Neuropeptides are commonly colocalized with classical neurotransmitters, yet there is little evidence for peptidergic neurotransmission in the mammalian CNS. We performed whole-cell patch-clamp recording from rodent thalamic brain slices and repetitively stimulated corticothalamic fibers to strongly activate NPY-containing GABAergic reticular thalamic (RT) neurons. This resulted in long-lasting (∼10 sec) feedforward slow IPSPs (sIPSPs) in RT cells, which were mimicked and blocked by NPY1 (Y1) receptor agonists and antagonists, respectively, and were present in wild-type mice but absent in NPY-/- mice. NPYergic sIPSPs were mediated via G-proteins and G-protein-activated, inwardly rectifying potassium channels, as evidenced by sensitivity to GDP-β-S and 0.1 mm Ba2+. In rat RT neurons, NPYergic sIPSPs were also present but were surprisingly absent in the major synaptic targets of RT, thalamic relay neurons, where instead robust GABAB IPSPs occurred. In vitro oscillatory network responses in rat thalamus were suppressed and augmented by Y1 agonists and antagonists, respectively. These findings provide evidence for segregation of postsynaptic actions between two targets of RT cells and support a role for endogenously released NPY within RT in the regulation of oscillatory thalamic responses relevant to sleep and epilepsy.
The coexistence of neuropeptides with classical neurotransmitters in neurons has been well established in a number of studies over the last two decades (Hökfelt, 1991; Hökfelt et al., 2000). Important roles for peptide neurotransmitters have been determined, especially from studies of central pattern generators in invertebrate systems (Nusbaum et al., 2001). However, our understanding of the precise physiological roles of endogenous neuropeptides in CNS neurons is largely incomplete for the following reasons: (1) postsynaptic responses of neuropeptides tend to be small, slowly rising, and long lasting (Jan and Jan, 1982; Weisskopf et al., 1993); (2) selective antagonists for peptidergic receptors are generally unavailable (Hökfelt et al., 2000); (3) most neuropeptides cannot be detected electrochemically (Chow and von Rüden, 1995; Whim and Moss, 2001); and (4) neuropeptides are packaged in large dense-core vesicles, the release of which is restricted by limited secretory granule mobility (Burke et al., 1997). Thus, although several dozen neuropeptides have been identified (Hökfelt et al., 2000), there are very few examples showing clear functional effects of endogenously released neuropeptides on neural network activities (for review, see Nusbaum et al., 2001).
It has been proposed that actions of neuropeptides can be classified as either “neuromodulatory” or “neurotransmitter” (Strand, 1999; Hökfelt et al., 2000; Nusbaum et al., 2001). Although a variety of studies in the mammalian CNS have documented that application of exogenous neuropeptides leads to both presynaptic and postsynaptic actions, evidence has been sparse for transmission mediated via naturally released peptides. In fact, compelling evidence so far exists for only a few neuromodulatory actions (Wagner et al., 1993; Weisskopf et al., 1993; Baraban et al., 1997; Kombian et al., 1997). In addition to the above endogenous modulatory actions that influence neurotransmitter release, in a few cases, endogenously released neuropeptides have been shown to directly evoke slow EPSPs in the PNS and spinal cord (Nicoll, 1978; Jan et al., 1979; Kupfermann, 1991). As yet, there is no direct evidence for peptidergic IPSPs in the mammalian CNS.
The intrathalamic circuit has a number of potential advantages for the study of functional responses resulting from release of neuropeptides. NPY is present in the GABAergic neurons of the nucleus reticularis [reticular thalamic (RT) nucleus] (Molinari et al., 1987; Morris, 1989), and NPY receptors (particularly NPY1 receptors) are abundantly expressed at postsynaptic sites on these cells (Molinari et al., 1987; Sun et al., 2001a,b). Reciprocal connections between inhibitory RT and excitatory relay neurons (von Krosigk et al., 1993) promote recurrent network oscillatory activity. Such oscillations are dependent on burst firing in reticular and relay neurons (Huguenard, 1999; McCormick and Contreras, 2001; Steriade, 2001) and can be used as a sensitive index response to detect peptidergic actions (Cox et al., 1997; Sun et al., 2002a). In the experiments reported here, we took advantage of this favorable circuitry, the availability of potent and selective Y1 antagonists (Doods et al., 1995; Sun et al., 2001b), and genetically NPY-deficient mice (Erickson et al., 1996) to provide a demonstration of peptidergic IPSPs, mediated by NPY. We show that peptidergic output of GABA/NPYergic RT neurons is segregated between recurrent and feedforward nerve terminals. Although both RT and thalamocortical (TC) neurons possess postsynaptic NPY1 receptors that are functionally coupled to G-protein-activated, inwardly rectifying potassium (GIRK) channel activation (Sun et al., 2001a,b), RT neurons preferentially use NPY at their recurrent synapses within the reticular nucleus to generate slow IPSPs (sIPSPs), whereas minimal NPYergic effects were detected at synapses on relay neurons, where instead GABAB receptor-mediated synaptic responses predominated. We also show that NPYergic feedback, presumably within RT, provides use-dependent negative feedback of network oscillations.
Materials and Methods
NPY knock-out mice. Mice deficient in NPY were generated by Erickson et al. (1996). All experiments were performed on F2 and F3 129SV/EVJR2448 hybrid males and were performed using a protocol approved by the Stanford Institutional Animal Care and Use Committee.
Thalamic slice preparation. Young [postnatal day 12 (P12)-P16] NPY-/- and NPY+/+ mice of the same background strain (129SV/EVJR2448) and young Sprague Dawley rats were deeply anesthetized with pentobarbital sodium (55 mg/kg) and decapitated. The brains were removed quickly and placed into cold (∼4°C) oxygenated slicing medium. The slicing medium contained (in mm): 2.5 KCl, 1.25 NaH2PO4, 10 MgCl2, 0.5 CaCl2, 26 NaHCO3, 11 glucose, and 234 sucrose. Tissue slices (300-400 μm) were cut in the horizontal plane using a vibratome (TPI, St. Louis, MO), transferred to a holding chamber, and incubated (35°C) for at least 1 hr before recording. Individual slices were then transferred to a recording chamber fixed to a modified microscope stage and allowed to equilibrate for at least 30 min before recording. Slices were continuously superfused with oxygenated physiological saline at the rate of 4.0 ml/min. The physiological perfusion solution contained (in mm): 126 NaCl, 2.5 KCl, 1.25 NaH2PO4, 1 MgCl2, 2 CaCl2, 26 NaHCO3, and 10 glucose. These solutions were gassed with 95% O2/5% CO2 to a final pH of 7.4 at a temperature of 35 ± 1°C.
Whole-cell patch-clamp recording. Whole-cell recordings were obtained using visualized slice patch techniques and a modified microscope (Zeiss Axioskop) with a fixed stage. A low-power objective (2.5 ×) was used to identify the various thalamic nuclei, and a high-power water-immersion objective (40 ×) with Nomarski optics and infrared video was used to visualize individual neurons. Recording pipettes were fabricated from capillary glass obtained from World Precision Instruments (M1B150F-4), using a Sutter Instrument P80 puller, and had tip resistances of 2-5 MΩ when filled with the intracellular solutions below. An Axopatch1A amplifier and an AxoClamp2B amplifier (Axon Instruments, Foster City, CA) were used for voltage-clamp and current-clamp recordings, respectively. Access resistance in whole-cell recordings ranged from 4 to 12 MΩ and was stable (<25% change) during experiments. Current and voltage protocols were generated using pClamp software (Axon Instruments). Patch pipette saline was modified according to Sodickson and Bean (1996) and was composed of (in mm): 100 K-gluconate, 13 KCl, 9 MgCl2, 0.07 CaCl2, 10 EGTA, 10 HEPES, 2 Na2-ATP, and 0.4 Na-GTP, pH adjusted to 7.4, and osmolarity adjusted to 280 mosMl-1. A sharpened bipolar tungsten electrode placed in the internal capsule near RT was used to deliver synaptic stimulation. To quantify sIPSPs, we measured the difference between resting membrane potential and that measured 1000 msec after the last stimulus in the train.
Extracellular multiple-unit activities were recorded using monopolar tungsten electrodes (0.2-2 MΩ; Frederick Haer, Brunswick, ME) and a Grass amplifier (bandwidth, 0.03-3 kHz). All data were digitized (6 kHz) and stored using Axotape software (Axon Instruments). Extracellular stimuli were delivered to the internal capsule through bipolar sharpened tungsten electrodes. To quantify the degree of synchrony and the duration of the intrathalamic oscillations, autocorrelograms were constructed from the extracellular multiple-unit data. The oscillatory activity in the autocorrelograms was quantified by four measures: number of well-defined peaks; time constant of oscillation decay; oscillatory index (OI); and the period of oscillation (Cox et al., 1997; Huntsman et al., 1999; Sun et al., 2002a). A modified Gabor function measured distinct autocorrelogram components, including a nonoscillatory component (Rand), an exponentially decaying cosine function (Osc, with a characteristic time constant of oscillation decay), and an oscillatory period (interburst period). The OI was calculated as (P - V)/P, where P is the amplitude of the autocorrelogram central peak, and V is the amplitude of its first valley.
Immunocytochemistry. Biocytin (0.5%) was included in the pipette solution to label RT neurons during electrophysiological recordings. Brain slices were removed from the recording chamber and fixed in 4% paraformaldehyde in 0.1 m phosphate buffer overnight and transferred to 0.1 m PBS, pH 7.4, for 10 hr. The brain slices were then sectioned at 100 μm with a vibratome. Sections were blocked in 10% normal goat serum and then incubated for 16 hr in buffer containing rabbit NPY1 receptor antibody (1:2000; Dia Sorin). After rinsing in PBS, the sections were incubated in a mixture of Alexa goat anti-rabbit FITC and Alexa goat anti-mouse Texas Red (5 μg/ml each; Chemicon) for 2 hr. Sections were mounted on slides and coverslipped using Vectashield mounting media (Vector Laboratories). Double immunofluorescence was assessed with a laser scanning confocal microscope (Zeiss LSM 510).
Drugs. Drugs were applied focally through a multibarrel microperfusion pipette that was positioned within 1 mm of the cell. Concentrated NPY, Leu31Pro34NPY, and BIBP3226 (Peninsula Laboratories, Belmont, CA) stock solutions were dissolved in ultra-pure water to a final concentration of 0.1 m and stored in a -70°C freezer. Stock solutions were diluted in physiological saline to final concentrations of 100 nm to 1 μm 1 hr before use. GDP-β-S, bicuculline methiodide (BMI), and biocytin apamin were purchased from Sigma (St. Louis, MO).
Statistics. All data are presented as means ± SEM, unless stated otherwise. Unless indicated otherwise, statistical significance was assessed with a two-tailed Student's t test. Probability values of <0.05 were considered statistically significant.
Synaptic stimulus trains elicit long-lasting IPSPs in RT neurons from wild-type, but not NPY, knock-out mice
RT neurons receive glutamatergic corticothalamic and TC inputs (Warren and Jones, 1997; Liu and Jones, 1999) and a disynaptic feedforward inhibition from neighboring RT cells via recurrent GABAergic connections (Scheibel and Scheibel, 1996; Zhang et al., 1997; Huntsman et al., 1999) (Fig. 1A1). Thus, recordings made from RT neurons in TC or horizontal slices reveal monosynaptic AMPA receptor-mediated and NMDA receptor-mediated EPSPs (Warren and Jones, 1997) and disynaptic feedforward GABAA IPSPs. The latter are blocked by BMI (Ulrich and Huguenard, 1995; Kim et al., 1997; Warren and Jones, 1997). We performed whole-cell patch-clamp recordings from visually and electrophysiologically identified (Fig. 1A2) RT neurons in horizontal mouse thalamic brain slices and used 10 μm BMI to block GABAA-mediated recurrent IPSPs (Zhang et al., 1997; Huntsman et al., 1999) and afterhyperpolarizations (AHPs) (Debarbieux et al., 1998). Under these conditions, extracellular stimuli applied to the internal capsule evoked spike bursts in a large population of RT neurons across widespread regions of the slice (Huguenard and Prince, 1994; Sun et al., 2002a).
Because RT cells express high levels of NPY and, thus, are a potential source of synaptically released peptide (Molinari et al., 1987; Morris, 1989), we determined whether repetitive activation of RT cells in the presence of BMI would lead to identifiable electrophysiological actions caused by release of NPY. Stimulus trains applied to the internal capsule (1-5 V, 40-50 μsec; four stimuli at 20-25 Hz; Fig. 1A1, experimental diagram) resulted in multiple spike bursts in RT neurons. Total spike counts increased dramatically as a function of the number of stimuli within the train, because of supralinear summation of the evoked compound EPSPs (Steriade, 2001; Sun et al., 2002a). Typically, a group of two or more of such short stimulus trains (Fig. 1B, individual trains indicated by Δ) were applied at 3.3 Hz to elicit rhythmic burst discharges in RT neurons. After the stimulus train-induced bursting, long-lasting membrane hyperpolarizations (Fig. 1B1, arrow; >5 sec duration; mean peak amplitude, -5.1 ± 0.7 mV; n = 17) were routinely observed in RT cells. These did not arise from the intrinsic properties of these cells, because prolonged action potential trains evoked by maintained or repetitive intracellular current injections evoked either undetectable (see Fig. 5A1) or small and brief AHPs (see Fig. 5A2, inset; amplitude, -1 ± 0.4 mV; duration, 156 ± 33 msec; n = 18). The directly evoked AHPs were small under these conditions, probably resulting from both the high amount of EGTA (10 mm; see Materials and Methods) in the patch pipette solution (Sodickson and Bean, 1996; Chen and Regehr, 1999; Rozov et al., 2001) and the application of BMI-containing perfusate that is known to block the slow AHP in these cells (Debarbieux et al., 1998). Because the stimulus-induced prolonged hyperpolarizations seem to be synaptic in nature, we refer to them here as sIPSPs. Their dependence on stimulus conditions and other physiological properties will be discussed in more detail below.
To determine whether the sIPSPs might result from NPY release, we next tested for their presence in RT neurons of NPY-deficient mice (Erickson et al., 1996). The long-lasting hyperpolarizations were either absent or much smaller in amplitude (-1.1 ± 0.2 mV; n = 16; p < 0.001 vs NPY+/+; Figs. 1B2, open arrow and inset, 3B1). In contrast, other neuronal properties, such as resting membrane potential (-67 ± 0.8 mV, n = 18 in NPY-/- vs -64 ± 1.1 mV, n = 30 in NPY+/+), input resistance (226 ± 10 MΩ, n = 18 in NPY-/- vs 205 ± 11 MΩ, n = 30 in NPY+/+), and spike/burst morphologies were not significantly different between NPY-/- (n = 18) and NPY+/+ (n = 30) mice, suggesting that intrinsic neuronal excitability was unaffected by the knock-out NPY.
These data suggest that the RT cell long-lasting hyperpolarization induced by trains of extracellular stimuli is not an AHP resulting from repetitive firing, but rather is an sIPSP resulting from the action of NPY released during the stimulus.
Exogenous activation of Y1 receptors hyperpolarizes mouse RT cells
We first confirmed that Y1 receptors were present in murine RT neurons and that Y1-dependent responses of these neurons were similar to those previously found in rat (Sun et al., 2001a). Recorded mouse RT neurons were filled with biocytin (Fig. 2A1), and the sections subsequently reacted with Y1 receptor antibody. As shown in Figure 2, A2 and A3, Y1 receptor immunoreactivity was present at high levels in RT, with prominent staining on somata and dendrites of reticular neurons, suggesting a predominantly postsynaptic site of action. In both NPY+/+ and NPY-/- mouse RT neurons, the Y1 agonist Leu31Pro34NPY (100-300 nm) consistently produced membrane hyperpolarizations (NPY+/+: -5 ± 1.5 mV; n = 11 of 11 responding; p < 0.001; Fig. 2B1,B3; NPY-/-: -5.1 ± 1.8 mV; n = 6 of 6 cells responding; p < 0.001; Fig. 2B2,B3) and inhibition of neuronal discharge (data not shown). The degree of hyperpolarization was not significantly different between NPY+/+ and NPY-/- RT cells (p > 0.5; Fig. 2B3), suggesting that the genetic deletion of the endogenous neuropeptide induced little compensatory change in Y1 receptor expression or function. Consistent with these electrophysiological data, the density and expression patterns of Y1 immunoreactivity in sections from thalamus in three NPY-/- mice were similar to those seen in NPY+/+ mice of similar age (data not shown). In other experiments, the Y1 agonist Leu31Pro34NPY (100-300 nm) activated an inwardly rectifying current in both NPY+/+ and NPY-/- RT neurons (Fig. 2C; n = 5 in each group). Therefore, Y1 receptors have similar expression and functional coupling in RT neurons of rats and mice.
sIPSPs result from activation of Y1 receptors coupled to GIRK channels
We have shown previously that the postsynaptic effects of exogenous NPY application on rat RT neurons are blocked by the Y1 antagonist BIBP3226 (Sun et al., 2001a,b). We next examined the effects of BIBP3226 on sIPSPs to determine whether they would be affected similarly, as expected for an NPY-dependent response. RT neuron sIPSPs were blocked reversibly by 1 μm BIBP3226 (control: -5.1 ± 0.7 mV; n = 11; BIBP3226: -1 ± 0.5 mV; n = 11; p < 0.001; Fig. 3A,C) in NPY+/+ mice. In contrast, BIBP3226 had no effect on the very small sIPSPs observed in RT cells of NPY-/- mice (control, -1.1 ± 0.4 mV vs BIBP3226, -0.7 ± 0.3 mV; n = 12; not significant; Fig. 3, compare B2 with B1,C1,C2).
Exogenous Y1 receptor activation has been shown to cause GIRK channel opening in rat thalamic neurons (Sun et al., 2001a,b). Therefore, we tested the effects of a GIRK channel blocker on mouse RT neuron sIPSPs and found that these were eliminated by 0.1 mm Ba2+ (p < 0.001; n = 6; data not shown). We next examined the temporal changes in membrane conductance associated with the NPY-mediated sIPSP in mouse RT neurons. Repetitive hyperpolarizing currents (-40 pA, 50 msec, 10 Hz) were injected intracellularly to provide an assay of input conductance before and after synaptic train stimuli (Fig. 4A1). There was an apparent increase in membrane conductance beginning after the first stimulus train and reaching peak amplitude ∼1 sec after termination of the stimuli. This was followed by a gradual conductance decline over several seconds (Fig. 4A1,A2). The membrane conductance measured 1 sec after the train was increased above control by 67 ± 8% (Fig. 4B1, left; n = 9; p < 0.01; paired t test vs prestimulation) and recovered within 5 sec (Fig. 4B1, left). In NPY+/+ mouse RT neurons, the increase in membrane conductance was abolished by perfusion with 1 μm BIBP3226 but recovered after BIBP3226 washout (Fig. 4B1; n = 9). Figure 4B shows these effects of NPY and BIBP3226 on membrane conductance in the nine RT neurons examined and indicates that Y1 receptors mediate the membrane conductance changes. In NPY-deficient mice, there was no significant membrane conductance increase 1 sec after train termination (Fig. 4B2,B3). Furthermore, BIBP3226 had no effect on the resting membrane resistance in NPY-/- mice (241 ± 13 MΩ in BIBP3226 vs 227 ± 10 MΩ in controls; p > 0.2; n = 9) or on the membrane resistance after the synaptic stimulation (Fig. 4B2,B3; 246 ± 16 MΩ; n = 9; p > 0.2 vs prestimulation).
To provide additional evidence that the effects of BIBP3226 on the sIPSPs were postsynaptic, 200 μm GDP-β-S (a G-protein hydrolysis inhibitor that blocks G-protein activation) was loaded into NPY+/+ mouse RT neurons via the patch pipette solution. In these neurons, sIPSPs were initially detectable (-4.9 ± 0.5 mV; n = 4; Fig. 5D1) but were nearly abolished after ∼30 min of whole-cell recording (0.9 ± 0.5 mV; n = 4; p > 0.1 vs controls; Fig. 4C1,C2). In contrast, RT neurons recorded with normal patch pipette solution revealed robust sIPSPs that could be activated repeatedly at 1 min intervals for at least 1 hr (data not shown). The increased membrane conductance associated with the sIPSP (Fig. 4C1; 3 min; n = 4; p < 0.001 vs prestimulation) was also blocked after 30 min (Fig. 4C3; 30 min; n = 4; p > 0.1 vs prestimulation). Application of BIBP3226 after 30 min of recording from GDP-β-S-loaded cells had no effect on either resting membrane potential or input resistance (Fig. 4C1-C3; n = 4). These results suggest that the BIBP3226-dependent increased membrane conductance associated with the sIPSP is mediated via activation of postsynaptic G-protein-mediated responses.
In summary, stimulation of corticothalamic fibers in vitro strongly activates mouse GABA/NPYergic RT neurons and results in long-lasting (∼10 sec) feedforward Y1 receptor-dependent and GIRK-dependent sIPSPs in the same cells. To directly compare responses mediated by a classical neurotransmitter (GABA) with a neuropeptide (NPY) coreleased from the same presynaptic neurons, we designed experiments to test whether the sIPSP was present in TC relay neurons, the major target of RT cell axons. In these cells, we could then compare the sIPSP with a previously characterized sIPSP mediated by GABAB receptors (Soltesz et al., 1988; Crunelli and Leresche, 1991; von Krosigk et al., 1993; Kim et al., 1997; Ulrich and Huguenard, 1999). Because such experiments would require GABAA receptor blockade with bicuculline to obtain robust synaptic activation of RT cells (see below) and GABAB-dependent IPSPs are small and functionally weak in mouse relay neurons (Warren et al., 1994, 1997; Warren and Jones, 1997), these experiments could not be performed in mice. Instead, we made this comparison in rat thalamic slices, which have a number of advantages, including: (1) thalamic relay neurons express well characterized RT cell-dependent GABAB-dependent IPSPs (Huguenard and Prince, 1994; Ulrich and Huguenard, 1999); (2) rat thalamic slices are amenable to in vitro studies of oscillatory network interactions and their modulation by peptides and other agents (Huguenard and Prince, 1994; Cox et al., 1997; Sun et al. 2002a;), and (3) we have demonstrated previously in rat thalamic RT and ventrobasal (VB) neurons that exogenous NPY application results in activation of both Y1 and Y2 receptors, which are coupled to postsynaptic GIRK channels and presynaptic Ca2+ channels, respectively (Sun et al., 2000a,b).
Target-specific postsynaptic NPYergic synaptic inhibition in rat thalamic slices
We performed simultaneous whole-cell patch-clamp recordings from visualized rat RT and TC neurons to measure postsynaptic responses evoked by the same stimulus delivered to the corticothalamic tract (Fig. 1A1). TC recordings were obtained from neurons in the somatosensory VB complex. As in the mouse, we determined whether repetitive activation of RT cells under conditions of GABAA receptor blockade would lead to identifiable electrophysiological actions because of the release of NPY.
As shown in Figure 5, A2 and B2, a single shock to the internal capsule elicited a spike burst (Fig. 5A2, gray trace) in an RT neuron, followed by a very small but long-lasting (∼2 sec) sIPSP (Fig. 5A2, large arrow). Increasing the numbers of stimuli within the train evoked progressively longer depolarizations, increased numbers of spikes (Fig. 5A2), and larger sIPSPs (Fig. 5A2, large arrow). Repeated stimulus trains (at 3 Hz) elicited repetitive burst discharges and sIPSPs of progressively larger amplitude and duration (Fig. 6A, large arrows). Amplitude and duration of the sIPSPs were linear functions of train number (Figs. 5A, 6A,A2, inset; n = 26 cells). As in the mouse, prolonged action potential trains evoked by either maintained or repetitive intracellular current injections (1 Hz, 200 msec, 300 pA; Fig. 5A2, inset) evoked either undetectable (Fig. 5A1) or small and brief (Fig. 5A2, inset) AHPs in these neurons (amplitude, 1 ± 0.4 mV; duration, 156 ± 33 msec; n = 18).
In contrast to the sIPSPs observed in RT cells, simultaneous intracellular recordings from VB neurons revealed that the same stimulation protocol elicited repetitive medium-duration IPSPs (∼200-300 msec) in relay neurons (Figs. 5B2,6B, open arrows vs Figs. 5A2, 6A, large black arrows denoting sIPSPs). The onset of the bursts in RT neurons immediately preceded the IPSPs in VB relay cells, suggesting that the latter were mediated by neurotransmitter released from RT neurons (Fig. 5, B2 vs A2, dotted vertical line denoting onset of RT burst). Previous studies have shown that such medium-duration IPSPs in relay neurons are mediated by GABAB receptors activated by the release of endogenous GABA from GABAergic neurons (Soltesz et al., 1988; Crunelli and Leresche, 1991; Kim et al., 1997). In contrast to the RT cell sIPSCs, which required several stimulus trains for their robust expression, large GABAB IPSPs occurred after single burst discharges of RT neurons (Fig. 5B2, gray trace). Increasing the number of stimuli within the train evoked progressively larger GABAB IPSPs in relay neurons (Fig. 5B2, open arrow). In contrast to the RT neuron sIPSPs, the GABAB IPSPs in relay neurons did not summate, but rather showed significant attenuation when evoked repetitively (Fig. 6, B2 vs A2, compare insets). Furthermore, a nearly maximal GABAB IPSP was usually evoked with a single train of only four extracellular stimuli (1-5 V, 40-50 μsec, 25 Hz; Fig. 5B2). Even a larger number of stimuli (6-10 trains; n = 6 cells; data not shown) failed to evoke a significant sIPSP. These results show that repetitive firing of RT neurons evoked differential postsynaptic effects in RT and VB cells.
As with mice, we found that the selective NPY1 receptor antagonist BIBP3226 blocked the rat RT neuron sIPSPs (Fig. 7A; n = 27; p < 0.001). Y1-sensitive responses, obtained by subtracting membrane responses recorded in BIBP3226 from those in control conditions, showed the characteristics of the sIPSP (Fig. 7A1, inset). BIBP3226-sensitive IPSPs were characterized by monoexponential decay with a mean time constant of 2.7 ± 0.3 sec (Fig. 7A1; n = 12). Other properties of the rat RT sIPSPs, such as reversal potential (-95 ± 4 mV; n = 7; data not shown), sensitivity to Ba2+ block (0.1 mm; n = 4; data not shown), and occlusion by Y1 agonist Leu31Pro34NPY (200 nm; n = 6; data not shown) were similar to those described for mouse RT neurons. The GABAB receptor antagonist CGP55485 (10 μm) had no significant effect on sIPSPs in four of four RT neurons (-5.3 ± 1.1 mV in control vs -5.0 ± 0.8 mV in CGP55845; p > 0.5; Fig. 7A3, inset), suggesting minimal contribution of GABAB receptors to the sIPSP.
We found little or no evidence for sIPSPs in rat VB neurons (Figs. 5B2, 6B, 7B), despite the fact that exogenous application of NPY and Y1 agonists effectively hyperpolarizes these neurons via the same mechanisms as in RT cells (Sun et al., 2001a). Instead, medium-duration IPSPs were the predominant response, with reversal potentials and time course expected for GABAB receptor-mediated events (Crunelli and Leresche, 1991; von Krosigk et al., 1993; Kim et al., 1997). The latter were blocked by the selective GABAB receptor antagonist CGP55845 (10 μm: n = 15, -0.6 ± 1 mV, p < 0.01 vs controls: -12 ± 1 mV, n = 15; Fig. 7B2,B3), often revealing corticothalamic EPSPs. In contrast to the complete suppression of the sIPSP by the Y1 antagonist BIBP3226 in RT neurons, this compound had little or no effect on the GABAB IPSPs in relay neurons (-10 ± 0.6 mV in BIBP vs -11.2 ± 1 mV in controls; n = 23; p < 0.01; Fig. 7B1,B3). Although no clear sIPSPs were seen in VB neurons, the net late (1000 msec after termination of the stimulus trains) hyperpolarization evoked by stimulus trains was not completely blocked by CGP55845 (Fig. 7B3), and BIBP perfusion produced a small but significant decrease in the response (Fig. 7B3), suggesting a minor (<10%) NPYergic contribution to postsynaptic responses in relay cells.
Endogenous release of NPY modulates thalamic paroxysmal oscillations
In rat thalamic slices bathed with 10 μm BMI to block GABAA receptor-mediated IPSPs, highly synchronized network oscillations (∼3 Hz), resembling the thalamic activity observed during spike-wave seizures (von Krosigk et al., 1993; Huguenard and Prince, 1994; Huguenard, 1999), were elicited by extracellular stimulation of the internal capsule. The synchronized network activities were relatively stable over a period of ≥1 hr under our recording conditions (Cox et al., 1997; Sun et al., 2002a). After recording for a control period of ∼10 min, NPY (100 nm) was applied via bath perfusion, resulting in shortened duration of evoked oscillatory responses (Fig. 8A1,A2,C,D2; Table 1). The number of bursts per oscillation was also reduced (Fig. 8A1,A2,C; Table 1), as were the total number of spikes per episode (Fig. 8C; Table 1) and synchrony as measured by OI (Cox et al., 1997; Huntsman et al., 1999; Sun et al., 2002a). The dominant frequency of the oscillation, measured by fast Fourier transform, dropped from 3.2 ± 0.5 Hz in control conditions to 2.5 ± 0.4 Hz after NPY application (Fig. 8A1,A2; Table 1). These effects were mimicked by the NPY1 receptor agonist Leu31Pro34NPY (100 nm; n = 4; data not shown). The decrease in bursts/oscillation elicited by Leu31Pro34NPY (32 ± 4%; p < 0.05; n = 4; data not shown) was similar to that after NPY application (Fig. 8D2). These results suggest that exogenous activation of NPY receptors, specifically NPY1 receptors, caused significant depression of epileptiform network activities.
The addition of BIBP3226 (30 nm) produced opposite effects on oscillations, including an increase in oscillation duration (Fig. 8B1,B2,C,D1; Table 2), an increase in the number of bursts (Fig. 8B1,B2,C,D1; Table 2), and an increase in total spikes per episode (Fig. 8B1,B2; Table 2). In 7 of 11 experiments, BIBP3226 increased the OI (Fig. 8C; Table 2), the number of spikes per episode, and the duration of oscillation (Fig. 8B1,B2). In the remaining four experiments, BIBP3226 enhanced background activities, so that it was not possible to quantify synchronous oscillatory activity (data not shown). Nevertheless, the total number of spikes per response in these four experiments also increased (Table 2). In general, BIBP3226 also shortened the interburst interval (Fig. 8, B2 vs A2, arrows; n = 5), an effect opposite to that of NPY. In summary, activation and blockade of Y1 receptors produced opposite effects on the thalamic epileptiform oscillations. These findings support the conclusion that NPY is endogenously released during synchronous network oscillations and provides a negative feedback for activity in the thalamic network.
We report three major findings with respect to the action of endogenous NPY on thalamic networks. First, NPY, most likely released from RT cells in which it is highly and uniformly expressed (see below), produces an sIPSP in the same cells. Second, the major output target of RT cells, TC relay neurons, does not generate significant NPYergic sIPSPs, even though Y1 receptors are present in these cells. Instead, in relay neurons, GABAB receptor-dependent medium-duration IPSPs are the predominant response to RT activation, suggesting segregation of neurotransmitter output of the RT cells. Third, antagonism of Y1 receptors results in enhanced thalamic oscillations. Because RT cells are apparently the major target of released NPY, the site of action of the antagonistic network effect is most likely on RT cells. Given that RT neurons are essentially the sole cellular source of NPY in the rodent thalamus (Morris, 1989), and that no NPY immunoreactive fiber staining is observed in the internal capsule in which extracellular NPY-IPSP-evoking stimuli were delivered (data not shown), these data suggest that the main source of endogenously released NPY is the RT neurons themselves.
The novel actions of endogenous NPY reported here differ from the mainly presynaptic actions of peptides described previously in the hippocampus (Wagner et al., 1993; Weisskopf et al., 1993; Baraban et al., 1997), cerebellum (Miyata et al., 1999), and hypothalamus (Kombian et al., 1997). In the thalamus, NPY causes a direct membrane potential change in postsynaptic neurons via activation of Y1 receptors and GIRK channels (Sun et al., 2001a). Although the activation of GIRK channels has been demonstrated to underlie nonpeptidergic classical metabotropic receptor-mediated IPSPs in a variety of CNS neurons (Soltesz et al., 1988; Crunelli and Leresche, 1991; Luhmann and Prince, 1991; Solis and Nicoll, 1992; Dutar et al., 2000), to our knowledge, these results are, thus, the first demonstration of IPSPs produced via an endogenously released neuropeptide in mammalian CNS.
Target-specific actions of endogenously released NPY
Recently, a number of laboratories have shown that synaptic transmission from a single presynaptic neuron can have divergent, target-specific features (Reyes et al., 1998; Toth and McBain, 2000; Thomson, 2000; Rozov et al., 2001). Our results, obtained using simultaneous intracellular recordings from excitatory neurons (VB) and inhibitory neurons (RT) in the thalamus, demonstrate that this target specificity can also occur for peptidergic neurotransmission (Figs. 1, 4). We show that activation of RT neurons generates prominent self-inhibitory NPYergic sIPSPs via their recurrent connections within the nucleus, but not on their extrinsic target neurons in VB, the TC relay cells. This indicates that the inhibitory effects of endogenously released NPY on intrathalamic oscillations occur largely through actions in RT. In contrast, the predominant effect of strong RT activation on relay cells is generation of GABAB receptor-mediated IPSPs that are not detected in RT cells. Such differential effects of classical neurotransmitters and neuropeptides released from the same neurons onto two targets represent a novel form of synaptic transmission in the mammalian nervous system.
This segregation of GABA/NPY output of RT cells is consistent with the concept that the recurrent inhibitory connections within the reticular nucleus may serve to desynchronize thalamic network activity (Huntsman et al., 1999). In thalamic circuits, both spindle-like and seizure-like oscillations depend on synaptic inhibition and rebound excitation (Kim et al., 1997; Ulrich and Huguenard, 1999). Thus, GABAA and GABAB receptor-dependent IPSPs in relay neurons promote oscillations (von Krosigk et al., 1993; Huguenard, 1999; Steriade, 2001; McCormick and Contreras, 2001). However, IPSPs in RT cells can have the opposite effect (Huntsman et al., 1999). The specific localization of functional NPY responses to RT cells would be antioscillatory and antiepileptic. Interestingly, TC cells have been shown to express high levels of functional Y1 receptors (Sun et al., 2001a) and can generate hyperpolarizations because of GIRK current activation after exogenous NPY applications (Sun et al., 2001b). We speculate that the minimally detectable NPYergic sIPSPs in these cells (Fig. 4B3) might be caused by failure to release the peptide in sufficient quantities from RT neuron axon terminals in relay nuclei. In this connection, it is also important to mention the presence of dendrodendritic synapses as potential sites of NPY release in RT, but not VB (Pinault et al., 1995).
Conditions for NPY release
NPY mRNA is present in the majority of the perikarya in the reticular nucleus of the thalamus (Morris, 1989), suggesting that GABAergic reticular neurons also synthesize NPY. Using confocal microscopy, we detected discrete NPY-positive fibers in the reticular nucleus (data not shown). Stimulation of corticothalamic fibers excites large numbers of RT neurons (von Krosigk et al., 1993; Huguenard, 1999), and we show here that this excitation results in presumed disynaptic sIPSPs mediated by endogenous NPY. A relatively high concentration of intraterminal calcium (10-100 μm) may be required to trigger neuropeptide release (Lang et al., 1997; Ohnuma et al., 2001; Whim and Moss, 2001). Therefore, trains of high-frequency stimuli (10-20 Hz) that can result in large increases in Ca2+ within terminals (Zucker and Regehr, 2002) are effective in eliciting neuropeptide release in the CNS (Hökfelt, 2000; Weisskopf et al., 1993; Wagner et al., 1993; Kupfermann, 1991; Muschol and Salzberg, 2000). Our results agree with this suggestion and provide additional information regarding the conditions under which endogenous neuropeptides may be released. We show that graded sIPSCs could be evoked by increasing the number of stimulus trains within a group. A single stimulus train (four stimuli applied to the internal capsule at 25 Hz) was apparently sufficient to release physiologically detectable levels of NPY.
The shortest latency from the onset of such stimuli to detectable actions of endogenous NPY was ∼200-400 msec (Figs. 4A1, 5A2). Considering the rapid release of large dense core vesicles measured directly in some studies (2-4 msec) (Bruns et al., 2000; Whim and Moss, 2001), the time frame for NPY actions in RT is actually quite slow. Several mechanisms might account for this: (1) the release process might be much slower at nerve terminals than in neuroendocrine cells in which previous measurements have been made (Bruns et al., 2000; Whim and Moss, 2001); (2) release of peptide from neurons may require multiple steps including translocation of the large dense core vesicles (Burke et al., 1997); and (3) Y1 receptors may be located distant from the release sites (Jan and Jan, 1982). Release of sufficient endogenous NPY to evoke sIPSCs normally required multiple trains applied at intervals of 300 msec. The significant enhancement of NPYergic postsynaptic actions by groups of stimulus trains (Fig. 6A) has not been reported in experiments on mammalian CNS neurons. In neuroendorine cells and non-neuronal tissues however, short-term enhancement of neuropeptide release was apparent when two or more depolarizing pulses were given at various intervals from 60 to 100 msec (Muschol and Salzberg, 2000; Whim and Moss, 2001).
Physiological consequences of endogenous release of NPY
Our data, obtained using intracellular recordings and synaptic stimulus trains that mimic repetitive burst firing in RT neurons, suggest that endogenously released NPY may modulate network oscillations in at least three ways. First, NPY released during a single synaptically evoked burst may provide negative feedback to limit the burst (data not shown). This could explain the observation that, during paroxysmal oscillations, the addition of BIBP3226 led to enhanced spiking (Fig. 8; Table 2). Second, Y1 receptors mediate a long-lasting sIPSP (duration, 2-10 sec) that may inhibit burst generation during a synchronized network oscillation. This seems to be the case, because BIBP3226 significantly increased the oscillation decay time constant by 46% and increased the number of bursts per oscillation by 26% (Fig. 8; Table 2). Third, persistently elevated NPY levels resulting from endogenous release could increase the resting membrane potentials of a large population of RT neurons, making it less likely that additional RT neurons would be recruited to generate propagating spike-waves or spindle waves (McCormick and Contreras, 2001).
After blockade of endogenous Y1 receptors, we found that trains of internal capsule stimuli evoked a slow membrane depolarization in RT cells (Fig. 7A2) that could trigger action potentials (data not shown). Such slow depolarizing potentials may be mediated via activation of metabotropic glutamate receptors (mGluRs) in RT cells (von Krosigk et al., 1993). We found that the duration of the slow depolarization was ∼200-300 msec after a single synaptic train (data not shown) and 300-2000 msec after three to six trains, each containing four stimuli (Fig. 7A2). These response durations are similar to those reported previously for activation of mGluRs in the thalamus (von Krosigk et al., 1993; Golshani et al., 1998). In addition, preliminary results indicate that the slow depolarizations are blocked by mGluR1 antagonists (Sun et al., 2002b). Together, these findings indicate that both glutamatergic metabotropic excitation and NPYergic inhibition may occur simultaneously at postsynaptic sites on RT cells. We suggest that the balance between these two postsynaptic events is critically important in maintaining the timing and strength of burst discharges in RT neurons and network oscillations.
In our previous studies in the thalamus, we found that Y2 and Y1 receptors are functionally segregated at presynaptic and postsynaptic sites, respectively. However, the physiological roles of Y2 receptors activated by endogenously released NPY, from both recurrent collaterals within RT and feedforward synapses onto TC relay neurons in dorsal thalamus, are not yet known. Exogenous activation of Y2 receptors by a Y2 agonist (NPY3-36) resulted in a reduction of spontaneous and evoked GABA release but had no direct effect on RT neuron excitability (Sun et al., 2001a,b). Because a selective antagonist for Y2 receptors is not generally available, the roles of these receptors, and their activation by endogenous release of NPY, cannot yet be tested. We speculate that Y2 receptors may also serve as autocrine sensors (i.e., when the extracellular NPY concentration is elevated, these receptors would be activated to inhibit additional release of NPY and GABA) (Sun et al., 2001b).
In summary, our results show that endogenous NPY can be released from GABAergic neurons of the RT cells under conditions in which repetitive burst responses are evoked in response to trains of stimuli (Fig. 1), or in the course of rhythmic, pathophysiological oscillations (Fig. 8). The released NPY exerts a direct postsynaptic inhibition (NPYergic IPSP) of RT neurons, but not target relay neurons, via activation of G-protein-coupled receptors and a potassium channel. Furthermore, NPY released endogenously by epileptiform network oscillations causes a self-suppression of such activities in vitro.
This work was supported by National Institutes of Health-National Institute of Neurological Disorders and Stroke Grant NS12151 and the Pimley Research and Training Funds. We are grateful to Michael Nusbaum for critical comments on a previous version of this manuscript and to Isabel Parada for excellent technical assistance in the immunocytochemistry experiments.
Correspondence should be addressed to Dr. John R. Huguenard, Department of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA 94305. E-mail:.
Copyright © 2003 Society for Neuroscience 0270-6474/03/239639-11$15.00/0