DNQX). For monophasic events, drug effects were measured as changes in average (at least 3 traces) peak current amplitude. For polyphasic events, evoked currents were integrated and drug effects were measured as changes in average charge. In instances where the stimulus artifact obscured the initial component of an evoked PSC, the latency to onset of the PSC was estimated by extrapolation of a single exponential fit to the uncontaminated PSC rising phase.

Perforated patch recording Perforated patch recordings were obtained as previously described (Owens et al. 1996). Briefly, patch pipettes were filled with a KClbased filling solution (as above), to which gramicidin (Calbiochem, San Diego, CA) was added on the day of the experiment (solution kept on ice). The tip of the electrode was filled with 40 mg/ml gramicidin and the remainder of the electrode was filled with 10 mg/ml gramicidin. Following tight seal formation, access was allowed to develop for 10 –30 min prior to recording.

Retrograde fluorescent microsphere labeling A solution of rhodamine-conjugated microspheres (FluoSpheres, Molecular Probes, Eugene, OR) was injected (1 ␮L) onto the surface

429

of the cortex beneath the meninges of P2/3 postnatal rat pups under hypothermic anesthesia. The surgical wound was closed using New Skin (Medtech, Jackson, WY), and the pups were killed at P6/7 to prepare coronal diencephalic slices for patch clamp recording. The zona incerta was identified at low power (4 ⫻ objective), and cells containing fluorescent microspheres were identified under high power (60 ⫻ objective) prior to patch clamp recording with a filling solution containing 1–2% Lucifer yellow (Sigma, St. Louis, MO). Lucifer yellow and FluoSphere (rhodamine) fluorescence were recorded with a cooled CCD camera (Dage, Stamford, CT) using epifluorescence and appropriate filters during recording.

Calcium imaging Measurements of relative changes in [Ca2⫹]i were made using epifluorescence microscopy of neocortical slices as previously described (Flint et al. 1999). Cells were loaded with the calcium indicator dye fluo-3-AM (Molecular Probes,15 ␮M) and imaged under epifluorescent illumination with appropriate filters. Excitation light exposure was minimized using neutral density filters and a shutter in the light path. Images were acquired with a cooled CCD camera (Dage) using NIH Image software on a Macintosh computer. Changes in fluorescence were converted to ⌬F/F format according to standard calculations (Owens et al. 1996). RESULTS

Whole-cell patch clamp recordings were made from pyramidal neurons in coronal cortical slices obtained from neonatal rats. Bipolar stimulation of layer 1 (Fig. 1A) induced fast monophasic PSCs in recordings from pyramidal cells in layers 2– 6 (n ⫽ 70 cells, 14 cells in each lamina). The layer 1-evoked PSC had a latency to onset of 3.9 ⫾ 0.2 ms (mean ⫾ SE), a latency to peak of 9.0 ⫾ 2 ms, and decay constant ␶ ⫽ 56.7 ⫾ 4.2 ms. The latency of this response is similar to that of the well-characterized, monosynaptic thalamocortical synapse at this early age recorded at room temperature (Kidd and Isaac 1999). During the first postnatal week (P0 –7), the layer 1-evoked PSC was completely and reversibly abolished (⬎95% reduction in peak amplitude) by the GABAA receptor antagonist bicuculline (50 ␮M, n ⫽ 35, Fig. 1A), demonstrating that, in developing layer 1, GABAergic fibers provide the major input to neocortical neurons. We found that the GABAergic layer 1-evoked PSC could follow repetitive stimulation in the 0.1–5 Hz range without signs of fatigue including increased failure frequency or diminished peak amplitude (Fig. 1A). In contrast to our findings in neonates, stimulation of layer 1 in the adult cortex has been reported to produce a predominantly glutamate-mediated response in cortical pyramidal neurons (Cauller and Connors 1994). We therefore examined developmental changes in the response to layer 1 stimulation. There was no glutamatergic contribution to the layer 1-evoked PSC from P0 –7, as the PSC was entirely abolished by bicuculline at these ages (n ⫽ 35, Fig. 1A) and was unaffected by the glutamatergic antagonists DNQX (10 ␮M) and D-AP5 (50 ␮M) (n ⫽ 10). By contrast, in recordings made after P11 (P11–35), we found a consistent contribution to the layer 1-evoked PSC by ionotropic glutamate receptors (n ⫽ 15) as previously described (Cauller and Connors 1994). Detection of a persistent bicuculline-sensitive GABAergic component, however, was possible at these later ages in the presence of the nonselective glutamate receptor antagonist, kynurenic acid (1

430

R. S. DAMMERMAN, A. C. FLINT, S. NOCTOR, AND A. R. KRIEGSTEIN

FIG. 1. During the first postnatal week, stimulation of neocortical layer 1 results in a GABAAmediated postsynaptic current (PSC) in neocortical pyramidal cells in layers 2– 6. A: a schematic drawing of a coronal neocortical slice indicates stimulation and recording sites. Stimulation of layer 1 in P5 slices produces monophasic PSCs that are completely and reversibly blocked by the GABAA receptor antagonist bicuculline methiodide (bic). Each trace is the average of 10 responses recorded in a layer 2/3 pyramidal cell. In the second postnatal week, layer 1 stimulation results in a polysynaptic, polyphasic response in P11 layer 5 pyramidal cells. A bicuculline-sensitive PSC is detected in the presence of AP-5 and DNQX to block ionotropic glutamate receptors. The right panel depicts the ability of the GABAergic PSC at P5 to sustain relatively high rates of repetitive stimulation. Shown above are superimposed consecutive traces (n ⫽ 7) from a single cell at 1 Hz stimulation. The average PSC amplitude (normalized to peak response) for 10 layer 2/3 pyramidal cells at 1 Hz stimulation is plotted below. B: at P5, evoked PSCs were recorded from the same layer 4 neurons (n ⫽ 6) in response to stimulation in layer 1 (stim1) and in layer 4 (stim2) as shown in the schematic. The response to layer 1 stimulation was reversibly blocked by bicuculline, while the response to layer 4 stimulation was reversibly abolished by the glutamate receptor antagonists, AP5 and DNQX.

mM) or the combination of DNQX and D-AP5 (n ⫽ 5 P11 pyramidal cells, n ⫽ 5 P35 pyramidal cells). To determine the relative contribution of ionotropic glutamate and GABAA receptor activation to the layer 1 evoked response, the sodium channel antagonist QX-314 (2-[Diethlyamino]-N-[2,6-dimethlyphenyl]acetamide) was added to the intracellular solution to prevent escape spikes, which were otherwise often observed even at low stimulus intensities. At P11 (n ⫽ 5 pyramidal cells in 3 slices) 82 ⫾ 4% (SD) of the evoked response was reversibly blocked by a combination of DNQX and D-AP5. The response that persisted in the presence of these glutamate antagonists was reversibly blocked (⬎95% reduction in peak amplitude) by bicuculline (Fig. 1A). Thus, during the second postnatal week, the GABAergic component of the layer 1 response represented approximately 18% of the total response. Electrical stimulation of cortical lamina other than layer 1 produces large polysynaptic currents predominantly mediated by glutamate (Burgard and Hablitz 1993; Flint et al. 1997; Kim et al. 1995). This was confirmed in recordings with the stimulation electrode placed within layer 3/4 (Fig. 1B). Evoked GABA receptor-mediated synaptic responses have generally not been described during the first postnatal week (Agmon and O’Dowd 1992; Burgard and Hablitz 1993; Kim et al. 1995; Luhmann and Prince 1991). Low intensity stimulation in the subcortical white matter during the first postnatal week produces a purely glutamatergic synaptic response (Agmon and O’Dowd 1992; Burgard and Hablitz 1993; Kim et al. 1995); however, more intense white matter stimulation can recruit a

GABAA receptor-mediated polysynaptic response which fatigues rapidly on repeated stimulation (Agmon et al. 1996). Unlike the inhibitory effect of GABAA receptor activation in adult cortical neurons, activation of GABAA receptors depolarizes developing postnatal pyramidal cells because immature neurons have relatively high [Cl⫺]i (Owens et al. 1996). The reversal potential of the layer 1-evoked PSC depends on the concentration of chloride in the whole-cell recording pipette (see Fig. 2A), as expected for a response mediated by GABAA receptors. To test whether the GABAergic layer 1-evoked PSC is excitatory, we used the gramicidin perforated patch recording method that leaves [Cl⫺]i undisturbed (Kyrozis and Reichling 1995) (Fig. 2B). Focal stimulation of layer 1 in the first postnatal week produced a depolarizing excitatory postsynaptic potential in all cells recorded (n ⫽ 6/6), and in several cells (4/6) stimulation of this pathway led to action potential discharge (Fig. 2C). Therefore the GABAergic fiber plexus in layer 1 provides an early excitatory input to pyramidal cells in the developing neocortex. To confirm that we were exclusively stimulating fibers in layer 1, we employed a slice preparation in which layer 1 was surgically isolated (Cauller and Connors 1994) (Fig. 3A). In this slice preparation, where only a band of layer 1 connected the stimulus site to the region of cortex containing the recorded cell, GABAergic PSCs persisted (n ⫽ 5/5, Fig. 3A). Stimulation of layer 1 at increasing distances from the recording site (up to 5 mm) in this preparation and in intact slices yielded GABAergic PSCs that were indistinguishable from those ob-

GABAERGIC PROJECTION TO DEVELOPING LAYER 1

431

FIG. 2. High intracellular chloride concentration renders synaptically released GABA excitatory. A: the postsynaptic potential (PSP) reversal depends on [Cl⫺]i. The average reversal potential is shown for the evoked PSP for both high (104 mM, n ⫽ 3) and low (24 mM, n ⫽ 3) chloride concentrations in the recording electrode. The calculated Nernst potential, assuming complete dialysis and ideal space clamp, is indicated by the dashed line. B: gramicidin perforated patch recording (perf. patch) leaves intracellular anion concentrations unperturbed. Stimulation of layer 1 produces a bicuculline-sensitive inward PSC at resting membrane potential (⫺60 mV). To confirm the integrity of the perforated patch, the synaptic response was compared before and after rupturing the seal to obtain a whole cell recording (WCR). The high concentration of chloride in the pipette causes the amplitude of the PSC to increase. C: in several perforated patch recordings as shown in a representative current-clamp recording here, the depolarization caused by a single PSP was sufficient to evoke one or more action potentials.

tained with closer (500 ␮m) stimulation (n ⫽ 10/10, Fig. 3B). This indicates that GABAergic fibers project over relatively long distances in layer 1. Intrinsic GABAergic layer 1 cells, including Cajal-Retzius and neurogliaform cells, have axons that ramify locally and extend up to approximately 700 ␮m within layer 1 (Hestrin and Armstrong 1996; Zhou and Hablitz 1996a). Thus intrinsic cortical neurons are unlikely to contribute significantly to the layer 1 response evoked by stimulation at distances of several millimeters. In a further attempt to examine the contribution of intrinsic

layer 1 neurons to the evoked layer 1 response, we used focal applications of glutamate to stimulate intrinsic cells of layer 1 but not incertocortical axons. We have previously demonstrated that focal application of glutamate in cortical layers other than layer 1 evokes a barrage of GABAergic PSCs in developing cortical pyramidal neurons (Owens et al. 1999). As expected, glutamate applied within 250 –500 ␮m of recorded P5– 6 layer 5 pyramidal cells (n ⫽ 7/7 from 4 slices) resulted in a barrage of PSCs (Fig. 3C). In these same cells, glutamate application at multiple sites in layer 1, however, failed to elicit

FIG. 3. Stimulation of GABAergic fibers of passage within layer 1 produces a PSC in pyramidal cells. A: surgical isolation of layer 1 with a radial cut through layers 2– 6 confirms that stimulation is restricted to horizontal fibers in layer 1. Left: a low power differential interference constrast (DIC) image shows the relative locations of the stimulating electrode (note position of tip), recording electrode, and surgical cut. Right: stimulation of layer 1 fibers in this slice produces PSCs that are completely and reversibly blocked by bicuculline. B: stimulation at two sites separated by 5 mm within layer 1 evokes similar PSCs. The schematic depicts the recording arrangement. The evoked synaptic current was similar at sites 1 and 2 and was reversibly blocked by bicuculline. C: electrical but not chemical stimulation of layer 1 results in a PSC in layer 5 pyramidal neurons (n ⫽ 7/7 cells in 4 slices). Left: the schematic depicts the recording arrangement. Center: voltage clamp recordings from a representative layer 5 pyramidal neuron are shown subsequent to glutamate stimulation at sites 1– 4 or electrical stimulation near site 4. Right: the average PSC frequency (PSC/15 s ⫾ SD, n ⫽ 7 cells) on glutamate stimulation, normalized to baseline frequency (0.2 PSC/s), shows no significant increase above baseline in PSC frequency with stimulation in layer 1 at the sites indicated. Inset: a current clamp recording from the layer 1 cell depicted demonstrates the ability of glutamate stimulation to trigger action potential firing in intrinsic cells of layer 1.

432

R. S. DAMMERMAN, A. C. FLINT, S. NOCTOR, AND A. R. KRIEGSTEIN

an increase in PSC frequency above the baseline frequency of 0.2 PSC/sec, while bipolar electrical stimulation of these same sites in layer 1 resulted in the characteristic monophasic PSC (Fig. 3C). This observation, in conjunction with the ability to evoke a response by distant electrical stimulation, suggests that the presynaptic source eliciting this PSC is provided largely by

GABAergic fibers of passage in layer 1 rather than by intrinsic layer 1 cells (Chai et al. 1988; Sandkuhler et al. 1988). The GABAergic fiber plexus in developing layer 1 contains axon projections from GABAergic neurons of the ZI (Imamoto et al. 1994; Lin et al. 1990). While the physiological properties of neonatal layer 1 neurons have been examined (Hestrin and Armstrong 1996; Kim et al. 1995; Zhou and Hablitz 1996b), the properties of neonatal ZI cells, including incertocortical projection cells, have not been studied. Cytochrome oxidase staining of the ZI suggests that this nucleus has high metabolic activity at early developmental stages (Nicolelis et al. 1995). We therefore assessed the physiological maturity of incertocortical projection neurons by patch clamp recording. Incertocortical projection neurons in the neonatal (P6/7) ZI were identified by two complementary approaches: 1) observation of their characteristic spindle-shaped morphology (Nicolelis et al. 1995) by IR-DIC microscopy and dye-filling with Lucifer yellow, and 2) retrograde labeling of individual incertocortical projection neurons with fluorescent microspheres (Figure 4A) (Katz et al. 1984). Whole cell recordings from spindle-shaped and fluorescent microsphere-labeled incertocortical projection neurons (Fig. 4B) revealed high rates of spontaneous synaptic events, spontaneous action potential discharge, and occasional burst firing (n ⫽ 22, Fig. 4C). ZI neurons had the following mean physiologic membrane properties: Vm ⫽ ⫺48.4 ⫾ 1.9 mV (mean ⫾ SE), Rinput ⫽ 595 ⫾ 60 M⍀, ␶mem ⫽ 73.3 ⫾ 8.9 ms, and dV/dT of the rising phase of the action potential ⫽ 32.4 ⫾ 3.9 mV/ms. Therefore projection neurons of the ZI, like those of the dorsal thalamus (Perez Velazquez and Carlen 1996), are spontaneously active and functionally well-developed by the first postnatal week. We used epifluorescence calcium imaging methods in neonatal brain slices to sample simultaneous activity in large numbers of incertal cells. Consistent with the high levels of local activity inferred from our patch clamp recordings, calcium imaging of fluo-3-loaded slices of the ZI showed high rates of spontaneous calcium transients (Fig. 4D, n ⫽ 200 cells showing multiple calcium FIG. 4. Incertocortical projection neurons were identified for patch clamp recording by retrograde labeling with fluorescent microspheres. A: rhodamineconjugated microspheres were surgically injected onto the surface of cortex and allowed to be transported retrogradely for 3– 4 days in vivo. Acute slice preparations of the diencephalon had high-density accumulations of rhodamine in the dorsal thalamus [primarily in the ventrolateral (VL) and ventrobasal (VB) nuclei] and in the zona incerta (ZI). DIC was used to visualize the diencephalic slice. B: high power micrographs of an incertocortical projection neuron before (left) and during (middle) recording with a Lucifer yellow (LY) containing patch pipette. The superimposed Rhodamine fluorescence and LY images (right) demonstrate co-localization of retrogradely transported rhodamine-conjugated microspheres and LY introduced via the patch pipette. Scale bar indicates 8 ␮m. C: physiology of first week postnatal incertocortical projection neurons. Incertocortical neurons displayed relatively mature membrane properties and a high degree of spontaneous activity. Left: several firing patterns were observed among ZI neurons, including regular spiking cells with variable spike widths (top 2 traces) and single-spiking cells with narrow spike widths (bottom trace). Right: ZI neurons were highly active at rest, displaying burst discharges and regular trains of action potentials (top two traces). A high degree of spontaneous synaptic activity was observed, as shown here with the cell voltage clamped at ⫺60 mV (bottom trace). D: calcium imaging also revealed a high degree of network activity present in the ZI. Upper left: a representative example of an acute slice of the ZI loaded with the calcium indicator dye fluo-3. Right: patterns of spontaneous change in epifluorescence (⌬F/F) are shown for the 3 cells circled on the left. Lower left: calcium transients were reversibly antagonized during bath application of extracellular solution containing 0 mM Ca2⫹, 3 ␮M TTX, 1 mM La3⫹ and 3 mM EGTA as depicted for a single incertal neuron.

GABAERGIC PROJECTION TO DEVELOPING LAYER 1

transients in 4 slices) and confirmed that the ZI is metabolically active in the perinatal period. These transients were completely and reversibly abolished (n ⫽ 100 cells in 2 slices) on bath application of the sodium channel antagonist tetrodotoxin and a combination of agents to prevent extracellular calcium entry (0 Ca2⫹, 1 mM La3⫹, and 3 mM EGTA) suggesting that these calcium transients are mediated by extracellular mechanisms rather than by release from intracellular stores. DISCUSSION

In the present study, we have shown that the GABAergic plexus in rodent developing layer 1 forms functional synaptic contacts with pyramidal neurons from birth. GABAergic synapses in layer 1 provide localized excitatory GABAergic drive onto the distal apical dendrites of immature cortical neurons. The layer 1 GABAergic plexus is composed of axons from intrinsic layer 1 cells and projecting fibers from the ZI. Because GABAergic layer 1 neurons and GABAergic incertocortical neurons exhibit features of physiological maturity by this stage of development, this excitatory drive may help to sculpt early synaptic connections. The relative contribution of incertal neurons and cortical layer 1 interneurons to the GABAergic synapse in layer 1 is difficult to assess. A slice preparation akin to the thalamocortical slice preparation (Agmon and Connors 1991), in which the intact incertocortical projection was preserved, would simplify this analysis. Unlike the thalamocortical projection, however, the more tortuous incertocortical projection cannot be preserved in a 500 ␮m acute slice (R.C.S. Lin, personal communication). Nevertheless, the ability to evoke the layer 1 GABAergic PSC at distances of several millimeters argues against a contribution of intrinsic layer 1 interneurons whose axons have not been reported to extend that far (Hestrin and Armstrong 1996; Zhou and Hablitz 1996a). Responses evoked at these distances appear to support an extrinsic source, in particular the incertocortical pathway, which courses over relatively long distances within layer 1 and is the only described extrinsic GABAergic projection to the neonatal cortex (Lin et al. 1990; Nicolelis et al. 1995). Similarly, failure of glutamatergic stimulation of layer 1 to evoke a PSC in pyramidal cells suggests that the PSC evoked by bipolar stimulation arises from activation of fibers of passage rather than of intrinsic layer 1 cells (Chai et al. 1988; Sandkuhler et al. 1988). Several lines of evidence suggest that the layer 1 GABAergic synapse may play a role in early stages of cortical development. This plexus provides an example of synaptically released GABA exciting neocortical neurons. The excitatory actions of GABA are generally confined to immature cortical neurons which have a high intracellular chloride concentration (Owens et al. 1996). A chloride extruding activity develops postnatally, rendering intracellular chloride low and GABA inhibitory (Rivera et al. 1999). GABAergic synapses in layer 1 develop during corticogenesis, as revealed by GABA immunohistochemistry, with a peak density during neonatal stages and subsequent decrease into adulthood (Lauder et al. 1986; Nicolelis et al. 1995). These observations suggest a possible functional role of GABA in layer 1 to provide a depolarizing influence limited to early postnatal stages of cortical development. Neuroanatomical studies show that incertocortical axons express multiple varicosities, presumed presynaptic specializa-

433

tions, as they course within layer 1 (Lin et al. 1997). Individual incertocortical fibers can thus contact the apical dendrites of multiple pyramidal neurons. This is in marked contrast with the intracortical ramification of specific dorsal thalamic afferents. These predominantly glutamatergic fibers originate in nuclei of the dorsal thalamus, course below the cortical plate, arborize within a restricted radial domain in layers 3 and 4 (CastroAlamancos and Connors 1997), and synapse predominantly onto pyramidal neurons (Elston et al. 1997). These two pathways develop prior to birth (Catalano et al. 1991; Nicolelis et al. 1995), and their nuclei of origin both receive somatosensory information from the trigeminal nucleus (Carstens et al. 1990). Thus activity of specific dorsal thalamocortical afferents will influence local cortical regions, while activity of incertocortical afferents could influence multiple pyramidal cells dispersed throughout the developing cortex. In the somatosensory cortex, the first postnatal week is a critical period for sensory plasticity in the formation of layer 4 “barrels,” the cortical representations of individual whiskers (Woolsey and Wann 1976). Sensory plasticity during the critical period has been shown to be dependent on the NMDA subtype of glutamate receptor (Schlaggar et al. 1993), which can act as a detector of coincident neuronal activity. Protocols shown to induce thalamocortical synaptic plasticity in neonates involve pairing of thalamocortical stimulation with direct experimental depolarization, presumably to relieve the magnesium blockade of NMDA receptors (Crair and Malenka 1995; Feldman et al. 1998; Isaac et al. 1997). However, the native source of depolarization is unknown, as many thalamocortical synapses at this age lack synaptic nonNMDA-type glutamate receptors and are therefore functionally silent synapses at negative membrane potentials (Isaac et al. 1997). In other brain regions it has been suggested that synaptic release of GABA in early development provides the necessary depolarization to induce long-term potentiation at silent glutamatergic synapses (Ben-Ari et al. 1997). During the first postnatal week GABAergic synapses in layer 1 are capable of depolarizing pyramidal cells. Further, the layer 1 evoked GABAergic excitatory postsynaptic current can sustain stimulation at rates exceeding those commonly used to induce synaptic plasticity at this age (Crair and Malenka 1995; Feldman et al. 1998). The GABAergic depolarization in layer 1 may, therefore, act in concert with thalamocortical activation during the critical period for somatosensory plasticity to produce an activity-dependent refinement of thalamocortical circuitry. We thank D. Owens for helpful comments. This work was supported by National Institute of Neurological Disorders and Stroke Grants NS-21223 and NS-35710 and a grant from the Robert Leet and Clara Guthrie Patterson Trust.

REFERENCES AGMON A AND CONNORS BW. Thalamocortical responses of mouse somatosensory (barrel) cortex in vitro. Neuroscience 41: 365–379, 1991. AGMON A, HOLLRIGEL G, AND O’DOWD DK. Functional GABAergic synaptic connection in neonatal mouse barrel cortex. J Neurosci 16: 4684 – 4695, 1996. AGMON A AND O’DOWD DK. NMDA receptor-mediated currents are prominent in the thalamocortical synaptic response before maturation of inhibition. J Neurophysiol 68: 345–349, 1992. BEN-ARI Y, KHAZIPOV R, LEINEKUGEL X, CAILLARD O, AND GAIARSA JL. GABAA, NMDA and AMPA receptors: a developmentally regulated ’menage a trois’. Trends Neurosci 20: 523–529, 1997.

434

R. S. DAMMERMAN, A. C. FLINT, S. NOCTOR, AND A. R. KRIEGSTEIN

BLANTON MG, LOTURCO JJ, AND KRIEGSTEIN AR. Whole cell recording from neurons in slices of reptilian and mammalian cerebral cortex. J Neurosci Methods 30: 203–210, 1989. BURGARD EC AND HABLITZ JJ. Developmental changes in NMDA and nonNMDA receptor-mediated synaptic potentials in rat neocortex. J Neurophysiol 69: 230 –240, 1993. CARSTENS E, LEAH J, LECHNER J, AND ZIMMERMANN M. Demonstration of extensive brainstem projections to medial and lateral thalamus and hypothalamus in the rat. Neuroscience 35: 609 – 626, 1990. CASTRO-ALAMANCOS MA AND CONNORS BW. Thalamocortical synapses. Prog Neurobiol 51: 581– 606, 1997. CATALANO SM, ROBERTSON RT, AND KILLACKEY HP. Early ingrowth of thalamocortical afferents to the neocortex of the prenatal rat. Proc Natl Acad Sci USA 88: 2999 –3003, 1991. CAULLER LJ AND CONNORS BW. Synaptic physiology of horizontal afferents to layer I in slices of rat SI neocortex. J Neurosci 14: 751–762, 1994. CHAI CY, LIN RH, LIN AM, PAN CM, LEE EH, AND KUO JS. Pressor responses from electrical or glutamate stimulations of the dorsal or ventrolateral medulla. Am J Physiol 255: R709 –R717, 1988. CLAYTON GH, OWENS GC, WOLFF JS, AND SMITH RL. Ontogeny of cation-Clcotransporter expression in rat neocortex. Brain Res Dev Brain Res 109: 281–292, 1998. COBAS A, FAIREN A, ALVAREZ-BOLADO G, AND SANCHEZ MP. Prenatal development of the intrinsic neurons of the rat neocortex: a comparative study of the distribution of GABA-immunoreactive cells and the GABAA receptor. Neuroscience 40: 375–397, 1991. CRAIR MC AND MALENKA RC. A critical period for long-term potentiation at thalamocortical synapses. Nature 375: 325–328, 1995. DEL RIO JA, SORIANO E, AND FERRER I. Development of GABA-immunoreactivity in the neocortex of the mouse. J Comp Neurol 326: 501–526, 1992. DIVAC I, KOSMAL A, BJORKLUND A, AND LINDVALL O. Subcortical projections to the prefrontal cortex in the rat as revealed by the horseradish peroxidase technique. Neuroscience 3: 785–796, 1978. ELSTON GN, POW DV, AND CALFORD MB. Neuronal composition and morphology in layer IV of two vibrissal barrel subfields of rat cortex. Cereb Cortex 7: 422– 431, 1997. FELDMAN DE, NICOLL RA, MALENKA RC, AND ISAAC JT. Long-term depression at thalamocortical synapses in developing rat somatosensory cortex. Neuron 21: 347–357, 1998. FLINT AC, DAMMERMAN RS, AND KRIEGSTEIN AR. Endogenous activation of metabotropic glutamate receptors in neocortical development causes neuronal calcium oscillations. Proc Natl Acad Sci USA 96: 12144 –12149, 1999. FLINT AC, MAISCH US, WEISHAUPT J, KRIEGSTEIN AR, AND MONYER H. NR2A subunit expression shortens NMDA receptor synaptic currents in developing neocortex. J Neurosci 17: 2469 –2476, 1997. HESTRIN S AND ARMSTRONG WE. Morphology and physiology of cortical neurons in layer I. J Neurosci 16: 5290 –5300, 1996. IMAMOTO K, KARASAWA N, ISOMURA G, AND NAGATSU I. Cajal-Retzius neurons identified by GABA immunohistochemistry in layer I of the rat cerebral cortex. Neurosci Res 20: 101–105, 1994. ISAAC JTR, CRAIR MD, NICOLL RA, AND MALENKA RC. Silent synapses during development of thalamocortical inputs. Neuron 18: 269 –280, 1997. KATZ LC, BURKHALTER A, AND DREYER WJ. Fluorescent latex microspheres as a retrograde neuronal marker for in vivo and in vitro studies of visual cortex. Nature 310: 498 –500, 1984.

KIDD FL AND ISAAC JT. Developmental and activity-dependent regulation of kainate receptors at thalamocortical synapses. Nature 400: 569 –573, 1999. KIM HG, FOX K, AND CONNORS BW. Properties of excitatory synaptic events in neurons of primary somatosensory cortex of neonatal rats. Cereb Cortex 5: 148 –157, 1995. KYROZIS A AND REICHLING DB. Perforated-patch recording with gramicidin avoids artifactual changes in intracellular chloride concentration. J Neurosci Methods 57: 27–35, 1995. LAUDER JM, HAN VK, HENDERSON P, VERDOORN T, AND TOWLE AC. Prenatal ontogeny of the GABAergic system in the rat brain: an immunocytochemical study. Neuroscience 19: 465– 493, 1986. LIN CS, NICOLELIS MA, SCHNEIDER JS, AND CHAPIN JK. A major direct GABAergic pathway from zona incerta to neocortex. Science 248: 1553– 1556, 1990. LIN RC, NICOLELIS MA, AND CHAPIN JK. Topographic and laminar organizations of the incertocortical pathway in rats. Neuroscience 81: 641– 651, 1997. LUHMANN HJ AND PRINCE DA. Postnatal maturation of the GABAergic system in rat neocortex. J Neurophysiol 65: 247–263, 1991. MARIN-PADILLA M. Cajal-Retzius cells and the development of the neocortex. Trends Neurosci 21: 64 –71, 1998. NEHER E. Correction of liquid junction potentials in patch clamp experiments. Methods Enzymol 207: 123–131, 1992. NICOLELIS MA, CHAPIN JK, AND LIN RC. Development of direct GABAergic projections from the zona incerta to the somatosensory cortex of the rat. Neuroscience 65: 609 – 631, 1995. OWENS DF, BOYCE LH, DAVIS MBE, AND KRIEGSTEIN AR. Excitatory GABA responses in embryonic and neonatal cortical slices demonstrated by gramicidin perforated patch recordings and calcium imaging. J Neurosci 16: 6414 – 6423, 1996. OWENS DF, LIU X, AND KRIEGSTEIN AR. Changing properties of GABA(A) receptor-mediated signaling during early neocortical development. J Neurophysiol 82: 570 –583, 1999. PEREZ VELAZQUEZ JL AND CARLEN PL. Development of firing patterns and electrical properties in neurons of the rat ventrobasal thalamus. Brain Res Dev Brain Res 91: 164 –170, 1996. RIVERA C, VOIPIO J, PAYNE JA, RUUSUVUORI E, LAHTINEN H, LAMSA K, PIRVOLA U, SAARMA M, AND KAILA K. The K⫹/Cl⫺ co-transporter KCC2 renders GABA hyperpolarizing during neuronal maturation. Nature 397: 251–255, 1999. SANDKUHLER J, HELMCHEN C, FU QG, AND ZIMMERMANN M. Inhibition of spinal nociceptive neurons by excitation of cell bodies or fibers of passage at various brainstem sites in the cat. Neurosci Lett 93: 67–72, 1988. SCHLAGGAR BL, FOX K, AND O’LEARY DD. Postsynaptic control of plasticity in developing somatosensory cortex. Nature 364: 623– 626, 1993. WOOLSEY TA AND WANN JR. Areal changes in mouse cortical barrels following vibrissal damage at different postnatal ages. J Comp Neurol 170: 53– 66, 1976. ZHOU FM AND HABLITZ JJ. Morphological properties of intracellularly labeled layer I neurons in rat neocortex. J Comp Neurol 376: 198 –213, 1996a. ZHOU FM AND HABLITZ JJ. Postnatal development of membrane properties of layer I neurons in rat neocortex. J Neurosci 16: 1131–1139, 1996b.